Chromatin-associated protein phosphatase 1 regulates aurora-B and histone H3 phosphorylation.

Proper chromosome condensation requires the phosphorylation of histone and nonhistone chromatin proteins. We have used an in vitro chromosome assembly system based on Xenopus egg cytoplasmic extracts to study mitotic histone H3 phosphorylation. We identified a histone H3 Ser(10) kinase activity associated with isolated mitotic chromosomes. The histone H3 kinase was not affected by inhibitors of cyclin-dependent kinases, DNA-dependent protein kinase, p90(rsk), or cAMP-dependent protein kinase. The activity could be selectively eluted from mitotic chromosomes and immunoprecipitated by specific anti-X aurora-B/AIRK2 antibodies. This activity was regulated by phosphorylation. Treatment of X aurora-B immunoprecipitates with recombinant protein phosphatase 1 (PP1) inhibited kinase activity. The presence of PP1 on chromatin suggested that PP1 might directly regulate the X aurora-B associated kinase activity. Indeed, incubation of isolated interphase chromatin with the PP1-specific inhibitor I2 and ATP generated an H3 kinase activity that was also specifically immunoprecipitated by anti-X aurora-B antibodies. Nonetheless, we found that stimulation of histone H3 phosphorylation in interphase cytosol does not drive chromosome condensation or targeting of 13 S condensin to chromatin. In summary, the chromosome-associated mitotic histone H3 Ser(10) kinase is associated with X aurora-B and is inhibited directly in interphase chromatin by PP1.

Mitotic phosphorylation of histone H3, one of the protein components of the nucleosome (1), has been observed in all eukaryotes so far tested (2)(3)(4)(5). Phosphorylation at Ser 10 of histone H3 is essential in Tetrahymena; cells bearing a mutant histone H3 where Ser 10 is replaced by alanine are unable to fully condense or segregate their chromosomes (6). In cultured cells phosphatase inhibitors stimulate histone H3 phosphorylation and visible chromosome condensation (7,8). Finally, addition of oligonucleosomes to Xenopus in vitro chromosome assembly reactions decreases chromosomal histone H3 phosphorylation and inhibits condensation (9). Taken together, these studies suggest that a balance of kinase and phosphatase activities regulates mitotic phosphorylation of histone H3 and chromosome condensation.
Recently protein kinases of the aurora/Ipl1 family have been implicated in regulating mitotic histone H3 phosphorylation. The aurora protein kinases are essential at various points to promote the proper execution of mitosis (10 -14). Mitotic histone H3 phosphorylation is reduced in a temperature-sensitive mutant of the Saccharomyces cerevisiae aurora homologue Ipl1 grown at the permissive temperature (15). Most multicellular organisms have two or three aurora homologues that may have separate, nonredundant functions (for reviews, see Refs. 16 -18). RNA-mediated interference (RNAi) 1 of the two aurora homologues in Caenorhabditis elegans, air-1 and air-2, has demonstrated that mitotic histone H3 phosphorylation requires AIR-2 but not AIR-1 (15). In Drosophila cultured cells, reduction of aurora-B by RNAi caused a reduction in the level of mitotic histone H3 phosphorylation (12). Thus, the role for aurora-B in mitotic histone H3 phosphorylation appears to be conserved throughout metazoans. However, the nature of this requirement is not yet clear. Mitotic histone H3 phosphorylation first appears in late G 2 phase or in prophase, initiating at centromeres and then progressing along the arms until it is detected along the whole arm in prophase (3,12). The existing data show that aurora-B first localizes to centromeres in prophase and remains concentrated at centromeres until mitosis when, like its binding partner INCENP, it relocalizes to the central spindle and subsequently the spindle midbody (12,19,20). Thus, it remains to be shown that aurora-B is indeed directly associated with the mitotic chromosomal histone H3 kinase or that, if once localized to the centromere, it regulates the activity of another kinase that localizes along the chromosome arm.
Cell cycle-regulated histone H3 Ser 10 phosphorylation must also require at least one protein phosphatase. In S. cerevisiae, genetic supressors of the temperature-sensitive ipl1-1 include specific alleles of glc7, a gene that encodes the catalytic subunit of protein phosphatase 1 (PP1) (21). In both yeast and worms, reduction of PP1 activity partially supresses defects in mitotic histone H3 phosphorylation in ipl1-1 or air-2(RNAi) (15,22). These results are consistent with the suggestion that mitotic histone H3 phosphorylation is the result of a balance of competing kinase and phosphatase activities but do not exclude the possibility that these enzymes regulate each other.
The role of histone H3 phosphorylation may be to modulate interaction of the basic N terminus of histone H3 with DNA. Incubation of isolated chromatin with cAMP-dependent protein kinase causes phosphorylation of histones H3 and H1 and causes an apparent relaxation of the canonical "zigzag" appearance of chromatin in the electron microscope (23). However, the importance of these structural changes in chromosome condensation remains to be determined. Additionally, histone H3 phosphorylation may represent one part of a multifaceted histone modification code (24,25). In this model, post-translational histone modifications might serve as ligands for factors that regulate chromatin dynamics (4,26). Consistent with this model, reducing aurora-B levels in Drosophila prevented the localization of the barren protein (12), a component of the pentameric condensin complex that is required for mitotic chromosome condensation (27). However, a recent biochemical analysis showed that phosphorylation of nucleosomes in vitro does not induce condensin binding, suggesting that condensin targeting may not solely depend on histone H3 phosphorylation (28).
The properties of a number of nonhistone chromosomal proteins are also regulated by phosphorylation. DNA topoisomerase II␣, a 170-kDa DNA strand-passing enzyme required for chromosome condensation, is hyperphosphorylated in G 2 /M phase cells, a modification that stimulates its activity in vitro (29 -32). 13 S condensin, a complex of five proteins required for chromosome condensation, is also specifically phosphorylated in mitotic cell extracts (27,33). Phosphorylation of 13 S condensin by cdc2/cyclin B in vitro stimulates its positive supercoiling activity. In short, mitotic phosphorylation activates condensation factors and correlates with the induction of chromosome condensation. Although numerous kinases and phosphatases have been localized to chromosomes (34 -38), the specific enzymes responsible for mitotic chromosome protein phosphorylation have not been identified.
To study the mechanism of chromosome condensation, we have made use of an in vitro chromosome assembly system based on cytoplasmic extracts of Xenopus laevis eggs (39). We have used this in vitro assembly system to explore the nature of mitotic phosphorylation of chromosomal proteins, in particular histone H3. We have characterized a mitotic chromosome-associated histone H3 kinase activity and shown that this activity is associated with Xenopus X aurora-B. We have also shown that the activity of the enzyme is regulated by phosphorylation and that in addition to antagonizing X aurora-B activity, interphase chromatin-associated PP1 directly inactivates X aurora-B.

EXPERIMENTAL PROCEDURES
Antibodies-Anti-XCAP-C and anti-XCAP-E antisera (kindly provided by Dr. Tatsuya Hirano) were diluted 1:5000 and used for immunoblotting (40). Anti-phospho histone H3 (3) was diluted to 1:1000 for immunoblotting and 1:500 for immunofluorescence. Affinity purified anti-human PP1␥-1 (a gift from Dr. P. T. W. Cohen) (41) was diluted 1:1000 for immunoblotting. Anti-Xenopus PP2A (kindly provided by Dr. T Lee and Dr. T. Stukenberg) was diluted 1:500 for immunoblotting (42). The anti-X aurora-A antibody (Rb788) was generated against a C-terminal peptide (CKNSQLKKKDEPLPGAQ) derived from the sequence of Eg2 (43,44) conjugated to keyhole limpet hemocyanin (Sigma) (45). This antibody recapitulated the published localization of Eg2 (43), recognized a single band of 46 kDa on blots of Xenopus egg cytoplasm, and recognized the in vitro expressed protein (data not shown). The anti-X aurora-B antibody (Rb958) was generated against a fusion of glutathione S-transferase with amino acids 1-95 of X aurora-B/AIRK2 (19). The antibody recognized a single 41-kDa protein on immunoblots of Xenopus egg cytosol and in anti-INCENP immunoprecipitates, localized to centromeres during metaphase and the spindle midbody after completion of mitosis, and recognized the recombinant protein (data not shown).
The clear cytosol was frozen immediately in liquid nitrogen in 50 -200-l aliquots and stored at Ϫ80°C until later use. No difference was noted between the activities of fresh and frozen cytosol. All characterization of the chromatin-and chromosome-associated kinase activities has been performed using both CSF-arrested cytosols and interphase cytosols converted to the mitotic state by the addition of a nondegradable cyclin B (⌬90-cyclin) (48). All results were identical using these two types of cytosols; therefore, we refer to CSF-arrested or cyclin-treated cytosols and chromosomes assembled in them as mitotic cytosol and chromosomes.
In Vitro Chromatin Assembly, Isolation, and Phosphorylation-Interphase chromatin and mitotic chromosomes were prepared as described previously (39,46). It should be noted that the interphase cytosol does not support formation of interphase nuclei because of the absence of membrane components required for nuclear envelope formation. For visualization and immunofluorescence, assembly reactions were fixed with 3.7% CH 2 O and then centrifuged onto polylysine-coated coverslips and stained with anti-Ser(P) 10 histone H3 and 4,6-diamidino-2-phenylindole (Sigma).
To assay chromatin-and chromosome-associated kinase activities, 25 l of mitotic cytosol was diluted with an identical volume of XBE5 (as XBE2, except 5 mM MgCl 2 ) supplemented with an ATP regeneration system to give a final concentration in diluted cytosol of 1 mM ATP (Sigma), 1 mM MgCl 2 , 10 mM phosphocreatine (Roche Molecular Biochemicals), and 0.1 mg/ml creatine kinase (Roche Molecular Biochemicals). Diluted cytosol was centrifuged at 16,000 ϫ g for 10 min at 4°C in a microcentrifuge. Interphase cytosol was treated identically except that it was diluted with XB (10 mM K ϩ -Hepes, pH 7.7, 50 mM sucrose, 100 mM KCl, 1 mM MgCl 2 , 0.1 mM CaCl 2 ) supplemented with the ATP regeneration system. Demembranated Xenopus sperm nuclei (47) were added to diluted cytosol at a final concentration of 6,000 nuclei/l cytosol. No difference in protein labeling or chromatin and chromosome kinase activity was detected in reactions containing 10 -100 l of cytosol or 1000 -6000 sperm nuclei/l cytosol. Assembled chromatin and chromosomes were then isolated through a 30% sucrose in XBE2 (w/v) cushion as described (39,46). The chromatin and chromosomes form an invisible, stable pellet at the bottom of the tube. 10 l of kinase buffer (XBE5 ϩ 0.2 mM ATP) was added to the isolated chromatin or chromosomes and incubated at 18°C for 10 min. For radioactive phosphorylation reactions, the kinase buffer was supplemented with 2 Ci of [␥-32 P]ATP (6000 Ci/mmol; PerkinElmer Life Sciences). Incubation was stopped by the addition of 10 l of 2ϫ SDS-PAGE sample buffer and immediately plunged into liquid nitrogen to partially shear the DNA. After thawing, the reactions were boiled and then separated on either a 15% or 7.5-15% gradient SDS-PAGE gel. The gels were either transferred to nitrocellulose and immunoblotted with anti-phosphohistone H3 antibody (3), or dried and labeled proteins were detected by autoradiography. Kinase inhibitors 6-dimethylaminopurine (Sigma), wortmannin, rapamycin, olomoucine (all from Alexis), PKI (5-24) (a gift of Dr. C. G. W. Smythe), roscovitine (Calbiochem), and butyrolactone I (a gift from M. Kitagawa) were added directly to the kinase reaction from stock solutions.
Isolated Chromatin and Chromosome Elution-Chromatin or chromosomes were assembled and isolated as described above, except that demembranated sperm nuclei were added at 3750 nuclei/l cytosol. Following 30 min of incubation at 18°C, the chromatin assembly reactions were chilled on ice for 10 min. All subsequent steps were carried out on ice or at 4°C. The assembly reaction was then overlaid onto a 30% sucrose cushion in XBE2 (w/v) and centrifuged at 10,000 ϫ g for 10 min at 4°C in a swinging bucket rotor in a microcentrifuge (Eppendorf). The cushion was then washed and removed, and the chromatin was washed with 50 l of CE buffer (0.1% Triton X-100, 0.4 M NaCl in XBE2). The chromatin pellet was recentrifuged, and the interphase chromatin eluate (ICE) or the mitotic chromosome eluate (MCE) was then removed and assayed.
To test the effect of I2 and ATP on chromatin-associated kinase activity, the isolated and washed chromatin pellet was incubated on ice for 30 min in XBE5 supplemented with 600 nM I2 (kindly provided by Dr. P. T. Cohen) (49). 45 l of XBE5 plus 0.4 mM ATP was subsequently added and then incubated at 18°C for 10 min. The supernatant was removed from the chromatin pellet following centrifugation at 10,000 ϫ g for 2 min at 4°C. The pellet was washed with XBE2 and then eluted with CE buffer.
H3 Kinase Assays-H3 kinase activity was determined using a peptide derived from the N terminus of histone H3 (KQTARKSTGGKA-PRK) or a mutant peptide that removes Ser 10 (KQTARKATGGKAPRK). The assays were performed in XBE2 with 0.2 mM ATP, 10 Ci/ml [␥-32 P]ATP, and 5 mM substrate peptide. The reactions were incubated at 22°C for 10 min, spotted onto P81 paper (Whatman), washed with 1% phosphoric acid, and then counted (50).
Immunoprecipitation and Kinase Reactions-For immunoprecipitations from cytosol, rabbit polyclonal antibodies were adsorbed on to protein A-agarose beads (AffiPrep, Bio-Rad), washed with XBE2, and then incubated in undiluted interphase or mitotic cytosols for 1 h at 4°C. The beads were then washed with XBE2 supplemented with leupeptin, chymostatin, and pepstatin (10 g/ml each) and then assayed for H3 kinase activity using a filter binding assay (50). Including microcystin LR or other phosphatase inhibitors in the bead washes gave no change in the level of kinase activity associated with the beads.
To test inactivation of H3 kinase, recombinant PP1␥ (kindly provided by Dr. P. T. Cohen) (41) at 2 milliunits/l was incubated on ice for 10 min in PP1 buffer (50 mM Tris, pH 7.5, 0.1 mM EGTA, 0.03% Brij, 1 mM MnCl 2 ) with or without 1 M microcystin LR. Affi-Prep beads loaded with X aurora-B immunoprecipitated from MCE were washed twice in CE buffer and then three times in PP1 buffer. 5 l of beads were resuspended in 10 l of PP1 with or without microcystin LR and shaken for 10 min at room temperature. The beads were washed, and filter binding assays were performed as described above in the presence of 1 M microcystin LR.

RESULTS
In Vitro Chromosome Assembly-We have used an in vitro chromosome assembly system based on cytoplasmic extracts from Xenopus eggs to study the role of protein phosphorylation in chromosome condensation. Xenopus eggs are arrested in metaphase of meiosis II by the action of CSF, an activity that results from activation of the c-Mos kinase and the mitogenactivated protein kinase pathway in meiosis II (51,52). This arrest is maintained until entry of the sperm nucleus causes a transient increase in intracellular Ca 2ϩ , destruction of CSF, and entry into anaphase of meiosis II. Cytoplasmic extracts prepared in the presence of EGTA preserve CSF and the meiotic metaphase state (47). Treatment of eggs with a Ca 2ϩ ionophore causes the destruction of CSF and entry into interphase; extracts made from these eggs are in interphase. Adding nondegradable cyclin B to the interphase extract generates mitotic extracts. We prepared clarified cytosols of CSF, mitotic, and interphase cytoplasmic extracts by centrifugation at 200,000 ϫ g and, for simplicity, refer to these as "cytosols." When supplemented with demembranated sperm nuclei, these cytosols support the formation of decondensed interphase chromatin or condensed mitotic chromosomes. In vitro assembled chromatin can be recovered quantitatively from these cytosols with a better than 20,000-fold purification from cytosol (40). 2 We have used this system to identify protein kinases and phosphatases present in interphase chromatin and mitotic chromosomes and study the functions these molecules play in chromatin and chromosome structure.
Mitotic Chromosomal Histone H3 Is Phosphorylated by Chromosome-associated Kinases-Paulson and Taylor (53) have shown that chromosomes purified from colchicine-arrested HeLa cells contain kinase activities that phosphorylate histones H1 and H3. To determine whether a similar activity copurified with interphase chromatin and mitotic chromosomes assembled in vitro, we added sperm nuclei to interphase and mitotic extracts, allowed chromatin and chromosomes to form, and then purified the reaction products through a sucrose cushion (Fig. 1A). [␥-32 P]ATP was added to the purified interphase chromatin and mitotic chromosomes to detect any copurifying kinase activity. Histones were then purified by acid extraction and separated using a Triton X-100/acetic acid/urea gel system (54). This method resolves the histones on the basis of charge and hydrophobicity. Comparison of the Coomassiestained gel and the autoradiogram of a representative Triton X-100/acetic acid/urea gel (Fig. 1B) shows that all histones are present in equal amounts in interphase chromatin and mitotic chromosomes, but histones H2A and H4 are phosphorylated in isolated interphase chromatin, whereas histones H3 and H1 are phosphorylated in isolated mitotic chromosomes. The core histones and histone B4, the Xenopus egg linker histone (55), are present in equal amounts in both interphase chromatin and mitotic chromosomes, but the kinase activities present in purified chromatin and chromosomes distinguish between these potential targets at different points in the cell cycle.
To characterize the site of phosphorylation on histone H3, we immunoblotted interphase chromatin and mitotic chromosomes with an anti-Ser(P) 10 histone H3 antibody (3). A significantly higher level of phosphorylation at Ser 10 was detected in isolated mitotic chromosomes than in interphase chromatin (Fig. 1C, lanes 1 and 2). Incubation of isolated chromatin and chromosomes at 22°C in the absence of ATP caused the loss of the mitotic-specific phosphorylation, suggesting the presence of a protein phosphatase in isolated mitotic chromosomes (Fig.  1B, lanes 3 and 4). Incubation in the presence of ATP increased the amount of anti-Ser(P) 10 signal in interphase chromatin but gave a much stronger increase in mitotic chromosomes (Fig.  1B, lanes 5 and 6). A small amount of interphase H3 phosphorylation was also observed by 32 P labeling (Fig. 1B). Neither the interphase chromatin-or mitotic chromosome-associated kinase activities were sensitive to the CDK inhibitors roscovitine or butyrolactone I, wortmannin, H-89, or rapamycin (data not shown). None of these treatments resulted in changes in the total amount of histone protein (Fig. 1C). These data confirm the specificity of histone phosphorylation in isolated chromosomes; not only is histone H3 phosphorylated in a cell cycledependent fashion in isolated mitotic chromosomes, but phosphorylation occurred at a site known to be phosphorylated in mitotic chromosomes in vivo. Most importantly, they indicate that both phosphatase and kinase activities are present in isolated mitotic chromosomes, and it is the sum of their respective activities that combine to generate the final steady state level of phosphorylation on histone H3.
Because PP1 modulates the mitotic phosphorylation of histone H3 in vivo (15,22) and specific isoforms of PP1 are localized to interphase chromatin and mitotic chromosomes (35), we hypothesized that chromatin-associated PP1 might modulate the phosphorylation of chromatin-associated histone H3. We immunoblotted isolated chromatin and chromosomes with an antibody generated against the highly conserved N terminus of human PP1␥ that cross-reacts with all isoforms of the human enzyme (41). The identity of the antigen and Xenopus PP1␥ is 98%, suggesting that the antibody might recognize the Xenopus PP1. A protein of 36 kDa, the correct size for PP1, cross-reacted with the anti-PP1 antibody in both interphase and mitotic cytosols and isolated interphase chromatin and mitotic chromosomes (Fig. 1D). One additional uncharacterized polypeptide of slower mobility was recognized in cytosol. Further inhibitor studies confirmed this localization (see below). In summary, these results show that a cell cycle-specific histone H3 kinase and a phosphatase known to modulate histone H3 phosphorylation are components of isolated mitotic chromosomes.
Elution of Chromosome-associated Kinase-To identify and characterize the mitotic chromosome-associated kinase activity and characterize its regulation, we developed a method for solubilizing chromatin-associated kinase activities compatible with biochemical fractionation and quantitative kinase activity assays. Fig. 2A shows that washing isolated chromosomes with 0.4 M NaCl and 0.1% Triton X-100 diminished the phosphorylation of chromosomal proteins by over 90% (Fig. 2A, lanes 1 and  2). Strikingly, most of the proteins detected by Coomassie staining are retained on the chromosomes ( Fig. 2A, lanes 3 and  4). There are two notable exceptions that are eluted from isolated mitotic chromosomes, histone B4 ( Fig. 2A, bracket), and DNA topoisomerase II ( Fig. 2A, dot; see Ref. 40 for gel band identification). These conditions elute all detectable histone B4 and reproducibly about half of the DNA topoisomerase II. No effect on the content of core histones or the components of the 13 S condensin complex was detected. To allow a quantitative analysis of the kinase activities eluted from chromatin and chromosomes, we assayed NaCl/Triton X-100 eluates from isolated interphase chromatin and mitotic chromosomes in filter binding assays using a peptide derived from the N terminus of histone H3 as a substrate. MCE phosphorylated a histone H3-derived peptide (Fig. 2B); ICE phosphorylated this peptide at least 20-fold less than MCE. Mutation of the serine at position 10 to alanine (S10A) reduced the incorporation of phosphate to background levels (Fig. 2B). Therefore, the activity in MCE contains the same site specificity for histone phosphorylation found in the mitotic chromosome-associated kinase. This characterization demonstrated that MCE possessed the cell cycle-dependent histone H3 kinase activities observed in isolated mitotic chromosomes.
Anti-X aurora-B Antibodies Immunoprecipitate the Chromosomal H3 Kinase-To identify the chromosomal protein kinase that phosphorylates histone H3, we generated rabbit polyclonal antibodies to two Xenopus aurora family members, Eg2/ XlAIRK1/X aurora-A (43) and XlAIRK2/X aurora-B (19) (see "Experimental Procedures"; in concert with recent suggestions (17,18), we will refer to these molecules as X aurora-A and X aurora-B, respectively). Both antibodies were bound to protein A-conjugated beads and used for immunoprecipitation reactions. We first incubated anti-X aurora-A or anti-X aurora-B To detect proteins phosphorylated by chromatin or chromosome-associated protein kinases, chromatin is assembled first in cytosol containing no label, isolated, and then incubated in buffer containing [␥-32 P]ATP (see "Experimental Procedures"). B, cell cycle-dependent histone phosphorylation in isolated interphase chromatin and mitotic chromosomes. Interphase chromatin and mitotic chromosomes were assembled in vitro, isolated, incubated with [␥-32 P]ATP, and then acid extracted to isolate histones. These were run on 15% Triton X-100/acetic acid/urea gel. The gel was stained with Coomassie Blue (lanes 1 and 2) and autoradiographed (lanes 3 and 4). The migration of standard histone markers is indicated. C, phosphorylation of Ser 10 histone H3 in isolated chromatin and chromosomes. Isolated chromatin and chromosomes were either boiled immediately after isolation in gel sample buffer (first and second lanes) or incubated at 22°C for 10Ј in the absence (third and fourth lanes) or the presence of ATP (fifth and sixth lanes). Isolated interphase chromatin and mitotic chromosomes were separated by SDS-PAGE, transferred to nitrocellulose, stained with Sypro Orange (Total Histones), and then immunoblotted with anti-Ser(P) 10 histone H3 (Phospho H3). In the gel system used, H2A, H3, and H2B comigrate in the upper, brighter band observed by Sypro Orange. Lanes I, interphase chromatin; lanes M, mitotic chromosomes. D, interphase chromatin and mitotic chromosomes contain PP1. Interphase (I) and mitotic (M) cytosol (Cytosol) and chromatin isolated from interphase and mitotic cytosols (Chromatin) were separated by SDS-PAGE and immunoblotted with an antibody specific for PP1 (41). beads in interphase and mitotic cytosols and then assayed the phosphorylation of histone H3 at Ser 10 by the immunoprecipitates. Fig. 3A shows that both antibodies immunoprecipitate cell cycle-specific histone H3 kinase activity from mitotic cytosol. We reproducibly immunoprecipitated moderate histone H3 kinase activity from interphase cytosol with both the crude anti-X aurora-A antiserum and the affinity purified anti-X aurora-A. By contrast we detected no H3 kinase activity in the anti-X aurora-B immunoprecipitate from interphase cytosol. We next tested whether these kinases were responsible for the H3 kinase activity observed in MCE by immunodepletion. High levels of H3 kinase were observed on anti-X aurora-B beads incubated in MCE but not on anti-X aurora-A beads (Fig. 3B). We also assayed the H3 kinase remaining in the eluates after the immunoprecipitation reactions were completed and found a significant complementary depletion of histone H3 kinase ac-tivity only in supernatants from the anti-X aurora-B immunoprecipitations (Fig. 3B). Finally, we assayed the amounts of both kinases in ICE and MCE by immunoblot (Fig. 3C). Our  1 and 2) and Coomassie stain (lanes 3 and 4). A combination of 0.1% Triton X-100 and 0.4 M NaCl efficiently eluted the chromosome kinase activity. The kinase activity is not eluted by NaCl concentrations up to 0.8 M in the absence of detergent or by detergent alone (data not shown). It is difficult to prevent chromatin contamination in eluates above this NaCl concentration. Migration of DNA topoisomerase II (q) and histone B4 (bracket) are indicated. B, histone H3 phosphorylation in chromatin eluates is cell cycle-and site-specific. The assay of histone H3 kinase activity in MCE and ICE was by filter binding assay. The activity is expressed as pmol of phosphate incorporated into substrate peptide/min/10 6 nuclei (see "Experimental Procedures") added to chromatin assembly reaction. The H3 kinase activity requires the presence of Ser 10 (H3); use of a peptide with a Ser 3 Ala mutation (H3(S10A)) causes no significant phosphorylation of substrate.
FIG. 3. H3 kinase activity coimmunoprecipitates with X aurora-B from MCE. A, H3 kinase assay of aurora immunoprecipitates from interphase and mitotic cytosol. Protein A beads were coated with either anti-X aurora-A or anti-X aurora-B, incubated with interphase or mitotic cytosol, and then assayed for bound histone H3 kinase activity. B, histone H3 kinase activity in immunoprecipitates from ICE and MCE. Protein A beads were coated with either anti-X aurora-A or anti-X aurora-B, incubated with ICE or MCE, and then assayed for H3 kinase activity. The supernatants from the immunoprecipitation reactions were also assayed. Total activity in the supernatant and beads are shown in the histogram. H3 kinase activity was only found in the anti-X aurora-B beads and only depleted by anti-X aurora-B beads. All immunoprecipitation assays are presented as pmol of phosphate/min incorporated into substrate peptide. Gray bars, total histone H3 kinase activity in supernatant; black bars, total histone H3 kinase activity in immunoprecipitate. C, immunodepletion of X aurora-A or X aurora-B from ICE and MCE. Supernatants from immunoprecipitation reactions (Fig. 3B) were separated by SDS-PAGE and immunoblotted with specific anti-X aurora-A and anti-X aurora-B antibodies. The top panel shows an immunoblot of ICE and MCE after incubation with anti-X aurora-A beads (Deplete) or control IgG beads (Control). The blot was probed with anti-X aurora-A. Rabbit IgG heavy chain (IgG HC) detected by secondary antibody is visible migrating slightly slower than the 46-kDa X aurora-A. The bottom panel shows an immunoblot of ICE and MCE after incubation with anti-X aurora-B beads (Deplete) or control IgG beads (Control) probed with anti-X aurora-B. The migration of the 41-kDa X aurora-B significantly separates it from the IgG heavy chain; therefore, this signal was not included in the figure.
immunoprecipitation reactions caused a significant depletion in X aurora-A and X aurora-B from ICE and MCE but only detected H3 kinase activity in the anti-X aurora-B immunoprecipitates. Taken together, these data suggest that both known Xenopus aurora kinases are associated with a histone H3 kinase activity in whole mitotic cytoplasm, but the major chromosomal histone H3 kinase activity is associated with X aurora-B. Interestingly, X aurora-A immunoprecipitated from MCE contained no detectable histone H3 kinase activity, suggesting that this protein kinase is either inactive or missing an associated component present in cytosol. Whether these kinases directly phosphorylate histone H3 or do so by an associated protein kinase remains to be determined.
Regulation of the Mitotic Histone H3 Kinase by PP1-Bearing in mind that isolated interphase chromatin contains a weak histone H3 kinase activity (Fig. 1) and having identified PP1 as a component of isolated chromatin assembled in vitro (Fig. 1), we tested whether PP1 activity regulated the ability of the chromatin-associated kinase activity to efficiently phosphorylate histone H3. We designed an assay that could determine whether PP1 modulates histone H3 phosphorylation by regulating a chromatin-associated H3 kinase as opposed to simply dephosphorylating H3 and antagonizing the activity of the H3 kinase. We incubated isolated interphase chromatin with ATP plus the PP1-specific inhibitor I2 (49). We then washed the chromatin extensively, prepared ICE, and assayed histone H3 kinase activity eluted from chromatin in the presence of the protein phosphatase inhibitor microcystin LR. Treatment of isolated interphase chromatin with either ATP or I2 alone gave little stimulation of H3 kinase activity, suggesting that providing substrates for protein kinases or inhibition of PP1 is insufficient to fully activate the histone H3 kinase (Fig. 4A). The highest activation was obtained when interphase chromatin was incubated with I2 and ATP. In six separate assays using chromatin prepared from different cytosol preparations, the activation of the H3 kinase by I2 and ATP together was always significantly more than the sum of the activities observed when each reagent was incubated alone with chromatin. Similar results were observed with the less specific protein phosphatase inhibitor microcystin LR (data not shown) (56). This synergistic activation required ATP and was not blocked by roscovitine or PKI (data not shown). We were only able to activate the histone H3 kinase activity when I2 was added to isolated chromatin; attempts to activate the H3 kinase after elution from chromatin by adding ATP and I2 to ICE were unsuccessful, suggesting that the preparation of ICE disturbs or prevents the association of the activating kinase with the H3 kinase and that the H3 kinase is not regulated by autophosphorylation.
Levels of aurora-B in cultured cells and Ipl1p in S. cerevisiae peak in G 2 /M, but there are still detectable amounts of aurora-B and Ipl1p in G 1 and S phase cells (57,58). We have also detected X aurora-B in interphase Xenopus cultured cells (data not shown) and in ICE (Fig. 3C), suggesting that this protein is present in interphase Xenopus egg cytosol and can assemble onto interphase chromatin. This implies that the activity of aurora-B and Ipl1p may be regulated in a cell cycle-dependent fashion. To determine whether the activity observed in Fig. 4A was associated with X aurora-B, we mixed ICE made from chromatin treated with I2 and ATP with anti-X aurora-B beads. This treatment effectively removed all histone H3 kinase activity from activated ICE, and all kinase activity was recovered on the beads (Fig. 4B). No detectable kinase activity was associated with anti-X aurora-A beads (Fig. 4B). Taken together, these results show that during interphase, PP1 inhibits the activation of aurora-B-associated kinase and suggest that a separate protein kinase is responsible for activating it.
They also suggest that the mitotic histone H3 kinase is activated by phosphorylation. To test this, we immunoprecipitated histone H3 kinase activity from MCE with anti-X aurora-B beads, incubated the beads with purified recombinant PP1, and then assayed the histone H3 kinase activity on the beads. To isolate the effect of PP1 to inactivation of the kinase, PP1 was preincubated in either buffer alone or 1 M microcystin LR, and all assays were performed in the presence of microcystin LR. Fig. 4C shows that treatment of the immunoprecipitated histone H3 kinase activity with active PP1 completely inhibits its activity. Thus, the activity of the X aurora-B-associated mitotic histone H3 kinase is directly regulated by phosphorylation.
Inhibition of Protein Phosphatases Induces Histone H3 Phos- FIG. 4. PP1 regulates the chromosome-associated kinase. A, activation of histone H3 kinase in interphase chromatin by PP1 inhibition. Interphase chromatin was isolated and then incubated with 0.2 mM ATP and 660 nM I2 alone or together (see "Experimental Procedures"). ICE was then prepared and assayed for H3 kinase activity. Activity is presented as pmol of phosphate/min/mg protein incorporated into substrate peptide in chromatin eluate. Incubation with I2 and ATP together produces a level of activation significantly higher than the sum of the separate incubations. B, activated H3 kinase from interphase chromatin is associated with X aurora-B. ICE from A was incubated with anti-X aurora-A and anti-X aurora-B beads. Beads and supernatant were then assayed for histone H3 kinase activity. The histogram shows the total histone H3 kinase activity in supernatant and immunoprecipitate. Gray bars, H3 kinase activity in supernatant; black bars, H3 kinase activity in immunoprecipitate. C, inactivation of mitotic histone H3 kinase by PP1. Histone H3 kinase was immunoprecipitated from MCE using anti-X aurora-B beads and incubated with PP1 that had been pretreated with buffer or microcystin LR. The beads were then assayed for H3 kinase activity in the presence of microcystin LR. The histogram shows histone H3 kinase activity present in immunoprecipitate.
phorylation but Not Chromosome Condensation-Previous studies have shown that addition of protein phosphatase inhibitors to cultured cells can stimulate histone H3 phosphorylation and chromosome condensation (7,8). To determine the significance of histone H3 phosphorylation in more detail, we added protein phosphatase inhibitors to in vitro chromatin assembly reactions and examined gross chromatin structure and phosphorylation of histone H3 at Ser 10 . Interphase cytosol converted demembranated sperm nuclei to decondensed balls of chromatin, and immunofluorescence with an anti-Ser(P) 10 histone H3 antibody revealed little detectable H3 phosphorylation (Fig. 5A, I). Mitotic cytosol converted demembranated sperm nuclei into condensed chromosomes with high levels of anti-phospho-H3 staining (Fig. 5A, M). By contrast, addition of the serine/threonine protein phosphatase inhibitor microcystin LR (56) (1 M) to interphase cytosol caused the formation of decondensed chromatin with high levels of anti-phospho-H3 staining (Fig. 5A, I ϩ Mc). Identical structures occurred whether we pretreated interphase cytosol with microcystin LR before adding sperm nuclei or added microcystin LR after the sperm decondensation reaction was complete (data not shown). Moreover, the chromatin formed in microcystin LR appeared more decondensed and, when fixed and centrifuged onto coverslips, spread out more than chromatin assembled in untreated interphase cytosol. This chromatin was also very sensitive to shear during sample handling, suggesting that it was more fragile than control interphase chromatin or mitotic chromosomes. The addition of microcystin LR to mitotic cytosol resulted in normally condensed mitotic chromosomes and no noticeable changes to chromosome structure or histone H3 phosphorylation (data not shown).
To determine whether interphase chromatin assembled in the presence of microcystin LR was missing an essential chromosome condensation factor, chromatin isolated from interphase, mitotic, or microcystin LR-treated interphase cytosol was immunoblotted with antibodies to XCAP-C and XCAP-E, two components of 13 S condensin (40). This protein complex only associates with chromosomes during mitosis and is required for proper chromosome condensation (59). As expected, XCAP-C and XCAP-E associated with mitotic chromosomes but not with interphase chromatin (Fig. 5B). However, despite the presence of mitotic levels of Ser 10 phosphorylation on histone H3 (Fig. 5B, P-H3), XCAP-C and XCAP-E did not associate with chromatin made in interphase cytosol supplemented with 1 M microcystin LR (Fig. 5B, XCAP-C and XCAP-E). Thus, induction of histone H3 phosphorylation on interphase chromatin is insufficient to cause binding of 13 S condensin to chromatin, in agreement with previous results showing that the 13 S condensin does not bind to nucleosomes (9,28). Our results suggest that one reason decondensed fragile chromatin is formed when protein phosphatases are inhibited in interphase cytosol is that 13 S condensin fails to load onto chromatin. 13 S condensin is phosphorylated and activated in vitro by active cdc2 kinase/cyclin B (33); therefore, it seemed possible that targeting of 13 S condensin required the presence of high H1 kinase levels. Cyclin degradation is induced during the preparation of the interphase extracts and addition of cycloheximide, and the preparation of cytosol makes it unlikely that any additional cyclin could be synthesized. Indeed, we detected little activation of H1 kinase by microcystin LR (Fig. 5C). DISCUSSION We have shown here that mitotic histone phosphorylation is mediated by a protein kinase and regulated by a phosphatase that are stably associated with chromatin and chromosomes. Consistent with events in living cells, histone H3 Ser 10 and the linker histone B4 are phosphorylated in mitotic chromosomes, whereas histones H2A and H4 are phosphorylated in interphase chromatin (Fig. 1). Eluates of isolated mitotic chromosomes contain the mitotic histone H3 kinase, and this activity is associated with the X aurora-B kinase (Figs. 2 and 3). The X aurora-B-associated activity is inactivated by treatment with phosphatase, showing that it requires phosphorylation for its activity. Furthermore, the X aurora-B-associated activity exists in interphase chromatin but is inactive. This form is activated by inhibition of PP1 and incubation in ATP, showing that PP1 activity directly inhibits X aurora-B in interphase chromatin (Fig. 4). Phosphorylation of chromosomal proteins has been implicated in controlling the formation of mitotic chromosomes (6 -8, 32, 33). Our work pinpoints the location of chromosomal protein phosphorylation, uncovers the mechanism of regulation of the kinase involved in this modification, and suggests the presence of signaling pathways that are stably associated with chromatin and chromosomes.
What does mitotic histone H3 phosphorylation do? Histones are subject to a wide range of post-translational modifications throughout the cell cycle including phosphorylation, acetylation, and methylation (26). Histone H3 phosphorylation was first identified in mitotic cells and has since been recognized as a ubiquitous mitotic modification (2,3,60). However, phosphorylation of H3 Ser 10 also occurs in more discrete chromatin regions during activation of transcription and coincides with histone H3 acetylation on the same nucleosomes (61)(62)(63). This kind of specific combinatorial modification has been proposed to form a "histone code" that would form a marked chromatin surface that could recruit chromatin remodeling and modification activities or factors mediating higher order chromatin structures (24). We do not yet know the full list of histone modifications that occur in mitosis, but it likely will include changes in ubiquitination and phosphorylation on other histones (4,15).
Despite the correlation between histone H3 phosphorylation and chromosome condensation, induction of global H3 phosphorylation is not sufficient to induce chromosome condensation or targeting of 13 S condensin to chromatin in Xenopus cytosol (Fig. 2). The dissociation of histone H3 phosphorylation and chromosome condensation observed in Fig. 5 may occur because of the absence of increased CDK activity or possibly because the chromatin-associated H1 kinase activity is not stimulated by inhibition of PP1 (data not shown). Without these activities condensation factors like 13 S condensin may not bind to chromatin. Regardless, mitotic levels of histone H3 phosphorylation are not sufficient to drive 13 S condensin targeting to chromatin (12).
Induction of histone H3 phosphorylation to mitotic levels in interphase cytosol did affect chromatin structure. We observed fragile chromatin that was deformed and sometimes even sheared during sample handling ( Fig. 5 and data not shown). This result is consistent with previous biophysical analyses showing that in vitro phosphorylation of isolated chromatin on histone H3 Ser 10 with cAMP-dependent protein kinase increases the flexibility of chromatin fibers (23,64). This increased flexibility likely results from changes in the charge density in the N terminus of histone H3 and thus the strength of electrostatic interactions between the positively charged H3 N terminus and the phosphate backbone of the DNA. However, we cannot eliminate the possibility that other chromatin modifications, undetected in our assays, may be combining with histone H3 phosphorylation to generate fragile chromatin (4,15). An increase in flexibility may make chromatin a better substrate for 13 S condensin and other condensation factors, perhaps by allowing supercoiling or other distortions to propagate along the chromatin fiber. Whether this structural change is required for condensation or to mediate sister chromatid cohesion (5) remains to be determined.
A number of recent studies have localized aurora-B kinase to chromosomes and demonstrated a role for it and PP1 in mitotic histone H3 phosphorylation (12,13,15,22). At the moment, we are not certain that Ipl1p/AIR-2/X aurora-B and Glc7p/Ceglc7/ PP1 account completely for mitotic histone H3 phosphorylation. We have identified a number of other protein kinases and phosphatases in ICE and MCE (data not shown), and determining their role in chromosome condensation will be a major goal for the future. Nonetheless, the X aurora-B-associated activity contains most of the mitotic histone H3 kinase activity, and this combined with genetic data (15,22) suggests that aurora-B is a bona fide mitotic histone H3 kinase. However, reducing aurora-B levels in Drosophila decreases the levels of phospho-H3 all along chromosome arms, even though aurora-B is concentrated at centromeres (12). Aurora-B may be the major mitotic histone H3 kinase, but it may concentrate at centromeres and occur at much lower and so far undetected levels along the chromosome arm. In addition, mitotic histone H3 phosphorylation in mouse cells initiates at centromeres in late G 2 and spreads throughout the chromatin before condensation is complete and before aurora-B is concentrated on chromosomes (3). Mitotic frog egg extracts approximate a metaphase state, but whether aurora-B or another kinase initiates histone H3 phosphorylation during G 2 remains to be determined. Studies to discern these possibilities are in progress. Most importantly for the study of chromosome formation, the approach presented here will permit a more detailed examination of the phosphorylation of DNA topoisomerase II (31,32,65), cohesin (66 -68), and 13 S condensin (33).
There is a marked specificity in the aurora family protein kinases. Inactivation of C. elegans air-1, a second C. elegans aurora family member, does not affect histone H3 phosphorylation (15). We show that the cytosolic forms of X aurora-A and X aurora-B both possess mitosis-specific histone H3 kinase activity, but we detect no histone H3 kinase activity in chromosome-associated X aurora-A. Further studies will be required to ascertain the nature of these activities, but we note that X aurora-B from MCE fractionates as a 1 Md protein complex (data not shown). This is consistent with a large amount of data suggesting that aurora kinases associate with specific cell cycle proteins, some of which may function as regulatory subunits. In Xenopus egg cytosol, X aurora-B is complexed with INCENP, a chromosome passenger protein that transits from centromeres to the mitotic spindle to the cleavage furrow as cells proceed through mitosis (19,69). Biochemical and genetic evidence exists for similar associations in yeast, worms, and human cells (20,70). This complex may be required for aurora-B function. Inactivation of the worm INCENP icp-1 by RNAi or introduction of a dominant negative N-terminal INCENP fragment into human cells disrupts aurora-B localization and causes defects in prometaphase chromosome alignment, sister chromatid segregation, and cytokinesis (19,20). In addition, mitotic phosphorylation of histone H3 in C. elegans may depend on the physical interaction of aurora-B and BIR-1, a homologue of survivin (22). One function of these interactions may be to target aurora-B to chromatin. Similarly, the association of aurora-A with Eg5, a mitotic spindle kinesinlike motor may localize this kinase to the spindle pole (43,71). Following this model, aurora-A from MCE does not phosphorylate histone H3 because it is missing a necessary binding partner required for localization and enzymatic activity, although cytosolic and recombinant X aurora-A both phosphorylate histone H3 (Fig. 5 and data not shown). This localizationdependent model is attractive but will require an analysis of the interactions of the aurora kinases and their binding partners. How the recognition of aurora-B-binding sites is regulated to generate the chromosome passenger localization will also be of interest.
In our in vitro system, blocking PP1 function allows activation of the aurora-B histone H3 kinase activity, suggesting that in interphase chromatin aurora-B activity is directly inhibited by PP1. This regulation is likely critical in vivo because Ipl1p and aurora-B are present during interphase in yeast and mammalian cells (57,58). In S. cerevisiae, defects in the aurorarelated kinase Ipl1p cause reduced levels of histone H3 phosphorylation, and these are rescued by mutations in glc7, the gene encoding the catalytic subunit of PP1 (15). In worms, air-2(RNAi) significantly decreases mitotic histone H3 phosphorylation in the worm gonad. Again, this defect is partially rescued by Ceglc7(RNAi) (15). Our data explain these genetic results by showing that PP1 functions both to antagonize the activity of the aurora-B or Ipl1p kinase and to directly inhibit the activity of aurora-B. Additionally, the requirement for ATP in our activation reaction demonstrates the presence of an aurora-B-activating kinase. We have been unable to activate X aurora-B in ICE, suggesting that aurora-B does not activate by autophosphorylation in solution (data not shown). Whether aurora-B simply requires chromatin or, more likely, a chromatindependent association with another protein kinase remains to be determined. We note that further fractionation of interphase and mitotic chromatin eluates shows that PP1 fractionates away from X aurora-B (data not shown). Therefore, PP1 is not bound to X aurora-B in MCE but may associate with it in chromatin.
Our results suggest that the phosphorylation of histones occurs by enzymes stably associated with chromatin and chromosomes. The insensitivity of mitotic chromosome protein phosphorylation to CDK inhibitors suggests that the regulation of these activities by CDKs may require at least one soluble factor, possibly the CDKs themselves, to communicate the status of the cell cycle. Alternatively insensitivity to CDK inhibition might occur because of defective cell cycle-dependent nuclear import of CDK-cyclin complexes in our simplified chromatin assembly system (34,72). Most membranes have been sedimented away from the cytosol, so that interphase chromatin is missing a functional nuclear envelope and is deficient in nuclear import (73,74). In this case, CDK-cyclin complexes could affect the activity of chromatin-associated kinases in cytosol, but their interaction may be transient, or they may be lost during chromatin purification in the absence of a nuclear envelope. Regardless, our data clearly show that the mitotic histone H3 phosphorylation associated with X aurora-B is distinct from CDK activity. Discovering how CDK-cyclin complexes communicate with the aurora-B/PP1 signaling system will be an important goal for future work.