pEg2 Aurora-A Kinase, Histone H3 Phosphorylation, and Chromosome Assembly in Xenopus Egg Extract*

In eukaryotes cell division is accompanied by phosphorylation of histone H3 at serine 10. In this work we have studied the kinase activity responsible for this histone H3 modification by using cell-free extracts prepared from Xenopus eggs. We have found that the Xenopus aurora-A kinase pEg2, immunoprecipitated from the extract, is able to phosphorylate specifically histone H3 at serine 10. The enzyme is incorporated into chromatin during in vitro chromosome assembly, and the kinetics of this incorporation parallels that of histone H3 phosphorylation. Recombinant pEg2 phosphorylates efficiently histone H3 at serine 10 in reconstituted nucleosomes and in sperm nuclei decondensed in heated extracts. These data identify pEg2 as a good candidate for mitotic histone H3 kinase. However, immunodepletion of pEg2 does not interfere with the chromosome assembly properties of the extract nor with the pattern of H3 phosphorylation, suggesting the existence of multiple kinases involved in this H3 modification in Xenopus eggs. This hypothesis is supported by at (cid:2) 20 °C. Nucleosome Reconstitution— In the reconstitution experiments, “bulk” nucleosomal DNA, prepared from native hen erythrocyte nucleosomes, was used. A strictly stoichiometric amount of the four histones (determined spectrometrically and checked on a SDS gel) in 10 m M HCl was dialyzed overnight at 4 °C against 2 M NaCl, 50 m M Tris-HCl, pH 7.8, 1 m M EDTA. The next morning a mixture of bulk nucleosomal DNA and 32 P-end-labeled 152-bp Eco RI- Rsa I fragment containing a Xenopus borealis somatic 5 S RNA gene was added to the dialysis tubing. The ratio of the added DNA to the core histone octamer was 1:0.8. Nucleo- some reconstitution was performed by successively lowering the concentration of the NaCl in the dialysis buffer. Finally, the reconstituted particles were dialyzed against 10 m M Tris, pH 7.8, 10 m M NaCl, 1 m M EDTA and used for kinase assay. The extent of reconstitution and the integrity of the nucleosomes were checked by electrophoretic mobility shift analysis (EMSA) and DNase I footprinting. carried out buffer. Upon completion of electro-phoresis, gel by stopped stop

In eukaryotes cell division is accompanied by phosphorylation of histone H3 at serine 10. In this work we have studied the kinase activity responsible for this histone H3 modification by using cell-free extracts prepared from Xenopus eggs. We have found that the Xenopus aurora-A kinase pEg2, immunoprecipitated from the extract, is able to phosphorylate specifically histone H3 at serine 10. The enzyme is incorporated into chromatin during in vitro chromosome assembly, and the kinetics of this incorporation parallels that of histone H3 phosphorylation. Recombinant pEg2 phosphorylates efficiently histone H3 at serine 10 in reconstituted nucleosomes and in sperm nuclei decondensed in heated extracts. These data identify pEg2 as a good candidate for mitotic histone H3 kinase. However, immunodepletion of pEg2 does not interfere with the chromosome assembly properties of the extract nor with the pattern of H3 phosphorylation, suggesting the existence of multiple kinases involved in this H3 modification in Xenopus eggs. This hypothesis is supported by in gel activity assay experiments using extracts from Saccharomyces cerevisiae.
Cell division requires accurate condensation and faithful segregation of chromosomes. Despite the great efforts invested, the mechanisms of these two processes still remain unclear. However, during the last few years an impressive progress has been made in their understanding by using two complementary approaches as follows: genetics in yeast and experiments in extracts prepared from Xenopus eggs (reviewed in Ref. 1). The biochemical manipulations of the Xenopus egg extract were extremely useful since they have led to the identification of multiprotein complexes, termed condensins, required for chromosome condensation (2). Targeting of condensins is mitoticspecific, and their phosphorylation may trigger chromosome condensation (3). Several lines of evidence indicate that the master kinase Cdc-2 might be involved in the phosphorylation and activation of condensins (3).
Chromosome assembly is also accompanied by phosphorylation of linker histone H1 (4,5) and core histone H3 (6 -9). However, the presence of linker histones is not necessary for chromosome and nucleus formation (10 -14), and consequently their phosphorylation should not be required for these processes. In contrast, the phosphorylation of serine 10 in the amino-terminal domain of histone H3 is essential for cell division. Phosphorylation of histone H3 has been observed and characterized in organisms as divergent as yeast (15), Tetrahymena thermophila (16), Aspergillus nidulans (17), Caenorhabditis elegans (15,18), plants (19), and vertebrates (20,21). It was also described during in vitro chromosome assembly in Xenopus egg extract (22). A mutant T. thermophila strain, containing a non-phosphorylatable histone H3, exhibited perturbed chromosome condensation, abnormal segregation, and chromosome loss during mitosis and meiosis (16), demonstrating the primary significance of this histone H3 modification. This was further supported by experiments on blocking of the mitotic histone H3 kinase and, thus, histone H3 phosphorylation, which has resulted in inhibition of chromosome assembly in vitro (22) and in cells in culture (21).
Nonetheless, the mechanism of action of histone H3 phosphorylation in cell division is still poorly understood. A real progress toward elucidation of this mechanism will be the identification of the enzymes involved in the regulation of mitotic H3 phosphorylation. Recently, reports from two different groups identified two distinct kinases Never in Mitosis A (NIMA) 1 in A. nidulans (17) and an aurora kinase Ipl1 (Increase in Ploidy) in budding yeast S. cerevisiae and air2 (aurora Ipl1-related kinase 2) in the worm C. elegans (15) as the mitotic histone H3 kinases.
In vitro NIMA phosphorylated histone H3 at serine 10, and the in vivo phosphorylation of the histone is dependent on the presence of the kinase activity (17). At the onset of mitosis, NIMA is detected on chromatin and subsequently colocalized with spindle microtubules and spindle pole bodies. The chromatin localization of this enzyme is tightly correlated with histone H3 phosphorylation (17).
The budding yeast genome encodes for a single aurora kinase Ipl1 that is required for cell cycle progression, and strains bearing genetics defects in this enzyme showed abnormal chromosome segregation and suffered severe nondysjunction (23). The enzyme expression peaks at mitosis, and when a temperature-sensitive lethal Ipl1 strain was grown at permissive tem-perature, a markedly reduced H3 phosphorylation was detected (15). In vitro, the kinase phosphorylated both H3 and H2B in a mixture of free histones and on nucleosomes. The C. elegans genome encodes for two aurora kinases, air1 and air2, and the last one was observed on chromosomes at mitosis and meiosis (15,18). However, only the disruption by RNA interference of air2 expression in C. elegans embryos led to undetectable histone H3 phosphorylation (15,18).
The situation is different in vertebrates where three aurora kinases, recently renamed as aurora-A, -B, and -C, have been described (reviewed in Refs. 24 -26) The Xenopus laevis kinase pEg2 belongs to the aurora kinase protein family, and according to the new nomenclature is the Xenopus aurora-A kinase. pEg2 was found associated with centrosomes in a cell cycle-dependent manner. It also binds to the spindle microtubules, and its activity is required for bipolar spindle assembly (27). In vivo pEg2 has been reported to associate with the kinesin-related protein XlEg5 and to the cytoplasmic polyadenylation element binding factor. Both proteins are phosphorylated by pEg2 in vitro in residues found phosphorylated in vivo (24,28).
In vertebrate cells the aurora kinase(s) that phosphorylates histone H3 is not known. Moreover, none of the X. laevis H3 mitotic kinases have been identified yet. The aim of this work was to search for such enzyme(s). To this end extracts from Xenopus eggs were used, and in vitro chromosome assembly was performed. We have identified the aurora-A kinase pEg2 as a potentially good candidate for histone H3 mitotic kinase in Xenopus eggs.

Mitotic Extract Preparation and Isolation of Demembraned Sperm
Nuclei-Mitotic extracts from Xenopus eggs were prepared essentially as described (49). Dejellied eggs were crushed by centrifugation for 15 min at 15,000 rpm in an SW41 rotor (Beckman Instruments) at 16°C in XBE2 buffer (100 mM KCl, 2 mM MgCl 2 , 0.1 mM CaCl 2 , 10 mM K-HEPES, pH 7.7, 5 mM K-EGTA, 0.05 M sucrose) supplemented with 10 g/ml leupeptin and aprotinin and 100 g/ml cytochalasin D. The cytoplasmic layer was collected using a 20-gauge needle via a side puncture and kept on ice. Protease inhibitors (leupeptin and aprotinin) and cytochalasin D at final concentration of 10 g/ml and 1/20 volume of 20ϫ energy mix (20 mM phosphocreatine, 2 mM ATP, and 5 g/ml creatine kinase, final concentration) were added, and the extract was clarified by centrifuging at the same speed as above, but at 4°C. The golden layer (low speed supernatant) was collected and transferred in 2 ml of polypropylene tubes for a TLS-55 rotor (Beckman Instruments) and spun at 52,000 rpm for 2 h at 4°C. The top lipid layer was aspirated under vacuum, and the clear cytoplasmic fraction was recentrifuged at 4°C for 30 min at 52,000 rpm to remove the residual membranes. The extract (high speed supernatant) was collected, aliquoted in 25-l fractions, immediately frozen in liquid nitrogen, and stored at Ϫ80°C. Heated extracts were prepared as described previously (35).
Demembraned sperm nuclei were isolated according to a protocol published previously (50). They were aliquoted in 5-l fractions at a concentration of 1 g/l DNA and stored at Ϫ80°C. Each aliquot was used only once, since after refreezing and a second thawing the demembraned sperm nuclei strongly tend to aggregate.
Expression and Purification of Full-length and Mutant Histones-Recombinant X. laevis full-length and globular domain histone proteins were made in bacteria and purified to homogeneity as described by Luger et al. (34). The mutations of serine 10 and of serine 28 to alanine in histone H3 were made according to the standard site-directed mutagenesis approach by using QuickChange TM site-directed mutagenesis kit (Stratagene). For the first mutation the oligonucleotides used were pet3aH3S10A (cagaccgcccgtaaagctaccggagggaagg) and pet3bH3S10A (ccttccctccggtagctttacgggcggtctg), and the mutagenesis at serine 28 was carried out with pet3aH3S28A (caccaaggcagccaggaaggctgctcctgctacc) and pet3bH3S28A (ggtagcaggagcagccttcctggctgccttggtg) oligonucleotides. The mutated histone H3 was expressed and purified exactly as the non-mutated one.
The GST-histone tail fusion proteins were prepared as described (22). The concentration of the recombinant proteins was determined by using both Bradford assay and spectrophotometrically.
Immunodepletion of pEg2 Kinase from the Egg Extract-pEg2 was depleted from the extract by the already characterized 1C1 monoclonal antibody (27). The immunodepletion was carried out essentially according to a protocol described previously (37). Briefly, protein A-agarose beads (Amersham Pharmacia Biotech) were washed with EB buffer (80 mM ␤-glycerophosphate, pH 7.3, 15 mM MgCl 2 , 20 mM EGTA, 1 mM dithiothreitol) and blocked with 5 times volume of bovine serum albumin at a concentration of 10 mg/ml. After three successive washings with EB, 500 l of the hybridoma supernatant was added to 50 l of pelleted beads, and the suspension was incubated for 1 h under rotation at 4°C. For the mock immunodepletion, 25 l of preimmune serum was diluted with EB to 500 l final volume, mixed with 50 l of settled beads, and incubated as above. Once the incubation completed, the beads were washed 3 times with 10 volumes of EB. The cytosol was depleted by adding 4 volumes of extract to 1 volume of settled beads followed by incubation under rotation at 4°C for 1 h. After centrifugation, the supernatant was treated with additional 50 l of fresh beads under the conditions described.
Quantitative measurements showed that this protocol removed 95-98% of pEg2 present in the extract. The native pEg2 immobilized to the beads was further used in kinase assay experiments.
Expression of pEg2 and Kinase Assay-Recombinant pEg2 was expressed and purified essentially as described (27). Briefly, cDNA of pEg2 was prepared by differential screening of Xenopus egg cDNA library. The coding sequence of the protein was amplified by polymerase chain reaction and inserted in NheI/XhoI restriction sites of the His tag expression vector pET21 (Novagen Inc.) using primers described previously (27). The catalytically inactive form of the enzyme pEg2KR was engineered with Transformer Site-directed Mutagenesis Kit (CLON-TECH) to change the lysine 169 of pEg2 to arginine. The kinases were expressed in Escherichia coli strain BL21(DE3) and the tagged proteins purified on a nickel column (Qiagen).
The purified enzymes were dialyzed overnight against EB, aliquoted in 20-l fractions, and stored at Ϫ20°C at concentration 0.2-0.5 g/l. For the kinase assay 1 l of recombinant pEg2 was added to 50 l of EB containing 1 g of recombinant histones or 5 g of reconstituted nucleosomes. After addition of 20 Ci of [␥-32 P]ATP, the reaction was allowed to proceed for 15 min at room temperature, and then the proteins were precipitated with 20% trichloroacetic acid (final concentration). The pellet was dissolved in 8 M urea, and after adding sample buffer the proteins were separated on 18% SDS-polyacrylamide gel, and the gel was dried and autoradiographed.
Immunoblotting-The immunoblotting protocol was already described (22). The dilution used for the anti-phosphorylated histone H3 antiserum was 1:3000, whereas the hybridoma supernatant was diluted 1:500. The filters were developed by using the enhanced chemiluminescence system (ECL, Amersham Pharmacia Biotech) following the instructions of the manufacturer.
Preparation of Yeast Protein Extract-The extract was prepared by using Saccharomyces cerevisiae strain GA59. This yeast strain has the advantage of exhibiting low protease activity. An overnight preculture, grown in YPD medium, was diluted to A 600 ϭ 0.2-0.3 with the same medium. The yeast were further grown at 30°C to A 600 ϭ 0.5 and then supplemented with nocodazole to 10 g/ml final concentration. The control and the nocodazole-containing cultures (50 ml each) were next incubated for 5 h and harvested by centrifugation. The pellets were washed twice with ice-cold water and resuspended in 1 ml of lysis buffer (10 mM MgCl 2 , 20 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 0.05% SDS, 1 M okadaic acid, 50 mM Tris-HCl, pH 8.0). The suspension was then transferred in 2-ml Eppendorf tubes containing glass beads, and the cells were lysed by 5 min of vortexing. The cell debris were removed by centrifugation, and the supernatant was used in gel activity assay experiments.
Assembly of Mitotic Chromosomes-Mitotic chromosomes were assembled in Xenopus egg extract at 23°C following a standard protocol (50). 20 -25 l of extract were used for 40 -60,000 demembraned sperm nuclei. In some experiments 1 l of cyclin B⌬90 (the nondegradable form of cyclin B) was added. To follow the kinetics of assembly, 5-l aliquots from the mock-depleted and pEg2-depleted extract were taken at different times after initial incubation and fixed immediately with 5 l of the fix/stain buffer (Hoechst 33258 at 1 g/ml in 200 mM sucrose, 10 mM HEPES, pH 7.5, 7.4% formaldehyde, 0.23% 1,4-diazabicycle-[2.2.2]octane, 0.02% B NaN 3 , and 70% glycerol).
The decondensation of sperm nuclei in heated extract was carried out as described previously (35).
Immunofluorescence-The immunofluorescence analysis was carried out as described by de la Barre et al. (50). The anti-phosphorylated histone H3 antibody was used at dilution 1:5000. Finally, the chromosomes were counterstained with 8 l of fix/stain buffer.
Preparation of Nuclei and Nucleosomes-Hen erythrocyte nuclei were isolated as described by Mirzabekov et al. (51). Oligosomes were prepared by digestion of the nuclei with micrococcal nuclease and linker histones, and non-histone proteins were removed by centrifugation of the oligosomes over 5-20% sucrose, containing 0.65 M NaCl (52). After overnight dialysis against a solution of 10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 1 mM EDTA, oligonucleosomes were aliquoted and stored frozen at Ϫ20°C.
Nucleosome Reconstitution-In the reconstitution experiments, "bulk" nucleosomal DNA, prepared from native hen erythrocyte nucleosomes, was used. A strictly stoichiometric amount of the four histones (determined spectrometrically and checked on a SDS gel) in 10 mM HCl was dialyzed overnight at 4°C against 2 M NaCl, 50 mM Tris-HCl, pH 7.8, 1 mM EDTA. The next morning a mixture of bulk nucleosomal DNA and 32 P-end-labeled 152-bp EcoRI-RsaI fragment containing a Xenopus borealis somatic 5 S RNA gene was added to the dialysis tubing. The ratio of the added DNA to the core histone octamer was 1:0.8. Nucleosome reconstitution was performed by successively lowering the concentration of the NaCl in the dialysis buffer. Finally, the reconstituted particles were dialyzed against 10 mM Tris, pH 7.8, 10 mM NaCl, 1 mM EDTA and used for kinase assay. The extent of reconstitution and the integrity of the nucleosomes were checked by electrophoretic mobility shift analysis (EMSA) and DNase I footprinting.
EMSA and DNase I Footprinting on Reconstituted Nucleosomes-The EMSA was carried out in 2% agarose gel at room temperature in 0.5ϫ TBE (Tris borate/EDTA) buffer. Upon completion of the electrophoresis, the gel was stained with ethidium bromide.
The footprinting of the reconstituted nucleosomes was performed by using DNase I. The digestion was carried out with 10 ng of DNase I per 10 l of nucleosome solution (10 ng/l) in 10 mM Tris-HCl, pH 7.6, 5 mM MgCl 2 for 2 min at room temperature. The reaction was stopped by adding 100 l of stop solution (10 mM EDTA, 0.1% SDS, 50 ng/l proteinase K) followed by 30 min of incubation at 37°C. Then the samples were phenol-extracted, ethanol-precipitated, and separated on 8% polyacrylamide sequencing gel containing urea. The dried gel was exposed overnight on a PhosphorImager screen.

Phosphorylation of Histone H3 at Serine 10 by a 40 -42-kDa Polypeptide Present in Yeast Mitotic
Extracts-We were interested in determining the Xenopus kinase(s) responsible for the mitotic-specific phosphorylation of histone H3 at serine 10. However, various bona fide candidates for histone H3 mitotic kinases were identified in different organisms (15,17), which suggests either that distinct kinases operate in different systems or that several kinases can act in the same organism. If the last suggestion is true, it will be important to find the relative contribution of each enzyme in the mitotic phosphorylation of histone H3. Since in vitro serine 10 of histone H3 can be phosphorylated by a multitude of kinases (protein kinase A (29), NIMA (17), Ipl1/aurora-B (15), Msk1 (30), mitogen-activated protein kinase-dependent kinases (31), and Rsk-2 (32)), the determination of the active mitotic kinases could be done more easily by using yeast, which is a somewhat simpler system, instead of Xenopus. As H3 phosphorylation is highly conserved in evolution, the major mitotic kinases modifying this protein should belong to the same families in different organisms. Thus, the identification of the yeast enzymes will undoubtedly be instrumental for the analysis of the Xenopus system. Since the yeast genome is already sequenced, the kinase molecular weight determination will allow their identification. Following this rationale, we carried out a series of in gel activity assays using crude extracts isolated from nocodazoletreated or control, non-treated S. cerevisiae cells (Fig. 1). As substrates for the kinases we have incorporated into the gels either the non-mutated GST-histone H3 tail fusion (GSTH3) or the same fusion, but mutated at serine 10 to alanine (GSTH3S10), or the globular domain of histone H3 (GH3). As a control, we have used gel-incorporated GST. When the mitotic extract (isolated from the nocodazole-treated cells) was subjected to the assay, three specific bands were detected by autoradiography in the gel containing GST-H3. The strongest band migrated with molecular mass of 40 -42 kDa. The incorporation of 32 P in these three positions was characteristic for the yeast mitotic extract in the gel comprising GST-H3; very faint or no 32 P-labeled bands with the above molecular masses were observed in all other cases (Fig. 1, compare lane 1 with  lanes 2-8). These data suggest that at least three different kinases from the yeast mitotic extract phosphorylate specifically histone H3 at serine 10 when using gel activity assay. Since among them the 40 -42-kDa enzyme should be the most active one, we have further concentrated on it.
Budding yeast has more than 100 genes coding for protein kinases (33). Twelve of these enzymes exhibited a molecular mass in the region of 40 -45 kDa as described in the YPD proteome data base (www.proteom.com). Nine of them are serine/threonine protein kinases. Among them, only the Ipl1 aurora kinase was shown to participate in chromosome assembly and cell cycle progression, suggesting strongly that this is the major yeast kinase participating in phosphorylation of histone H3 at serine 10. Thus, our in gel activity assay results are in agreement with the available data identifying this enzyme as a H3 mitotic kinase in S. cerevisiae (15).
In higher eukaryotes multiple homologues to Ipl1 were described (reviewed in Ref. 24). In Xenopus, the aurora-A kinase pEg2 is active during mitosis and is shown to be involved in microtubule dynamics (27). Sequence alignment of pEg2 and FIG. 1. Activity gel assay detects a mitotic-specific increase of phosphorylation of histone H3 at serine 10 by a 40 -42-kDa polypeptide. Crude extracts, isolated from nocodazole-treated (Nϩ), and control, non-treated yeast cells (NϪ) were subjected to electrophoresis in 12% polyacrylamide gel containing SDS in which either GST, GST-H3 tail fusion, or GST-H3 tail fusion singly mutated at serine 10 (GSTH3S10A) or tail-less histone H3 (GH3) were incorporated prior to polymerization. Following electrophoresis, the gels were washed to remove SDS, the separated proteins renatured, and the kinase assay carried out in the gels. The arrowheads (lane 1) showed the 32 P incorporated into GST-H3 in regions of the gel corresponding to three different molecular masses.

pEg2 Kinase and H3 Phosphorylation
Ipl1 demonstrates a significant sequence similarity (50% identity) of the catalytic domains of both enzymes. Since Ipl1 is involved in H3 phosphorylation in S. cerevisiae, we hypothesized that its homologue pEg2 could play similar role in Xenopus.
Phosphorylation of Free Histones in Solution by Recombinant pEg2-Our first step in identifying the role of pEg2 in H3 phosphorylation was to prepare a recombinant active pEg2 (rpEg2) and to produce and purify to homogeneity the four recombinant core histones and a large number of their mutants. Next we have checked whether the enzyme can act specifically on serine 10 of histone H3. Fig. 2 and Fig. 3 show the results of the kinase assay of rpEg2 on different histone substrates. As seen, full-length histones H3 and H2B are phosphorylated (Fig. 2, lanes 2-4), whereas H2A and H4 are not substrates for the enzyme (Fig. 2, lanes 1 and 6). The radioactivity is incorporated in the tail of histone H3, since its globular domain GH3 is not labeled (Fig. 2, lane 9), whereas the GSThistone H3 tail (GST-H3) is efficiently marked (Fig. 3A, lane 4, and Fig. 3C, lane 2). The phosphorylation is specific for serine 10, because a mutation of serine 10 of H3 to alanine abolishes the incorporation of 32 P in the protein (Fig. 2, compare lane 3 with lane 5, and Fig. 3C, lanes 2-4). The same type of mutation, but in the other serine of H3 tail (serine 28), has no effect on the labeling by the kinase (Fig. 2, lane 4, and Fig. 3C, lane 4). Catalytically inactive rpEg2 does not phosphorylate the tail of histone H3 (Fig.  3C, lane 6), further confirming that H3 labeling is due to the kinase itself and not to a bacterial contaminant.
rpEg2 also phosphorylates histone H2B. The site(s) of phosphorylation is located in the globular domain of the protein, since by using kinase assay GH2B is radioactively labeled (Fig.  2, lane 8), whereas the H2B tail is not (Fig. 3A, lane 3).
Specific Phosphorylation of Histone H3 at Serine 10 on Nucleosome Templates by rpEg2 Kinase-We have demonstrated that rpEg2 phosphorylates histone H3 at serine 10 and histone H2B when they are free in solution. However, in the nucleus these histones are not free but are instead organized into nucleosomes. Besides, during chromosome assembly histone H2B is not phosphorylated. 2 Thus, if pEg2 is associated with chromosome condensation, it should be able to phosphorylate on nucleosomal templates specifically histone H3 at serine 10, but not histone H2B. To check this we have reconstituted "chimerical nucleosomes" by using bulk nucleosomal DNA and different combinations of recombinant full-length and tail-less histones. In the reconstitution reactions a trace amount of 32 P-end-labeled 152-bp EcoRI-RsaI fragment containing an X. borealis somatic 5 S RNA gene was added. This has allowed us to follow both the efficiency of reconstitution and the integrity of the reconstituted nucleosomes ( Fig. 4 and Fig. 5A). As seen, under the conditions used, complete reconstitution is achieved; the EMSA did not detect free DNA (Fig. 5A). The DNase I 2 V. Gerson and S. Dimitrov, unpublished observations.

FIG. 2. Recombinant pEg2 phosphorylates the tail of histone H3 and the "histone fold" domain of H2B.
In vitro kinase assays were carried out by using [␥-32 P]ATP, recombinant pEg2, recombinant intact histones H2A, H2B, H3, and H4, their globular domains GH2A, GH2B, GH3, and GH4, and singly mutated histone H3 at serine 10 (H3S10A) and at serine 28 (H3S28A). Each reaction contained 200 ng of pEg2 and 1 g of intact or mutated histone. After completion of the reaction, the proteins were trichloroacetic acid-precipitated and separated on 18% SDS-polyacrylamide gel. Phosphorylation was analyzed by autoradiography (A). The Coomassie-stained gel is shown in B.
FIG. 3. Specific phosphorylation of the tail of histone H3 at serine 10 by recombinant pEg2. The kinase assays were carried out with recombinant wild and mutated pEg2 by using as substrates GST fusions (GSTH2A, GSTH2B, GSTH3, and GSTH4) with the non-mutated tail of the core histones and the fusions of the tail of histone H3 but mutated at serine 10 (GSTH3S10A) and at serine 28 (GSTH3S28A). The autoradiographs and the Coomassie-stained gels are shown in A and C and B and D, respectively. pEg2 Kinase and H3 Phosphorylation footprinting analysis showed the nucleosome 10-base pairs repeat with essentially no background between the different bands (Fig. 4). Thus, the particles prepared with recombinant histones exhibited overall structure closely similar to that of native nucleosomes, a result in agreement with the data in the literature (34). Once the reconstituted nucleosomes were characterized, we asked if they could be phosphorylated by rpEg2. As shown on Figs. 5 and 6, histone H3 was a good substrate for the kinase in nucleosomes comprising the four full-length histones. Moreover, this phosphorylation was specific for serine 10, since particles containing histone H3 with mutated serine 10 to alanine were not phosphorylated by the kinase (Fig. 6B,  lane 3), whereas the phosphorylation of nucleosomes with histone H3 mutated at serine 28 was not affected (Fig. 6B, lane 2). Phosphorylation of histone H2B was observed neither in these particles nor on H2B in nucleosomes, reconstituted with intact H2B and the globular domains of the other three histones (Fig.  5B, lane 3). In all other cases the globular domain of histone H2B was accessible and efficiently phosphorylated by the kinase. Therefore, the tail of histone H2B impeded the phosphorylation of GH2B within nucleosomal templates.
However, in decondensed sperm nuclei the real substrates of the mitotic H3 kinase are the nucleosomal fibers. Thus, we next asked if rpEg2 could phosphorylate histone H3 at serine 10 within such complex structures. To address this problem we have prepared heat-treated Xenopus egg extract, which has allowed us to decondense sperm nuclei without histone H3 phosphorylation. Upon heating at 80°C the kinase activities of the extract are completely eliminated (35), but its sperm decondensation ability is preserved due the thermostability of nucleoplasmin, the protein responsible for the sperm nucleus decondensation (36). As shown and in agreement with previously published data (35,36), the incubation of sperm nuclei in the heated extract resulted in its impressive decompaction (Fig. 7, column 1). Addition of recombinant pEg2 led to the phosphorylation of histone H3 (Xenopus sperm nuclei contain an amount of H3-H4 tetramer identical to the one of somatic nuclei, for details see Ref. 37), thus demonstrating the ability of rpEg2 to penetrate decondensed sperm and act on histone H3.
The Incorporation of pEg2 in Chromatin during Chromosome Assembly Parallels the Phosphorylation of Histone H3-All the In lanes 1 and 2 are shown the pattern of the guanines (G) and the DNase I digestion products of the free 152-bp DNA fragment. The arrow shows the dyad axis of the particle. The other types of reconstituted nucleosomes used in this study exhibited the same DNase I footprinting pattern, but for simplicity only the patterns of the above three ones are shown.
FIG. 5. Recombinant pEg2 phosphorylates only histone H3 in nucleosomes reconstituted with intact histones. Nucleosomes were reconstituted by using bulk nucleosomal DNA and different combinations of recombinant histones; ⌺H, particles containing intact histones; ⌺GH, nucleosomes reconstituted with the globular domains of the histones; H2A, "chimerical" particles containing intact histone H2A only and the globular domains GH2B, GH3, and GH4 of the other histones; H2B, H3, and H4, "chimerical" nucleosomes, comprising either intact H2B or H3 or H4 and the globular domains of the three other histones, respectively. The kinase reaction using the different nucleosomal samples was carried out as described in Fig. 2. A, EMSA of the reconstituted nucleosomes in 2% agarose gel. The staining was performed with ethidium bromide; B and C represents the autoradiograph and the Coomassie-stained gel of the proteins from the respective kinase assays.
pEg2 Kinase and H3 Phosphorylation above data show that histone H3 is a good substrate for rpEg2. The next question addressed was whether the native enzyme, present in the extract, is also able to phosphorylate efficiently H3 at serine 10. To this end we have immunoprecipitated pEg2 with a monoclonal antibody from the extract and carried out a kinase assay by using as substrates H3, H3S10A, and H3S28A. As seen on Fig. 8A, the radioactive label is incorporated only in H3 and H3S28A. The mutation of serine 10 completely abolished the labeling. This fact clearly shows that native pEg2 specifically modifies histone H3 at serine 10.
In addition, incubation of sperm nuclei in Xenopus mitotic extract resulted in the uptake of the kinase from the extract and in its association with the remodeled nuclei structures (Fig. 8B). In other words during chromosome assembly native pEg2 binds to chromatin. Importantly, the kinetics of histone H3 phosphorylation strictly parallels the kinetics of uptake of pEg2 from the extract; the more pEg2 is found bound to chromatin, the more histone H3 phosphorylation is observed. These results suggest that pEg2 participates in the phosphorylation of histone H3. However, quantitative estimation showed that the chromatin-bound enzyme represented no more than 1-2% of the enzyme present in the extract (data not shown). This low amount of the chromatin-associated enzyme could explain the failure of previous attempts to localize pEg2 on chromosomes by using indirect immunofluorescence (27). FIG. 6. Specific phosphorylation of the tail of histone H3 at serine 10 in reconstituted nucleosomes. Nucleosomes were reconstituted with recombinant histones and used as substrates for a rpEg2 kinase assay. ⌺H and ⌺GH, particles comprising the four intact histones or their globular domains only; H3S10A and H3S28A, nucleosomes reconstituted with wild H2A, H2B, and H4 and histone H3 mutated at serine 10 or at serine 28, respectively; H3, "chimerical" nucleosomes containing the three globular domains GH2A, GH2B, and GH4 of the respective histones and the intact H3. A, EMSA of the reconstituted particles in 2% agarose. B, autoradiography visualization of the 32 P incorporation in the reconstituted particles. The Coomassie stained gel is shown in C.  8. A, native pEg2 phosphorylates specifically histone H3 at serine 10. Monoclonal antibodies against pEg2 were immobilized on protein A-Sepharose and incubated for 1 h in mitotic extract isolated from Xenopus eggs. Then the protein A-Sepharose beads were pelleted and washed, and the activity of the associated native pEg2 was analyzed by using as substrates either histone H3 or histone H3 mutated at serine 10 (H3S10A) or at serine 28 (H3S28A), respectively. The efficiency of phosphorylation was visualized by autoradiography and Western blotting. B, association of pEg2 with chromatin during in vitro chromosome assembly. Sperm nuclei were incubated in mitotic Xenopus egg extract for the time indicated and pelleted by centrifugation. The pelleted material was then loaded on 15% polyacrylamide SDS gel, and the presence of pEg2 and the degree of H3 phosphorylation was detected by Western blotting.

pEg2 Kinase and H3 Phosphorylation
Depletion of pEg2 and Histone H3 Phosphorylation-We have found that in vitro pEg2 selectively and efficiently phosphorylated serine 10 of histone H3 when using as substrates reconstituted nucleosomes or decondensed sperm in heated extracts. Furthermore, during chromosome assembly in Xenopus extract the kinase associates with chromatin, and the time dependence of the amount of bound kinase is very similar to the kinetics reflecting the increase of phosphorylation of histone H3 at serine 10. Taken together, these data strongly suggest that pEg2 could be involved in mitotic-specific phosphorylation of histone H3. We have addressed this hypothesis by immunodepleting pEg2 from the extract followed by chromosome assembly. The structural transitions of the sperm nuclei and the presence and distribution of phosphorylated histone H3 at serine 10 were visualized by using indirect immunofluorescence.
pEg2 was very efficiently removed from the extract at the conditions used for immunodepletion (Fig. 9A). Quantitative measurements show that more than 95-98% of pEg2 present in the extract was depleted by the monoclonal antibody (data not shown). Histone H3 is efficiently phosphorylated in the sperm nucleus structures formed in both control and pEg2-depleted extracts (Fig. 9B). Incubation of sperm nuclei in the different extracts (Fig. 9C) results in the already described sperm nucleus rearrangements (22) and culminates in the formation of condensed chromosomes. The time course of H3 phosphorylation pattern as well as the specific structures of the chromosome intermediates assembled in the control and pEg2depleted extracts were very similar (Fig. 3C). In some experiments a delay in the kinetics of chromosome assembly was observed, but the final chromosome structures as well as the histone H3 distribution were essentially the same (data not shown).

DISCUSSION
During cell division in all eukaryotic organisms studied chromosome assembly is accompanied by a phosphorylation of the flexible tail of histone H3 at serine 10 (38). The evolutionary conserved character of this modification suggests that it should play an important role in both mitosis and meiosis. Since the discovery of histone H3 phosphorylation in the late seventies (6) numerous studies have addressed this problem, but it still remains an enigma (for recent reviews see Refs. 38 and 39). The identification of the mitotic kinase(s) responsible for the phosphorylation of histone H3 at serine 10 will undoubtedly be of great help in the understanding of the function of this modification. It is quite possible that these enzymes exist in the cell as high molecular mass protein complexes. The kinase partners in the complexes should exhibit the property to target the kinases to chromatin, to help them in a local remodeling of chromatin structure, and to make available histone H3 tails for efficient phosphorylation. Thus, the understanding of the mechanism of histone H3 kinase action during cell division is inherently related to the isolation and characterization of such complexes. The extracts isolated from Xenopus eggs represent the perfect reagent for these types of experiments, since they contain very strong kinase activities that phosphorylate serine 10 of histone H3 during in vitro chromosome assembly (22).
pEg2, a Good Candidate for Histone H3 Mitotic Kinase-Our long term goal is to characterize the H3 mitotic kinase complex(es) and to understand the relevance of histone H3 phosphorylation for cell division. As a first step toward this goal, we have tried to identify the kinase phosphorylating histone H3 at serine 10 in Xenopus egg extract. Since kinases from divergent families were found to exhibit mitotic histone H3 phosphorylation activity (15,17), initially we have concentrated on a much simpler system, the yeast S. cerevisiae. By using in gel activity assay, we have confirmed that in S. cerevisiae the Ipl1 aurora kinase is likely to be the major H3 mitotic kinase. We further hypothesized that in Xenopus, enzymes from the same family should be able to phosphorylate histone H3 and to play a similar role in mitosis. We have focused on pEg2, the major Xenopus mitotic aurora-A kinase, and have studied in vitro its histone phosphorylation capability. We have shown that recombinant pEg2 phosphorylates both H3 and H2B when free in solution, but in nucleosomes only histone H3 phosphorylation at serine 10 was detected. Moreover, decondensation of sperm nuclei in heated extract made histone H3 accessible to rpEg2 and phosphorylatable by the enzyme. Native pEg2, isolated by immunoprecipitation from the Xenopus egg extract, was also able to phosphorylate specifically histone H3 at serine 10. In addition, during in vitro chromosome assembly pEg2 was recruited from the extract and associated with the remodeled sperm nuclei. Importantly, the kinetics of histone H3 phosphorylation parallels closely that of pEg2 association with chromatin. All these data designate pEg2 as a good candidate for an H3 mitotic kinase in Xenopus eggs.
Multiple Kinases Might be Involved in the Mitotic Phosphorylation of Histone H3-Mitotic chromosome assembly is a very FIG. 9. Effect of pEg2 depletion on in vitro chromosome assembly and histone H3 phosphorylation. Xenopus egg mitotic extracts were depleted from pEg2, and chromosomes were assembled in control and pEg2-depleted extracts. A, the monoclonal anti-pEg2 antibody depleted very efficiently the extract from pEg2. Lane 1, Western blotting with anti-pEg2 of control; lane 2, mock-depleted; and lane 3, pEg2depleted extract. B, phosphorylation of histone H3 in sperm nuclei structures incubated in control (lane 1), mock-depleted (lane 2), and pEg2-depleted extracts (lane 3) for the time indicated. Following incubation the chromosome intermediates were pelleted by centrifugation, and histone H3 phosphorylation was visualized by Western blotting using anti-phosphorylated histone H3 antibody. C, time course of chromosome assembly in control and pEg2-depleted extracts. Sperm nuclei were incubated in the extracts, and at the time indicated, aliquots from the assembly reactions were removed and immediately fixed. DNA was visualized by Hoechst 33258 and phosphorylation of histone H3 (H3P) by indirect immunofluorescence. complex process, and undoubtedly it should require the active participation of numerous protein factors. Two well defined chromosome assembly factors are topoisomerase II and the condensin complex (1). A good candidate for a third one is the kinase phosphorylating histone H3 in mitosis (22,38,39). The data presented in this study strongly suggest that pEg2 is a histone H3 mitotic kinase in Xenopus eggs. However, immunodepletion of pEg2 from the extract has no considerable effect on both chromosome formation and histone H3 phosphorylation. Thus, if pEg2 is a bona fide histone H3 mitotic kinase, how could this result be explained? A plausible explanation is the existence of several and redundant in function kinases involved in the mitotic phosphorylation of H3. This hypothesis is supported by our in gel activity assay data that suggest the existence in S. cerevisiae of three H3 mitotic kinases with different molecular masses. After chromatographic fractionation of Xenopus extract, several distinct activities able to phosphorylate nucleosomes and free histone H3 were detected. 3 Recently, the Xenopus aurora-B kinase was cloned (40). Keeping in mind the similarity of the catalytic domain of aurora-B and pEg2 (72% identity), one could expect that aurora-B will also be able to phosphorylate efficiently histone H3 at serine 10. Interestingly, aurora-B is stored in Xenopus eggs in a complex with the chromosomal passenger protein INCENP. At metaphase, this protein is localized on centromeres, and at anaphase it migrates to the central spindle and the equatorial cortex (41). The specific localization of this kinase suggests its participation in centromeric histone H3 phosphorylation.
It is also conceivable that kinases from evolutionarily divergent families could operate in different organisms. For example, two recent reports identified the aurora kinases Ipl1 in S. cerevisiae and air2 in C. elegans (15) but also NIMA kinase in A. nidulans (15) as histone H3 mitotic kinases. Importantly, the NIMA murine homologue Nek2, contrary to AIRK2, was shown to localize close to the ends of meiotic metaphase chromosomes (42) further supporting the hypothesis of the existence of different kinases functioning on distinct chromosomal domains.
"Targeted" Versus "Bulk" Histone H3 Phosphorylation?-The suggested histone H3 phosphorylation scenario is reminiscent of the mechanism of histone acetylation. Within the same organism histones can be acetylated by a large number of histone acetyltransferases and deacetylated by numerous histone deacetylases. The various histone acetyltransferases and histone deacetylases belong to different families and form high molecular mass complexes with other proteins in the cell (43)(44)(45)(46). The histone acetyltransferases could affect a "targeted" (promoter specific or chromatin domain specific) or "bulk" acetylation of histones (47,48). In vivo, the potential mitotic H3 kinases also exist as high molecular mass complexes (18,40). By analogy with histone acetylation, one could think that the targeted H3 phosphorylation is determined by a distinct partner of the complex, which recognizes a specific domain of the chromosome. For example, in Xenopus in the case of aurora-B, INCENP could target the kinase to centromeres (40). RNA interference data suggest that in C. elegans the survivin-like BIR-1 protein might form a complex with the presumed mitotic H3 kinase AIR2 (18) and play a role analogous to INCENP.
How many chromosome domains could be modified by "targeted" phosphorylation? Analysis of the literature (see above and Ref. 39) shows only one candidate for the moment, the pericentromeric chromatin. However, in most of the studied systems the chromosomes are phosphorylated along their entire length. Thus, it could be that some mitotic kinase activities are involved in "bulk" phosphorylation of histone H3. Such suggestion implies some accessibility of histone H3 tail for phosphorylation by the non-chromosomal region-specific enzymes. Our data on the phosphorylation of decondensed sperm in heat-treated extracts by recombinant pEg2 are in agreement with this proposal.