The Xenopus laevis Aurora-related Protein Kinase pEg2 Associates with and Phosphorylates the Kinesin-related Protein XlEg5*

We have previously reported on the cloning of XlEg5, a Xenopus laevis kinesin-related protein from thebimC family (Le Guellec, R., Paris, J., Couturier, A., Roghi, C., and Philippe, M. (1991) Mol. Cell. Biol. 11, 3395–3408) as well as pEg2, an Aurora-related serine/threonine kinase (Roghi, C., Giet, R., Uzbekov, R., Morin, N., Chartrain, I., Le Guellec, R., Couturier, A., Dorée, M., Philippe, M., and Prigent, C. (1998) J. Cell Sci. 111, 557–572). Inhibition of either XlEg5 or pEg2 activity during mitosis in Xenopus egg extract led to monopolar spindle formation. Here, we report that inXenopus XL2 cells, pEg2 and XlEg5 are both confined to separated centrosomes in prophase, and then to the microtubule spindle poles. We also show that pEg2 co-immunoprecipitates with XlEg5 from egg extracts and XL2 cell lysates. Both proteins can directly interactin vitro, but also through the two-hybrid system. Furthermore immunoprecipitated pEg2 were found to remain active when bound to the beads and phosphorylate XlEg5 present in the precipitate. Two-dimensional mapping of XlEg5 tryptic peptides phosphorylatedin vivo first confirmed that XlEg5 was phosphorylated by p34 cdc2 and next revealed that in vitro pEg2 kinase phosphorylated XlEg5 on the same stalk domain serine residue that was phosphorylated in metabolically labeled XL2 cells. The kinesin-related XlEg5 is to our knowledge the first in vivo substrate ever reported for an Aurora-related kinase.

Mitotic spindle assembly is a complex and essential event in the cell division process because it is a prerequisite for chromosome segregation (1). During this process, motor proteins and the microtubules of the bipolar spindle ensure even distribution of the genomic information present in daughter cells (2). In order to assemble the bipolar mitotic spindle in metaphase, cells re-organize their entire microtubule network upon entering in mitosis primarily by means of phosphorylation reactions (3,4).
Motor protein activities such as association with microtubules, interaction with other proteins, and subcellular localization are regulated by phosphorylation reactions (18 -21).
The X. laevis pEg2 centrosome protein kinase was first isolated through a differential screening undertaken during early development (22). pEg2 mRNA belongs to the Eg family of mRNA, which is adenylated, recruited in polysomes, and translated in unfertilized eggs and deadenylated after fertilization (23). The kinase was also recently isolated during a screen designed to isolate early players in the progesterone-induced maturation pathway of the Xenopus oocyte. In the oocyte, the overexpression of active pEg2 accelerates the appearance of the germinal vesicle breakdown after contact with progesterone (24).
In Xenopus egg extracts, pEg2 kinase activity is required for mitotic spindle assembly (16). pEg2 is also a microtubule-associated protein (16). In vitro pEg2 binds to paclitaxel-stabilized microtubules independently of its kinase activity (25), whereas in vivo pEg2 binds to mitotic microtubule structure and not to interphasic microtubule networks (25).
pEg2 belongs to a family of protein kinases related to Drosophila Aurora (26) and Saccharomyces cerevisiae Ipl1 (increase in ploidy) (27). Aurora gene mutation is characterized by a centrosome separation defect, a mechanism that needs to be completed in prophase to allow bipolar spindle assembly to occur (26). The mutation manifestations in the Ipl1 gene show a chromosome segregation defect (27). Three different Aurorarelated kinases were found in mammalian cells (mouse and human) (28 -31). Two of the human Aurora-related kinases were found to be overexpressed in several human cancer types (30 -33), and one was shown to be transforming when overexpressed (30,31).
The X. laevis XlEg5 microtubule-based motor protein is a kinesin-related protein that was isolated through the same differential screening as pEg2 (5), and its mRNA behaves in the same manner of pEg2 mRNA during early development (23). Like pEg2, XlEg5 activity is required for mitotic spindle assembly in Xenopus egg extract (34). The XlEg5 sequence ranked it in the bimC family of kinesin-related protein, which includes yeast, fungus, Drosophila, and human proteins (35)(36)(37)(38)(39). All members of the bimC family have a highly preserved motor domain (N-terminal domain) and a small preserved domain in the tail domain (C-terminal domain) containing a threonine residue in a phosphorylation consensus sequence for p34 cdc2 . The mutations of the potential phosphorylation site for p34 cdc2 in XlEg5 inhibited its binding to the mitotic spindle (18). In addition, p34 cdc2 phosphorylation of the human Eg5 (HsEg5) threonine 1067 was shown to control any association with the spindle in vivo (19) and interaction with the dynactin component p150 Glued (20). XlEg5 was suspected to be required for centrosome separation because the phenotypes obtained after XlEg5 inhibition are reminiscent of those obtained after cytoplasmic dynein inhibition, which is involved in centrosome migration (19,40).
In Drosophila, mutations in both Aurora (related to pEg2) and KLP61F (related to XlEg5) genes are characterized by improper centrosome positioning, leading to the formation of monopolar mitotic spindles (26,39,41). In Xenopus egg extracts, both the inhibition of XlEg5 through the addition of antibodies (34) and the inhibition of pEg2 through the addition of an inactive dominant negative pEg2 form (16) provoke an inhibition of the mitotic spindle assembly. These results suggest that pEg2 may act on a substrate involved in centrosome separation, and one of the obvious candidates is the kinesinrelated protein XlEg5.
In Xenopus cultured cells, pEg2 associates with the centrosome in G 2 and binds to microtubules at the mitotic spindle poles (16). XlEg5 also binds to the microtubules over the entire spindle (18). In interphase XlEg5 staining has been described as a "weak cytoplasmic staining" and was not detected specifically around centrosomes (18). Here, we re-examined XlEg5 subcellular localization in Xenopus XL2 cultured cells and showed that the protein started to accumulate between the duplicated centrosomes at the end of S phase and associated with the centrosome in prophase where pEg2 was localized.
In this study, we looked for evidence of direct interaction between pEg2 and XlEg5 and demonstrated that pEg2 associates with XlEg5. In addition, we demonstrated that pEg2 is capable of phosphorylating a serine in the stalk domain of XlEg5. Our results definitely suggest that XlEg5 is phosphorylated in vivo on two residues with two different kinases; a threonine residue is phosphorylated by p34 cdc2 , and a serine is phosphorylated by pEg2.

EXPERIMENTAL PROCEDURES
Xenopus Eggs and Cultured Cells-X. laevis oocytes and eggs were obtained from laboratory-reared females. The embryonic X. laevis cells XL2 (42) were grown at 25°C, without CO 2 in L15 Leibovitz medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Bio Times) and antibiotic-antimycotic (Life Technologies, Inc.).
Antibodies-Anti-pEg2 1C1 antibody is a mouse monoclonal antibody described by Roghi et al. (16). Anti-XlEg5 antibodies are rabbit polyclonal antibodies described by Sawin et al. (34), with an anti-stalk/ tail domain and an anti-tail domain. The antibodies raised against the C-terminal end of XlEg5 were purified on a nitrocellulose membrane containing the tail domain of XlEg5 fused to glutathione S-transferase. Briefly, 5 g of GST-XlEg5T protein were run on an SDS-polyacrylamide gel and transferred onto nitrocellulose. The proteins were subsequently stained with Red Ponceau, and the bands corresponding to GST-XlEg5T protein were cut out. The various pieces of the membrane (10 cm 2 ) were first incubated at 4°C in TBST (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% Tween 20) containing 5% dry skim milk for 2 h and then incubated overnight at 4°C in 10 ml of TBST containing 2.5% milk and 1 ml of anti-tail domain serum. The pieces of the membrane were next washed with TBST, and the anti-tail antibodies were eluted with 2 ml of 100 mM glycine-HCl, pH 2.9, and immediately neutralized in 200 l of 1 M Tris-HCl, pH 8. The purified antibodies were concentrated and washed with PBS in a Centricon 30 (Amicon) following the manufacturer's instructions.
Indirect Immunofluorescence Microscopy-X. laevis XL2 cells were grown on round coverslips in 12-well plates (Corning Inc.) for 48 h, washed with phosphate-buffered saline (PBS: 136 mM NaCl, 26 mM KCl, 2 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.2) and fixed by immersion in cold (Ϫ20°C) methanol for 6 min. Following successive washes in PBS, the cells were blocked in PBS containing 3% BSA for 30 min and then incubated with both mouse anti-pEg2 monoclonal antibody 1C1 (20 g/ml) and rabbit anti-XlEg5 purified polyclonal antibody (anti-tail) (dilution 1/60). The antibodies were sequentially revealed by fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (dilution 1/75) (Interchim) and Texas Red-conjugated goat anti-mouse IgG (dilution 1/35) (Interchim). All antibody reagents were diluted in PBS containing 1% BSA. Incubations were performed at room temperature for 60 min. The cells were rinsed in PBS containing 1% BSA between each incubation. The coverslips were rinsed in PBS and mounted in a Mowiol containing antifade. All samples were observed using a Zeiss fluorescent microscope (Axiovert 35) and photographed using a Nikon 601 camera.
Immunoprecipitation-The egg extracts were prepared as described by Lohka and Maller (44) and diluted 10 times in a TBS-IP buffer (20 mM Tris-HCl, pH 8, 150 mM NaCl, complemented with an IP buffer composed as followed: 0.5% Nonidet P-40, 5 mM EDTA, 3 mM EGTA, 5 mM glycerophosphate, 0.5 mM sodium vanadate, 0.5 g/ml each leupeptin, pepstatin, and chymostatin). The XL2 cells were lysed in a PBS-IP buffer at 4°C for 30 min. After centrifugation at 13,000 ϫ g at 4°C for 15 min, the supernatant was used for immunoprecipitation. Anti-pEg2 1C1 monoclonal antibodies were conjugated to protein G-Sepharose (Amersham Pharmacia Biotech), whereas anti-XlEg5 polyclonal antibodies were conjugated to protein A-Sepharose (Amersham Pharmacia Biotech). In both cases, the beads were saturated by incubation at 4°C for 1 h in PBS containing an excess of antibodies and extensively washed with PBS. For the immunoprecipitation of XlEg5 from Xenopus XL2 cells, 50 l of purified XlEg5 anti-tail domain antibodies (10 g) were covalently bound to 50 l of CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech) following the manufacturer's instructions.
Regarding immunoprecipitations, 10 l of beads was added both to 50 l of diluted egg extract and to 1.5 ml of cell lysate (2 ϫ 10 6 cells) and incubated at 4°C for 1 h on rotating wheel. After centrifugation, the beads were washed three times with 1 ml of a corresponding buffer (TBS-IP buffer or PBS-IP buffer). An additional 30-min wash on a rotating wheel was performed when labeled proteins were immunoprecipitated. The beads were then heat-denatured by incubation at 90°C for 10 min in 20 l of Laemmli sample buffer, and immunoprecipitated proteins were analyzed on SDS-polyacrylamide gels, followed by electrotransfer onto nitrocellulose membranes and then immunodetection (45,46).
If the beads were to be used for phosphorylation reactions, they were further washed twice with 1 ml of kinase buffer: 50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol.
Affinity Chromatography on a Nickel-NTA-Agarose Column-10 g of purified histidine-tagged protein (pEg2-(His) 6 or Sup35-(His) 6 ) were incubated at 4°C with 10 l of dried Ni-NTA-agarose beads in IMAC20 (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 10% glycerol, 20 mM imidazole) for 30 min. After extensive washing with IMAC20 buffer, the beads were incubated at 4°C for 30 min in 100 l of interaction buffer, which was IB (50 mM Tris-pH 8, 100 mM KCl, 5 mM MgCl 2 , 0.1% Triton X-100, 20% glycerol) containing 1 l of rabbit reticulocyte lysate previously programmed with XlEg5. The beads were then washed three times with 1 ml of IB containing 20 mM imidazole. Histidine-tagged proteins and proteins bound to histidine-tagged proteins were eluted with 15 l of IB containing 250 mM imidazole. After heat denaturation in a Laemmli sample buffer (45) for 10 min at 90°C, the proteins were analyzed by SDS-polyacrylamide gel electrophoresis.
Western Blot Analysis-Electrophoresis on SDS-polyacrylamide gel was performed according to Laemmli (45) and gels transferred onto nitrocellulose membranes as described by Towbin et al. (46). The membranes were blocked in TBST containing 5% skim milk for 2 h at 4°C and incubated at 4°C for 1 h with antibodies diluted in TBST containing 2.5% skim milk. Immunocomplexes were revealed with antibodies coupled with alkaline phosphatase (Sigma) using either nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate as substrates or a chemiluminescence according to the manufacturer's instructions (NEN Life Science Products).
In Vitro Transcription and Translation-1 g of pBluescript containing XlEg5 (minus 39 nucleotides encoding 13 missing amino acids in the N-terminal end; Refs. 5 and 34) was transcribed and translated in 25 l of reaction mix containing 20 Ci of [ 35 S]methionine (Amersham Pharmacia Biotech) using a TNT-T7 quick coupled transcription translation system following the manufacturer's instructions (Promega).
Protein Kinase Assay-The assays were performed in 10 l of 50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM dithiothreitol, 10 mM MgCl 2 , 10 M ATP containing 3000 Ci/mmol [␥-32 P]ATP (Amersham Pharmacia Biotech). The reactions were incubated at 37°C for 15 min, stopped by addition of 10 l of 2 ϫ Laemmli SDS sample buffer and denatured at 90°C for 10 min. The proteins were then separated by SDS-polyacrylamide gel electrophoresis, electrotransferred onto nitrocellulose membranes and analyzed by autoradiography.
Metabolic Labeling of XL2 Cells-After serum starvation, the XL2 cells were blocked in the G 1 /S border by aphidicolin as described by Uzbekov et al. (49). They were then released in the cycle and cultured for 2 h in a phosphate-free Leibowitz L-15 medium (Life Technologies, Inc.) containing 10% fetal calf serum dialyzed at 4°C overnight against TBS. The cells were then cultured for 5 h in 1 ml of the same medium containing 2 mCi of [ 32 P]orthophosphorus (Amersham Pharmacia Biotech). When the cells reached the end of G 2 or the beginning of M, they were washed three times with cold PBS and lysed on ice in 1 ml of 20 mM phosphate buffer, pH 7.5, 1% Nonidet P-40, 500 mM NaCl, 0.25% SDS, 2 mM EDTA, 5 mM MgCl 2 , 0.5 mM sodium vanadate, 0.5 g/ml each leupeptin, pepstatin, and chymostatin. After centrifugation at 13,000 ϫ g at a temperature of 4°C for 15 min, the supernatant was submitted to immunoprecipitation.
Phosphoamino Acid Analysis-The immunoprecipitated proteins were separated on SDS-polyacrylamide gels and transferred onto Immobilon membranes (Applied Biosystems). A piece (0.4 cm 2 ) of membrane containing the radioactive protein was cut and incubated at 110°C for 1 h in 200 l of 6 N HCl under nitrogen atmosphere for amino acid hydrolysis. The HCl was evaporated in a Speed-Vac. The pellet was washed three times with water and dissolved in 10 l of 2.5% v/v formic acid, 7.8% v/v acetic acid, pH 1.9. The amino acids were loaded together with 2.5 g of nonradioactive phosphoserine, phosphothreonine, and phosphotyrosine (Sigma) on a thin layer cellulose (TLC) plate (Sigma) and submitted to high voltage electrophoresis (50) using a Hunter thin layer peptide mapping system (model HTLE-7000). Non radioactive amino acids were revealed with ninhydrin (Sigma) and radioactive amino acids by autoradiography.
Phosphopeptide Mapping-The immunoprecipitated proteins were separated on SDS-polyacrylamide gels and transferred onto nitrocellulose membranes (Amersham Pharmacia Biotech). A piece of membrane (0.4 cm 2 ) containing the radioactive protein was cut out and incubated in 0.5% polyvinylpyrrolidone and 10 mM acetic acid at 37°C for 30 min, then washed twice with water followed by the addition of 50 mM ammonium bicarbonate. The protein was digested by incubating the piece of membrane in 400 l of 50 mM ammonium bicarbonate containing 10 g of trypsin (Roche Molecular Biochemicals) for 2 h, followed by an overnight incubation at 37°C after the addition of another 10 g of trypsin. The tryptic peptides released from the membrane were washed twice in 100 l of water and lyophilized. The peptides were then oxi-dized at 0°C for 1 h in 50 l of performic acid (10% v/v 30% H 2 O 2 and 90% v/v formic acid). The tryptic peptides were washed three times in water and lyophilized again. Next they were dissolved in 5 l of water and analyzed by two-dimensional separation on a TLC plate (20 cm ϫ 20-cm polyester silica gel, 100 m thick; Sigma). The tryptic peptides were then first submitted to high voltage electrophoresis at pH 8.6 in a Hunter thin layer peptide mapping system (model HTLE-7000) (1000 V for 20 min) and then submitted to ascendant chromatography in a buffer containing 37.5% butanol, 25% pyridine, and 7.5% acetic acid (51). The plate was dried and the phosphotryptic peptides detected in a PhosphorImager (Molecular Dynamics).

RESULTS
pEg2 and XlEg5 Localize in the Centrosome and at the Poles of the Mitotic Spindle-The first evidence that pEg2 might interact with XlEg5 came from the indirect immunodetection of the proteins in Xenopus XL2 cultured cells. We used the specific monoclonal antibody 1C1 directed against pEg2 (16) and an affinity-purified polyclonal antibody directed against the specific C-terminal domain of XlEg5 (34). Both antibodies are highly specific. In XL2 cell lysate, anti-Eg2 antibody detected a single protein with a relative molecular mass of 46 kDa (Fig.  1A, lane 1), while anti-XlEg5 antibodies detected a doublet at 130 kDa (Fig. 1B, lane 1).
In XL2 cells, no protein is apparent prior to the end of S phase. pEg2 appears as two bright spots on the duplicated centrosomes at the end of S phase (Fig. 2B) (16). XlEg5 can only be detected between the two duplicated centrosomes after pEg2 appears on the centrosome (Fig. 2C). At this stage, the two proteins localize differentially (Fig. 2D).
XlEg5 and pEg2 were both detected among the centrosomes in prophase, when they had reached their position at opposite side of the nucleus (Fig. 3, B and C). However, XlEg5 seemed to occupy a larger area than pEg2 in the centrosome (Fig. 3D, upper left) as revealed by double staining with a yellow dot in the centrosome and a green halo around each centrosome (Fig.  3D). Indirect electron microscopy showed that both proteins are localized around the pericentriolar material (data not shown). From that time on, both proteins localized on same structures throughout mitosis. During prometaphase and metaphase, both proteins slipped from the centrosome position to invade the spindle microtubules. pEg2 appeared as "cup shape" staining only at the spindle poles (Fig. 3, F and J), whereas XlEg5 was also found along the spindle microtubules (Fig. 3, G and K). In anaphase, XlEg5 appeared to move back to the pole of the spindle and was characterized by a weaker staining of spindle microtubules (Fig. 3O). From prometaphase to anaphase, both pEg2 and XlEg5 localized on the microtubules that form the poles region of the spindle. A proportion of pEg2 remained associated with the centrosome, while XlEg5 was rather found only on the spindle microtubules. Double staining always showed a red dot at the centrosome, a yellow pole, and green microtubules (Fig. 3, H, L, and P). In telophase, both proteins pEg2 Binds to and Phosphorylates XlEg5 still localized in the centrosome with a dramatic decrease in XlEg5 staining. Moreover, XlEg5 relocated on the microtubules in the midzone without any association with pEg2 (Fig. 3, R, S, and T). To summarize, both proteins are present around the centrosome in prophase and on microtubules during metaphase and anaphase.
pEg2 and XlEg5 Co-immunoprecipitate-Given the fact that both proteins localized on the same structures in the cell during mitosis, we searched for evidence of direct interaction between the two proteins in immunoprecipitation. pEg2 and XlEg5 were immunoprecipitated from protein extracts prepared from Xenopus unfertilized eggs using the same antibodies intended for indirect immunofluorescence. The precipitates were subjected to Western blot analysis. Control immunoprecipitations did not reveal any unspecific interactions (Fig. 4, lanes 1 and 3). XlEg5 was found among pEg2 immunoprecipitates and pEg2 among XlEg5 immunoprecipitates (Fig. 4, lanes 2). In the case of both XlEg5 and pEg2 precipitates, XlEg5 appeared as a doublet, as had already been reported for HsEg5 (19), suggesting that different forms of XlEg5 associate with pEg2. The comprehensive immunodepletion of pEg2 from Xenopus extracts revealed that 5-10% of total Xenopus egg XlEg5 molecules was present in the pEg2 immunoprecipitate (data not shown). pEg2 and XlEg5 were also co-immunoprecipitated from unsynchronized Xenopus XL2 cell lysate, but the signals were very weak (data not shown), which would suggest that the association process depends on the cell cycle. This would tend to agree with and confirm the previous immunofluorescence studies, which have indicated that both proteins interact only in mitosis.
pEg2 Interacts Directly with XlEg5-Although pEg2 and XlEg5 turned out to be co-immunoprecipitated, we wished to demonstrate that pEg2 interacted directly with XlEg5. Purified bacterially expressed pEg2-(His) 6 was affinity-bound to nickel beads (Ni-NTA-agarose) and mixed with reticulocyte lysateexpressed [ 35 S]methionine labeled XlEg5. Reticulocyte lysate was used because XlEg5 is known to be highly insoluble when produced in bacteria. The various controls were performed with beads alone (Fig. 5, lane 2) or beads containing another histidine-tagged protein: Sup35-(His) 6 (the eukaryotic release factor eRF3) (43) (data not shown). After incubation at a temperature of 4°C for 60 min, the histidine-tagged proteins together with associated proteins were eluted with 250 mM imidazole. [ 35 S]Methionine-labeled XlEg5 was detected only in the fraction eluted from the nickel column that contained pEg2-(His) 6 (Fig. 5, lane 1). Although 90% of [ 35 S]methionine-labeled XlEg5 loaded onto the column was recovered in the flow-through, 10% was specifically retained by pEg2-(His) 6 , demonstrating that XlEg5 produced in reticulocyte lysate can bind to pEg2. Such poor interaction may be explained by the fact that pEg2-(His) 6 produced in bacteria might not be appropriately folded. Although the purified kinase is active, its activity is much lower than that of immunoprecipitated pEg2 (data not shown).
pEg2 Interacts with XlEg5 in the Two-hybrid System-Interaction between pEg2 and XlEg5 was assayed using the twohybrid system. Because it had been previously reported that kinesin-related proteins could form dimers (52), we used XlEg5 dimerization as a positive test for the assay. Two XlEg5 molecules associated through their stalk domain to form homodimers. Then, we inserted an XlEg5 cDNA encoding the stalk (S) and the tail domain (T) in both vector (pGADGH and pGBT9) used for the two-hybrid. The XlEg5 motor domain was not present in any of the construct because it has been previously reported that production of full-length HsEg5 in yeast was lethal (20). Strong ␤-galactosidase activity was obtained when homodimer formation was assayed and when the XlEg5ST construct proved to be present in both two-hybrid vectors (Fig. 6, lane 1). When pEg2 interaction with XlEg5ST was assayed, significant ␤-galactosidase activity was detected, indicating that both proteins interacted (Fig. 6, lane 2). This activity, however, was much lower than that of XlEg5ST/ XlEg5ST interaction. This significant difference in ␤-galactosidase activities may be explained by the generally low affinity of protein kinases for their substrates, which is much lower than the affinity of proteins that form stable dimers. However, the assay clearly showed that pEg2 interacts with XlEg5. Interestingly enough, this experiment also indicates that the motor domain of XlEg5 is not necessary for pEg2 interaction.
pEg2 Phosphorylates a 130-kDa Protein in the Immunoprecipitated Complex-Because the 1C1 monoclonal antibody is not an inhibitory antibody, immunoprecipitated pEg2 remains active and can be assayed for kinase activity. In this undertaking, pEg2 was immunoprecipitated from Xenopus egg extracts, and the precipitate was incubated with [␥-32 P]ATP. The radioactive proteins were analyzed by means of SDS-polyacrylamide gel electrophoresis (Fig. 7A). At least four phosphorylated proteins were detected in the immunoprecipitate: a 46-kDa protein that was proved to be autophosphorylated pEg2, a 90-kDa protein, a 130-kDa protein that was highly phosphorylated, and a very high molecular mass protein that migrated above the 200-kDa marker (Fig. 7A, lane 3). The highly phosphorylated 130 kDa protein migrated at the same position as the XlEg5 identified by Western blot analysis and present only in the immunoprecipitate containing pEg2 (Fig. 7B, lane 3, top  panel).
In order to confirm that the phosphorylation observed in the immunoprecipitate was due to pEg2 kinase activity, we added to the reaction an excess of the bacterially expressed dominant negative mutant of pEg2, pEg2-K/R-(His) 6 in which the lysine 169 was replaced by an arginine residue (16). The addition of 5 g of pEg2-K/R-(His) 6 inhibited 90% of XlEg5 phosphorylation (Fig. 7C, lane 4), while the addition of 5 g of BSA exerted no effect (Fig. 7C, lane 3). The addition of pEg2-K/R-(His) 6 specifically inhibited pEg2 activity, presumably by titrating its substrates. This result demonstrated that pEg2 phosphorylated a 130-kDa protein in the immunoprecipitate that migrated at the same position as XlEg5.
pEg2 Phosphorylates XlEg5 in Vitro in the Stalk Domain-Like the other kinesin-related proteins from the bimC family, XlEg5 is composed of three distinct domains, an N-terminal motor domain or head (H), a stalk domain (S), and a C-terminal tail (T) (Fig. 8A). Three different bacterially expressed XlEg5 truncated proteins were expressed in bacteria as GST fusion proteins, purified, and used in vitro as substrates for recombinant pEg2-(His) 6 protein kinase (Fig. 8B). The HS construct contained the head and stalk domain of XlEg5. The ST construct contained the stalk and tail domain, and the T construct contained only the tail domain. Because the GST-XlEg5 proteins were highly insoluble, the phosphorylation reactions were performed directly on GST proteins fixed to glutathione-agarose beads. Western blot analysis with anti-GST antibodies (Sigma) was performed to ensure that equal amounts of the three proteins were present in the reactions (data not shown). The two XlEg5 fusion proteins, HS and ST, which had the stalk domain in common, were phosphorylated only when pEg2 was present in the reaction (Fig. 8C, compare lanes 1 and 3 with  lanes 4 and 6). The third T protein, which contained only the tail domain, was not found to be phosphorylated even in the presence of pEg2, indicating that the tail is not the substrate (Fig. 8C, lanes 2 and 5). In order to determine if the label was located in the stalk domain or in the motor domain, two-dimensional mapping of the in vitro phosphorylated tryptic peptides of both proteins was performed. Both maps were identical. One major and one minor common phosphorylated tryptic peptide were detected, which demonstrated that only the stalk domain contained the in vitro phosphorylation site for pEg2 (data not shown). The XlEg5 tail domain contained a phosphorylation site for p34 cdc2 (18,19). Our results indicated that in vitro pEg2 phosphorylated XlEg5 in the stalk domain. In addition, phosphoamino acid analysis by high voltage electrophoresis revealed that in vitro pEg2 phosphorylated XlEg5 on a serine residue (Fig. 8D).
Phosphorylation State of XlEg5 in Vivo-The human protein HsEg5 was found to be phosphorylated in vivo on two distinct tryptic peptides, with one phosphorylated on a serine residue and the other on a threonine residue (19). p34 cdc2 phosphorylated the threonine residue located in the tail domain that is conserved within the bimC family of kinesin-related protein.
The kinase that phosphorylated the serine, which was the major phosphorylated site in HsEg5 in vivo, has not been identified yet (19).
In order to investigate the phosphorylation state of XlEg5 in vivo, the Xenopus XL2 cells were metabolically labeled with [ 32 P]orthophosphorus. Because p34 cdc2 is a mitotic kinase and because pEg2 and XlEg5 localized on the same structure only after duplication of the centrosomes in G 2 and M, the XL2 cells were labeled and harvested in the course of G 2 and M phases (49). 32 P-Labeled XlEg5 was then immunoprecipitated with affinity-purified polyclonal antibodies and digested with FIG. 4. pEg2 and XlEg5 co-immunoprecipitates. pEg2 and XlEg5 were immunoprecipitated from Xenopus egg extract. Immunoprecipitations were performed using either anti-pEg2 1C1 monoclonal antibody affixed to protein G-Sepharose (lane 2, left panel) or anti-XlEg5 purified polyclonal antibodies affixed to protein A-Sepharose (lane 2, right panel). Control immunoprecipitations were done either without antibody (lanes 1) or without extract (lanes 3). The presence of both proteins in different immunoprecipitates were detected using specific antibodies for XlEg5 (upper panels) or pEg2 (lower panels) and revealed by chemiluminescence. trypsin. The resulting phosphotryptic peptides were resolved with thin layer electrophoresis, followed by ascendant chromatography.
Like HsEg5, XlEg5 was phosphorylated in vivo on two tryptic peptides (Fig. 9B), one of which migrated to the same location as the peptide phosphorylated in vitro by p34 cdc2 (Fig. 9A,  black arrow), thereby demonstrating that XlEg5 was indeed phosphorylated in vivo by p34 cdc2 as had been suggested by Sawin and Mitchison (18). The second phosphotryptic peptides migrated to the same location as a peptide phosphorylated in vitro by recombinant pEg2 (Fig. 9, C and D), which strongly suggested that pEg2 phosphorylates XlEg5 in vivo. DISCUSSION pEg2 and XlEg5 are two proteins that have been reported to be necessary for bipolar mitotic spindle assembly in Xenopus egg extracts (16,34). The mutation of the genes encoding the Drosophila orthologues of Xenopus pEg2 and XlEg5, Aurora and KLP61F, respectively, causes a centrosome separation defect (26,39,41). These observations have led us to posit two questions: (1) is there a physical association between pEg2 and XlEg5, and (2) does pEg2 phosphorylate XlEg5?
We first investigated whether XlEg5 and pEg2 could be found on the same subcellular localization in Xenopus XL2 cells during the cell cycle. First, pEg2 appeared in the centrosome at the end of S phase. Then, 2 h later XlEg5 started to accumulate between the two duplicated centrosomes. In prophase, when both centrosomes had reached their position on each side of the nucleus, both pEg2 and XlEg5 were present in the separated centrosomes. During metaphase and anaphase, the area shared by pEg2 and XlEg5 moved to the poles of the spindle. In metaphase, while pEg2 was localized in the centrosome and on the microtubules present in the spindle poles, XlEg5 was found only on the spindle microtubules. In telophase, both proteins were found together in the centrosome, but XlEg5 was also found in the midzone in the absence of detectable pEg2. The localization of XlEg5 between the duplicated centrosome during the S phase and on the centrosome during prophase was in compliance with its presumptive role in centrosome separation (18,19). Its localization on the spindle microtubules was in compliance with its presumptive role in regulating the stability of the spindle (34). The localization of XlEg5 on the microtubule in the midzone may indicate a role in cytokinesis that remains to be demonstrated.
Although pEg2 and XlEg5 showed a shared area of localization on the mitotic spindle during the prophase-to-anaphase stages of the cell cycle, the localization process proved temporally different, suggesting that the molecular mechanisms of localization are different for the two proteins.
Furthermore, the physical association of the two proteins  6 as described under "Experimental Procedures" and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. Lanes 1-3, control reactions without pEg2 kinase added; lanes 4 -6, with pEg2-(His) 6 . D, phosphoamino acid analysis of XlEg5 phosphorylated in vitro by pEg2-(His) 6 . Amino acids obtained after hydrochloric acid hydrolysis were separated on TLC plate and autoradiographed. was detected by immunoprecipitation from metaphase-arrested Xenopus egg extract. Only partial co-immunoprecipitation was detected, indicating that the localization of pEg2 may depend on the localization of XlEg5 but also on another mechanism. When the same experiment was performed with lysates prepared from unsynchronized Xenopus XL2 cells, interaction was very weak. This phenomenon suggests that the interaction only occurs at a specific point in the cell cycle.
Recombinant pEg2 protein can directly interact with in vitro translated XlEg5. Accordingly, one may deduce that there is a direct physical interaction of the two proteins. It cannot be ruled out that a protein in the reticulocyte lysate may have helped physical interaction, although this is very unlikely because in vitro purified recombinant pEg2 phosphorylates purified recombinant XlEg5 which indicates direct interaction. Furthermore, both proteins interact through the two-hybrid system, demonstrating that pEg2 interacts with XlEg5. The XlEg5 motor domain is not necessary for interaction purposes. XlEg5 phosphorylation through pEg2 kinases is a physiological reaction, because the same phosphopeptide is found to be phosphorylated both in vivo and in vitro as a result of pEg2.
The site phosphorylated by pEg2 was found to be located in the stalk domain, which is the domain also involved in proteinprotein association and dimerization. The kinesin-related protein members of the bimC family can form homotetramers (52,53), with the dimer formed head-to-head via the stalk domain, and two dimers associated head-to-tail to form the tetramer. The alignments of the various kinesin-related proteins stalk domains related to XlEg5 did not reveal any conserved phosphorylation domain that might be the target of pEg2, thereby suggesting that the recognition motif of the substrate depends largely upon its secondary and tertiary structure.
We also showed that, like HsEg5, XlEg5 was phosphorylated by p34 cdc2 in vivo. The p34 cdc2 phosphorylation site located in the tail domain of XlEg5 was shown to be preserved within the bimC family of kinesin-related proteins (20). Although the conservation of this site throughout the bimC family is real in terms of sequence, it is not necessarily a p34 cdc2 target. In Cut7 from Schizosaccharomyces pombe (36,54) the mutation of the threonine 1011 (the potential p34 cdc2 phosphorylation site) to an alanine remained ineffectual on spindle assembly (55). Conversely, mutation of the threonine 927 in HsEg5 eliminated spindle association (19) and any binding to the dynactin subunit p150 (20). Mutation of threonine 937 in XlEg5 also disrupted mitotic spindle association (18).
Potential orthologues of XlEg5 were also found in fission yeast (36,54), and in Drosophila (39,41,52,53). Candidate proteins that could be orthologues of pEg2 were pinpointed in rat (56), in mouse (57)(58)(59)(60), in Drosophila (26), and in human (28 -30, 60 -62). It was obvious that both the Aurora-related kinases and the bimC kinesin-related proteins were conserved throughout evolution. Accordingly, we showed that pEg2 kinase associated with and phosphorylated the kinesin-related protein XlEg5, which suggests that the Aurora-related kinase family might be involved in the regulation of kinesin-related protein activities. To our knowledge, XlEg5 proved to be the first reported physiological substrate for an Aurora-related kinase.