Aurora B is required for mitotic chromatin-induced phosphorylation of Op18/Stathmin

Abstract

Oncoprotein 18/Stathmin (Op18) is a microtubule-destabilizing protein that is inhibited by phosphorylation in response to many types of signals. During mitosis, phosphorylation of Op18 by cdc2 is necessary but not sufficient for Op18 inhibition. The presence of mitotic chromosomes is additionally required and involves phosphorylation of Ser-16 in Xenopus Op18 (and/or Ser-63 in human). Given that Ser-16 is an excellent Aurora A (Aur-A) kinase consensus phosphorylation site and the Aurora kinase inhibitor ZM447439 (ZM) blocks phosphorylation in the activation loop of Aur-A, we asked whether either Aur-A or Aurora B (Aur-B) might regulate Op18. We find that ZM blocks the ability of mitotic chromatin to induce Op18 hyperphosphorylation in Xenopus egg extracts. Depletion of Aur-B, but not Aur-A, blocks hyperphosphorylation of Op18, and chromatin assembled in the absence of Aur-B fails to induce hyperphosphorylation. These results suggest that Aur-B, which concentrates at centromeres of metaphase chromosomes, contributes to localized regulation of Op18 during the process of spindle assembly.

Both centrosomes and mitotic chromosomes play important roles in organizing microtubules into the bipolar mitotic spindle (reviewed in refs. 1 and 2). In most cells, centrosomes appear to play the dominant role, as microtubules emanating from centrosomes are captured at their plus ends by chromosomes. In large cells or cells without centrosomes, chromosomes provide the dominant spindle generating force: microtubules are first nucleated around chromosomes and then become organized into a bipolar spindle. Experimentally, the addition of centrosome-free nuclei or chromatin-coated beads to Xenopus eggs or egg extracts results in the nucleation and stabilization of microtubules that then become ordered into bipolar arrays (3–5). In both centrosome-containing and centrosome-free cells, formation of bipolar spindles in which sister chromatids become bioriented and under equal tension from opposite poles is due to a balance of (i) microtubule stabilizing activities of certain microtubule-associated proteins, (ii) sets of motor proteins that cross-link microtubules and bundle them into focused poles, and (iii) the localized inhibition of plus-end microtubule depolymerases, such as MCAK/XKCM1 (MCAK), which allows polar microtubules to make stable connections with kinetochores, and Oncoprotein 18/Stathmin (Op18), which has both tubulin sequestering and microtubule depolymerizing activities (reviewed in refs. 1, 6, and 7). The kinase Aurora B (Aur-B) is required for the localized inhibition of MCAK activity in the vicinity of chromosomes (see below). Here we have asked whether Aur-B, or the related kinase Aur-A, regulates Op18 as well.

Op18 connects numerous signaling pathways to changes in microtubule dynamics. It was originally identified as a small protein that is phosphorylated on up to four serine residues in response to extracellular signals and is highly expressed in many cancers (reviewed in ref. 8). It was independently found as a factor that promotes the catastrophic shrinkage of microtubules at their plus ends (9) whose activity can be suppressed during mitosis by phosphorylation (10). It is now known that phosphorylation of Op18 on Ser-16, Ser-25, and Ser-39 (Xenopus numbering) both reduces Op18’s capacity to sequester tubulin and its ability to depolymerize microtubules (reviewed in refs. 6, 11, and 12). Expression of nonphosphorylatable mutants results in the accumulation of mitotic cells with short spindles and misaligned chromosomes (reviewed in ref. 13). During mitosis, phosphorylation of one of two sites (Ser-25 or Ser-39) by cytoplasmic cdc2 or mitogen-activated protein kinase (MAPK) is required but not sufficient for inhibition of Op18. An additional phosphorylation on Ser-16 (and/or Ser-63 in human Op18) is required to inhibit Op18. This “hyperphosphorylation” of Op18 is induced by the presence of mitotic chromatin (10, 14, 15) and appears to be required for the localized inhibition of Op18 and stabilization of microtubules near mitotic chromosomes (15–18). Thus, either chromatin-associated factors and/or chromatin-generated kinase activities are required for both hyperphosphorylation and localized inhibition of Op18 during mitosis. The polo-like kinase Plx1 is required for inhibition of Op18 in egg extracts (18) but its role may be indirect, because neither Ser-16, Ser-25, nor Ser-39 fit known polo consensus phosphorylation sites (19). However, Ser-16 does form a good match with Aurora kinase consensus sites (20, 21).

Aur-B and its yeast counterpart Ipl1 are essential for processes that resolve inappropriate kinetochore-microtubule attachments and lead to correct bipolar alignment of sister chromatids on the metaphase spindle (reviewed in refs. 22–24). Aur-B is part of a multisubunit “chromosomal passenger” complex that associates with chromosomes during early mitosis and then, in somatic cells, concentrates at inner centromeres. Inhibition of Aur-B interferes with the localization of several proteins at centromeres and kinetochores, including MCAK (reviewed in ref. 25). In vitro, Aur-B phosphorylates and inhibits MCAK. Localized phosphorylation of MCAK by centromeric Aur-B is thought to block MCAK’s microtubule plus-end depolymerizing activity in the vicinity of kinetochores, suggesting that cycles of MCAK phosphorylation and dephosphorylation are important for achieving microtubule stabilization and attachment to kinetochores that accompany the correct bipolar orientation of sister chromatids (26–31).

Aur-A is required for the establishment and/or maintenance of spindle bipolarity in many cells (reviewed in refs. 22, 32). During mitosis, Aur-A is concentrated at centrosomes and nearby spindle microtubules. Phosphorylation of Aur-A within the activation loop (Thr-295 Xenopus, Thr-288 human) is required for its activity (33, 34). Signals originating from the mitotic chromosomes themselves strongly increase the activity of Aur-A. During mitosis, chromatin-associated guanine nucleotide exchange factor RCC1 generates an increase in the local concentration of Ran-GTP (35–38), which displaces the spindle assembly factor TPX2 from inhibitory association with importins (39, 40). TPX2 interaction with Aur-A strongly enhances its kinase activity (41, 42). Thus, it is possible that a small population of active Aur-A exists near mitotic chromosomes, where it could participate in the regulation of interactions of microtubules with chromosomes.

Using Xenopus egg extracts, we find that the Aurora kinase inhibitor ZM447439 (ZM) blocks the chromatin-induced hyperphosphorylation of Op18 previously associated with localized inhibition of Op18 in the vicinity of chromosomes. Although ZM has been regarded as a highly selective inhibitor of Aur-B in cells, we find that ZM also blocks phosphorylation of the Aur-A activation loop residue Thr-295. Immunodepletions show that Aur-A, which is required for early centrosome-induced microtubule nucleation, is not essential for either the subsequent formation of bipolar spindles organized around mitotic chromatin or Op18 hyperphosphorylation. Depletion of Aur-B blocks both spindle assembly and the ability of mitotic chromatin to induce Op18 hyperphosphorylation. These findings argue that, in addition to its role in regulating MCAK activity, Aur-B is required for the localized inhibition of Op18 during spindle assembly.

Results

Like ZM, Depletion of Aur-B Permits Aster Formation but Blocks Subsequent Chromosomally Organized Spindle Formation: Depletion of Aur-A Interferes with Aster Formation but Not Subsequent Spindle Formation.

Extract was prepared from Xenopus eggs, which are naturally arrested in metaphase of meiosis II (“CSF extract”), and immunodepleted by using Aur-A, Aur-B, or control antibodies. A separate portion of the extract was treated with ZM. Demembranated sperm nuclei and rhodamine-labeled tubulin were added to follow changes in chromatin and spindle structure, respectively. Samples were taken at 10 min, when centrosome-nucleated microtubule aster formation is occurring, and at 45 min, when chromosome-mediated microtubule nucleation and stabilization predominates. Fig. 1A shows that the depletions of Aur-A and Aur-B were both effective and highly specific (lanes 4 and 6). Purified recombinant his-tagged Aur-A and Aur-B proteins (lanes 1 and 2) served as controls for antibody specificity, and the phospho-MAPK signals served as extract loading controls (lanes 3–7).

Fig. 1.

Depletion of Aur-B or ZM treatment, but not Aur-A, prevents mitotic chromatin-induced Op18 hyperphosphorylation. (A) Xenopus egg (CSF) extracts were subjected to two rounds of immunodepletion using Aur-A preserum (Δ Mock Pre-serum), Aur-A serum (Δ Aurora A), IgG (Δ Mock IgG), or anti-Aur-B (Δ Aurora B) antibodies. A portion of undepleted extract was incubated on ice with 20 μM ZM. Samples were taken and analyzed by SDS/PAGE followed by immunoblotting with antibodies directed against Aur-B, phospho-Ser-10 of Histone H3, Aur-A, phospho-Thr-295 of Aur-A, and phospho-Thr-202/Tyr-204 of MAPK. Recombinant histidine-tagged Aur-A and B were run on lanes 1 and 2. (B) Extracts were depleted as in A or treated with ZM. Depleted and ZM-treated extracts were supplemented with sperm nuclei and rhodamine-labeled tubulin on ice. The extracts were warmed to 21°C, samples were taken at 10 and 45 min, and microtubule assembly and chromatin morphology were analyzed by fluorescence microscopy. The images shown represent the most abundant structures observed. (Scale bar, 15 μm.)

By 10 min, microtubule asters emanating from compact sperm nuclei were seen in the control extracts (Fig. 1B). After Aur-A depletion, >50% of nuclei failed to form centrosomal asters, and the remaining ones formed asters with much smaller microtubule arrays. Despite this early defect, by 45 min the Aur-A-depleted extract had assembled microtubules into bipolar spindle-type arrays associated with chromosomes. However, those spindles were frequently irregular and often had broader poles. Thus, in this system, Aur-A is required for two temporally distinct aspects of spindle formation: initial aster formation and subsequent focusing of the spindle poles. However, Aur-A does not appear to be required for the formation of microtubule arrays surrounding the mitotic chromosomes.

Aur-B depletion did not block formation of early centrosome-associated microtubule arrays at 10 min, but completely inhibited the later chromatin-driven development of spindles (Fig. 1B). These effects are very similar to those of ZM (Fig. 1B) as described in ref. 43, and unlike those of removing Aur-A. These findings support the idea that ZM selectively targets Aur-B and does not affect Aur-A. However, as shown below, that conclusion is probably incorrect.

ZM Blocks Phosphorylation of Aur-A at the Activation Loop Residue Thr-295.

When tested on purified kinases, ZM strongly inhibits both Aur-A and Aur-B, but its effects on cells have appeared to be the result on inhibiting only Aur-B (43, 44). Because ZM is a reversible inhibitor that is lost during washes of immunoprecipitates, it was not possible to directly test for its inhibition of Aur-A’s kinase activity in extracts. However, phosphorylation of Aur-A on Thr-295 (Thr-288 human) in the activation loop is required for high kinase activity (33, 34). In both mammalian somatic cells (45) and Xenopus egg extracts (46), ZM blocks phosphorylation of this residue (Fig. 1A, lane 7). It leads also to a slight downshift in Aur-A’s electrophoretic mobility (lane 7), which generally correlates with the inactive form (33, 47, 48). This result suggests that ZM could also inhibit Aur-A in cell extracts and cells.

If ZM does indeed inhibit Aur-A and depletion of Aur-A blocks early aster formation, why doesn’t ZM block aster formation? One possibility is that Aur-A’s role in centrosome-mediated aster formation depends on Aur-A protein itself rather than its kinase activity. This seems unlikely because beads coated with wild-type Aur-A (but not inactive Aur-A) can function as microtubule-organizing centers (49). Another, as recently proposed (45), is that a requirement for Aur-A may be bypassed when Aur-B kinase activity is inhibited. Further work will be required to explain these observations.

Inhibition of MCAK Partially Rescues Spindle Assembly in ZM-Treated Extracts.

When somatic cells are treated with ZM, chromosome alignment is defective, but bipolar spindle formation does occur and at least some microtubule-kinetochore attachments are seen (44). In Xenopus egg extracts, ZM completely blocks spindle formation (43). This difference is probably because of the fact that, in eggs, mitotic chromosomes play the major role in organizing spindles (1, 2). Localized phosphorylation of MCAK by Aur-B inhibits MCAK’s microtubule-depolymerizing activity. Moreover, immunodepletion of Aur-B-associated chromosomal passenger complex proteins blocks spindle formation and codepletion of MCAK restores microtubule assembly (26–30). We were curious to know whether simple inhibition of Aurora kinase activity by ZM, rather than depletion of the entire complex, could also be overcome solely by inhibiting MCAK. To do this, we made use of an MCAK inhibitory antibody (29). CSF extract containing sperm nuclei and rhodamine-labeled tubulin was treated with calcium to induce exit into interphase of the first mitotic cell cycle. Sixty minutes later, DMSO or ZM was added and the incubation was continued for 20 min. A half volume of CSF extract preincubated with DMSO or ZM, respectively, was then added to drive the extract into M phase and metaphase arrest. Control or inhibitory MCAK antibody was added 60 min after M phase induction, and chromosomes and spindles were visualized 20 min later. As shown previously (29, 50, 51), the MCAK antibody prevented formation of a bipolar spindle and, instead, resulted in the formation of large microtubule asters (Fig. 2A). Strikingly, when ZM-treated extracts (which lack microtubules) were supplemented with MCAK inhibitory antibody, microtubule polymerization around chromatin was restored; moreover, in ≈20% of these cases, the spindles were obviously bipolar (Fig. 2B). These results provide further support to the idea (30) that among all of the potential substrates of Aurora kinase activity that are required for spindle formation, MCAK plays a major role.

Fig. 2.

Inhibition of MCAK in ZM-treated extracts rescues microtubule polymerization around chromatin. (A) CSF extract was supplemented with rhodamine-labeled tubulin and sperm nuclei. Extracts were driven into interphase by the addition of calcium. After 60 min, DMSO (control) or 20 μM ZM was added. After 20 min, control and ZM-treated extracts were driven into mitosis by the addition of CSF extract containing DMSO or 20 μM ZM, respectively. Sixty minutes later, 300 μg/ml anti-IgG or anti-MCAK antibodies were added to the extracts. Microtubules (red) were visualized with rhodamine-labeled tubulin and DNA (blue) with Hoechst dye 33042. (Magnification, ×40; scale bar, 20 μm.) The images represent the most abundant structures observed. (B) Spindle structures were counted after 20 min of antibody addition. “Bipolar spindles” contained microtubules organized into bipolar structures; “DNA with microtubules” contained microtubules that were not organized into obvious bipolar spindles; “No spindle” showed no microtubule staining. Greater than 100 structures were counted. The graph shown represents one of three independent experiments.

However, the bipolar spindles that formed when MCAK was inhibited in ZM-treated extracts were about half the size of the controls (Fig. 2A). This result seemed especially interesting because overexpression of Op18 or addition of nonphosphorylatable Op18 mutants also results in smaller spindles (reviewed in ref. 13). Thus, it seemed possible that Op18’s microtubule-depolymerizing activity might also be inhibited by Aurora kinase activity.

ZM Blocks Mitotic Chromatin-Induced Op18 Hyperphosphorylation.

Op18 Ser-16 resides within the sequence Lys-Arg-Ala-Ser-16-Gly, a strong consensus sequence for both Ipl1 and Aur-A (20, 21). Therefore, we examined ZM’s effects on Op18 phosphorylation, which is readily monitored by shifts in electrophoretic mobility (16). By itself, recombinant his-tagged Op18 ran as a single broad band (Fig. 3A, band 1). In the absence of chromatin, incubation of Op18 with CSF extract converted it into two more slowly migrating forms (Fig. 3A, bands 2 and 3). As reported (18), these bands were not seen after incubation with interphase extract (not shown), indicating that the shifts are specific for M phase. ZM almost completely inhibited phosphorylation of histone H3 on Ser-10, a target of Aur-B, but had no effect on the level of cdc2 or MAPK activity (Fig. 3A), showing that the extract remained in M phase. Notably, ZM did not block the two M phase-specific shifts in Op18 (Fig. 3A, lane 3), indicating that Aurora kinase activity is not responsible for the chromatin-independent shifts in Op18 during mitosis.

Fig. 3.

ZM inhibits mitotic chromatin-induced Op18 phosphorylation. (A) Control or ZM-treated CSF extract lacking or containing nuclei (10,000 per μl) were incubated at 20°C for 60 min. Recombinant His-tagged wild type or Op18-AAA (12 μM) was added and the reaction continued for 20 min. Samples were processed for histone H1 kinase assays or immunoblot analysis using antibodies against the histidine-tag on Op18, phospho-MAPK, and phospho-Ser-10 on Histone H3. (B) Chromatin was assembled in CSF extract in the presence of 1% DMSO, 20 μM ZM, or 10 μg/ml nocodazole for 60 min at 21°C. (C) Chromatin was assembled in CSF extract in the presence of 1% DMSO or 20 μM ZM for 60 min at 21°C. IgG or MCAK antibodies (300 μg/ml) were added, followed by a 10-min incubation. (D) Control- or ZM-treated CSF extracts containing nuclei were incubated at 20°C for 60 min. Chromatin was pelleted and resuspended in kinase buffer containing either DMSO or ZM. His-tagged wild-type or Op18AAA (12 μM) and [γ-³²P]ATP were added and incubated at 30°C for 20 min. Samples were analyzed by SDS/PAGE followed by autoradiography (Upper). Coomassie-stained Op18 served as a loading control (Lower).

To ask whether ZM affected chromatin-induced phosphorylation of Op18, nuclei were added to control or ZM-treated CSF extracts and incubated for 60 min to allow formation of metaphase chromatin, and wild-type Op18 was added. The presence of mitotic chromatin induced the appearance of an additional “hyperphosphorylated” form (Fig. 3A, band 4). Only a small fraction of total Op18 underwent this shift, presumably reflecting localized inhibition of Op18 in the vicinity of chromosomes (16, 17). ZM almost completely blocked the appearance of band 4 (lane 5). As seen earlier (43), metaphase chromosomes assembled in ZM-treated extracts resembled those in controls (data not shown), arguing that the lack of chromatin-induced Op18 hyperphophorylation was not simply due a gross defect in chromatin assembly (data not shown). As shown earlier (18), the mobility of the triple Ser 16, 25, 39 Ala mutant AAA-Op18 did not shift after incubation with CSF extract, even in the presence of mitotic chromatin (Fig. 3A Right). Thus, it appears that Aurora kinase activity is required for the ability of chromatin to induce Op18 hyperphosphorylation.

Mitotic Chromatin Itself, Rather than Microtubules, Is Responsible for Inducing the Majority of Op18 Hyperphosphorylation.

Op18’s microtubule depolymerizing activity can also be negatively regulated by polymerized microtubules. Moreover, microtubule-dependent Op18 phosphorylating activity cosediments with microtubules, suggesting that chromatin-induced microtubules could be responsible for some or all of the inhibitory phosphorylation on Op18 rather than chromatin itself (52). Because ZM completely blocks spindle formation in egg extract, it seemed possible that the microtubules themselves, rather than chromatin, induced Op18 hyperphosphorylation.

We approached this question in two ways. First, chromatin was assembled in CSF extracts in the presence of DMSO (control), ZM, or nocodazole, and Op18 was added (Fig. 3B). The intensities of the phosphorylated histone H3 and MAPK bands serve as ZM specificity and extract loading controls, respectively. As above, ZM did not interfere with the appearance of the mitotic-specific bands 2 and 3 but did block appearance of band 4. Addition of nocodazole, which completely inhibited spindle assembly (data not shown), only slightly reduced the appearance of band 4. This result suggests that, at least in this system, microtubules do not appear to play a large role in driving Op18 hyperphosphorylation. Second, we took advantage of our finding (Fig. 2) that inhibition of MCAK can restore a large amount of microtubule polymerization to extracts treated with ZM. Chromatin was assembled in the presence or absence of ZM. To rescue the spindle assembly defect caused by ZM, control or inhibitory MCAK antibodies were added, followed by addition of Op18 (Fig. 3C). Neither antibody blocked Op18 hyperphosphorylation in the control extract (lanes 1 and 3) or stimulated Op18 hyperphosphorylation in the ZM-treated extract (lanes 2 and 4). These results argue that mitotic chromatin itself, rather than the spindle microtubules organized by mitotic chromatin, is responsible for most Op18 hyperphosphorylation.

ZM Inhibits a Chromatin-Associated Kinase Activity Required for Op18 Hyperphosphorylation.

Op18 could be phosphorylated by a ZM-sensitive kinase that is associated with mitotic chromosomes, such as Aur-B, or by a ZM-sensitive kinase that is activated by a factor generated by mitotic chromatin, such as Aur-A. To address this, metaphase chromatin was first isolated from control or ZM-treated extracts, and then incubated with recombinant Op18 protein in the presence of [γ-³²P]ATP. Chromatin isolated from control extract induced strong ³²P incorporation into Op18 (Fig. 3D, lane 1). In contrast, exposure to chromatin that had been isolated from the ZM-treated extract induced much less phosphorylation of Op18 (lane 2). Addition of ZM to chromatin after it had been isolated from control extracts also strongly reduced the ability of mitotic chromatin to stimulate ³²P incorporation into Op18 (lane 3). Finally, chromatin-induced incorporation of ³²P into the AAA-Op18 mutant was absent, suggesting that those three residues are the major phosphorylation sites targeted by the metaphase chromatin-associated kinase activities (lane 4). These results argue that ZM directly or indirectly inhibits one or more chromatin-associated kinase activities that are required to repress Op18 activity near chromatin. Chromosome-associated Aur-B was an obvious candidate.

Depletion of Aur-B Blocks Chromatin-Induced Hyperphosphorylation of Op18, Depletion of Aur-A Does Not.

Aur-A and Aur-B each phosphorylate recombinant Op18, but not AAA-Op18 in vitro (data not shown). To directly test whether either is required for Op18 hyperphosphorylation, CSF extracts were immunodepleted of Aur-A or Aur-B. Nuclei were added and chromatin-induced changes in Op18 mobility were assayed. As before, the addition of ZM to control extract completely blocked the appearance of hyperphosphorylated band 4 (Fig. 4, lane 7). Depletion of Aur-A (lane 4) had no detectable effect on band 4, whereas depletion of Aur-B (lane 6) completely blocked the appearance of band 4. This result argues that chromatin-associated Aur-B is required for the vast majority of the Op18 hyperphosphorylation that is induced by mitotic chromatin.

Fig. 4.

Depletion of Aur-B prevents chromosome-induced Op18 hyperphosphorylation. CSF extracts were immunodepleted by using Aur-A preserum (Δ Mock Pre-serum), Aur-A serum (Δ Aurora A), IgG (Δ Mock IgG), or Aur-B (Δ Aurora B) antibodies. A portion of undepleted extract was incubated on ice with ZM. Depleted and ZM-treated extracts were incubated for 60 min with 10,000 nuclei per μl at 21°C. His-tagged Op18 (12 μM) was added to the extract, followed by a 30-min incubation. Samples were immunoblotted as indicated. Analysis of Op18 samples from a separate experiment is shown at Bottom.

ZM’s Inhibition of Aur-A Thr-295 Phosphorylation Is Independent of Its Effects on Aur-B.

As seen earlier (Fig. 1), ZM blocked phosphorylation of Aur-A on Thr-295. ZM also affected the gel mobility of Aur-A, reducing it mainly to a single band that usually correlates with the inactive form (Fig. 4, lane 7). Although Thr-295 is a site of Aur-A autophosphorylation, it can be phosphorylated by other kinases (53, 54). Thus, we asked whether depletion of Aur-B from egg extract reduced or blocked Thr-295 phosphorylation. Removal of Aur-B had no obvious effect on Aur-A Thr-295 phosphorylation (see Fig. 1A, lane 6) or gel mobility (Fig. 4, lane 6), indicating that ZM does not prevent phosphorylation of Aur-A on Thr-295 by inhibiting Aur-B. These considerations strongly argue that ZM inhibits Aur-A directly by interfering Aur-A’s autophosphorylation-dependent activation pathway, or by inhibiting an upstream, ZM-sensitive kinase yet to be identified.

Discussion

This work shows that (i) Aur-B, which is highly concentrated at the centromeric regions of metaphase chromosomes, is required for the ability of mitotic chromatin to induce the hyperphosphorylation of Op18, and (ii) the presence of Aur-B on mitotic chromatin is required for Op18 hyperphosphorylation. Because Ser-16 is an excellent Aurora kinase consensus site (20, 21), and phosphorylation of Ser-16 is required for the inhibition of Op18’s microtubule-depolymerizing activity, these findings argue that chromatin-associated Aur-B is required for inhibiting Op18 during mitosis. They also suggest a model in which a subset of Op18 that has already been phosphorylated by cdc2 on Ser-25 and/or Ser-39 (human Ser-38) is locally inhibited through phosphorylation by centromere-associated Aur-B, thus participating in the generation of stable microtubule-kinetochore attachments during spindle formation. Moreover, after cdc2 inactivation at anaphase onset, the inhibitory effect of Ser-16 phosphorylation might be expected to persist until mitotic exit owing to phosphorylation of Ser-25 and/or Ser-39 by mitotic MAPK activity (55).

Both MAPK and other cdc2-related kinases have been demonstrated to phosphorylate Ser-25 and Ser-39 in a many pathways activated by extracellular signals (56). During mitosis, Ser-39 appears to be the major site phosphorylated by cdc2, in both human somatic cells and Xenopus eggs (16, 57). Although we have not investigated exact requirements for cdc2, MAPK, or Aurora kinases in phosphorylation of the specific, individual sites, those previous studies suggest that cdc2 is probably responsible for the nearly complete conversion of unphosphorylated recombinant Op18 (band 1) to the phosphorylated band 2 after incubation with mitotic cytoplasm. A portion of Op18 is modified by mitotic cytoplasm to form band 3, perhaps as a result of further phosphorylation of Ser-25 by MAPK. Because mitotic chromatin induces an additional hyperphosphorylation of Op18 and Ser-16 is the only other residue of Xenopus Op18 that is phosphorylated during mitosis, it seems very likely Ser-16 is phosphorylated by Aur-B associated with mitotic chromosomes. Each of these predictions can now be tested.

Mitotic chromatin that is assembled in Xenopus egg extracts depleted of Plx1 significantly reduces the overall incorporation of [γ-³²P]ATP incorporation into Op18; moreover, Plx1 depletion causes defects in microtubule polymerization and spindle assembly (18). These results implicate Plx1 in the regulation of Op18, although they do not distinguish between direct and indirect roles. Given known Plx1 (Plk1) phosphorylation consensus motifs (19), Plx1 is not expected to phosphorylate either of the two cdc2/MAPK consensus sites Ser-25 and Ser-39. Plx1 could phosphorylate Ser-16 directly, but the match with known sites of phosphorylation by polo-like kinases is not strong. In view of the facts that (i) the Ser-16-containing sequence matches known Aurora kinase consensus sequences, (ii) the Aurora kinase inhibitor ZM blocks the ability of mitotic chromatin to induce Op18 hyperphosphorylation, and (iii) mitotic chromatin assembled in the absence of Aur-B does not induce Op18 hyperphosphorylation, Aur-B could be directly responsible for phosphorylation of Ser-16. Perhaps the requirement for Plx1 relates to its association with INCENP (58), an activator of Aur-B (reviewed in ref. 32).

ZM inhibits the phosphorylation of Aur-A on the activation loop Thr-295 that is essential for its activity, as shown here and elsewhere (45, 46). The results of immunodepletion experiments strongly argue that Aur-A is not required for the chromatin-induced hyperphosphorylation of Op18. However, given that a portion of Op18 has been detected at spindle poles (52) and in vitro Op18 can also depolymerize microtubules at their minus ends (59), the possibility that Aur-A might regulate Op18 in the vicinity of centrosomes is worth considering. Finally, the Ser-16-containing Aur-A consensus sequence it is highly conserved in five other Op18 family members, where it is phosphorylated in response to numerous extracellular signals (56). This might be of special concern in the many types of cancer in which Aurora kinases are overexpressed (60).

Methods

Egg Extracts.

CSF extracts were prepared from metaphase II-arrested Xenopus eggs and supplemented with demembranated sperm nuclei (600 nuclei per μl) and 50 ng/μl rhodamine-labeled tubulin (Cytoskeleton, Denver; TL331M-A) as described (43). Extracts were warmed to 21°C, and metaphase exit was induced by addition of 0.5 mM calcium. Eighty minutes later, extracts were driven into M phase by the addition of a half volume of CSF extract. Samples were analyzed by microscopy as described (43). ZM447439 (AstraZeneca) was made to a stock of 10 mM in dimethyl sulfoxide (DMSO), stored at −20°C, and added to extracts (20 μM) as indicated. Cdc2 activity was assayed by phosphorylation of histone H1 (43). Chromatin was prepared from CSF extract as described (61), suspended in 10 mM Hepes-KOH, pH 7.7/50 mM sucrose/2 mM MgCl₂/0.1 mM CaCl₂/2 mM EGTA-KOH/50 μM ATP/100 mM KCl/2 μCi of [γ-³²P]ATP (1 Ci = 37 GBq), and incubated with 12 μM His-Op18 at 30°C for 20 min.

Antibodies.

Samples were analyzed by SDS/PAGE followed by immunoblotting (43). Blots were incubated with antibodies for Aurora A (1:2,000 dilution; ref. 62), phospho-Aurora A T295 (1:1,000; Cell Signaling Technology, Beverly, MA), Aurora-B (1:2,000; ref. 29), phospho-histone H3 (S10) (1:2,000; Upstate Biotechnology catalog no. 06-570), phospho-MAPK (1:5,000; Cell Signaling Technology Catalog no. 9101S), or polyHistidine (1:3,000; Clone HIS-1, Sigma H 1029). For depletions, 8 μg of affinity purified anti-Aur-B or control IgG was incubated with 30 μl of protein A-Dynabeads for 1 h in 600 μl of PBS. Antibody-coupled beads were washed twice in TBS and twice in sperm dilution buffer (1 mM MgCl₂/100 mM KCl/150 mM sucrose/100 μg/ml cytochalasin B). Beads were mixed with 30 μl of CSF extract and rotated at 4°C for 80 min, and then removed by using a magnetic particle separator. For Aur-A, serum or control preserum was incubated with 30 μl of protein A-Dynabeads for 1 h in 600 μl of PBS.

Recombinant Op18, Aur-A, and Aur-B Proteins.

Xenopus Op18 and the triple mutant Ala-16, Ala-25, Ala-39 cDNAs, a gift of Rebecca Heald (University of California, Berkeley), were expressed as described (18). His-tagged Aur-A was prepared as described (33). Aur-A was purified as described (33). Recombinant Aurora B was a gift of Ryoma Ohi (29) (Harvard Medical School).

Abbreviations

MCAK

MCAK/XKCM1

Aur

Aurora

ZM

ZM447439.