Phosphorylation of Maskin by Aurora-A Is Regulated by RanGTP and Importin β*

Mitotic spindle assembly in Xenopus egg extracts is regulated at least in part by importin β and its regulator, the small GTPase, Ran. RanGTP stabilizes microtubules near the chromosomes during spindle assembly by selectively releasing spindle assembly factors from inhibition by importin α/β in the vicinity of the chromosomes. Several spindle assembly factors are regulated in this manner. We identified maskin, the Xenopus member of the transforming acidic coiled coil family of proteins, as a potential candidate in a two-step affinity chromatography approach designed to uncover additional downstream targets of importin α/β in mitosis. Here, we show that although maskin lacks a canonical nuclear localization sequence, it binds importin β in a RanGTP-regulated manner. We further show that importin β inhibits the regulatory phosphorylation of maskin by Aurora-A. This suggests a novel mechanism by which importin β regulates the activity of a spindle assembly factor.

XMAP215/TOG protein family to the centrosome (32,33). Recombinant Xenopus maskin TACC domain alone is sufficient to rescue centrosome defects brought about by depleting maskin from egg extracts (27,28), suggesting that the TACC domain plays an important role in centrosome function. The localization and activity of maskin are regulated by Aurora-A (28,29,34). Aurora-A is an essential protein kinase required for mitosis and centrosome maturation, a process in which centrosomes acquire additional MT nucleating material at the onset of mitosis (35)(36)(37). Phosphorylation of maskin by Aurora-A on three residues (Ser 33 , Ser 620 , and Ser 626 ) is required to localize maskin to the centrosome and activate its function (Refs. 28 and 29, see also Ref. 38). The Caenorhabditis elegans and Drosophila Aurora-A homologs are similarly required to target TACC proteins to centrosomes and activate them (38 -41), suggesting that the regulation of TACC protein function by Aurora-A kinase is evolutionarily conserved.
Here, we report isolating maskin in an assay designed to identify MT-binding proteins that are regulated by importin ␤. We show that maskin interacts with importin ␤ in vitro, and this interaction is regulated by RanGTP. The direct interaction between maskin and importin ␤ was unexpected because maskin lacks a canonical NLS. Using truncation mutants to map the importin ␤ binding domain of maskin, we found that the central portion of maskin is important for importin ␤ binding, but efficient binding requires the full-length protein. Most importantly, we show that importin ␤ binding inhibits the phosphorylation of maskin by Aurora-A. These findings suggest a role for importin ␤ in regulating maskin activity at the centrosome.
Maskin truncations were generated using the following primer pairs (the cloning vector for each construct is indicated in parentheses, restriction sites introduced by the primer are underlined and are identified by name following the sequence, and stop codons are indicated in bold): 1-774 (pGEX4-T2rTEV; see Ref. 27): 5Ј-CGTTGAAATCGACTATCTAGAG (XbaI), 3Ј-GTGAG-CTCCTAGCCTTCAAACTCTGC (SacI); the fragment amplified by these primers was used to replace the XbaI/SacI fragment of full-length maskin (27) 27; 1-363 was generated by digesting full-length maskin (27) with AgeI, blunt-ending, and digesting with BamHI.
Antibodies-Antibodies against recombinant full-length maskin were described in Ref. 27 and for immunofluorescence were directly labeled with Oregon Green 488 maleimide according to the manufacturer's instructions (Molecular Probes, Eugene, OR). TPX2 antibodies were described in Ref. 43. Importin ␣ antibodies were a kind gift from M. Dasso (National Institutes of Health). GST antibodies were a kind gift from S. Bednarek (University of Wisconsin, Madison, WI). Control preimmune IgG was from the serum of rabbits before inoculation with maskin. Importin ␤ antibodies were purchased from Transduction Laboratories (Lexington, KY). Secondary antibodies used for Western blotting were purchased from Sigma.
Xenopus Egg Extract Preparation and Light Microscopy-Cytostatic factor-arrested (44,45) or interphase (46) extracts were prepared from Xenopus eggs as described. Maskin (0 -40 M) and/or importin ␤ (as indicated in Fig. 4) were added to mitotic extract on ice; the reaction was then moved to 22-25°C and incubated for 15 min. The volume of protein solution added never exceeded 20% of total extract volume. Images were taken with a Photometrics CoolSnap HQ cooled CCD camera (Roper Scientific, Inc.) through a ϫ60/1.4 NA plan apo objective mounted on a Nikon Eclipse E800 fluorescence microscope. Images were obtained using MetaMorph software and processed using Adobe Photoshop.
Importin ␤ Affinity Chromatography-Mitotic MAPs were incubated with XB buffer and importin ␣, importin ␤, or both importin ␣ plus importin ␤ for 3 h at 4°C. To retrieve importin ␤, S-protein beads were added to the mixture, which was incubated for an additional 60 min. The beads were collected by a brief spin, and washed three times with XB and once with BRB80. Bound proteins were eluted by boiling the beads in SDS sample buffer. Eluted proteins were separated by SDS-PAGE and visualized by Coomassie staining or Western blotting.
Pull-out Assays-Purified recombinant proteins were incubated together or with mitotic egg extract (as indicated in Figs. 2 and 5) for 30 min at 4°C. Glutathione-agarose was added to retrieve GST-maskin, or S-protein-agarose was added to retrieve importin ␤. For the experiment shown in Fig. 5B, GSTtagged maskin truncation mutants were first covalently linked to Dynabeads M-270 Carboxylic Acid (Dynal Biotech, Oslo, Norway) according to the manufacturer's instructions, and then incubated in mitotic egg extracts. The beads and associated proteins were collected by a brief spin and washed 3 times with H100 (50 mM HEPES, 1 mM EGTA, 1 mM MgCl 2 , 100 mM NaCl, pH 7.6) and once with the same buffer containing 250 mM NaCl and 0.1% Triton X-100. Proteins were eluted by boiling in SDS sample buffer and analyzed by SDS-PAGE. For Western blotting, proteins were transferred to nitrocellulose membranes. The equilibrium dissociation constant was calculated using GraphPad Prism Software.
Immunoprecipitations-To immunoprecipitate maskin, antimaskin antibodies were conjugated to Affi-prep protein A beads (Bio-Rad) as described (48). Antibody-coupled beads (50 l) were added to the extract and incubated for 1 h at 4°C. Beads were collected by brief centrifugation, the immunoprecipitates were washed with H100, and analyzed by SDS-PAGE and Coomassie staining, or were transferred to nitrocellulose for Western blotting, as indicated in Fig. 3.
Kinase Assays-Kinase assays were performed as described (28,29), with the following modifications: maskin (0.6 M), importin ␤ (3-12 M), importin ␣ (0.6 M), and RanL43E (24 M) were incubated in kinase buffer for 30 min at 4°C. Aurora-A (0.06 M) and [␥-32 P]ATP (8 mM; Redivue, 3,000 Ci/mmol; obtained from Amersham Biosciences) were added and the mixture was incubated at 22-25°C for 15 min. Reactions were stopped by addition of SDS sample buffer and separated by SDS-PAGE. Incorporation of label was detected by autoradiography using a PhosphorImager and was quantitated using Adobe Photoshop software.

RESULTS
The Xenopus Spindle Assembly Factor Maskin Is a Potential Importin ␤ Target-Several proteins that function to regulate mitotic spindle assembly or MT bundling have been shown to interact with importin ␤ in mitosis (7,49). To identify additional potential targets for regulation by importin ␤, we reasoned that at least a subset of proteins that regulate mitotic spindles through importin ␤ should directly bind to both MTs and importin ␤. Other potential importin ␤ binding targets, which were not selected for in our assay, might be regulatory factors that interact with MTs transiently, or not at all. To identify candidate MT binding targets, we isolated mitotic MAPs from Xenopus egg extracts (see "Experimental Procedures"), supplemented them with importin ␣ (which in some cases acts as an adaptor between NLS-containing proteins and importin ␤), and loaded these onto an importin ␤ affinity chromatography column. A distinct subset of mitotic MAPs specifically interacted with importin ␤, either directly or in the presence of importin ␣ (marked with asterisks in Fig. 1A). These proteins were excised from the gel, digested with trypsin to generate peptides, and subjected to matrix-assisted laser desorption ionization time-of-flight mass spectrometry to analyze peptide masses. Data base analysis of the peptide masses revealed that 11 peptides of the ϳ100-kDa protein matched the expected masses of tryptic peptides of TPX2, covering 17% of the TPX2 sequence (not shown). Western blotting confirmed the presence of TPX2 in our final fraction (Fig. 1B). This result is consistent with previous reports that TPX2 is a downstream target of the Ran pathway in mitotic spindle assembly (12,43).
Matrix-assisted laser desorption ionization time-of-flight analysis further identified the ϳ150-kDa protein as the Xenopus TACC protein, maskin, and this result was confirmed by Western blotting (Fig. 1B; 12 peptides matched). Maskin is involved in centrosome and spindle assembly in Xenopus egg extracts (27)(28)(29). Interestingly, sequence analysis revealed that maskin lacks a canonical importin ␤ binding domain. Maskin binding to importin ␤ appeared to be independent of importin ␣, but more efficient in its presence ( Fig. 1B; see also Fig. 2). We concluded that despite the absence of an NLS, maskin is a potential importin ␤ target during mitosis.
Maskin Interacts with Importin ␤ in Vitro-Two possibilities could explain why the non-NLS-containing maskin was identified in an assay designed to uncover proteins that are regulated by importin ␤: 1) maskin interacts with importin ␤ via an unconventional importin ␤ binding domain; or 2) maskin interacts with importin ␤ indirectly via other, NLS-bearing proteins. To distinguish between these possibilities, we tested the ability of maskin to interact with importin ␣ and/or ␤ in vitro, in the absence of the other MAPs. For this, recombinant proteins were incubated in vitro. Complexes were selectively isolated by affinity chromatography using the unique tag on one of the proteins ("pull-out" experiments). In one set of pull-out experiments, we incubated protein S-tagged importin ␤ with importin ␣ and/or maskin, isolated importin ␤ using its protein-S tag, and probed for bound maskin ( Fig. 2A). These experiments showed that maskin co-purified with importin ␤, and this binding was independent of importin ␣. Moreover, the interaction between importin ␤ and maskin was abolished in the presence of RanL43E, a Ran point mutant that mimics its GTP-bound state ( Fig. 2A). In another set of experiments, we incubated GST-tagged maskin with importin ␣ and/or importin ␤, re-isolated GST-maskin using glutathione beads, and probed for bound importins (Fig. 2B). We found that maskin interacted directly with importin ␤ as well as importin ␣, and the interaction between maskin and importin ␤ was enhanced by the presence of importin ␣. Importantly, the maskin/importin ␤ interaction was abolished by the addition of RanL43E but not by RanT24N (a RanGDP mimic) (Fig. 2B), suggesting that the in vitro interaction between maskin and importin ␤ is regulated by RanGTP.
The interaction between importin ␤ and maskin was saturable: pull-out experiments with GST-maskin and increasing amounts of importin ␤ (in the presence of constant levels of importin ␣) showed that the amount of importin ␤ that copurified with maskin reached a maximum that was not increased by an excess of importin ␤ (Fig. 2C). Nonlinear regression analysis determined that the apparent equilibrium dissociation constant (K D ) for the interaction between maskin and importin ␤ was 1.3 Ϯ 0.2 M (Fig. 2D).
Maskin Interacts with Importin ␤ in Xenopus Egg Extracts-The result that maskin binds to importin ␤ in vitro supported the idea that maskin might possess an unconventional importin ␤ binding domain. We used four criteria to test this hypothesis: 1) we examined a potential maskin/importin ␤ interaction in egg extracts by co-immunoprecipitation; 2) we examined the subcellular localization of maskin in interphase extracts; 3) we examined the effect of adding exogenous maskin into mitotic extracts; and 4) we mapped the importin ␤ binding region of maskin.
To determine whether maskin and importin ␤ interacted in extracts, we immunoprecipitated maskin from mitotic Xenopus egg extracts and probed for the presence of importin ␤ (Fig.  3A). We found that importin ␤ co-immunoprecipitated with maskin, suggesting that importin ␤ binds maskin in mitotic extracts (see also GST pull-downs in Fig. 5). Importantly, addition of RanL43E, but not RanT24N, to the extract before immunoprecipitation weakened the maskin/importin ␤ interaction but had no effect on the interaction between maskin and XMAP215 or Aurora-A (Fig. 3A).
Further support for the notion that maskin and importin ␤ interact came from experiments using interphase Xenopus egg extracts. These extracts readily assemble nuclei around exogenously added sperm chromatin (50). If maskin interacts with importin ␤ in the extracts, we expect maskin to accumulate in nuclei. Nuclei assembled in egg extracts were probed for maskin using directly labeled affinity purified anti-maskin antibodies (Fig. 3B). This analysis showed that maskin accumulated in nuclei assembled in vitro. There are two possibilities to explain this result: 1) maskin was imported into the nucleus, presumably in an importin ␤ dependent pathway; or 2) maskin was brought into the nucleus with the sperm chromatin. To address this point, we both stained demembranated sperm chromatin with anti-maskin antibodies (Fig. 3C) and assembled nuclei in maskin-depleted egg extracts (not shown). Immunofluorescence analysis showed that sperm chromatin was mostly devoid of maskin staining (Fig. 3C). Similarly, only a very weak maskin signal was detected in nuclei assembled in extracts mostly depleted (ϳ70%) of maskin (not shown). We concluded that maskin was imported into nuclei in interphase Xenopus egg extracts. Because importin ␤ is the major nuclear import receptor in vertebrate cells (7,10), this strongly suggests that maskin interacted with importin ␤.
Maskin Addition to Xenopus Egg Extracts Induces Aster and Spindle Formation-As another approach to test whether maskin is downstream of RanGTP and importin ␤, we tested the effect of adding high concentrations of recombinant maskin to mitotic egg extracts. We reasoned that if maskin was able to FIGURE 1. Several MAPs bind specifically to importin ␤. Mitotic MAPs isolated from Xenopus egg extract were incubated with importin ␤ and/or importin ␣. S-protein beads were added to retrieve importin ␤ and associated proteins (affinity chromatography). One set of controls (as indicated) contained no importin ␤ and thus measures nonspecific binding to S-protein beads. A, Coomassie-stained gel of importin-␤ affinity chromatography. interact with importin ␤ in mitotic egg extracts, it might be able to induce MT assembly by liberating spindle assembly factors from the inhibitory effect of importin ␤ by competing for importin ␤ binding, as was previously reported for the importin ␤-binding protein of NuMA (8), and the artificial importin ␤-binding protein, BSA-NLS (12). As expected, no MT structures formed in mitotic extracts treated with buffer alone, and RanL43E addition resulted in formation of asters and spindles (Fig. 4,  A and B). Significantly, addition of 10 M bacterially expressed maskin to mitotic extracts also resulted in the formation of asters and spindlelike structures (Fig. 4, A and B), although at 10-fold reduced efficiency. Increasing the amount of maskin added did not increase the number of structures formed, but instead exerted a dominant negative effect: the asters were unfocused at intermediate maskin concentrations (20 M), and were disrupted at high maskin concentrations (40 M) (Fig. 4A). Addition of importin ␤ to the reaction was able to overcome the maskin-induced MT assembly in a dose-dependent manner (Fig.  4C), suggesting that the effect of maskin on MT assembly was mediated by importin ␤. We concluded that exogenous maskin induced MT assembly by competing with spindle assembly factors for binding to importin ␤, supporting the notion that maskin interacts with importin ␤ in mitosis.
The Importin ␤ Binding Domain of Maskin Maps to Amino Acids 501-636-Next, we sought to map the importin ␤ binding region of maskin. For this, we generated a series of truncation mutants of maskin (as N-terminal GST fusion proteins to aid in purification and detection; Fig. 5A). The truncations were assayed for importin ␤ binding using two approaches: 1) truncations were incubated in Xenopus egg extract, re-isolated using glutathione-agarose beads, and the amount of importin ␤ present in the pull-outs was quantitated by Western blotting (Fig. 5B); or 2) the truncations were added to Xenopus egg extracts (20 M final concentration) supplemented with rhodamine tubulin, and their activity was scored as the number of asters formed in 50 fields. The results of these assays (summarized in Fig. 5) showed that whereas some truncation mutants bound importin ␤ more efficiently or induced more asters than others, none of the fragments did so as well as the full-length protein. The truncation mutants behaved similarly in both assays, i.e. those mutants that bound more importin ␤ (e.g. the fragment encompassing amino acids 364 -774 or 501-636) also induced more asters. A fragment derived from the central portion of maskin (amino acids 501-636) appeared to be the smallest fragment that showed activity in both assays (Fig. 5). These results suggested that the importin ␤ binding domain of maskin most likely includes residues in the central portion of maskin, but efficient binding of importin ␤ might be dependent on a particular three-dimensional conformation  involving more than one linear domain not preserved in the truncation mutants.
Importin ␤ Regulates the Phosphorylation of Maskin by Aurora-A-Maskin is phosphorylated by Aurora-A kinase in mitosis, and this phosphorylation is involved in maskin function and localization to centrosomes (28,29). Two of the Aurora-A phosphorylation sites of maskin (Ser 620 and Ser 626 ) fall within the portion of maskin that contains the putative importin ␤ binding domain (Fig. 5A). We therefore wondered whether importin ␤ might have an effect on the phosphorylation of maskin. To examine this possibility, we performed in vitro kinase assays in the presence or absence of importin ␤ (Fig.  6A). When importin ␤ was added to a kinase reaction containing maskin, Aurora-A, and [ 32 P]ATP, importin ␤ inhibited phosphorylation of maskin in a dose-dependent manner (Fig.  6B). Consistent with our observation that maskin binds to importin ␣, phosphorylation of maskin was also inhibited by importin ␣, but to a lesser extent than importin ␤ (Fig. 6C). Inhibition by importins was partially reversed by the addition of RanL43E, but not RanT24N, to the reaction, suggesting that phosphorylation of maskin by Aurora-A is regulated by RanGTP.

DISCUSSION
Mitotic spindle assembly requires the precise spatial and temporal regulation of proteins involved in modulating MT nucleation and dynamics. The Ran pathway plays an essential role in this process. Considering the compositional and functional complexity of the mitotic spindle, it is reasonable to expect that the activities of many proteins are modulated by Ran and its downstream effectors. This is underscored by the observation that Ran stabilizes MTs in Xenopus egg extracts by increasing their rate of "rescue" (51,52), yet none of the eight Ran-regulated proteins identified to date affect rescue frequency. It is also interesting to note that with the exception of HURP (18,19), all Ran-regulated spindle assembly factors localize to the spindle poles during mitosis. Paradoxically, the spindle poles are the part of the spindle that is farthest away from the chromosomes, where the RanGTP concentration is highest. Thus, it is clear that important components of this emerging picture are still missing.
We set out to isolate additional downstream targets of Ran, and designed a two-step affinity purification scheme to identify potential candidates. To our surprise, our assay identified maskin as a candidate. Maskin is a centrosomal protein whose phosphorylation by the Aurora-A kinase was recently shown to be essential for its activity (28,29). Here, we describe the characterization of maskin as a novel target of Ran-regulated spindle assembly. Using five criteria we show that maskin and importin ␤ interact directly and specifically: 1) recombinant maskin and importin ␤ interact directly in vitro, and this interaction can be reversed by RanGTP; 2) maskin localizes to the nucleus in interphase extracts; 3) maskin addition to mitotic Xenopus egg extracts induces aster formation; 4) the importin ␤ binding domain of maskin maps to its central portion; and 5) importin ␤ inhibits the phosphorylation of maskin by Aurora-A in a RanGTP-reversible manner.
The Importin ␤ Binding Domain of Maskin-Importin ␤ interacts with a wide variety of cargos, usually by binding to short stretches of basic amino acids, or, in some cases, two basic patches that are separated by 10 -12 amino acids (53). Import signals that arise by bringing together basic patches from different regions of the molecule have also been described (53).
Maskin is a highly acidic protein with an overall pI of ϳ4.5. Not surprisingly, clusters of basic amino acids that could serve as "classical" or "bipartite" NLSs are very rare in maskin (Fig.  5A). Three such "clusters" are found between amino acids 40 and 52 ("RPSILRPSQKDNLPPKPALKSTK"; Fig. 5A; see also supplemental Fig. S1), 378 and 391 ("RKPPPKKLGKRPLLK-TAAKKPSPK"), and 617 and 641 ("KESVLRKQSLYLKFD-PLLRESPKK") of the "non-TACC" portion of maskin, and additional clusters are found in the TACC domain (Fig. 5A). Surprisingly, our experiments suggested that a truncation mutant of maskin comprising amino acids 501-636 was able to interact with importin ␤ and induce asters when added to egg extract. Importantly, however, none of the truncation mutants we tested scored as well as the full-length protein in our assays. This suggests that either the local secondary structure or the global tertiary structure of maskin is required for efficient importin ␤ binding. Taking into consideration that maskin is composed of several highly negatively charged regions in its non-TACC portion, whereas the TACC domain is positively charged (the pI of the TACC domain is ϳ8.7), we imagine that maskin perhaps folds back on itself. This idea is supported by the observation that the TACC domain of maskin can localize to the centrosome on its own, but full-length maskin-3A (a mutant in which three important serine residues have been changed to alanines; Ref. 38) fails to accumulate at centrosomes even though the TACC domain is still present. This indicates that the non-TACC portion of maskin can obscure the centrosome targeting signal in the TACC domain. This hypothesis is further supported by the observation that the three human TACC proteins (hTACC1-3) have distinct subcellular localizations despite closely related TACC domains (54).
Maskin binds importin ␤ directly, although the interaction is more efficient in the presence of importin ␣ (Fig. 2). Thus, maskin joins HURP and Rae1 as a third Ran-regulated spindle assembly factor that interacts with importin ␤ directly (16,19). Both HURP and Rae1 shuttle through the nucleus in interphase (19,55), and we wondered if maskin might also shuttle. We found that maskin accumulated in nuclei assembled in the Xenopus egg extract in vitro. However, we have been unable to find evidence in tissue culture cells that maskin shuttles through the nucleus. It is possible that maskin behaves differently in embryonic systems and somatic cells. However, it is also possible that maskin does shuttle but its lower concentration makes it undetectable once it is dispersed over the nuclear volume. Consistent with this, maskin is ϳ100-fold less abundant in tissue culture cells than in Xenopus egg extracts. A third possibility is that maskin is actively retained in the cytoplasm in tissue culture cells, as has previously been reported for Xnf7 (17).
Maskin is unlikely to be unique among TACC proteins in its ability to interact with importin ␤. For example, all three human TACC proteins show nuclear staining in HeLa cells during interphase, and hTACC3 localizes strongly to the nucleus (54). It will be interesting to know whether human TACC proteins also interact with importin ␤.
Ran and the Centrosome-Although its involvement in spindle assembly was initially discovered in studies of the centrosome-independent pathway of spindle assembly, several lines of evidence are beginning to support a role for the Ran pathway in the regulation of centrosome function. For example, RanGTP associates with centrosomes throughout the cell cycle, and dissociation of AKAP450, the protein that anchors Ran, from the centrosome results in defects in MT nucleation and anchoring (24). The Ran-binding protein, RanBP1, localizes to centrosomes, and its disruption causes centrosome splitting and multipolar spindles (56 -58). Furthermore, importin ␤ (as well as importin ␣) localizes to spindle poles in mitosis in a TPX2-and dynein/dynactindependent manner, and its overexpression causes spindle pole fragmentation (25). Evidence for a connection between the centrosome and the Ran pathway also comes from the observation that several centrosomal proteins cycle through the nucleus, including pericentrin, RanBP1, and centrin-1 (but not ninein, ␥-tubulin, or AKAP450; Ref. 24). Last, Aurora-A kinase local- Positions of positive and negative charge clusters are shown in the bottom two traces, respectively. The graph was generated with the Protean module of the DNAstar software. B, GST maskin truncations were incubated in Xenopus egg extract and re-isolated using glutathione-agarose beads. a, quantitation of the amount of importin ␤ that co-purifies with maskin truncation mutants. The graph represents an average of four independent experiments Ϯ S.E.; *, p Յ 0.05; **, p Յ 0.01 (Student's t test). b, importin ␤ Western blot of a typical pull-out experiment. C, quantitation (average of three independent experiments Ϯ S.D.) of the number of asters formed upon addition of 20 M maskin truncation mutants (as indicated) to Xenopus egg extract, normalized to the number of asters formed in reactions containing full-length maskin.
izes to centrosomes and is required for centrosome function during mitosis. Although Aurora-A kinase is not a direct target of Ran or importin ␤, its activator, TPX2, is a downstream effector of the Ran pathway (14,43). Here, we show that one of its targets (maskin) is unable to be phosphorylated in the presence of importin ␤.
Although many spindle assembly factors regulated by RanGTP and importins that have now been identified (7), mechanistically, these interactions are still poorly understood. Importin ␣ and/or ␤ inhibit the ability of XKid, XCTK2, and HURP to bind to MTs or bundle them, but they do not affect the MT binding or bundling activity of Xnf7 and TPX2 (14,15,17). On the other hand, the ability of TPX2 to activate Aurora-A is inhibited by importin ␣/␤ binding (14,43). We show here that importin ␤ inhibits the phosphorylation of maskin by Aurora-A (Fig. 6), whereas it has little or no effect on the interaction between maskin, Aurora-A, and XMAP215 (Fig. 3). The ability of importins to inhibit phosphorylation by Aurora-A constitutes a novel mechanism for regulating a spindle assembly factor.
Maskin is phosphorylated on three Aurora-A consensus sites, and mutations of these three sites result in disruption of its localization to the centrosome and its activity (28,29). Phosphorylated TACC proteins are exclusively centrosomal, whereas un-phosphorylated TACCs also localize along spindle MTs (29,38). Based on these findings, it has been suggested that phosphorylation by Aurora-A might alter the conformation of maskin or release maskin from interacting proteins that inhibit its recruitment to the spindle (28). Here, we show that importin ␣ and ␤ prevent maskin phosphorylation, suggesting that importins need to be released from maskin to allow its recruitment. We propose a mechanism whereby importin ␤ ensures that maskin is only phosphorylated at the centrosome. The presence of RanGTP at the centrosome leads to the local release of maskin from importin ␤, and to phosphorylation of maskin by Aurora-A. We further propose that phosphorylation increases the affinity of maskin for one or more centrosomal proteins thus anchoring it there, perhaps by exposing the previously inaccessible TACC domain. Although maskin is likely to also be released from importins in the vicinity of chromosomes, this release does not result in phosphorylation because Aurora-A is located exclusively at the centrosome and spindle poles.
In summary, we show that importin ␤ interacts with maskin. Most strikingly, this interaction regulates the phosphorylation of maskin by Aurora-A. The results described here are a first account of an effect of importin ␤ on the phosphorylation state of a spindle assembly factor. Aurora-A phosphorylation is important for TACC function and for its association with the centrosome. Thus, these data strengthen the emerging view that the regulation of centrosome function by importin ␤ and Ran contributes to spindle assembly.