Microtubule assembly by the Apc protein is regulated by importin-β—RanGTP

Adenomatous polyposis coli (Apc) is a tumour suppressor that is mutated in most colorectal cancers. Apc is a large multidomain protein with many binding partners. It controls the Wnt-regulated turnover of β-catenin and stabilises MTs at specific cellular locations. The β-catenin-binding sites (15 and 20 amino acid repeats) in the middle of the Apc molecule are necessary for the downregulation of β-catenin (McCartney and Nathke, 2008). The interaction of Apc with MTs is achieved primarily via a direct, C-terminal MT-binding domain that is rich in basic residues and has been implicated in Apc-mediated microtubule assembly and bundling in vitro (Deka et al., 1998), in MT stabilisation in vivo (Zumbrunn et al., 2001) and in spindle formation (Dikovskaya et al., 2004). However, this domain is dispensable for the association of Apc with MTs, suggesting that other sites are also involved (Zumbrunn et al., 2001). Apc can also associate with MTs indirectly, via several binding partners. EB1, a MT plus-end-binding protein, binds to Apc at its extreme C terminus (Su et al., 1995). Binding of Apc to EB1 is required for EB1-induced MT polymerisation in vitro and in permeabilised cells (Nakamura et al., 2001). The formin mDia2 binds to the C-terminal basic domain of Apc, and a complex between Apc, EB1 and mDia2 is involved in Rho-induced MT stabilisation (Wen et al., 2004). In addition, the kinesin adaptor protein Kap3, which binds to the N-terminal armadillo repeats of Apc, can target Apc to MT clusters in migrating epithelial cells (Jimbo et al., 2002).

Cancer-associated mutations in Apc result in expression of a truncated molecule that is unable to interact with partners that bind to the C-terminal half of Apc. Such mutations usually lead to loss of the majority of the β-catenin-binding sites, as well as all C-terminal MT-targeting sites. Accordingly, the misregulation of both Wnt signalling and cell migration, which results from Apc mutation, is implicated in the pathogenesis of colorectal cancers (Polakis, 2007; McCartney and Nathke, 2008). Additionally, mutations in or lack of Apc cause chromosomal instability (Fodde et al., 2001; Kaplan et al., 2001; Tighe et al., 2004; Dikovskaya et al., 2007), a hallmark of colorectal cancers, which could be partially due to the defects in mitotic spindles induced by the absence of functional Apc (Green and Kaplan, 2003; Dikovskaya et al., 2004; Draviam et al., 2006; Dikovskaya et al., 2007).

Not much is known about how the different interactions and activities of Apc are coordinated in cells. Here, we identify a new regulatory mechanism for Apc function and describe our discovery that importin-β can bind to Apc and modulate its activity, specifically its effect on MTs.

Importin-β is a transport factor that is involved in nuclear import of various cargo molecules in interphase and contributes to mitotic spindle formation by ‘delivering’ spindle-promoting activities to specific cellular regions during mitosis. The ability of the abundant importin-β protein to bind its cargo is inhibited by the small GTPase Ran in its GTP-bound state. In interphase, the chromatin-bound exchange factor for Ran, RCC1, is responsible for high concentrations of RanGTP in the nucleus (Bischoff and Ponstingl, 1991). At the same time, the conversion of RanGTP to RanGDP by cytoplasmic Ran GTPase-activating protein RanGAP1 and the Ran-binding protein RanBP1 is responsible for high concentrations of RanGDP in the cytoplasm (Bischoff et al., 1994; Bischoff et al., 1995). During mitosis, a diffusion-limited RanGTP concentration gradient around chromatin is produced by this system. This RanGTP gradient directs the localised release of spindle-assembly factors from importin-β and their activation near chromatin (Clarke and Zhang, 2008; Kalab and Heald, 2008). Cytoplasmic areas of elevated RanGTP generated by RanBP10 have also been implicated in re-organising peripheral MTs in interphase (Schulze et al., 2008).

We discovered that importin-β and Apc interact in a RanGTP-regulatable manner. Furthermore, we found that interaction sites between Apc and importin-β overlap with functionally important regions of Apc, including binding sites for β-catenin, EB1 and MTs. Importin-β reduced the association of Apc with β-catenin, but not with EB1 in vitro. Significantly, importin-β inhibited the ability of Apc to assemble and bundle MTs in vitro, and to promote the formation of MT structures in Xenopus egg extracts. Furthermore, it negatively regulated the spindle-promoting activity of Apc in Xenopus egg extracts. Thus, importin-β has emerged as a potent regulator of the MT-related function(s) of Apc.

In an attempt to identify potential regulators of the tumour suppressor Apc, we probed material co-precipitated with endogenous Apc from Xenopus egg extracts using a panel of antibodies against known proteins. To account for the possible contribution of mitotic modifications, we used both interphase extract (I) and extract naturally arrested at metaphase of meiosis II by a cytostatic factor (CSF), which mimics a mitotic environment (M). Importin-β specifically co-immunoprecipitated with Apc (Fig. 1A) from both CSF and interphase extracts. Furthermore, recombinant GST—importin-β, but not GST alone, bound Apc from both CSF and interphase Xenopus egg extracts (Fig. 1B).

The association of endogenous Xenopus Apc with endogenous (Fig. 1A) or recombinant (Fig. 1B) importin-β was reduced in the presence of RanQ69L, a mutant of RanGTPase that mimics the GTP-bound state (Hughes et al., 1998). Thus, the Apc—importin-β interaction is dissociated by RanGTP, similar to other importin-β cargo.

We further confirmed a direct interaction between Apc and importin in vitro. We found that a purified recombinant fragment of Apc comprising the middle and C-terminal regions of Apc (referred to as APC4; for a schematic, see Fig. 2A) bound to purified recombinant GST—importin-β immobilised on glutathione beads, but not to GST alone (Fig. 1C). This binding was not mediated by the His-HA-Flag tag of the Apc fragment, because substitution of this tag with a glutamic-acid-rich (Glu) tag did not alter the binding of APC4 to importin-β (supplementary material Fig. S1C). This binding was significantly reduced in the presence of RanQ69L, but not nucleotide-free wild-type Ran (Fig. 1C; supplementary material Fig. S1D), confirming a regulatory role for RanGTPase in the Apc—importin-β interaction. Consistently, depletion of the Ran-interaction domain of importin-β, located in the first 45 amino acids (Gorlich et al., 1996), did not reduce binding of importin-β to Apc, but completely abolished the regulation of this interaction by RanGTP (supplementary material Fig. S1D).

A direct interaction between Apc and importin-β was further confirmed by our finding that binding to importin-β protected APC4 against degradation by several proteases (supplementary material Fig. S2; data not shown).

To define the domain(s) of Apc involved in binding to importin-β, we used a panel of Apc fragments (Fig. 2A; supplementary material Fig. S1A) in in vitro binding assays. Deletion of the direct MT-binding domain of APC4 (fragment referred to as APC4ΔMT) did not diminish binding to importin-β (supplementary material Fig. S1B), indicating that this domain is not required for the binding between APC4 and importin-β. Therefore, we used APC4 and APC4ΔMT interchangeably in some experiments (as indicated in the figure legends).

To refine the mapping, APC4 was divided into M-APC and C-APC fragments (see Fig. 2A). In vitro binding assays revealed that both M- and C-APC bound importin-β directly (Fig. 2B). This was further confirmed by the pattern of Apc fragments that were protected by importin-β against proteolysis (supplementary material Fig. S2). Detailed analysis of the Apc fragments generated by partial proteolysis in the presence of importin-β narrowed the interaction sites in the middle of Apc to residues 1266-2040, suggesting that almost the entire middle region of Apc is involved in binding to importin-β (supplementary material Fig. S2). C-APC was further subdivided into C1, C2 and C3 fragments (see Fig. 2A). Using these fragments, we identified C2 (residues 2166-2547) and C3 (residues 2544-2843) as additional regions that can bind importin-β independently of each other (Fig. 2C). C3m, a C3 fragment that lacks the last 18 C-terminal amino acids and contains two amino acid substitutions at positions 2805 and 2806, which eliminates EB1 binding (Honnappa et al., 2005; Li et al., 2008), still bound importin-β (Fig. 2C, lowest panel). Both M-APC and C3-APC bound importin-β in a Ran-dependent manner (Fig. 2D,E).

Thus, Apc can bind importin-β through several binding sites, one located in the middle of the Apc molecule and two others in C-terminal regions. There was a notable difference between the relative binding of these different Apc fragments to importin-β: whereas the middle of Apc, M-APC, required a large amount of importin-β for detectable binding in vitro, smaller amounts of importin-β were able to bind C2- and C3-APC efficiently (compare the amount of importin-β in the Coomassie gels between Fig. 2D and Fig. 2E, and between Fig. 3C and Fig. 3D,E).

In order to map the site(s) in importin-β that interact with Apc, we used a panel of importin-β truncation mutants (Fig. 3A) and measured their ability to bind APC4 in vitro under non-saturating conditions (when approximately half of the Apc in the reaction is bound to full-length importin-β). Truncated importin-β comprising residues 1-462 bound APC4ΔMT at least as well as full-length importin-β (residues 1-876) (Fig. 3B). However, shorter fragments containing residues 1-380 or 1-365, which lack the armadillo (also known as β-catenin-like) repeat domain (Arm) of importin-β located between residues 367 and 453 (UniProtKB/Swiss-Prot entry Q14974), bound APC4ΔMT (Fig. 3B) and APC4 (supplementary material Fig. S3A) poorly. Nevertheless, residues 1-380 of importin-β were sufficient to pull down endogenous Apc from CSF Xenopus egg extracts (supplementary material Fig. S3B). This suggested that the Arm-like domain of importin-β is a major, but not the only binding site for Apc.

We further examined whether the three different importin-β-binding sites in the Apc molecule bind to different positions in importin-β. M-APC bound to importin-β containing the Arm-like region, with some residual binding to shorter importin-β fragments lacking this region, similar to APC4 (Fig. 3C). C3-APC, on the other hand, bound equally well to all truncated forms of importin-β tested, including residues 1-365 (Fig. 3D). Likewise, truncated importin-β containing residues 1-365 bound the same amount of C2-APC as full-length importin-β (Fig. 3E), indicating that C-terminal regions of Apc bind to N-terminal domains of importin-β. Consistently, truncated importin-β containing residues 1-365 protected only C-terminal fragments of Apc against proteolysis (supplementary material Fig. S2A)

Thus, we identified several connections between Apc and importin-β: a strong interaction between two C-terminal APC fragments and the N-terminal half of importin-β; and a weaker interaction of M-APC with the middle (residues 380-462) of importin-β (Fig. 3F).

Residues 380-462 of importin-β are involved in binding to at least one other cargo protein, rpL23a (Jakel and Gorlich, 1998). BLAST alignment of the Apc and rpL23a protein sequences revealed four fragments of Apc with more than 50% similarity to the importin-β-binding domain (BIB domain) of rpL23a (supplementary material Fig. S4A). Three of these fragments are located in the middle portion of Apc and one is in C2-APC (asterisks in Fig. 3F). Furthermore, a recombinant BIB fragment reduced both binding of M-APC to importin-β in vitro (supplementary material Fig. S4B) and the association of endogenous Apc and importin-β in Xenopus egg extract (supplementary material Fig. S4C). This indicates a potential similarity between regions of Apc and rpL23A in their mode of interaction with importin-β.

Our mapping data indicated that the importin-β-binding site in the middle of Apc overlaps with the region that binds β-catenin (Fig. 3F). Therefore, we tested whether β-catenin and importin-β compete for binding to Apc. Soluble importin-β reduced the association of M-APC with immobilised β-catenin (Fig. 4A), indicating that importin-β and β-catenin can compete for Apc in vitro. However, β-catenin appears to bind Apc more tightly than importin-β: 2 μg (17.1 pmol) of β-catenin were sufficient to completely bind 1.5 μg (6.1 pmol) of APC4ΔMT in vitro (supplementary material Fig. S5A), whereas only a fraction of the same amount of APC4ΔMT was bound by ten times more importin-β (Fig. 3B).

One of the C-terminal regions of Apc that binds to importin-β overlaps with the EB1-binding site, which resides between residues 2673 and 2843 (Askham et al., 2000). To test the possibility that the Apc-EB1 interaction could be affected by importin-β, we measured the effect of recombinant importin-β (lacking a GST tag) on the binding of C3-APC to GST-EB1 immobilised on glutathione beads. Importin-β did not change the ability of Apc to bind EB1 (Fig. 4B). Furthermore, soluble importin-β was effectively recruited to EB1 bound to C3-APC. These data show that the C terminus of Apc can bind importin-β and EB1 simultaneously.

Another functionally important domain of Apc that overlaps with one of the importin-β-binding sites is the MT-binding domain, located within the C2 region of Apc. We therefore tested whether the binding of Apc to MTs in vitro was altered by importin-β. We found that importin-β did not affect the ability of Apc to bind MTs: the amount of APC4 or C2-APC that co-sedimented with taxol-stabilised MTs did not change in the presence of importin-β (Fig. 5A,B). By contrast, β-catenin reduced binding of Apc to MTs in vitro and in vivo (supplementary material Fig. S5B; data not shown).

The MT-binding fragment of Apc was previously reported to promote assembly of MTs from purified tubulin (Deka et al., 1998). Indeed, addition of APC4 to a solution of pure tubulin that was too dilute for self-assembly (Fig. 6A, grey line) resulted in a rapid increase in light scattering (Fig. 6A, red lines). Microscopy confirmed that MT structures were formed under these conditions (Fig. 6B, top panel). Importantly, we found that importin-β significantly reduced the assembly of MTs induced by APC4 (Fig. 6A, blue lines; Fig. 6B, bottom panel). Notably, importin-β also reduced aggregation of APC4 in BRB80 buffer in the absence of tubulin (Fig. 6A, compare orange and purple lines).

To measure this effect under more physiological conditions, we measured the ability of Apc to promote MT formation in Xenopus egg extracts. In clarified CSF extract, APC4 on its own did not induce MT nucleation; however, it greatly enhanced the formation of MT asters in extracts supplemented with DMSO and taxol (Fig. 6C,D). This effect was noticeably reduced when APC4 was stably bound to importin-β mutant protein lacking the Ran-interaction domain (Fig. 6C). We used this mutant to prevent its dissociation from Apc in extract by endogenous RanGTP. Addition of the same amount of importin-β mutant on its own did not reduce the amount of DMSO-nucleated MTs, but promoted their elongation (Fig. 6C,D). Thus, importin-β reduces Apc-mediated assembly of MT structures in vitro and in cytosolic extracts.

The ability of Apc to bundle MTs both in vivo and in vitro has been described previously (Zumbrunn et al., 2001). To determine whether importin-β regulates this function of Apc, we determined the effect of importin-β on Apc-induced bundling of taxol-stabilised MTs in vitro. Incubating APC4 with taxol-stabilised MTs induced MT bundles (Fig. 6E, middle panel). When importin-β was added to MTs together with Apc, the density of bundles was measurably reduced (Fig. 6E-G). Thus, importin-β reduces the ability of Apc to form thick MT bundles.

The surprising discovery that importin-β reduced MT assembly and bundling by Apc, but not binding of Apc to MT, prompted us to investigate how Apc bundles MTs. The exact mechanism of MT assembly and bundling by Apc is unknown. One possibility is that Apc crosslinks MTs through several MT-binding sites. However, only one direct MT-binding site, located in the C2-APC region, has been described so far (Munemitsu et al., 1994; Deka et al., 1998). In search of additional MT-binding sites, we tested the ability of other fragments of C-APC (which is sufficient for MT bundling in vitro and in vivo) to bind MTs. We found that C3, but not C1, also bound MTs in MT co-pelleting assays (Fig. 5C). In the presence of importin-β, the binding of C3-APC to MTs was slightly reduced, but only at a low concentration of MTs: the association of C3 with 0.8 μM, but not 1.6 μM, MTs was partially inhibited by an excess of importin-β (Fig. 5C).

To validate the physiological relevance of an interaction between Apc and importin-β, we took advantage of the unique property of Xenopus egg extract to serve as a functional model for spindle assembly while allowing biochemical manipulation of molecular complexes. We previously described the spindle-promoting activity of Apc in CSF Xenopus egg extracts and showed that it depends on the direct interaction of Apc with MTs (Dikovskaya et al., 2004). So we asked whether importin-β could regulate such activity.

Spindles formed in CSF-arrested Xenopus extract, supplemented with rhodamine-labelled tubulin and Xenopus sperm-derived chromatin, were chilled to 12°C for 6-8 minutes. The proportion of cold-resistant spindles and spindles that were destroyed by exposure to the low temperature (representative images are shown in Fig. 7A) was counted. Spindles formed in extracts depleted of Apc (Fig. 7B) were disrupted more readily by exposure to cold than spindles in mock-depleted extracts (Fig. 7C,E,F), consistent with the idea that Apc is required for the stabilisation of K-fibers, cold-resistant bundles of MTs attached to kinetochores (Salmon and Begg, 1980). Addition of recombinant APC4 to endogenous Apc levels (Fig. 7B) partially restored the resistance of Apc-depleted spindles to cold (Fig. 7C,E,F). These results mirrored our previous data showing that APC4 rescues the aberrant spindle morphology of APC-depleted CSF extract (Dikovskaya et al., 2004). Similar to our previous results, excess Apc was detrimental to spindles. Thus, for the first time, we established a functional assay to measure the contribution of Apc to spindle formation in Xenopus egg extract. This assay provided a more robust read-out of the spindle-promoting activity of Apc than the Apc-dependent morphological changes described previously (Dikovskaya et al., 2004).

In order to measure the effect of importin-β on the spindle-promoting activity of Apc, we pre-bound recombinant Apc to a Ran-insensitive importin-β mutant (supplementary material Fig. S1B). This prevented the dissociation of Apc from importin-β by the high concentration of RanGTP near chromatin.

Pre-incubation of APC4 with Ran-insensitive importin-β significantly reduced the ability of APC4 to rescue the Apc-depleted phenotype (an example is shown in Fig. 7E; average data shown in Fig. 7F). We confirmed that Apc remained bound to Ran-insensitive importin-β in extract for the duration of the assay (supplementary material Fig. S6A). The amount of Ran-insensitive importin-β mutant used was minute compared to that of endogenous importin-β (Fig. 7D) and on its own did not destabilise Apc-deficient (Fig. 7E) or control (supplementary material Fig. S6B) spindles at the concentrations used in our experiments: addition of importin-β mutant alone resulted in an average 1.1±0.1-fold increase in the proportion of cold-stable spindles in extracts. This shows that the negative effect of importin-β was due to its effect on Apc and not another importin-β cargo protein. We concluded that, when bound to importin-β, Apc is inhibited from supporting formation of stable spindles and that Ran-mediated release from importin-β is required for this Apc activity.

Thus, importin-β is a novel binding partner for Apc that negatively regulates the ability of Apc to assemble and bundle MTs and promote spindle stabilisation.

Apc performs many functions in cells through its interactions with multiple binding partners. Here, we identified importin-β as a direct binding partner and regulator of Apc. Most prominently, importin-β reduces the ability of Apc to promote MT assembly and bundling. Apc binds, bundles and stabilises MTs in vitro and in vivo (Zumbrunn et al., 2001). This feature of Apc is crucial to its spindle-promoting activity, because the direct MT-binding domain is required for this function (Dikovskaya et al., 2004). Furthermore, Apc is involved in cell migration and polarity (McCartney and Nathke, 2008), two other phenomena that depend on MT network regulation by Apc.

To allow the localised regulation of MT structures by Apc, its MT-stabilising activity must be spatially controlled. Importin-β is an excellent candidate to provide such control: it is itself part of a dynamic machinery that mediates the transport and activation of different cargo molecules. Similar to other cargo proteins, Apc binding to importin-β is inhibited by RanGTP. Thus, repression of the MT-stabilising activity of Apc by importin-β can be locally released in areas where concentrations of RanGTP are high, for example, near chromatin in the mitotic spindle. Consistent with this idea, Apc bound to a Ran-insensitive importin-β mutant cannot promote spindle formation, presumably because its MT-stabilising activity is inhibited by the stably bound importin-β.

Interestingly, importin-β inhibits the ability of Apc to assemble and bundle MTs without significantly affecting the binding of Apc to MTs per se. This is in contrast to the effect of β-catenin, which reduces the binding of Apc to MTs (supplementary material Fig. S5B). How Apc bundles MTs is not known and only one direct MT-binding domain (the ‘basic domain’) had previously been described for Apc. Here, we identified a second MT-binding site in Apc, which is located C terminal to the classical basic domain. Consistently, Apc that lacks the basic domain still binds MTs, but its ability to stabilise MTs is strongly reduced (Zumbrunn et al., 2001). We found that importin-β slightly diminished the MT association of this newly identified MT-binding domain of Apc at low concentrations of MTs. However, this effect is much weaker than that described for other cargo molecules of importin-β, such as XCTK2 (Ems-McClung et al., 2004).

Although we cannot exclude that a minor reduction in MT binding contributes to the negative effect of importin-β on the MT-bundling activity of Apc, we favour an alternative explanation. It is likely that, for efficient bundling or assembly of MTs, Apc needs to oligomerise. Notably, large clusters of Apc are found at the tip of MT bundles in Apc-dependent cellular protrusions (Zumbrunn et al., 2001; Nathke et al., 1996), suggesting that oligomerisation is involved in Apc-mediated MT stabilisation at these sites. We found that importin-β reduces the aggregation of APC4 in vitro (Fig. 6A) consistent with the idea that importin-β acts to reduce multimerisation of Apc.

Importin-β is known to act as a chaperone to prevent ionic aggregation of basic proteins such as rpL23a (Jakel et al., 2002). We found significant similarity between the importin-β-binding domains of Apc and rpL23a, and detected competition between Apc and the importin-binding domain of rpL23a for importin-β (supplementary material Fig. S4). Furthermore, one of the importin-β-binding sites in Apc, the classical MT-binding domain, is rich in basic residues, which could potentially mediate aggregation and require chaperone protection. Thus, it is possible that importin-β functions as a chaperone to inhibit self-association specifically of charged Apc domains.

It is tempting to speculate that, for multimerising MT-bundling proteins, such chaperone function could serve to inhibit MT assembly or bundling by reducing the clustering of such proteins. We hypothesise that importin-β inhibits the self-association of Apc required for efficient MT bundling, resulting in the formation of smaller Apc clusters or single Apc molecules. When such importin-Apc complexes encounter a MT, importin-β is released and Apc binds the MT; however, the number of available MT-binding domains remains too low to bundle or assemble MTs efficiently. This hypothesis explains how importin-β can influence MT assembly by Apc without either reducing the binding of Apc to MTs or forming a ternary complex with Apc and MTs. Alternatively, importin-β could bind tubulin dimers (Gache et al., 2005) and reduce their incorporation into MT structures formed by Apc.

In conclusion, we have identified importin-β as a potent modulator of Apc function. Importin-β binds to Apc at several sites in a Ran-regulatable manner and inhibits the ability of Apc to bundle and assemble MTs, but not MT binding per se. We further showed that the spindle-promoting activity of Apc in mitosis is negatively regulated by importin-β. Thus, the importin-β—Ran system can provide a control mechanism for cellular functions of Apc that involve its MT-stabilising activity.

His-HA-Flag-tagged APC4 and APC4ΔMT, fragments of Apc containing the middle and C-terminal portions (residues 1015-2843 of human Apc), were produced using baculovirus from pFB-NHis10HA plasmids containing indicated constructs as described (Dikovskaya et al., 2004). Glu-tagged APC4 (residues 1034-2844 of human Apc) (Rubinfeld et al., 1993) was a kind gift of Paul Polakis (Genentech, South San Francisco, CA, USA). His-tagged APC4 (residues 1015-2843) was expressed in Escherichia coli from pET28a plasmids containing APC4 that was recloned from the above pFB-NHis10HA plasmid using SacI and BamHI restriction sites. His-HA-Flag-tagged C-APC fragment (residues 2038-2843 of human Apc) was produced using baculovirus expressing pFB-NHis10HA plasmids containing C-APC (a kind gift from Aurora Burds, MIT, Cambridge, MA, USA). His-tagged C-APC (residues 2035-2843), C1-APC (residues 2035-2182), C2-APC (residues 2166-2547), C3-APC (residues 2545-2843) and C3m-APC (residues 2166-2826, with the substitution of Ile2805 and Pro2806 to serines) fragments were produced in E. coli expressing the indicated fragments in pET28a plasmids as described (Li et al., 2008). M-APC (residues 1015-2041) was recloned into the BamHI site of pET15b plasmid from M-APC-containing pEGFP plasmid (Zumbrunn et al., 2001) and produced in E. coli as a His-tagged protein. GST-tagged importin-β constructs, including full-length importin-β (1-876), were expressed in E. coli from pGEX4T1 (Hutchins et al., 2004). Importin-β truncations comprising residues 1-462 and 1-380 were generated from a GST—importin-β construct (Hutchins et al., 2004) by introducing a STOP codon at the relevant position using the QuikChange kit (Stratagene). N-terminal truncation (45-876) was carried out using QuikChange to generate an internal EcoRI restriction site immediately 5′ to the desired start codon; the 5′ EcoRI fragment was then excised and the DNA religated. For the GST-free importin-β, full-length importin-β was amplified from pGEX4T1 by PCR and recloned into pGEX-6P1, and the GST tag was cleaved from purified protein using PreScission protease (GE Healthcare). Truncated importin-β comprising residues 1-365 was amplified from full-length importin-β by PCR and cloned into pGEX-6P1. GST-tagged RanQ69L and wild-type Ran (Hughes et al., 1998) were expressed in E. coli from pGEX-5X2 plasmid containing the appropriate sequence. Purified RanQ69L protein was loaded with GTP as described (Hughes et al., 1998). GST-tagged EB1 was produced in E. coli from an EB1 construct in pGEX-4T3 (Askham et al., 2000), which was a kind gift from Ewan Morrison (St. James's University Hospital, Leeds, UK). Proteins were purified on Ni-agarose or glutathione-sepharose as appropriate and, in some cases, further purified by ion exchange chromatography and gel filtration. GST—β-catenin was expressed and purified as described (Huber et al., 2001). BIB fragment (the importin-β-binding domain of rpL23a, containing residues 32-74 of rpL23a) was amplified from HeLa cDNA (a kind gift from Daniel Klotz, University of Dundee, UK) by PCR and cloned into pGEX-6P1. GST tag was cleaved from the purified protein using PreScission protease (GE Healthcare).

CSF-arrested Xenopus egg extracts were prepared as described (Desai et al., 1999; Swedlow, 1999). Interphase extract was derived from CSF extract by supplementing it with 0.8 mM CaCl₂ followed by incubation for 30-40 minutes, monitored by examining decondensation of exogenously added demembranated sperm nuclei using DAPI-stained samples. For immunodepletion or immunoprecipitation of Apc from the extracts, 100 μl Dynabeads slurry (Dynal, Invitrogen) were loaded with 26 μg affinity-purified anti-APCII antibody (Nathke et al., 1996) or the equivalent amount of rabbit IgG (Sigma). Antibody-bound beads were washed in buffer containing 10 mM HEPES pH 7.7, 1 mM MgCl₂, 100 mM KCl, 150 mM sucrose and drained. 100-120 μl freshly prepared Xenopus egg extract were incubated with the antibody-bound beads for 1 hour rotating at 4°C. Beads were subsequently removed from extracts using a Dynal MPC-S magnet. For immunoprecipitation, the isolated beads were further washed in PBS supplemented with 150 mM NaCl and 0.1% Tween 20. The precipitated material was separated on 4-12% PAGE (Invitrogen), transferred onto Protran nitrocellulose membrane with 0.1 μm pore size (Schleicher & Schuell) and analysed by immunoblotting with appropriate antibodies. For GST pull-down, GST or GST-tagged importin-β immobilised on glutathione-sepharose (Amersham) were incubated at 4°C for 1 hour with CSF or interphase extract before washing the beads with PBS supplemented with 150 mM NaCl and 0.1% Tween 20, and analysing the precipitated material as above. When required, the extract was preincubated for 40 minutes with RanQ69L before immunoprecipitation or GST pull-down. For pull-down with the truncated importin-β, an N-terminal fragment of importin-β containing residues 1-380 was covalently bound to activated Affi-Prep 10 beads (BioRad), before incubating with CFS extract.

CSF extract was clarified by centrifugation at 210,000 g for 2 hours at 4°C. MT structures were induced by addition of 2 μl of buffer containing 10 mM HEPES pH 7.2, 100 mM KCl, 5% glycerol, 1 mM GTP, 16.25% DMSO, 0.2 μM taxol and 62.5 μM TubulinTracker (Oregon Green 488 taxol, bis-acetate, Invitrogen), supplemented, where indicated, with 0.45 μg APC4 (0.1 μM final concentration) and/or 0.7 μg Ran-insensitive importin-β mutant (0.38 μM final concentration) to 14 μl of clarified extract. When added together, APC4 and mutant importin-β were pre-incubated for 30 minutes on ice before addition. The mixture was spotted onto a slide, covered with a coverslip and the formation of MT structures was followed at room temperature using 40×/0.60 NA lens on a Leica DMIRB microscope operated by Improvison Openlab 5.02 software. Images were captured with a Hamamatsu camera.

Only freshly prepared extracts were used for spindle formation. Spindles were formed in the presence of 4-6 × 10⁵ demembranated sperm nuclei and approximately 0.02 μg rhodamine-labelled bovine tubulin (Cytoskeleton) per 1 μl of extract (~0.2 μM) by incubating the mix for 50-65 minutes at room temperature. When needed, APC4 and/or importin-β mutant in buffer containing 10 mM HEPES pH 7.7, 1 mM MgCl₂, 100 mM KCl and 150 mM sucrose, or the buffer alone, were added to the spindle-assembly reaction in a volume not exceeding one-twentieth of the extract volume. APC4 and mutant importin-β were pre-incubated for 30 minutes on ice before addition to the extract. For the cold-sensitivity assay, spindles formed in a small volume (20-30 μl) of CSF extract were incubated for 6-8 minutes at 12-16°C in a thermocycler before spotting a sample onto a glass slide and fixing with 4% formaldehyde solution containing 60% glycerol, 1 μg/ml DAPI and 1 × MMR buffer (100 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 0.1 mM EGTA, 5 mM HEPES pH 7.8). Only a small number of samples were processed in each individual experiment, to keep consistent timing between control and experimental samples (this precluded duplicate samples in individual experiments). Spindle structures were examined using a 20×/0.7 NA lens on a Leica DMIRB microscope operated by Improvison Openlab 5.0 software. Images were captured with a Hamamatsu ORCA camera.

Recombinant GST-tagged importin-β, EB1 or β-catenin immobilised on glutathione-sepharose 4B (Amersham) was incubated for 1 hour with Apc fragments at 4°C in binding buffer containing 50 mM HEPES pH 7.5, 100 mM KCl, 2 mM EGTA and 0.05% CHAPS. Bound Apc fragments were co-sedimented with glutathione beads by centrifugation for 3 minutes at 8000 rpm at 4°C in a centrifuge with a swing-out rotor. After collecting the unbound material, the beads were washed in binding buffer and the bound proteins were directly dissolved in 2 × LDS sample buffer (Invitrogen). Both bound and unbound fractions were separated on the two 4-12% PAGE gels (Invitrogen), with half of each fraction analysed by Colloidal Coomassie (Invitrogen) and another quarter of the fraction analysed by western blot with appropriate antibodies. When required, Ran mutants were added together with Apc fragments at a final concentration of 10 μM. For the competition assay, the recombinant GST-free importin-β was added together with APC fragment at a final concentration of 0.73 μg/μl (7.5 μM).

A total of 1 μg (4.3 pmol) of purified APC4ΔMT pre-incubated for 1 hour with or without 10 μg (83.3 pmol) of purified GST-tagged importin-β in binding buffer containing 50 mM HEPES pH 7.5, 100 mM KCl, 2 mM EGTA and 0.05% CHAPS was digested with different amounts of endoproteinase Asp-N (Roche) for 30 minutes at room temperature and directly resolved on 4-12% PAGE (Invitrogen). Alternatively, the reaction was stopped with a protease inhibitor cocktail containing 1 mM PMSF and 100 μg/ml each of leuptin, pepstatin A and chymotrypsin, and the proteolytic fragments were immunoprecipitated with anti-HA antibodies immobilised on protein G-sepharose before resolving on the gel as above. The resolved fragments were directly transferred onto Protran nitrocellulose membrane with 0.1 μm pore size (Schleicher & Schuell) and analysed by western blot with various anti-Apc antibodies.

MT binding was measured as previously described (Haren and Merdes, 2002). Taxol-stabilised MTs were incubated with Apc fragments with or without importin-β for 15 minutes at 37°C in BRB80 buffer (80 mM K-Pipes pH 6.8, 1 mM MgCl₂, 1 mM EGTA) and spun through the under-layered cushion of 30% glycerol in BRB80 at room temperature at 14,000 rpm for 13 minutes. The supernatant was precipitated in 10 volumes of cold 100% methanol. Both pellet and precipitated supernatant were resolved on PAGE, and immunoblotted with appropriate antibodies.

Taxol-stabilised MTs produced in the presence of a small amount of rhodamine-labelled tubulin were incubated with Apc fragments and importin-β as above, and fixed in two volumes of 0.5% glutaraldehyde, before spotting onto glass slides. MT bundles were examined under a Leica microscope as above.

A total of 1.2 mg/ml (12 μM) tubulin in 1 × BRB80 containing 1 mM DTT and 1 mM GTP was placed into 37°C warm ultra-micro Quartz SUPRASIL cuvette, light path 10 mm (Hellma), and up to 40 ng/μl (0.14 μM) of APC4 and/or 188 ng/μl (1.57 μM) of importin-β, or equal volumes of corresponding buffers was added. Light scattering of the mixture was measured at 37°C for 20 minutes immediately after mixing. Absorbance at a 350 nm wavelength was recorded using a temperature-controlled UV-VIS recording spectrophotometer UV-1601 with CPS controller (Shimadzu, Japan). When APC and importin-β were added together, they were pre-incubated on ice for 20 minutes before addition to tubulin solution.

We thank Ewan Morrison for providing the EB1 expression construct, Bill Hunter and members of the Näthke laboratory for helpful discussions, Paul Appleton for help with imaging, Holly Goodson for advice with the light scattering assay, and Alan Fairlamb for providing a spectrophotometer. This work was funded by a program grant from Cancer Research UK to I.S.N.