Transportin Regulates Major Mitotic Assembly Events: From Spindle to Nuclear Pore Assembly

Recombinant Protein Cloning, Expression, Purification, and Antibodies

GST-human transportin (pGEX6P-Trn) was cloned by ligating a BamHI and XhoI fragment containing full-length transportin (Transportin 1; AAH40340) from pET28a-Trn (S. Shah and D. Forbes, unpublished data) into pGEX6P-3 vector (GE Healthcare, Piscataway, NJ). Cloning of GST-Xenopus importin beta was described in Delmar et al. (2008). Glutathione S-transferase (GST), GST-tagged Xenopus importin beta, and GST-tagged human transportin proteins were expressed in BL21 competent cells (EMD Chemicals, Gibbstown, NJ) by inducing with 0.1 mM IPTG and grown overnight at 17°C. Glutathione-Sepharose 4B beads (GE Healthcare) were used to purify the GST-tagged protein according to manufacturer's instructions. To obtain untagged importin beta and transportin, the GST-tagged protein was cleaved using GST-Precision Protease (GE Healthcare) and incubated at 4°C for 4 h. The cleaved protein was eluted and dialyzed into 5% glycerol in PBS (8 g/l NaCl, 2 g/l KCl, 1.44 g/l Na₂HPO₄, and 0.24 g/l KH₂PO₄, pH 7.4) and stored at −80°C. All the Xenopus importin beta and human transportin recombinant proteins added to the reactions are untagged, except in the case of pulldown experiments where GST-tagged importin beta and transportin were used. We note that GST-tagged and 6xHis-tagged transportin behaves indistinguishably from untagged transportin.

6xHis- and GST-tagged RanQ69L-GTP were expressed, purified, and loaded with GTP as previously described (Harel et al., 2003a). GST-ELYS fragments were expressed and purified as previously described (Rasala et al., 2008).

Antibodies used in this study included rabbit anti-gp210 (Harel et al., 2003a), anti-hNup133 and anti-xNup160 (Harel et al., 2003b), anti-xNup43 (Orjalo et al., 2006), anti-Nup93 (Miller and Forbes, 2000), anti-Nup53 and anti-xELYS (Rasala et al., 2008), anti-Orc2 (generous gift from Dr. John Newport, University of California, San Diego), anti-Xenopus importin beta (Rasala et al., 2006), and mouse anti-human transportin and anti-Nup62 (BD Biosciences, San Jose, CA).

Membrane Fusion and Nuclear Pore Assembly Assays

Xenopus high-speed cytosol and membranes, both from interphase Xenopus eggs, and Xenopus sperm chromatin were prepared as described in Harel et al. (2003a). Membrane fusion and nuclear pore assembly assays were performed as described previously (Harel et al., 2003a). Nuclear membrane assembly was analyzed (Figure 1A) using the fluorescent membrane dye 3,3-dihexyloxacarbocyanine iodide (DHCC; Eastman Kodak, Rochester, NY) and visualized with an Axiovert 200M confocal microscope (Carl Zeiss, Thornwood, NY) and a 63× objective. Images from the confocal microscope were recorded using a Coolsnap HQ (Photometrics, Tucson, AZ) camera and Metavue software (Molecular Devices, Downingtown, PA). Images were processed using ImageJ (available at http://rsb.info.nih.gov/ij/) and Adobe Photoshop (San Jose, CA). For dextran diffusion analysis of membrane integrity (Figure 1B), in vitro nuclear reconstitution reactions were carried out as above, except that at the beginning of the reaction additions were made as follows: GST, as a control (25 μM); GTPγS (2 mM); BAPTA (7.5 mM); RanQ69L-GTP (37.5 μM); GST-Trn (25 μM); or RanQ69L-GTP (37.5 μM) plus GST-Trn (25 μM). Assembly was allowed to proceed for 60 min, and then WGA (100 μM; EY Laboratories, San Mateo, CA) was added for 10 min to further ensure a tight seal of any nuclear pores present (although nuclear pores are already expected to exclude 70-kDa dextran). Reactions were stopped on ice, and then rhodamine-labeled dextran (70 kDa; 2.5 μg; Molecular Probes, Eugene, OR) was added to a 25-μl reaction, followed by 15-min incubation on ice. Reactions were diluted 1:1 and fixed with egg lysis buffer containing 7.4% paraformaldehyde. Dextran exclusion was visualized with an Axioskop2 microscope (Carl Zeiss) at a magnification of 100× using an oil objective (Carl Zeiss). Nuclear pore assembly (Figure 2) was visualized with the same microscope and a 63× objective as determined by staining for FG nucleoporins with mAb414 antibody (Covance, Berkeley, CA) which was directly labeled with Alexa 488 (Invitrogen, Eugene, OR) per manufacturer protocol. BAPTA [1,2-bis(2-aminophenoxy)ethane-N,N,N,N-tetraacetic acid; EMD Chemicals] was added at 8 μM.

Annulate Lamellae Assembly

Xenopus cytosol (40 μl), membranes (5 μl), glycogen (5 μl), and the indicated amount of recombinant proteins were incubated together for 90 min at room temperature. Ten microliters of the reaction were diluted in 190 μl ELB (10 mM HEPES, pH 7.6, 50 mM KCl, and 2.5 mM MgCl₂) and overlaid onto a 300-μl sucrose cushion (0.5 M) in ELB. The samples were spun at 25,000 rpm for 20 min at 2°C in a TL-100 tabletop ultracentrifuge (Beckman Coulter, Fullerton, CA). The membrane pellet was collected, rinsed with ELB, and resuspended in 100 μl 1× SDS-PAGE sample loading buffer. One tenth of the volume was loaded for SDS-PAGE and processed for immunoblotting.

For GST pulldowns, recombinant GST, GST-Xenopus importin beta, and GST-human transportin were incubated with glutathione-Sepharose beads (GE Healthcare) without cross-linking. After blocking with 20 mg/ml bovine serum albumin (BSA) in PBS for 30 min, Xenopus egg cytosol was spun for an additional 150,000 rpm for 10 min at 4°C to remove residual membrane contamination and then added to the beads and incubated for 2 h at 4°C (25 μL cytosol in 500 μL PBS). After washing the beads with PBS, the bound proteins were eluted with 0.1 M glycine, pH 2.5, and neutralized with 1 M Tris, pH 8.0. One fifth of each reaction was loaded for SDS-PAGE and processed for immunoblotting.

Chromatin Binding for Immunofluorescence Microscopy

Xenopus cytosol, sperm chromatin, and recombinant proteins were incubated together for 20 min at room temperature. The reactions were diluted in 800 μl ELB, overlaid on a 300 μl 25% sucrose cushion in ELB, and centrifuged at 750 rpm (100 × g) for 15 min onto poly-l-lysine–treated coverslips. Coverslips were then fixed in 4% formaldehyde in PBS for 10 min at room temperature and processed for immunofluorescence microscopy. Nuclei were visualized with an Axioskop 2 microscope (63× objective; Carl Zeiss).

Anchored Chromatin Assay for Immunoblotting

Xenopus cytosol was heat inactivated at 100°C for 3 min and then spun at 14,000 rpm for 20 min to collect the supernatant. Twenty-five microliters of heat-inactivated cytosol, which contains primarily the protein nucleoplasmin, was used to decondense 500,000 Xenopus sperm chromatin at room temperature for ∼15 min, and the state of chromatin decondensation was monitored by DAPI staining and fluorescence microscopy. The decondensed sperm chromatin were diluted in 300 μl ELB (10 mM HEPES, pH 7.6, 50 mM KCl, and 2.5 mM MgCl₂) and allowed to bind to poly-l-lysine–coated coverslips by gravity for 2 h at room temperature. The chromatin covered coverslips were blocked with 4% BSA in ELB. The blocked chromatin coated coverslips were then incubated with cytosol or cytosol in the presence of 20 μM Xenopus importin beta, 20 μM human transportin, or 30 μM RanQ69L-GTP recombinant protein for 20 min at room temperature. After three washes in ELBK (10 mM HEPES, pH 7.6, 100 mM KCl, and 2.5 mM MgCl₂) the chromatin was lysed in 30 μl 1× SDS-PAGE sample loading buffer, and the bound proteins were subjected to SDS-PAGE and immunoblotting.

ELYS Direct Binding Assay

To assess direct binding, 1.0 μM untagged Xenopus importin beta or human transportin was incubated with 1 μM GST, GST-RanQ69L, GST-ELYS AT-hook+, GST-ELYS ΔAT-hook, or GST-ELYS short AT-hook+ in 1× binding buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.2 mM EDTA, 1 mM DTT, 0.2% BSA, 1 μg/ml μg/ml pepstatin, 1 μg/ml leupeptin, and 1 μg/ml protease inhibitor TPCK, 1 mM benzamidine, and 1 mM PMSF) with or without 5 μM RanQ69L-GTP on ice for 1 h. Meanwhile, glutathione-Sepharose beads (GE Healthcare) were washed and blocked with 20 mg/ml BSA (Sigma-Aldrich, St. Louis, MO) for 30 min at room temperature in 1× binding buffer. Ten microliters of the glutathione beads were added to the recombinant proteins in each reaction and incubated for additional 30 min at 4°C in a total volume of 100 μl, before washing three times with 1× binding buffer (without BSA) and two times with PBS. The bound proteins were eluted with 27 μl 0.1 M glycine, pH 2.5, followed by neutralization with 3 μl 1 M Tris, pH 8.0, and addition to 10 μl 4× SDS-PAGE sample loading buffer. Samples were boiled and loaded onto 12% SDS-PAGE gels, which were silver-stained, stained with Coomassie blue, or transferred to PVDF for immunoblotting.

Spindle Assembly Experiments

Frogs for interphase extract preparation were induced to ovulate with injection of 500 U of human chorionic gonadotropin (Sigma-Aldrich) and put into individual buckets containing 100 mM NaCl for 18 h. The frogs laid eggs overnight at 18°C, and the eggs were collected the next morning. After removing most of the NaCl solution from the best batches, eggs were dejellied with a 2% cysteine solution, pH 7.7. Activated and dead eggs were constantly removed using a transfer pipette (Fisher Scientific, Pittsburg, PA). When dejellied, eggs were washed extensively three times with XB buffer (100 mM KCl, 0.1 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, and 50 mM sucrose, pH 7.7). Using a cut transfer pipette, eggs were transferred in 2-ml polypropylene tubes (Beckman) and crushed by centrifugation using a TL55S rotor in a Beckman TL100 ultracentrifuge for 15 min at 15,000 rpm. The cytoplasmic layer was removed with a 18-guage needle on a 3-ml syringe, gently drawn out, and supplemented with cytochalasin B (50 μg/ml) and a protease inhibitor cocktail (10 μg/ml aprotinin and 10 μg/ml leupeptin). The cytoplasmic extract was then centrifuged a second time for 15 min at 15,000 rpm. The light layer was recovered and used as the interphase extract.

The mitotic extract (cytostatic factor or CSF extract) was made exactly as the interphase extract, except for the use of the buffer CSF-XB (100 mM KCl, 0.1 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 50 mM sucrose, and 7 mM EGTA, pH 7.7) instead of XB.

For spindle assembly, ∼6.5 × 10⁴ sperm chromatin DNA (3000 SpC/μl) were added to 20 μl of interphase extract, as well as ATP energy mix (10 mM phosphocreatine, 80 μg/ml creatine kinase, 1 mM ATP, 2 mM MgCl₂, and 100 μM EGTA). At t = 90 min after the DNA addition, the extract now containing assembled nuclei was supplemented with rhodamine-tubulin (final concentration 20 μg/ml) and 20 μl of mitotic extract to induce entry into mitosis. Where indicated, GST-transportin was added to a final concentration of 20 μM and RanQ69L-GTP to a final concentration of 5 μM. At t = 15 or 60 min after mitotic extract addition, aliquots were withdrawn and fixed with 1 μl of fixation buffer (48% glycerol, 11% formaldehyde, and 10 mM HEPES, pH 7.5) supplemented with 5 μg/ml Hoechst DNA dye. Spindle formation was monitored using a Zeiss Axioskop 2 microscope using a 63× objective.

Transportin Blocks Nuclear Membrane Formation In Vitro in a Ran-reversible Manner

Importin beta has been shown to act as a negative regulator of both nuclear membrane and nuclear pore assembly (Harel et al., 2003a; Walther et al., 2003b; Delmar et al., 2008). We asked whether another nuclear import receptor, transportin, could regulate nuclear envelope assembly in the Xenopus laevis in vitro nuclear assembly system. In a control reaction, when chromatin was mixed with Xenopus egg cytosol and membranes, completely fused nuclear membranes were observed to encircle the chromatin by 60 min, as indicated by a smooth membrane stain surrounding each nucleus (Figure 1A, Ctrl). The addition of GTPγS, a known inhibitor of vesicle–vesicle fusion, inhibited nuclear membrane formation as evidenced by the discontinuous membrane profile, indicative of unfused membrane vesicles (Figure 1A, GTPγS; Newport and Dunphy, 1992; Macaulay and Forbes, 1996). Importin beta gave, as expected, a similarly discontinuous membrane profile around the chromatin (data not shown; Harel et al., 2003a; Delmar et al., 2008). When excess transportin was added, unfused membranes were also observed (Figure 1A, Trn). Importantly, this membrane fusion defect was counteracted by the positive regulator Ran in the form of RanQ69L-GTP (a form of Ran that cannot hydrolyze GTP; Hughes et al., 1998; Figure 1A, compare Trn+Ran with Trn), as previously seen with importin beta (Harel et al., 2003a; Delmar et al., 2008). We also measured membrane fusion (or lack thereof) by determining whether the structures formed around chromatin after 60 min of reconstitution were able to exclude 70-kDa rhodamine-labeled dextran; 70-kDa dextran is known to be unable to diffuse through nuclear pores in completely formed nuclear membranes, but can diffuse into nuclei if there are gaps in the nuclear membranes or if nuclear membrane assembly never proceeds beyond the vesicle-binding stage. We found that when either transportin or GTPγS was added, then exclusion of rhodamine 70-kDa dextran was not observed (Figure 1B, red). Observation of a discontinuous nuclear rim by differential interference contrast (DIC) microscopy confirmed this lack of fusion when transportin or GTPγS were added (Figure 1B, gray). When Ran was added with transportin, dextran exclusion and a continuous rim were now observed. Taken together, these results showing a discontinuous rim by DIC microscopy (Figure 1B) and by DHCC membrane labeling (Figure 1A), as well as an absence of dextran exclusion (Figure 1B, red) all indicated that an intact nuclear membrane does not form in the presence of excess transportin. We conclude that transportin negatively regulates nuclear membrane fusion and does so in a Ran-reversible manner.

Transportin Negatively Regulates Nuclear Pore Assembly

To determine whether transportin separately blocks mature nuclear pore assembly, we formed nuclear intermediates that contained complete nuclear membranes, but which lacked nuclear pores. This was done by performing nuclear assembly in the presence of the calcium chelator BAPTA (Macaulay and Forbes, 1996; Harel et al., 2003a; Delmar et al., 2008). BAPTA nuclear intermediates contain a fused nuclear envelope, but no mature nuclear pores, as previously shown by electron microscopy and a lack of detectable FG nucleoporins by immunofluorescence with mAb414 (Macaulay and Forbes, 1996; Harel et al., 2003a; Delmar et al., 2008). When such BAPTA nuclear intermediates are diluted 1:10 into fresh cytosol lacking BAPTA, nuclear pore assembly quickly ensues (Macaulay and Forbes, 1996; Harel et al., 2003a; Delmar et al., 2008). This rescue was typically apparent from the acquisition of FG-nucleoporin staining by the rescued nuclei as shown in Figure 2A (+Buffer). If the added Xenopus cytosol instead contained a second addition of BAPTA, rescue was not observed (Figure 2A, +BAPTA). However, if BAPTA nuclei were diluted into cytosol containing excess transportin at either 10 or 20 μM concentrations, the incorporation of FG-nucleoporins was substantially reduced by 10 μM transportin (Supplemental Figure S1) and was completely blocked by 20 μM transportin (Figure 2A, +h-Trn; Supplemental Figure S1). We found that the ability of transportin to block nuclear pore assembly was partially reversed by RanQ69L-GTP: FG-staining pores were now observed, albeit fewer in number (Figure 2A, +h-Trn+Ran; see also 5× high-magnification inset) than in the control (+buffer). This partial rescue of pore assembly by Ran-Q69L was also observed when importin beta was the inhibitory agent (Figure 2A, +x-beta +Ran). When RanQ69L-GTP was added alone, it did not have any negative effect on nuclear pore assembly (Figure 2A, +Ran). To quantitate these effects, seventy-five nuclei for each condition in each of three different experiments were counted for a strong FG nucleoporin rim, a partial FG rim, or the absence of FG staining. The quantitation, as shown in Figure 2B, clearly confirms that transportin blocks nuclear pore assembly and that this block is partially reversed by Ran-GTP.

Annulate Lamellae Assembly Is Negatively Regulated by Transportin

AL are stacked membranes within the ER membrane network that contain cytoplasmic mimics of intact nuclear pore complexes (Merisko, 1989; Dabauvalle et al., 1991; Meier et al., 1995). In most ways these AL pores are identical to nuclear pores, but they are assembled differently in that they do not require the pore-targeting protein ELYS or chromatin in order to form (Kessel, 1989; Merisko, 1989; Dabauvalle et al., 1991; Meier et al., 1995; Miller and Forbes, 2000; Rasala et al., 2008). A previous study found that 10 μM transportin did not block AL pore assembly (Walther et al., 2003b). However, in that electron microscopic experiment, transportin was only tested in the presence of 5 μM RanQ69L-GTP and thus was likely lower in effective concentration. We observed in our experiments above that 20 μM transportin was required to completely block nuclear pore assembly. [We note that our assembly extracts are about twice as concentrated as those used in Walther et al. (2003b)]. We next set out to ask whether transportin (20 μM, with no added RanQ69L-GTP) could block in vitro AL pore assembly, as determined by immunoblotting of membrane pellets for nucleoporins. To form ALs, Xenopus egg cytosol was mixed with Xenopus membranes in the absence of chromatin and incubated for 90 min. The AL were isolated by centrifugation and subjected to immunoblot analysis, and the incorporation of multiple key nucleoporins into ALs was assessed. When ALs were formed under control conditions, all the individual nucleoporins tested including Nup133, Nup43, Nup93, and Nup53 were detected as present, indicating the formation of AL pores (Figure 3C, lane 1). When we added 20 μM importin beta to an AL assembly reaction, as expected from previous work, it blocked the incorporation of these soluble nucleoporins into membranes (Figure 3C, lane 2). Addition of 20 μM transportin also clearly blocked AL assembly (Figure 3C, lane 4). Indeed, importin beta and transportin inhibited the incorporation of all tested nucleoporins, which in addition to the ones shown in Figure 3C, included Nup214, Nup155, Nup153, Nup98, and Nup62 (data not shown). The block to AL formation by importin beta and transportin was significantly reversed by the addition of RanQ69L-GTP (Figure 3C, lanes 3 and 5). These results demonstrate that transportin is mirroring its affect on nuclear pore assembly and negatively regulating AL pore assembly.

A Subset of Nucleoporins Binds to Both Transportin and Importin Beta

We next wanted to determine the nucleoporin subunit targets of transportin and importin beta regulation of pore assembly. Both direct and indirect interactions between transportin or importin beta and nucleoporins could affect pore assembly. During import in intact cells, both karyopherins are known to bind to FG-containing nucleoporins that are essential for nucleocytoplasmic transport (Madrid and Weis, 2006; Terry et al., 2007; Lim et al., 2008). One hypothesis would be that transportin and importin beta regulate pore assembly by sequestering the essential FG-nucleoporins, except in the region of the chromosomes where RanGTP is high. A broader hypothesis might predict that all nucleoporin subunits are targets of regulation. Lastly, it is conceivable that transportin and importin beta might have overlapping and nonoverlapping nucleoporin targets that can be regulated differentially.

To address the potential nucleoporin subunit targets of receptor inhibition in the Xenopus system in a comprehensive manner, we performed GST pulldowns from Xenopus cytosol using GST, GST-importin beta, or GST-transportin as bait. We probed for the different pore subunits by immunoblotting using antibodies to 12 of the 14 soluble nuclear pore complex subunits with the exception of Aladin and Tpr (Rasala et al., 2008). We found that both importin beta and transportin clearly interacted with the FG-nucleoporins Nup358, Nup214, Nup153, Nup98, Nup62, and Nup50 (Figure 3, A, lanes 3 and 4, and B, lane 4), as expected (Moroianu et al., 1995; Shah and Forbes, 1998; Shah et al., 1998; Yaseen and Blobel, 1999; Fontoura et al., 2000; Ben-Efraim and Gerace, 2001; Lindsay et al., 2002). Nup53, which contains a low number of FGs (Marelli et al., 1998; Devos et al., 2006) also associated with importin beta and transportin (Figure 3A, lanes 3 and 4).

The major subunit of the nuclear pore, the Nup107-160 complex, exists in two complexes: one containing nine core nucleoporins (Nup107, Nup160, Nup133, Nup96, Nup85, Nup43, Nup37, Sec13, and Seh1) and a second complex that also contains the pore-targeting nucleoporin ELYS/MEL-28. A previous study has shown that an ELYS/Nup107-160 complex immunoprecipitated by anti-ELYS antibody from human cell lysates does not contain importin beta (Rasala et al., 2006). In contrast, immunoprecipitates of ELYS/Nup107-160 complex from Xenopus egg extracts with anti-ELYS antibodies contain importin beta (Franz et al., 2007 and our unpublished observations). When GST-importin beta was used here to perform a pulldown assay from Xenopus egg cytosol, all of the tested members of the Nup107-160 complex were observed (Figure 3A, lane 3). The complex was also observed in GST-transportin pulldowns (Figure 3, A, lane 4, and B, lane 4). Strikingly, the two newly discovered Nup107-160 complex-associated proteins, ELYS and centrin (Rasala et al., 2006; Franz et al., 2007; Gillespie et al., 2007; Resendes et al., 2008), were also found to bind to both GST-importin beta and GST-transportin pulldowns (Figure 3A, lanes 3 and 4).

Not all nucleoporin subunits were found to bind to importin beta and transportin: members of the key Nup93/188/205 scaffold subunit (Grandi et al., 1997; Hawryluk-Gara et al., 2005), as well as another subunit of the pore scaffold, Nup155 (Aitchison et al., 1995; Franz et al., 2005), did not show interaction with importin beta or transportin (Figure 3A, lanes 3 and 4).

The above results show that exogenously added importin beta and transportin can bind to a specific and similar subset of nucleoporin complexes in Xenopus egg cytosol. To ask whether importin beta or transportin is found naturally in endogenous nucleoporin subcomplexes, we performed immunoprecipitation with several different anti-Nup antibodies and probed for the presence of importin beta and transportin. We found that endogenous importin beta and transportin were each present in immunoprecipitates with anti-Nup62, anti-Nup50, and anti-Nup98 antibodies (Supplemental Figure S2, A and B). Significantly, these nucleoporins are all known to be present in Xenopus extracts in separate Nup subcomplexes (see, for example, Rasala et al., 2008). We conclude from this sampling of subcomplexes that importin beta and transportin are clearly present in endogenous Nup subcomplexes in vivo.

We looked further at the effect of Ran-GTP on the nucleoporin targets of transportin. The interaction between GST-transportin and the large Nup107-160 complex (Nup160, Nup133, and Nup43) was abolished by the presence of RanQ69L-GTP (Figure 3B, lane 5). Similarly, the binding of transportin to the FG-nucleoporins Nup214 and Nup98 was also prevented by the presence of RanQ69L-GTP (Figure 3B, lane 5), and the interaction with Nup62 was significantly decreased. In contrast, the interaction between transportin and the FG-nucleoporins Nup358, Nup153, and Nup50 remained the same or increased in the presence of RanQ69L-GTP (Figure 3B, lane 5). Interestingly, these latter three proteins are unique among all nucleoporins in that they have the ability to bind RanGTP on their own (Wu et al., 1995; Yokoyama et al., 1995; Saitoh et al., 1996; Nakielny et al., 1999; Lindsay et al., 2002).

Together, the above data demonstrate that importin beta and transportin interact with a near identical and specific subset of nucleoporins. This suggests a model where importin beta and transportin regulate nuclear pore assembly by a similar mechanism. It is interesting that neither importin beta nor transportin interact with the Nup93/188/205 subcomplex or with Nup155. We conclude that the recruitment of these latter nucleoporins into the forming nuclear pore is not likely to be regulated by importin beta or transportin. Thus, it appears that transportin and importin beta regulate not all, but a subset, of nucleoporins and that this subset is substantially similar between the two karyopherins.

Transportin and Importin Beta Regulate Chromatin Binding of ELYS in the First Step of Nuclear Pore Assembly

The first known step of nuclear pore complex assembly involves recruitment of the pore-targeting nucleoporin ELYS/MEL-28 to AT-rich regions of chromatin. Importantly, this recruitment can occur either in the absence or presence of the Nup107-160 complex in Xenopus egg extracts (Rasala et al., 2006, 2008; Franz et al., 2007; Gillespie et al., 2007). However, the Nup107/160 complex cannot bind to chromatin in the absence of ELYS (Franz et al., 2007). ELYS thus initiates nuclear pore assembly. Because we have found that both transportin and importin beta interact with ELYS and the Nup107-160 complex (Figure 3A), it is feasible that these transport receptors regulate the initial step of nuclear pore assembly. We tested this hypothesis using a chromatin-binding assay. For this assay, Xenopus cytosol and sperm chromatin were mixed in solution with or without the recombinant receptor proteins and incubated for 20 min at room temperature. The chromatin templates were centrifuged onto polylysine-coated coverslips and processed for immunofluorescence. In the control, where cytosol was added to chromatin without any added importin beta or transportin, the chromatin binding proteins ELYS and Orc2 both bound to the chromatin (Figure 4A, +Buffer). In contrast, ELYS no longer associated with chromatin if excess importin beta was present in the reaction (Figure 4A, +beta). ELYS was also greatly reduced on chromatin in the presence of transportin (Figure 4A, +Trn). The binding of Orc2 to chromatin was unaffected by either of these karyopherins (Figure 4A, + beta; +Trn). The inhibition of ELYS binding to chromatin by transportin was substantially prevented if recombinant RanQ69L-GTP was included (Figure 4A, +Trn+Ran). A similar effect was observed with the simultaneous addition of importin beta and Ran-GTP (Figure 4A, +beta+Ran). These effects on ELYS binding were quantitated in four different experiments and are shown in Figure 4B. We conclude that importin beta and transportin negatively regulate the binding of ELYS to chromatin, as judged by this immunofluorescence assay.

The binding of ELYS to chromatin was also tested biochemically, using instead an anchored chromatin assay. In this assay, decondensed sperm chromatin was bound to poly-l-lysine coverslips, then incubated with Xenopus cytosol. The unbound cytosolic proteins were washed away, whereas the chromatin-bound proteins were collected in SDS sample buffer, subjected to SDS-PAGE, and tested for the presence of ELYS and the Nup107-160 complex by immunoblotting. In the absence of any added recombinant receptor proteins, ELYS and members of the Nup107-160 complex bound to chromatin, as expected (Figure 5, lane 5; ELYS, Nup160, Nup133, and Nup43). We have previously shown that AT-rich DNA sequences capture ELYS (Rasala et al., 2006, 2008; Franz et al., 2007; Gillespie et al., 2007). ELYS then recruits the Nup107-160 complex (Rasala et al., 2006, 2008; Franz et al., 2007; Gillespie et al., 2007). We found that association of the Nup107-160 complex with chromatin was greatly reduced in the presence of importin beta (Figure 5, lane 6) or transportin (Figure 5, lane 9). The Orc2 signal present on the chromatin remained unchanged whether or not importin beta or transportin were added (Figure 5, cf. lanes 6 and 9 with lane 5). The block of ELYS and Nup107-160 complex binding to chromatin could, in large part, be prevented by the addition of RanQ69L-GTP (Figure 5, lanes 7 and 10). Taken together, the data indicate that importin beta and transportin can negatively regulate the recruitment of the pore targeting protein ELYS to chromatin and do so in a Ran-regulated manner.

Transportin and Importin Beta Bind to ELYS Directly

As shown above, transportin and importin beta negatively regulate nuclear pore assembly, prevent ELYS from binding to chromatin, and interact with ELYS in pulldown assays from Xenopus egg cytosol. To determine whether importin beta or transportin can bind to ELYS directly, we performed a direct binding assay using different regions of the C-terminus of ELYS which are known to associate with chromatin (Rasala et al., 2008). The ELYS fragments used were GST-tagged versions of ELYS AT-hook+ (aa 2281-2408), ELYS ΔAT-hook (aa 2359-2408), and ELYS-short AT-hook+ (aa 2281-2359; Figure 6A). All three fragments bind to chromatin, although GST-ELYS AT-hook+ binds somewhat more strongly, as it possesses two chromatin binding domains (Rasala et al., 2008; and data not shown). The GST-tagged versions of each of these ELYS fragments were incubated with untagged importin beta or untagged transportin in the absence of any cytosolic proteins. The bound fraction was analyzed by silver stain to detect importin beta (Figure 6B). GST-RanQ69L-GTP, a positive control, clearly bound importin beta (Figure 6B, lane 1). GST alone, a negative control, did not bind importin beta (Figure 6B, lane 2). We found that both ELYS AT-hook+ and ELYS ΔAT-hook directly bound importin beta (Figure 6B, lanes 3 and 4, respectively). However, ELYS-short AT-hook+ did not bind to importin beta (Figure 6B, lane 5).

We further found that both ELYS AT-hook+ and ELYS ΔAT-hook bound to transportin directly (Figure 6C, lanes 3 and 4), whereas GST-ELYS-short AT-hook+ did not bind transportin (Figure 6C, lane 5), as determined by immunoblotting. RanGTP substantially prevented the binding of transportin to ELYS AT-hook+ and ELYS ΔAT-hook (Figure 6E, lanes 8 and 6, respectively), a significant reversal also observed with importin beta and Ran (Figure 6D, lanes 4 and 6). These data indicate that transportin and importin beta can bind directly to the extreme C-terminus of ELYS in a region that does not include the AT-hook.

Transportin Blocks Spindle Assembly in a Ran-sensitive Manner

Importin beta has been shown in multiple studies using mitotic Xenopus egg extracts to regulate spindle assembly. Mitotic spindles readily assemble in vitro around chromatin added to such an extract (see Kalab and Heald, 2008; Clarke and Zhang, 2008 for review). It has been proposed that the location of spindle assembly specifically in the area of mitotic chromosomes in cells is orchestrated by importin beta and its adaptor protein importin alpha, which act to inhibit spindle assembly factors elsewhere (Clarke and Zhang, 2008; Kalab and Heald, 2008). RanGTP, produced by its chromatin-bound RanGEF, releases spindle assembly factors from inhibition in the vicinity of chromosomes; thus Ran acts as a positive regulator of spindle assembly and importin alpha and/or beta act as negative regulators. When excess exogenous Ran is added to Xenopus mitotic cytosol, abundant asters then are induced to form throughout the cytosol, even in the complete absence of chromatin (Carazo-Salas et al., 1999; Kalab et al., 1999; Ohba et al., 1999; Wilde and Zheng, 1999).

Here we asked whether the import receptor transportin would show a similar ability to regulate spindle assembly. We reasoned that this could very well be the case since, like importin beta, transportin negatively inhibits both nuclear membrane assembly and nuclear pore assembly (Figures 1–5).

To assess effects of transportin on mitotic spindle assembly, nuclei were assembled by adding sperm chromatin to a Xenopus interphase egg extract (20 μl) and allowing them to assemble into nuclei for 90 min (Figure 7A, top left panel, 0 min). During this time, it has been demonstrated that the DNA present faithfully replicates (Newport, 1987; Tutter and Walter, 2006). Mitotic egg extract (20 μl) was then added to convert the reaction to a mitotic state. Rhodamine-labeled tubulin was simultaneously added to allow the monitoring of mitotic spindle (or aster) assembly. Transportin (20 μM) was added with or without RanQ69L-GTP (5 μM) in parallel reactions to test for effects on spindle assembly. At 15 min after the addition of mitotic cytosol, essentially all of nuclei in the reactions had been converted to condensed chromatin figures that had an attached rhodamine-labeled half spindle, as is typical of this time point in spindle assembly assays in Xenopus mitotic extracts (Mitchison et al., 2004; Maresca and Heald, 2006; Orjalo et al., 2006; Figure 7A, 15 min). We noted that when transportin was present the half spindles appeared somewhat smaller. At 30 min, small bipolar spindles were beginning to form in both the control and transportin reactions (data not shown). By 60 min, the control showed abundant robust bipolar spindles organized around condensed chromatin (66%; Figure 7A, 60 min). However, the chromatin in reactions with added transportin were for the most part now devoid of associated microtubules (89%; Figure 7A, 60 min). In contrast, when RanQ69L-GTP was included with the transportin, robust (40%) and weak (22%) bipolar spindles formed (Figure 7A, 60 min). Addition of RanGTP alone caused large numbers of asters to form independent of chromatin (52%) as well as the formation of multipolar spindles in the vicinity of chromatin (38%). The data from three such experiments has been quantitated in Figure 7B. The 60-min Control results are shown in blue, transportin results are in red, and transportin plus RanGTP results shown in yellow. We conclude that transportin, like importin beta, is an inhibitor of mitotic spindle assembly in vitro and this inhibition can be counteracted by RanGTP.

Dueling Regulators

Ran reversal of importin beta inhibition is, for the most part, a hallmark of the studies of importin beta regulation to date. When tagged human importin beta was shown to block mitotic spindle assembly and nuclear membrane assembly in Xenopus extracts, both processes were reversed by RanGTP (Nachury et al., 2001; Wiese et al., 2001; Harel et al., 2003a; Ems-McClung et al., 2004; Blower et al., 2005; Albee et al., 2006; Ribbeck et al., 2006; Tahara et al., 2008). An exception to this was the inhibition of nuclear pore assembly by tagged human importin beta, where RanGTP did not reverse the inhibition (Harel et al., 2003a). It was later found, however, that when an untagged version of either human or Xenopus importin beta was used to inhibit nuclear pore assembly, the inhibition was RanGTP-reversible (Delmar et al., 2008). In the present study, we used untagged human transportin, but found that its block to pore assembly was only partially reversed by excess RanGTP. One possibility is that human transportin may be less sensitive to RanGTP than importin beta is, specifically in the area of nuclear pore assembly. An alternate explanation is that there may be a separate effector of transportin in addition to the GTPase Ran involved in pore assembly. Potentially consistent with this, we did observe that although RanGTP was able to release transportin from certain nucleoporins (the FG nucleoporins Nup98 and Nup214 and the Nup107-160 complex), RanGTP left intact transportin's binding to the FG nucleoporins Nup358, Nup153, and Nup50 (Figure 3B), perhaps arguing for the existence of a second positive regulator for transportin. It is known that these latter proteins are unique among nucleoporins in that they contain known RanGTP-binding domains (Wu et al., 1995; Yokoyama et al., 1995; Saitoh et al., 1996; Nakielny et al., 1999; Lindsay et al., 2002). It is possible that their regulation by transportin is reversed via a different mechanism.

Multiple Points of Regulation in Pore Assembly

In mammalian cells, the order of assembly of the Nups has been very broadly categorized using GFP-tagged nucleoporins into early, intermediate, and late-assembling nucleoporins (Bodoor et al., 1999; Dultz et al., 2008). Assuming this order is followed in Xenopus cells as seems likely (Walther et al., 2001; Harel et al., 2003b; Walther et al., 2003b; Rasala et al., 2006, 2008), we conclude that transportin and importin beta bind to nucleoporin subunit targets in each of the temporal stages of pore assembly. Transportin and importin beta bind to ELYS and the Nup107-160 complex, which assemble early; Nup98, which has an intermediate assembly time (Dultz et al., 2008); and Nup50 and Nup214, which assemble late, as well as Nup153 which has both early- and late-assembling populations (Bodoor et al., 1999; Dultz et al., 2008). Thus, it appears that there could be sequential points of regulation by transportin and importin beta during nuclear pore assembly.

Multiple Regulators

Why is more than one nuclear import receptor used to regulate nuclear assembly? One explanation would be that, although transportin and importin beta bind to a similar subset of nucleoporins, they may bind with different affinities. The nucleoporins could then be differentially regulated in a more controlled manner. Another possibility is that the availability or concentration of importin beta accessible for regulation in vivo might be quite different from that of transportin, providing another mechanism for fine tuning the regulation of steps in nuclear assembly. Lastly, we do not exclude the possibility that additional import receptors may regulate nuclear assembly in a similar way.

Regulation of the Initial Step in Nuclear Pore Assembly

The binding of ELYS to chromatin is the initial step in pore assembly. We show this to be regulated by both transportin and importin beta (Figures 4 and 5). In the absence of the recruitment of ELYS to chromatin, pore complexes no longer form in the nuclear envelope, but instead form as annulate lamellae pores in the cytoplasm (Rasala et al., 2006, 2008; Franz et al., 2007). We would predict that if an excess of ELYS were allowed to bind to chromatin, then excessive nuclear pores would form that would lead to defects in nuclear assembly including, for example, aberrant assembly of intranuclear annulate lamellae as previously observed in the presence of excess Ran (Harel et al., 2003a) and to defects in pore spacing or nuclear lamina formation. A disproportion of nuclear pore numbers could also adversely affect the novel functions recently recognized to be mediated by the nuclear pore complex, such as transcriptional activation of certain genes (Casolari et al., 2004; Cabal et al., 2006; Taddei et al., 2006; Brown et al., 2008) and DNA break repair (Nagai et al., 2008). In consequence, we believe it is likely that pore number would be tightly regulated. Interestingly, most vertebrate cultured cell lines have a very similar number of nuclear pores (∼2000–5000), whereas inactive B cells have many fewer nuclear pores (∼400; Maul, 1977). In reality, little is known of the mechanism of pore number regulation. Regulation of the targeting of ELYS to chromatin by transportin and importin beta could be one such mode of regulation.

Karyopherin-binding Sites on the C-Terminus of ELYS

The C-terminal 50 aa of Xenopus ELYS binds both importin beta and transportin directly (aa 2359-2408, Figure 6A, yellow/red). Sequence analysis reveals that the last 18 amino acids contain 10 positively charged residues in two clusters (RRTRRRIIAKPVTRRKMR_COOH; Figure 6A, red). Importin beta is well known to bind to positively charged regions in proteins ranging from its adaptor, importin alpha, to cargoes that importin beta binds directly, such as HIV Rev and ribosomal proteins L23a and S7 (Gorlich and Kutay, 1999). Thus, each basic cluster may bind importin beta directly. The most recently reported transportin NLS consensus sequences contain a PY dipeptide preceded by various alternate upstream amino acid sequences, one of which contains 5–8 basic amino acids (basic-enriched_5–8-X_8–10-PY; Lee et al., 2006). Although this abundant class of transportin NLSs was preselected by computer analysis on the basis of containing a PY element, the protein HuR, a known transportin cargo, contains a PG in place of PY in its NLS (Lee et al., 2006). Also, of two known cargoes for the S. cerevisiae transportin homologue kap104, Hrp1 contains a PY, whereas Nab2 has a PY-type NLS where a PL replaces the PY (aa 200-250). Each is essential for kap104 binding of these cargoes (Siomi et al., 1998; Truant et al., 1998; Lee and Aitchison, 1999; Marfatia et al., 2003; Lange et al., 2008). Here we found that the C-terminus of Xenopus ELYS directly binds transportin, even though it does not contain a PY; it does, however, contain a PV dipeptide that maps between the two basic clusters at the ELYS C-terminus (Figure 6A, red). It is possible that this PV serves in place of the PY. Indeed, the ELYS sequence RRTRRRIIAKPVTRRKMR_COOH resembles somewhat the predicted transportin NLS of the human metal response element-binding transcription factor 2 (KKGKKKSVGRPPGPYTRKM; Lee et al., 2006).

For vertebrate ELYS, transportin and importin beta have both been shown to bind to the last 50 amino acids (Figure 6), although they may also bind elsewhere on ELYS. If it is found that they bind to the identical sequence or to adjacent sequences within the positively charged ELYS C-terminus, it may be that they contribute to the negative regulation of ELYS binding to chromatin in an interchangeable manner. Alternatively, one of the karyopherins may be present in greater abundance in vivo or be higher in its binding affinity and thus act as the dominant regulator of that particular site. It is also highly possible that importin beta and transportin bind to additional regions of ELYS and that multiple binding events are required for controlling ELYS. In any case, their goal presumably would be to achieve negative regulation of ELYS except in areas of high RanGTP, such as at the surface of chromatin.

Multiplex Regulation of Spindle Assembly

Lastly, we demonstrated that transportin has the ability to negatively regulate an entirely different cell cycle structure, the mitotic spindle, and does so in a Ran-sensitive manner (Figure 7). Our data thus support the idea that, like importin beta, transportin has diverse functions during different parts of the cell cycle, namely as an import receptor during interphase and as a negative regulator of mitotic spindle and nuclear assembly during mitosis. Importin beta was demonstrated to regulate mitotic spindle assembly by binding and sequestering SAFs in areas away from the mitotic chromosomes (reviewed in Harel and Forbes, 2004; Funabiki, 2005; Kalab and Heald, 2008). RanGTP, produced on the mitotic chromosomes by the RanGEF RCC1, acts as a positive regulator for mitotic spindle formation by triggering dissociation of importin beta from its SAF binding partners (Kalab and Heald, 2008). An interesting question will be whether transportin binds to the same SAFs as importin alpha and beta or different ones. At present, at least 10 SAFs have been identified that are direct targets of importin beta and/or alpha (including NuMA, TPX2, and the kinesin/MAP XCTK2; Clarke and Zhang, 2008; Kalab and Heald, 2008). A cursory analysis of these SAFs reveals that several contain a PY dipeptide (D. Forbes, unpublished data), but whether these are authentic transportin binding sites is not yet known. Significantly, the S. cerevisiae kap104 mutant kap104-E604K has been found to speed up mitosis in vivo, consistent with a negative regulatory role for transportin in mitosis (Asakawa and Toh-e, 2002). However, because yeast retain an intact nucleus during their closed mitosis, this has also been suggested to result from a defect in the import role of kap104.

An interesting question is whether other karyopherins similarly sequester SAFs. At present it is known that the export karyopherin Crm1 acts during mitosis but in a different manner, targeting to kinetochores and aiding in the binding of a set of proteins important for kinetochore function (Arnaoutov et al., 2005; Arnaoutov and Dasso, 2005; Knauer et al., 2006). Because export receptors are stabilized by Ran, one might expect that only import receptors could duel with RanGTP to regulate assembly of the large mitotic cellular structures around the chromatin. However, export receptors could also regulate these assembly events if they used an entirely different way to reverse target sequestration such as, for example, being dissociated by RanGAP.

Summary

In conclusion, we have demonstrated that transportin is a negative regulator of nuclear membrane formation and pore assembly in the Xenopus in vitro system and that this regulatory complexity extends to spindle assembly. Focusing on nuclear pore assembly, we showed that both transportin and importin beta negatively regulate the initial step in pore assembly, ELYS binding to chromatin. RanGTP acts in an opposite manner from that of transportin to balance the effects of this new negative regulator. Both transportin and importin beta bind directly to the C-terminus of ELYS, a region critical for ELYS recruitment to chromatin. The precise choreography of regulation provided by the new regulatory karyopherin, transportin, reveals yet another level of complexity to the regulation of mitotic spindle and nuclear assembly.