Nuclear export of the yeast mRNA-binding protein Nab2 is linked to a direct interaction with Gfd1 and to Gle1 function

(Amersham Pharmacia Biotech) for 20 min at 4ÚC. The beads were washed 3x in binding buffer (20 mM Hepes, pH 6.8, 150 mM potassium acetate, 2 mM magnesium acetate, 2 mM DTT, 0.1% Tween-20), 2x with binding buffer containing 0.1 mM ATP and 2x with binding buffer containing 1 M NaCl. The beads were incubated with 1 ml yeast lysate or binding buffer alone for 2 h at 4ÚC and then washed 6x with binding buffer. Bound proteins were eluted with 100 µl of 1 M NaCl followed by boiling in SDS-sample buffer or eluted directly by boiling in SDS-sample buffer. Proteins eluted with 1 M NaCl were precipitated with trichloroacetic acid/sodium deoxycholate. Blots were probed with monoclonal anti-Nab2 antibodies (kindly provided by J. Aitchison) diluted 1:4000, rabbit polyclonal anti-Gle1 antibodies diluted 1:200 or rabbit anti-Gfd1 peptide antibodies diluted 1:200 (overnight, 4ÚC). Bound proteins were detected using peroxidase-labeled anti-mouse IgG or anti-rabbit IgG (1h, 23ÚC) and developed by enhanced chemiluminescence (Amersham Pharmacia Biotech) for anti-Nab2 and anti-Gfd1 or SuperSignal West Femto chemiluminescence (Pierce) for anti-Gle1. GST, glutathione S-transferase; hnRNP, heterogeneous ribonucleoprotein; Kap, karyopherin; MBP, maltose NES, NLS, localization NPC, nuclear pore complex; Nup, nucleoporin; PMSF, phenylmethylsulfonyl fluoride; SC, synthetic complete; TAP-tagged, tandem affinity purification tagged.


INTRODUCTION
Trafficking of molecules between the nucleus and cytoplasm proceeds through portals known as nuclear pore complexes (NPCs). NPCs are embedded in a pore formed by the fusion of the inner and outer membranes of the nuclear envelope. Yeast and vertebrate NPCs are highly conserved in architecture with a characteristic nuclear basket, 8-fold central spoke-ring structure that forms an aqueous channel, and cytoplasmic filaments (1). Proteomic analysis has revealed EXPERIMENTAL PROCEDURES Yeast Strains and Plasmids-All yeast strains used in this study are listed in Table I Polymerase chain reaction products were generated with oligonucleotides and a template containing sequences encoding GFP and the Schizosaccharomyces pombe HIS5 (pGFP-HIS5; kindly provided by J. Aitchison). The resulting DNA fragment was transformed into SWY518 using the lithium acetate method and colonies were selected on media lacking histidine. Correct integration was confirmed by immunoblotting with affinity-purified rabbit polyclonal anti-GFP antibodies (kindly provided by M. Linder). The resulting strain was back-crossed twice and the GFP tagged progeny was used in this study. The plasmids used in this study are listed in Table II and were maintained in either BL21 (Nab2-GST (18), pSW1279 and pSW1296) and or DH5α (all others). Expression vector references: pGAD-C1 and pGBD-C1 (51); pGEX-5X (Amersham Pharmacia Biotech); pMAL-cR1 (New England Biolabs).
Yeast Two-hybrid Analysis-Gal4 AD -Gfd1and Gal4 BD -Nab2 constructs were cotransformed into the two-hybrid reporter strain PJ69-4A. Transformants were selected on SC medium lacking leucine and tryptophan and assayed for interaction by growth on SC medium lacking leucine, tryptophan, histidine and adenine. Growth was scored at 30ÚC. All plasmids were tested for specificity and ability to self-activate using Gal4 AD -Snf4 and Gal4 BD -Snf1.
For the binding assay, E.coli extracts were incubated with Glutathione-Sepharose beads (Amersham Pharmacia Biotech) for 20 min at 4ÚC. The beads were washed 3x in binding buffer (20 mM Hepes, pH 6.8, 150 mM potassium acetate, 2 mM magnesium acetate, 2 mM DTT, 0.1% Tween-20), 2x with binding buffer containing 0.1 mM ATP and 2x with binding buffer containing 1 M NaCl. The beads were incubated with 1 ml yeast lysate or binding buffer alone for 2 h at 4ÚC and then washed 6x with binding buffer. Bound proteins were eluted with 100 µl of 1 M NaCl followed by boiling in SDS-sample buffer or eluted directly by boiling in SDSsample buffer. Proteins eluted with 1 M NaCl were precipitated with trichloroacetic acid/sodium deoxycholate. Blots were probed with monoclonal anti-Nab2 antibodies (kindly provided by J. Antibodies-MBP-Gle1 was expressed and purified as described previously (37). The antigen was sent to Cocalico Biologicals for production of rabbit antiserum WU851. Antiserum to Gle1 was purified by affinity chromatography over a GST-Gle1 Affi-Gel 10 column (Bio-Rad) as described previously (37). Anti-Gfd1 peptide antibodies (Gfd1D/1) were generated by Bethyl Laboratories. Bethyl Laboratories synthesized the peptides Gfd1D9R25 (CDAPDEEPIKKQKPSHKR) and Gfd1K85K101 (CKISPVSESLAINPFSQK) for production of rabbit anti-serum, and purified the antiserum against the Gfd1D9R25 peptide.
Immunoprecipitations were conducted as described previously (53). Briefly, 2µl of anti-GFP antibody was mixed with 200 µl of cell extract and 50 µl of Protein A-sepharose beads for 1.5 hours at 4ÚC. The beads were washed six times with 50 mM Tris pH 6.5, 150 mM NaCl, 0.05% Tween-20 before elution. Blots were probed with anti-Nab2 monoclonal antibodies at a 1:3000 dilution (overnight, 4ÚC) and peroxidase-labeled anti-mouse IgG (1h, 23ÚC). The blots were developed by enhanced chemiluminescence.

Purification of Protein and Soluble Binding Assay-E. coli strains containing pGEX-5X (GST)
or pSW1296 (GST-Gfd1) were grown in 1 liter of LB media containing 100µg/ml carbenicillin and 2% glucose at 37ÚC. When the OD 600 was ~1, cultures were induced for 3 hours with 0.3mM IPTG and cell pellets frozen at 70ÚC. Thawed pellets were resuspended in ice-cold lysis buffer (20 mM sodium phosphate pH7.3, 150 mM NaCl, 10 mM EDTA, 0.1 mM DTT) supplemented with complete protease inhibitor cocktail and 0.1 mM PMSF and lysed in a French Press. The suspension was centrifuged at 9000 x g for 30 min, and Triton X-100 was added to a final concentration of 1%. The supernatant was diluted in lysis buffer containing 1% Triton X-100 and incubated with Glutathione-Sepharose resin for 2 h at 4ÚC. After washing the resin with 10 ml lysis buffer containing 1% Triton X-100 and 25 ml of lysis buffer the beads were packed into a column. The column was washed with lysis buffer until the OD 280 of the flowthrough was ~0. The protein was eluted in 50 mM Tris pH 9.0, 10 mM glutathione.
MBP and MBP-Nab2 were purified as described above with the following modifications.
The lysis buffer was 10 mM sodium phosphate pH 7.2, 30 mM NaCl, 0.25% Tween-20, 10 mM EDTA, 10 mM EGTA and 10 mM β-mercaptoethanol supplemented with PMSF and complete protease inhibitor cocktail. Following lysis, NaCl was added to a final concentration of 0.5 M.
The supernatant was incubated with amylose resin (New England Biolabs) overnight at 4ÚC.
After washing with 25 ml column buffer (10 mM sodium phosphate pH 7.2, 0.5 M NaCl, 10 mM β-mercaptoethanol) containing 0.25% Tween-20) the resin was packed in a column and washed with column buffer until the OD 280 of the flow through was ~0. The bound protein was eluted with column buffer containing 10mM maltose. MBP-Gle2 (kindly provided by L. Strawn) was purified as described previously (29).
Stationary cultures grown in SC media lacking uracil (-ura) were diluted and grown overnight at 23ÚC to early log phase. Half the culture was maintained at 23ÚC and the other half shifted to 37ÚC for 1 hour. Cells were then labeled with Hoechst 33258 at a final concentration of 10 µg/mL, washed once in SC-ura and viewed by direct fluorescence microscopy. Images were collected with a 100x objective on an Olympus BX50 microscope using a Photometric CoolSNAP HQ camera (Roper Scientific) and MetaVue software.

Nab2 interacts with Gfd1 in two-hybrid and biochemical affinity assays
In a recent report of a genome-wide two-hybrid interaction analysis, an interaction between Nab2 and Gfd1 was reported (54). To test whether this two-hybrid result was potentially physiologically significant, we further analyzed the specificity of the two-hybrid interaction. In-frame fusions were generated for Gfd1 to the transcriptional activation domain of Gal4 (Gal4 AD ), and for Nab2 to the DNA binding domain of Gal4 (Gal4 BD ). These plasmids were co-transformed into reporter strains and the presence of an interaction was assayed by growth on media lacking histidine and adenine. As controls for non-specific interactions, we used fusions of the transcription factors Snf1 to Gal4 BD and Snf4 to Gal4 AD . As shown in Figure 1A, a two-hybrid interaction was observed specifically between Nab2 and Gfd1, and not between Nab2 and Snf4 or Gfd1 and Snf1.
Next, we used an affinity chromatography assay to determine whether a physical interaction between Nab2 and Gfd1 could be detected in vitro. Gfd1 was expressed in bacteria as a fusion to glutathione-S-transferase (GST) and bound to glutathione sepharose beads. The immobilized GST-Gfd1 was then incubated with buffer alone or total yeast lysate prepared under non-denaturing conditions. Bound proteins were sequentially eluted, first with high salt (1M NaCl) followed by boiling in SDS buffer. Eluted fractions were separated by SDS-PAGE and analyzed by immunoblotting with monoclonal anti-Nab2 antibodies. Nab2 in the yeast steady state in wild-type cells, Nab2-GFP localizes to the nucleus whereas Nab2RGG-GFP localizes throughout the cell (Fig. 5A and 6A). Therefore, the localization of Nab2RGG-GFP is a useful tool to study Nab2 export. Wild type and gfd1 cells were grown at 23ÚC and shifted to 37ÚC for 1 hour or 3 hours (data not shown). To visualize the nuclei, cells were labeled with Hoechst dye after the temperature shift. In gfd1 cells, the localization of Nab2-GFP and Nab2RGG-GFP was not perturbed at 23ÚC or after incubation at 37ÚC. Thus, Gfd1 was not required for efficient Nab2 export correlating with the finding that gfd1 cells do not demonstrate a detectable defect in viability or polyA + RNA export (40,48).
Since Gfd1 physically associates with Gle1 (48), we speculated that Gfd1 could facilitate an interaction between Nab2 and Gle1. To determine whether Gle1 was also present in the Nab2-Gfd1 complex, we performed an affinity chromatography assay. Immobilized recombinant Nab2-GST or GST alone was incubated with yeast lysate from cells expressing Gle1-TAP-tagged. As shown in Figure 7, Gle1-TAP-tagged and Gfd1 were co-isolated in the bound fraction with Nab2-GST (lane 2), but not with GST alone (lane 3). Interestingly, a

DISCUSSION
To delineate the mechanism underlying mRNA transport through the NPC, it is essential to understand the interactions between shuttling hnRNPs and mRNA export factors. In this study, we report that the shuttling hnRNP Nab2 forms a complex with Gfd1. A role for Gfd1 in mRNA export has been previously implicated due to its interactions with Nup42 and the mRNA export factors Gle1 and Dbp5 (40,48). We have assembled a series of biochemical data documenting direct Nab2-Gfd1 binding. This includes yeast two hybrid assays, isolation of endogenous Nab2 using recombinant GST-Gfd1, Nab2 isolation by coimmunoprecipitation with Gfd1-GFP from yeast cell lysates, and soluble binding assays using purified recombinant Nab2 and Gfd1. Since previous in vivo and in vitro evidence has documented an interaction between Gfd1 and Gle1 (40, 48), Gfd1 could link Nab2 export to Gle1 function. Consistent with this conclusion, Nab2∆RGG export is blocked at the restrictive temperature in a gle1-4 mutant. Moreover, endogenous Gfd1 and Gle1-TAP-tagged are co-isolated using recombinant Nab2-GST. These results highlight a connection between Nab2 and Gle1-mediated mRNA export.
To date, Kap104 and Mlp1 are the only reported protein binding partners for Nab2 (18,26). Incorporating our findings with this work, a model for Nab2 shuttling is presented in Figure 8. Import is mediated by direct binding of Kap104 to the RGG domain of Nab2, and Nab2 accumulates in the cytoplasm of kap104-16 mutant cells (18,19). Since Nab2RGG enters the nucleus, it is possible that other factors may also mediate efficient import (23). However, a role for Kap104 in Nab2RGG import has not been formally excluded. The FG Nups on both faces of the NPC likely represent the main docking sites for Kap104-mediated import. Nab2 import is perturbed when FG domains in specific Nups on both NPC faces are deleted (55). In the nucleus, the dissociation of Nab2 from Kap104 requires the dual interaction of Kap104 with RanGTP and Nab2 with RNA (18).
For the export of Nab2, one of the first steps at the nuclear NPC face is predicted to be an interaction with Mlp1 (26). However, the Nab2 domain required for Mpl1 interaction has not yet been defined. Our results indicate that a step in the Nab2 export pathway also requires Gle1 function. Based on the lack of a two-hybrid interaction between Nab2 and Gle1 (data not shown), we propose that Gfd1 serves as a bridging factor between Nab2 and Gle1. We found that the Gfd1 binding region in the N-terminal Nab2 domain is distinct from the Nab2 RNA binding domains. Thus, Nab2 may bind mRNA and Gfd1 simultaneously at some point during the export pathway. As shown in Figure 8, reported binding partners to Gle1 are Nup42, Nup159, Dbp5 and Gfd1 (40,48). The DEAD-box RNA helicase Dbp5 also interacts with Nup159 and Gfd1, and Nup42 interacts with Gfd1 (40). Since Nup42 and Nup159 localize to the cytoplasmic NPC face (2), the Gle1-Gfd1-Nab2 complex may form at a terminal step in export. However, a Gle1-ProtA fusion protein has been localized to both faces of the NPC (2), and hGle1 is known to shuttle between the nucleus and cytoplasm (49). Therefore, the Nab2-Gfd1 complex could interact with Gle1 at both NPC faces. Once in the cytoplasm, Nab2 will be released from the mRNA upon binding to Kap104, and then imported into the nucleus for further rounds of transport.
Throughout the mRNA export pathway and as shown in part for Nab2 shuttling in Figure   8, a number of individual protein-protein interactions have been documented (reviewed in 4, 13). However, both the formation of an export-competent hnRNP and the NPC translocation mechanism are likely based on combinatorial and/or overlapping protein-protein interaction networks. For example, we predict additional factors may bind to the N-terminal domain of Nab2 and compensate for the absence of Gfd1. The growth and mRNA export defects observed in nab2N mutant cells are not present in gfd1 mutant cells (40,48), and we found that Nab2 export is not impaired in gfd1 mutant cells. We speculate that Gfd1 and the other functionally redundant proteins serve as bridging factors between Nab2 and Gle1, other mRNA export factors, or Nups. We have also previously shown that inositol hexakisphosphate (IP 6 ) production is required for efficient Gle1-mediated mRNA export (56). Any of these steps along the Gle1 pathway may be modulated by IP 6 .
It is uncertain whether yeast and human Gle1 utilize similar mechanistic pathways to mediate mRNA export. The ability of yeast-human Gle1 chimera proteins to complement gle1 mutant cells suggests that the pathways may be functionally conserved (56). Four putative functional domains have been identified in hGle1 (49, 57, 58). Both the N and C-terminal domains are required for NPC localization (49, 58). In addition, there is a coiled-coil domain and a region required for nucleocytoplasmic shuttling (49). Thus far, the only binding partner identified for hGle1 is the NPC-localized hNup155 (58). If the yeast and human pathways are analogous, adaptor proteins similar to Gfd1 could facilitate interactions between hGle1 and shuttling hnRNPs in vertebrate cells.
It has been the long-standing view in the field that Mex67/Tap/NXF1 serves as the transporter for exporting hnRNPs (13). This is based on Mex67/Tap/NXF1 nucleocytoplasmic shuttling and essential interactions with both the RNA-bound Yra1 and nucleoporins. The studies in this paper, combined with previous work, have now linked Gle1 both with an RNAbound protein (Nab2) and nucleoporins (36, 40,46,47,48,58), and shown Gle1 shuttling or having access to both NPC faces (2,49). Thus, Gle1 is also positioned to also play an active role in the translocation mechanism. However, Gle1 and Mex67 clearly function differently in the mRNA export pathway. In cells overexpressing the Nup116 GLFG region, Mex67-GFP accumulates in the nucleus while Gle1-GFP remains predominantly localized at NPCs (29).
Moreover, they interact with distinct domains in Nup42; Mex67 binding the N-terminal FG repeat region of Nup42 and Gle1 exclusively the C-terminal non-FG domain (28,48). We have also not observed any synthetic lethal genetic interactions between gle1 and mex67 mutants (S.    Extracts from wild-type or Gfd1-GFP cells were immunoprecipitated with anti-GFP antibodies.

FIGURE LEGENDS
Bound proteins were separated by SDS-PAGE and immunoblotted with anti-Nab2 antibodies.
Proteins were detected using anti-mouse IgG coupled to horseradish peroxidase. Samples loaded represent 1/10 th of the total input compared to the total bound fraction.

Figure 3: Nab2 and Gfd1 interact directly in vitro.
Purified GST or GST-Gfd1 immobilized on glutathione-sepharose beads was incubated with   Wild-type, gfd1 and gle1-4 mutant cells harboring a Nab2-GFP plasmid were grown at 23ÚC in synthetic media lacking uracil to log phase. Cells were maintained at 23ÚC (data not shown) or shifted to 37ÚC for 1 hour before visualization by direct fluorescence microscopy.
Corresponding Hoechst staining (middle) and Nomarski (right) are shown. Wild-type, gfd1 and gle1-4 mutant cells harboring a Nab2RGG-GFP plasmid were grown at 23ÚC in synthetic media lacking uracil to early log phase. Cells were maintained at 23ÚC or shifted to 37ÚC for 1 hour before visualization by direct fluorescence microscopy.

Figure 7: Co-isolation of endogenous Gfd1 and Gle1-TAP-tagged with Nab2-GST
Nab2-GST or GST immobilized on glutathione-sepharose beads was incubated with Gle1-TAP-tagged yeast cell extract. Bound proteins were eluted with SDS-sample buffer, and analyzed by SDS-PAGE and immunoblotting with anti-Gle1 or anti-Gfd1 antibodies. Proteins were detected using anti-rabbit IgG coupled to horseradish peroxidase. Samples loaded represent 1/25 th of the total input compared to the total bound fraction. Nab2 import is mediated by Kap104 binding to the RGG domain and is also dependent on FG containing Nups (18,19,55). These FG Nups are localized on both faces of the NPC and could represent docking sites for Kap104. In the nucleus, Nab2 binds RNA during transcription and then shuttles to the cytoplasm transporting its cargo mRNA (20,21). One of the first steps in its exit path is predicted to be binding with Mlp1 that localizes to the nuclear face of the NPC (26).
The N-terminal domain of Nab2 is required for its export (23), and we report Gfd1 as a binding partner to this domain. Since Gfd1 and Gle1 interact (48) and Gle1 is present in Nab2-Gfd1 complexes, we propose that Nab2 could interact with Gle1 through Gfd1 to facilitate a terminal step in export.