INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Newly synthesized precursor RNAs become associated with hnRNP proteins and components of the splicing and 3'-end-processing machinery to undergo a series of well-coordinated processing steps resulting in the formation of export-competent mRNP complexes. As a number of hnRNP proteins remain associated with the mRNA during its translocation through the nuclear pore complex (NPC), they were previously proposed to contain signals recognized by specific mRNA export receptors (8, 11, 25, 43).

The Saccharomyces cerevisiae Mex67p-Mtr2p heterodimer and its human orthologue TAP-p15 play a central role in mRNA export and represent so far the best-characterized mRNA export receptors (12, 16, 17, 33). Indeed, conditional mutations in Mex67p induce a rapid nuclear accumulation of poly(A)⁺ RNA as well as of specific transcripts, consistent with an essential role in mRNA export (13, 36). The human TAP protein was identified as the cellular factor interacting with the constitutive transport element (CTE) present in RNAs from simple retroviruses (12). Because the injection of CTE RNA into Xenopus oocytes competes with the export of cellular mRNAs, TAP was also described as a component of the normal mRNA export pathway (26, 31). Mex67p and TAP present the main features of mRNA export receptors, or nuclear export factors, since both shuttle between the nucleus and the cytoplasm, cross-link to poly(A)⁺ RNA in vivo, and directly interact with FG-nucleoporins at the NPC (2, 4, 16, 17, 39).

TAP binds the viral CTE very strongly and specifically (5). In contrast, the interaction of TAP and Mex67p with cellular RNAs is nonspecific and of low affinity, suggesting that these export factors associate with poly(A)⁺ RNA through protein-protein interactions rather than direct binding to RNA (2, 5, 17, 33). Consistently, Yra1p, an hnRNP-like protein which directly interacts with Mex67p, was identified in a genetic screen for mutations synthetically lethal with a MEX67 mutation as well as by affinity purification with tagged Mex67p from yeast extracts (40, 42). Yra1p had initially been purified as a protein with strong RNA-RNA annealing activity, but it is still unclear whether this activity is relevant to Yra1p function in vivo (27). The interaction between Mex67 and Yra1p is direct and occurs in the presence or absence of RNA. Yra1p belongs to an evolutionarily conserved family of hnRNP-like proteins, called REF (RNA and export factor binding proteins), which has more than one member in several species. The REF proteins exhibit nonspecific affinity for RNA, and they are all characterized by a central RNP-motif RNA-binding domain (RBD) and two highly conserved N- and C-terminal boxes (N box and C box). These conserved regions are separated by two more variable domains (N-vr and C-vr) (42). In higher eucaryotes, the variable regions are rich in glycines and asparagines, reminiscent of the RGG domain of certain RNA binding proteins (7). Consistent with a direct role of Yra1p in mRNA export, both Yra1p depletion and analysis of a YRA1 conditional mutant under nonpermissive conditions result in nuclear accumulation of poly(A)⁺ RNA. Mammalian REF proteins interact with TAP, and Aly, one of the mouse REF proteins, also called REF1-I, is able to rescue a YRA1 gene disruption in yeast, indicating that the function of the REF proteins in mRNA export has been conserved (40). For these reasons, REF proteins have been proposed to participate in mRNA export by facilitating the recruitment of Mex67p/TAP to cellular mRNPs (40, 42).

Yra1p was found earlier in a screen for genes causing overexpression-mediated cell growth arrest (9). Similarly, Mlo3, the Schizosaccharomyces pombe homologue of Yra1p, was also isolated as a gene causing cell cycle arrest when overexpressed (15). The mouse homologue Aly was identified in a yeast two-hybrid screen as a protein interacting with LEF1, a transcription factor regulating the activity of the T-cell receptor alpha enhancer. In that study, Aly was proposed to facilitate the functional collaboration of multiple proteins and their assembly into a complex (6). Aly was also described previously as a chaperone protein regulating the activity of transcription factors containing a leucine zipper (44). These latter data suggest that REF proteins may function at multiple steps in mRNA biogenesis and expression.

In this study, we pursue the functional characterization of Yra1p in mRNA export. We describe a nonessential homologue of Yra1p, Yra2p, which functionally overlaps with Yra1p. The phenotypic and biochemical analyses of Yra1p deletions show that the RBD is dispensable for viability as well as RNA binding and indicate that the C-terminal and N-terminal domains of the protein are functionally overlapping and redundant for both RNA and Mex67p binding. Yra1 mutants compromised in Mex67p binding have parallel defects in poly(A)⁺ RNA export. As lower amounts of Mex67p are associated with mRNP complexes in these Yra1 mutant strains, the data support the view that Mex67p is recruited to the mRNP through an interaction with Yra1p.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids and constructs. The yeast plasmids used in this study are summarized in Table 1. The Myc-tagged YRA2 gene construct expresses an N-terminally tagged protein and was obtained in two PCR amplification steps. First, 5' and 3' flanking sequences (500 bp) were amplified as BamHI-SalI and SalI-BamHI fragments, respectively. The 3' primer for the 5' flanking fragment was designed so that the SalI site is preceded by the ATG codon followed by a Myc-tag sequence. The 5' primer for the 3' flanking fragment contained a SalI site just upstream of the YRA2 stop codon. The 5' and 3' flanking fragments were digested with SalI, ligated, and amplified again with the external 5' and 3' primers. The final PCR product was digested with BamHI and cloned into YCpLac22-SalI cut with BamHI from which the unique SalI cloning site had been eliminated; this generated the cassette construct Myc-YRA2 ± 500 (pFS2131). The YRA2 coding sequence was amplified from genomic DNA as a SalI fragment and cloned in frame into the Myc-YRA2 ± 500 cassette linearized with SalI to generate YCpLac22-Myc-YRA2 (pFS2262). For Myc-Yra2p overexpression, the BamHI insert of pFS2262 was subcloned into the high-copy-number plasmid YEpLac112 (TRP1, 2µm) to generate YEpLac112-Myc-YRA2 (pFS2261). The same strategy was used to generate the cassette construct YCpLac22-HA-YRA1 ± 500 (pFS2128) and HA-YRA2 ± 500 (pFS2343). The construct expressing hemagglutinin (HA)-tagged Yra2p was obtained by subcloning the YRA2 genomic SalI fragment from pFS2262 into pFS2343 cut with SalI to generate pFS2345. The constructs expressing wild-type HA-YRA1 (pFS2146) or HA-YRA1 truncations (14-227, 1-210, 14-210, 77-227, 1-167, and 14-167; pFS2151, pFS2154, pFS2155, pFS2156, pFS2145, and pFS2148, respectively) were obtained by PCR amplification of the corresponding sequences as SalI fragments on the YRA1 cDNA and cloning into pFS2128 cut with SalI. The HA-RBD construct was obtained by ligating a 5' BamHI-XhoI PCR fragment containing 500-bp YRA1 5' flanking sequences and codons 1 to 77 (with the HA tag) to a 3' XhoI-BamHI fragment containing codons 167 to 227 and 500-bp 3' flanking sequences. The ligated products were PCR amplified with the BamHI external primers and cloned into YCpLac22 cut with BamHI to generate pFS2268. Construct pHA-YRA1+int (pFS2233) was obtained by cloning the YRA1 genomic coding sequence as a SalI PCR fragment into pFS2128 linearized with SalI.

Yeast strains. Yeast strains used in this study are summarized in Table 2. The YRA1 shuffle strain yra1::HIS3 <pURA3-YRA1> (FSY1026) in the W303 background (MATa ade2 his3 leu2 trp1 ura3) was described earlier (42). The shuffle strain yra1::HIS3/yra2::KAN^r <pURA3-YRA1> (FSY1135) was obtained by backcrossing the yra2::KAN^r Euroscarf deletion strain (BY4742, mat his31 leu21 lys20 ura30 YKL214c::KanMX4; accession no. Y15064) three times with the YRA1 shuffle strain FSY1026. Strains expressing functional Yra1 truncations were obtained by transformation of the YRA1 shuffle strain with wild-type or mutant YCpLac22-HA-YRA1 plasmid followed by selection against the pURA3-YRA1 plasmids on 5-fluoroorotic acid (5-FOA) plates. The strains used to prepare extracts for pull-down assays were constructed as follows. The strain expressing Gle2p with a C-terminal HA tag (GLE2-HA, FSY1121) was obtained by genomically tagging GLE2 with an HA epitope through insertion of a DNA cassette at the 3' end of the gene in a haploid W303 strain using KAN^r as a selectable marker (22). Strains expressing GFP-tagged Mex67p and Pab1p (MEX67-GFP and PAB1-GFP, FSY1113 and FSY978, respectively) were obtained by chromosomal insertion of a DNA cassette at the 3' end of the genes using HIS3 as a selectable marker in the W303 wild-type strain for MEX67 and KAN^r as a selectable marker in the protease-deficient strain BJ2168 for PAB1. Strains expressing GFP-Yra2, GFP-Yra1, GFP-Yra1 deletion mutants, or GFP-Npl3 were obtained by transforming W303 with pGAL-GFP-YRA2 (pFS1941), pGAL-GFP-YRA1 (pFS1915), pGAL-GFP-YRA1 deletion mutants (see above, pPS808 constructs), and pGAL-GFP-NPL3 (20). Cells transformed with these plasmids were grown to mid-log phase in Ura medium containing 2% glucose, washed, induced for 4 to 5 h in selective medium containing 3% galactose-1% raffinose, and examined under the microscope or processed for extract preparation.

Expression and purification of recombinant proteins. Yra1-GST and Yra2-GST fusions, ProtA-Pse1-6xHis, and 6xHis-RanQ69L were expressed in the Escherichia coli M15 pRep4 strain. Cultures were grown to an optical density at 600 nm (OD₆₀₀) of 0.8 and induced with 0.5 mM isopropyl--D-thiogalactopyranoside (IPTG) for 4 h at 37°C. Cells were collected by centrifugation, and pellets were resuspended in lysis buffer (150 mM NaCl, 20 mM HEPES-HCl [pH 8.0], 10% glycerol) plus 1 tablet of complete EDTA-free protease inhibitor cocktail (Roche Diagnostics) per 40 ml and lysed with a one-shot cell disrupter (Constant System Ltd.) set to 1.75 × 10⁵ kPa (1.75 kilobars). Cell debris were removed by centrifugation for 20 min at 12,000 × g, and GST fusions were affinity purified on glutathione-agarose beads (Pharmacia) as described by the manufacturer. For the RNA binding assays, Yra1- and Yra2-GST fusions were purified in lysis buffer containing 500 mM NaCl, eluted from beads in 1× phosphate-buffered saline-10 mM glutathione-10% glycerol, and concentrated with a Centricon 30 (Amicon) concentrator. GST-Mex67 was expressed in BL21(DE3) cells. Induction was performed at 16°C for 48 h; cells were lysed in 100 mM KOAc-10 mM HEPES (pH 7)-2 mM MgOAc-10% glycerol and purified as described above. RanQ69L-GTP and ProtA-Pse1-6xHis were prepared as described previously (10).

In vitro binding with recombinant Pse1p. The E. coli lysate containing ProtA-Pse1-6xHis was adjusted to 0.1% Triton X-100 and incubated with different GST fusions immobilized on beads in the presence or absence of 1 mM GTP and ~10 µg of RanQ69L loaded with GTP (Pharmacia). Each binding reaction mixture contained 200 µl of E. coli lysate and 5 µg of GST fusion in a final volume of 300 µl. Binding reactions were performed for 1 h at 4°C on a rotating wheel; beads were washed three times with 500 µl of lysis buffer plus 0.1% Triton X-100, resuspended in 2× sample buffer, and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by Coomassie blue staining or Western blot analysis.

GST pull-down assays with yeast extracts. Yeast lysates for pull-down assays were prepared from cells grown in selective or yeast extract-peptone-dextrose (YEPD) medium at 30°C to an OD₆₀₀ of 0.8 to 1.0. When galactose inductions were needed (extracts expressing GFP-Yra1, GFP-Yra2, and GFP-Npl3), cells were grown overnight in selective medium containing 2% glucose, washed, and induced for 5 h in selective medium containing 3% galactose and 1% raffinose. Cell pellets corresponding to 80 OD₆₀₀ units of cells were resuspended in 1 ml of lysis buffer (150 mM NaCl, 20 mM HEPES-KOH [pH 7.6], 5 mM MgCl₂, 10% glycerol) containing complete EDTA-free protease inhibitor cocktail and lysed with a one-shot cell disrupter set to 2 × 10⁵ kPa (2 kilobars). Lysates were cleared by centrifugation at 12,000 × g for 20 min. Extract corresponding to 4 OD₆₀₀ units of cells (50 µl) was incubated with 5 µg of GST-tagged recombinant protein immobilized on 20 µl of packed glutathione agarose beads in a final volume of 200 µl of lysis buffer containing 0.1% Triton X-100 for 90 min at 4°C on a turning wheel. Beads were then washed three times with 300 µl of lysis buffer plus 0.1% Triton X-100, resuspended in 30 µl of 2× sample buffer, boiled, and analyzed by SDS-PAGE and Western blotting. RNase-treated extracts were obtained by incubating 100 µl of extract (8 OD₆₀₀ units of cells) with 15 µg of RNase A for 20 min at 25°C prior to the binding experiment.

Affinity purification of HA-Yra1p and HA-Mex67p with ProtA-Pab1p expressed in yeast. Strain FSY1246 shuffled with wild-type or mutant YRA1 constructs on LEU2/CEN was grown in YEPD-2% glucose overnight at 30°C to an OD₆₀₀ of 1. Cells were spun, resuspended in 1 ml of lysis buffer (150 mM NaCl, 20 mM HEPES-KOH [pH 7.6], 5 mM MgCl₂, 10% glycerol, 0.1% Triton X-100, protease inhibitor mix) per 100 OD₆₀₀ units of cells, and lysed by vortexing in the presence of glass beads. RNase A treatment was done as described above. Binding reactions were performed in a final volume of 300 µl containing 150 µl of extract, 50 µl of 50% immunoglobulin G (IgG)-Sepharose (Pharmacia), and 150 µl of lysis buffer on a turning wheel for 90 min at 4°C. Beads were washed four times with 400 µl of lysis buffer for 5 min. Bound proteins were eluted from the beads by addition of 200 µl of 2 M KCl-20 mM HEPES-KOH, pH 7.9, for 15 min at 4°C, followed by precipitation with trichloroacetic acid. The eluted samples were analyzed by SDS-PAGE and Western blotting with anti-HA and anti-Npl3p antibodies.

Western blot analysis. Western blot analyses were performed according to standard procedures. The anti-Npl3p and the anti-GFP antibodies, a kind gift of Pam Silver (34), the anti-Gle1p, a kind gift from Laura Davis, and the anti-HA antibody (Roche Diagnostics) were used at a 1:2,000 dilution. The protein signals were revealed with Super Signal West Pico or West Femto Chemiluminescent substrate (Pierce) whenever stated in the figure legends.

Electrophoretic mobility retardation assay. An 80-nucleotide RNA probe was prepared by in vitro transcription with T3 RNA polymerase (Promega) using pBluescript KS(+) digested with SmaI as the template. Reactions were performed in binding buffer (15 mM HEPES [pH 7.9], 100 mM KCl, 0.2 mM EDTA, 5 mM MgCl₂, 1 mM dithiothreitol, 0.05% bovine serum albumin, 0.05% NP-40, 10% glycerol) in a final volume of 10 µl containing 1 µl of GST fusion eluted from beads (50 or 500 ng/µl) and the RNA probe (10,000 cpm). After 20 min at room temperature, samples were separated on a 5% native polyacrylamide gel. Electrophoresis was carried out at a constant voltage of 120 V at 4°C in 0.5× Tris-borate-EDTA buffer. Protein-RNA complexes were visualized by autoradiography.

In situ hybridization. Shuffled yeast strains (yra1::HIS3 <pTRP1-HA-YRA1>) expressing wild-type or truncated versions of HA-Yra1p were grown in YEPD-2% glucose to mid-log phase at 25°C. Cells were shifted to 37°C for 1 h or kept at 25°C before being fixed and processed for in situ hybridization with a digoxigenin-labeled oligo(dT) probe and fluorescently labeled antidigoxigenin antibodies as described earlier (38).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Yra2p, a nonessential homologue of Yra1p, rescues a YRA1 deletion. Yra1p belongs to the evolutionarily conserved REF family of hnRNP-like proteins. At least two REF members have been identified in Caenorhabditis elegans, Xenopus laevis, mouse, human, S. pombe, and S. cerevisiae (40, 42). Yra2p (YKL214c) corresponds to the second protein in yeast sharing the same domain organization as members of the REF family. Yra2p shows 58% overall homology to Yra1p (19% identity and 39% similarity), and most conserved residues are located within the N and C boxes and the RBD (data not shown).

View larger version (48K): [in this window] [in a new window]   FIG. 1.   (A) Overexpression of Yra2p complements the YRA1 deletion. The YRA1 shuffle strain (yra1::HIS3 <pURA3-YRA1; pFS1876>) was transformed with an empty TRP1 vector (YCpLac22), and constructs carrying an N-terminally Myc-tagged YRA2 gene on a centromeric (Myc-YRA2; pFS2262) or high-copy-number (Myc-YRA2, 2µm; pFS2262) plasmid and N-terminally HA-tagged YRA1 cDNA (HA-YRA1; pFS2146) or gene plus intron (HA-YRA1+int; pFS2233). Both YRA1 and YRA2 genes were expressed from their own promoters. Complementation of the YRA1 deletion by these constructs was assessed by growing the transformants for 5 days at 30°C on medium containing 5-FOA. (B) Growth phenotypes of FOA+ shuffled strains described for panel A. The indicated strains were grown in rich medium, and 10-fold serial dilutions were spotted on YEPD-2% glucose at 25 and 37°C. (C) Yra2p is 10-fold less abundant than Yra1p. Shown are the results of Western blot analysis of the yra1::HIS3 knockout strain containing low-copy-number HA-YRA1 cDNA (pFS2146; lane 1) or HA-YRA1 gene (pFS2233; lane 2) plasmids and a yra2::KANr knockout strain containing a low-copy-number HA-YRA2 plasmid (pFS2345; lane 5). Samples in lanes 3 and 4 are 10-fold dilutions of the protein extracts loaded in lanes 1 and 2. The blot was probed with anti-HA antibody and anti-Gle1p as an internal control for protein loading. The protein signals were revealed with Super Signal West Femto Chemiluminescent substrate.

Yra1p, Yra2p, and Mex67p associate with mRNPs in vitro. The proposed roles of Yra1p and Mex67p in mRNA export imply that these two proteins associate with mRNP complexes. To initiate the biochemical characterization of these complexes, the interaction of Yra1/2p and Mex67p with mRNPs was indirectly tested by incubating these proteins as E. coli GST fusions with total yeast extracts and testing their ability to select defined components of mRNP complexes. As these interactions may be indirect and mediated by RNA, they were examined in untreated extracts or extracts subjected to a preliminary RNase treatment (Fig. 2).

View larger version (60K): [in this window] [in a new window]   FIG. 2.   GST-Yra1, GST-Yra2, and GST-Mex67 associate with mRNPs in total extracts. Whole-cell extracts of strains expressing different RNA binding proteins as HA or GFP fusions were incubated with GST-Yra1, GST-Yra2, GST-Mex67 fusions, or GST alone immobilized on beads. Extracts were first incubated with (+) or without () RNase for 20 min at 25°C. Bound proteins were fractionated by SDS-PAGE followed by Western analysis using anti-GFP (34) or anti-HA antibodies (lanes 3 to 10). A fraction (1/20) of the total extracts (with or without RNase A treatment) was analyzed in parallel to control for input protein levels (lanes 1 and 2). The protein signals were revealed by normal enhanced chemiluminescence.

The growth phenotypes of Yra1p truncations indicate a functional redundancy between the N- and C-variable regions. The conserved domain organization of Yra1p, Yra2p, and other members of the REF family indicates a high evolutionary pressure on the function and overall architecture of these proteins. To define the importance of the Yra1p conserved domains in cell growth, various N- and C-terminal deletions were generated within the HA-YRA1 cDNA and the constructs were tested for their ability to rescue a YRA1 deletion (Fig. 3A). Since Yra2p functionally overlaps with Yra1p (Fig. 1A), the phenotypes of the Yra1 mutants were tested in the presence or absence of Yra2p. The HA-YRA1 mutant plasmids were transformed into the YRA1 shuffle strain (yra1::HIS3 <pURA3 YRA1>) as well as in a YRA1 shuffle strain from which YRA2 had been deleted (yra1::HIS3 yra2::KAN^r <pURA3-YRA1>). The Yra1 mutant proteins were examined for their ability to rescue the YRA1 deletion in either genetic background by streaking the transformants on 5-FOA-containing medium (Fig. 3A). The growth phenotypes induced by the Yra1 truncations were subsequently examined by spotting serial dilutions of the FOA⁺ strains at 25 or 37°C (Fig. 3B).

View larger version (46K): [in this window] [in a new window]   FIG. 3.   (A) Phenotypic analysis of Yra1p deletion mutants in the presence or absence of Yra2p. The YRA1 and YRA1/YRA2 shuffle strains (yra1::HIS3 <pURA3-YRA1; pFS1876> and yra1::HIS3/yra2::KANr <pURA3-YRA1; pFS1876>) were transformed with constructs coding for the indicated Yra1p truncations. Transformants were grown on 5-FOA plates at 25°C for 5 days (bottom). Growth on 5-FOA indicates complementation of the YRA1 deletion. Duplicates were grown at 37°C (data not shown), and growth phenotypes at 25 and 37°C are summarized at the top. Yra1 truncations indicated by an asterisk have a dominant negative phenotype when expressed in a wild-type strain background. (B) Growth of Yra1 wild-type (WT) and mutant strains at 25 and 37°C. The yra::HIS3 shuffled strains (FOA+) expressing wild-type or mutant Yra1p from a plasmid (TRP1/CEN) were grown in rich medium, spotted as 10-fold serial dilutions on YEPD-2% glucose, and grown for 3 days at 25 or 37°C. (C) Wild-type and mutant HA-Yra1p protein levels expressed in the presence (top) or absence (bottom) of the wild-type pURA3-YRA1 gene. The YRA1 shuffle strain (yra1::HIS3 <pURA3-YRA1>) was transformed with low-copy-number plasmids (TRP1/CEN) expressing HA-tagged wild-type or truncated Yra1p proteins as indicated. Total protein extracts were prepared before (top) or after (bottom) elimination of pURA3-YRA1 on 5-FOA and examined by Western blotting with an anti-HA antibody. The blots were simultaneously incubated with an anti-Gle1p antibody as an internal control for protein loading. The strains expressing Yra1 mutants 14-210 and 14-167 are not viable on 5-FOA. (D) CLUSTALW alignment of the N- and C-terminal regions of Yra1p (N-box+N-vr, aa 1 to 77; C-vr+C-box, aa 167 to 227) reveals a homology of about 50%. The sequences corresponding to the conserved boxes are indicated.

Yra1p contains an NLS in its N-vr region and is imported by the importin--like factor Pse1p. Yra1p is located within the nucleus at steady state. To map the nuclear localization signal (NLS) in Yra1p, various portions of the protein were expressed as GFP fusions in the yeast wild-type strain W303 and examined for localization (Fig. 4A). A fusion containing the N-terminal region of Yra1p (aa 1 to 77) efficiently targeted GFP into the nucleus (panel m). The deletion of the N box from full-length Yra1p (14-227) did not interfere with its nuclear localization (panel b), indicating that the NLS lies within the N-vr region of the protein. Consistently, all the truncations containing the N-vr appeared nuclear (panels c, d, k, and l). Notably, the GFP fusions missing either one or both highly conserved boxes (14-227, 1-210, and 14-210) had the tendency to form aggregates within the nucleus or at the nuclear periphery (panels b, c, and d; see Discussion). GFP fusion 77-227, which lacks the N-vr region, exhibited weak nuclear accumulation as well as cytoplasmic staining consistent with a nuclear import defect (panel e). The nuclear fraction of this mutant protein may result from passive diffusion and is sufficient for function since the Yra1 77-227 truncation is perfectly viable (Fig. 3A). Finally, GFP fusions containing the RBD (77-167) or the C-terminal region (167-227) exhibited a homogenous distribution throughout the cell presumably due to free diffusion of these small proteins between the nuclear and cytoplasmic compartments (panels n and o). It cannot be excluded that the functional truncation 77-227 contains a second weaker NLS. This NLS could be revealed by fusing the C-terminal region of Yra1p to multiple GFP moieties to generate a chimeric protein that is too large to cross the nuclear envelope by diffusion. In the absence of an additional NLS, the fusion protein is expected to be restricted to the cytoplasm and unable to rescue a YRA1 deletion.

View larger version (48K): [in this window] [in a new window]   FIG. 4.   (A) The NLS of Yra1p maps within the N-vr region. Wild-type (wt) yeast cells were transformed with plasmids expressing full-length or truncated versions of Yra1p fused to GFP from a galactose-inducible promoter. Transformants were grown at 30°C to mid-log phase in selective medium containing glucose, washed, induced for 4 h in selective medium containing galactose, and examined for GFP fluorescence signal under a Zeiss Axioplan microscope equipped with a 100× objective lens and a cooled charge-coupled device camera. Cells were photographed for GFP fluorescence (panels a to e and k to o) and by Nomarski optics (panels f to j and p to t). The localization of each GFP fusion is indicated as N (nuclear) or N+C (nuclear and cytoplasmic) below each set of pictures. (B) The importin--like transport factor Pse1p mediates the nuclear import of Yra1-GFP and Yra2-GFP fusions. Plasmids expressing GFP fused to the N terminus of Yra1p (GFP-Yra1-1-77 containing the NLS) or Yra2p (GFP-Yra2p) from a galactose-inducible promoter were transformed into a wild-type strain (WT) or a strain containing a conditional mutation in PSE1 and a deletion of KAP123 (pse1-1 kap123), two genes encoding functionally related importin--like import factors. Transformants were grown to mid-log phase in selective medium plus glucose, washed, induced for 4 h in medium containing 3% galactose-1% raffinose, and shifted to 37°C for 1 h. Cells were then photographed for GFP fluorescence (left) and by Nomarski optics (right). (C) Pse1p directly interacts with Yra1p and Yra2p in vitro. The indicated Yra1-GST and Yra2-GST fusions or GST alone immobilized on glutathione beads was incubated with an E. coli lysate containing the ProtA-Pse1-6xHis fusion protein in the presence or absence of RanQ69L-GTP. After washing, proteins bound to the beads were fractionated by SDS-PAGE and either subjected to Western blot analysis with an anti-ProtA antibody to detect ProtA-Pse1-6xHis bound to the beads (top) or Coomassie blue stained to visualize the GST fusions present in the binding reactions (bottom). Full-length GST fusions are indicated by an asterisk.

The N-vr and C-vr regions of Yra1p mediate binding to Mex67p. Yra1p directly interacts with Mex67p (40, 42). To investigate which region of Yra1p binds Mex67p, different Yra1p deletion mutants fused to GFP were expressed in yeast and whole-cell extracts were used in pull-down assays with recombinant GST-Mex67 (Fig. 5A). As already shown in Fig. 2, GFP-Yra1 as well as GFP-Yra2 binds to GST-Mex67 (lanes 1 and 2). Deletion of the C-terminal box of Yra1p (fusion 1-210) did not affect binding to GST-Mex67 (lane 4); however, deletion of the N-terminal box (fusion 14-227) substantially reduced binding, indicating that this conserved sequence contributes to the interaction with Mex67p (lane 3). Deletion of both conserved boxes (fusion 14-210) did not affect binding efficiency any more than did deletion of the N box alone (lane 5).

View larger version (35K): [in this window] [in a new window]   FIG. 5.   The N- and C-terminal variable regions of Yra1p bind to Mex67p and to RNA. (A) GST-Mex67 pull-downs of Yra2- and Yra1-GFP fusions from extracts. The indicated Yra1- and Yra2-GFP fusions were expressed in vivo from a galactose-inducible promoter. Whole-cell extracts were incubated with GST-Mex67 purified from E. coli and immobilized on beads. Bound proteins were analyzed by SDS-PAGE followed by Western blot analysis using an anti-GFP antibody (upper panel). A fraction (1/20) of the total extracts was analyzed in parallel to confirm that all the GFP fusions were expressed to comparable levels (lower panel). (B) Yra1p and Yra2p bind RNA in vitro. An electrophoretic mobility retardation assay was performed with an 80-nucleotide 32P-labeled RNA probe (lane 1) and the purified GST fusions indicated above the lanes. For each protein, binding was carried out in the presence of 50 (lanes 2, 4, 6, 8, 10, 12, 14, 16, and 18) or 150 (lanes 3, 5, 7, 9, 11, 13, 15, 17, and 19) ng of tRNA competitor per µl. The binding reaction mixtures contained 500 ng of GST fusion 77-167 (RBD) and 50 ng of all the other fusion proteins. Protein-RNA complexes were resolved on a 5% native polyacrylamide gel and visualized by autoradiography. (C) Summary diagram of Yra1p domains and partners.

The N-vr and C-vr regions of Yra1p bind RNA. Yra1p was first identified as a protein with RNA-RNA annealing activity (27) and was more recently shown to bind RNA (40, 42). The central regions of Yra1p and Yra2p show strong homology to an RNP-like RBD (RRM), although the RNP-1 and RNP-2 motifs are not highly conserved. To test whether Yra2p binds RNA in vitro and to define the region(s) in Yra1p mediating the interaction with RNA, various GST fusion proteins purified from E. coli were tested for their ability to bind an 80-nucleotide-long ³²P-labeled RNA probe in an electrophoretic mobility shift assay (Fig. 5B). Yra2p and Yra1p exhibited comparable affinities for the RNA probe, and the complexes formed could be challenged with an excess of tRNA, reflecting a nonspecific interaction between the GST fusions and the RNA. Strikingly, the RBD (77-167) exhibited no RNA binding activity, even when added to the binding reactions in 10-fold-higher amounts than the other proteins. In contrast, both the N-terminal (1-77) and C-terminal (167-227) domains bound the RNA. Consistently, deletion of the RBD (RBD) did not affect interaction with RNA. Furthermore, deletion of the conserved C box (1-210) or the complete C-terminal domain (1-167) did not substantially affect RNA binding. Finally, a fusion containing just the N-vr region and the RBD (14-167) was still able to induce an RNA shift. These observations taken together indicate that the highly conserved N box, C box, and RBD are not required for RNA binding but that at least one variable region, N-vr or C-vr, is necessary to interact with RNA. These data show that the two variable regions of Yra1p not only interact with Mex67p but also are implicated in RNA binding (Fig. 5C). Earlier data have shown that Yra1p can interact simultaneously with Mex67p and RNA (42).

Yra1 truncations induce poly(A)⁺ RNA export defects. To assess the importance of individual Yra1p domains in poly(A)⁺ RNA export in vivo, in situ hybridizations with oligo(dT) were performed on strains expressing functional truncations of Yra1p as described for Fig. 3B. The strains were grown at 25°C to mid-log phase and subjected to analysis by fluorescent in situ hybridization at 25°C or after shifting the cells for 1 h to 37°C (Fig. 6). No nuclear accumulation of poly(A)⁺ RNA was detected in strains expressing wild-type Yra1p or the C-box deletion mutant (1-210). Deletion of the C-terminal domain (mutant 1-167) induced a modest poly(A)⁺ RNA export defect at 37°C. Deletion of the N box (14-227) or the complete N-terminal region (77-227) induced a more substantial export defect at 37°C. As the N-terminal deletions showed a lower affinity for Mex67p in vitro (Fig. 5A), the data indicate a correlation between the abilities of Yra1 mutants to interact with Mex67 and to promote poly(A)⁺ RNA export. Because the regions of Yra1p involved in Mex67p interaction also mediate RNA binding, it cannot be excluded that the export defect in strains lacking one of the two RNA binding regions (1-167 or 77-227) is due, in part, to a lower affinity of these Yra1 truncations for RNA in vivo, although they did not show substantially different RNA binding activities in vitro (Fig. 5B). Finally, inefficient nuclear localization of mutant 77-227, which lacks the NLS, may also contribute to the stronger export defect in this mutant (Fig. 4A). Although the RBD is required for neither Mex67p nor RNA binding in vitro (Fig. 5), the RBD mutant strain exhibited a substantial poly(A)⁺ RNA export defect at 37°C, indicating that this conserved domain is important for optimal Yra1p function in mRNA export.

View larger version (97K): [in this window] [in a new window]   FIG. 6.   Localization of poly(A)+ RNA in strains expressing wild-type (wt) Yra1p or Yra1p truncations as indicated at the top. Strains were grown at 25°C or shifted for 1 h to 37°C. Poly(A)+ RNA was detected by in situ hybridization with a digoxigenin-labeled oligo(dT) probe and a fluorescein isothiocyanate-conjugated antidigoxigenin antibody (left panels). The location of the nuclei was determined by staining the same cells with 4',6'-diamidino-2-phenylindole (DAPI; right panels). Cells were photographed as described in the legend to Fig. 4.

The association of Mex67p with mRNPs is weakened in strains expressing mutant Yra1p. We have shown that recombinant Yra1p, Yra2p, and Mex67p associate with mRNP complexes in extracts (Fig. 2), and in situ hybridizations on Yra1 mutant strains support the idea that efficient mRNA export depends on the interaction of Mex67p with Yra1p. To assess whether Yra1p influences the association of Mex67p with mRNPs in vivo in a more direct assay, we examined the levels of Mex67p associated with mRNP complexes in the presence of wild-type or mutant Yra1p (Fig. 7). For these experiments, the poly(A) binding protein Pab1p was tagged with ProtA at its C terminus in strains expressing HA-tagged Yra1p (wild type or mutant) as well as HA-tagged Mex67p (Materials and Methods). Pab1p is a very abundant, mainly cytoplasmic protein, but a fraction of Pab1p is nuclear and interacts with components of the 3'-end formation complex CF1A (23); Pab1p is therefore likely to be bound to nuclear and cytoplasmic mRNP complexes.

View larger version (44K): [in this window] [in a new window]   FIG. 7.   Yra1p facilitates the recruitment of Mex67p to the mRNP. mRNP complexes associated with Pab1p-ProtA in extracts expressing wild-type (wt; lanes 1 to 3) or mutant (lanes 4 to 7) HA-tagged Yra1p and HA-tagged Mex67p were affinity purified on matrix-bound IgG. Equal amounts of each extract were used for affinity purification. In lanes 1 and 2, extract containing wild-type Yra1p was preincubated for 20 min at 25°C with or without 30 µg of RNase A before purification on IgG beads. Proteins bound to Pab1-ProtA were eluted by high-salt treatment (2 M KCl) of the beads. IgG bead eluates (top) or total input extracts (bottom) were analyzed by Western blotting. Anti-HA antibodies were used to detect Mex67p and Yra1p; the hnRNP-like protein Npl3p was used as an internal control for mRNP purification and revealed with anti-Npl3p antibodies. Pab1-ProtA was detected in the input samples by cross-reaction of the anti-HA primary and anti-mouse secondary antibodies. For all samples, 1/10 of the eluted samples (IgG bead eluate) was used for the analysis of Mex67p and Npl3p and 1/4 was used for Yra1p. One-twentieth of the material used in each pull-down experiment was analyzed in parallel to determine the levels of these proteins in the starting extracts (input). Western blots were developed with Super Signal West Femto Chemiluminescent substrate in the case of IgG bead eluate samples and Super Signal West Pico Chemiluminescent substrate in the case of input samples.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Yra1p and other members of the REF family of hnRNP-like proteins have been proposed to participate in mRNA export by recruiting the mRNA export factor Mex67p/TAP to the mRNP. This study presents a functional analysis of Yra1p conserved domains and shows that Yra1 deletion mutants defective in Mex67p interaction have parallel poly(A)⁺ RNA export defects and a decrease in the amounts of Mex67p associated with mRNP complexes. These results confirm the model that Yra1p facilitates the recruitment of Mex67p to the mRNP to promote mRNA export. Overexpression of Yra2p, a second member of the REF family in S. cerevisiae, fully complements a YRA1 deletion, indicating overlapping or redundant functions for these two proteins (Fig. 1).

REF proteins contain a central conserved RNP-type RBD and two highly conserved N and C boxes separated by more variable regions, N-vr and C-vr. Surprisingly, the RBD of Yra1p exhibits neither RNA nor Mex67p binding activity in vitro and is not essential for growth. In contrast, both the N-terminal and C-terminal domains of Yra1p mediate RNA and Mex67p binding in vitro. Consistent with the functional overlap between these two domains, Yra1 truncations lacking either the N-terminal or C-terminal domain are viable (Fig. 3 and 5). Neither the N box nor the C box is absolutely required for RNA or Mex67p binding in vitro, but at least one of these sequences is necessary for viability. One possibility is that these conserved boxes optimize the binding activities of the adjacent N-vr or C-vr region in vivo. Solving the three-dimensional structure of Yra1p, in a cocrystal with Mex67p and/or RNA, will be necessary to understand the function of its conserved domains.

Deletion of the N box or the complete N-terminal or C-terminal domain induces nuclear accumulation of poly(A)⁺ RNA. As there was a good correlation between the effect of the Yra1 mutations on Mex67p binding and the degree of nuclear poly(A)⁺ RNA accumulation (Fig. 5 and 6), the data support the idea that mRNA export depends on an efficient interaction between Yra1p and Mex67p.

The Yra1 RBD mutant efficiently binds RNA in vitro and contains both Mex67p binding domains; this mutant nevertheless induces a substantial poly(A)⁺ RNA export defect, indicating that the RBD is important for efficient mRNA export. The RBD may have a structural role which is important for the proper folding of Yra1p in vivo. Alternatively, it may engage in additional nonessential interactions which stimulate export. In that respect, the conserved RBD of Aly, the Yra1p mouse homologue, has been shown to directly interact with the activation domain of a subset of transcription factors, suggesting that Aly, and by analogy Yra1p, may contribute to multiple steps in mRNA biogenesis and possibly enhance export by establishing a functional link with the transcription machinery (6, 44).

The amounts of Mex67p associated with mRNP complexes were substantially reduced in the Yra1 mutant strains deficient in mRNA export (Fig. 7). In the case of the Yra1 RBD mutant, these observations indicate that the RBD, although not directly involved, may nevertheless optimize the interaction between Mex67p and Yra1p. The impaired association of Mex67p with mRNP complexes in the Yra1 mutant strains may also be due, in part, to the lower levels of expression of the Yra1 mutant proteins in the starting extracts. For unclear reasons, the levels of the N- and C-terminal truncations (77-227 and 1-167) were particularly low in the Pab1-ProtA/Mex67-3HA double-tagged strain. As proportionally less Mex67p copurified with mRNPs in these samples, the data are fully consistent with a direct role of Yra1p in recruiting Mex67p to the mRNP. The poorly expressed Yra1 mutant proteins 1-167 and 77-227 were not detectable in the affinity-purified mRNP complexes, whereas Mex67p was reduced but clearly visible in these samples. These observations may suggest that additional interactions stabilize Mex67p on the mRNP after its recruitment by Yra1p. Alternatively, a fraction of Mex67p may associate with the mRNP independently of Yra1p. Finally, Mex67p may be recruited to different mRNP populations, some of which are naturally devoid of Yra1p. The results taken together show that poly(A)⁺ RNA export depends on the binding of Mex67p to the mRNP through an interaction with Yra1p. Whether the export of all or only a subset of mRNP complexes depends on this interaction is an interesting question for the future.

At least one highly conserved N box or C box is required for growth. However, the N box and C box are unlikely to have equivalent roles, as the deletion of the C box, but not the N box, induces a temperature-sensitive phenotype (Fig. 3). In addition, Yra1 mutants lacking the C box (or both the N box and C box) have a substantial dominant negative growth phenotype in a wild-type strain background even when expressed from a low-copy-number plasmid (data not shown). This phenotype may be related to the tendency of these mutants to form aggregates when expressed as GFP fusions (Fig. 4A). The C-box deletion mutant does not generate a clear poly(A)⁺ RNA export defect when expressed on its own (Fig. 6). As the C box is not playing a major role in Mex67p or RNA binding (Fig. 5), it may be required for a different aspect of Yra1p function. The C box could engage in nonessential interactions with additional partners or mediate conformational changes important for association-dissociation reactions between Yra1p and the mRNP or between Yra1p-containing complexes and other components along the export pathway.

The functional organization of the Yra1p domains has been conserved in REF proteins from higher eucaryotes. In mouse REF1-II and REF2-II, the central RBD similarly exhibits no RNA or TAP binding activity, whereas the N-terminal and C-terminal domains are implicated in both of these interactions. However, as in Yra1p, the RBD is important for the activity of REF1-II in mRNA export in vivo (30). Yra1p and REF2-II contain an NLS within the N-vr region, and the two proteins are imported through conserved pathways, since nuclear import of REF2-II is promoted by RanBP5, the mammalian homologue of Pse1p. REF2-II import is also mediated by transportin and a combination of importin- and RanBP7 (D. Görlich and E. Izaurralde, personal communication). Nuclear import of Yra1p by additional import receptors could explain the partial delocalization of GFP-Yra1p in the pse1-1 kap123 mutant (Fig. 4B).

The central M domain of Mex67p mediates heterodimerization with the essential export factor Mtr2p, and this interaction is important for the localization of Mex67p at the pore. Consistently, a recent report indicates that Mex67p interacts with FG-nucleoporins in vitro only as a Mex67p-Mtr2p heterodimer (33, 39). Another report proposes that Mtr2p is not absolutely required but may enhance the interaction of Mex67p with FG-nucleoporins in vitro (41). Here, we examined the interaction of a GST-Mex67 fusion with wild-type or mutant GFP-Yra1 fusions expressed in yeast extracts. Mtr2p present in extracts could associate with GST-Mex67p and enhance its ability to interact with GFP-Yra1. However, our earlier data showed that Mex67p and Yra1p interact in vitro in the absence of Mtr2p. We have also shown that the N-terminal domain of TAP including the leucine-rich repeat region (aa 1 to 372) mediates the interaction with Yra1p/REF proteins (42). This interaction is conserved in yeast, as the corresponding N-terminal domain of Mex67p (aa 1 to 264) interacts with GST-Yra1/2 fusions (D. Zenklusen, unpublished data). It is therefore unlikely that Mtr2p is required for the Mex67p-Yra1p interaction, but it could modulate this association.

Both mouse REF2-II and Aly (REF1-I) proteins shuttle between the nucleus and the cytoplasm even in the absence of RNA (30, 47). Based on the extensive functional conservation between Yra1p and REF2-II, Yra1p is likely to shuttle as well, although a standard assay for protein shuttling in yeast (20) initially scored Yra1p as a nonshuttling protein (40, 42). As an indication that Yra1p is shuttling between the nucleus and the cytoplasm, localization of a functional ProtA-tagged Yra1p by immunoelectron microscopy has detected the fusion protein on both sides of the pore, in association with the nuclear basket and the cytoplasmic fibrils (B. Fahrenkrog and E. Izaurralde, personal communication).

Yra2p overexpression fully rescues a YRA1 deletion, and GFP-Yra2 was selected from extracts by GST-Mex67 in an RNA-independent manner, consistent with a direct interaction between these two proteins (Fig. 2). Although the activities of Yra2p and Yra1p appear to overlap broadly, the two proteins may have distinct roles under physiological conditions. To investigate a possible role for Yra2p in the export of a specific class of transcripts, we examined heat shock RNA export in a YRA2 strain. However, analysis of heat shock protein synthesis in this strain after a shift to 42°C did not reveal an essential role for Yra2p in the export of these regulated transcripts (Y. Strahm, unpublished data). Further studies will be required to determine whether Yra1p and Yra2p are involved in the export of different types of transcripts.

The GST pull-down experiments also identified an RNA-independent interaction between Mex67p and the RNA export factor Gle2p (Fig. 2). An interaction between the human homologues TAP and hGle2 has been described earlier, which involves the C-terminal half of TAP, indicating that distinct domains of TAP bind REF and hGle2 (2). The corresponding proteins in S. pombe, spMex67p and spRae1p, were also shown to associate in a complex in vivo (46). Gle2p, spRae1p, and hGle2 accumulate at the NPC, and hGle2 shuttles between the nucleus and the cytoplasm and cross-links to poly(A)⁺ RNA in vivo (3, 19, 24, 28). Gle2p, spRae1p, and hGle2 are thought to function at the pore, but the exact role of these homologous proteins in mRNA export is not defined. They may interact with the mRNP via a yet unidentified mRNP component and promote export in conjunction with Mex67p/TAP. Alternatively, they may contribute to the interaction of Mex67p/spMex67p/TAP with the pore. Gle2p was not detected in the pull-down assays with GST-Yra1p and GST-Yra2p, suggesting that these proteins do not directly associate. These observations also suggest that Mex67p bound to Gle2p in extracts may not be recognized by GST-Yra1/2, either because the fraction of Mex67p associated with Gle2p is too low or because the interaction of Yra1p, Yra2p, and Gle2p with Mex67p is mutually exclusive. Further studies will be required to elucidate the functional relationship between Mex67p and Gle2p.

YRA1 was initially isolated in a screen for GAL-inducible yeast cDNAs causing overexpression-mediated growth arrest (9). We observed a similar phenotype when overexpressing the YRA1 cDNA from its own promoter. Interestingly, overexpression of the YRA1 gene, which contains an unusually long intron, had no dominant negative phenotype, whether it was expressed from GAL1 or its own promoter. The levels of Yra1p were 5 to 10 times lower in strains expressing the YRA1 gene than in those expressing the YRA1 cDNA (Fig. 1C). These observations suggest that the Yra1p levels are tightly controlled by a mechanism involving the YRA1 intron. It will be interesting to determine whether Yra1p affects its own production by interfering with the transcription, splicing, or stability of the corresponding pre-mRNA.

Recent reports showed that mouse Aly (REF1-I) and several other proteins, including SRm160, DEK, RNPS1, and Y14, are recruited to mRNP complexes during spliceosome assembly and become tightly associated with the spliced mRNPs. Splicing was proposed to be required for efficient export by ensuring a tight coupling between the processing and export machineries (18, 21, 47). However, anti-REF antibodies were shown to inhibit the export of mRNAs from Xenopus oocyte nuclei, whether these were injected as in vitro-transcribed pre-mRNAs or as non-intron-containing mRNAs, indicating that REF associates with both types of transcripts in vivo (30). In yeast, only a small fraction of genes contains introns and splicing is therefore unlikely to be a necessary step for export. Yra1p and/or Yra2p could nevertheless stimulate the export of spliced transcripts through preferential association with these mRNAs. It will be interesting to determine whether Yra1 mutants primarily affect the export of spliced transcripts and whether Yra1p and/or Yra2p associates with spliceosomes and splicing products in vitro.