Cytoplasmic Inositol Hexakisphosphate Production Is Sufficient for Mediating the Gle1-mRNA Export Pathway*

Production of inositol hexakisphosphate (IP6) by Ipk1, the inositol-1,3,4,5,6-pentakisphosphate 2-kinase, is required for Gle1-mediated mRNA export in Saccharomyces cerevisiae cells. To examine the network of interactions that require IP6 production, an analysis of fitness defects was conducted in mutants harboring both an ipk1 null allele and a mutant allele in genes encoding nucleoporins or transport factors. Enhanced lethality was observed with a specific subset of mutants, including nup42, nup116, nup159, dbp5, and gle2, all of which had been previously connected to Gle1 function. Complementation of the nup116Δipk1Δ and nup42Δipk1Δ double mutants did not require the Phe-Gly repeat domains in the respective nucleoporins, suggesting that IP6 was acting subsequent to heterogeneous nuclear ribonucleoprotein targeting to the nuclear pore complex. With Nup42 and Nup159 localized exclusively to the nuclear pore complex cytoplasmic side, we speculated that IP6 may regulate a cytoplasmic step in mRNA export. To test this prediction, the spatial requirements for the production of IP6 were investigated. Restriction of Ipk1 to the cytoplasm did not block IP6 production. Moreover, coincident sequestering of both Ipk1 and Mss4 (an enzyme required for phosphatidylinositol 4,5-bisphosphate production) to the cytoplasm also did not block IP6 production. Given that the kinase required for inositol 1,3,4,5,6-pentakisphosphate production (Ipk2) is localized in the nucleus, these results indicated that soluble inositides were diffusing between the nucleus and the cytoplasm. Additionally, the cytoplasmic production of IP6 by plasma membrane-anchored Ipk1 rescued a gle1-2 ipk1-4 synthetic lethal mutant. Thus, cytoplasmic IP6 production is sufficient for mediating the Gle1-mRNA export pathway.

The nucleus is the defining structure of a eukaryotic cell and houses the genetic information that characterizes the organism. The nuclear compartment is separated from the cytoplasm by the nuclear envelope (NE), 1 two lipid bilayer membranes that join to form pores containing nuclear pore complexes (NPCs). The 60-MDa NPC structure is assembled from multiple copies of ϳ30 distinct proteins designated nucleoporins (Nups) (1,2). The central region of the NPC consists of inner and outer rings connected by a series of spokes. Fibrils extend from either face of the NPC, and the filaments on the nuclear side are joined at the distal end to form a basket structure (3). The NPCs create aqueous semipermeable portals across the NE that allow the passive diffusion of small molecules. However, translocation of macromolecules through the NPC requires facilitated transport mediated by Nups, other NPC-associated proteins, and shuttling transport factors (4 -6). One class of macromolecules that must traverse the NPC is mRNAs. The movement of mRNA through the NPC is a regulated process in the progression from DNA transcription in the nucleus to the production of a translated protein in the cytoplasm (7,8). To ensure proper gene expression, there are many steps contributing to this pathway and potential points of control by signaling and checkpoint mechanisms. The pre-mRNA is processed by polyadenylation, 5Ј-cap addition, and potentially splicing; packaged into a heterogeneous nuclear ribonucleoprotein (hnRNP) complex; and targeted to the NPC. Translocation through the NPC is followed by coordinated delivery of the transcript to the ribosome and recycling of RNA-binding proteins and shuttling transport factors to the nucleus. A conserved heterodimer, Mex67/Mtr2 in budding yeast and NXF1/ p15 in vertebrates, is a central player in this mechanism (9 -11). Mex67/Mtr2 has been linked to both the transcription and splicing machineries. It also is thought to serve as a transport mediator by interacting both with hnRNPs and with Nups that harbor domains with phenylalanine-glycine (FG) repeats (10,(12)(13)(14)(15)(16). The essential factor Gle1 also associates with Nups (17)(18)(19)(20) and, in yeast cells, is functionally connected to the hnRNP Nab2 (21). The dual association of shuttling transport factors with both cargo and Nups is a fundamental tenet of NPC translocation models. However, how all these events are coordinated for the seamless delivery of mRNAs for translation remains poorly defined.
The overall mechanism for nuclear mRNA export is likely based on coordinated changes in the hnRNP composition and associated factors (8). Some proteins are specifically recruited cotranscriptionally or coincident with execution of different processing steps. For example, association of the RNA-binding protein Yra1/ALY/REF is promoted by Sub2/UAP56 and transcription elongation factors (22)(23)(24). In addition, the association of some proteins is temporal, with release prior to export, whereas others remain assembled until after transport to the cytoplasm (25). The DEAD box protein Dbp5 may play a role in hnRNP remodeling during export (19,26). Regulated post-translational modifications may also trigger changes in protein-protein interactions. In budding yeast, the SR (serine/arginine-rich)-like protein Npl3 undergoes specific phosphorylation and dephosphorylation cycles linked to proper mRNA export (12,27,28).
The efficient export of mRNA in Saccharomyces cerevisiae is also influenced by the production of inositol hexakisphosphate (IP 6 ) (29). The budding yeast metabolic pathway from phosphatidylinositol 4,5-bisphosphate (PI(4,5)P 2 ) to IP 6 requires three enzymes. PI(4,5)P 2 is hydrolyzed by phospholipase C (Plc1) to produce soluble inositol 1,4,5-trisphosphate (IP 3 ) (30). Ipk2 is a dual specificity IP 3 6-kinase and inositol-1,4,5,6tetrakisphosphate (IP 4 ) 3-kinase that generates both IP 4 and inositol 1,3,4,5,6-pentakisphosphate (IP 5 ) (31). Finally, Ipk1 is an IP 5 2-kinase that generates IP 6 (29). All evidence to date indicates that the IP 3 -to-IP 6 pathway produces distinct molecules involved in regulating multiple and diverse nuclear processes, including mRNA export, transcription, and chromatin remodeling (29,(31)(32)(33). Work also suggests that specifically IP 6 production plays a role in mRNA export. A saturating synthetic lethal screen with a gle1 mutant specifically identified mutant alleles of PLC1, IPK2, and IPK1 (29,31). The plc1, ipk2, and ipk1 mutants independently have defects in poly(A ϩ ) mRNA export (29). Moreover, the common loss-of-function biochemical defect among the three mutants is a loss of IP 6 production, and Ipk1 is required for suppression of the temperature sensitivity of a gle1 mutant by Plc1 overexpression (29). We have previously characterized ipk1 point mutants that are nonfunctional with regard to both IP 6 production and Gle1 function (34). This observation and the fact that an ipk2 mutant is defective even though the Ipk1 protein is present (29) are collectively strong evidence that IP 6 itself is the required mediator versus a nonkinase-related role for the Ipk1 protein. IP 6 also serves as a precursor for diphosphorylated inositols such as diphosphoinositol pentakisphosphate (IP 7 ) and bisdiphosphoinositol tetrakisphosphate (IP 8 ) (35). However, the enzyme required for IP 7 production is not apparently required for mRNA export (34,36). Taken together, these results indicate that IP 6 acts to regulate efficient mRNA export mediated by Gle1.
The presence of inositide-modifying enzymes and independent lipid pools within the nucleus also supports the hypothesis that inositide pathways function within the nucleus (37,38). Pik1, an essential phosphatidylinositol 4-kinase, and Mss4, an essential phosphatidylinositol-4-phosphate (PI(4)P) 5-kinase, are both required for PI(4,5)P 2 production in budding yeast (39,40). Recent studies have shown that Pik1 and Mss4 localize in the nucleus and that Mss4 shuttles between the nucleus and the cytoplasm (41,42). In mammalian cells, PI(4)P 5-kinases also localize in the nucleus (43). Fluorescence microscopy and subcellular fractionation studies showed that, in budding yeast, Ipk2 concentrates in the nucleus (31,44), whereas Ipk1 is localized in the nucleus and at the nuclear rim (29). These connections suggest that production of inositols in the nucleus might be critical for regulating events in the nuclear compartment, including mRNA export. However, the spatial constraints for soluble inositide production in the cytoplasmic versus nuclear compartments have not been investigated. The mechanism by which Ipk1 activity and IP 6 production regulate mRNA export is also unknown. In this study, we have taken multiple approaches to further investigate the requirements for IP 6 production in nuclear transport and NPC function. We speculate that the production of soluble IP 6 influences events at the cytoplasmic face of the NPC during mRNA export.
Unless indicated otherwise, the null (⌬) mutant strains used in these studies were obtained from Research Genetics. Table I summarizes the  novel yeast strains used in this study, and Table II summarizes the plasmids. The ipk1⌬::KAN R strain from the null collection (Research Genetics) was altered to generate the starting strain for the screen. The antibiotic resistance was changed from kanamycin to nourseothricin (ipk1⌬::NAT R ) using an MX4 switcher cassette (46). The ipk1⌬::NAT R strain was then manipulated to generate a MAT␣ strain in which the CAN1 locus was disrupted with HIS3 behind the MFA1 promoter (can1::MFA1 pr-HIS3) to allow expression of HIS3 specifically in MATa haploid cells (46). Some of the haploid strains were generated by classic sporulation and dissection of the heterozygous diploids. Alternatively, the synthetic lethal testing strategy developed by Boone and co-workers was used (46). In the latter situation, strains with null KAN R -marked mutations in individual genes related to transport were arrayed onto a single 96-well plate. The transport factor plate was pinned onto a lawn of the ipk1⌬::NAT R strain or an isogenic strain with an independent NAT R gene. The resulting diploids were selected with kanamycin and nourseothricin. After sporulation, the first round of selection was for HIS3 to select only MATa spores. Additive rounds of selection were completed to select for cells with kanamycin and nourseothricin resistance, generating a corresponding array of double mutant strains. Strains generated in the array format were not individually isolated as permanents. For double mutants that showed a potential growth defect, haploid strains were independently generated by sporulation and tetrad dissection of the heterozygous diploids. Equivalent mutations in the W303 background were also tested.
Cryo-immunoelectron Microscopy-Yeast nuclei from wild-type diploid cells in the W303 background were purified as described (47). Samples containing nuclei in ϳ2.5 M sucrose in polyvinylpyrrolidone buffer were diluted with 1 volume of 4% paraformaldehyde in phosphate-buffered saline and incubated at room temperature for 2 h. The fixed nuclei were pelleted at 55,000 rpm in a TLA55 rotor for 1 h at 4°C. The pellet was rinsed two times in phosphate-buffered saline, embedded in 10% gelatin, and processed for ultracryotomy as described (48). Ultrathin sections were prepared and incubated with blocking buffer containing 10% goat serum. Immunolabeling with the primary antibodies was carried out for 2 h. For Gle1 labeling, affinity-purified rabbit polyclonal antibodies raised against Gle1 (WU851) were used at a dilution of 1:20 (21). For Nup159 labeling, a tissue culture supernatant of monoclonal antibody 165C10 (49) was used at 1:5. Incubation with gold-labeled secondary antibodies (12-nm gold-labeled goat anti-rabbit IgG or goat anti-mouse IgG) was carried out for 1 h. After washing, sections were stained with uranyl acetate and embedded in methyl cellulose (48). Specimens were visualized with a Zeiss 902 electron microscope, and photographs were recorded with Eastman Kodak electron microscopy film.
In Vivo Imaging-Fluorescence and differential interference contrast microscopy studies were performed on an Olympus BX50 microscope using a UPlan ϫ100/1.3 objective. Images were collected in MetaVue TM (Universal Imaging Corp.) using a Photometrics Cool-Snap HQ TM camera (Roper Scientific). All images were captured under identical exposure conditions.
Radiolabeling and Determination of Inositide Profiles in S. cerevisiae-Yeast cells were grown in the presence of 30 Ci/ml [2-3 H]myoinositol (ICN) to late logarithmic phase. Soluble inositide polyphosphates were isolated as described previously (51). Inositides were separated by HPLC on a Partisphere SAX (strong anion exchange) column (4.6 ϫ 125 mm; Whatman). Inositides were eluted with a 25-min linear gradient from 10 mM to 1.7 M ammonium phosphate (pH 3.5), followed by 1.7 M for an additional 20 min. Peaks were identified by comparison with known peaks in isolates from wild-type or ipk1⌬ yeast.

The ipk1⌬ Mutant Has Genetic Interactions with a Specific Subgroup of Nucleoporin and Nuclear Transport Mutants-
The role of IP 6 production in mRNA export was discovered based on the synthetic lethality of gle1-2 ipk1 double mutants (29). Because the exact role of Gle1 in mRNA export is not yet defined, we wanted to broadly evaluate what other mRNA export factors, Nups, or nucleocytoplasmic transport factors are influenced by IP 6 production. We speculated that functions dependent on IP 6 would be perturbed by its absence in the ipk1⌬ strain and result in double mutants with phenotypes more severe than either single mutant. Such synthetic lethal or synthetic fitness defects have historically helped to define roles for numerous nonessential genes (52).
To examine this, we systematically combined the ipk1⌬ mutant with a panel of selected yeast mutants. The tester yeast mutant strains included null mutant alleles of nonessential genes as well as conditional mutant alleles of essential genes. We were particularly interested in the mex67, dbp5/rat8, and gle2 mutants, which have specific mRNA export defects. Mu- This study tants of genes known to encode Nups and karyopherins (Kaps) were also tested. Kaps are shuttling transport factors that mediate the import and export of proteins, tRNA, and ribosomes in a mechanism that is dependent on the small GTPase Ran (4,53). A direct role for a Kap in the mRNA export pathway has not been fully resolved (5,8). Strains from the Research Genetics MATa collection harboring null mutant alleles in genes encoding factors related to nucleocytoplasmic transport were tested in an array format (see "Materials and Methods"). Each mutant was crossed with either ipk1⌬::NAT R or a control strain with an independent NAT R mutant. Additionally, an ipk1⌬::KAN R mutant strain was used to generate double mutants by a traditional mating, sporulation, and dissection strategy. Viable double mutant strains were tested for growth at 23, 30, 34, and 37°C on YPD medium and compared directly with the corresponding single mutant strains. Under these conditions, the ipk1⌬ strain showed colony formation. For situations wherein viable double mutant haploid strains were not obtained, a URA3/CEN plasmid harboring a wild-type gene was included. The resulting plasmid-rescued double mutants were tested for growth on medium containing 5-FOA, with only those that could live without the URA3/CEN plasmid scored as viable.
Results from testing the double mutants are summarized in Table III. The gle1-2 ipk1⌬ and dbp5-2 (rat8-2) ipk1⌬ mutants showed strong synthetic lethality with no viability on 5-FOA at any growth temperature. A subset of the other double mutants showed enhanced lethality wherein the double mutant was lethal at a lower growth temperature than either individual single mutant. Such synthetic fitness defects were observed for ipk1⌬ combined with nup42⌬, gle2⌬, nup116⌬, and nup159-1 (rat7-1). Remarkably, every gene found to have a genetic interaction with the ipk1⌬ mutant had previously been connected by genetic, molecular, or biochemical tests to Gle1 function (17, 19, 20, 54 -56). Interestingly, the majority of the double mutants did not exhibit any enhanced growth defects. This reinforced our hypothesis that IP 6 production is specifically linked to the Gle1-mediated mRNA export pathway.
There was not a complete genetic interaction overlap between gle1 and ipk1 synthetic lethal phenotypes. Some mutants that show fitness defects with gle1 mutants did not have enhanced defects with the ipk1⌬ mutant. There was no observable growth difference in the double nup100⌬ ipk1⌬ double mutant, although nup100⌬ gle1 mutants are synthetically le-thal (17). Furthermore, not every mutant with documented mRNA export or nucleocytoplasmic transport defects was affected by combination with the ipk1⌬ mutant. The mex67 ipk1 double mutants were the same as the mex67 single mutants. None of the kap mutant alleles tested showed any genetic interaction, and the majority of the nup mutants tested did not. Overall, only a relatively small subgroup of the mutants tested (5 of 36) showed a synthetic fitness defect with the ipk1⌬ mutant. The diagram in Fig. 1A summarizes the results. The analysis identified genes that encode a specific subset of proteins, all of which have documented interactions at the NPC cytoplasmic face. The affected mutants included the genes encoding two of the Nups localized exclusively to the NPC cytoplasmic face (Nup42 and Nup159) (1,49). In contrast, none of the genes encoding Nups that are localized exclusively to the NPC nuclear face (Nup1, Nup2, and Nup60) were linked (1,(57)(58)(59). These results potentially pinpoint a general site of action for IP 6 to the cytoplasmic face of the NPC during mRNA export.
The Gle1-binding Domain of Nup42 Is Required to Complement the nup42⌬ ipk1⌬ Double Mutant-The nup116⌬ ipk1⌬ and nup42⌬ ipk1⌬ double mutants were of particular interest. We originally identified GLE1 in a synthetic lethal screen with the nup100⌬ and nup116⌬ mutants (17,54) and demonstrated a physical interaction between Gle1 and Nup42/Rip1 (17). Others have found that the C-terminal 68 amino acid residues of Nup42 are necessary and sufficient for mediating the interaction between Gle1 and Nup42 and for complementing a gle1 nup42/rip1 synthetic lethal mutant (56). The N-terminal region of Nup42 is an FG repeat domain that binds Mex67 and Kaps (14,60). To test whether the nup42⌬ ipk1⌬ synthetic fitness defect is linked to Gle1 function, we identified the region of Nup42 that is required to complement the phenotype. Plasmids expressing full-length Nup42 (pFS398) or only the C-terminal Gle1-binding domain (lacking all of the FG repeats; pFS730) were transformed into the nup42⌬ ipk1⌬ strain. The growth defect at 37°C was complemented by both plasmids (Fig. 1B). Thus, the FG domain was not necessary, and the Gle1 domain was sufficient. This suggests that the role of IP 6 may be related to the interaction between Gle1 and Nup42 or the mRNA export step mediated by their association.
To test whether IP 6 production is required for mediating the Gle1-Nup42 protein interaction, the two-hybrid assay was used. We generated a two-hybrid reporter strain harboring the   : nup2⌬, nup53⌬, nup59⌬, nup60⌬, nic96-1, nup120⌬,  nup145⌬ N, nup157⌬, nup170⌬, nup188⌬, kap104-16, kap108⌬, kap114⌬, kap119⌬, kap120⌬, kap122⌬, kap142⌬, gfd1⌬, los1⌬, mlp1⌬, mlp2⌬,  pom34⌬, pom152⌬, sac3⌬, seh1⌬, snl1⌬, and srn2⌬. All strains were grown on YPD medium except for the indicated strains for the GLE1 and DBP5 genes, which were tested on medium containing 5-FOA. ϩ, growth at the indicated temperature; Ϫ, no growth at the indicated temperature. 1. Genetic interactions between the ipk1⌬ mutant and nup mutants. A, shown is a diagram of the NPC structure and protein substructural localizations with potential sites for IP 6 action highlighted. Along the 8-fold rotational axis (vertical arrow), the relative surfaceaccessible localizations for the proteins at the NPC are shown based on data summarized in Ref. 1. For clarity, the proteins are shown only once, although all are present in at least eight copies along the axis. Genes encoding those proteins on the left side in light shading were linked to IPK1 function and thus IP 6 production. Nup82 did not have connections, but is shown on the left to highlight its structural role in interacting with the other affected Nups. Genes encoding proteins on the right side in darker shading did not have any connections to IPK1. FG domain-containing Nups are shown as ovals; structural Nups or non-FG domain-containing Nups are rectangles/squares; and mRNA export factors are hexagons. Nup100 and Nup145N are not included; however, both are partially redundant with Nup116 function (75). B and C, growth defects in ipk1⌬ nup strains were complemented by expression of defined regions of the respective Nups. Double mutant strains were transformed with CEN plasmids bearing genes expressing the Nup in the wild-type or deleted form. The diagrams show the respective structural domains: Nup42 has an N-terminal FG domain and a C-terminal Gle1-binding domain, whereas Nup116 has an N-terminal Gle2-binding domain, a middle GLFG domain, and a C-terminal Nup82-binding domain. Growth was compared after spotting equal numbers of cells onto YPD medium in serial 5-fold dilutions at the designated temperatures for 3-4 days. Rescue with IPK1 was used as a positive control. In B, the enhanced lethality of ipk1⌬ nup42⌬ at 37°C was rescued by NUP42 and nup42⌬FG plasmids. Nup42⌬FG has a deletion of residues 2-364. In C, the enhanced lethality of ipk1⌬ nup116⌬ at 34°C was rescued by NUP116, nup116⌬2, and nup116⌬4, but not by nup116⌬1 or nup116⌬3. Nup116⌬1 has a deletion of residues 6 -180; Nup116⌬2 has deletion of residues 186 -725; Nup116⌬3 is truncated at residue 731; and Nup116⌬4 is truncated at residue 923.
ipk1⌬ mutant (SWY2109). Plasmids expressing Gal4 DNAbinding domain and Gal4 activation domain fusions of Gle1 and Nup42, respectively, were transformed into wild-type and ipk1⌬ two-hybrid reporter strains. The interaction between Gal4 DNA-binding domain-Gle1 and Gal4 activation domain-Nup42 was not changed in the ipk1⌬ reporter strain (data not shown). Thus, IP 6 production is not required for the Gle1-Nup42 physical interaction.
The Glycine-Leucine-Phenylalanine-Glycine (GLFG) Region of Nup116 Is Not Required for Complementation of Either the nup116⌬ ipk1⌬ or nup116⌬ gle1-1 Double Mutant-Several domain-specific protein interaction partners have been defined for Nup116. The N-terminal region of Nup116 contains FG repeats as well as a Gle2-binding site (61,62). The middle domain harbors an extensive span of predominantly GLFG repeats and binds Mex67 and Kaps (13,14,60,63,64). Meanwhile, the C-terminal region, conserved with the C-terminal regions of Nup100 and Nup145N, interacts with Nup82 and mediates Nup116 assembly in the NPC (65,66). Using a panel of TRP1/CEN plasmids expressing deletion mutants of NUP116, we identified the region required for complementation of both the nup116⌬ gle1-1 and nup116⌬ ipk1⌬ phenotypes. The nup116⌬ gle1-1 mutant harbors a wild-type NUP116/URA3/CEN plasmid, and complementation was tested by evaluating growth on medium containing 5-FOA. 5-FOA is toxic to cells containing the wild-type Ura3 protein, and therefore, growth will be observed only when cells can lose the NUP116/URA3/CEN plasmid and be maintained by the mutant nup116/TRP1/CEN plasmid. As shown in Fig. 2, complementation was observed upon expression of Nup116 lacking the GLFG domain (Nup116⌬2). Growth was partially rescued by Nup116 lacking the last half of its C-terminal region (Nup116⌬4). The partial rescue by expression of the Nup116⌬4 protein may be due to the lower expression level of the altered protein. In contrast, complementation was not supported by deletion of the N-terminal (Nup116⌬1) or C-terminal (Nup116⌬3) domain of Nup116. The Nup116⌬1 and Nup116⌬3 proteins were stably expressed. Similarly, expression of Nup100 lacking the GLFG region was sufficient to rescue the nup100⌬ gle1-1 phenotype (data not shown). These findings indicate a role for the non-GLFG domains of Nup100 and Nup116 in Gle1 function.
When the same panel of plasmids was tested with the nup116⌬ ipk1⌬ double mutant, an identical set of complementation results was obtained. Growth of the nup116⌬ ipk1⌬ strain at the non-permissive temperature of 34°C was rescued by Nup116⌬2, which lacks the GLFG region (Fig. 1C). Deletion of either the N-terminal (Nup116⌬1) or C-terminal (Nup116⌬3) region resulted in lack of complementation. Thus, growth in the absence of IP 6 or with a gle1-1 mutant was dependent on the presence of both terminal regions of Nup116 and did not require the GLFG region. The common genetic and functional connections of IP 6 production with Nup42-Nup116 and of Gle1 with Nup42-Nup116 further solidified our hypothesis that Gle1-mediated mRNA export requires IP 6 .
Cryo-immunolocalization of Gle1 to Both the Cytoplasmic and Nucleoplasmic Faces of the NPC-Nup42 and Nup159, both of which have connections to Ipk1 and Gle1 function, are localized exclusively to the NPC cytoplasmic face (1,49). The other two Nups with Ipk1 and Gle1 links, Nup116 and Nup100, are localized to both NPC faces; however, their distribution is biased toward the cytoplasmic face (1,66). None of the Nups localized exclusively to the NPC nuclear face were connected to IP 6 production (Fig. 1A). To further analyze IP 6 roles at the NPC, we felt it was important to be confident of the subcellular localization of Gle1. We had previously localized Gle1 in yeast at the NPC/NE by indirect immunofluorescence microscopy (17). Additional studies by two other groups have reported substructural NPC localization for Gle1 using immunoelectron FIG. 2. The GLFG region of Nup116 is not required to complement the nup116⌬ gle1-1 mutant phenotype. Plasmids expressing each of the Nup116 deletions were transformed into the nup116⌬ gle1-1 strain containing the NUP116/URA3/CEN plasmid. Transformants were selected on synthetic complete medium lacking tryptophan. Growth was assayed at 23°C for 5 days on synthetic complete medium lacking Trp or on synthetic complete medium containing 5-FOA. Only strains that were viable and capable of losing the wild-type NUP116/URA3 plasmid showed growth on 5-FOA-containing medium. microscopy. One study concluded that Gle1 was present exclusively on the NPC cytoplasmic fibrils (20), whereas the other provided evidence for Gle1 distribution to both sides of the NPC (1). Both of these studies localized epitope-tagged versions of Gle1, but used different methods of sample preparation.
To resolve this discrepancy, we employed an independent approach to visualize endogenous Gle1 at yeast NPCs. Affinitypurified anti-Gle1 polyclonal antibodies were used to probe purified wild-type yeast nuclei by cryo-immunoelectron microscopy. Cryosections were incubated with an affinity-purified polyclonal antibody raised against Gle1, followed by a secondary antibody coupled to 12-nm gold particles. A parallel control experiment was completed using a monoclonal antibody recognizing Nup159 (49). The samples were viewed by electron microscopy, and the results were quantified by counting gold particles according to their distance from the mid-plane of the nuclear pore membrane. An example micrograph for the anti-Gle1 labeling is shown in Fig. 3A. Negative distance values reflect localization toward the nucleus and positive values toward the cytoplasm. Gold particles labeling Gle1 were localized to NPCs on both sides of the NE (Fig. 3C), with 42% of the particles concentrated in a 50-nm region spanning both sides from Ϫ25 to ϩ25 nm from the mid-plane. In contrast, Nup159 labeling was exclusively on the cytoplasmic side (Fig. 3B), as reported previously (49). The presence of Gle1 on both the nuclear and cytoplasmic sides of the NPCs, with a greater proportion on the cytoplasmic face, is consistent with the previous report of Rout et al. rified wild-type yeast nuclei was performed. The localization of Gle1 was detected by binding of affinity-purified rabbit anti-Gle1 polyclonal antibodies and 12-nm gold-labeled anti-rabbit secondary antibody. No labeling was observed after incubation with secondary antibodies alone. Gold particles (black arrows) were present on both the cytoplasmic and nuclear sides of the NE at sites of NPCs (asterisks). n, nucleus; c, cytoplasm; ne, nuclear envelope; SPB (black arrowhead), spindle pole body. Scale bar ϭ 0.4 m. B and C, quantification of gold particle location (n ϭ total number counted) for nuclei labeled with either anti-Gle1 antibody (C) or a mouse monoclonal antibody recognizing Nup159 (B) (49). The location of the gold particles was measured from the mid-plane (distance ϭ 0) of the nuclear membrane pores in perpendicularly sectioned NEs. Positive distances are noted for the cytoplasmic side, and negative distances are noted for the nuclear side. The particles were grouped into 25-nm zones for the histograms.
pre-embedded labeling of yeast cells versus cryosectioning (our study) or purified NEs (1). We concluded that Gle1 is present on both sides of the NPC.
Sequestering Ipk1 and Mss4 in the Cytoplasm Allows IP 6 Production-Our previous indirect immunofluorescence microscopy studies have shown that Ipk1 is located in the nucleus and at the nuclear rim (29). Given that Gle1 was present on both NPC faces, whereas two of the Nups connected to Ipk1 function were exclusively cytoplasmic (Nup42 and Nup159), we investigated whether there is a spatial requirement for IP 6 production within the nucleus. To do this, Ipk1 was restricted to the cytoplasm by fusion to a truncated form of Ste2 that lacks its cytoplasmic tail domain (Ste2⌬tail). Ste2 is a plasma membrane-localized integral membrane protein with seven transmembrane domains and functions as the ␣-mating factor receptor (67). A similar fusion protein, Ste2⌬tail-GFP, is targeted to the plasma membrane and localizes to membrane structures within the cytoplasmic compartment (50). To visualize the localization of Ste2⌬tail-Ipk1, we made a further in-frame fusion of GFP to the Ipk1 C terminus. By direct fluorescence microscopy, cells expressing Ste2⌬tail-GFP and Ste2⌬tail-Ipk1-GFP showed highly enriched plasma mem-brane localization (Fig. 4). Low levels of signal were detected at the nuclear rim, likely reflecting endoplasmic reticulum or outer NE localization during synthesis. Most important, Ste2⌬tail-Ipk1-GFP was completely restricted from the nucleoplasm. Thus, when expressed in an ipk1⌬ cell, Ste2⌬tail-Ipk1-GFP activity and any resultant IP 6 production would be restricted to membranes facing the cytoplasmic compartment.
To examine whether ipk1⌬ cells expressing only Ste2⌬tail-Ipk1-GFP produce IP 6 , the strain was grown in the presence of [ 3 H]inositol, and the inositol polyphosphates were isolated and analyzed by HPLC (Fig. 5). As has been reported previously (29), in wild-type cells, the major soluble inositide peak was IP 6 (Fig. 5A). This peak was absent in ipk1⌬ cells, and instead, peaks representing IP 4 , IP 5 , and PP-IP 4 were observed (Fig. 5,  B and E) (29). Based on the HPLC analysis, in the ipk1⌬ strain producing Ste2⌬tail-Ipk1-GFP, PP-IP 4 and IP 4 were absent, and IP 5 levels were diminished (Fig. 5, compare C and D). Most important, a significant increase in the IP 6 level was observed (Fig. 5D). The production of IP 6 indicated that the kinase function of Ste2⌬tail-Ipk1-GFP was intact. Some IP 5 was still present; however, we have reported previously that plasmidbased expression of IPK1 may not be fully effective (34). These results also meant that cytoplasmic Ipk1 had access to its IP 5 substrate in vivo. A GFP-tagged form of Ipk2, the kinase that produces IP 5 , displays a predominant nuclear localization (31). Taken together, these results suggest that IP 5 diffuses from the nucleus to the cytoplasm for phosphorylation by the cytoplasmically sequestered Ste2⌬tail-Ipk1-GFP fusion protein (Fig. 6A).
To further investigate the dynamics of soluble inositides in cells, we analyzed cells expressing forms of Ipk1 and the PI(4)P 5-kinase Mss4 that are restricted to the cytoplasm. Recently, it has been shown that Mss4 can be membrane-anchored by fusion to the N-terminal region of Psr1, a dipalmitoylated, plasma membrane-localized phosphatase (42). Localization studies show that Psr-Mss4-GFP is localized predominantly to the plasma membrane, with low levels on the outer NE and/or vacuole. In mss4⌬ cells expressing Psr-Mss4-GFP, HPLC analysis of soluble inositides showed a profile similar to that of wild-type cells, with a single IP 6 peak present (Fig. 5, compare A and F). When Psr-Mss4-GFP alone was expressed in an ipk1⌬ mss4⌬ double mutant, the profile was similar to that of ipk1⌬ cells, with elevated levels of IP 4 , IP 5 , and PP-IP 4 (Fig. 5, compare G and E). However, coexpression of both Psr-Mss4-GFP and Ste2⌬tail-Ipk1-GFP in the ipk1⌬ mss4⌬ double mutant resulted in minimal IP 4 , decreased IP 5 , and minimal PP-IP 4 levels and a significant increase in IP 6 levels (Fig. 5, compare G and H). The key observation was that IP 6 was now present. As shown in Fig. 6C, these results suggest that both IP 3 and IP 5 can diffuse between the cytoplasm and the nucleus.
Cytoplasmic IP 6 Production Is Sufficient for Gle1 Function-To understand whether the cytoplasmically restricted Ste2⌬tail-Ipk1-GFP fusion protein is functional with regard to facilitating Gle1-mediated mRNA export, we tested whether the expression of Ste2⌬tail-Ipk1-GFP would complement the gle1-2 ipk1-4 synthetic lethal phenotype. To maintain viability, the gle1-2 ipk1-4 strain harbors a URA3 plasmid expressing wild-type GLE1. An empty TRP1/CEN vector and TRP1/CEN plasmids expressing ste2⌬tail-GFP or ste2⌬tail-IPK1-GFP were transformed into the gle1-2 ipk1-4 mutant. Complementation was assayed by growth on 5-FOA medium. Growth was not observed for the strains containing the empty vector or the ste2⌬tail-GFP plasmid (Fig. 7). However, the strain expressing Ste2⌬tail-Ipk1-GFP was viable, indicating that the synthetic lethality was rescued. Thus, Ipk1 does not have to enter the nucleus to positively impact the role of Gle1 within the mRNA export pathway. These results further indicate that IP 6 produced in the cytoplasm by Ste2⌬tail-Ipk1-GFP is sufficient for Gle1 function. DISCUSSION Here, we have further investigated how IP 6 is involved in mRNA export and addressed the role for Ipk1 localization in IP 6 production. By combining the ipk1⌬ mutation with a panel of mutant genes encoding Nups or other factors involved in nucleocytoplasmic transport, we have revealed a network of Production of substrate IP 5 by Ipk2 is presumably localized in the nucleus (31). As such, in the Ste2⌬tail-Ipk1-GFP cells, IP 5 would need to diffuse across the NPC to allow IP 6 production. In the Ste2⌬tail-Ipk1-GFP/Psr-Mss4 cells (C), both IP 3 and IP 5 would need to diffuse. In A, the site of IP 3 production could be either cytoplasmic or nuclear. In B, endogenous Ipk1 is localized in the nucleus and at the nuclear rim (29), and IP 6 may be produced in the nucleus or the cytoplasm. Although Ste2⌬tail-Ipk1-GFP and Psr-Mss4 cannot enter the nucleus, a fraction of the fusion proteins are present at the endoplasmic reticulum/outer NE and may allow generation of IP 6 near the cytoplasmic face of the NPC (Fig. 4) (42). specific functions influenced by IP 6 production. These include NUP42, GLE2, DBP5, NUP159, and NUP116. The fact that each of these genes has known roles in mRNA export and connections to GLE1 function strengthens our conclusion that IP 6 is involved in regulating events required for Gle1-mediated trafficking of mRNA from the nucleus. Furthermore, the encoded proteins are all linked to a substructure on the NPC cytoplasmic face (Fig. 1A). These results, combined with the ability of Ste2⌬tail-Ipk1-GFP in the cytoplasm to produce IP 6 and complement a gle1-2 ipk1-4 mutant, provide evidence that IP 6 mediates steps in the mRNA export mechanism performed by Gle1, Gle2, or Dbp5 potentially at the cytoplasmic side of the NPC.
In this study, in addition to Gle1, the functions of Nup42, Nup116, Nup159, Gle2, and Dbp5 have now also been linked to requirements for IP 6 production. If these various protein associations perform complementary functions at the cytoplasmic face of the NPC, IP 6 might be involved in only one function, and a lack of IP 6 could increase the dependence on the others to allow cell viability. Alternatively, IP 6 may play roles in the functions of multiple proteins. For example, Nup116 has three defined functional regions: 1) an N-terminal Gle2-binding domain, 2) a central GLFG repeat domain for Kap and Mex67 binding, and 3) a C-terminal Nup82-binding region for NPC targeting. The nup116⌬ ipk1⌬ growth defect was not rescued by Nup116 proteins that independently lack either the N-or C-terminal regions. Thus, discrete regions of Nup116 recruit components of the mRNA export pathway that are each influenced by IP 6 . The role of IP 6 may be related to the integrity of the physical interactions between Nups at the NPC, between mRNA export factors, and/or between Nups and recruited mRNA export factors. The absence of either IP 6 or Nup116 causes some defect in the function of different aspects of such subcomplexes, but the combination of these effects results in a more severe phenotype. Interestingly, ipk1⌬ did not show a fitness defect with the nup82⌬108 mutant. This seems to indicate that the defect caused by nup82⌬108 is independently more severe (with mislocalization effects on both Nup159 and Nup116 (66, 69)) and not further compromised by removal of IP 6 .
The roles for IP 6 and Gle1 in mRNA export appear to be independent of the FG repeat domains of the Nups. NUP42, NUP159, and NUP116 belong to the FG domain family of Nups, and these domains serve as binding sites for shuttling Kaps and for Mex67 (reviewed in Ref. 68). However, we found that the FG domain of Nup42 and the GLFG domain of Nup116 were not required to complement the respective nup42⌬ ipk1⌬ and nup116⌬ ipk1⌬ phenotypes. The lack of genetic interaction between ipk1⌬ and mex67 or kap mutants also suggests that the role of IP 6 is at an mRNA export step independent or downstream of FG domain-mediated interactions. IP 6 likely regulates the other unique regions of the proteins rather than FG domain function. Specifically absent from the list of genes linked to IPK1 are those whose gene products reside exclusively at the nuclear face of the NPC, such as NUP60 and NUP2, or those that serve primarily a structural role, such as NIC96 and POM152 (71).
The cytoplasmic production of IP 6 by Ste2⌬tail-Ipk1-GFP has interesting implications for the physiology of inositides in cells. In particular, these results give us a new understanding of the diffusive capacity of endogenously produced soluble inositol polyphosphates (Fig. 6). IP 5 (and presumably, the other highly charged inositol polyphosphates) appears capable of freely dispersing throughout cells. Although unable to enter the nucleus, Ste2⌬tail-Ipk1-GFP had access to its substrate, IP 5 , generated by nuclear localized Ipk2 (Fig. 6). Moreover, in the cells with forms of both Ipk1 and Mss4 restricted to the cytoplasm, the production of IP 6 suggests that IP 3 is also diffusive. With a resting diameter of ϳ9 nm, the NPC aqueous channel is thought to allow diffusive passage of macromolecules up to 40 -50 kDa (72,73). Thus, these soluble inositides, in free form or in molecular assemblies below this limit, likely distribute between the nucleus and the cytoplasm by passive diffusion through NPCs. Although the fraction of free versus protein-bound inositides in the cell is unknown, these results also suggest that a significant fraction of IP 3 and IP 5 is either free or in small complexes.
These studies have also clarified the spatial relevance of IP 6 production for regulation of mRNA export. Surprisingly, Ste2⌬tail-Ipk1-GFP that was excluded from the nuclear compartment produced IP 6 and rescued the gle1-2 ipk1-4 synthetic lethal phenotype. This indicates that nuclear production of IP 6 is not necessary for its action in Gle1-mediated mRNA export and that cytoplasmic production is sufficient. This cytoplasmic production could result either from the predominant plasma membrane-localized Ste2⌬tail-Ipk1-GFP pool or from the low levels of Ste2⌬tail-Ipk1-GFP presumably on the outer NE. Moreover, that IP 3 and IP 5 are likely able to cross the NE infers that IP 6 may also freely cross. Thus, cytoplasmically produced IP 6 may diffuse into the nucleus to execute its function. This is especially important given that Gle1 is present on both the NPC nuclear and cytoplasmic faces (Fig. 3) (1) and that Dbp5 has been linked to nuclear functions (8). Our genetic results showing explicit linkages between IPK1 and genes encoding proteins with exclusively NPC cytoplasmic localizations (NUP42 and NUP159) provide independent evidence for the hypothesis that IP 6 regulates mRNA export events occurring at the NPC cytoplasmic face. Overall, the location of IP 6 production does not seem to be the most important parameter. The presence of IP 6 for interaction with specific NPC components and associated factors is the more critical factor.
The capacity for soluble inositide movement between the nucleus and the cytoplasm greatly expands the scope for IP 5 and IP 6 in mediating cellular functions besides intranuclear transcription, mRNA export, DNA repair, and chromatin remodeling. This also correlates with the products of additional IP 6 phosphorylation (IP 7 and IP 8 ) having roles in proper endocytic trafficking (74). IP 6 could be converted in either the cytoplasm or the nucleus, and IP 7 and IP 8 may also have diffusive properties. Further study of how these soluble inositides regulate discrete functions without being spatially produced/confined will be critical for understanding both the inositol polyphosphate signaling pathway and the cellular processes they regulate.