|Home | About | Journals | Submit | Contact Us | Français|
The assembly, distribution, and functional integrity of nuclear pore complexes (NPCs) in the nuclear envelope (NE) are key determinants in the nuclear periphery architecture. However, the mechanisms controlling proper NPC and NE structure are not fully defined. We used two different genetic screening approaches to identify Saccharomyces cerevisiae mutants with defects in NPC localization. The first approach examined green fluorescent protein (GFP)-Nic96 in 531 strains from the yeast Tet-promoters Hughes Collection with individual essential genes expressed from a doxycycline-regulated promoter (TetO7-orf). Under repressive conditions, depletion of the protein encoded by 44 TetO7-orf strains resulted in mislocalized GFP-Nic96. These included STH1, RSC4, RSC8, RSC9, RSC58, ARP7, and ARP9, each encoding components of the RSC chromatin remodeling complex. Second, a temperature-sensitive sth1-F793S (npa18-1) mutant was identified in an independent genetic screen for NPC assembly (npa) mutants. NPC mislocalization in the RSC mutants required new protein synthesis and ongoing transcription, confirming that lack of global transcription did not underlie the phenotypes. Electron microscopy studies showed significantly altered NEs and nuclear morphology, with coincident cytoplasmic membrane sheet accumulation. Strikingly, increasing membrane fluidity with benzyl alcohol treatment prevented the sth1-F793S NE structural defects and NPC mislocalization. We speculate that NE structure is functionally linked to proper chromatin architecture.
The nuclear envelope (NE) double lipid bilayer is a defining feature of the eukaryotic cell, imparting spatial separation between the nuclear chromatin and the cytoplasm. As such, knowing how communication across the NE is mediated will be critical to resolving regulation of gene expression and nucleocytoplasmic signaling. Nuclear pore complexes (NPCs) constitute the site of exchange for all macromolecules between the nucleus and cytoplasm. Each NPC spans a NE pore and consists of a central channel, cytoplasmic and nuclear ring structures, cytoplasmic fibrils, and a nucleoplasmic basket-like structure (Beck et al., 2004 ). The composition of the metazoan and budding yeast NPC has been analyzed by multiple groups, and overall both are built from a similar complexity of ~30 total conserved proteins, referred to as nucleoporins (Nups) and pore membrane proteins (Poms) (Rout et al., 2000 ; Cronshaw et al., 2002 ; reviewed in Tran and Wente, 2006 ). Some Nups are present exclusively on one face of the NPC, and others on both faces (Rout et al., 2000 ; Fahrenkrog and Aebi, 2003 ). Recent studies have revealed connections between nuclear face Nups and chromatin (reviewed in Capelson and Hetzer, 2009 ) and between NE dynamics and NPCs (Scarcelli et al., 2007 ). Understanding the structural organization and biogenesis of the NE and NPCs is required to more fully define functional events at the nuclear periphery.
In higher eukaryotes, NPCs assemble at the end of an open mitosis as the NE reforms (Hetzer et al., 2005 ). Importantly, NPCs also are generated de novo in the existing NE during interphase with the number of NPCs nearly doubling (Maul et al., 1971 ). In organisms with a closed mitosis, such as the budding yeast Saccharomyces cerevisiae, an intact NE is maintained throughout the entire cell cycle and all NPC biogenesis requires de novo insertion into this preexisting NE (Winey et al., 1997 ). Therefore, the NE must be plastic and dynamic for these de novo events of NPC assembly, while simultaneously functioning to preserve the structural integrity of the nucleus. Remarkably, the NE in S. cerevisiae lacks the structural support provided by the nuclear lamins in metazoans and still retains a spherical nuclear shape with a nonrandom distribution of NPCs (Winey et al., 1997 ).
Recent evidence suggests that several factors converge to control NE dynamics at sites of de novo NPC assembly. Such new NPCs arise by insertion and not by the duplication and division of existing NPCs (D'Angelo et al., 2006 ). First, reorganization and fusion of the NE to form a pore is probably initiated from both sides of the double membrane by the Poms: Pom34, Pom152, and Ndc1 in S. cerevisiae and Pom121, gp210, and Ndc1 in higher eukaryotes (Aitchison et al., 1995 ; Lau et al., 2004 ; Antonin et al., 2005 ; Campbell et al., 2006 ; Madrid et al., 2006 ; Mansfeld et al., 2006 ; Miao et al., 2006 ; Stavru et al., 2006 ; Dawson et al., 2009 ; Onischenko et al., 2009 ). Second, several Nups with predicted COPII/coatomer-like domains are implicated in stabilizing these pore membranes, including the yeast Nup84 (metazoan Nup107-160) subcomplex (Siniossoglou et al., 1996 ; Harel et al., 2003 ; Walther et al., 2003 ; D'Angelo et al., 2006 ; Devos et al., 2006 ; Drin et al., 2007 ; Hsia et al., 2007 ; Brohawn et al., 2008 ; Debler et al., 2008 ), yeast Nup53-Nup59 (metazoan Nup32) (Marelli et al., 2001 ; Hawryluk-Gara et al., 2008 ; Onischenko et al., 2009 ), and yeast Nup170-Nup157 (Flemming et al., 2009 ; Makio et al., 2009 ). Notably, Nup53-Nup59 and Nup170-Nup157 also have discrete connections to the Poms. Nup53-Nup59 interact physically with Ndc1 (Mansfeld et al., 2006 ; Onischenko et al., 2009 ) and genetically with Pom34 (Miao et al., 2006 ); whereas Nup170-Nup157 exhibits both genetic and physical interactions with Pom34 and Pom152 (Aitchison et al., 1995 ; Tcheperegine et al., 1999 ; Miao et al., 2006 ; Flemming et al., 2009 ; Makio et al., 2009 ). Known to maintain endoplasmic reticulum (ER) tubules (De Craene et al., 2006 ; Voeltz et al., 2006 ; Hu et al., 2008 ), yeast RTN1 and YOP1 also have genetic linkages to both the POMs and genes encoding the yeast Nup84 subcomplex (Dawson et al., 2009 ). Moreover, loss of Rtn1 and Yop1 results in dramatic alterations of NPC morphology and localization and reduced pore formation in vitro. These discoveries underscore the importance of controlling NE dynamics for NPC assembly.
Several ER/NE integral membrane proteins that affect NE composition or fluidity also impact NPC structure. NPCs are mislocalized into NE herniations in brr6 and apq12 mutants (de Bruyn Kops and Guthrie, 2001 ; Scarcelli et al., 2007 ), and the membrane fluidizing agent benzyl alcohol rescues the apq12 phenotype (Scarcelli et al., 2007 ). Interestingly, flares of NE-containing NPCs develop in yeast strains lacking the Spo7/Nem1 holoenzyme, a negative regulator of phospholipid synthesis (Siniossoglou et al., 1998 ; Campbell et al., 2006 ). These NE/NPC flares expand directly from the NE region nearest the nucleolus, suggesting that both phospholipid composition and chromatin interactions impact NE and NPC dynamics.
For postmitotic NE and NPC assembly, recent studies have suggested that the chromatin-associated factor MEL-28/ELYS is required for Nup107-160 complex targeting (Rasala et al., 2006 ; Franz et al., 2007 ; Gillespie et al., 2007 ; Liu et al., 2009 ). The AT-rich hook of MEL-28/ELYS binds to AT-rich chromatin, and Nup107-160 binding facilitates recruitment of vesicles containing Pom121 and Ndc1 (Rasala et al., 2008 ). This might reflect the recruitment of Nups to condensed chromatin and formation of a “prepore” structure. Moreover, such prepores could trigger nuclear pore formation coincident with postmitotic NE reformation (Anderson and Hetzer, 2008 ). A similar requirement for Nup–chromatin interactions in biogenesis during de novo NPC insertion into intact NEs has not been reported.
Here, we used a combination of innovative genetic approaches in S. cerevisiae to comprehensively assess the role of essential factors in NPC localization, structure, and potentially assembly into the NE. The genes identified encode factors involved in nuclear transport, chromatin remodeling, secretion, lipid anchoring, protein degradation, and lipid biosynthesis. Strikingly, multiple components of the RSC chromatin remodeling complex were identified including the essential ATPase catalytic subunit Sth1 (Du et al., 1998 ). In S. cerevisiae, the RSC complex is composed of 15 subunits, several of which are essential for cell viability (Cairns et al., 1996 ; Martens and Winston, 2003 ; Sahaa et al., 2006 ). Although RSC was first identified for its roles in chromatin remodeling and has been linked to transcriptional activation and inhibition (Cairns et al., 1996 ; Angus-Hill et al., 2001 ; Damelin et al., 2002 ; Ng et al., 2002 ; Kasten et al., 2004 ; Soutourina et al., 2006 ), RSC has also been linked to a wide range of chromatin-based functions such as kinetochore function and cohesin association (Hsu et al., 2003 ; Baetz et al., 2004 ; Huang et al., 2004 ) and double-strand break repair with the DNA damage response (Chai et al., 2005 ; Shim et al., 2005 , 2007 ; Liang et al., 2007 ). Several reports suggest connections between NPCs and RSC. A nup84Δ rsc7Δ double mutant is synthetically lethal (Wilson et al., 2006 ), and an rsc9 mutant has altered Kap121-GFP localization (Damelin et al., 2002 ). In this report, we present evidence for the role of the RSC complex in maintaining proper NE and NPC structure.
All S. cerevisiae strains used in this study are listed in Table 1. The original npa18-1 strain (SWY3201) was backcrossed with the parental strain SWY2090 to yield SWY3202 (temperature sensitive at 34°C and GFP-Nup mislocalization). A LEU2/CEN library (American Type Culture Collection, Manassas, VA) was transformed into the SWY3202 strain, and colonies were incubated at the permissive temperature, 23°C, for 36 h and then shifted to 34°C. Plasmid DNA was recovered from each resulting colony and analyzed by restriction digest. The library plasmid inserts from two independent isolates were sequenced. The minimal overlapping region harbored only two complete open reading frames (ORFs), STH1 and YIL127C. Wild-type STH1 and YIL127C, with respective flanking promoter regions, were independently subcloned into the XbaI and XhoI sites of pRS315 (Sikorski and Hieter, 1989 ) by polymerase chain reaction (PCR) amplification using library plasmid template and the following forward and reverse primers, respectively: STH1, 5′-CAAGTCTAGACCTGTCGATTAACTGAGC-3′ and 5′-GTAACTCGAGCTAGAAAGAGTATTAGAGG-3′ and YIL127C, 5′-ACGTTCTAGACGAACAACTTAAGGAGGGAG-3′ and 5′-GCAACTCGAGTTCACATTGATGAGCACGTG-3′. The resulting pSTH1 (pSW3051) and pYIL127C (pSW3049) plasmids were transformed into SWY3202. To analyze the sth1 allele in SWY3202, genomic DNA from the mutant strain was amplified using STH1 flanking oligonucleotides and the high-fidelity polymerase Pfu (Stratagene, La Jolla, CA). Products from two independent PCR reactions were purified and sequenced.
All strains were cultured in either rich (YPD: 1% yeast extract, 2% peptone, and 2% dextrose) or synthetic minimal (SM) media lacking appropriate amino acids and supplemented with 2% dextrose. All yeast genetic techniques and molecular cloning were performed according to standard procedures (Sherman et al., 1986 ; Sambrook et al., 1989 ). Cell viability assays were performed on treated and untreated sth1-F793S and the TetO7-STH1 mutant strains. After growth under permissive and nonpermissive conditions (3 and 12 h, respectively), the mutant strains were plated onto YP plates at 100 cells per plate, incubated at 23°C for 2 d, and quantified for colony-forming units. Serial dilutions of mid-log phase W303, SWY4143, S288C, and BLY49 were spotted onto YP plates supplemented with 2% glucose, 2% galactose, 2% raffinose or 2% ethanol/2% glycerol. These strains were also spotted onto YPD plates containing thiabendazole (TBZ; 60 μg/ml) or hydroxyurea (HU; 50 mM). The plates were imaged after 3 d incubation at the semipermissive temperatures of the respective mutant alleles. Multicopy suppressor plasmids from were obtained from the Yeast Genomic Tiling Collection through Open Biosystems (Huntsville, AL) (Jones et al., 2008 ).
The yeast Tet-promoters Hughes Collection (referred to here as the TetO7-orf strain collection) was obtained from Open Biosystems (Mnaimneh et al., 2004 ). This collection contains 813 strains of the 1105 reported total essential genes. By a series of strain crosses and selections, GFP-nic96 was incorporated into each TetO7-orf strain that was reported as having a slow growth phenotype on doxycycline. Strain Y3656 was crossed with SWY2090 (Table 1). The resulting strain, SWY3191, was crossed with strains from the TetO7-orf strain collection. Strains were mated on YPD for a minimum of 6 h, and diploids were selected by pinning three successive times onto SM Lys−His− media. For sporulation, strains were incubated on YPD for 15 h at 30°C and then transferred by pinning to SPO media (1% potassium acetate, 0.1% yeast extract, 0.05% glucose, 14 mg/l histidine, and 71 mg/l leucine). Diploids were allowed to sporulate at 23°C for at least 4 d. MATα haploids were selected by streaking each strain to SM Arg−Leu−Can+ (60 mg/l canavanine sulfate) media. Strains with the TetO7 promoter were selected by streaking on YPD media containing G418 (200 μg/ml active units). Strains expressing the tetracycline transactivator (tTA) and GFP-nic96 were further identified by growth on SM Ura−His−Leu− media. Resulting strains had the genotype MATα can1Δ::MFA1pr-HIS3::MFαpr-LEU2 GFP-Nic96:HIS3 URA3::CMV-tTA gene::kanR-tetO7-TATA leu2 his3 (LYS or lys; TRP or trp; ADE2 or ade2-1::ADE2:ura3). Some GFP-nic96 TetO7-orf strains were not obtained due to apparent technical difficulties with incorporating GFP-nic96 into the given background.
GFP-Nic96 localization was screened visually in 531 GFP-nic96 TetO7-orf strains after growth in doxycycline containing media. Specifically, the strains described as having constitutive slow growth (CSG) or having a weak, moderate, or severe growth defect in media containing 10 μg/ml doxycycline (Table 2) were inoculated directly into YPD media containing 10 μg/ml doxycycline and cultured overnight (13–15 h) at 30°C. For strains with a growth phenotype described as “very severe” or “very severe/(almost) no growth on doxycycline” (Mnaimneh et al., 2004 ), log phase cultures in YPD were treated with 10 μg/ml doxycycline for ~5 h. Some of the strains with “very severe” growth defects grew sufficiently in the presence of doxycycline overnight, and were screened under these conditions.
Yeast strains with GFP-tagged Nups were examined from cultures by direct fluorescence microscopy. For cycloheximide, thiolutin, and benzyl alcohol experiments, logarithmically growing cultures were treated with 10 μg/ml cycloheximide, 3 μg/ml thiolutin, or 0.4% benzyl alcohol and then temperature shifted for 5 h at 34°C or treated with 10 μg/ml doxycycline for 8 to 12 h. Cell cycle arrest experiments included a 2 d preincubation with nocodazole (15 μg/ml) followed by a 3 h shift to 34°C. Arrest was monitored with quantification of the percentages of G2-arrested cells in treated and untreated cultures, both before and after the temperature shift. For indirect immunofluorescence microscopy, cells from logarithmically growing cultures were pelleted; fixed for 10 min at room temperature with 3.7% formaldehyde, 10% methanol in 100 mM potassium phosphate, pH 6.5; and processed as described previously (Wente et al., 1992 ). Samples were incubated with affinity-purified, rabbit anti-Nup116 C-terminal polyclonal antibody (Iovine et al., 1995 ) (1:50). Bound antibody was detected by incubation with Alexa 594 goat anti-rabbit secondary antibody (1:400). Additional samples were incubated with mouse anti-Nup159 monoclonal antibody (1:10; a gift from G. Blobel and M. Rout (Rockefeller University, New York, NY), and bound antibody was detected with Alexa 594 goat anti-mouse secondary antibody (1:200).
A final stain for 5 min with 0.1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) in phosphate-buffered saline (PBS), 1% bovine serum albumin was conducted before mounting onto slides with 90% glycerol, 1 mg/ml p-phenylenediamine, and PBS, pH 9.0. Light microscopy was performed with a BX50 microscope (Olympus, Tokyo, Japan) with a UPlanF1 100×/1.30 oil immersion objective. Images were collected with a CoolSNAP HQ camera and MetaVue version 4.6 software (Photometrics, Tucson, AZ) and processed with Photoshop 9.0 software (Adobe Systems, Mountain view, CA). For electron microscopy, 2 × 108 logarithmically growing cells were harvested from the specific culture conditions and processed as described previously (Wente and Blobel, 1993 ). Samples were analyzed on a CM-12 120-keV electron microscope (FEI, Hillsboro, OR). Images were acquired with an Advantage HR or MegaPlus ES 4.0 camera (Advanced Microscopy Techniques, Danvers, MA) and processed with Photoshop 9.0 software.
Cells were prepared as described previously (Ryan and Wente, 2002 ), except that 20 μl of cell suspension was used for each assay. Strains assayed included SWY2089 (parental), SWY3378 [sth1-F793S (npa18-1)], SWY2324 [sec13-G176R (npa2-1)], and SWY2325 [sec23-S383L (npa1-1)]. The percentage of activity in each sample was calculated relative to the activity of the wild-type control strain. All assays were performed on three replicate cultures.
Cultures were grown to early log phase at 23°C and then shifted to growth at 34°C in the presence or absence of 0.4% benzyl alcohol. Total cell lysates were prepared by bead beating in lysis buffer (20 mM Tris, pH. 6.5, 5 mM MgCl2, 2% Triton X-100, and 150 mM NaCl) and resolved by SDS-polyacrylamide gel electrophoresis (PAGE). The blots were incubated with either affinity purified rabbit anti-Dbp5 polyclonal antibody (1:1000; Bolger et al., 2008 ) (as a loading control) or a rabbit anti-Sth1 polyclonal antibody (1:100; Saha et al., 2002 ), followed by incubation with horseradish peroxidase-conjugated anti-rabbit antibodies (Jackson ImmnoResearch Laboratories, West Grove, PA) and detection via SuperSignal West Pico enhanced chemiluminescence substrate (Pierce Chemical, Rockford, IL).
Cells were grown to early log phase and shifted to 34°C with the addition of thiolutin (3 μg/ml). After 3 h, cells were rinsed with ice-cold sterile water and frozen in liquid nitrogen. RNA was isolated from equivalent cell numbers with hot phenol (Geng and Tansey, 2008 ). Oligo(dT) reverse-transcription was performed with TaqMan reverse-transcription kit (Applied Biosystems, Foster City, CA), and quantitative PCR was performed in triplicate using iCycler and iQ SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA). The comparative CT method was used to quantify fold changes in NUP-GFP transcripts relative to ACT1. Gene-specific primers for GFP and ACT1 were validated across 6 logs of input cDNA: ACT1, 5′-CTCCACCACTGCTGAAAGAGAA-3′ and 5′-CGAAGTCCAAGGCGACGTAA-3′ and GFP, 5′-AGTGGAGAGGGTGAAGGTGA-3′ and 5′-GTTGGCCATGGAACAGGTAG-3′.
To identify essential factors required for NPC localization, structure, and/or assembly, we designed a genetic screening approach in the budding yeast S. cerevisiae. The rationale for the screen was based on extensive genetic evidence showing that mutants with defects in NPC assembly or stability have GFP-Nup mislocalization (Bucci and Wente, 1998 ; Ryan and Wente, 2002 ; Ryan et al., 2003 , 2007 ; Madrid et al., 2006 ; Miao et al., 2006 ). This can be due to the inability of the GFP-Nup to incorporate into newly forming NPCs or the disassembly of existing NPCs. We hypothesized that the genes encoding regulators of the essential NPC structure would themselves be essential for viability. A collection of yeast strains has been generated wherein 813 of the 1105 reported essential genes in S. cerevisiae were individually placed under the control of a doxycycline-regulated promoter, TetO7 (Mnaimneh et al., 2004 ). The TetO7-promoter allows regulated transcription of the respective gene (orf) with specific repression in the presence of doxycycline. The availability of this collection enabled the design of a direct genome-wide strategy to analyze the effective null or hypomorph phenotype of known essential genes for defects in NPC structure/assembly.
To conduct the screen, a GFP-tagged allele of the essential nucleoporin NIC96 (GFP-nic96) was systematically incorporated into individual doxycycline-sensitive strains of the yeast TetO7-orf strain collection (see Materials and Methods). Specifically, the screen used only the TetO7-orf strains with a reported slow growth phenotype in the presence of doxycycline (Mnaimneh et al., 2004 ). Perturbations in growth rate indicated that the essential gene was indeed down-regulated. We speculated that if the gene played a role in NPC structure/assembly, then the GFP-Nic96 localization should be perturbed when the given TetO7-orf strain was grown in doxycycline. The resulting GFP-nic96 TetO7-orf strains were individually examined for GFP-Nic96 localization based on direct fluorescence microscopy of live cells. Strains were cultured in the presence of doxycycline for 5 h or overnight. In total, GFP-Nic96 localization was evaluated in 531 strains and compared with that in a parental control strain without a TetO7-orf. GFP-Nic96 localization was scored as wild type if the fluorescent signal was detected at the nuclear rim and as mislocalized if all or a portion of the fluorescent signal was not at the nuclear rim. Mislocalization phenotypes were further ranked as weak, moderate, or severe. In addition, some strains were scored as having speckles (small foci of fluorescent signal in the cytoplasm) or as having foci/clusters of fluorescent signal at the nuclear rim.
We identified 44 TetO7-orf strains with mislocalized GFP-Nic96 and/or distorted nuclear rim structure (Figure 1A and Table 2). Based on functional analysis in published studies, these genes were classified into eight major categories. This included genes encoding known Nups as well as factors required for nuclear transport (Ran/Kap), chromatin remodeling, secretion, protein degradation, glycosylphosphatidyl inositol (GPI) anchoring, and lipid biosynthesis. Previous studies have also documented NPC and NE perturbations in mutants with defective Nups/Poms (Wente and Blobel, 1993 ; Bogerd et al., 1994 ; Doye et al., 1994 ; Wente and Blobel, 1994 ; Aitchison et al., 1995 ; Heath et al., 1995 ; Siniossoglou et al., 1996 ; Kosova et al., 1999 ; Madrid et al., 2006 ; Miao et al., 2006 ), secretion factors (Nanduri et al., 1999 ; Nanduri and Tartakoff, 2001 ; Ryan and Wente, 2002 ), lipid biosynthetic enzymes (Schneiter et al., 1996 ), the RanGTPase cycle (Ryan et al., 2003 ), and Kap95 (Ryan et al., 2007 ). A small subset of the components known to affect NPC structure or assembly were not identified by our screen, including the Nups NDC1, NUP1, NUP159, and NUP192, as well as the RAN cycle members NTF2 and RNA1. KAP95 and KAP121 were unresponsive to doxycycline treatment, whereas PRP20 and GSP1 were absent from the collection; therefore, these candidates were not included in the screen data set.
Interestingly, the screen identified genes encoding several essential components of the RSC chromatin remodeling complex: STH1, RSC8, RSC58, and ARP9. RSC4, RSC9, and ARP7 were also identified after direct testing. Each of these strains showed GFP-Nic96 mislocalization to varying extents (Figure 1B and Table 2), which generally correlated with the growth defect of the strain in doxycycline-containing media. The level of growth in the presence of doxycycline is thought to reflect the level of transcriptional repression for the respective TetO7-orf (Mnaimneh et al., 2004 ). Mislocalization and growth defects were severe in the TetO7-RSC58, TetO7-RSC8, and TetO7-STH1 strains. Mislocalization of GFP-Nups in TetO7-STH1 cells was first apparent after 6 h of culturing in the presence of doxycycline. This mislocalization became more extensive after 12 h and was detected in >90% of the cells. At this time point, viability assays confirmed that mislocalization was not an indirect effect of doxycycline toxicity or cell death (data not shown).
To further analyze the localization of NPC proteins in the TetO7-orf strains for the RSC complex, the respective strains were processed for indirect immunofluorescence microscopy for Nup116 (Figure 1C). The TetO7-RSC8, TetO7-RSC58, and TetO7-STH1 strains showed severe mislocalization of Nup116 when grown in the presence of doxycycline. The TetO7-RSC4 and TetO7-RSC9 strains were again less markedly altered. Defects in NPC structure/assembly have not been documented previously in RSC complex mutants. STH1 encodes the essential ATPase catalytic subunit of the RSC complex, whereas RSC4, RSC8, RSC9, and RSC58 encode core or accessory RSC complex components (Sahaa et al., 2006 ). Overall, this genome-wide screening strategy identified several essential RSC components that were required for normal Nup localization.
Previously, in an independent approach for identifying factors required for NPC structure/assembly, we conducted a visual screen for temperature-sensitive strains with defective GFP-Nic96 and Nup170-GFP localization (Ryan and Wente, 2002 ; Ryan et al., 2003 , 2007 ). This screen isolated 121 NPC assembly (npa) mutant strains in numerous complementation groups, including those with defects in secretion factors, Ran-cycle factors, and Kap95. Here, we selected one unidentified npa complementation group, npa18, to further characterize. The npa18-1 mutant showed some GFP-Nic96/Nup170-GFP mislocalization at 23°C, and had severe mislocalization at the nonpermissive temperature (34°C) (Figure 2A). The GFP-Nic96/Nup170-GFP signal was no longer localized around the nuclear rim; instead, the fluorescent signal was detected in large, nonuniform foci throughout the cytoplasm and surrounding the nucleus. This mislocalization was first observed after 3 h at 34°C in ~40% of cells (data not shown) and was maximal by 5 h. Cell viability assays found that mislocalization was not due to cell death. Indirect immunofluorescence detection of Nup116, Nup159, and Pom152 also showed similar mislocalization (Figure 2B and Supplemental Figure S1). Thus, multiple distinct Nup subcomplexes were perturbed in the npa18-1 mutant.
Backcrossing the npa18-1 mutant with the parental strain revealed 2:2 linked segregation of temperature sensitivity and GFP-Nup mislocalization. This indicated that the defects were due to the mutation of a single gene. To identify the mutated gene, a yeast CEN genomic library was used to select for complementation of the recessive temperature-sensitive phenotype. The inserts from two unique plasmids that rescued the temperature-sensitive growth defect were isolated from yeast and sequenced. Both contained nucleotide sequence corresponding to a portion of chromosome IX that contained the complete ORF for STH1 and a putative ORF YIL127C. Expression of YIL127C alone did not complement the growth defect (Figure 2D). However, an expression plasmid with STH1 alone was necessary and sufficient for restoration of growth (Figure 2D). Furthermore, STH1 expression also restored nuclear rim localization of GFP-Nic96 and Nup170-GFP at 34°C (Figure 2C). Sequencing the chromosomal DNA from the npa18-1 mutant strain revealed a single point mutation in the STH1 nucleotide sequence, which resulted in a single amino acid substitution, F793S, in the ATPase domain. Thus, we designated this npa18-1 mutant as sth1-F793S and refer to it as such henceforth. Complementation analysis among the remaining unidentified npa mutant strains identified sth1-F793S as the only allele representing this npa18 complementation group.
Previous studies of STH1 have reported four temperature-sensitive sth1 alleles (sth1-1, sth1-2, sth1-3, and sth1-L1346A) (Du et al., 1998 ; Huang et al., 2004 ). The sth1-1, sth1-2, and sth1-3 alleles each have mutations in the sequence region corresponding to the ATPase domain, although distinct from the sth1-F793S allele. To determine whether these other sth1 alleles perturb Nup localization, we conducted indirect immunofluorescence microscopy for Nup116 localization. After 4 h at 37°C, Nup116 remained predominantly at the nuclear rim in each of these strains (Figure 3A), whereas Nup116 mislocalized under similar conditions in the strain expressing sth1-F793S (Figure 2B). Similar results were obtained after 9 h at 37°C, with only slight mislocalization of Nup116 detectable in cells expressing sth1-3 (data not shown). Therefore, the sth1-F793S allele had a specific effect on Nup localization.
We further characterized the sth1-F793S mutant by testing for whether known multicopy suppressors of sth1-3 allele also suppressed the temperature sensitive phenotype and Nup mislocalization of the sth1-F793S allele. Genes encoding members of the cell wall integrity pathway (MID2, RHO2, ROM2. PKC1, and WSC1) have been shown previously to multicopy suppress the temperature-sensitive growth phenotype of the sth1-3 allele (Chai et al., 2002 ). However, the growth defect (data not shown) and Nup60-GFP mislocalization in the sth1-F793S mutant were not rescued by overexpression of any of these genes (Supplemental Figure S3). Therefore, the sth1-F793S allele may be affecting distinct or multiple functions of RSC that are not compensated for by the cell wall integrity pathway alone.
Next, we compared the sth1-F793S allele and the sth1-3 allele for growth on different carbon sources and in the presence of TBZ (microtubule-depolymerizing agent) or HU (ribonucleotide reductase inhibitor) (Figure 3B). Although the parental strains of each mutant exhibit slightly different growth phenotypes, growth of the sth1-F793S mutant was dramatically enhanced on nonglucose carbon sources compared with both respective parental strains and to the sth1-3 mutant. The enhanced growth phenotype specific to the sth1-F793S mutant might be due to changes in transcription as a result of RSC depletion. Similar to the previously described effects on other sth1 mutant alleles (Koyama et al., 2002 ; Hsu et al., 2003 ), the sth1-F793S mutant showed enhanced sensitivity to HU, whereas TBZ was less effective on the sth1-F793S mutant (Figure 3B, bottom two rows). The allele-specific drug sensitivities indicate separable functions for RSC in double-strand break repair, microtubule function and kinetochore structure (Tsuchiya et al., 1998 ; Chai et al., 2002 , 2005 ; Shim et al., 2005 , 2007 ; Liang et al., 2007 ).
Given the similarities between the Nup mislocalization in the sth1-F793S and TetO7-sth1 mutants, we evaluated protein stability in the sth1-F793S cells by immunoblotting. Wild-type Sth1 protein levels were unchanged after shifting to growth at 34°C for 5 h; however, the sth1-F793S protein was not detectable after temperature shifting (Figure 3C). Others report that the sth1-3 protein is stable and has wild-type ATPase activity (Du et al., 1998 ). Thus, at 34°C, the sth1-F793S allele is an effective null with distinct cellular perturbations.
By the nature of our genetic screening strategies, all of the RSC components identified represented essential genes. To investigate other subunits, we directly examined the available null strains for nonessential RSC components (Supplemental Figure S1). Indirect immunofluorescence microscopy for anti-Nup116 and anti-GLFG Nups was conducted. Nups localized in a normal perinuclear punctate pattern in rsc1Δ, rsc2Δ, and rsc14Δ mutant cells. In htl1Δ cells, moderate mislocalization was detected after shifting to the nonpermissive temperature. Visual scanning of the Z-plane showed severe nuclear morphology perturbations coincident with the pattern of Nup mislocalization (Supplemental Figure S1). The most striking mislocalization was observed in the rsc7Δ mutant, where Nups were markedly redistributed to cytoplasmic foci after shifting to growth at the nonpermissive temperature (Figure 1D). Overall, multiple independent members of the RSC complex were linked to proper NPC localization.
To further investigate the NPC defects in these TetO7-RSC and sth1-F793S mutants, thin section transmission electron microscopy (TEM) was conducted. The sth1-F793S mutant and wild-type parental strains were evaluated before and after growth for 5 h at 34°C, whereas the TetO7-STH1 and TetO7-RSC58 strains were processed after 10 h of growth in the absence and presence of doxycycline. In the wild-type parental strain and before temperature shifting (data not shown) or doxycycline treatment, the nuclei, NEs, and NPCs of all the strains were not perturbed (Figure 4). In the control cells, the NPCs appeared as electron-dense structures spanning the NE of a single distinct nucleus (Figure 4, A, D, and G). In contrast, striking ultrastructural perturbations were observed in the temperature-arrested sth1-F793S cells (Figure 4, B and C) and the doxycycline-treated TetO7-STH1 (Figure 4, E and F) and TetO7-RSC58 cells (Figure 4, H and I). Relative to parental or control cells, in all three mutants, there was significant cytoplasmic membrane proliferation that seemed to originate from the ER and/or NE. Extensive sheets of membrane were present, often in multiple layers, around the cell periphery/plasma membrane, and in intertwined honeycombs. There was also an accumulation of distinct 40- to 50-nm cytoplasmic vesicles. The nucleus itself was often difficult to clearly identify. When an apparent nuclear cross section was observed, a few electron-dense structures representing NPCs were detected. The time frame after temperature or doxycycline shifting for the appearance of these ultrastructural defects was coincident with the Nup mislocalization defects described above (Figures 1 and and22).
As a test for defects in new NPC assembly versus perturbations in the stability of existing NPCs, we have previously assayed the effect of cycloheximide treatment on Nup mislocalization in npa mutants (Ryan et al., 2003 , 2007 ). Mutants that perturb preexisting factors or NPC components will not require translation for the phenotype and will show mislocalization in the presence of cycloheximide. In contrast, mislocalization due to perturbations in de novo NPC or NE biogenesis will require translation of assembly or structural factors for accumulation of perturbed GFP-Nups, and thus will not show GFP-Nup mislocalization in cycloheximide. This is true for the NPC assembly defects documented in the prp20-G282S (npa14-1), ntf2-H104Y (npa11-1), rna1-S116F (npa13-1), gsp1-P162L (npa15-1), kap95-E126K (npa16-1), and apq12Δ mutants (Ryan et al., 2003 , 2007 ; Scarcelli et al., 2007 ). In sth1-F793S (npa18-1) and rsc7Δ mutant cells treated with cycloheximide, the GFP-Nups remained associated in a predominantly nuclear rim localization after incubation at the nonpermissive temperature (Figure 5A). Marked mislocalization was not detected. Similarly, treatment of TetO7-RSC8 cells with cycloheximide during nonpermissive growth conditions also prevented Nup mislocalization (Figure 5B). These data indicate that the defects in the sth1-F793S, rsc7Δ, and TetO7-RSC8 mutant strains required ongoing translation.
Because the RSC complex is functionally linked to gene expression (Angus-Hill et al., 2001 ; Damelin et al., 2002 ; Ng et al., 2002 ; Kasten et al., 2004 ; Soutourina et al., 2006 ; Badis et al., 2008 ; Parnell et al., 2008 ; Hartley and Madhani, 2009 ; Mas et al., 2009 ), we speculated that some of the defects in the sth1-F793S mutant might be linked to altered expression of RSC-controlled genes that encode proteins involved in NE and/or NPC biogenesis. To globally assess the role of transcription in the sth1-F793S Nup mislocalization phenotype, we used a RNA polymerase II temperature-sensitive mutant. The RBP4 gene encodes a nonessential RNA polymerase II subunit (Woychik and Young, 1989 ); however, the rbp4Δ is temperature sensitive for growth above 32°C and after 45 min at 37°C, 96% of RNA polymerase II transcription is lost (Woychik and Young, 1989 ; Miyao et al., 2001 ). The sth1-F793S rbp4Δ double mutant was evaluated for NPC localization by monitoring GFP-tagged Nic96, Nup60, or Nup133 (Figure 6). After shifting to growth at 34°C for 5 h, the respective GFP-tagged Nups remained localized at the nuclear rim, and mislocalization was not detected. GFP-tagged Nups also remained rim localized in the rpb4Δ single mutant (data not shown). This observation was further confirmed using thiolutin, an inhibitor of global RNA synthesis. Treatment with thiolutin blocked GFP-tagged Nic96 mislocalization in TetO7-STH1 cells grown in the presence of doxycycline (Supplemental Figure S4) and GFP-tagged Nic96, Nup60, Nup133 mislocalization in the sth1-F793S mutant (data not shown). Together, both ongoing transcription and translation were required for the NPC/NE defects.
Control experiments were also conducted to assay for effects on mRNA stability in the sth1-F793S Nup mislocalization phenotype. Quantitative PCR was used to evaluate NUP and ACT1 relative mRNA levels between wild-type and sth1-F793S mutant cells. At the permissive growth temperature, NUP60-GFP and NIC96-GFP mRNA levels did not vary > 1.5-fold between wild-type and sth1-F793S cells. After a 3-h shift to 34°C in the presence of thiolutin, the NUP mRNAs examined were actually stabilized relative to ACT1 in the sth1-F793S cells (NUP60-GFP up to 5-fold and NIC96-GFP up to 21-fold). Therefore, the lack of Nup mislocalization upon transcriptional shutoff was not due to decreased mRNA stability of the NUP transcripts tested.
To evaluate whether the transcriptional and translational shut-off were acting indirectly to block Nup mislocalization by inhibiting sth1-F793S cell division, we tested for mislocalization in nocodazole-arrested cells. The sth1-F793S mutant was treated with 15 μg/ml nocodazole for 2 h, resulting in >90% of the cells as large budded and held in G2-M. At this time point, the cultures were shifted to 34°C for 3 h. The cell population remained at >65% large-budded/G2-M. Importantly, Nup60-GFP was mislocalized to the same level in both arrested and unarrested control cultures (Supplemental Figure S5). This suggested that Nup mislocalization in sth1-F793S cells does not require cell division and confirmed that the lack of mislocalization in the cycloheximide, rpb4Δ and thiolutin experiments is linked to inhibition of translation or transcription.
Nup mislocalization and NE/ER defects have been reported in mutants defective in the RanGTPase cycle (Ryan et al., 2003 ), in the COPII complex for ER/Golgi trafficking (Ryan and Wente, 2002 ), in NPC proteins (Doye and Hurt, 1995 ), in lipid biogenesis factors (Siniossoglou, 2009 ), and NE/ER membrane proteins (Scarcelli et al., 2007 ; Dawson et al., 2009 ). We also identified additional components in some of these pathways in the TetO7-orf screen reported here (Figure 1A and Table 2). To directly test for links to secretion in sth1-F793S cells, we assayed for secreted invertase activity. The sth1-F793S cells displayed 53% of wild-type invertase activity relative to our parental control strain. In comparison, sec23-S383L (npa1-1) and sec13-G176R (npa2-1) mutants had 3 and 30% of wild-type invertase activity levels, respectively. We also tested for genetic interactions between the sth1-F793S mutant and the sec13-G176R or sec23-S383L mutant alleles. Of note, a sth1-F793S sec13-G176R double mutant and the sth1-F793S sec23-S383L double mutant were both viable and showed no synthetic fitness defects (SWY3436 and SWY3437, Table 1). The same results were found for a sth1-F793S prp20-G282S double mutant that was viable and showed growth identical to the sth1-F793S mutant (SWY3409, Table 1). We concluded that the defects in the sth1-F793S mutant were not due to indirect severe perturbations on the levels of secretory or RanGTPase cycle factors.
We used an independent assay to investigate whether NE membrane composition or fluidity was connected to the sth1-F793S mechanism of perturbation. Benzyl alcohol (BA) is an established membrane fluidizer (Colley and Metcalfe, 1972 ; Gordon et al., 1980 ) that has recently been used in S. cerevisiae to examine the role of Apq12 in NPC assembly (Scarcelli et al., 2007 ) and in Aspergillus nidulans to analyze functional roles for the An-Nup84-120 complex at the NE (Liu et al., 2009 ). To test this with the sth1-F793S mutant, 0.4% BA was added to the cells coincident with the shift to the nonpermissive growth temperature. Nuclear rim localization of GFP-tagged Nic96, Nup170, Nup60, Nup133, and Pom34 were independently evaluated in respective strains by direct fluorescence microscopy (Figure 7). Strikingly, no Nup mislocalization was observed in the BA treated sth1-F793S cells. GFP-Nic96 was also not mislocalized when TetO7-STH1 cells were treated with BA during growth in the presence of doxycycline (Supplemental Figure S4). Moreover, TEM examination of the BA-treated, temperature-shifted sth1-F793S cells revealed that the ultrastructural NE defects were also absent (Figure 8). Immunoblotting was conducted and showed that the sth1-F793S protein was still unstable in the BA-treated cells (Figure 3C). Thus, the RSC role in mediating proper NE morphology and NPC localization was compensated for by alteration in NE dynamics.
In our independent TetO7-orf and npa genetic screens, we find that perturbation of Sth1 and several other RSC components results in altered Nup localization, perturbed NE organization and significant cytoplasmic membrane proliferation. The comparable phenotypes between the sth1-F793S (npa18-1), the TetO7-STH1, the TetO7-RSC, and the rsc7Δ mutant strains indicate that the Nup/NE perturbations result from RSC complex loss-of-function. This conclusion is further corroborated by the loss of detectable sth1-F793S protein at the nonpermissive temperature in the mutant strain. Such defects in NE/NPC structure have not been previously documented in RSC mutants. Others have found that the rsc7(npl6) mutant allele leads to defective localization of nuclear proteins and also have reported a genetic interaction between rsc7 and nup84 mutants (Bossie and Silver, 1992 ; Damelin et al., 2002 ; Wilson et al., 2006 ). We speculate that the RSC complex mutant phenotypes reflect a functional connection between proper chromatin remodeling and NE/NPC structure.
On a more general level, we have demonstrated the utility of the TetO7-orf collection for GFP-based screening of perturbations in specific cell functions. Our prior npa mutant screen was not to saturation, and it would be technically challenging to achieve full genomic coverage based on the number of genes we have found with indirect perturbations in NE/NPC structure (e.g., the secretory pathway; Ryan and Wente, 2002 ). Taking the TetO7-orf and npa screens together, we have now repeatedly identified genes in the same functional classes, indicating a nearly comprehensive assessment of the role of essential factors. In this study, we have further identified components of the lipid biosynthesis and secretory pathways for proper Nup localization. Others have shown that mutation of FAS3/ACC1, a gene required for long-chain fatty acid synthesis, results in NE/NPC defects (Schneiter et al., 1996 ). The same lipid–membrane effects might be the basis for the TetO7-LCB2, TetO7-FAS2, and TetO7-CDS1 defects in GFP-Nic96 localization. We also identified connections here to the proteasome and enzymes required for GPI anchoring. Future analysis of the NE and NPC defects in these mutants could give insight into the mechanisms by which the global nuclear architecture is coordinated and regulated.
Our results with the RSC complex mutants also potentially impact on prior interpretations of RSC-associated functions. Multiple studies have shown that RSC functions in DNA double-strand break repair (Chai et al., 2005 ; Shim et al., 2005 , 2007 ; Liang et al., 2007 ). Interestingly, the functional integrity of two different Nup subcomplexes is required for double-strand break repair by homologous recombination (Palancade et al., 2007 ) and at least the Nup84 subcomplex is also required for anchoring telomeres and efficient DNA double-strand break repair (Therizols et al., 2006 ). Studies also report that nup170 mutants have defects in chromosome segregation (Kerscher et al., 2001 ; Iouk et al., 2002 ). Such striking NE and NPC perturbations, and severely perturbed nuclear morphology, in the sth1-F793S and TetO7-RSC cells could have indirect effects on DNA damage responses and gene expression. Additional work will be required to reveal whether some of the RSC-associated phenotypes are due to altered NE/NPCs.
We propose that there are at least two possible mechanistic explanations for the NE/NPC defects in the RSC complex mutants. First, the lack of RSC activity could result in decreased expression of a factor(s) directly required for proper NE/NPC structure and/or biogenesis, or in decreased expression of a factor(s) that maintains membrane fluidity. Others have reported that defects in the RSC complex result in pleiotropic effects attributed to either misregulated transcription or lack of chromatin access for other proteins (reviewed in Sahaa et al., 2006 ). RSC controls the transcriptional activation and repression of a broad subset of genes, with different RSC mutants having different transcriptional defects (Angus-Hill et al., 2001 ; Damelin et al., 2002 ; Ng et al., 2002 ; Kasten et al., 2004 ; Soutourina et al., 2006 ; Badis et al., 2008 ; Parnell et al., 2008 ; Hartley and Madhani, 2009 ). We observed that both new protein synthesis and ongoing transcription were required for the GFP-Nup perturbation, suggesting that the defects were not caused by loss of gene expression. Furthermore, we find similar NE/NPC defects in several different RSC mutants, and the TetO7-orf screen also identified the TetO7-SPT16 and TetO7-TAF6 strains as having weak Nup localization defects. An independent study has examined strains with deleted nonessential genes and identified nuclear morphology defects in arp5Δ and bre1Δ mutants (affecting components of histone remodeling and modifying complexes) and the seh1Δ mutant (affecting the NPC) (Teixeira et al., 2002 ). A common silencing defect was identified among the deletion strains with altered nuclear morphology, pointing toward an interdependence between maintenance of silenced chromatin and NE structure. This indicates that the NE/NPC perturbation could be a function of the global chromatin state as opposed to a specific transcriptional defect. Our biochemical and genetic analysis of potential transcriptional targets with NPC/NE connections also suggested that the sth1-F793S mutant is not linked to severe indirect defects in secretion or the RanGTPase cycle. Furthermore, to date our tests of known multicopy suppressors of sth1 mutants have not found any that rescue the altered nuclear morphology or temperature sensitivity of the sth1-F793S mutant. Therefore, although we cannot rule out specific changes in gene expression, we speculate that the NE/NPC defects are not simply indirect perturbations due to altered transcription levels.
As an alternative model, the RSC complex activity might be required for generating the correct chromatin state for contacts with the NE and/or association with a NE/NPC assembly factor. It has recently been shown that post-mitotic NPC assembly requires the chromatin-interacting factor MEL-28/ELYS for recruitment of the metazoan Nup107-160 complex (Rasala et al., 2006 , 2008 ; Franz et al., 2007 ). In yeast, the RSC complex has been connected to the yeast Nup84 complex by its shared link to nonhomologous end-joining (NHEJ) with Nup133 and Nup120 (as well as Nup60) (Palancade et al., 2007 ). In addition, the reported synthetic lethality of a nup84Δ rsc7Δ double mutant (Wilson et al., 2006 ) further suggests that proper function of the Nup84 complex is dependent on the integrity of RSC. In this light, the connection of the RSC chromatin-remodeling complex to proper NE structure is especially intriguing. We speculate that the loss of RSC function could decouple the chromatin/NE interface, leading to a chromatin or NE stress response. Structural and/or chromatin-associated roles of Nups and Poms might be inhibited, whereas lipid biosynthetic pathways might signal to the NE to expand to reestablish chromatin connections. Indeed, several reports have shown that the nucleosome occupancy of RSC changes in response to stress (Damelin et al., 2002 ; Ng et al., 2002 ; Mas et al., 2009 ). This hypothesis is supported by our observation that increasing membrane fluidity prevented the NE and NPC perturbations in the sth1-F793S cells, even though the sth1-F793S protein was still absent.
Recent studies have documented connections between NPCs/Nups and transcriptional regulation (Ishii et al., 2002 ; Casolari et al., 2004 ; Rodriguez-Navarro et al., 2004 ; Dilworth et al., 2005 ; Schmid et al., 2006 ; Brown and Silver, 2007 ). For example, genome-wide analysis of protein:DNA binding interactions has shown that Nups preferentially bind to transcriptionally active genes and induction of GAL genes results in their translocation to the nuclear rim (Casolari et al., 2004 ). Two NPC nuclear basket Nups (Nup2 and Nup60) have been linked to this transcriptional regulation by their association with chromatin-bound Prp20, the RanGEF (Dilworth et al., 2005 ). Interestingly, the membrane perturbations in the sth1-F793S and TetO7-RSC mutants are similar to that reported previously for nup1 mutant cells (Bogerd et al., 1994 ), which are defective for a NPC nuclear basket Nup (Rout et al., 2000 ). There are also reported genetic interactions among components of the Nup84 complex and the Rap1 transcriptional activation complex, and most components of the Nup84 complex have the capacity to activate transcription (Menon et al., 2005 ). These data suggest that RSC might activate transcription of genes at the NPC through interactions with the Nup84 complex. Together, we conclude that a general mechanism may exist whereby the RSC complex generates a correct chromatin state for NE/NPC association, whether for transcriptional activation and/or for NE/NPC structure and biogenesis.
We are indebted to G. Winfrey and G. Olson (Vanderbilt University, Nashville, TN) and to E. Woodruff (Magnify, Inc., Nashville, TN) for assistance and expertise with the electron microscopy experiments. We thank B. Laurent (Mount Sinai, New York, NY) for yeast strains, B. Cairns (University of Utah Health Sciences, Salt Lake City, Utah) for the Sth1 antibody, G. Blobel and M. Rout (The Rockefeller University, New York, NY) for the Nup159 antibody, and P. A. Weil, W. P. Tansey, and members of the Wente laboratory for critical discussion. This work was supported by National Institutes of Health grant R01 GM-57438 (S.R.W.), F32 GM-072272 (to D.J.R.), and T32 CA-09582-21 (to L.C.T.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-07-0615) on January 28, 2010.