|Home | About | Journals | Submit | Contact Us | Français|
Nuclear membrane fusion is the last step in the mating pathway of the yeast Saccharomyces cerevisiae. We adapted a bioinformatics approach to identify putative pheromone-induced membrane proteins potentially required for nuclear membrane fusion. One protein, Prm3p, was found to be required for nuclear membrane fusion; disruption of PRM3 caused a strong bilateral defect, in which nuclear congression was completed but fusion did not occur. Prm3p was localized to the nuclear envelope in pheromone-responding cells, with significant colocalization with the spindle pole body in zygotes. A previous report, using a truncated protein, claimed that Prm3p is localized to the inner nuclear envelope. Based on biochemistry, immunoelectron microscopy and live cell microscopy, we find that functional Prm3p is a peripheral membrane protein exposed on the cytoplasmic face of the outer nuclear envelope. In support of this, mutations in a putative nuclear localization sequence had no effect on full-length protein function or localization. In contrast, point mutations and deletions in the highly conserved hydrophobic carboxy-terminal domain disrupted both protein function and localization. Genetic analysis, colocalization, and biochemical experiments indicate that Prm3p interacts directly with Kar5p, suggesting that nuclear membrane fusion is mediated by a protein complex.
Nuclear fusion (or karyogamy) is the final step in the fusion of two haploid yeast cells to become a diploid zygote. Mating is initiated by secreted pheromones that trigger a signaling cascade in cells of the opposite type, causing arrest in G1 and the development of mating proficiency (Sprague and Thorner, 1994 ). The two cells grow in a polarized manner toward each other until they come into contact. The intervening cell walls break down and the plasma membranes fuse, forming a zygote with two nuclei. In Saccharomyces cerevisiae, as in other fungi, the nuclear envelope never breaks down. Thus, during mating the nuclear envelopes must fuse to form a diploid nucleus. Karyogamy proceeds in two steps: nuclear congression and nuclear envelope fusion (for review, see Marsh and Rose, 1997 ).
Nuclear envelope fusion is an example of homotypic fusion, involving two membranes of the same composition. The nuclear envelope is itself composed of two membranes (outer and inner), which must fuse in a coordinated manner. Although the detailed mechanism is not well understood, recently it has been demonstrated that nuclear membrane fusion proceeds in three distinct steps (Melloy et al., 2007 ). After the nuclei are pulled together by microtubules emanating from the spindle pole bodies (SPBs), first the outer membranes fuse, followed by inner membrane fusion, and finally by SPB fusion. It is likely that a fusogenic complex made up of proteins localized near the initial site of fusion is responsible for coordinating some or all of these events.
Three proteins have been found to be strongly required for nuclear membrane fusion: Kar5p, Kar2p, and Kar8p. Kar2p is the yeast BiP/GRP78, a protein of the 70-kDa heat-shock protein family localized in the endoplasmic reticulum (ER) lumen (Normington et al., 1989 ; Rose et al., 1989 ). Kar2p performs several functions in mitotic cells, including protein translocation into the ER, folding of secretory proteins, and transport of misfolded proteins out of the ER for degradation (Vogel et al., 1990 ; Sanders et al., 1992 ; Brodsky and Schekman, 1993 ; Brodsky et al., 1995 ; Simons et al., 1995 ). The role of Kar2p in nuclear fusion is separate from its role in translocation (Vogel et al., 1990 ). Kar8p/Jem1p is a lumenal DnaJ-homologue that is required for nuclear fusion (Nishikawa and Endo, 1997 , 1998 ; Brizzio et al., 1999 ) and likely interacts directly with Kar2p. Overexpression of Kar8p suppresses the mating defect of kar2-1 (Brizzio et al., 1999 ). Mutations affecting several other proteins that interact with Kar2p (Sec63p, Sec71p/Kar7p, and Sec72p) also cause mild or temperature-sensitive nuclear fusion defects (Ng and Walter, 1996 ). Last, Kar5p is a novel pheromone-induced integral membrane protein localized to the nuclear envelope near the SPB (Beh et al., 1997 ). Protease protection (Beh et al., 1997 ) and His4C hybrid experiments (Erdeniz and Rose, unpublished observations) demonstrated that the bulk of Kar5p resides within the lumen of the nuclear envelope.
Given that Kar2p, Kar8p, and most of Kar5p reside within the nuclear envelope (NE)/ER lumen, it is unlikely that they would act directly in the initial steps of outer nuclear envelope fusion. Similar to other membrane fusion events, we expect that the initial events would be mediated by proteins such as soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs), anchored on the cytoplasmic face of the nuclear envelope by a transmembrane domain with the bulk of the protein facing outward (for review, see Burri and Lithgow, 2004 ). Accordingly, we extended a strategy used by Heiman and Walter (2000) to identify genes encoding putative membrane proteins that are specifically induced during the pheromone response. Examination of disruptions in these genes identified one, PRM3, which is strongly required for nuclear fusion during mating. Previous work (Beilharz et al., 2003 ) suggested that Prm3p is an integral membrane protein on the inner nuclear membrane envelope. Instead, we show that Prm3p is a peripheral membrane protein residing on the cytoplasmic face of the nuclear envelope.
Yeast media preparation and general yeast methodology was done as described previously (Rose et al., 1990 ; Adams et al., 1997 ). Transformation of yeast was done using the TRAFO method (Gietz and Woods, 2002 ). Limited plate and filter matings were performed as described previously (Gammie and Rose, 2002 ). Equal numbers of MATa and MATα cells were mixed together on a 0.45-μm nitrocellulose filter and allowed to mate for 2.5 h on an YPD plate. Mating mixtures were fixed in 3:1 methanol:acetic acid and stained with 1 μg/ml 4′,6′-diamidino-2-phenylindole (DAPI). The number of nuclei in each zygote was counted to determine the presence of a karyogamy defect. Shmoos were obtained by inducing cells in early exponential phase with 10 μg/ml α-factor for 1.5–2 h.
All strains used are listed in Table 1. All strains are isogenic to S288C. Indicated MATa and MATα KanMX marked deletion strains were obtained from Open Biosystems (Huntsville, AL). The PRM3 gene disruption was constructed in a diploid strain (MS1556 × MS2290) by a polymerase chain reaction (PCR)-based gene disruption method using HIS3 as the selective marker (Baudin et al., 1993 ). The gene disruption in the His+ transformants was confirmed by PCR. Diploids containing the PRM3 disruption were sporulated and dissected. MS7590, MS7591, MS7592, MS7593, MS7594, and MS7595 are all haploid strains derived from the diploid transformants. Plasmids pMR5062, pMR5063, pMR5064, and pMR5065, carrying various tagged forms of PRM3, were constructed using primers designed to generate a product with regions of homology for in vivo recombination (Ma et al., 1987 ). In general, purified PCR products and linearized vector plasmids were cotransformed into yeast. Plasmids from Leu+ transformants were recovered from yeast, transformed into Escherichia coli, and checked by restriction digest and/or PCR.
The dut ung site-directed mutagenesis method was used to produce the N-terminal point mutations (Kunkel, 1985 ). Oligonucleotides used for mutagenesis were designed with 10 base pairs of homology on each side of the mutated residue, with each mutated codon changed to alanine. N-terminal, C-terminal, and internal deletions of PRM3 were made using in vivo recombination. PCR primers were used to generate DNA fragments with the region of interest deleted and regions of homology to the vector pRS415. Gel-purified PCR products and digested vector were transformed into the yeast strain MS7590 for recombination.
PCR mutagenesis was performed using 10× error-prone PCR buffer (100 mM Tris-HCl, pH 8.3, 500 mM KCl, 70 mM MgCl2, and 0.1%, wt/vol gelatin), 0.25 mM each dATP and dGTP, 1 mM each dCTP and dTTP, and 0.3 mM final concentration of MnCl2, 50 pmol of each primer (T3 and T7), 10 ng of template DNA (MR5063), and 5 U of Taq polymerase (Hoffman-La Roche, Nutley, NJ). One 100-μl reaction was prepared, separated into 10-μl aliquots to run independently, and mixed back together after PCR reactions were complete. The PCR cycle conditions were 94°C for 30 s, 55°C for 1 min, and 72°C for 1 min, repeated for 30 cycles. PCR products were transformed with digested vector (pRS415) into a prm3Δ strain (MS7590) for in vivo recombination and plated out. Mutant colonies were mated to a prm3Δ (MS7591) lawn for 2.5 h at 23, 30, and 37°C. Colonies unable to mate were selected for further study. Plasmids were isolated from these strains and transformed back into yeast to check whether mutant phenotypes were still present. Plasmids were then sequenced for the specific mutation.
Mutants were checked for proteins levels using Western blotting. Proteins were precipitated using 50% trichloroacetic acid (TCA), separated by SDS-polyacrylamide gel electrophoresis (PAGE), and transferred to nitrocellulose. Blots were blocked in 5% milk in 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.05% Tween 20. Green fluorescent protein (GFP) was detected with a mixed mouse monoclonal α-GFP at 1:1000 dilution (Hoffman-La Roche), and the FLAG epitope was detected with a mouse monoclonal antibody at 1:2500 dilution (Sigma-Aldrich, St. Louis, MO). Sec22p and Sec34p were detected using affinity-purified rabbit antibodies used at a 1:2000 dilution (gift from Daniel Ungar and Angela Chan, Princeton University, Princeton, NJ). Secondary antibody (anti-mouse-horseradish peroxidase [HRP]; GE Healthcare, Chalfont St. Giles, Buckinghamshire, United Kingdom) was diluted 1:2500 in blocking buffer and visualized with the enhanced chemiluminescence chemiluminescent system (GE Healthcare).
Indirect immunofluorescent staining of fixed cells was performed with modifications as described previously (Rose et al., 1990 ; Adams et al., 1997 ). Cells were fixed with formaldehyde for 30 min at 30°C; spheroplasts were obtained using 10 mg/ml Zymolyase 100T (MP Biomedicals, Aurora, OH) and 5 μl of β-mercaptoethanol for 30 min at 30°C. Mouse monoclonal anti-GFP (Hoffman-La Roche) was used at a 1:50 dilution and mouse monoclonal anti-FLAG (Sigma-Aldrich) was used at a 1:100 dilution in blocking buffer (10 mg/ml bovine serum albumin and 0.1% NaN3 in phosphate-buffered saline). The secondary antibody goat anti-mouse fluorescein isothiocyanate (FITC) was used at a 1:500 dilution. Cells were stained with DAPI for localization of the nuclei. For colocalization experiments, two primary antibodies, anti-GFP and anti-Tub4p (a gift from Mark Winey, University of Colorado, Boulder, CO), were incubated with sample concurrently. Anti-GFP was used at 1:50 dilutions, and rat anti-Tub4p was used at a dilution of 1:1000. Secondary antibodies, FITC-conjugated anti-mouse and rhodamine-conjugated anti-rabbit, were used at dilutions of 1:500.
Membrane isolation, protease protection, and protein extractions were performed as described by Beh et al. (1997) . Exponential phase cells were grown to OD600 = 0.5 at which time α-factor was added to 10 μg/ml for 2.5 h. Cells were pelleted and washed twice in 10 ml of spheroplast buffer (0.1 M potassium phosphate, pH 6.5, 1.2 M sorbitol, and 1 mM phenylmethylsulfonyl fluoride [PMSF]). To remove the cell walls, cells were resuspended in 10 ml of spheroplast buffer plus 0.5 ml of 10 mg/ml Zymolyase 100,000T and 50 μl of β-mercaptoethanol. After 1 h at 30°C, spheroplasts were centrifuged at 2K rpm in a centrifuge (Beckman Coulter, Fullerton, CA). After two washes with spheroplast buffer, spheroplasts were resuspended in 25 ml of lysis buffer (20% Ficoll 400, 20 mM potassium phosphate, pH 6.5, 1 mM MgCl2, and 1 mM PMSF) and lysed with 25 strokes of a stainless steel Dounce homogenizer. The lysate was chilled on ice for at least 10 min and centrifuged at 2500 × g for 5 min in 30-ml Corex tubes in a JA-17 rotor (Beckman Coulter) at 4°C to pellet unlysed cells. The supernatant was then centrifuged twice at 13,000 × g first for 5 min and then for 10 min. The supernatant was then diluted fivefold in cold 20 mM potassium phosphate, pH 6.5, 1 mM MgCl2, and 1 mM PMSF, and the membranes were pelleted in 500-ml bottles (Beckman Coulter) or multiple plastic tubes by centrifugation for 10 min at 11,700 × g in a JA-14 or JA-17 rotor (Beckman Coulter) at 4°C.
For protease protection experiments, membrane pellets were resuspended in 2 ml of ice-cold protease buffer (0.3 M mannitol, 0.1 M KCl, 50 mM Tris-HCl, pH 7.5, and 1 mM EGTA) and divided into two equal fractions. To one fraction, Triton X (TX)-100 was added to a final concentration of 1%. Before, and at various times after, the addition of proteinase K (final concentration, 0.3 mg/ml), 150-μl aliquots were taken and mixed with 15 μl of 100% TCA. Samples were chilled for an additional 20 min after the last time point. The samples were centrifuged in a microfuge and washed with 90% acetone. Proteins were resolved by 10% SDS-PAGE.
For protein extractions, the membrane pellet was resuspended in 3 ml of buffer. Aliquots were added to 500 μl of 2× extraction buffer to bring the final concentrations to 1 M NaCl, 1% Triton X-100, 1% Triton X-100/Plus/0.5 M NaCl, or 0.1 M Na2CO3, pH 11.0. Samples were extracted on ice for 1 h. Extracted proteins were separated from the membranes by centrifugation at 70,000 × g (model TL-100 ultracentrifuge and TLA100.2 rotor; Beckman Coulter) for 1 h. Samples were layered on top of 200 μl of 10% sucrose cushions made in each extraction buffer. Supernatants were removed and precipitated with TCA. Proteins were boiled in sample buffer, and equivalent volumes of pellet and supernatant were separated by 10% SDS-PAGE.
Lysates for coimmunoprecipitations were prepared as described previously (Jaspersen et al., 2006 ). Mid-log phase cells (100 ml) were treated with α-factor for 1.5 h. Cells were harvested, washed, and resuspended in 1 ml of spheroplast buffer (50 mM Tris-HCl, pH 7.5, 1.2 M sorbitol, and 10 mM NaN3). Spheroplasts were prepared by treatment with 100 μl of Zymolyase 100T (2.5 mg/ml stock) and 5 μl of β-mercaptoethanol for 30 min at 30°C. Spheroplasts were washed three times in spheroplast buffer, resuspended in lysis buffer (20 mM HEPES-KOH, pH 7.4, 100 mM K-acetate, 5 mM Mg-acetate, 1 mM EDTA, and 1 mM PMSF), and lysed by Dounce homogenization. For cross-linking experiments, extracts were treated with dithiobis(succinimidyl propionate) (0.2 mg/ml) for 10 min on ice and quenched by addition of 50 mM Tris-HCl, pH 8.0. Lysates were centrifuged (11,000 × g; 10 min; 4°C), and pellets were resuspended in immunoprecipitation (IP) lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, and 1% NP-40) to solubilize proteins. The clarified supernatant was taken for immunoprecipitation.
Immunoprecipitations were performed by incubating lysates with either 60 μl of anti-FLAG M2 agarose (Sigma-Aldrich) or 30 μl of Dynabeads protein G (Invitrogen, Carlsbad, CA), with 1:400 dilution anti-hemagglutinin (HA) for 3 h at 4°C. Beads were washed three times in IP lysis buffer and boiled in 100 μl of 2× sample buffer (90 mM Tris-HCl, pH 6.8, 20% glycerol, 2% SDS, 0.02% bromophenol blue, and 0.1 M dithiothreitol). We analyzed 1/80 each of the input lysate and flow-through and one fourth of the bound protein by SDS-PAGE and Western blotting. Primary antibodies were diluted in blocking buffer as follows: 1:1000 mouse monoclonal anti-GFP (Hoffman-La Roche), 1:5000 mouse monoclonal anti-FLAG (Sigma-Aldrich), and 1:2500 mouse anti-HA. Secondary antibody (anti-mouse HRP; GE Healthcare) was diluted 1:2500 in blocking buffer.
Shmoos expressing GFP-tagged Prm3p were collected on 0.45-μm nitrocellulose filters for fixation. Cells were fixed by cryoimmobilization in liquid nitrogen under high pressure using an EM Pact High Pressure Freezing Unit (Leica Microsystems, Deerfield, IL). The samples were freeze substituted in 0.25% glutaraldehyde and 0.1% uranyl acetate in acetone for 84 h at −85°C, and warmed to −50°C over an 8-h period in an AFS Freeze Substitution Machine (Leica Microsystems). Samples were embedded in Lowicryl HM20 resin at −50°C using UV polymerization. Thin sections of 70 nm were cut on a UC6 ultramicrotome (Leica Microsystems). Sections were probed with mouse monoclonal anti-GFP (Hoffman-La Roche) at a 1:2 dilution for 2 h at room temperature. Secondary antibody was at 1:10 dilution (Ted Pella, Redding, CA) for 2 h at room temperature. Images were observed at 80 kV on a 912AB transmission electron microscope (Carl Zeiss, Thornwood, NY) equipped with an Omega energy filter. Micrographs were captured using a digital camera (Advanced Microscopy Techniques, Danvers, MA).
Many of the genes currently known to be involved in mating are induced by pheromone; moreover, the mechanisms of plasma and nuclear membrane fusion are likely to involve membrane-bound proteins. To identify pheromone-inducible membrane proteins, we extended an approach used by Heiman and Walter (2000) to screen published databases to identify putative pheromone-induced genes with predicted transmembrane domains. To increase the range of potential genes, we identified the set of genes showing at least threefold induction upon exposure to pheromone at various times after treatment (Roberts et al., 2000 ). The presence of transmembrane domains was predicted using the SOSUI program (Hirokawa et al., 1998 ) and TMHMM program (Krogh et al., 2001 ). In total, 55 pheromone-induced genes were predicted to encode transmembrane proteins. Half of the candidates had been characterized previously and were demonstrated to have roles in cell wall synthesis and/or mating, including PRM1, AGA1, STE2, and FUS1. The effects of deletions in the 27 remaining genes not described previously as having a role in mating were assayed for potential effects on mating by using a quantitative microscopic mating assay (Gammie et al., 1998 ). This set included six genes identified previously as encoding potential pheromone-induced membrane proteins (Heiman and Walter, 2000 ). Eight mutants showed a significant cell or nuclear fusion defect (Supplemental Table 1). However, only one deletion mutant exhibited a strong defect (YPL192C/PRM3) and was characterized further.
To understand the nature of the mutant phenotype, prm3 zygotes were examined microscopically for mating defects. Karyogamy mutants are classified as either unilateral or bilateral; unilateral mutants are defective when mated to wild-type cells, bilateral mutants show a defect only when both mating partners are mutant. The prm3 mutant zygotes showed a strong bilateral karyogamy defect (Figure 1A and Table 2). No defect was apparent in unilateral matings, indicating that Prm3p need be expressed in only one cell for nuclear fusion to occur (Table 2). The nuclear fusion defect in prm3 mutant zygotes was complemented by a plasmid carrying a copy of PRM3, as well as N-terminal FLAG and GFP-tagged PRM3 (Table 2). Epitope tags at the C terminus of Prm3p resulted in nonfunctional constructs, indicating that the C terminus is important for protein function.
The close apposition of the nuclei in prm3 zygotes is consistent with a class II nuclear fusion defect (Kurihara et al., 1994 ), in which the nuclei have congressed but nuclear envelope fusion is blocked. To detect a defect in membrane fusion, we examined prm3 zygotes in which the nuclear envelope lumen was labeled with 3XGFP-HDEL and the spindle pole bodies were labeled with Spc42p-red fluorescent protein (RFP) (Melloy et al., 2007 ). Consistent with a defect in nuclear envelope fusion, the two spindle bodies moved close together, but distinct nuclear envelopes remained surrounding the two unfused haploid nuclei (Figure 1B). In wild-type zygotes, the nuclear envelopes fuse quickly to form a diploid nucleus (Figure 1C). We conclude that Prm3p is required for nuclear envelope fusion.
To understand the role Prm3p plays in nuclear membrane fusion, we investigated its structure, expression, and localization within the cell. Prm3p is a small protein of 133 residues, with a predicted transmembrane domain at the carboxy terminus between residues 107-127. Comparison with homologous genes from related yeast species (Cliften et al., 2003 ; Kellis et al., 2003 ) showed that the carboxy-terminal region of Prm3p is highly conserved (Figure 2). This region includes the sole predicted transmembrane domain. The presence of conserved charged residues spaced three or four residues apart in the amino-terminal region suggested a coiled-coil structure that may interact with other proteins. In the middle of the protein are two small highly conserved regions consisting of positively charged residues (Figure 2).
The upstream region of PRM3 contains a putative pheromone response element, consistent with its regulation by pheromone. To confirm the results from the genome-wide study of pheromone-induced genes, expression of FLAG-Prm3p in both mitotic and pheromone-treated cells was determined using Western blot analysis. Expression of FLAG-Prm3p was only detected in pheromone-treated cells (Figure 3A), consistent with a role in karyogamy.
Although membrane fusion is typically catalyzed by cytoplasmic proteins, all of the proteins identified previously to be strongly required for nuclear membrane fusion (Kar5p, Kar2p, and Kar8p) are located within the nuclear envelope/ER lumen. Initially predicted to have a single transmembrane domain, Prm3p could be oriented with its amino-terminal domain either exposed to the cytoplasm or contained within the lumen of the membrane. To determine the orientation of Prm3p, membrane fractions isolated from cells expressing FLAG-Prm3p were exposed to exogenous protease with and without detergent. Exposed portions of the protein are sensitive to proteases, whereas proteins contained within the lumen are resistant. The majority of FLAG-Prm3p was sensitive to protease in the absence of detergent (Figure 3B). Kar2p, a lumenal protein, remained resistant to protease, demonstrating that the membranes remained intact throughout the course of the experiment (Figure 3B). Comparison of FLAG-Prm3p and Kar2p levels showed that FLAG-Prm3p was degraded with a half-life of ~3 min. Approximately 15% of the protein was not digested within the course of the experiment, due either to protection within a protein complex or the formation of a small amount of inside-out vesicles that protected a fraction of the protein. After addition of detergent to disrupt the membranes, lumenal proteins became fully accessible to proteases, and both proteins were rapidly degraded. We conclude that the Prm3p is oriented such that it is exposed on the outside of the nuclear envelope and not contained within the lumen.
Protein extraction was performed to determine whether Prm3p is an integral membrane protein. Crude membrane fractions were treated with various reagents to extract Prm3p from the membrane. As controls, we also examined the extraction of Sec34p, a peripheral membrane protein and Sec22p, an integral membrane protein (Figure 3C). Typically, peripheral proteins can be extracted with salt or high pH, which disrupt protein–protein interactions, whereas integral membrane proteins require detergents for extraction. As observed previously, the peripheral protein Sec34p was partially extracted under all conditions, but most strongly by high pH (VanRheenen et al., 1999 ). The integral membrane protein Sec22p was not extracted by either 1 M NaCl or high pH. Sec22p was partially extracted by detergent alone and more completely by detergent plus 1 M NaCl. Unlike Sec22p, Prm3p was substantially extracted by 1 M NaCl and not at all by detergent alone. The partial extraction of Prm3p by 1 M NaCl was similar to Sec34p, and all three proteins were more completely extracted by detergent plus 1 M NaCl (Figure 3C). Unlike Sec34p, Prm3p was not extracted by high pH. However, Prm3p has an isoelectric point of 11.1, and many proteins are not soluble near their isoelectric point. Furthermore, Sec72p, another peripheral ER membrane protein is also not extracted at high pH (Feldheim and Schekman, 1994 ; Kaiser et al., 2002 ). Together with the protease protection data, the protein extraction data indicate that Prm3p is actually a peripheral membrane protein. Unlike the other proteins required for nuclear membrane fusion, Prm3p seems to reside entirely outside of the ER lumen.
Prm3p may reside on the cytoplasmic or nuclear face of the nuclear envelope. Fusion of GFP or the FLAG epitope to the N terminus resulted in a fully functional protein (Table 2). Both the GFP fusion (Figures 4–6 and and8)8) and the FLAG-tagged protein (Figures 6 and and7)7) localized to the nuclear envelope when expressed in either mating or mitotic cells. Similar localization results for the full-length protein in mitotic cells were reported previously (Beilharz et al., 2003 ). Previously, Prm3p was reported to be a “tail-anchored” integral membrane protein resident on the inner nuclear membrane (Beilharz et al., 2003 ). In that study, GFP was fused to a truncated form of Prm3p lacking the hydrophobic C-terminal domain and was found to localize in the nucleus. Mutation of a putative nuclear localization sequence (NLS) in the truncated protein resulted in mislocalization throughout the cytoplasm. In addition, mutations that disrupt nuclear import (rna1-1 and prp20-1) by failing to regulate the Ran-GTPase caused increased localization of full-length GFP-Prm3p to more peripheral ER (Beilharz et al., 2003 ; King et al., 2006 ). Based on these results, the authors concluded that native full-length Prm3p is an inner nuclear membrane protein. However, localization to the inner nuclear envelope would be surprising, given its role in nuclear envelope fusion. As we show in the next section, the carboxy-terminal region is critical for Prm3p function and its deletion leads to a nonfunctional protein. When the same NLS mutation was introduced into full-length GFP-Prm3p, or the region containing the putative NLS was deleted, the proteins localized to the nuclear envelope (Figure 6F) and supported wild-type levels of nuclear fusion during mating (Figure 6G). Thus, it is unlikely that the putative NLS plays a critical role in Prm3p function. In addition, mutations that affect nuclear import are likely to have a major effect on the localization of many proteins and the mislocalization of Prm3p may result from indirect effects on nuclear structure.
Given its role in nuclear fusion, it is likely that Prm3p is located on the cytoplasmic face of the nuclear envelope. We therefore used two approaches to determine on which face of the nuclear envelope Prm3p resides. Because light microscopy does not provide sufficient resolution to distinguish between the outer and inner nuclear membranes, we used immunoelectron microscopy to localize GFP-Prm3p in shmoos. GFP-Prm3p was detected with mouse anti-GFP and F(ab′)2 anti-mouse conjugated to 10-nm gold particles. Gold particles were enriched in the vicinity of the nuclear envelope, as well as sporadically elsewhere in the cell (Figure 4A). The positions of the gold particles relative to the center of the NE lumen were measured (Figure 4B); positive numbers indicate distances toward the cytoplasm, negative numbers indicate distances toward the nucleus. The maximum distance between a gold particle and the center of the ER lumen should be roughly 50 nm, including the length of the primary and secondary antibodies, Prm3p itself, and distance from the membranes to the center of the lumen. Gold particles >80 nm away from the center of the lumen were not measured and were taken to be background staining. In total, 79 gold particles within 80 nm on either side of the nuclear envelope were observed in 36 different cells. In addition, 104 gold particles whose distance from the NE lumen was >80 nm were observed scattered over the entire cross sections of the cells. We estimate that the region of the nuclear envelope showed ~20-fold enrichment in gold particles. The histogram (Figure 4B) shows that the gold particles were highly enriched in a region centered on the outer nuclear membrane. The number of gold particles located within 50 nm of the center of the NE nuclear envelope was significantly higher on the outer membrane than on the inner nuclear membrane (p = 0.0022; chi-square test). Assuming that the particles are symmetrically distributed about the outer membrane, then the excess of gold particles on the inner membrane (11 particles) would constitute no more than ~15% of the total.
For an independent test of Prm3p localization, we used live cell microscopy to measure the relative kinetics of transfer of Prm3p between nuclei during nuclear fusion. Because outer nuclear membrane fusion occurs before inner nuclear envelope fusion (Melloy et al., 2007 ), if Prm3p is located on the cytoplasmic face of the nucleus, then it should transfer to the recipient nucleus before a nucleoplasmic protein is able to transfer. Conversely, if Prm3p is only located on the inner nuclear membrane, then transfer of Prm3p could only occur after inner nuclear membrane fusion, at the same time as the nucleoplasmic protein. For this experiment, time-lapse microscopy was performed on wild-type zygotes in which one parent expressed GFP-Prm3p and the other parent expressed Pap1p-mCherryFP as the nucleoplasmic marker. Images of zygotes were taken at 1-min (n = 4) or 30-s intervals (n = 15) for 15 min, starting immediately after completion of cell fusion. In the example shown, GFP-Prm3p began to transfer to the recipient nucleus after 4 min, whereas the nucleoplasmic marker did not transfer until after 5 min (Figure 4C). Fluorescence intensity in the recipient nuclei was measured and graphed as a percentage of total fluorescence in the zygotes (Figure 4, D and E). The median delay between GFP-Prm3p and Pap1p-cherryFP transfer at half-maximal transfer was ~45 s (n = 19). The median delay for the initial time of entry was ~23 s. In four of 19 zygotes, no delay was observed; transfer of Pap1p-mCherryFP before GFP-Prm3p was not observed. Thus, both the immuno-EM and live cell microscopy demonstrate that Prm3p is primarily located on the cytoplasmic face of the nuclear envelope.
In contrast to the short-term experiments, longer term time-lapse microscopy provided a likely explanation for why prm3 mutants are bilateral, that is, why Prm3p is not required to be expressed in both mating cells for nuclear fusion to continue. GFP-Prm3p was expressed from the GAL1 promoter, and cells were mated to a kar1-1 strain to prevent nuclear congression. Matings were performed on glucose media to shut off synthesis of GFP-Prm3p. As expected, GFP-Prm3p was restricted to one nucleus immediately after cell fusion. After 30 min, GFP-Prm3p was detected on the other nucleus, eventually reaching comparable levels (Figure 5A). By 60 min, the zygote has reentered the mitotic cycle, and two nuclei have moved together near the incipient bud. Thus, one explanation for the prm3 bilateral phenotype is that Prm3p is only needed in one partner because the protein can shuttle between the two nuclei. Interestingly, when GFP-Prm3p was expressed from the GAL1 promoter in mitotic cells before mating, it caused striking alterations in the shape of the nuclear envelope (Figure 5B), possibly indicating a role for Prm3p in the structure of the nuclear membranes during karyogamy.
To identify the regions of Prm3p required for nuclear envelope-specific localization and function, we analyzed the effects of point and deletion mutations in the gene. To start, we conducted an in vitro mutagenesis screen of PRM3 to identify point mutations that cause defects in mating. GFP-PRM3 and FLAG-PRM3 on centromere plasmids were mutagenized using error-prone PCR, transformed into a prm3Δ strain, and the transformants were mated to a prm3Δ strain at 23, 30, and 37°C to identify mutants unable to form diploids. Eight point mutations (indicated in Figures 2 and 6A and Table 3) were identified, all located within the conserved hydrophobic C-terminal region. Two point mutations caused temperature-sensitive defects, losing nuclear fusion function at 37°C; growth and cell fusion were not affected. Thirteen nonsense mutations were identified by sequencing and not studied further.
All eight point mutations affected the function, stability, and localization of Prm3p to different extents. Nuclear fusion efficiency decreased to 10–25% of wild type (Figure 6G). Protein levels in prm3-1, prm3-2, prm3-3, prm3-5, and prm3-8 strains were greatly decreased (Figure 6C). Prm3p was mislocalized in prm3-2, prm3-3, prm3-5, prm3-6, and prm3-8 cells (Figure 6B). Mislocalization of the mutant protein and decreased protein levels were correlated, except for mutant prm3-6, in which the protein ran at a lower molecular weight. Only in mutant prm3-4 was the protein stable and localized normally.
Two point mutants, prm3-1 and prm3-7, were temperature sensitive. In both strains, protein function was only slightly reduced at 23°C; nuclear fusion efficiency decreased with increased temperature (Figure 6G). Protein levels in both mutants decreased and Prm3p became increasingly mislocalized with increased temperature (Figure 6, D and E). However, Prm3p became mislocalized in the prm3-7 mutant at a lower temperature; some mislocalization was seen at 23°C. The presence of stable but mislocalized protein at 23°C suggests that mislocalization of the Prm3p protein by itself does not lead to destabilization. The isolation of point mutations causing defects in nuclear fusion only in the conserved C-terminal region underscores the essential role of this part of the protein for stability, localization, and function.
We used a directed mutagenesis approach to probe the function of the other conserved regions of Prm3p. For example, each of the conserved charged residues in the N-terminal region suggestive of a coiled-coil structure were changed to alanine (Figure 2 and Table 3, prm10-prm17); none affected Prm3 function, localization, or stability (data not shown). Because it was possible that multiple residues would need to be mutagenized to cause a defect in function or localization, a deletion of the entire N terminus (residues 1-46) was constructed. Deletion of the N-terminal region (NΔ) had no affect on protein function, stability, or localization (Figure 6, F and G). The internal section of Prm3p contains two conserved regions largely made up of positively charged residues. Deletion of region I3 (residues 86-96) had a significant effect on nuclear fusion efficiency and protein stability (Figure 6, F and G). Deletion of region I2 (residues 63-76), which includes the putative NLS sequence, had no effect on function or localization (Figure 6, F and G). As expected, deletion of the C-terminal region (residues 104-128) had severe affects on protein function and localization. Together, the mutational analysis indicates that the C-terminal half of Prm3p is the most critical part of the protein, responsible for function, localization, and stability.
Although Prm3p localized to the nuclear envelope in shmoos, immunofluorescent detection of both FLAG-Prm3p and GFP-Prm3p showed distinctly nonuniform staining. In 78% of shmoos, one or more concentrated dots of Prm3p staining were present along the nuclear envelope (Figure 7A). One of the Prm3p dots was adjacent to or colocalized with the SPB in 61% of wild-type shmoos (Figure 7A) and 96% of zygotes (n = 151; Figure 8Kar5p, another protein important in karyogamy, is localized adjacent to the SPB (Beh et al., 1997 ). When Kar5p was overexpressed, Prm3p localization at or adjacent to the SPB increased to 93% and Prm3p colocalized with Kar5p in 79% of the shmoos (Figure 7B). Localization of Prm3p to the vicinity of the SPB, the site of nuclear envelope fusion, is consistent with Prm3p's important role in karyogamy.
The colocalization of Prm3p with Kar5p in shmoos suggested that these proteins may be part of a protein complex that facilitates fusion of the nuclear envelope. To explore the interaction between Prm3p and Kar5p, we examined Prm3p localization in zygotes blocked at the initial stage of nuclear envelope fusion. GFP-Prm3p was enriched at the SPB in 96% (n = 151) of KAR5 zygotes (Figure 8A). In the kar5Δ zygotes, the concentration of GFP-Prm3p at the SPB was reduced (Figure 8B) or disappeared altogether in 29% (n = 96) (Figure 8C). Measurement of the fraction of total GFP-Prm3p at the SPB (Figure 8D) showed that the amount of GFP-Prm3p at the SPB was significantly decreased in the kar5Δ zygotes to 54% of wild-type levels (p = 7 × 10−52; Student's t test). In contrast, localization of Kar5p to the nuclear periphery was not affected by prm3Δ (data not shown). Furthermore, GFP-Prm3p localization in shmoos was not affected by mutations in KAR7, KAR8, SEC72, or PRM5. The increased localization of GFP-Prm3p at the SPB when Kar5p is overexpressed in shmoos, together with the decreased localization in kar5Δ zygotes, suggests that Kar5p is responsible for recruiting or stabilizing Prm3p to the vicinity of the SPB.
In a second approach to probing the interactions between Kar5p and Prm3p, we examined the genetic interactions between various mutations. Both prm3Δ and kar5Δ are “bilateral” mutations, requiring both parents to be mutant for nuclear fusion to be defective. When a prm3Δ cell was mated to a kar5Δ cell, a partial nuclear fusion defect resulted, resulting in 28% Kar− zygotes (Table 4). This “synthetic” bilateral defect was not observed when prm3Δ was mated to other kar mutants. A stronger defect was observed when prm3Δ was crossed to a partially dominant allele, kar5-1162, increasing to 72% (Table 4). Overexpression of Kar5p suppressed the prm3Δ nuclear fusion defect; nuclear fusion increased sixfold when KAR5 was overexpressed in one mating partner and 10-fold when KAR5 was overexpressed in both mating partners (Table 4). Overexpression of Prm3p did not suppress the mating defect in kar5Δ zygotes. Thus, both localization and genetic data suggest that Prm3p interacts with Kar5p.
To determine whether Prm3p interacts directly with Kar5p, we used coimmunoprecipitation (coIP) of proteins tagged with FLAG, HA, or GFP. Because coIPs of membrane proteins can be complicated by the harsh conditions required to extract the proteins, a membrane permeable cross-linker, dithiobis(succinimidyl propionate) (DSP) was used in some experiments. DSP is a cleavable cross-linker that reacts with the amine group on lysine residues (Walleczek et al., 1989 ). Each experiment was performed with and without DSP; although coprecipitation was detected without the cross-linker, the efficiency was reduced.
When GFP-Prm3p and HA-Kar5p were coexpressed and anti-HA antibody was used to immunoprecipitate HA-Kar5p, a small fraction of the GFP-Prm3p coprecipitated (Figure 9A), consistent with Kar5p's restricted localization near the SPB. As a control, the abundant 3XGFP-HDEL protein in the NE lumen did not coprecipitate with HA-Kar5p. The reciprocal experiment was performed with cells expressing FLAG-Prm3p and HA-Kar5p, by using anti-FLAG antibody to immunoprecipitate FLAG-Prm3p. HA-Kar5p was coprecipitated with FLAG-Prm3p (Figure 9C), confirming the physical interaction between Prm3p and Kar5p.
The bulk of Kar5p resides within the NE lumen and only ~20 residues between the second and third transmembrane domains are predicted to be exposed to the cytosol (Beh et al., 1997 ; Erdeniz and Rose, unpublished). Thus, the cytosolic loop is the only region of the protein that may be in contact with Prm3p. To determine the role of this region of Kar5p in interactions with Prm3p, coimmunoprecipitations were performed using cells expressing FLAG-Prm3p and HA-tagged kar5 deletion mutants (Erdeniz and Rose, unpublished). The kar5-hyd2Δ mutation is a deletion of the second transmembrane domain, whereas the kar5-hyd3Δ mutation is a deletion of the third transmembrane domain (Figure 9B). The topologies of the kar5 deletion mutant proteins were determined using C-terminal His4C fusions (Sengstag, 2000 ; Erdeniz and Rose, unpublished). In the kar5-hyd2Δ mutant, the third transmembrane domain is oriented such that that the cytosolic loop is within the NE lumen; in the kar5-hyd3Δ mutant, the loop remains cytosolic. The mutant proteins were expressed at levels comparable to the wild type (Figure 9C). Comparison of coIPS of the kar5 mutants with the wild-type strain showed that the amount of Kar5p protein pulled down was significantly reduced in the kar5-hyd2Δ mutant (10% wild type) but not in the kar5-hyd3Δ mutant (90% wild type; Figure 9C). Therefore, we conclude that the cytosolic loop of Kar5p is required for interaction with Prm3p, consistent with interaction on the cytoplasmic face of the nuclear envelope.
Prm3p is required for nuclear fusion during mating in S. cerevisiae. PRM3 has several properties that indicate it plays a key role in nuclear fusion during karyogamy. First, prm3Δ mutant zygotes have a severe karyogamy defect in which the nuclear membranes fail to fuse. Second, like other proteins important in mating, Prm3p is only expressed in response to pheromone. Third, although Prm3p is localized diffusely over the nuclear envelope, it becomes concentrated near the SPB, the initial site of fusion, in shmoos and zygotes. Finally, PRM3 shows genetic and physical interactions with KAR5, which encodes a protein previously shown to play a key role in nuclear fusion.
The role of Prm3p in nuclear envelope fusion leads to an expectation that the protein is located on the cytoplasmic face of the nuclear envelope. Protein extraction experiments demonstrated that Prm3p is a peripheral membrane protein and sensitivity to protease indicates that the amino-terminal FLAG-epitope, at least, is exposed on the surface of the nuclear envelope. Immuno-electron microscopy and live cell microscopy showed that Prm3p resides largely, if not exclusively, on the cytoplasmic face of the nuclear envelope. Based on these data, we conclude that Prm3p resides on the cytoplasmic side of the nuclear envelope, where it can mediate the nuclear membrane fusion.
A previous study proposed that Prm3p localized to the inner nuclear envelope, based on the nuclear localization of a truncated protein, lacking the hydrophobic carboxy terminus (Beilharz et al., 2003 ). The localization of the truncated protein was mediated by a region of basic residues, similar to a nuclear localization sequence, but which very likely has a different function in the intact protein.
One striking aspect of Prm3p localization is its exclusive localization to the nuclear envelope and not the peripheral endoplasmic reticulum. Restriction to the inner nuclear envelope (Beilharz et al., 2003 ) provided a natural explanation for Prm3p's localization pattern. However, in light of the recognition that Prm3p is a peripheral outer membrane protein, its localization must be governed by interactions with components of the membrane specific to the nuclear envelope. Association with nuclear pore proteins would be one way to restrict Prm3p to the nuclear envelope. For example, Ndc1p, an integral membrane protein, and, Cdc31p, a soluble centrin homologue, are localized to the nuclear envelope at both the SPB and the nuclear pore (Chial et al., 1998 ; Fischer et al., 2004 ). However, except near the SPBs, Prm3p localization is relatively uniform over the surface of the nuclear envelope, making interactions with the nuclear pores less likely. It is, of course, possible that Prm3p localization is dependent on interaction with some other as yet unidentified nuclear membrane specific protein.
Based on its primary sequence, it is tempting to speculate that Prm3p is a lipid binding protein and that nuclear localization is governed by interactions with nuclear-specific lipids. Although it does not share strong homology, Prm3p contains several features similar to many lipid-binding proteins (Cho and Stahelin, 2005 ). In addition to the extended hydrophobic C-terminal region, Prm3p contains a region significantly enriched in lysines and arginines. In the 28 residues encompassing the two internal conserved regions (I2 and I3), there are 12 lysines and arginines, comparable with the level of charge found in the well characterized phosphatidylinositol bisphosphate (PIP2) binding myristoylated alanine-rich C-kinase substrate effector domain (13 lysines within 25 residues) PIP2 (Wang et al., 2001 ). Moreover, the sequence KAVSPGRVRKHK found in region I2 is reminiscent of the conserved PIP2 binding motif (KATTHEVMGPKKKH found in the adaptin AP180 and related proteins; Ford et al., 2001 ).
The hydrophobic region is predicted to form an amphipathic helix, with all charged and polar residues resident on the one side (Supplemental Figure 1), consistent with a lateral association with the membrane. In general, mutations that disrupt localization changed residues from hydrophobic to polar or charged on the hydrophobic side, and polar residues to hydrophobic on the polar side. Interestingly, mutation prm3-4, which changed serine to alanine on the polar side, did not affect localization, suggesting that this part of Prm3p is specifically important for function or protein–protein interaction.
Finally, we note that peripheral membrane proteins often contain aromatic residues that insert into the lipid bilayer (Cho and Stahelin, 2005 ). When the highly conserved phenylalanine at residue 105 was mutated to either alanine or tyrosine, the mutant protein was nonfunctional and failed to localize (data not shown), suggesting that this residue has a specific function separate from simple hydrophobic interactions. We speculate that Prm3p is a novel lipid-binding protein and that interactions with an as yet unidentified nuclear envelope-specific lipid may account for its unique localization pattern.
The prm3 mutant zygotes contain nuclei that become closely apposed but do not fuse. Based on Prm3p's localization on the cytoplasmic surface of the nuclear envelope, we predict that Prm3p would act at an early step of nuclear membrane fusion. Electron tomography indicates that Prm3p is required before outer membrane fusion, possibly helping to initiate fusion (Melloy, Shen, White, and Rose, unpublished). Like other membrane fusion events, nuclear fusion is likely to be mediated by SNARE proteins on the nuclear envelope (Burri and Lithgow, 2004 ), and Prm3p might act as a novel nuclear fusion SNARE or SNARE accessory factor. Superficially, Prm3p resembles a SNARE protein with a region near the N terminus similar to coiled-coils and a hydrophobic domain at the C terminus. However, the N-terminal region is not required for nuclear fusion, making it unlikely that it is acting as a SNARE in this pathway.
Given that such a small region of the protein is essential for nuclear fusion (possibly less than half of the 133 residue protein), the question arises as to what function Prm3p might have in nuclear fusion. Possibly Prm3p interacts with other proteins as part of a fusogenic complex assembled during mating. Alternatively, many peripheral membrane proteins can alter the intrinsic curvature of the membrane (Cho and Stahelin, 2005 ). Interactions between Prm3p and the nuclear envelope may change the physical properties of the membrane to aid fusion. Consistent with this hypothesis, expression of Prm3p in mitotic cells caused proliferation of curved nuclear membranes.
Proteins important in nuclear membrane fusion are expected to interact with each other in a fusogenic complex. Localization data suggested that Kar5p helps recruit or stabilize Prm3p at the SPB. Coimmunoprecipitation experiments showed that Prm3p and Kar5p physically interact, most likely through a small region of Kar5p residing outside the NE lumen. PRM3 and KAR5 also showed genetic interactions (synthetic bilateral defects and dosage suppression) consistent with their physical interaction. The genetic interactions between PRM3 and KAR5 were not reciprocal; overexpression of Kar5p partially suppressed prm3Δ, but overexpression of Prm3p did not suppress kar5Δ. This result suggests that Prm3p may play an accessory role, enhancing the activity or stability of proteins such as Kar5p that may act more directly in nuclear fusion.
We thank the members of the Rose laboratory for extended discussions. We thank Naz Erdeniz for constructing the kar5-hyd deletion mutations. We thank Peggy Bisher for extensive help with the immuno-EM experiments. This work was supported by National Institutes of Health grant GM-37739 (to M.D.R.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E08-10-0987) on March 18, 2009.