Ergosterol depletion independently inhibits two aspects of yeast mating: pheromone signaling and plasma membrane fusion. In signaling, ergosterol participates in the recruitment of Ste5 to a polarized site on the plasma membrane. Ergosterol is thought to form microdomains within the membrane by interacting with the long acyl chains of sphingolipids. We find that although sphingolipid-free ergosterol is concentrated at sites of cell–cell contact, transmission of the pheromone signal at contact sites depends on a balanced ratio of ergosterol to sphingolipids. If a mating pair forms between ergosterol-depleted cells despite the attenuated pheromone response, the subsequent process of membrane fusion is retarded. Prm1 also participates in membrane fusion. However, ergosterol and Prm1 have independent functions and only prm1 mutant mating pairs are susceptible to contact-dependent lysis. In contrast to signaling, plasma membrane fusion is relatively insensitive to sphingolipid depletion. Thus, the sphingolipid-free pool of ergosterol promotes plasma membrane fusion.
The fusion of two or more cells to form a larger hybrid is a fundamental process required for sexual reproduction and the development of multinuclear cells including muscle fibers, placental trophoblasts, and osteoclasts (Chen and Olson, 2005). Emerging results indicate that cell fusion also contributes to the progression of malignant diseases and to tissue regeneration by stem cells (Duelli and Lazebnik, 2003; Ogle et al., 2005). The defining event of cell fusion is the merger of two plasma membranes. Although the mechanisms of membrane fusion during intracellular transport and viral infection have been intensively investigated, there is a relative paucity of information about how membranes fuse from their extracellular surfaces in the absence of viral fusion proteins. The Caenorhabditis elegans protein Eff-1 is currently the most promising candidate fusogen. Eff-1 is essential for fusion of epithelial cells during development (Mohler et al., 2002), and ectopic expression of Eff-1 in naive cells promotes cell fusion (Podbilewicz et al., 2006). However, Eff-1 homologues have not been identified in other species. Mating in the yeast Saccharomyces cerevisiae provides an amenable genetic system that may reveal features common to diverse types of cell fusion.
Yeast mating begins with an exchange of pheromone signals between haploid cells of the opposite mating type (Elion, 2000; Bardwell, 2005). The mating pheromones bind to specific receptors that transmit their signals via a common heterotrimeric G protein. G protein activation leads to polarized recruitment of signaling proteins to the cell surface. These proteins include Cdc42, Far1, Bni1, Ste20, and the components of a MAPK cascade comprising the scaffold protein Ste5 and the kinases Ste11, Ste7, and Fus3. Among the targets of the Fus3 MAPK are Far1, which arrests the cell cycle in G1, and Ste12, the transcription factor that activates expression of mating-associated genes. After a 30-min delay, cells of the opposite mating type bind to each other to form mating pairs, which are also referred to as prezygotes. A carefully orchestrated program of cell wall remodeling then begins. The cell walls of the mating pair are first joined into a unified structure, and then the cell walls at the junction between the two cells are selectively degraded (Gammie et al., 1998). Once the intervening cell walls have been removed, the plasma membranes of the two apposing cells come into contact and fuse to form a zygote. Mating is subsequently completed by fusion of the two nuclei followed by budding of a diploid daughter cell.
In cell fusion mutants, mating pairs form but fail to fuse, leading to an accumulation of prezygotes (White and Rose, 2001). Accumulation of early prezygotes with intact cell walls separating the two partner cells indicates a cell wall remodeling defect, whereas accumulation of late prezygotes with plasma membranes in contact indicates defective membrane fusion. Although many genes are known to be involved in cell wall remodeling, the pheromone-regulated membrane protein Prm1 was the first and, until recently, the only protein implicated in plasma membrane fusion (Heiman and Walter, 2000). In addition to accumulating late prezygotes, the two cells in a prm1 mutant mating pair are prone to simultaneous lysis once their plasma membranes come into contact, suggesting that Prm1 stabilizes the assembly of nascent fusion pores (Jin et al., 2004; Aguilar et al., 2006). Once a fusion pore forms, it must expand to permit the nuclei to fuse. Fusion pore expansion is regulated by Fus1, which also has a critical but independent role in cell wall remodeling (Nolan et al., 2006).
Although phospholipid bilayer membranes are typically viewed as passive participants in protein-mediated membrane fusion, the lipid composition of a membrane has profound effects on biophysical properties that may affect a membrane's fusability, including intrinsic curvature, thickness, stiffness, and permeability. Compared with intracellular membranes, the yeast plasma membrane is highly enriched in ergosterol, just as mammalian plasma membranes are highly enriched in cholesterol (Schneiter et al., 1999). Within a membrane, sterols can interact with the long saturated acyl chains of sphingolipids to dynamically partition into membrane microdomains, which are often referred to as lipid rafts (Mukherjee and Maxfield, 2004; Hancock, 2006). Rafts are thought to form by dense packing of the flexible acyl chains of sphingolipids against the flat rigid sterol molecule to produce a thickened liquid-ordered phase membrane, which still permits rapid lateral diffusion. Association of proteins with a membrane fraction that is resistant to detergent extraction at 4°C is commonly cited as evidence that the proteins are concentrated in lipid rafts, but it is now understood that chilling cells and extracting phospholipids can induce interactions that do not exist in living cells (Lichtenberg et al., 2005). Furthermore, the large (micrometer scale) and stable liquid-ordered microdomains found in artificial membranes at reduced temperatures do not exist in most biological membranes. Instead, lipid raft–associated glycosylphosphatidylinositol-anchored proteins have an apparently uniform cell surface distribution by confocal microscopy, and sophisticated fluorescence resonance energy transfer (FRET) techniques were required to detect <5-nm clusters of three to four proteins (Sharma et al., 2004). Indeed, the difficulty of unambiguously detecting nanometer-scale domains in living cells has led some to question whether lipid rafts actually exist (Munro, 2003; Douglass and Vale, 2005). One emerging model is that functional membrane microdomains are formed via cooperative interactions between nanoscale lipid domains, membrane-associated proteins, and the actin cytoskeleton (Viola and Gupta, 2007).
We uncovered two ergosterol biosynthesis genes in a visual screen for yeast mutants arrested at the plasma membrane fusion stage of mating. Plasma membrane ergosterol promotes rapid fusion and acts independently of the Prm1 protein. Ergosterol depletion also interfered with the response to mating pheromones, but robust pheromone signaling was not essential for membrane fusion. Sphingolipids were depleted to investigate the potential involvement of lipid rafts in signaling and fusion. Signaling depends on a balanced ratio of ergosterol to sphingolipids, whereas fusion is more dependent on the total amount of ergosterol, indicating that signaling and fusion are regulated by different pools of ergosterol.
Discovery of the erg6 mating defect
The yeast knockout collection was screened for mutants that were defective at the plasma membrane fusion stage of mating by crossing pairs of MATa and MATα strains with the same gene deleted in each mating partner (Fig. S1). Most cell fusion mutants accumulate only early prezygotes, but late prezygotes that were identical to those originally described for prm1 were readily detected in an erg6 mating, implicating ergosterol in plasma membrane fusion (Fig. 1).
Electron microscopy confirmed that the two plasma membranes of an erg6 mating pair could be in intimate contact over an extended zone of apposition (Fig. 2).
For comparison, many prm1 mating pairs also had an extended zone of membrane apposition, whereas the two plasma membranes in fus1 mating pairs were separated by cell walls. Two additional features are documented in the prm1 mating pair: clustered vesicles adjacent to the cell wall remnants and a myelin sheath-like whorl formed from the two plasma membranes at one point within the zone of plasma membrane apposition. Similar features were described in an earlier study of yeast mating (Gammie et al., 1998) and were also found in some erg6 mating pairs. Finally, the erg6 and prm1 mating pairs both have cell wall fragments near the base of the cytoplasmic finger that lie perpendicular to the remnant cell wall separating the two plasma membranes. Thus, the cell wall may be able to regenerate at a later time if plasma membrane fusion is inhibited.
Phenotypic differences between erg6 and prm1
Similar to prm1, the erg6 mating phenotype is heterogeneous, containing a mixture of fused mating pairs and early and late prezygotes (Fig. 3).
However, erg6 matings had a higher proportion of early prezygotes as well as an increased percentage of haploid cells that did not engage a mating partner, suggesting that ergosterol is also involved in an earlier step in the mating pathway (see Fig. 5). A further distinction between erg6 and prm1 is that the percentage of erg6 mating pairs with cytoplasmic projections declined over time with an accompanying increase in fused mating pairs (unpublished data). Thus, plasma membrane fusion is delayed rather than blocked by altering the sterol composition of cellular membranes.
The dynamics of individual cell fusion events were examined by time-lapse imaging of MATaerg6 GFP cells mating to MATα erg6 RFP cells. Similar to previous results with prm1 (Nolan et al., 2006), fusion pore permeance calculated from the rate of GFP diffusion between cells was not strongly reduced in erg6 mating pairs (unpublished data). Under standard time-lapse imaging conditions, the two cells of a prm1 mating pair often lyse after achieving plasma membrane contact (Jin et al., 2004; Nolan et al., 2006). The lysis/fusion ratio was >50 in prm1 mating pairs but <0.1 in wild-type mating. In the erg6 videos, there were 29 fusions and 5 simultaneous lysis events. Thus, we conclude that the two plasma membranes of an erg6 mating pair are susceptible to lysis once they come into contact, but they are substantially more stable during fusion than prm1 membranes. The differences between the erg6 and prm1 phenotypes suggest that ergosterol and Prm1 might function in different processes leading to plasma membrane fusion.
Plasma membrane ergosterol promotes fusion
To confirm the importance of ergosterol during plasma membrane fusion, wild-type mating pairs were treated with antibiotics that inhibit ergosterol biosynthesis or bind to plasma membrane ergosterol. Fluconazole (FLZ) is an azole antibiotic that interferes with lanosterol demethylation, an essential step in the ergosterol biosynthetic pathway (Fig. 3 A). Treatment with 1 mg/ml FLZ, a dose which is 200-fold above the ID50, has no effect on the growth rate of a log-phase culture for the first 6 h, indicating that the preexisting pool of ergosterol is sufficient for essential functions until it is turned over and/or diluted by expansion of the culture (Fig. S2 A). Nevertheless, ergosterol synthesis is immediately inhibited, leading to lanosterol accumulation within 30 min (Fig. S2 B). Prezygotes were not detected when yeast were mated on FLZ plates, indicating that ongoing ergosterol synthesis is not essential for mating. However, late prezygotes accumulated when MATa and MATα cells were individually pretreated with FLZ for 3 h before mating (Fig. 3 B). The cellular ergosterol concentration that promotes plasma membrane fusion must be higher than that required for growth because a 3-h FLZ pretreatment inhibits fusion but has no effect on the growth rate.
Nystatin is a polyene antibiotic that binds to ergosterol in the yeast plasma membrane and eventually forms channels in the membrane leading to cell lysis (Silva et al., 2006). Yeast treated with 32 μg/ml nystatin failed to form mating pairs, but late prezygotes were found when mating pairs were allowed to assemble during a 30-min preincubation and then transferred to a nystatin plate. Importantly, the two cells of these late prezygotes maintained their cytoplasmic fluorescence, which is an indication that they had not yet lysed. The FLZ and nystatin mating results indicate that the plasma membrane pool of ergosterol contributes to cell fusion and argue against the alternative possibility that newly synthesized ergosterol in the secretory pathway is needed to target a fusion protein to sites of plasma membrane contact.
Structural features of ergosterol that modulate membrane fusion
Although zymosterol synthesis is essential for aerobic growth, later steps in the ergosterol synthesis pathway are not, and the late enzymes do not obligatorily act in a linear pathway (Parks and Casey, 1995; Heese-Peck et al., 2002). To identify structural features of ergosterol that are important for cell fusion, MATa GFP and MATα RFP strains with deletions in each of the nonessential erg genes were mated and scored for prezygote accumulation. Mutations in erg2, 3, and 6 inhibited plasma membrane fusion, whereas mutations in erg4 and 5 did not (Fig. 3 C). Thus, plasma membrane fusion appears to depend on both a proper double bond configuration in the B ring (erg2 and 3) and methylation of the tail (erg6), although it is possible that one or more of the erg mutations inhibits fusion indirectly by altering the activity of other enzymes in the ergosterol biosynthesis pathway. Some of the erg mutants have actin polarity, endocytosis, and/or homotypic vacuole fusion phenotypes (Kato and Wickner, 2001; Heese-Peck et al., 2002), but the subset of erg mutants with mating defects is unique. In particular, erg3, which had the strongest plasma membrane fusion defect, does not interfere with α-factor binding, localization and endocytosis of the α-factor receptor, or the polarized distribution of actin patches and cables (Heese-Peck et al., 2002). We conclude that the mating phenotype is unlikely to be an indirect consequence of defects in these other processes.
Interactions between PRM1 and ergosterol
The prm1 and erg mutations have low penetrance, allowing a significant level of plasma membrane fusion even when they are deleted from both cells in a mating pair. Fusion was normal in erg6 cross wild-type matings, regardless of whether the mutation was in the MATa or MATα cell, as was previously shown for prm1 and many other cell fusion mutants (Heiman, and Walter, 2000). In contrast, there was essentially no plasma membrane fusion and an increased accumulation of late prezygotes when two prm1 erg6 double-mutant strains were mated (Fig. 4 A).
Similar results were obtained with double-mutant combinations between prm1 and erg2 or 3. The additive effect of the prm1 and erg mutations supports the conclusion that Prm1 and ergosterol function in independent processes leading to plasma membrane fusion.
One implication of the double-mutant results is that ergosterol depletion does not inhibit mating by interfering with Prm1 targeting to sites of cell–cell interaction. This inference was directly tested by depleting ergosterol with a FLZ pretreatment and then observing the localization of GFP-Prm1 in arrested mating pairs (Fig. 4 B). GFP-Prm1 was concentrated at sites of cell–cell contact in 71.4% of the FLZ-pretreated early prezygotes (n = 388) compared with 74.9% of the untreated controls (n = 339).
To examine the effect of varying Prm1 expression on plasma membrane fusion, an HA epitope-tagged form of the PRM1 gene was placed under the control of a series of constitutively active promoters (Mumberg et al., 1995). Western blotting with an anti-HA antibody confirmed that the GPD promoter yielded the highest HA-Prm1 expression, with progressively lower expression from the TEF, ADH1, and CYC promoters (unpublished data). When these plasmids were transformed into both mating partners, HA-Prm1 expression from the weak CYC1 promoter was sufficient to restore normal mating to prm1 mutant mating pairs. In contrast, a progressive increase in Prm1 expression yielded a progressive increase in cell fusion in prm1 erg6 double-mutant mating pairs (Fig. 4 C). Thus, ergosterol depletion enhances the dependence of plasma membrane fusion on high Prm1 expression. Interestingly, only the highest level of PRM1 expression driven by the GPD promoter was sufficient to restore mating to the efficiency found when PRM1 is expressed from its native promoter in the erg6 mutant.
Ergosterol promotes pheromone signaling
erg6 matings had a high percentage of haploid cells that failed to interact with a mating partner. The erg6 mutant also had a diminished morphogenic response to pheromones, with a lower percentage of cells extending mating projections to form the pear-shaped cells known as shmoos. These observations suggested that sterols modify the response to mating pheromones. To further investigate this possibility, quantitative measurements of the transcriptional response to mating pheromones in erg mutant strains were made with a PFUS1-lacZ reporter construct (Fig. 5 B).
The results showed a positive correlation between reduced FUS1 induction and defective plasma membrane fusion, with erg3 showing the strongest defect in both processes.
Because ergosterol is concentrated in the plasma membrane (Schneiter et al., 1999), we tested the hypothesis that ergosterol depletion inhibits membrane-localized events in the pheromone signaling pathway. One critical signaling event is recruitment of the Ste5 MAPK scaffold to polarized sites on the plasma membrane. As illustrated in Fig. 5 A, Ste5 binds to Gβγ and Cdc24 (a nucleotide exchange factor for Cdc42) and interacts with membrane lipids via an N-terminal amphipathic helix known as the plasma membrane domain and a cryptic pleckstrin homology domain, both of which are specific for phosphatidylinositol-4-phosphate (PI(4,5)P2; Whiteway et al., 1995; Pryciak and Huntress, 1998; Winters et al., 2005; Garrenton et al., 2006). As a MAPK scaffold, Ste5 recruits the Ste11, Ste7, and Fus3 kinases to the membrane. The ultimate effect of recruiting Ste5 to the membrane is to facilitate phosphorylation of Ste11 by Ste20, thereby activating the MAPK cascade.
Ste5 recruitment was examined using a GFP-Ste5 fusion protein. In wild-type MATa cells, α-factor triggers rapid translocation of a portion of the intracellular pool of GFP-Ste5 to a focused spot on the plasma membrane that corresponds to the future site of mating projection growth (Pryciak and Huntress, 1998; Mahanty et al., 1999). In mating pairs, GFP-Ste5 was found at sites of cell–cell contact until the moment of fusion, when it diffused throughout the cytoplasm of the fused zygote (Fig. 5 C). The percentage of nonbudded cells with a polarized GFP-Ste5 spot was reduced in the erg3 mutant (Fig. 5 D), suggesting that ergosterol promotes recruitment of a signaling complex to the plasma membrane. The bni1Δ mutant was used as a control for this experiment because the actin cable nucleation activity of Bni1 was previously shown to facilitate GFP-Ste5 translocation (Qi and Elion, 2005). In contrast to bni1, the erg3 mutant has normal actin cables (Heese-Peck et al., 2002), indicating that the failure to recruit GFP-Ste5 is not caused by an underlying defect in cell polarization. In conclusion, altering the sterol composition of the plasma membrane interferes with recruitment of Ste5 to the site of signaling.
The critical role of Ste5 recruitment was further defined by an epistasis experiment with Ste5-CTM, a chimeric protein in which the transmembrane anchor of Snc2 is fused to the C terminus of Ste5 (Pryciak and Huntress, 1998). Targeting of Ste5-CTM to the plasma membrane restored pheromone signaling to ergosterol-depleted cells (Fig. S3), confirming that ergosterol depletion inhibits membrane-localized events in the pheromone signaling pathway.
The relationship between pheromone signaling and plasma membrane fusion
An identical subset of ergosterol biosynthesis mutants reduced both pheromone signaling and plasma membrane fusion (Figs. 3 C and 5 B). Given the central role of pheromones in regulating the overall mating process, a reduction in pheromone responsiveness might indirectly cause the plasma membrane fusion defect. To investigate this possibility, cell fusion was assayed in the temperature-sensitive ste5ts mutant, which fails to mate at 34°C (Hartwell, 1980). Adjusting the temperature of ste5ts cells acts as a rheostat to control the degree of pheromone-induced FUS1 expression without creating a subpopulation of nonresponsive cells (Fig. S4). Thus, this mutant provides an ideal system for examining the effect of reduced pheromone responsiveness. In a 24°C mating reaction, <10% of ste5ts mating pairs arrested before fusion (Fig. 6 A).
Early prezygotes accumulated at 30°C, potentially because of reduced expression of FUS1 and other pheromone-regulated genes that are involved in cell wall remodeling, but there was not a significant accumulation of late prezygotes. Apparently, a higher level of signaling is required for the completion of cell wall remodeling than for plasma membrane fusion. A similar defect in cell wall remodeling, but not plasma membrane fusion, was previously found in mutants with reduced a-factor synthesis (Brizzio et al., 1996). The more modest pheromone signaling defect of a bni1 mutation (Qi and Elion, 2005) did not result in accumulation of either early or late prezygotes in our standard mating conditions. Because a robust pheromone response is not essential for plasma membrane fusion, reduced pheromone signaling cannot be the sole cause of the membrane fusion defect associated with ergosterol depletion.
Because the pheromone-regulated protein Prm1 had to be expressed at high levels to promote fusion in erg6 mating pairs, we examined the effect of boosting the pheromone response to above normal levels (Fig. 6 B). erg6 cells induced with a combination of α-factor and STE5-CTM had twofold higher PFUS1-lacZ expression than wild-type cells induced with α-factor alone. In a mating reaction, STE5-CTM expression reduced the number of erg6 cells that could form mating pairs by 70% (unpublished data), possibly by binding to Gβγ and thereby competitively inhibiting Gβγ–Far1 interactions (Butty et al., 1998; Winters et al., 2005). However, the mating pairs that were able to form between STE5-CTM–expressing erg6 cells were more likely to fuse and less likely to arrest as either early or late prezygotes. Only a small fraction of this increased fusion was recapitulated by PRM1 overproduction, indicating that additional pheromone-regulated processes contribute to the efficiency of both cell wall remodeling and plasma membrane fusion. These processes could include posttranslational activation and polarized recruitment of fusion proteins and/or synthesis of additional pheromone-regulated genes. STE5-CTM expression also promoted fusion of wild-type cells that were mated in suboptimal conditions (synthetic complete [SC] galactose plates for 3 h at 30°C), raising the percentage of fused pairs from 92 to 98%. In conclusion, the membrane fusion defect resulting from ergosterol depletion can be overcome by enhancing the efficiency of other processes leading to fusion.
Polarized targeting of free ergosterol in mating pairs
It was previously proposed that lipid rafts enriched in ergosterol and sphingolipids contribute to mating by facilitating the targeting of Fus1 and other membrane proteins to mating projections (Bagnat and Simons, 2002). The proposal that lipid rafts contribute to membrane protein targeting has been challenged by others (Valdez-Taubas and Pelham, 2003), and we found no obvious defect in Fus1-GFP targeting to mating projections in the erg mutants (unpublished data). If lipid rafts were required for Fus1 targeting, ergosterol depletion should inhibit the Fus1-dependent processes of cell wall remodeling and fusion pore expansion. In contrast, ergosterol depletion inhibits pheromone signaling and plasma membrane fusion, as shown in Figs. 1–3,23 and 5. We therefore conclude that the plasma membrane fusion defect in erg mutant mating pairs is not caused by a primary defect in lipid raft–mediated membrane protein targeting.
An important observation, which was originally used to support the concept that lipid rafts promote polarized transport, is that filipin, a sterol ligand, stains the tip of the mating projection in shmoos (Bagnat and Simons, 2002). We confirmed this observation using a more rapid filipin staining procedure (see Materials and methods) to preferentially stain the plasma membrane and minimize the time available for sterol redistribution (Fig. 7 A).
The bright filipin staining at the shmoo tip does not represent a general increase in the density of plasma membrane because the plasma membrane protein Sso2-GFP is not concentrated there. In genuine mating pairs, filipin stained sites of cell–cell contact (Fig. 7 B). Polarized filipin staining was maintained in arrested fus1 prezygotes and redistributed to the zygotic bud after fusion. This filipin staining pattern is consistent with a role for polarized ergosterol in pheromone signaling and plasma membrane fusion.
We next used erg mutant shmoos to examine the effect of sterol structure on filipin staining (Fig. 8 A).
The percentage of shmoos with polarized filipin staining was strongly reduced in the erg2, 3, and 6 mutants (Fig. 8 B), with a corresponding reduction in the ratio of shmoo tip to cell body filipin intensity. Filipin formed bright speckles on erg6 cells that were randomly distributed over the surface of the cell and its mating projection. A lesser degree of speckling was found in the erg2, 3, and 5 mutants (unpublished data). Speckling might result from filipin-induced sterol redistribution in strains with ergosterol precursors that can diffuse more rapidly in the plasma membrane (Valdez-Taubas and Pelham, 2003). In summary, the absence of a concentration of smoothly polarized filipin staining in mating projections correlates with defective pheromone signaling and plasma membrane fusion.
Surprisingly, filipin does not stain lipid rafts as was previously assumed. Mitotic cells stained poorly with filipin (Fig. 8 A), despite the fact that ergosterol represents 40% of plasma membrane lipids (Zinser et al., 1991; Schneiter et al., 1999). A potential explanation for this phenomenon is that ergosterol binds avidly to sphingolipids (Xu et al., 2001), which are also enriched in the plasma membrane, and that sphingolipids impede the access of filipin to ergosterol. This model was tested in lcb1ts cells, which have a 50% reduction in sphingolipid synthesis when grown under permissive conditions (Zanolari et al., 2000; Hearn et al., 2003) and also have a modest reduction in the concentration of plasma membrane ergosterol (Baumann et al., 2005). Mitotic lcb1ts cells had bright uniform filipin staining on their plasma membranes (Fig. 8 A). Equally bright filipin staining was found after treating lcb1ts cells with α-factor, but the filipin was modestly polarized toward the shmoo tip (shmoo tip to cell body fluorescence ratios: wild type, 2.5 ± 0.7 [n = 40]; lcb1ts, 1.5 ± 0.5 [n = 41]).
We considered several alternative explanations for the bright filipin staining of mitotic lcb1ts cells. One possibility is that a compromised cell wall allows enhanced access of filipin to the plasma membrane. However, disrupting the cell wall integrity MAPK cascade with an mpk1 deletion (Levin, 2005) does not enhance filipin staining (Fig. S5 A). Another possibility is that the bright filipin staining of lcb1ts cells is a secondary consequence of defects in endocytosis and actin organization. These defects can be suppressed by overproducing the Pkh1 or Ypk1 kinases, which are activated by sphingoid base intermediates in the sphingolipid biosynthesis pathway (Sun et al., 2000; Friant et al., 2001; deHart et al., 2002; Liu et al., 2005). However, PKH1 and YPK1 overproduction in lcb1ts cells had no effect on filipin staining (Fig. S5 B). These results suggest that bright filipin staining of the lcb1ts plasma membrane is a direct consequence of alterations in the lipid composition of the membrane.
Various steps in the sphingolipid biosynthetic pathway (Dickson et al., 2006) were inhibited to identify structural features that enable sphingolipids to inhibit the binding of filipin to ergosterol. The first step, conjugation of palmitoyl-CoA to serine to form sphingoid bases, was inhibited by myriocin (ISP-1). Addition of the second acyl chain, a C-26 very long chain fatty acid, was inhibited by fumonisin BI. Cells treated with either inhibitor stained brightly with filipin, which is consistent with the possibility that filipin staining is competitively inhibited by hydrophobic interactions between ergosterol and the long flexible acyl chains of sphingolipids (Fig. S5 C). The myriocin result was expected because LCB1 encodes a subunit of serine palmitoyltransferase, the enzyme inhibited by myriocin. The fumonisin result further confirms that bright filipin staining is not a secondary consequence of reduced sphingoid base signaling because sphingoid bases accumulate in fumonisin-treated cells (Wu et al., 1995). In contrast to inhibiting acylation, inhibiting conjugation of mannose and phosphatidylinositol to the hydrophilic headgroups of sphingolipids by deleting the CSG2 and IPT1 genes did not give rise to bright filipin staining (Fig. S5 D). We conclude that acylated sphingolipids inhibit the interaction between filipin and ergosterol. Thus, the bright filipin staining at the tips of mating projections indicates a polarized accumulation of accessible sterols.
Ergosterol promotes PI(4,5)P2 polarity
Because Ste5 binds to PI(4,5)P2 (Winters et al., 2005; Garrenton et al., 2006), we wondered if PI(4,5)P2 might also have a polarized distribution in mating yeast. Compared with ergosterol and sphingolipids, PI(4,5)P2 is a minor component of the plasma membrane. It is concentrated on the cytoplasmic leaflet of the plasma membrane by virtue of local synthesis by Mss4 and degradation during endocytosis by lipid phosphatases homologous to synaptojanin (Stefan et al., 2002). PI(4,5)P2 has been reported to associate with lipid rafts, but this proposal is controversial. PI(4,5)P2 from mammalian cells floats with detergent-resistant membranes (Pike and Casey, 1996). In contrast, PI(4,5)P2 has a negligible association with cholesterol by FRET, although the FRET signal can be substantially enhanced by addition of as little as 0.01% Triton X-100 (van Rheenen et al., 2005). Although PI(4,5)P2 does not possess the long flexible acyl chains required for hydrophobic interactions between sphingolipids and sterols, interactions between PI(4,5)P2 and sterols can be promoted by lipid raft–associated acidic proteins (Epand et al., 2004). Intracellular PI(4,5)P2 was detected with 2×PHPLCγ-GFP, a fusion of GFP to two copies of the pleckstrin homology domain of phospholipase Cγ (Stefan et al., 2002). In pheromone-treated yeast, 2×PHPLCγ-GFP fluorescence was modestly concentrated on the surface of mating projections (Fig. 9 A).
This polarized PI(4,5)P2 localization was not an illusion resulting from the shape of the plasma membrane within the optical section because Sso2-GFP was not polarized under identical conditions. Interestingly, the intensity of 2×PHPLCγ-GFP fluorescence was somewhat reduced at the very tip of the mating projection, where GFP-Ste5 is found. PI(4,5)P2 could be less concentrated at the tip of the mating projection if this site is a target for exocytosis of PI(4,5)P2-depleted secretory vesicles or for endocytosis and its associated PI(4,5)P2-directed lipid phosphatases. Alternatively, an appearance of PI(4,5)P2 depletion could result from competition for PI(4,5)P2 binding between GFP-Ste5 and 2×PHPLCγ-GFP (Balla et al., 2000), with GFP-Ste5 winning the contest at the shmoo tip because its localization there is reinforced by interactions with other polarized proteins. With respect to the potential role of lipid rafts in PI(4,5)P2 localization, the erg6 mutant had a 25% decrease (P < 0.01) in the percentage of shmoos with polarized 2×PHPLCγ-GFP (Fig. 9 B). We conclude that a reduction in PI(4,5)P2 polarization may contribute to reduced GFP-Ste5 recruitment and pheromone signaling upon ergosterol depletion.
A balanced ergosterol to sphingolipid ratio promotes signaling
Signaling events at the cell surface are often confined within membrane microdomains enriched in both sterols and sphingolipids, which serve as platforms for protein complex assembly (Golub et al., 2004). In mammalian cells, these microdomains range in size from 10 to 200 nm and are therefore too small to be resolved by wide-field light microscopy (Jacobson et al., 2007). Thus, microdomains of sterol–sphingolipid interaction could be present at the tip of mating projections and at contact sites in prezygotes, despite our previous conclusion that these sites are enriched in sphingolipid-free ergosterol. As an alternative method to address the potential role of membrane microdomains in signaling, we measured pheromone responsiveness in lcb1ts cells. Because the lcb1ts mutation reduces the rate of sphingolipid synthesis (Zanolari et al., 2000; Hearn et al., 2003), lcb1ts cells should have fewer ergosterol–sphingolipid complexes and an excess of free ergosterol. FUS1 reporter expression was reduced by 70% in the lcb1ts mutant, suggesting that the sphingolipid-associated pool of ergosterol is required for optimal signaling. More importantly, a 3-h FLZ pretreatment to deplete ergosterol enhanced pheromone-induced PFUS1-lacZ expression to near wild-type levels in the lcb1ts mutant but had little effect on control cells (Fig. 10 A).
These results were confirmed using myriocin as an alternative method to deplete sphingolipids at both 25 and 30°C (unpublished data). Thus, a balanced ergosterol to sphingolipid ratio is more critical for FUS1 induction than the overall amount of either lipid. Quantitative measures of polarized morphogenesis in FLZ-treated cells support the PFUS1-lacZ expression results. After a 3-h FLZ pretreatment, the percentage of cells that formed mating projections when challenged with 6 μM α-factor decreased by 40% in the wild type and increased by 20% in the lcb1ts mutant, and the ratio of shmoo tip to cell body filipin fluorescence decreased by 30% in the wild type and increased by 20% in the lcb1ts mutant. In conclusion, these results suggest that ergosterol and sphingolipids must assemble into stoichiometric complexes to promote pheromone signaling, which is consistent with the possibility that ergosterol/sphingolipid-enriched membrane microdomains serve as a platform to promote the association of Ste5 with Ste20 and other signaling proteins.
Sphingolipids have a minor role in plasma membrane fusion
Microdomains enriched in ergosterol and sphingolipids could potentially promote fusion by concentrating and activating fusion proteins. However, sphingolipid depletion with either the lcb1ts mutation or the biosynthetic inhibitor myriocin did not interfere with plasma membrane fusion at normal total ergosterol levels. Thus, sphingolipids have a more critical role in pheromone signaling than in plasma membrane fusion. To further investigate the possible participation of membrane microdomains in fusion, wild-type and lcb1ts cells were treated with FLZ before mating (Fig. 10 B). A 3-h FLZ pretreatment inhibited fusion to a similar extent in lcb1ts mutant and control matings. However, after a 5-h FLZ pretreatment, the lcb1ts mutant had a dramatically stronger fusion defect than the control. In summary, normal levels of total ergosterol promote efficient plasma membrane fusion even if sphingolipid synthesis is inhibited, but simultaneous depletion of ergosterol and sphingolipids revealed a secondary requirement for a low level of ergosterol–sphingolipid interaction. These results confirm that ergosterol has distinct functions in signaling and plasma membrane fusion.
Ergosterol promotes at least two independent processes during mating. In response to mating pheromones, ergosterol promotes recruitment of Ste5 to the site of signaling on the plasma membrane. After mating pair assembly and cell wall remodeling, ergosterol facilitates plasma membrane fusion. Ergosterol is thought to interact with sphingolipids to promote the formation of membrane microdomains (lipid rafts) that concentrate the activity of associated membrane proteins. Pheromone signaling is highly sensitive to sphingolipid depletion, suggesting the involvement of lipid rafts, whereas sphingolipid depletion only interfered with plasma membrane fusion if ergosterol was also depleted. Thus, pheromone signaling and membrane fusion depend on different pools of ergosterol.
Ergosterol polarity in mating yeast
Ergosterol assumes a polarized distribution during mating. Filipin-accessible ergosterol is concentrated at the tips of mating projections and at sites of cell–cell contact in mating pairs. Although originally interpreted as a lipid raft marker (Bagnat and Simons, 2002), filipin actually stains sphingolipid-free ergosterol because staining is brighter in the lcb1ts sphingolipid synthesis mutant. A recent study found that the general polarization of Laurdan fluorescence is strongest in mating projections (Proszynski et al., 2006). Laurdan provides an indication of lipid order by measuring water penetration into the lipid bilayer. In liposomes, lipid rafts have a high general polarization value, but it is not certain that this correlation extends to living cells. The filipin and Laurdan results clearly indicate that the tip of the mating projection has different lipid composition and packing than the cell body, but the exact nature of these differences requires further study. Nevertheless, the positive correlation among erg mutants between smoothly polarized filipin staining, strong pheromone signaling, and efficient plasma membrane fusion suggests that the local membrane environment must be properly controlled for efficient mating.
Sterols and sphingolipids promote pheromone signaling
Given that lipid rafts have long been considered as potential signaling platforms (Simons and Ikonen, 1997; Simons and Toomre, 2000), it is somewhat surprising that this study provides the first evidence that membrane lipids influence signal transduction in yeast. Pheromone-induced PFUS1-lacZ transcription was attenuated by the erg2, 3, and 6 and lcb1ts mutations and also by inhibiting ergosterol synthesis with FLZ or inhibiting sphingolipid synthesis with myriocin. The restoration of normal signaling when ergosterol and sphingolipids are both depleted provides compelling evidence that signaling depends on interactions between ergosterol and sphingolipids rather than on the function of either lipid in isolation. Two independent results indicate that ergosterol promotes plasma membrane–localized events in the signal transduction pathway. First, the erg3 mutant had reduced recruitment of GFP-Ste5 to shmoo tips. Second, artificially targeting Ste5 to the plasma membrane partially suppressed the signaling defect resulting from FLZ pretreatment. These results do not exclude the possibility that ergosterol promotes membrane-associated signaling interactions before Ste5-GFP recruitment. The pheromone response pathway has multiple components whose interactions could be modulated by the local lipid environment (Fig. 5 A). These include seven transmembrane domain receptors (Ste2 and 3), lipid-anchored proteins (Ste18/Gγ and Cdc42), and proteins with lipid-binding motifs (Ste5 and Far1). In addition, interactions between PI(4,5)P2 and ergosterol, as documented by reduced PI(4,5)P2 polarization in the erg6 mutant, may influence the localization and activity of PI(4,5)P2 binding proteins such as Ste5 and Far1. Further investigation of the role of ergosterol, sphingolipids, and PI(4,5)P2 in promoting interactions between signaling proteins should be conducted using methods, such as FRET, that can detect in vivo interactions on a submicroscopic scale (Jacobson et al., 2007).
Plasma membrane fusion in yeast mating pairs
The mechanism of plasma membrane fusion has been difficult to analyze because there are so few reagents that inhibit this step in the mating process. We have now identified three new mutations, erg2, 3, and 6, that cause an accumulation of mating pairs with plasma membranes that are in contact but not fused. This mating defect was documented by the presence of GFP- or RFP-labeled cytoplasmic fingers, which can only extend from a cell into its mating partner after the cell wall has been degraded, and by electron microscopy, where it is possible to directly observe an extensive zone of intimate contact between the two plasma membranes. The erg mutant phenotypes pointed to the involvement of ergosterol in plasma membrane fusion, and this was confirmed by the accumulation of late prezygotes after inhibiting ergosterol synthesis with FLZ or sequestering membrane ergosterol with nystatin. None of these mutations or treatments completely inhibits membrane fusion, possibly because ergosterol biosynthetic intermediates can partially replace the missing ergosterol. Two earlier studies reported mating defects for the erg6 mutant but did not describe the critical contributions of ergosterol to signaling and membrane fusion (Tomeo et al., 1992; Bagnat and Simons, 2002).
The prm1 and erg6 mutations each inhibit plasma membrane fusion but they do so in different ways, as highlighted by the additive effect of deleting both genes. prm1 mating pairs have a high propensity to lyse once the two membranes come into contact, whereas erg6 mating pairs do not. We previously proposed that prm1 lysis occurs via uncoordinated activation of the normal fusion machinery, but a definitive test of this model awaits the identification of a fusion protein (Jin et al., 2004). In our previous study, lysis was found to occur more frequently in time-lapse videos. The recent finding that extracellular Ca2+ increases the likelihood that prm1 mating pairs will fuse rather than lyse (Aguilar et al., 2006) provides an explanation for this phenomenon. The optically clear agarose used for microscopy has a lower Ca2+ concentration than the crude agar used for plate mating assays. Ca2+ has been proposed to promote fusion by activating a membrane repair process that protects against lysis (Aguilar et al., 2006), but this model fails to explain why fusion of prm1 mating pairs is also promoted by increasing membrane tension with a hypotonic shock (Nolan et al., 2006).
Recent reports have described two other mutations, kex2 and fig1, that enhance the prm1 fusion defect (Aguilar et al., 2006; Heiman et al., 2007). Kex2 is a Golgi-localized endoprotease involved in the processing of α-factor and a variety of other substrates. This protease activity is essential for the Kex2 plasma membrane fusion function but the relevant substrates are unknown. Arrested kex2 mating pairs had membrane blebs and giant barren vacuoles that were not found in erg6 or prm1 mating pairs, suggesting that kex2 defines a third independent function leading to membrane fusion (Heiman et al., 2007). Fig1 is a pheromone-inducible membrane protein that promotes Ca2+ influx during mating and is required for rapid cell death in response to high doses of α-factor (Erdman et al., 1998; Muller et al., 2003; Zhang et al., 2006). Because fig1 mating pairs were originally found to arrest before cell wall remodeling (Erdman et al., 1998), we reexamined the fig1 mating phenotype in both the BY4741 and W303 genetic backgrounds. After a 3-h mating, 3% of fig1 mating pairs had arrested as late prezygotes. Thus, Fig1 appears to be a minor participant in the plasma membrane fusion process.
Sterols, sphingolipids, and membrane fusion
Sterols have many functions within membranes. In addition to their critical role in establishing membrane microdomains, they also modify membrane thickness, permeability, fluidity, and curvature. Which of these properties is relevant to plasma membrane fusion in mating yeast remains to be discovered, but the low sensitivity to sphingolipid depletion suggests that interactions between ergosterol and sphingolipids play a minor, although still potentially significant, role. Sterols are essential for many viral and intracellular membrane fusions (Salaun et al., 2004; Teissier and Pecheur, 2007). In contrast, immature sperm actually have higher cholesterol levels than the optimum for acrosome exocytosis (Belmonte et al., 2005). Sterol-dependent clustering of viral fusion proteins, cellular receptors, and SNAREs is critical for fusion in various systems, but these clusters can be distinct from biochemically defined lipid rafts (Lang et al., 2001; Percherancier et al., 2003; Takeda et al., 2003; Fratti et al., 2004; Yi et al., 2006). In addition, a protein clustering–independent role for cholesterol is supported by the partial restoration of fusion after adding lipids with negative curvature to cholesterol-depleted cortical granules (Churchward et al., 2005) and also by the observation that the optimal concentration of sterols and sphingolipids for protein-free liposome fusion matches the lipid composition of synaptic vesicles (Haque et al., 2001).
The data presented in this paper support a model whereby the sterol content of the plasma membrane determines its propensity to be fused by a Prm1-regulated protein complex. Inhibiting ergosterol synthesis increases the potential energy cost of fusion, but this barrier can be overcome by increasing the mating time or by amplifying the pheromone response. In the absence of Prm1, uncoordinated activity of the currently unknown fusion proteins is insufficient to fuse ergosterol-depleted membranes.
Materials And Methods
Strains, reagents, and plasmids
The yeast strains used in this study were derived from strains produced by the Saccharomyces Genome Deletion Project (http://www-sequence.stanford.edu/group/yeast_deletion_project/deletions3.html) in BY4741 and BY4742 unless otherwise noted. Strains from the quality control collection of knockout strains were provided by M. Snyder (Yale University, New Haven, CT). The parental deletion strains were verified by PCR. MATa strains were transformed by the lithium acetate method with cytoplasmic GFP or plasma membrane–localized GFP-Sso2. MATα strains were transformed with either of two RFPs: DsRed or mCherry. The prm1 erg double-mutant strains were constructed by transformation of single mutants with a prm1∷HIS3 disruption plasmid. The MATa ste5ts strain PPY423 (MATa ste5-3ts cry1 his4 leu2 lys2 tyr1 ura3 sup4-3ts) was obtained from P. Pryciak (University of Massachusetts Medical Center, Worcester, MA). A MATα ste5ts strain was constructed by switching the mating type of PPY423 with a plasmid encoding the HO endonuclease. The MATa lcb1ts BAR1 strain EGMY600 was constructed by crossing RH2607 (MATa lcb1-100 his4 ura3 leu2 bar1; obtained from H. Reizman, Université de Genève, Geneva, Switzerland) to BY4742 prm1 and backcrossing twice with BY4741. RH2607 (lcb1ts bar1) failed to mate to an erg6 partner, as previously shown (Bagnat and Simons, 2002). However, separating the lcb1ts and bar1 alleles revealed that lcb1ts mates normally at 25°C, whereas mutations in the Bar1 α-factor protease cause a mating defect. The mpk1 strain DL454 (MATampk1∷TRP1 leu2 trp1 ura3 his4 can1R, EG123) was obtained from D. Levin (Johns Hopkins Bloomberg School of Public Health, Baltimore, MD).
FM4-64 was purchased from Invitrogen. FLZ, nystatin, filipin, fumonisin BI, and myriocin were purchased from Sigma-Aldrich. α-Factor was synthesized by the Johns Hopkins Synthesis and Sequencing facility.
Plasmids are listed in Table I.
pEG361 (prm1∷HIS3) was constructed by inserting segments from the 5′ and 3′ UTRs of the PRM1 gene into the XbaI and SphI sites of pRS303. The 5′ UTR segment from −526 to −207 was amplified with primers having 5′ SpeI and SphI extensions, and the 3′ UTR segment from 135 to 538 was amplified with primers having XbaI and SpeI extensions. pEG387 (PTEF1-GFP-PRM1) was constructed by PCR amplifying the 2.3-kb coding sequence of PRM1 with primers having 5′ EcoRI and SalI extensions, and then inserting the PCR product into pEG311 between the EcoRI and SalI sites at the 3′ end of the GFP coding sequence. pEG427 (PGPD-HA-PRM1) was constructed by inserting PCR products encoding a 3×HA tag (Schneider et al., 1995) and a 5′ BamHI 3′ PstI-flanked PRM1 open reading frame into p415GPD (Mumberg et al., 1995). The GPD promoter was then replaced by SacI–XbaI fragments containing the CYCI, ADHI, and TEFI promoters from p415CYC, p415ADH, and p415TEF (Mumberg et al., 1995) to construct pEG454, 455, and 456. pPP1551 was digested with SmiI to direct integration of PFUS1-GFP to the 5′ UTR of FUS1. All PCR-generated plasmids were verified by DNA sequencing.
Epifluorescent light microscopy was performed at room temperature with a motorized microscope (Axioplan 2; Carl Zeiss, Inc.) outfitted with a mercury arc lamp, band pass filters (Chroma Technology Corp.), differential interference contrast optics, and a digital camera (Orca ER; Hamamatsu). Single images were collected with a 100×/1.40 Plan Apochromat objective. Image fields were selected in an unbiased manner using differential interference contrast optics. Images were collected and their contrast was optimized with Openlab software (Improvision), using identical linear adjustments for all related images.
Time-lapse images of mating yeast were collected as previously described (Nolan et al., 2006). Mating mixtures were preincubated on filters over SC agar plates for 45 min. Cells were collected from the filters into 1 ml SC medium and concentrated to 20 μl by centrifugation. A 1.6-μl aliquot was then pipetted onto a 1.5-mm-thick pad of SC medium with 3% agarose on a microscope slide. Application of an 18-mm2 coverslip caused the cell suspension to spread into an even layer. After excess agar was trimmed away, the slides were sealed with VALAP (a 1:1:1 mixture of petrolatum [Vaseline], lanolin, and paraffin) and observed during the period from 1 to 2 h after mixing. Time-lapse images were collected with a 63× Plan Apochromat objective lens. Both the objective lens and microscope stage were heated to 30°C, and binning (2×) was used to reduce exposure times and minimize photobleaching, with sets of GFP, DsRed, and differential interference contrast images collected sequentially at 15-s intervals.
Cells were fixed as previously described (Heiman and Walter, 2000) with minor modifications. In brief, cells were scraped off and fixed in 3% gluteraldehyde contained in 100 mM cacodylate, pH 7.4, with 5 mM Ca2+ for 60 min at room temperature. The cells were then washed twice with 100 mM cacodylate, once with water, and once with 3% KMnO4 (Mallinckrodt). Cells were then fixed in 3% KMnO4 for 60 min at room temperature, dehydrated through a graded series of ethanol (5′ washed with 50, 70, 80, 90, and 95% ethanol and 3 × 100% ethanol, 15 min each), and stored in a final wash of 100% ethanol overnight. Cells were then washed two times for 15 min each with propylene oxide (PO); placed into a 1:1 mixture of PO and Spurr resin; and subsequently placed under vacuum overnight. The next day, cells were transferred to 100% Spurr resin, left under vacuum for 24 h, and subsequently placed into beem capsules and allowed to polymerize at 60°C for 24–48 h. 80-nm sections were cut on an ultramicrotome (UCT; Leica), stained with lead citrate (Ted Pella, Inc.), and imaged with a transmission electron microscope (EM 410; Philips) equipped with a camera (Megaview III; Soft Imaging System). Figures were assembled in Photoshop (Adobe), with only linear adjustments in brightness and contrast.
Screening for cell fusion mutants
Strains from the quality control set of yeast deletion mutants were preferred for this screen because the MATa and MATα strains with a given mutation are arrayed in the identical position on two different sets of 96-well plates. Additional screening was performed on strains that were obtained from Invitrogen, which had to be rearrayed for bilateral mating tests. The strains were grown to saturation as a 96-well array in a 2-ml TiterBlock filled with a 3-mm glass bead and 1 ml of yeast peptone dextrose (YPD). Cells were then transferred to a fresh 96-well YPD TiterBlock using a pinning tool and grown in a shaker for 10 h at 30°C. Mating was initiated by pinning sequentially from the MATa and MATα TiterBlocks onto a nitrocellulose filter layered over a rectangular YPD plate. After incubating for 2.5 h at 30°C, the mating reaction was stopped by using a pinning tool to scrape the cells off the filter and then to mix them into 100 μl of ice-cold TAF buffer (20 mM Tris, pH 8.0, 20 mM NaN3, and 20 mM NaF) in the wells of a round-bottom 96-well plate. Mating reactions could be stored for up to 2 d in TAF at 4°C before scoring. To score for mating defects, 2 ml of cells aspirated from the loose pellet at the bottom of each well were mixed into 2 μl FM4-64 (80 μM in H2O on ice), and then loaded on a microscope slide. A coverslip was carefully layered over the FM4-64–stained cell suspension to wick the yeast into a monolayer without crushing the cells. The reaction was then visually scored for prezygote accumulation with reference to wild-type, fus1, and prm1 standards (Fig. S1). No mutant that could form mating pairs had a complete block in cell fusion. Because several previously described cell fusion mutants had a low level of prezygote accumulation even in bilateral matings, the threshold for scoring prezygote accumulation was set at ∼5% of mating pairs. False positives were isolated at a frequency approaching 2% because of this low threshold. Isolated strains that grew to high density also contributed to the high rate of false positives. Many false positives were removed from the collection of putative mutants after repeating the primary screen with cells at closer to mid-log phase before the initiation of mating. For the secondary screen, the MATa strains of the putative cell fusion mutants were transformed with pEG311 for cytoplasmic GFP expression. MATa GFP strains were then mated to the corresponding MATα strains on 2.5-cm filters, following the standard mating assay procedure described in the following paragraph. When GFP transferred between two cells of a mating pair that appeared by FM4-64 staining to be arrested as a prezygote, we inferred that the fusion pore that allowed GFP transfer was either too small or too transient to be detected as a gap in the FM4-64–stained plasma membranes.
Mating assays were performed as previously described (Jin et al., 2004). 106 each of MATa and α cells growing in log phase were mixed and then collected on 2.5-cm-diam cellulose ester filters (Millipore). The filters were placed cell side up on SC agar plates and incubated for 100 min at 30°C unless otherwise indicated. Mated cells were collected from filters into ice-cold TAF buffer. The cells were concentrated by centrifugation for 5 s, resuspended in 20−30 μl TAF buffer, and analyzed by epifluorescent microscopy. At least 200 mating pairs were scored for all quantitative assays.
The standard mating conditions had to be adjusted to test for suppression of the erg6 mating defect because Ste-CTM was expressed from a galactose-regulated promoter. Each pair of erg6 strains was transformed with two plasmids containing different selectable markers: PGPD-PRM1 LEU2 or an empty vector control and PGAL1-STE5-CTM HIS3 or an empty vector control. The strains were grown to log phase in selective raffinose medium and then mated for 3 h at 30°C on galactose plates.
HA-Prm1 expression level comparisons
Yeast strains expressing the four HA-PRM1 constructs were cultured to log phase in SC-leucine medium. Protein extracts were prepared by glass bead lysis from one OD600 unit of cells. Four 1:2 serial dilutions were prepared from each extract by dilution with an extract from an sso1Δ strain. Samples were resolved on a 10% SDS-PAGE gel, and a Western blot was cut into molecular weight-range strips that were separately probed with the 12CA5 anti-HA monoclonal antibody (Covance) and with an anti-Sso1 polyclonal antibody (Grote and Novick, 1999) as a loading control. The blot was developed by chemiluminescence with exposure times ranging from 5 s to 5 min. The films were digitized on a flatbed scanner, and band intensities were measured using Image software (National Institutes of Health).
MATa cells expressing GFP-PRM1 from the TEF1 promoter (pEG387) were mated for 1.5 h to MATα fus1 fus2 RFP cells to accumulate early prezygotes. To deplete ergosterol, the MATaGFP-PRM1 cells were preincubated in YPD medium supplemented with 0.5 mg/ml FLZ for 3 h at 30°C and then mated to untreated MATα fus1 fus2 RFP cells on an SC + FLZ plate.
Pheromone response assays
Cells expressing PFUS1-FUS1(1–254)-lacZ from pSB234 were grown to log phase in SC-uracil medium. The cells were pelleted and resuspended at OD600 0.5 in medium supplemented with 6 μM α-factor and incubated for 90 min at 30°C unless otherwise indicated. For β-galactosidase assays, 0.4 OD600 units of cells were collected by centrifugation, resuspended in 100 μl Z buffer (82 mM NaPO4, pH 7.0, 10 mM KCl, 1 mM MgSO4, and 40 mM β-mercaptoethanol), and permeabilized by three rounds of freezing in liquid N2 and thawing in a 37°C waterbath. Reactions were started by mixing 5–30 μl of the homogenate into 150 μl o-nitrophenyl-β-D-galacotpyranoside (1 mg/ml in Z buffer), incubated at 37°C for 10–90 min, stopped by the addition of 50 μl of 1 M Na2CO3, and read at OD410 in a 96-well plate reader (PerkinElmer).
To assay pheromone-induced GFP expression, PFUS1-GFP–transformed cells were grown overnight in SC-uracil medium, treated in YPD medium with myriocin and/or FLZ as indicated, induced with 6 μm α-factor for 90 min at 30°C, and then washed with ice-cold TAF buffer. The GFP fluorescence of 20,000 cells was quantified in the FL1 channel of a FACSCalibur flow cytometer (BD Biosciences).
Cells transformed with pSKM21 were grown to log phase in SC-uracil medium. GFP-Ste5 expression was induced with 0.5 mM CuSO4 for 2 h at 30°C. 2 × 106 cells were collected by centrifugation, resuspended in 400 μl SC-uracil/CuSO4 + 6 μM α-factor, and incubated for an additional 30 min at 30°C. After α-factor treatment, the cells were again collected by centrifugation, resuspended in 10 μl SC-uracil/CuSO4 + 30 μM α-factor, and immediately imaged. Live cells without buds were scored for GFP-Ste5 polarization. Varying degrees of polarization were observed, and strong polarization was found in only a small percentage of the cells. Thus, any cell with a detectable concentration of fluorescence associated with an arc spanning <45° on the cell surface was scored as positive for GFP-Ste5 polarization. The results are presented as mean ± 95% confidence intervals for four independent experiments, with n > 150 for each mutant in each experiment.
The filipin staining procedure was based on a method developed for the study of Schizosaccharomyces pombe cytokinesis (Takeda and Chang, 2005). Filipin was added to live cells at a final concentration of 2.5 mg/ml in 0.5% DMSO. Cells were then concentrated by a brief centrifugation and imaged live within 1–5 min after filipin addition. The tips of Saccharomyces cerevisiae mating projections had somewhat brighter filipin staining than the growing end of mitotic S. pombe cells. Under these conditions, filipin did not compromise the viability of wild-type S. cerevisiae. This technique is therefore superior to previous methods for staining S. cerevisiae with filipin, which are prone to toxicity and artifacts (Valdez-Taubas and Pelham, 2003). Imaging filipin-stained cells was challenging because filipin is rapidly bleached by UV excitation and its staining pattern became more speckled over time. To facilitate direct quantitative comparisons of filipin intensity and polarity, populations of wild-type and mutant cells marked by expression of either cytoplasmic GFP or Sso2-GFP were mixed before pheromone induction, staining, and imaging. For each mutant, at least 400 shmoos were scored blindly for filipin polarization and then categorized as wild-type or mutant.
Cells expressing 2×PHPLCγ-GFP were induced with 6 μM α-factor for 90 min. For quantification, multiple fields of wild-type or erg6 cells were scored blindly for mating projections with polarized fluorescence.
Cells in log-phase growth were pelleted, resuspended at low density (OD600 = 0.05) in appropriate growth medium, divided into 1-ml aliquots, and then grown in a shaking incubator at 30°C before α-factor treatment or at 25°C before mating (because lcb1ts cells failed to form mating pairs at 30°C). 1 mg/ml FLZ was added to individual aliquots at the indicated times. Despite a significant amount of lysis leading to a slower apparent growth rate, the lcb1ts mutation does not significantly alter the rate of ergosterol depletion in FLZ-treated cells. In a dose–response assay, 10 μg/ml FLZ was sufficient to maximally inhibit growth in both LCB1 control and lcb1ts mutant strains. In addition, FLZ treatment led to a 50–60% reduction in cellular ergosterol levels in both lcb1ts mutant and control strains after 3 h at 30°C or 5 h at 25°C (Fig. S2).
Online supplemental material
Fig. S1 presents an overview of the genetic screening procedure that led to the identification of erg6 and a plasma membrane fusion mutant and examples of mating pairs arrested at various stages of the cell fusion pathway. Fig. S2 presents critical controls related to the use of FLZ to deplete ergosterol, including growth curves and sterol analysis of wild-type and lcb1ts mutant yeast. Fig. S3 presents results quantifying the relative activity in FLZ-treated cells of five plasmids that activate FUS1 expression at distinct stages of the pheromone-response signal transduction pathway. Fig. S4 presents flow cytometry data for ste5ts cells illustrating the gradual reduction in FUS1 expression at elevated temperatures. Fig. S5 presents filipin-staining results demonstrating that bright filipin staining of sphingolipid-depleted plasma membranes is a direct consequence of alterations in the lipid composition of the membrane and that acylation, but not head-group, modification is required for sphingolipids to compete for ergosterol binding.
Thanks to John Burg and Peter Espenshade for sterol analysis and to Qing Huang for strain construction. Flow cytometry was performed in the laboratory of Dianne Griffin. Thanks to Peter Pryciak, Leonid Chernomordik, Elaine Elion, Charles Martin, Fred Chang, and members of the Grote Laboratory for advice and discussion, to Christopher Stefan for communicating results before publication, and to Peter Pryciak, David Levin, Michael Edidin, and Elizabeth Chen for comments on the manuscript.
This work was supported by a Research Scholar Award from the American Cancer Society.
Abbreviations used in this paper: FLZ, fluconazole; FRET, fluorescence resonance energy transfer; PI(4,5)P2, phosphatidylinositol-4-phosphate; PO, propylene oxide; SC, synthetic complete; YPD, yeast peptone dextrose.