A role for yeast oxysterol-binding protein homologs in endocytosis and in the maintenance of intracellular sterol-lipid distribution

Sterol lipids, such as cholesterol and the fungal sterol ergosterol, are essential for plasma membrane integrity and structure. Changes in sterol levels and composition have extensive effects on membrane function. Ergosterol is important for yeast membrane structure, but it also influences membrane dynamics and the activities of membrane proteins (Gaber et al., 1989; Welihinda et al., 1994). Perturbations in cellular ergosterol and sterol lipid levels affect yeast cell fusion during mating (Gaber et al., 1989; Tomeo et al., 1992), vacuolar homotypic membrane fusion (Kato and Wickner, 2001), and secretory transport to the cell surface (Hardwick and Pelham, 1994; Bagnat et al., 2000). Yeast endocytosis is particularly sensitive to sterol composition (Munn et al., 1999).

Cholesterol and yeast ergosterol are both primarily localized in the plasma membrane. To maintain constant levels, these sterols are specifically transported to the plasma membrane from their site of synthesis in the endoplasmic reticulum (ER). Once in the plasma membrane, cholesterol and ergosterol are recycled back to the plasma membrane if internalized into the cell after endocytosis. In animals, cholesterol is delivered to the plasma membrane by at least two different transport pathways. Cholesterol transport can follow either the established pathway for glycoprotein secretion, or an undefined nonvesicular transport pathway (Kaplan et al., 1985; Urbani et al., 1990; Heino et al., 2000). Pathways for recycling internalized sterols back to the plasma membrane after endocytosis are poorly understood.

Oxysterol-binding proteins (OSBPs) constitute a conserved family of lipid-binding proteins that are implicated in the maintenance of sterol and sphingolipid composition in membranes (Dawson et al., 1989; Jiang et al., 1994; Ridgway et al., 1998; Storey et al., 1998; Beh et al., 2001). The canonical mammalian OSBP was identified based on its high affinity for oxysterols, which are oxygenated derivatives of sterols (Taylor et al., 1984). Oxysterols, such as 25-hydroxycholesterol, are more potent than cholesterol itself as feedback regulators of sterol synthesis, suggesting that they are the sterol metabolites whose levels control the de novo synthesis of sterols (Kandutsch et al., 1978). These findings implied that OSBPs mediate the effects of oxysterols, and OSBPs were once considered candidates for regulators of sterol synthesis feedback control (Taylor et al., 1984). However, other proteins are now known to mediate feedback regulation (Brown and Goldstein, 1997; Brown and Goldstein, 1999) and recent data point to other sterol-related regulatory roles for OSBP and its homologs.

The exact function of the OSBP protein family is unknown, but genetic analysis of the seven yeast OSBP orthologs (the OSH genes) affirmed that OSBPs are involved in some aspect of sterol regulation (Jiang et al., 1994; Beh et al., 2001). Although none of the OSH genes encodes an essential product, the elimination of the entire OSH gene family is lethal (Beh et al., 2001). When all seven Osh proteins are depleted from yeast, a drastic change in cellular sterol composition results and ergosterol levels increase by 3.5-fold (Beh et al., 2001). These findings indicate that the OSH gene family shares at least one essential overlapping function, potentially linked to sterol regulation. As part of the genetic analysis of the OSH genes, all 127 possible combinations of OSH deletion alleles were constructed and 125 of these combinations were in fact viable (Beh et al., 2001). Many of these viable OSH deletion combinations were characterized by specific sterol and membrane defects. These results underscore the regulatory link between Osh proteins and sterol lipids.

The effects of Osh proteins are not restricted to sterols but also include phospholipids. Several Osh proteins have been demonstrated to bind directly to phosphoinositide lipids (Levine and Munro, 1998; Li et al., 2002). Many of these Osh proteins also bind Scs2p, a yeast VAP (VAMP-associated protein) homolog that appears to recruit a variety of known phospholipid regulators to the ER membrane (Loewen et al., 2003). This is a conserved interaction because mammalian OSBP and VAP also physically interact (Wyles et al., 2002). Another regulator of lipid metabolism, encoded by SEC14, genetically interacts with OSH4. SEC14 encodes a phospholipid transfer protein, which creates a membrane lipid composition that promotes Golgi-derived secretory transport (Bankaitis et al., 1990). The deletion of OSH4/KES1, but none of the other OSHs, bypasses the essential requirement for SEC14 (Fang et al., 1996; Beh et al., 2001). This indicates that this SEC14 interaction is specific to OSH4, and that OSH4 is a negative regulator of SEC14-dependent secretion. The implication of these studies is that yeast Osh proteins are recruited to specific compartments together with other lipid regulators to adjust local lipid composition and, thereby, to influence membrane dynamics. Specifically how this mechanism might affect normal cellular functions is not yet understood.

In this paper, we investigated the role of the OSH gene family in cell growth to determine cellular processes that are dependent on all OSH genes. Our data indicated that the OSHs are required for regulating cellular sterol-lipid distribution, and for maintaining plasma membrane and vacuolar morphology. Daughter cell budding was defective in yeast depleted of Osh proteins, and lipid droplets and sterol lipids accumulated within cells. We found that endocytosis was disrupted in OSH mutants and also in other previously described mutants that perturb ergosterol distribution within cells. These findings established that, collectively, the yeast OSH gene family facilitates bud formation, endocytosis and affects intracellular sterol distribution.

Culture media and genetic manipulations were as described (Adams et al., 1997). To select for the kan-MX4 gene, yeast were grown on yeast rich medium (YPD) containing 200 μg ml^-1 geneticin sulfate (G418) (Gibco BRL Life Technologies, Rockville, MD) (Wach et al., 1994).

Yeast strains used in this study are listed in Table 1. The seven OSH genes, OSH1-OSH7, corresponded to the open reading frames YAR042w, YDL019c, YHR073w, YPL145c, YOR237w, YKR003w and YHR001w, respectively. OSH1 is a fusion of the open reading frames previously designated YAR042w and YAR044w in the completed yeast genome sequence (Schmalix et al., 1994; Goffeau et al., 1996; Beh et al., 2001; Levine et al., 1998; Levine et al., 2001).

DNA cloning techniques and bacterial transformations were performed by standard procedures (Sambrook et al., 1989). Restriction enzymes were obtained from New England Biolabs (Beverly, MA). Oligonucleotide primers for polymerase chain reaction (PCR) were purchased from Operon Technologies (Alameda, CA).

To disrupt the VPS4 gene in SEY6210, a fragment containing vps4Δ::kan-MX4 with flanking sequences amplified by PCR was used for one-step gene replacement (Adams et al., 1997). The vps4Δ disruption fragment was amplified from genomic DNA derived from a S288C vps4Δ::kan-MX4 strain (Research Genetics, Huntsville, AL) (Winzeler et al., 1999) and used as a template with the oligonucleotide primer combination: 5′-GTCCGATTCACATGTCGCCACTCC-3′ and 5′-CAAGCCGAATTGAACCAAATGAGCGC-3′.

To subclone the wild-type OSH4/KES1 gene, a 2.4 kb SacI-ClaI OSH4/KES1 genomic fragment was inserted into the complementary sites of pRS316, a URA3-marked ARS/CEN plasmid (Sikorski and Hieter, 1989) thereby generating pCB231.

To generate conditional mutant alleles of OSH4/KES1, pCB231 was mutagenized in vitro with hydroxylamine as described previously (Adams et al., 1997), and transformed into JRY6326, a strain whose growth on methionine-containing medium requires OSH function (Beh et al., 2001). From independent pools of mutagenized plasmids transformed into yeast, two transformants were isolated that rescued JRY6326 methionine-sensitivity at 23°C, but not 37°C. The fragments containing these two temperature-sensitive alleles, osh4-1 and osh4-2, and the wild-type OSH4/KES1 gene, were each subcloned into pRS314, a TRP1-marked ARS/CEN plasmid (Sikorski and Hieter, 1989), generating the plasmids pCB255, pCB256 and pCB254 respectively. At 23°C, these plasmids conferred growth to yeast devoid of all OSH genes but at 37°C, only transformants containing the plasmid with wild-type OSH4/KES1 grew.

To examine sterol-lipid distribution using filipin, 1 ml 37.5% formaldehyde was added to 9 ml of cell culture grown in synthetic medium to a density of 0.7 OD₆₀₀ units ml^-1 (early midlog). After 10 minutes of constant mixing at 23°C, the fixed cells were centrifuged and the pellet was washed twice with 10 ml distilled water. The washed cells were resuspended in 1 ml of water from which 0.2 ml was mixed with 4 μl of freshly made 5 mg ml^-1 filipin complex (Sigma Chemical, St Louis, MO) in ethanol (on ice). After incubating in the dark with filipin for 15 minutes at 23°C, cells were spotted directly on slides and filipin fluorescence was observed with a UV filter set using neutral density filters.

For all fluorescence microscopy, samples were mounted on polylysine coated slides, sealed under coverslips with nail polish, and imaged on an Eclipse E600 microscope (Nikon, Tokyo, Japan) equipped with a CCD digital camera (Hamamatsu Phototonics, Hamamatsu-City, Japan).

Fluid-phase endocytosis was observed in 2.5 OD₆₀₀ units of cells harvested from cultures grown in synthetic medium at 23°C to a density of 0.25 OD₆₀₀ units ml^-1. Cells were resuspended in 200 μl of synthetic medium and split into two samples. One sample was incubated for 60 minutes at 37°C, the other for 60 minutes at 23°C, prior to the addition of 50 μl 40 mg ml^-1 lucifer yellow CH (Sigma Chemicals). After lucifer yellow addition, the samples were incubated at the same temperatures for an additional 120 minutes. The cells were then centrifuged and pellets were washed three times with 200 μl ice-cold 50 mM succinic acid, 20 mM NaN₃ pH 5.0 and resuspended in 30 μl of the same buffer before viewing by fluorescence microscopy.

Ste6p stability was analyzed as described previously (Wendland et al., 1999). Yeast transformed with a plasmid expressing HA-epitope-tagged Ste6p, pSM672 (Berkower et al., 1996), were grown to early log phase, harvested, and then resuspended to a density of 12.5 OD₆₀₀ units ml^-1 in 2.5 ml pre-warmed 37°C synthetic complete medium lacking uracil. The cultures were incubated for 1 hour at 37°C before the addition of 12.5 μl of 1 mg ml^-1 cycloheximide. At the times indicated, 0.5 ml of each culture was removed and added to an equal volume of ice-cold 20 mM KF, 20 mM NaN₃, 50 μg ml^-1 BSA. After centrifugation, each pellet was resuspended in 1 ml ice-cold distilled water and 150 μl of a 2 N NaOH, 80 μl ml^-1 2-mercaptoethanol solution and incubated on ice for 20 minutes. 150 μl 50% trichloroacetic acid was then added and the mixture was incubated on ice for an additional 30 minutes before centrifugation. After complete removal of supernatants, pellets were resuspended in 20 μl 0.5 M Tris-HCl pH 7.5 and 20 μl 2X sodium dodecyl sulfate (SDS) sample buffer, and incubated at 37°C for 30 minutes before loading on a 6% SDS polyacrylamide gel. After electrophoresis, resolved proteins were transferred to nitrocellulose (N-Hybond, Amersham, Arlington Heights, IL) and Ste6-HAp was detected using a mouse anti-HA antiserum (1:1000 dilution; Boehringer Mannheim, Indianapolis, IN) and a horseradish peroxidase (HRP)-conjugated goat anti-mouse secondary antiserum (1:3000 dilution). Ste6-HAp bands were visualized by chemiluminescence (ECL, Amersham, Arlington Heights, IL).

FM4-64 labeling was performed as previously described, but with minor modifications (Vida and Emr, 1995). FM4-64 internalization was examined in 5.0 OD₆₀₀ units of cells harvested from cultures grown in synthetic medium at 23°C, resuspended in 500 μl of synthetic medium, and cultured for 30 minutes at 37°C. Cells were then treated with 32 μM FM4-64 for 30 seconds and immediately diluted with 10 ml of synthetic medium prewarmed at 37°C. After centrifugation, and a wash with warmed medium, cell pellets were resuspended in 0.5 ml of medium and incubated for an additional 30 minutes at 37°C before viewing by fluorescence microscopy. The use of FM4-64 to examine vacuolar morphology was performed as previously described (Kato and Wickner, 2001).

Pulse-chase analyses of carboxypeptidase Y (CPY), Hsp150p and Gas1p transport were performed on 20 OD₆₀₀ units of cells harvested from mid-log cultures grown in synthetic medium. For metabolic labeling, the cells were resuspended in 2.5 ml of synthetic medium prewarmed to 37°C and incubated at 37°C for 60 minutes before the addition of 250 μCi of Tran[³⁵S]-label (ICN Radiochemicals). After a 10 minute incubation, 25 μl of chase solution (200 mM (NH₄)₂SO₄, 2 mg ml^-1 methionine, 2 mg ml^-1 cysteine) was added and incubated at 37°C (for CPY maturation, a 5 minute incubation with Tran[³⁵S]-label was followed by the addition of chase solution). For each time point, 0.5 ml of culture was removed and transferred to a microfuge tube on ice containing 50 μl of stop solution (100 mM NaN₃, 1 mg ml^-1 cycloheximide). Cells and medium were separated by centrifugation.

For the analysis of Hsp150p secretion, radiolabeled glycoproteins secreted into the medium were precipitated using Concanavalin-A (Con-A) Sepharose as described (Holthuis et al., 1998). Cells were disrupted with glass beads and intracellular glycoproteins were precipitated from the cleared lysate after binding to Con-A Sepharose. Secreted glycoproteins, of which Hsp150p is the most abundant (Russo et al., 1992; Gaynor and Emr, 1997), were visualized by autoradiography following SDS-PAGE and fluorography.

CPY and Gas1p were immunoprecipitated from the cleared lysate of cell extracts using specific antibodies (both antibodies were gifts from the Schekman lab, University of California, Berkeley, CA) and protein A-Sepharose as previously described (Rose et al., 1989). Radiolabeled immunoprecipitates were analyzed by SDS polyacrylamide electrophoresis and visualized following fluorography.

Extracellular secretion of CPY from yeast grown on solid synthetic medium was analyzed as previously described (Roberts et al., 1991) with a few modifications. Cells were streaked in patches onto solid synthetic medium that lacks methionine, grown overnight at 30°C, and then transferred by replica printing onto solid media containing varying methionine concentrations. The methionine concentrations used repressed expression of OSH2 in the oshΔ P^MET3-OSH2 strain (JRY6326) to varying degrees as manifested in growth-rate differences. After another night of growth at 30°C, nitrocellulose filters were placed on top of the patched cells in direct contact with the solid medium, and incubated for 12 hours at 30°C. Once removed from the solid medium, the nitrocellulose filters were washed several times with Buffer A (20 mM Tris-HCl pH 7.6, 150 mM NaCl, 2% dry milk) to dislodge bound cells. The filters were probed with the rabbit polyclonal anti-CPY antiserum in Buffer A (1:2000 dilution), washed with Buffer A, and then incubated with a HRP-conjugated goat anti-rabbit secondary antiserum (1:5000 dilution; Amersham) in Buffer A.

Cells were prepared for electron microscopy as described (Wright, 2000). Cells were fixed overnight at 4°C in 0.5% glutaraldehyde followed by a 45 minute postfixation at room temperature with 2% KMnO₄ (low mercury; Aldrich Chemical, Milwaukee, WI). Samples were en bloc stained with 1% uranyl acetate, dehydrated with absolute ethanol, infiltrated and embedded in Spurr's resin (Electron Microscopy Sciences, Ft Washington, PA). Thin sections were stained with uranyl acetate and Reynold's lead citrate. Sections were viewed at 100 kV on a FEI Tecnai 12 transmission electron microscope.

Deletion of all seven yeast OSH genes causes lethality. This inviability can be rescued by expression of any of the individual OSH genes indicating that the OSH genes share at least one essential overlapping function (Beh et al., 2001). To define the essential functions mediated by the OSH family, we investigated the cellular defects associated with the elimination of all OSH function. We created two independent strains in which six OSH genes were deleted and the seventh was replaced with a construct permitting exogenous control of its expression. In one strain, OSH function was gradually depleted by repressing the synthesis of the sole remaining OSH gene, Osh2. In the other strain, the single remaining OSH gene, Osh4, contained a temperature-sensitive mutation and could be rapidly inactivated when cultured at higher temperatures. The effects of inactivating all the OSHs were then observed in both strains.

The first strain examined lacked six of the seven OSH genes, and the remaining OSH2 gene was regulated by the MET3 promoter (JRY6326) (Beh et al., 2001). In the presence of added methionine, OSH2 expression was repressed and the oshΔ P^MET3-OSH2 strain ceased to grow (oshΔ refers to the deletion of all OSH genes other than those indicated). Growth arrest was gradual, taking about four culture doublings. The second yeast strain analyzed also lacked all chromosomal copies of the OSH genes but was viable because it carried a plasmid with one of two KES1/OSH4 temperature-sensitive mutations, osh4-1(ts) and osh4-2(ts). At 37°C, the mutant proteins encoded by these OSH4 alleles were inactivated and the oshΔ osh4(ts) strains stopped growing. After the shift to the restrictive temperature, growth was retarded within 15 minutes and completely arrested within an hour. After culturing oshΔ osh4(ts) cells under these conditions, their inviability could be reversed by returning the cells to the lower permissive temperature for growth. When oshΔ osh4(ts) cells were incubated at 37°C for 1 hour and then micromanipulated onto solid medium and grown at 23°C, 88% (14 out of 16) recovered and formed colonies (92% of wild-type and 96% oshΔ OSH4 cells recovered after the same treatment). Using these strain constructions we monitored specific changes in cellular morphology that resulted from rapid inactivation of OSH4, or gradual depletion of Osh2p, in cells that otherwise lacked all OSH family genes.

After OSH2 expression was repressed, the oshΔ P^MET3-OSH2 cells exhibited drastic morphological changes. The effect of eliminating all OSH genes was analyzed by transmission electron microscopy using fixation methods designed to accentuate membranous structures (Fig. 1). We observed defects consistent with abnormal cell wall deposition (Fig. 1B). Along the perimeter of the cell surface the thickness of the cell wall varied, whereas in wild-type cells it was uniform. The abnormal depositions seen by electron microscopy were probably irregular deposits of chitin (other than bud scars). In Osh-depleted cells stained with calcofluor and examined by fluorescence microscopy, chitin was often mislocalized at the bud tip in budding daughter cells and randomly dispersed over the cell surface (Fig. 2). Defects in cell size and budding were also apparent. Many of the arrested cells were significantly larger than wild type. Septation of dividing cells was often incomplete, and cells with multiple buds were observed with buds oriented randomly with respect to one another, indicating a budding pattern defect. The gradual depletion of Osh proteins also caused observable changes within the cell. Many cells accumulated vesicles and the number of cytoplasmic lipid droplets increased (Fig. 1B). In nearly all cells, significant vacuolar fragmentation and collapse was observed, and within these vacuolar remnants numerous small lipid droplets were evident. The defects associated with the gradual depletion of Osh proteins indicated that the OSHs were important for processes both at the cell surface and within the cell itself.

Many, but not all, of the defects observed after gradual Osh depletion were seen when OSH function was inactivated more rapidly. One hour after the shift to the restrictive temperature of growth the oshΔ osh4(ts) cells were still wild type in size, and no cell wall abnormalities were apparent (Fig. 1C,D). Within these cells, however, cytoplasmic lipid droplets accumulated and extensive vacuolar fragmentation was seen. As previously observed after the gradual depletion of Osh2p, vacuolar remnants were often filled with small lipid droplets in oshΔ osh4(ts) cells. Thus, regardless of the strain, vacuolar morphology and lipid droplet distribution were disrupted in cells depleted of OSH function. At the permissive temperature, oshΔ osh4(ts) strains grew more slowly than wild type and exhibited many of the same defects (although less severe) as those observed after growth arrest at 37°C. Thus, regardless of how OSH function was eliminated, loss of OSH function resulted in vacuolar fragmentation and lipid droplet accumulation.

The elimination of all OSH function caused aberrant vacuolar morphology, and in some cases vesicle accumulation, suggesting that secretory transport to the vacuole or to the cell surface might be disrupted. Moreover, several reports suggest that OSBP homologs regulate vesicular trafficking through the Golgi (Fang et al., 1996; Xu et al., 2001). To test whether OSH genes affected vesicular trafficking of proteins from the ER to the vacuole, the maturation of CPY precursors in OSH mutants was analyzed. In wild-type cells, the p1 precursor form of CPY is generated when newly synthesized CPY is translocated into the ER and then core glycosylated. After transport from the ER to the Golgi, CPY is further glycosylated to the larger p2 precursor form. Finally, upon arrival in the vacuole, the smaller mature (m) CPY protein is produced through proteolytic removal of the pro-CPY leader peptide. To inactivate OSH function before pulse-chase labeling, the oshΔ osh4-1(ts) strain (and the other control strains examined) were incubated for 60 minutes at 37°C. After pulse-chase labeling, CPY was immunoprecipitated from extracts of wild-type and the OSH mutant strains. CPY processing showed similar kinetics in all the strains tested (Fig. 3A) indicating that CPY maturation was similar to wild-type in oshΔ osh4-1 and oshΔ OSH4 mutant strains. The only apparent difference was a slight delay in the conversion of the p1 precursor into the p2 form in the oshΔ osh4-1 strain, when compared with the oshΔ OSH4 or wild-type strains. In fact, a similar subtle lag in CPY processing has been reported in other mutants that affect sterols, including viable ERG mutants (Munn et al., 1999). These results indicated that the OSH genes do not significantly affect CPY transport from the ER to the Golgi, or from the Golgi to the vacuole.

Some secretory mutants do not affect CPY delivery to the vacuole, but rather secrete a large proportion of CPY out of the cell. In these mutants, the p2 form of CPY is mis-sorted to the plasma membrane and some is released into the medium (reviewed by Bryant and Stevens, 1998). To evaluate whether any CPY was secreted from the cell after OSH gene inactivation, we took advantage of the gradual growth arrest caused when OSH2 expression was repressed in the oshΔ P^MET3-OSH2 strain. Patches of oshΔ P^MET3-OSH2 cells were transferred onto solid medium containing a range of methionine concentrations. In the presence of added methionine, expression of the integrated P^MET3-OSH2 construct was repressed and growth of this strain slowed, especially at higher methionine concentrations. Extracellular secretion of CPY was monitored by overlaying nitrocellulose filters directly onto patches of yeast strains grown on solid media (Fig. 3B). Using anti-CPY antibodies, the filters were probed for the presence of secreted CPY. As a control, duplicate filters were probed with antiserum that recognizes actin. Although polymerized actin comprises the basic structure of the cytoskeleton, lower levels of monomeric actin are present in the cell (Karpova et al., 1998) and can be nonspecifically released by cell lysis with other cytoplasmic proteins. The nitrocellulose immunoblots were taken from comparable patches of cells grown on solid medium (bottom panel of Fig. 3B,). As a control, CPY extracellular secretion was also monitored in a vps4Δ strain. VPS4 is required for the sorting and trafficking of CPY from the trans-Golgi to the vacuole (Robinson et al., 1988; Rothman et al., 1989). The vps4Δ strain secreted a large amount of CPY onto nitrocellulose, but relative to wild type the vps4Δ mutant did not release actin indicating specific CPY release (Fig. 3B). In comparison, the oshΔ P^MET3-OSH2 strain did not release actin or CPY, regardless of the extent of OSH2 repression. Based on these experiments, reducing OSH function had no adverse effect on CPY sorting at the trans-Golgi.

The exocytosis of several other secreted proteins were also unaffected by OSH inactivation. To determine whether secretion to the plasma membrane was affected by OSH mutants, transport of the glycoprotein Hsp150p and the glycosyl-phosphatidylinositol (GPI)-anchored protein Gas1p were examined following pulse-chase radiolabeling (Fig. 4). Hsp150p is the most abundant glycoprotein secreted into the medium (Russo et al., 1992; Gaynor and Emr, 1997), and Gas1p is transported to the plasma membrane where it is associated with membrane regions rich in ergosterol (Bagnat et al., 2000). To examine Hsp150p secretion, glycoproteins were purified from the extracellular medium and from solubilized cell extracts after a 60 minute incubation at 37°C, and pulse/chase labeled as described in the Materials and Methods. In wild-type, oshΔ OSH4 and oshΔ osh4-1 strains, Hsp150p was secreted into the medium with equal kinetics and there was no detectable intracellular accumulation (Fig. 4). By contrast, in a sec1-1 mutant, which is defective for the transport of secretory proteins from the Golgi to the plasma membrane, Hsp150p was not secreted into the medium but remained within the cell (Fig. 4).

Gas1p transport to the plasma membrane was also unaffected by OSH inactivation. After a 60 minute incubation at 37°C and a 10 minute pulse/chase labeling, the conversion of the 105 kDa Gas1p precursor to the plasma membrane-associated 125 kDa mature form was similar in wild-type, oshΔ OSH4 and oshΔ osh4-1 strains (Fig. 5). In contrast, mature Gas1p was not observed in the sec18(ts) mutant, which is defective in exocytosis at 37°C. These results indicate that Gas1p transport to the plasma membrane was not defective in the OSH mutants. In similar pulse/chase experiments, invertase secretion to the plasma membrane was also unaffected in oshΔ osh4-1 cells (C.T.B., unpublished data).

We tested whether endocytosis, another major trafficking pathway, was affected by the elimination of OSH gene function. As demonstrated by several independent assays, Osh depletion had a significant effect on endocytosis from the cell surface into the cell. In the first endocytosis assay, the internalization of a marker dye for fluid-phase endocytosis, lucifer yellow, was measured by fluorescence microscopy (Fig. 6). In almost all wild-type cells grown at either 23°C or 37°C, lucifer yellow was efficiently internalized from the medium and accumulated in the vacuole. By contrast, lucifer yellow uptake was blocked in oshΔ osh4(ts) cells after a 60 minute incubation at the restrictive temperature (Fig. 6). In the majority of oshΔ osh4(ts) cells, internal lucifer yellow staining was reduced and only in a few cells was the fluorescence in vacuolar fragments equal in intensity to wild-type vacuoles. The percentage of faintly stained Osh-depleted cells (∼85%) was similar to the well-characterized endocytosis mutant, las17Δ (87%), indicating a comparable defect. LAS17/BEE1 encodes the Saccharomyces cerevisiae homolog of the mammalian Wiskott-Aldrich syndrome protein, which is required for cortical actin assembly and endocytosis (Li, 1997; Naqvi et al., 1998). Although some endocytosis genes are essential, LAS17 is only required for growth at high temperatures. Lesser defects in lucifer yellow uptake were detected in both oshΔ OSH4 cells grown at 23°C or 37°C, and also in osh1Δ osh4(ts) cells grown at the permissive temperature of 23°C. These findings indicated that OSH genes are required for fluid-phase endocytosis.

To determine whether Osh depletion affected other aspects of endocytosis, we analyzed the endocytic internalization and subsequent proteolysis of Ste6p, a plasma membrane protein required for export of a-factor mating pheromone (Kuchler et al., 1989). In wild-type cells, Ste6p is constitutively internalized from the plasma membrane and routed to the vacuole where it is degraded (Berkower et al., 1994). The stability of Ste6p is markedly increased in endocytosis mutants, which effectively block the delivery of Ste6p to vacuolar proteases. The stability of epitope-tagged Ste6p-HA was analyzed in the oshΔ osh4-1 temperature-sensitive mutant, the las17Δ endocytosis mutant, and their respective wild-type control strains (Fig. 7A). After 60 minutes at 37°C, nearly all Ste6p-HA was degraded in the wild-type strains, but significant levels of Ste6p-HA persisted in both the oshΔ osh4-1 and las17Δ mutants (Fig. 7A). Even after 90 minutes, Ste6p-HA was detectable in the oshΔ osh4-1 and las17Δ samples. These findings indicated a comparable disruption of Ste6p-HA internalization in oshΔ osh4-1 and las17Δ mutants. Moreover, the defect in Ste6p degradation was not due to a failure in vacuolar proteolysis because the proteolytic processing of CPY was unaffected in oshΔ osh4-1 cells (see above).

OSH mutant cells were also defective in the endocytic internalization of the fluorescent lipophilic dye, FM4-64. FM4-64 associates with the yeast plasma membrane and, following endocytosis, it is delivered to the vacuole where it stains the vacuolar membrane (Vida and Emr, 1995). In wild-type cells, FM4-64 was internalized and concentrated in the vacuole (Fig. 7B). In oshΔ OSH4 and oshΔ osh4-1 cells, FM4-64 fluorescence was less intense indicating a reduction in dye internalization. Despite comparable defects in lucifer yellow uptake, the FM4-64 endocytosis defect in las17Δ mutants was more pronounced than in oshΔ osh4-1 cells (see Fig. 10B for comparison) suggesting that LAS17 and the OSH genes affect endocytosis in a different manner. Taken together, the results of all endocytosis assays established that an important function of the OSHs is related to endocytosis.

In Osh-depleted cells, ergosterol levels increase 3.5-fold relative to wild type (Beh et al., 2001). This result suggests that ergosterol levels might be elevated at the plasma membrane, the normal cellular destination of unesterified ergosterol. However, many viable OSH deletion mutants are resistant to the ergosterol-binding antibiotic nystatin (Jiang et al., 1994; Beh et al., 2001) even though many nystatin-resistant strains have reduced levels of ergosterol exposed on the cell surface (Woods, 1971; Walker-Caprioglio et al., 1989). To resolve this apparent paradox, we analyzed in vivo the intracellular distribution of sterol lipids in OSH mutant cells.

In mammalian cells, the distribution of free cholesterol can be visualized by fluorescence microscopy using the cholesterol probe, filipin (Severs, 1997). Although filipin binds cholesterol with greater affinity than ergosterol (Brittman et al., 1974), filipin fluorescence was successfully adapted for analyzing yeast sterol-lipid distribution. Filipin/sterol fluorescence was observed primarily at the cell surface (Fig. 8A), where the bulk of free ergosterol is located in yeast. Fluorescence was also observed in small cytoplasmic spots, which might correspond to ergosterol-rich endocytic vesicles. Limited overlap between the filipin fluorescent spots and Nile Red stained lipid droplets was observed indicating that some, but not all filipin-fluorescent spots, corresponded to lipid droplets (C.T.B., unpublished data).

To evaluate the specificity of filipin/sterol staining in yeast, filipin fluorescence was examined in cells after sterol synthesis was blocked. ERG9 is an essential gene that encodes squalene synthase, the first enzyme in the ergosterol biosynthetic pathway that is specific for sterol-lipid synthesis. A strain was created in which the wild-type ERG9 gene was replaced with an integrated P^MET3-ERG9 construct. In the presence of methionine, expression of squalene synthase was repressed and the P^MET3-ERG9 erg9Δ cells did not grow. Compared with filipin-stained wild-type cells, filipin fluorescence was diminished in growth-arrested P^MET3-ERG9 erg9Δ cells (Fig. 8A). This result indicated that filipin was a specific probe for yeast sterols because filipin fluorescence was significantly reduced in yeast compromised for sterol synthesis.

To determine whether filipin fluorescence could be used to detect aberrant sterol distribution, an arv1Δ mutant was treated with filipin. ARV1 encodes a predicted zinc-binding integral membrane protein involved in the maintenance of normal ergosterol distribution in yeast cells (Tinkelenberg et al., 2000). In arv1Δ and wild-type cells, most ergosterol is present at the plasma membrane. However, in arv1Δ cells some ergosterol can be found in fractions corresponding to the endoplasmic reticulum (ER) and the vacuole (Tinkelenberg et al., 2000), organelles that are normally sterol deficient (Zinser et al., 1991). In arv1Δ cells filipin fluorescence was always observed at the cell periphery but in some cells fluorescence was also observed within the cell (Fig. 8B), which was rarely observed in wild-type cells. These results demonstrated that filipin fluorescence could be used in yeast to detect even subtle changes in intracellular sterol distribution.

Our previous analysis of all OSH deletion combinations identified two viable strains that were particularly resistant to nystatin (Beh et al., 2001). To examine sterol distribution in these strains, oshΔ OSH1 OSH3 and oshΔ OSH3 OSH6, cells were treated with filipin and viewed by fluorescence microscopy (Fig. 9A). The staining patterns observed in these OSH mutants differed from that in wild type. Overall, fluorescence was greater in the OSH mutants than in wild-type cells. Within these cells, sterol distribution was most affected in the oshΔ OSH3 OSH6 mutant, where a majority of cells exhibited aberrantly intense cytoplasmic filipin fluorescence. Unlike arv1Δ cells, the internal staining was characterized by bright staining spots and sinuous strands. Moreover, compared with wild type, a reduced number of oshΔ OSH3 OSH6 cells were observed with intense plasma membrane staining. Only a subtle increase in cytoplasmic fluorescence was observed in oshΔ OSH1 OSH3 cells and, between wild-type and oshΔ OSH1 OSH3 cells, there was no discernible difference in plasma membrane fluorescence. In this context, nystatin resistance appeared to be a more sensitive measure of ergosterol defects at the plasma membrane than filipin staining. These findings indicated that in at least one OSH mutant strain the maintenance of normal sterol levels in the plasma membrane was defective and that sterols accumulated in internal membranes.

To determine whether sterol distribution was affected by loss of all the Osh proteins, oshΔ P^MET3-OSH2 cells, and OSH temperature-sensitive oshΔ osh4-1 and oshΔ osh4-2 cells were examined by microscopy after filipin treatment (Fig. 9B,C). When OSH2 was expressed, oshΔ P^MET3-OSH2 cell morphology and filipin/sterol staining were comparable to wild-type cells although some internal filipin fluorescence was visible. OSH2 may be insufficient by itself to rescue the ergosterol distribution defect completely because oshΔ OSH2 strains (where the wild-type OSH2 is the sole OSH gene) exhibit many ergosterol-related defects (Beh et al., 2001). However, when OSH2 expression was repressed in oshΔ P^MET3-OSH2 cells, Osh protein depletion led to significant changes in morphology and sterol distribution in about half the cells (Fig. 9B). In these Osh-depleted cells, overall cell fluorescence increased due to an increase in cytoplasmic filipin fluorescence, and plasma membrane fluorescence decreased. In oshΔ osh4(ts) mutant strains at 37°C, filipin fluorescence at plasma membrane was comparable with wild-type cells but in these mutants significant cytoplasmic filipin staining was evident (Fig. 9C). These results indicated that OSH mutants affect intracellular sterol distribution.

To date, ARV1 is the only gene reported other than the OSHs that affects sterol levels and distribution within yeast cells, without simply blocking sterol synthesis (Tinkelenberg et al., 2000). To determine if, like the OSHs, ARV1 also plays a role in endocytosis, we evaluated lucifer yellow and FM4-64 uptake in arv1Δ cells. After a 60 minute incubation at 37°C, lucifer yellow uptake was reduced in arv1Δ cells relative to wild-type cells (Fig. 10A) indicating that fluid-phase endocytosis was defective. This defect was comparable to the internalization defect observed in erg2/end11Δ cells, which in addition to blocking ergosterol production also caused a pronounced endocytosis defect (Fig. 10A) (Munn et al., 1999). A modest defect was also observed for FM4-64 uptake in arv1Δ cells (Fig. 10B). The arv1Δ uptake defect was not as striking as that observed in las17Δ cells, but FM4-64 uptake was less efficient in the arv1Δ mutant than wild type. Thus, ARV1 affected both sterol distribution and endocytosis.

To determine whether, like OSH mutants, the deletion of ARV1 also had an effect on vacuole morphology, arv1Δ cells were stained for an extended period with FM4-64 to visualize vacuolar membranes. Under these conditions, FM4-64 was internalized and the dye-stained vacuole membranes revealed vacuolar fragmentation in arv1Δ cells (albeit minor compared with erg2Δ or OSH mutants; Fig. 10C). Despite these defects in vacuolar morphology, CPY processing and transport to the vacuole was unaffected in arv1Δ mutants when compared with wild-type cells (C.T.B., unpublished data). These results demonstrated that ARV1 and OSH mutants shared similar defects, and suggested a connection between endocytosis, vacuolar morphology and the maintenance of sterol distribution.

Our results suggested a link between endocytosis and the maintenance of sterol distribution, so we examined whether all endocytosis mutants affect the cellular pattern of filipin/sterol staining. The essential gene SLA2/END4 is involved in actin-dependent endocytosis (Raths et al., 1993; Holtzman et al., 1993). To determine whether SLA2/END4 was required for the normal maintenance of sterol distribution, an end4-1 endocytosis mutant was treated with filipin after incubation at 37°C, its nonpermissive growth temperature. The end4-1 endocytosis mutant had no perceptible accumulation of internalized sterols and no change was detected at the plasma membrane (Fig. 11). These findings indicated that endocytosis defects per se do not cause sterol accumulation within yeast cells.

Actin is essential for yeast endocytosis (Kubler and Riezman, 1993; Munn et al., 1995), and many genes required for endocytosis affect actin assembly and dynamics (reviewed by D'Hondt et al., 2000). To examine whether OSH-associated endocytosis defects were actin dependent, we examined the actin cytoskeleton in OSH mutants. Wild-type and mutant cells were fixed and treated with rhodamine conjugated phalloidin, a fluorescent actin-binding probe. In both the oshΔ osh4-1 and

oshΔ osh4-2 temperature-sensitive mutants at 37°C, the polarization of actin patches and the number and orientation of actin cables were comparable to wild-type cells (C.T.B., unpublished data). In addition, the actin cytoskeleton appeared unaffected in cells containing deletions of either ERG2 or ARV1 (C.T.B., unpublished data). Thus, the endocytosis defects associated with erg2Δ, arv1Δ, and temperature-sensitive OSH mutants were not manifested through any obvious effects on polymerized actin.

From our previous studies of OSH mutants, we determined that the OSH genes perform many nonessential roles and that combined they share at least one essential function (Beh et al., 2001). In this paper we focused on the overlapping functions of all seven OSH genes, and found that the OSHs are involved in daughter cell budding, endocytosis and the maintenance of cellular sterol distribution. These findings imply a specific association between the intracellular transport of sterols and these other processes.

The deletion of KES1/OSH4 suppresses the lethality associated with loss of SEC14, an essential gene required for Golgi vesicle budding, even though deletion of OSH4 has no detectable effect on Golgi secretory transport (Fang et al., 1996). The implication is that OSH4 does not facilitate Golgi secretion but rather acts as a repressor of SEC14-dependent Golgi transport. This suggests that OSH4 overexpression might completely block Golgi secretion. Although the overexpression of any single OSH gene has no affect on growth, and thus no appreciable secretory defect (C.T.B., unpublished data), mammalian OSBPs expressed at high levels in yeast cause growth arrest and CPY-Golgi transport defects (Xu et al., 2001). These results indicate that the OSHs and OSBPs are negative regulators of Golgi secretion, although the purpose and mechanism is unclear.

Our results showed that the entire OSH gene family is dispensable for the transport of several secretory proteins to the plasma membrane. Turning off the last remaining OSH gene, in cells that otherwise lacked all other OSHs, had little impact on Hsp150p, Gas1p, or invertase secretion to the plasma membrane. OSH inactivation also had insignificant effects on CPY transport to the vacuole, and CPY was not mis-sorted at the Golgi to the cell surface. These results indicated that the OSHs are not facilitators of exocytosis or trafficking from the Golgi to the vacuole, but do not preclude a role for the OSHs as negative regulators of secretion. Even so, our data indicated that the OSHs have other important roles in membrane trafficking and organelle morphology, which might be tied to Osh activities in affecting cellular lipid distribution.

The link between OSH mutants and sterols was established through the phenotypic analysis of all OSH deletion combinations (Jiang et al., 1994; Beh et al., 2001). Deletion of specific subsets of the OSH genes cause sterol-related defects affecting tryptophan uptake, ion transport and osmotic stress. In addition, many mutant combinations manifested altered sensitivities to the sterol-specific drugs lovastatin and nystatin. Consistent with these findings, in OSH mutants we observed defects in the cellular distribution of sterols. In addition to the intracellular accumulation of filipin-staining sterols, a decrease in plasma membrane filipin fluorescence was observed in some OSH mutants. Depletion of OSH function also caused lipid droplet accumulation. Following the gradual depletion of Osh proteins the level of cellular ergosterol increases 3.5-fold (Beh et al., 2001). Much of this excess ergosterol presumably resides in the lipid droplets observed by electron microscopy. Lipid droplets and filipin-stained sterols also accumulated after the rapid inactivation of OSH genes in oshΔ osh4(ts) cells at nonpermissive temperatures but total levels of cellular ergosterol were unchanged. Unlike gradual depletion of Osh proteins, rapid inactivation of OSHs in oshΔ osh4(ts) cells did not alter sterol composition or sterol levels when lipid extracts were compared with wild type (C.T.B. and L. Cool, unpublished). These findings indicated that defects in sterol distribution observed in oshΔ osh4(ts) cells were not the result of alterations in sterol synthesis. These results also suggested that the OSH gene family is not directly involved in sterol lipid metabolism but rather in maintaining ergosterol localization at the plasma membrane. During the limited period of growth that follows when Osh proteins are gradually removed, ergosterol mislocalization might perturb regulatory sensing of sterols in a particular membrane and sterol synthesis is thereby erroneously induced.

The disruption of cellular ergosterol distribution by Osh-depletion appears to have specific consequences on vacuolar morphology. In our experiments, regardless of how OSH function was turned off, the loss of all OSH function resulted in vacuolar fragmentation and collapse, and lipid droplets amassed within the fragmented remains of vacuoles. In wild type, lipid droplets were rarely seen within vacuoles, and ergosterol is not normally an abundant vacuolar lipid (Zinser et al., 1991). Substantial vacuolar fragmentation is also observed in ergosterol biosynthesis mutants (Munn et al., 1999; Kato and Wickner, 2001) in which the role of ergosterol in vacuolar integrity appears to be independent of vacuolar transport. Deletion mutants in the ergosterol biosynthetic genes ERG2 or ERG6 perturb vacuolar integrity, but neither deletion blocks CPY transport to the vacuole (Munn and Riezman, 1994). Likewise, in Osh-depleted cells, CPY delivery to the vacuole was essentially unaffected despite vacuolar collapse and fragmentation. Other vacuolar protein-sorting mutants also disrupt vacuole morphology and they can affect transport pathways other than that defined by CPY trafficking (Wickner and Haas, 2000). This raises the possibility that Osh proteins facilitate an aspect of vacuolar or endosomal trafficking other than that involving CPY. As discussed below, the OSHs are in fact necessary for endocytosis.

Inactivation of OSH function in a temperature-sensitive mutant disrupted lucifer yellow and FM4-64 uptake, and retarded Ste6p internalization from the plasma membrane. These results indicate that the OSHs are required for the internalization step of endocytosis. Similar to the OSH genes, we found that another gene that affects cellular ergosterol distribution, ARV1, also affected endocytosis. In arv1Δ cells, levels of plasma membrane sterols are reduced and sterols accumulate in intracellular membranes (Fig. 9B) (Tinkelenberg et al., 2000). We also observed endocytosis defects by lucifer yellow and FM4-64 uptake assays. Although the pattern of sterol accumulation within arv1Δ and Osh-depleted cells appeared to differ, the shared role in endocytosis suggests links between ergosterol and endocytic transport.

The elimination of OSH function disrupted several cellular processes. Possibly the OSHs contribute to each of these processes independently, and the lethality of deleting all OSH genes is a consequence of a mixture of defects that cumulatively compromise growth. However this is unlikely given the rapid growth arrest observed after shifting oshΔ osh4(ts) cells to their nonpermissive temperature. The results suggest that the cellular processes affected by the OSHs are functionally integrated.

If there is a causal link between endocytosis, vacuole integrity and intracellular sterol distribution, then the primary defect caused by Osh protein depletion may be associated with any one of these functions. The evidence suggests that OSH inactivation blocks the recycling of ergosterol back to the cell surface and, in turn, this might explain all phenotypes generated by the loss of OSH function. This implies that plasma membrane sterols are required for endocytosis. Indeed, mutations in several genes required for ergosterol production are required for endocytosis; ERG2, which encodes a late-acting enzyme in the ergosterol biosynthetic pathway, was independently identified as the endocytosis gene END11 (Munn and Riezman, 1994; Munn et al., 1999). Either ergosterol is required for endocytosis, or the precursor sterol that accumulates in ERG2 mutants inhibits internalization. Our results are consistent with the model that ergosterol is necessary for endocytosis because both ARV1 and OSH mutants perturb plasma membrane ergosterol and also disrupt endocytosis.

An alternative possibility is that the Osh proteins are themselves part of a complex directly involved in endocytosis. The rapid halt in endocytosis following OSH inactivation suggests that OSHs are directly involved. We did not detect sterol accumulation in an end4-1 endocytosis mutant indicating that sterol trafficking defects are not ubiquitous aspects of endocytosis mutants. Thus, the OSH genes represent a specific class of lipid-related endocytosis genes.

What is the mechanism by which endocytosis is affected by changes in sterol levels or distribution? The interaction between actin complexes and the plasma membrane is important for yeast endocytosis (Kubler and Riezman, 1993; Munn et al., 1995) and may be sensitive to membrane lipid composition. In ARV1, ERG2 or OSH mutants, however, the actin cytoskeleton and the distribution of actin patches on the plasma membrane appeared unaffected (C.T.B., unpublished data), indicating that the endocytosis defects observed in these mutants were actin independent. Alternatively, changes in plasma membrane sterol composition might disrupt endocytosis by altering membrane structure. In animal cells, the depletion of cholesterol from the plasma membrane alters the morphology of clathrin-coated pits and inhibits budding of clathrin-coated endocytic vesicles (Rodal et al., 1999; Subtil et al., 1999). Thus, ARV1, ERG2 or OSH mutants, all of which affect plasma membrane lipid composition, might also affect the deformation of the plasma membrane required for endocytic vesicle formation. In this respect, the yeast OSHs appear to affect multiple transport processes by changing local membrane lipid composition.

Special thanks to members of the Drubin/Barnes laboratory for antibodies, strains and helpful suggestions, and to the Schekman laboratory for antibodies and strains. We gratefully acknowledge Robin Wright for technical advice and support, and Nancy Hawkins and Keith Kozminski for comments and suggestions on the manuscript. Thanks also to Tim Levine, Sean Munro and Robert Yang Hongyuan for sharing results before publication. This work was funded by a National Institutes of Health grant to J.R. (GM35827). C.T.B. was supported by Leukemia and Lymphoma Society Special Fellows grant #3042-00.