A Novel Membrane Protein, Ros3p, Is Required for Phospholipid Translocation across the Plasma Membrane in Saccharomyces cerevisiae *

Ro09-0198 (Ro) is a tetracyclic peptide antibiotic that binds specifically to phosphatidylethanolamine (PE) and causes cytolysis. To investigate the molecular basis of transbilayer movement of PE in biological membranes, we have isolated a series of budding yeast mutants that are hypersensitive to the Ro peptide. One of the most sensitive mutants, designated ros3(Ro-sensitive 3), showed no significant change in the cellular phospholipid composition or in the sensitivity to amphotericin B, a sterol-binding polyene macrolide antibiotic. These results suggest that the mutation of ros3affects the PE organization on the plasma membrane, rather than PE synthesis or overall organization of the membrane structures. By functional complementation screening, we identified the geneROS3 affected in the mutant, and we showed that the hypersensitive phenotype was caused by the defective expression of theROS3 gene product, Ros3p, an evolutionarily conserved protein with two putative transmembrane domains. Disruption of theROS3 gene resulted in a marked decrease in the internalization of fluorescence-labeled analogs of PE and phosphatidylcholine, whereas the uptake of fluorescence-labeled phosphatidylserine and endocytic markers was not affected. Neither expression levels nor activities of ATP-binding cassette transporters of the ros3Δ cells differed from those of wild type cells, suggesting that Ros3p is not related to the multidrug resistance activities. Immunochemical analyses of the structure and subcellular localization showed that Ros3p was a glycosylated membrane protein localized in both the plasma membrane and the endoplasmic reticulum, and that a part of Ros3p was associated with the detergent-insoluble glycolipid-enriched complexes. These results indicate that Ros3p is a membrane glycoprotein that plays an important role in the phospholipid translocation across the plasma membrane.

Ro09-0198 (Ro) is a tetracyclic peptide antibiotic that binds specifically to phosphatidylethanolamine (PE) and causes cytolysis. To investigate the molecular basis of transbilayer movement of PE in biological membranes, we have isolated a series of budding yeast mutants that are hypersensitive to the Ro peptide. One of the most sensitive mutants, designated ros3 (Ro-sensitive 3), showed no significant change in the cellular phospholipid composition or in the sensitivity to amphotericin B, a sterol-binding polyene macrolide antibiotic. These results suggest that the mutation of ros3 affects the PE organization on the plasma membrane, rather than PE synthesis or overall organization of the membrane structures. By functional complementation screening, we identified the gene ROS3 affected in the mutant, and we showed that the hypersensitive phenotype was caused by the defective expression of the ROS3 gene product, Ros3p, an evolutionarily conserved protein with two putative transmembrane domains. Disruption of the ROS3 gene resulted in a marked decrease in the internalization of fluorescence-labeled analogs of PE and phosphatidylcholine, whereas the uptake of fluorescence-labeled phosphatidylserine and endocytic markers was not affected. Neither expression levels nor activities of ATP-binding cassette transporters of the ros3⌬ cells differed from those of wild type cells, suggesting that Ros3p is not related to the multidrug resistance activities. Immunochemical analyses of the structure and subcellular localization showed that Ros3p was a glycosylated membrane protein localized in both the plasma membrane and the endoplasmic reticulum, and that a part of Ros3p was associated with the detergentinsoluble glycolipid-enriched complexes. These results indicate that Ros3p is a membrane glycoprotein that plays an important role in the phospholipid translocation across the plasma membrane.
Phospholipids in most biological membranes are arranged asymmetrically between the two leaflets of the bilayer. In eukaryotic plasma membrane, aminophospholipids such as phosphatidylethanolamine (PE) 1 and phosphatidylserine (PS) reside predominantly in the inner leaflet, whereas phosphatidylcholine (PC) and sphingolipids are enriched in the outer leaflet (1)(2)(3). This transbilayer distribution of membrane lipids is not a static situation but is likely to be a result of the balance between the inward and outward translocation of phospholipids across the bilayer membranes (4). The rapid translocation of phospholipids between the outer and the inner leaflets of plasma membranes has been detected in various mammalian and yeast cells by fluorescence-labeled or short-chain analogs of phospholipids (5)(6)(7).
In budding yeast, it is shown that the fluorescence-labeled analogs of phospholipids, including PC, PE, and PS, are translocated across the plasma membrane and that this translocation is not significantly prevented in the end and vps yeast mutants in which intracellular vesicular transport, including endocytosis, is impaired (8 -11). Because the translocation reaction requires ATP and is sensitive to sulfhydryl-modifying reagents (8,9), the movements of fluorescence-labeled phospholipids are likely to be facilitated by proteins. Several molecules have been suggested to mediate the transbilayer movements of phospholipids in yeast cells. Some members of yeast ABC (ATP-binding cassette) transporters, such as Ste6p (12), Pdr5p, and Yor1p (13), are shown to facilitate the ATP-dependent outward movement of phospholipid analogs as well as amphipathic drugs. For the inward-directed translocation, an integral membrane P-type ATPase, called Drs2p, was proposed to function as an aminophospholipid translocase, because disruption of yeast DRS2 gene resulted in a significant decrease of the internalization of fluorescence-labeled PS across the plasma membrane (14). However, several groups have failed to detect a difference between wild type and drs2⌬ cells in the translocation of fluorescence-labeled phospholipids across the plasma membrane (10,15). Thus, at present, the molecular identity of the proteins involved in the inward-directed translocation of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Ro is a 19-amino acid tetracyclic polypeptide that strictly recognizes the structure of PE and forms a tight equimolar complex with PE in biological membranes (16,17). The Ro peptide has become a useful tool in monitoring the transbilayer movement of PE in biological membranes (18,19) and in studying the functional role of PE in cytokinesis (20,21) and membrane protein folding (22). Because Ro peptide specifically binds the cell surface PE and subsequently induces cytolysis (18), the peptide is also useful for isolation of mutants with defective PE synthesis as variants that are resistant to the cytolytic activity of the peptide. We have previously isolated a peptide-resistant CHO-K1 cell mutant with specific decrease in cellular PE content, and we have shown that the mutant is defective in intramitochondrial transport of phosphatidylserine (23).
In this study, we have isolated a yeast mutant, designated as ros3, that shows hypersensitivity to the Ro peptide. We found that the ros3 mutation was caused by defective expression of the ROS3 gene product, Ros3p, an evolutionarily conserved protein with two putative transmembrane domains. We have performed a detailed analysis of the structure, function, and subcellular localization of Ros3p, and we have shown that Ros3p is a unique transmembrane protein present in lipid rafts. Evidence that Ros3p plays an important role in regulating phospholipid translocation across the bilayer membranes is presented in this paper.

EXPERIMENTAL PROCEDURES
Yeast Strains and Culture-Saccharomyces cerevisiae strain MHY501 (MAT␣ his3 leu2 lys2 trp1 ura3) was used for the construction of strains deleted for the ROS3 gene. The media used are as follows: YPD, medium containing 1% yeast extract, 2% polypeptone, and 2% glucose; SD, medium containing 0.17% yeast nitrogen base without amino acids and ammonium sulfate, 0.5% ammonium sulfate, 2% glucose, and the required amino acids; SG, SD lacking glucose but containing 2% galactose; SC, SD lacking glucose but containing 2% sorbitol; SCNaN 3 , SC containing 20 mM sodium azide. Unless otherwise noted, the indicated strains were grown to early to mid-log phase at 30°C. Yeast transformations were performed by the lithium acetate method (24).
Isolation of Ro-sensitive (ros) Mutant-Mutagenesis with ethyl methanesulfonate was carried out as described previously (25). Mutagenized cells were plated for single colonies at 23°C. Colonies were replicated onto YPD plates containing 10 M Ro peptide and cultivated for 3 days at 30°C. The negative colonies on the Ro-containing plate were isolated from the master YPD plates.
Drug Sensitivity Assays-For assays with lipid-binding toxins, midlog phase cultures were diluted to 1 ϫ 10 6 cells/ml in YPD containing Ro peptide or amphotericin B (Sigma) at indicated concentrations. After 24 h incubation at 30°C, sensitivity was determined by measuring A 600 of cultures. Sensitivity to other agents such as SDS, Triton X-100, and ethanol was also determined as described above.
Phospholipid Composition Analysis-Mid-log phase cultures in YPD were washed and resuspended in phosphate-buffered saline. Phospholipids were extracted from the harvested cells (2-3 ϫ 10 9 cells) and separated by two-dimensional thin layer chromatography on silica plates. Solvent systems used for chromatography are as follows: first dimension, chloroform/methanol/acetic acid, 65:25:10 (v/v); and second dimension, chloroform/methanol/formic acid, 65:25:10 (v/v). The amounts of phospholipids were determined as described previously (26).
Cloning of the ROS3 Gene-The ROS3 gene was identified by its ability to complement the sensitivity to Ro peptide of the ros3 mutant. The ros3 mutant was transformed with the yeast genomic library, YCp50, which contains the selection marker URA3 and the CEN element. Candidate ROS3 clones were identified based on their ability to recover growth in the presence of 10 M Ro peptide. Only one plasmid was isolated, and it was designated pS3. The pS3 contained an ϳ6.4-kb insert that was a portion of chromosome XIV spanning the upstream of the ORF YNL323W through approximately two-thirds of the ORF YNL320W. There are three full-length ORFs within this region, YNL323W (LEM3), YNL322W (KRE1), and an uncharacterized ORF YNL321W. To identify the complementary region of the ros3 mutant, the 6.4-kb insert was cut into two fragments. One was the 2.2-kb fragment containing the full-length LEM3 gene and half of the KRE1 gene by digestion with SalI and KpnI. Another was the 4.2-kb fragment containing the other half of the KRE1 gene, the full-length YNL321W, and a portion of the ORF YNL320W by digestion with KpnI and Hin-dIII. The pS3-I carries the SalI/KpnI fragment from pS3 in the pRS416 (URA3 and CEN) vector. The pS3-II carries the KpnI/HindIII fragment from pS3 in the pRS416 vector. These plasmids were transformed into the ros3 mutant, and the ability to restore the growth of the ros3 mutant on the plates containing 10 M Ro peptide was examined.
Production of Anti-Ros3p Antibody-Polyclonal antibodies against Ros3p were raised in New Zealand White female rabbits against the synthetic peptides corresponding to the amino-terminal sequence MVNFDLGQVGEVFRRKDKGC, according to the method described previously (27). Antibodies were isolated from the immune sera of the rabbits by affinity chromatography on a synthetic peptide-conjugated SulfoLink column (Pierce).
Deletion of the ROS3 Gene-The SalI/KpnI fragment from pS3 was subcloned into a pUC119 vector. The HpaI fragment between nucleotides 210 and 740 was replaced with the HIS3 gene, which was digested with BamHI from pYAC3 vector and the end blunted with T4 polymerase. The resulting plasmid was cut with AccIII and NcoI, and then the digest was transformed into MHY501. The deletion of chromosomal ROS3 gene was confirmed by PCR analysis using oligonucleotides specific to the 3Ј-and 5Ј-ends of the open reading frame within ROS3 gene. The DNA amplification product from the ROS3-deleted strain (ros3⌬ strain) was ϳ1.2-kb pair larger than the product obtained from the parental strain. The PCR product from the ros3⌬ strain was digested by HindIII into three fragments, whereas the product from the wild type was not digested (Fig. 4A), confirming that the ROS3::HIS3 fragment was integrated into the predicted site of chromosome.
Internalization of Fluorescence-labeled Phospholipids into Yeast Cells-Fluorescence-labeled phospholipids internalization experiments were performed as described by Kean et al. (9). In brief, cultures in SD (A 600 ϭ 0.3-0.8) were diluted to 10 7 cells/ml in fresh SD medium. Cells were then incubated with vesicles containing 40% C 6 -NBD-phospholipids and 60% DOPC (50 M total lipid concentration) with shaking for 30 min at 30°C. For ATP depletion experiments, cells were resuspended in SCNaN 3 containing required amino acids and incubated for 30 min at 30°C before addition of vesicles. After washing the cells three times with ice-cold SCNaN 3 on ice, they were observed in a Zeiss Axioplan microscope equipped with 100ϫ Planneofluar oil immersion objective (Carl Zeiss Co., Ltd., Oberkochen, Germany). Micrographs were taken with Fuji NEOPAN 1600 film. Flow cytometric analyses of the internalization of C 6 -NBD-phospholipid into cells was performed with a FACScan cytometer and CellQuest software (BD PharMingen) as described previously (29).
Vital Dyes Uptake Studies-FM4-64 and Lucifer yellow carbohydrazide (LY-CH) were obtained from Molecular Probes Inc. (Eugene, OR). Cell staining with these dyes was examined following the methods by Vida and Emr (30) and Dulic and Riezman (31). A Zeiss confocal laser scanning microscope using a 543-nm laser line for FM4-64 fluorescence and a 458-nm laser line for LY-CH was used for the observation. Images were analyzed with LSM510 version 2.02 (Carl Zeiss Co., Ltd.).
Activities of ABC Transporters-Total RNA was extracted using RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. First strand cDNA was synthesized using SuperScript TM Choice System for cDNA Synthesis (Invitrogen, Groningen, Netherlands) with the supplied oligo(dT) [12][13][14][15][16][17][18] primer. 25 ng of cDNA was used for each subsequent PCR. For drug resistance assay, cycloheximide, 4-nitroquinoline-N-oxide, miconazole, and ketoconazole were from Sigma. 10 mg/ml stock solutions were made by dissolving the drugs in dimethyl sulfoxide or ethanol. In the case of liquid medium, drug sensitivity was determined as described above.
Protein Extraction and Endoglycosidase H Treatment-Harvested cells were washed in phosphate-buffered saline and resuspended in SDS buffer containing 1/20th volume of 2-mercaptoethanol, and glass beads were added. Disruption of cells was carried out with vortex mixer 7 times for 30 s, and solutions were kept on ice. Lysates were boiled for 5 min, centrifuged to clear the cell debris, and analyzed by 10% SDS-PAGE and Western blotting analysis. Protein concentration was determined by BCA protein assay reagent (Pierce). For endoglycosidase H treatment, the total cell lysates were added to and equal volume of 0.15 M citrate/phosphate buffer (pH 5.0) and 0.1 unit/ml endoglycosidase (Seikagaku Corp.). The reaction mixtures were incubated for 20 h at 37°C.
Subcellular Fractionation-The plasma membrane was isolated as described (32). The isolation of detergent-insoluble glycolipid-enriched complexes (DIGs) was performed according to the method described by Bagnat et al. (33) with some modifications. In brief, mid-log phase cultures (equal to ϳ1 ϫ 10 9 cells) were disrupted with glass beads. The lysates were mixed with an equal volume of TNE buffer (25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA) containing 2% Triton X-100 and incubated for 30 min on ice. After removal of cell debris, the detergent-treated lysates were added with 50% Opti-Prep (Sigma), TNE, 1% Triton X-100, to adjust to 35% Opti-Prep, and overlaid with 30% Opti-Prep, TNE, 1% Triton X-100, and TNE, 1% Triton X-100. The samples were centrifuged at 100,000 ϫ g for 7.5 h at 4°C in a Beckman SW55Ti rotor. Nine fractions of equal volume were collected from the top, precipitated by adding trichloroacetic acid, and resuspended in SDS buffer. Western blotting analysis was performed as described above. Rabbit anti-Pma1p was a gift from R. Serrano (Universidad Politecnica de Valencia, Valencia, Spain). Rabbit anti-Gas1p, -Wbp1, and -Emp47p were provided from H. Riezman (University of Basel, Basel, Switzerland). Mouse anti-Por1p was purchased from Molecular Probes.

Isolation of the ros3 Mutant in a Screen for Mutants
Hypersensitive to Ro09-0198 -Ro09-0198 is a tetracyclic peptide that binds specifically to PE in biological membranes and subsequently induces cytolysis (18,23). To isolate mutants with an altered PE organization on the plasma membrane, we selected a series of S. cerevisiae mutants that were hypersensitive to the peptide-induced cytolysis. We screened ϳ55,000 mutagenized colonies and identified 17 ros (Ro sensitive) mutants. One of the mutants, designated ros3, was most sensitive to the peptideinduced cytolysis, and the LD 50 of the peptide for the ros3 mutant was approximately one-tenth compared with wild type cells (Fig. 1A). There was no significant difference between the mutant and wild type cells in their cellular phospholipid compositions (Table I) or in the sensitivities to various agents such as amphotericin B, a polyene macrolide antibiotic that binds to membrane ergosterol and induces cellular leakage (34) (Fig.  1B), detergents (SDS and Triton X-100), and ethanol (data not shown). These results suggest that the hypersensitivity of the ros3 mutant to the peptide resulted from an altered organiza-tion of PE on the plasma membrane and not from an increased cellular PE content nor from cell wall leakage caused by impaired cell wall integrity.
Cloning of the ROS3 Gene-The gene affected in the ros3 mutant was cloned by complementation of the peptide hypersensitive phenotype after transformation with a genomic library (35). The DNA fragment that complemented the ros3 mutant contains four ORFs ( Fig. 2A). Subcloning experiments indicated that the short fragment derived from KpnI digestion was able to complement, but the other long fragment was not ( Fig. 2A). These results demonstrate that YNL323W (Gen-Bank TM accession number, Z71599), which is also named as LEM3 (36) and BRE3 (37), is the gene that complements the ros3 mutant. In this study, we refer to the gene as ROS3. ROS3 encodes a protein of 414 amino acids that contains two putative transmembrane domains (Fig. 2B). Data base search revealed two genes with high similarity to ROS3, CDC50 (GenBank TM accession number, X59720, 40% identity in deduced amino acid sequence) and YNR048W (GenBank TM accession number, Z71663, 38% identity) in yeast genome (Fig. 2C). In addition, the ROS3 genes are conserved in various organisms including worm (GenBank TM accession number, U61953, 30% identity in deduced amino acid sequence), fly (GenBank TM accession number, AE003502, 33% identity), and mammals (GenBank TM accession number, AA238841, 27% identity), suggesting that ROS3 homologs constitute a large gene family that has been  well conserved during evolution.
To examine further whether the ros3 mutant shows impaired expression of the ROS3 gene product, we produced a polyclonal antibody against a synthetic peptide corresponding to the amino-terminal amino acids 1-19 of Ros3p. In immunoblotting, the affinity-purified anti-Ros3p antibody bound specifically to a 62-kDa protein in the wild type cells, whereas no protein band was detected in the ros3 mutant (Fig. 3A). The observed relative molecular mass was larger than that predicted by the primary structure (47.4-kDa), but the treatment of cell lysate with endoglycosidase H resulted in the reduction of molecular mass to 47.5-kDa which is fully compatible with the predicted mass (Fig. 3B). Because the 62-kDa protein was not present in the lysate of the ROS3-deleted strain (see below) (Fig. 3A), these results clearly demonstrate that the 62-kDa protein is Ros3p and that the mutation of ros3 is caused by the defective expression of Ros3p. Ros3p was not extracted with 1 M NaCl, 2 M urea, and 0.1 M NaCO 3 (pH 11.0) but was partially extracted with 1% (w/v) Triton X-100 (Fig. 3C). These results indicate that Ros3p is a glycosylated transmembrane protein, which is consistent with the predicted structure of the protein that has two putative membrane-spanning domains and six possible glycosylation sites (Fig. 2B).

Ros3p Is Essential for Internalization of C 6 -NBD-PE and C 6 -NBD-PC via a Non-Endocytic
Pathway-To examine the cellular function of Ros3p, we disrupted the ROS3 gene in S. cerevisiae by the replacement of a 531-bp internal HpaI fragment with 1,768-kb fragment encoding the yeast HIS3 gene as shown in Fig. 4A. The deletion of chromosomal ROS3 gene was confirmed by PCR analysis using oligonucleotides specific to the 3Ј-and 5Ј-ends of the open reading frame within ROS3. The ROS3-deleted strain (ros3⌬ strain) could not grow on the YPD plate containing 10 M Ro peptide, and the ROS3 gene on a single copy plasmid suppressed this sensitivity (Fig.  4B). The ros3⌬ strain also showed similar phospholipid composition (Table I) and amphotericin B sensitivity with those observed with the ros3 mutant (data not shown).
To understand the role of Ros3p in PE organization in the plasma membrane, we examined the transport and localization of C 6 -NBD-PE, a fluorescence-labeled analog of PE molecule, in the ros3⌬ strain. It was shown previously that yeast cells incubated with lipid vesicles containing C 6 -NBD-PE internal- ized the fluorescence-labeled lipids from the outer leaflet of the plasma membrane (8 -10, 15). When the cells were incubated at 30°C in the presence of vesicles containing 40 mol % C 6 -NBD-PE and 60 mol % DOPC, the intracellular organelles of the wild type cells were strongly labeled, whereas the intracellular fluorescence of the ros3⌬ strain was much fainter (Fig.  5A, panels a and b). Phospholipid extraction and separation by thin layer chromatography revealed no degradation products of C 6 -NBD-PE in the cell extract (data not shown), indicating that only intact C 6 -NBD-PE was internalized. Flow cytometric analyses showed that the uptake of C 6 -NBD-PE in the ros3⌬ strain was reduced to ϳ10% that of the wild type cells (Fig. 5B). In the ros3⌬ strain, the uptake of C 6 -NBD-PC was also significantly decreased, whereas no significant change in the uptake of C 6 -NBD-PS was observed (Fig. 5B). Depletion of intracellular ATP reduced the uptake of C 6 -NBD-PE in the wild type cells to the level of the ros3⌬ strain (Fig. 5B), suggesting that the Ros3p is responsible for the ATP-dependent uptake of the fluorescence-labeled phospholipids. The uptake and intracellular distribution of the endocytic markers, Lucifer yellow (31) and FM4-64 (30), was not affected in the ros3⌬ strain (Fig. 5A, panels d and f), indicating that the ros3⌬ strain was not defective in endocytosis of soluble and amphipathic molecules. Because the expression of the ROS3 gene by using centromeric plasmid restored the uptake of both C 6 -NBD-PE and C 6 -NBD-PC of the ros3⌬ strain (Fig. 5B), these results suggest that Ros3p is required for the translocation of both C 6 -NBD-PE and C 6 -NBD -PC across the plasma membrane through the non-endocytic pathway.
Ros3p Is Unrelated to Multidrug Resistance Activity-Overexpression of the yeast ABC transporters gives the yeast cells a multidrug resistance phenotype, which results from the accelerated efflux of various amphipathic drugs (38,39). It was shown recently (9, 13) that some of the yeast ABC transporters, such as Pdr5p and Yor1p, exhibit outward-directed phospholipid translocase activity. To examine whether the reduced uptake of the fluorescence-labeled phospholipids in the ros3⌬ strain resulted from the enhanced outward-directed translocation of the phospholipid analogs through ABC transporters, we investigated the mRNA levels of yeast ABC transporters as well as the sensitivity of the ros3⌬ strain against various cytotoxic compounds that are shown to be substrates of the yeast ABC transporters (40,41). Reverse transcriptase-PCR analyses showed no significant difference in the mRNA levels of both Pdr5p and Yor1p between the ros3⌬ strain and wild type cells (Fig. 6A). In addition, there was no significant difference between the ros3⌬ strain and wild type cells in the sensitivities against various amphipathic cytotoxic drugs, such as miconazole, ketoconazole, 4-nitroquinoline-N-oxide, and cycloheximide (Fig. 6, B-E). These results suggest that the uptake deficiency of the fluorescence-labeled phospholipids in the ros3⌬ strain is not caused by the enhanced outward movement of the lipids mediated by the multidrug resistance activity of the transporters.
Subcellular Localization of Ros3p-To assess the subcellular localization of Ros3p, Ros3p tagged at the COOH terminus with EGFP was introduced into the ros3⌬ strain. Chimeric protein Ros3p-EGFP complemented the Ro peptide sensitivity phenotype of the ros3⌬ strain on the centromeric plasmid (data not shown), indicating that Ros3p-EGFP is fully functional. Logarithmically growing cells expressing Ros3p-EGFP by its own promoter showed the fluorescent staining depicted in Fig.  7A, panel a. Ros3p-EGFP was uniformly localized in the nuclear periphery and in discrete patches associated with the cytoplasmic membrane. This pattern coincides with the typical ER markers such as Sec63p (42). In addition to ER, a part of Ros3p-EGFP appeared to be localized at the plasma membrane (Fig. 7A, panel b). To examine further Ros3p localization, a fraction enriched in plasma membranes was prepared by sucrose density gradient fractionation. Ros3p was concentrated in the plasma membrane fraction, which was enriched for the plasma membrane marker protein Pma1p. Other organelle marker proteins including Wbp1p (ER), Emp47p (Golgi), and Por1p (mitochondria) distributed differently in the sucrose gradient (Fig. 7B). The results indicate that Ros3p is localized in both the plasma membrane and the ER membrane.
Bagnat et al. (33) recently isolated the detergent-insoluble glycolipid-enriched complexes (DIGs) from yeast cells and showed that DIGs were composed of phosphoinositol-based sphingolipids, ergosterol, and glycosylphosphatidylinositol-anchored proteins, suggesting that this yeast DIGs is identical to the lipid rafts of mammalian cells. Because Ros3p was partially resistant to Triton X-100 extraction (Fig. 3C), we examined whether Ros3p was localized in the yeast lipid rafts using the Opti-Prep density gradient (33,44). As shown in Fig. 7C, a significant portion of Ros3p was present in the low density DIGs fraction (fraction number 2 and 3) that was enriched with Gas1p and Pma1p, the major components of the yeast lipid rafts (33,44), whereas the non-DIG-associated protein Wbp1p was absent from the DIGs fraction. These results suggest that a part of Ros3p is associated with the yeast lipid rafts and raise the possibility that Ros3p is involved in phospholipid translocation across the bilayer membranes in the plasma membrane and the ER. indicate the ER and the plasma membrane staining, respectively. B, the plasma membrane fraction was purified as described under "Experimental Procedures." The total cellular lysate (L) and plasma membrane fraction (PM) were subjected to SDS-PAGE, followed by immunoblotting analyses with anti-Ros3p, anti-Pma1p (plasma membrane marker), anti-Wbp1 (ER marker), anti-Emp47p (Golgi marker), and anti-Por1p (mitochondria marker). C, detergentinsoluble glycolipid-enriched complexes (DIGs) were isolated as described under "Experimental Procedures." The presence of Ros3p and marker proteins Gas1p and Pma1p, for the major DIG-associated proteins, and Wbp1p for non-DIG-associated protein in each fraction collected from the top of the Opti-Prep density gradient was examined by immunoblotting.

DISCUSSION
In the present study, we have screened yeast mutants that showed hypersensitivity to the PE-binding antibiotic peptide, Ro09-0198. We have isolated 17 clones of mutants and present our initial investigation of a mutant, named ros3, which showed the most significant increase in the sensitivity to the peptide. The ros3 mutant did not show any significant change in cellular phospholipid composition or sensitivity to various agents such as amphotericin B, a sterol-binding polyene macrolide, and detergents. These data suggest that the ros3 mutation affects the PE organization in the plasma membrane, rather than PE synthesis or overall organization of the membrane structures. We first assumed that the hypersensitivity of the ros3 mutant to the peptide resulted from cell surface exposure of PE, which is normally localized in the inner leaflet of the plasma membrane. Hence, we determined the bilayer distribution of PE by using the membrane-impermeable reagent, 2,4,6-trinitrobenzenesulfonic acid, that chemically modifies the polar head group of PE (45). We could, however, not detect more than a slight increase in the 2,4,6-trinitrobenzenesulfonic acid-reactive PE in the ros3 mutant compared with that observed with wild type cells (data not shown). This could not account for the 10 times increase in sensitivity to the peptide. The results suggested that the asymmetric distribution of PE in the plasma membrane was not totally disrupted in the ros3 mutant. Instead a disturbance in either the dynamic movement or surface distribution of PE in the plasma membrane of the ros3 mutated cells could cause the Ro09-0198 hypersensitivity. These observations have prompted us to undertake a detailed analysis of the ros3 mutant. We identified the gene defective in the ros3 mutant and studied the role of Ros3p in the regulation of dynamic movement of PE in the plasma membrane.
Role of Ros3p in Transbilayer Movements of Phospholipids-The gene ROS3 was determined as a defective gene in the ros3 mutant by the functional complementation screening of the genomic library. The ROS3 gene product, Ros3p, was not expressed in the ros3 mutant, and disruption of the ROS3 gene caused the same hypersensitive phenotype as observed in the ros3 mutant. These data further support that the mutation of ros3 was caused by a defective expression of the ROS3 gene product, Ros3p. We then demonstrated that Ros3p is an essential component in the phospholipid translocation at the plasma membrane. The following lines of evidence support our conclusion. First, disruption of the ROS3 structural gene resulted in a marked decrease in the uptake of the fluorescence-labeled PE analog, C 6 -NBD-PE (Fig. 5). Second, the uptake of both soluble and lipophilic endocytic markers, Lucifer yellow and FM4-64, was not affected in the ros3⌬ strain (Fig. 5). This indicates that Ros3p is not involved in the lipid internalization via the conventional endocytic pathway. Finally, a significant portion of Ros3p is localized in the plasma membrane (Fig. 7). It has been shown that ATP depletion treatment at low temperature impairs the C 6 -NBD-phospholipid internalization from the outer leaflet of the yeast plasma membrane but not insertion into the outer leaflet after transport from the donor vesicles (9,11). In the present study, we also confirmed that the C 6 -NBD-PE internalization was ATP-dependent. As the disruption of the ROS3 gene greatly reduced the uptake of C 6 -NBD-PE, it is likely that Ros3p is responsible for the ATP-dependent internalization of the lipid analogs from the outer leaflet of plasma membrane, rather than the ATP-independent insertion of lipid analogs into the outer leaflet of plasma membrane. Taken together, Ros3p is likely to play a role in the phospholipid translocation across the plasma membrane, probably by the transbilayer transport.
Disruption of the ROS3 gene also caused total inhibition of the uptake of C 6 -NBD-PC but showed no significant effect on the uptake of C 6 -NBD-PS, suggesting that Ros3p is involved in the transmembrane movement of PE and PC but not PS. Based on the following discussions, we conclude that Ros3p is a novel type of phospholipid transporter or a regulator of phospholipid translocation, which is distinct from ABC transporters and P-type ATPase. Two protein families have been implicated in phospholipid transport across the yeast plasma membrane so far. The S. cerevisiae genome encodes 15 full-size putative ABC transporters, of which Pdr5p, Yor1p, and Ste6p are suggested to exhibit outward-directed phospholipid translocase activity as well as multidrug resistance activity (12,13). On the other hand, a novel P-type ATPase Drs2p is suggested to mediate the inwarddirected translocation of C 6 -NBD-PS in yeast (14). Ros3p exhibits no significant homology with ABC transporters and Ptype ATPase (Fig. 2). Neither the expression levels nor the cellular activities of ABC transporters were changed in the ros3⌬ strain (Fig. 6), suggesting that Ros3p is unrelated to ABC transporters. Furthermore, C 6 -NBD-PS internalization was not affected by the loss of Ros3p function (Fig. 5), and Ros3p function is thereby not related to Drs2p.
The transbilayer movements of the aminophospholipids such as PE and PS across the plasma membrane is well established and appears to be a common feature in eukaryotic cells. In contrast, an inward-directed translocation of PC at the plasma membrane has only been observed in few cell types, such as SV40-transformed WI-38 (46) and MDCK-II (47). Transbilayer transport of C 6 -NBD-PC across the yeast plasma membrane was also demonstrated previously (8 -11). The translocation of C 6 -NBD-PC, as well as C 6 -NBD-PE and C 6 -NBD-PS, in the yeast plasma membrane requires ATP and is sensitive to sulfhydryl-modifying reagents. Ros3p is likely to mediate translocation of both C 6 -NBD-PE and C 6 -NBD-PC, but not C 6 -NBD-PS (Fig. 5), strongly suggesting that in budding yeast C 6 -NBD-PE and C 6 -NBD-PC are internalized by the same mechanism in which Ros3p is involved and that C 6 -NBD-PS internalization is mediated by another transporter, such as Drs2p.
Intracellular Localization of Ros3p and Its Homolog-Intracellular localization of EGFP-tagged Ros3p suggests that Ros3p is localized in the ER membrane as well as in the plasma membrane (Fig. 7). There are several reports of protein-mediated phospholipid transport across yeast microsomal membrane (48) and mammalian ER membrane (49 -52). These translocase activities are less specific for head group structure and do not require energy. Although the molecular nature of ER phospholipid translocase has not been well defined, it is possible that Ros3p may contribute to the transbilayer movements of phospholipids in the ER membrane. Subcellular fractionation studies indicate that a part of Ros3p is associated with the yeast lipid rafts (Fig. 7). Because the yeast lipid rafts are involved in protein sorting from ER to plasma membrane (33,43), Ros3p may be transported from ER to cell surface through the lipid rafts on demand, though most of Ros3p is retained in the ER membrane during its biosynthesis. Alternatively, Ros3p may be involved in phospholipid translocation in the yeast lipid rafts.
Recent genetic analyses of yeast mutants suggest that the ROS3 gene is involved in the glucocorticoid signal transduction pathway (36) and in the brefeldin A sensitivity (37). Although no obvious explanation for the correlations between these functions of Ros3p has been obtained, these studies imply that the mutation of the ROS3 gene affects diverse cellular functions. The ROS3 gene shows a significant homology with two other yeast genes, CDC50 and YNR048W (Fig. 2). Although it has been reported (53) that the CDC50 mutation gives a coldsensitive phenotype, the functions of these Ros3p homologs are unknown. Despite such remarkable similarities (ϳ60% similarity in amino acid level), deletion of either CDC50 or YNR048W had little effect on the internalization of the C 6 -NBD phospholipids from the plasma membrane (data not shown). It was shown recently that Cdc50p appears to be predominantly localized in late endosome membranes rather than in the ER or the plasma membrane, and the cdc50 mutant shows defects in the uptake of C 6 -NBD-PS, -PC, and -PE at low temperature. 2 Thus, it is possible that the members of Ros3p protein family have distinct characteristics of cellular localization and ability to regulate phospholipid transport in various intracellular organelle membranes.
In conclusion, we have identified a novel membrane protein Ros3p involved in the phospholipid translocation on the yeast plasma membrane. In addition, we have shown that Ros3p is localized in multiple organelles including the plasma membrane and the ER. Further analyses of Ros3p and the ros3⌬ strain will help us understand the regulation of the phospholipid translocation across the bilayer membranes as well as its role in the regulation of various cellular functions.