Opt2 mediates the exposure of phospholipids during cellular adaptation to altered lipid asymmetry

A common feature of the eukaryotic plasma membrane is the difference in lipid composition between the inner (cytosolic) and outer (extracellular) leaflets, which is called lipid asymmetry (Devaux, 1991; Verkleij and Post, 2000). For instance, phosphatidylserine (PtdSer) and phosphatidylethanolamine (PtdEtn) are mostly confined to the inner leaflet, whereas sphingolipids and phosphatidylcholine (PtdCho) are enriched in the outer leaflet. This lipid asymmetry is generated and maintained by ATP-dependent membrane proteins termed flippases and floppases that mediate inward (flip) and outward (flop) movement of lipids between the leaflets, respectively (Axelsen and Palmgren, 1998; Hua et al., 2002; Ikeda et al., 2006; Pomorski et al., 2003; Seigneuret and Devaux, 1984). Proper lipid asymmetry is required for various biological processes, including the generation of membrane potential, establishment of cell polarity, vesicular transport, cytokinesis, blood coagulation and removal of apoptotic cells (Chen et al., 1999; Emoto and Umeda, 2000; Fadok et al., 1992; Furuta et al., 2007; Gurtovenko and Vattulainen, 2008; Saito et al., 2007; Toti et al., 1996). Yeast cells lacking all known flippases are found to be inviable (Hua et al., 2002), suggesting that lipid asymmetry is essential for cell viability. In humans, several mutations in flippases and floppases have been implicated in various diseases including cholestasis, Stargardt macular dystrophy and Scott syndrome (Allikmets et al., 1997; Bull et al., 1998; Toti et al., 1996).

Yeast cells have five phospholipid flippases of the P4-type ATPase family (Dnf1, Dnf2, Dnf3, Drs2 and Neo1), of which Dnf1 and Dnf2 are the main flippases in the plasma membrane and Drs2 functions primarily in the Golgi (Hua et al., 2002). Each flippase forms a complex with a regulatory subunit from the Cdc50 family for its activity; for example, Dnf1 and Dnf2 require complexation with Lem3 in order to exit from the endoplasmic reticulum (ER) and to function as the plasma membrane flippase (Saito et al., 2004). It was recently suggested that phospholipids are flipped along the protein–lipid interface of P4-type ATPases and not through their interior (Baldridge and Graham, 2012). As for the phospholipid floppases, two members of the ABC transporter family (Pdr5 and Yor1) are reported in yeast (Decottignies et al., 1998). In addition, Rsb1 has been identified in yeast as a putative floppase or translocase for sphingoid long-chain bases (Kihara and Igarashi, 2002; Kihara and Igarashi, 2004). It is currently unknown how these proteins flop lipids across the bilayer.

We previously reported that the Rim101 pathway, known as an alkaline-responsive pathway, also senses altered lipid asymmetry caused by the deletion of LEM3 and/or PDR5 (Ikeda et al., 2008). In this pathway, the plasma membrane protein Rim21 acts as the sensor molecule for both altered lipid asymmetry and external alkalization (Obara et al., 2012) and transmits the signal at the plasma membrane (Obara and Kihara, 2014), leading to the proteolytic activation of the transcription factor Rim101 (Peñalva and Arst, 2004). In the alkaline response, the processed Rim101 in yeast induces transcription of alkaline-responsive genes by suppressing the expression of transcription repressors such as Nrg1 and Smp1, whereas in filamentous fungi the processed PacC (a Rim101 homolog) directly induces alkaline-expressed genes (Lamb and Mitchell, 2003; Peñalva and Arst, 2002). Several permeases, secreted enzymes and proteins involved in intracellular pH homeostasis are induced by changes in external pH for adaptation (Causton et al., 2001; Lamb et al., 2001; Peñalva and Arst, 2002; Peñalva and Arst, 2004). In contrast to the pH response, the cellular response to altered lipid asymmetry in the plasma membrane remains largely unknown.

In the present work, we have comprehensively analyzed the cellular response to altered lipid asymmetry using DNA microarray and have identified Opt2 as a novel key factor in cellular adaptation to altered lipid asymmetry.

To evaluate the importance of the Rim101 pathway in adaptation to altered lipid asymmetry, the RIM21 gene encoding the sensor protein in this pathway was deleted in lem3Δ cells, in which lipid asymmetry is disturbed. The resultant rim21Δ lem3Δ double-mutant cells suffered a severe synthetic growth defect (Fig. 1), indicative of the involvement of the Rim101 pathway in the response to altered lipid asymmetry. Although Rsb1, a putative floppase for sphingoid long-chain bases, is induced by alterations in lipid asymmetry (Kihara and Igarashi, 2002; Kihara and Igarashi, 2004), the growth of rsb1Δ lem3Δ cells was comparable to that of lem3Δ cells, indicating that Rsb1 is not important for the adaptation.

In order to comprehensively characterize the cellular response to altered lipid asymmetry, DNA microarray analysis of the yeast genome was performed on wild-type, lem3Δ, pdr5Δ (floppase mutant) and rim21Δ lem3Δ cells. Genes induced by alterations in lipid asymmetry (i.e. genes induced in both lem3Δ and pdr5Δ cells) were selected; twenty-seven of these were further selected as Rim101-pathway-dependent genes because their elevated expression was not observed in rim21Δ lem3Δ cells (supplementary material Table S1). These genes are expected to encode proteins that are inducible through the Rim101 pathway and that function in adaptation to altered lipid asymmetry, and the list included genes encoding transporters, enzymes for glycogen and trehalose metabolism and an arrestin-related protein.

Among the 27 genes extracted, we focused on those encoding, or presumably encoding, integral membrane proteins that might function as flippases or floppases to repair altered lipid asymmetry: MUP3, OPT2, SSU1, MAL31, YDR089W and YJL163C. Each of these genes was then deleted in lem3Δ cells, and only opt2Δ lem3Δ cells were found to exhibit a severe growth defect similar to that of rim21Δ lem3Δ cells (Fig. 2A). Thus, Opt2 appears to play an important role in the adaptation to altered lipid asymmetry. Consistent with the results of DNA microarray analysis (Fig. 2B), Opt2 levels were elevated in both lem3Δ and pdr5Δ cells but were reduced in lem3Δ rim21Δ cells (Fig. 2C). In addition, Opt2 levels in rim21Δ cells were comparable to that in wild-type cells, suggesting that the dependency of Opt2 expression on the Rim101 pathway becomes prominent when lipid asymmetry is altered. Deletion of OPT2 did not affect the activation of the Rim101 pathway as monitored by proteolytic processing of Rim101 (Fig. 2D).

To further confirm the importance of Opt2 in adaptation to altered lipid asymmetry, Opt2 was overexpressed in rim21Δ lem3Δ and opt2Δ lem3Δ cells from the Rim101 pathway-independent ADH1 promoter. As expected, overexpression of Opt2 suppressed the severe growth defect in both strains (Fig. 2A). Given that the Rim101 pathway was originally reported as an alkaline-responsive pathway, the involvement of Opt2 in alkaline response was next investigated. The Rim101-pathway-defective rim21Δ cells were found to be hypersensitive to alkaline pH (Fig. 2E), as others have reported (Castrejon et al., 2006). Interestingly, cells deleted for LEM3 (lem3Δ cells) also exhibited similar sensitivity to alkaline pH. Deletion of OPT2 in wild-type and lem3Δ cells did not affect alkaline tolerance. Therefore, Opt2 appears to have no significant role in alkaline response.

Intracellular localization of Opt2 was monitored by using the N-terminally GFP-fused Opt2. The GFP-tagged Opt2 (GFP–Opt2) was mainly localized to punctate structures, which were colocalized with the late Golgi marker Sec7–mCherry, with limited localization to the plasma membrane (Fig. 3A). When endocytosis was blocked by transient degradation of the actin assembly factor Las17 using the auxin-inducible degron system (Nishimura et al., 2009; Obara and Kihara, 2014), the GFP–Opt2 signal was detected primarily in the plasma membrane (Fig. 3B). This result indicates that Opt2 cycles between the late Golgi and the plasma membrane by endocytosis and secretion of vesicles.

Because Opt2 is implicated in the regulation of vacuolar morphology (Aouida et al., 2009), we investigated the potential involvement of the Rim101-pathway-dependent response to altered lipid asymmetry in vacuolar homeostasis. In contrast to the normal wild-type morphology in lem3Δ, rim21Δ and opt2Δ cells, the vacuoles in rim21Δ lem3Δ and opt2Δ lem3Δ cells were highly fragmented and formed small spheres that were attached to each other (Fig. 4A). Overexpression of Opt2 from the ADH1 promoter could, however, restore the normal vacuole morphology in both double-mutant cells. These observations suggest that Opt2 is involved in the maintenance of vacuole morphology when lipid asymmetry is altered.

Abnormal vacuole morphology is often attributed to defects in vesicular trafficking to the vacuole (Raymond et al., 1992); thus, we examined two known Golgi–vacuole transport pathways [the carboxypeptidase Y (CPY) and AP-3 pathways] in wild-type, lem3Δ, opt2Δ and opt2Δ lem3Δ cells. A soluble vacuolar protein such as CPY is transported from the Golgi to the vacuole through the late endosome (the CPY pathway) (Valls et al., 1987), whereas a vacuolar protein such as the alkaline phosphatase Pho8 is transported to the vacuole directly from the Golgi (the AP-3 pathway) (Klionsky and Emr, 1989). These two pathways were monitored by following the processing of CPY and Pho8, respectively, using immunoblot analysis (Fig. 4B). In vps21Δ cells in which the CPY pathway is defective, some of the Golgi CPY (p2-CPY) was secreted to the extracellular medium as reported in the literature (Robinson et al., 1988). By contrast, in lem3Δ, opt2Δ, opt2Δ lem3Δ and wild-type cells, both CPY and Pho8 were processed to their normal vacuolar forms, i.e. the mature form of CPY (m-CPY) and the mature and soluble forms of Pho8 (m- and s-Pho8), respectively. Next, the endocytic pathway was monitored using the lipophilic dye FM4-64, which is first incorporated into the plasma membrane and then transported to the vacuole by endocytosis. Cells were treated with FM4-64 for 15 min and chased for 10 min. In all the cells, the FM4-64 signal was detected mainly in the vacuolar membrane and partially at the endosome (Fig. 4C), indicating that endocytosis is not affected in opt2Δ lem3Δ cells. Taken together, it can be concluded that Opt2 does not play a role in vesicular trafficking to the vacuole.

To gain further insight into how Opt2 mediates adaptation to altered lipid asymmetry, we measured the degree of lipid asymmetry in cells lacking or overexpressing Opt2 based on their sensitivity to the toxic peptides duramycin and papuamide B, which specifically bind to cell surface PtdEtn (Zhao et al., 2008) and PtdSer (Parsons et al., 2006), respectively. In lem3Δ cells, PtdEtn and PtdSer, normally confined to the inner leaflet of the plasma membrane, are exposed to the cell surface (the outer leaflet) owing to the defective flippase activity; as a result, lem3Δ cells were hypersensitive to both duramycin and papuamide B (Fig. 5A) (Noji et al., 2006; Parsons et al., 2006).

Deletion of either OPT2 or RIM21 in lem3Δ cells (opt2Δ lem3Δ or rim21Δ lem3Δ cells, respectively) conferred some tolerance to both peptides, although the growth of the double mutants on plates without the peptides was severely retarded; however, these cells became sensitive again when Opt2 was overexpressed from the ADH1 promoter. Thus, Opt2 induced by the Rim101 pathway seems to be involved in the exposure of PtdEtn and PtdSer. However, it must be noted that deletion of RIM21 had a greater effect than the deletion of OPT2 on the sensitivity of lem3Δ cells to papuamide B. Furthermore, overexpression of Opt2 had a milder effect on papuamide B sensitivity in rim21Δ lem3Δ cells than in opt2Δ lem3Δ cells (see Discussion). In rim21Δ cells, double deletion of DNF1 and DNF2, both of which encode phospholipid floppases regulated by Lem3, caused the same effects as deletion of the LEM3 gene – the dnf1Δ dnf2Δ rim21Δ cells exhibited slow growth and acquired tolerance to duramycin (supplementary material Fig. S1A), confirming that the effect of the LEM3 deletion is indeed the result of a defect in flip.

The surface-exposed PtdEtn was then detected by using PtdEtn-specific biotinylated Ro09-198 (Bio-Ro) and was visualized by using FITC-conjugated streptavidin (Iwamoto et al., 2004) (Fig. 5B). The structure of Ro-09-198 closely resembles that of duramycin (Noji et al., 2006). As described in the literature (Iwamoto et al., 2004; Saito et al., 2007), PtdEtn was highly exposed in lem3Δ cells but little PtdEtn was exposed in wild-type cells. Notably, the exposed PtdEtn in lem3Δ cells was significantly reduced by deletion of OPT2 (opt2Δ lem3Δ cells), and overexpression of Opt2 was found to enhance the surface exposure of PtdEtn in opt2Δ lem3Δ cells (Fig. 5C). We confirmed that the levels and localization of Drs2 and Dnf3 were comparable in these mutant cells and in wild-type cells (supplementary material Fig. S1B,C) and thus eliminated the possibility of increased levels of Lem3-independent flippases causing the reduction in the amount of exposed PtdEtn in opt2Δ lem3Δ cells. Interestingly, levels of Pdr5, the plasma membrane floppase, were significantly elevated in opt2Δ lem3Δ cells. These findings substantiate the involvement of Opt2 in the exposure of PtdEtn.

The flip-flop-mediated transfer of fluorescently labeled phospholipids (NBD–PtdEtn, NBD–PtdCho, and NBD–PtdSer) was next assessed in cells overexpressing Opt2. After back-extraction of NBD-labeled phospholipids from the outer leaflet with bovine serum albumin (BSA), the increase in flop activity was estimated by the decrease in the levels of NBD-phospholipids in the inner leaflet using flow cytometry. Overexpression of Opt2 resulted in an ∼25% decrease in incorporation of NBD-labeled phospholipids into the inner leaflet (Fig. 5D), indicative of the direct or indirect involvement of Opt2 in phospholipid flop.

The yeast bud grows apically during the early phase of budding, during which PtdEtn is localized to the outer leaflet at the tip of the elongating bud. When the apical growth switches to an isotropic growth, the exposed PtdEtn is flipped back to the inner leaflet. This local alteration in lipid asymmetry regulates polarized cell growth; hence, the flippase-defective lem3Δ cells have an elongated morphology (Saito et al., 2007). We next investigated whether Opt2 plays any role in apical growth by measuring the long∶short axis ratio of the cell. The ratio of lem3Δ cells was significantly larger than that of wild-type cells as predicted, whereas opt2Δ cells were found to have a smaller ratio than that of wild-type cells (Fig. 6), thus being more spherical than wild-type cells. In addition to its role in adaptation to altered lipid asymmetry, Opt2 appears to play an important role in apical growth.

We have identified Opt2, originally reported as a member of the oligopeptide transporter family (Lubkowitz et al., 1998), as a novel factor required for the Rim101-pathway-dependent adaptation to altered lipid asymmetry. Although the Rim101 pathway is known as an alkaline-responsive pathway (Peñalva and Arst, 2004), we have demonstrated that Opt2 is not involved in the alkaline response (Fig. 2). It is also known that the OPT2 gene is not induced by external alkalization (Causton et al., 2001; Lamb et al., 2001; Serrano et al., 2002). Interestingly, only one putative gene (YHR214W-A) among the 27 genes extracted from our microarray data (supplementary material Table S1) is reported as an alkaline-responsive gene. Thus, although both external alkalization and altered lipid asymmetry activate the Rim101 pathway, the set of genes induced by each perturbation is not likely to be determined solely by the Rim101 pathway, but rather is determined in cooperation with other signaling pathways.

Both trehalose and glycogen are known to accumulate in yeast during stress (François and Parrou, 2001). This seems to be also the case in response to altered lipid asymmetry, because most of the genes encoding enzymes for trehalose and glycogen metabolism were induced in cells with altered lipid asymmetry (lem3Δ and pdr5Δ cells; supplementary material Fig. S2). Of these, three genes (TPS2, GSY1 and GIP2) were induced in a Rim101-pathway-dependent manner (supplementary material Table S1). The ART2 and ART4 genes encoding arrestin-related proteins were also induced by altered lipid asymmetry (supplementary material Fig. S3) with the induction of ART2 being Rim101 pathway dependent (supplementary material Table S1). Arrestin-related proteins are involved in the endocytic turnover of plasma membrane proteins such as nutrient transporters and receptors. In this regard, it is interesting that several genes encoding transporters (or putative transporters) were upregulated in lem3Δ and pdr5Δ cells in a Rim101-pathway-dependent manner, including MUP3, OPT2, MAL31 and SSU1 (supplementary material Table S1). It is an interesting subject to investigate the endocytic turnover of plasma membrane proteins and the involvement of arrestin-related proteins in adaptation to altered lipid asymmetry. It should be noted that the expression of ART9 (the gene encoding the arrestin-related protein Art9, also known as Rim8 and essential for the Rim101 pathway) was downregulated in lem3Δ cells and upregulated in lem3Δ rim21Δ cells (supplementary material Fig. S3), which is consistent with the fact that its expression is negatively regulated by the Rim101 pathway (Lamb and Mitchell, 2003).

We have provided evidence for the involvement of Opt2 in phospholipid exposure on the extracytoplasmic leaflet. One simple explanatory hypothesis is that Opt2 is a floppase that directly mediates trans-bilayer movement of phospholipids. In this case, Opt2 would represent a novel type of floppase, because it does not belong to the ABC transporter family to which all known floppases (except for Rsb1) belong. An alternative possibility is that Opt2 mediates the exposure of phospholipids indirectly, e.g. by inhibiting or activating unknown flippases or floppases, respectively, that function during adaptation to altered lipid asymmetry. It is also possible that Opt2 regulates the localization of unknown flippases and floppases. Therefore, one of the most important directions for future work would be the establishment of an in vitro system that allows direct evaluation of Opt2-mediated translocation of phospholipids.

Our results demonstrate that Opt2 is involved in the exposure of phospholipids, especially in lem3Δ cells, in which phospholipid flip in the plasma membrane is largely impaired. In this regard, it is reasonable to predict that induction of Opt2 would exacerbate the phenotype of lem3Δ cells; however, contrary to this assumption, we showed that induction of Opt2 is an important process in Rim101-pathway-mediated adaptation to altered lipid asymmetry caused by LEM3 deletion. These apparently paradoxical results might be explained follows. Disturbance in lipid asymmetry likely affects events throughout the cell, because lipid asymmetry is involved in a wide range of processes including the generation of membrane potential, vesicular trafficking, establishment of cell polarity and cytokinesis. Therefore, the cellular response to altered lipid asymmetry must be fairly complex. Regulated local activation or suppression of flip-flop might be essential for a wide range of cellular events, and several factors involved in flip-flop (including unknown factors) are likely to be induced, activated or repressed when lipid asymmetry is altered. Indeed, in some cases, NBD–PtdSer is flipped more in lem3Δ cells than in wild-type cells (Saito et al., 2004), suggesting that some unknown compensatory mechanism is induced in these cases. Therefore, it is unlikely that effects caused by a reduction in flip (in lem3Δ cells) can be cancelled by a simultaneous reduction in flop. Opt2 is likely to be induced as an important factor in such global responses.

Overexpression of Opt2 completely counteracted the duramycin resistance of lem3Δ rim21Δ cells (Fig. 5A). By contrast, overexpression of Opt2 in wild-type cells caused a rather small reduction in the incorporation of NBD-labeled phospholipids into the inner leaflet of the plasma membrane (Fig. 5D). These somewhat inconsistent results might be due to the presence of the active plasma membrane flippases Dnf1 and Dnf2 in wild-type but not in lem3Δ rim21Δ cells. Alternatively, the seeming inconsistency between the results of the two assays might be due to the fact that Opt2 is primarily localized to the Golgi, with only a minor fraction located at the plasma membrane (Fig. 3). Alternatively the differences might be due to the dramatic difference in the incubation times used for the assays – 20 min for the NBD-labeled phospholipid transport assay versus 50 h for the duramycin-sensitivity assay. Thus, the results from the NBD-labeled phospholipid assay reflect low levels of Opt2 in the plasma membrane; by contrast, during the longer incubation period, the Opt2-mediated lipid asymmetry generated at the Golgi could travel to the plasma membrane through secretory vesicles, making the cells highly sensitive to duramycin. It could also be possible that if Opt2 functions as a subunit of the floppase complex, the majority of overexpressed Opt2 would not fully function because the other subunit is absent.

The effect of overexpression of Opt2 in rim21Δ lem3Δ cells was found to be much greater on the duramycin sensitivity than the papuamide B sensitivity, and opt2Δ lem3Δ cells showed much weaker tolerance to papuamide B than rim21Δ lem3Δ cells (Fig. 5A). Therefore, Opt2 might not be the sole factor that determines the exposure of PtdSer during adaptation to altered lipid asymmetry; instead, some other key protein(s) might be induced by the Rim101 pathway.

We have observed vacuole fragmentation in opt2Δ lem3Δ cells, indicative of the possible involvement of Opt2 in the maintenance of vacuole morphology when lipid asymmetry is altered (Fig. 4). Given that Opt2 shuttles between the Golgi and the plasma membrane (Fig. 3), it is conceivable that the lipid asymmetry generated by Opt2 at the Golgi and the plasma membrane might be propagated to the vacuole membrane through the vesicular transport pathway. The fragmentation of vacuoles could also be the result of the improper localization and activity of proteins in opt2Δ lem3Δ cells required for vacuole maturation, such as those involved in the homotypic fusion of vacuoles. However, the latter possibility is less likely because opt2Δ lem3Δ cells did not affect the maturation of vacuolar proteins such as CPY and Pho8, which is known to be impaired in cells defective in the homotypic fusion of vacuoles (Wada et al., 1992). In addition, the morphology and distribution of vacuoles appear to be different between opt2Δ lem3Δ and homotypic fusion mutant cells – they form small spherical vacuoles that are attached to each other (opt2Δ lem3Δ cells) versus much smaller discrete vacuoles dispersed throughout the cytoplasm (cells lacking the Rab-type small GTPase Ypt7 essential for the homotypic vacuole fusion) (Wada and Anraku, 1992).

We have also demonstrated that Opt2 is involved in the apical growth of yeast cells. This finding is supported by the previous analysis of yeast cell morphology (Ohya et al., 2005) that revealed a smaller long∶short axis ratio of opt2Δ cells than that of wild-type cells at all growth stages investigated [for details, refer to the Saccaromyces cerevisiae Morphological Database (http://scmd.gi.k.u-tokyo.ac.jp/datamine/)]. Interestingly, the larger ratio was reported for cells carrying a deletion of either PDR5 or YOR1, both of which encode the phospholipid floppase. It seems that Opt2 might be specifically involved in PtdEtn flop at the bud tip during apical growth. Future studies should be aimed at direct in vitro examination of the flop activity of Opt2, identification of proteins interacting with Opt2 and elucidation of the regulation of Opt2 function, which would greatly deepen our understanding of cellular adaptation to altered lipid asymmetry.

Saccharomyces cerevisiae strains used in this study are listed in supplementary material Table S2. Yeast cells were grown at 30°C to log phase in YPD (1% yeast extract, 2% bactopeptone and 2% D-glucose) or synthetic complete (SC) medium (2% D-glucose and 0.67% yeast nitrogen base without amino acids) with appropriate supplements. Alkaline treatment was performed by adding 1 M Tris-HCl (pH 8.0) to culture medium at a final concentration of 100 mM. A 500 mM stock solution of 3-indoleacetic acid (IAA; Nacalai Tesque, Kyoto, Japan) was prepared in ethanol and added to the medium at a final concentration of 500 µM.

Gene disruption was performed by replacing the entire coding region of the gene with a marker gene. Chromosome fusion of mCherry or Myc to the C-terminus of Sec7 or Opt2, respectively, was performed using PCR-based gene disruption and modification (Longtine et al., 1998). The sequence encoding mCherry or Myc, the ADH1 terminator and a marker sequence was amplified by PCR from the pFA6a vector series (Longtine et al., 1998) with a primer set specific to each gene. For the chromosomal fusion of GFP to the N-terminus of Opt2, the sequence encoding a marker sequence, the ADH1 promoter and the GFP tag was amplified by PCR from the pYM-N9 (Janke et al., 2004) vector with a primer set specific to OPT2. Amplified cassettes were inserted directly into the chromosome by homologous recombination. Integration of the P_ADH1-HA-OPT2 gene into the URA3 or TRP1 locus was performed as follows. The P_ADH1-HA-OPT2-T_OPT2 sequence was amplified by PCR from genomic DNA of YOK3260 (P_ADH-HA-OPT2) to have the SacI and XhoI sites at the 5′ and 3′ ends, respectively. The amplified fragment was cloned into the SacI-XhoI site of pRS306 or pRS304 (Sikorski and Hieter, 1989) to generate pOK563 or pOK571, respectively. The pOK563 and pOK571 constructs were linearized by digestion with StuI and HindIII, respectively, and inserted at the URA3 and TRP1 loci, respectively. The plasmid for the expression of HA–Rim101 (pFI1) was a gift from Dr. Tatsuya Maeda (University of Tokyo, Japan).

Proteins were separated by SDS-PAGE and transferred to an Immobilon^TM polyvinylidene difluoride membrane (Millipore, Billerica, MA) as described previously (Yamagata et al., 2011). The membrane was incubated with antibody against HA (16B12; Covance, Princeton, NJ), CPY (Molecular Probes, Eugene, OR), GFP (598; Medical & Biological Laboratories, Nagoya, Japan), Myc (PL-14; Medical & Biological Laboratories), Pho8 (a gift from Prof. Yoshinori Ohsumi, Tokyo Institute of Technology, Japan) or Pgk1 (Molecular Probes). Immunodetection was performed using Western Lightning ECL Pro system (PerkinElmer Life Sciences, Waltham, MA) with a bioimaging analyzer (LAS4000; Fuji Photo Film, Tokyo, Japan) or X-ray film.

Total RNA was prepared from yeast cell homogenates using the RNeasy Mini kit (Qiagen, Hilden, Germany). Poly (A)⁺ RNA was purified using the mRNA Purification Kit (Amersham Pharmacia Biotech, Piscataway, NJ). The quality of RNA was verified by electrophoresis using Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). A total of 200 ng of poly (A)⁺ RNA was labeled with Cy3 using Low Input Quick Amp Labeling Kit (Agilent Technologies), and hybridized with the Yeast (V2) Gene Expression Microarray (Agilent Technologies) according to the manufacturer's instructions. The microarray was scanned with the Agilent G2565BA microarray scanner (Agilent Technologies), and the fluorescence intensity for each spot was quantified using Feature Extraction software (Agilent Technologies). RNA from four independent cultures for each strain was subjected to DNA microarray assay and their mean values were analyzed using Subio Platform software (Subio, Kagoshima, Japan).

Exposed PtdEtn was visualized using Bio-Ro as reported previously (Iwamoto et al., 2004; Saito et al., 2007) with slight modifications. A 1-ml culture of cells in log phase was harvested and incubated in 20 µl of YPD medium containing 80 µM Bio-Ro at 4°C for 3 h. Cells were washed once with PBS and fixed with 5% formaldehyde in PBS at room temperature for 1 h. Cells were then washed with spheroplast buffer (1.2 M sorbitol, 0.1 M potassium phosphate, pH 7.4) and resuspended in 100 µl of spheroplast buffer containing 100 µg/ml zymolyase 100T (Seikagaku Kogyo, Tokyo, Japan). After addition of β-mercaptoethanol to a final concentration of 28 mM, cells were incubated at 30°C for 10 min and washed twice with spheroplast buffer. Spheroplasted cells were attached to poly-L-lysine-coated multiwell slides and incubated in PBS containing 0.1% BSA at room temperature for 20 min. Cells were washed three times with PBS, and incubated in PBS containing 5 µg/ml Fluorescein–streptavidin (Vector Laboratories, Burlingame, CA) at room temperature for 1 h. After five washes with PBS, cells were suspended in ProLong Gold Antifade reagent (Life Technologies, Carlsbad, CA) and subjected to fluorescence microscopy.

NBD–PtdEtn (1-myristoyl-2-{6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-phosphoethanolamine), NBD–PtdCho (1-myristoyl-2-{6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-phosphocholine) and NBD–PtdSer (1-palmitoyl-2-{6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-phosphoserine) were purchased from Avanti Polar Lipids (Alabaster, AL). For each strain, cell solutions with an OD₆₀₀ of 1.0 were harvested and the cells were resuspended in SC medium. After the addition of NBD–PtdEtn, NBD–PtdCho or NBD–PtdSer at a final concentration of 50 µM, cells were incubated at 30°C for 20 min, washed with SSA medium (0.67% yeast nitrogen base without amino acids, 2% sorbitol, 0.5% casamino acids, 20 mg/l tryptophan, 20 mg/l adenine sulfate, 20 mg/l uracil, 0.067% sodium azide), and transferred to a new tube. Cells were further washed twice with SSA medium containing 4% BSA and once with SSA medium and resuspended in 100 µl of SSA medium. Each sample was diluted with SSA medium at 1∶100 and analyzed by flow cytometry using a FACS Calibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ). For each strain, values were corrected by subtracting the background values for the unstained cells.

Cells were grown in SC medium to log phase, and cell solutions with an OD₆₀₀ of 0.5 were harvested and the cells were resuspended in 400 µl of SC medium. Formaldehyde and potassium phosphate buffer (pH 6.6) were added at a final concentration of 3.7% and 100 mM, respectively, and cells were incubated at 25°C for 30 min to fix the cells. Cells were collected by centrifugation and treated again with 100 mM potassium phosphate buffer (pH 6.6) containing 4% formaldehyde at room temperature for 45 min. The fixed cells were washed and resuspended in 500 µl of 100 mM potassium phosphate buffer (pH 6.6), stained with 1.6 µg/ml FITC–concanavalin-A (Sigma, St Louis, MO) at room temperature for 10 min, and observed under fluorescence microscopy. The long and short axes of each cell were measured using Image J software (National Institutes of Health, Bethesda, MD) and their ratio was calculated. For each strain, 125 individual cells were analyzed.

Fluorescence was visualized using a fluorescence microscope (DM5000B, Leica Microsystems, Wetzlar, Germany) equipped with 100× HCX PL FLUOTAR NA 1.30 oil-immersion objective. Images were acquired with a cooled CCD camera (DFC365FX, Leica Microsystems) controlled with LAS AF software (version 2.60, Leica Microsystems) and archived using Photoshop CS3 (Adobe; San Jose, CA). In some cases, a linear adjustment was applied to enhance the image contrast using the level adjustment function of Photoshop. To visualize the vacuole, cells in log phase were stained with 1 µM FM4-64 (Molecular Probes) for 30 min, washed and resuspended in the medium, and incubated for an additional 30 min. To monitor the progression of endocytosis, cells in log phase were treated with 1 µM FM4-64 for 15 min, washed and resuspended in the same medium, and incubated for an additional 10 min. After addition of sodium azide to a final concentration of 20 mM, cells were kept on ice until examined by microscopy.

We are grateful to Prof. M. Umeda and Dr. U. Kato (Graduate School of Engineering, Kyoto University) for providing the biotinylated Ro 09-0198 (Bio-Ro), to Dr. T. Maeda (Institute of Molecular and Cellular Biosciences, University of Tokyo) for providing the HA-Rim101 plasmid (pFI1) and to Prof. Y. Ohsumi (Tokyo Institute of Technology) for providing the anti-Pho8 antibody. The template plasmid for N-terminal tagging with GFP (pYM-N9) was provided by the European S. cerevisiae Archive for Functional Analysis (Euroscarf, Germany) and the AID system was from the National Bio-Resource Project (NBRP) of the MEXT, Japan. We also thank Dr. T. Toyokuni for editing the manuscript.