Loss of Drs2p does not abolish transfer of fluorescence-labeled phospholipids across the plasma membrane of Saccharomyces cerevisiae.

The yeast DRS2 gene, which is required for growth at 23 degreesC or below, encodes a member of a P-type ATPase subgroup reported to transport aminophospholipids between the leaflets of the plasma membrane. Here, we evaluated the potential role of Drs2p in phospholipid transport. When examined by fluorescence microscopy, a drs2 null mutant showed no defect in the uptake or distribution of fluorescent-labeled 1-palmitoyl-2[6-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl (NBD))aminocaproyl]phosphatidylserine) or 1-myristoyl-2[6-NBD-aminocaproyl]phosphatidylethanolamine. Quantification of the amount of cell-associated NBD fluorescence using flow cytometry indicated a significant decrease in the absence of Drs2p, but this decrease was not restricted to the aminophospholipids (phosphatidylserine and phosphatidylethanolamine) and was dependent on culture conditions. Furthermore, the absence of Drs2p had no effect on the amount of endogenous PE exposed to the outer leaflet of the plasma membrane as detected by labeling with trinitrobenzene sulfonic acid. The steady state pool of Drs2p, which was shown to reside predominantly in the plasma membrane, increased upon shift to low temperature or exposure to various divalent cations (Mn2+, Co2+, Ni2+, and Zn2+ but not Ca2+ or Mg2+), conditions that also inhibited the growth of a drs2 null mutant. The data presented here call into question the identification of Drs2p as the exclusive or major aminophospholipid translocase in yeast plasma membranes (Tang, X., Halleck, M. S., Schlegel, R. A., and Williamson, P. (1996) Science 272, 1495-1497).

The yeast DRS2 gene, which is required for growth at 23°C or below, encodes a member of a P-type ATPase subgroup reported to transport aminophospholipids between the leaflets of the plasma membrane. Here, we evaluated the potential role of Drs2p in phospholipid transport. When examined by fluorescence microscopy, a drs2 null mutant showed no defect in the uptake or distribution of fluorescent-labeled 1-palmitoyl-2[6-(7-nitrobenz-2-oxa-1,3-diazol-4-yl (NBD))aminocaproyl]phosphatidylserine) or 1-myristoyl-2[6-NBD-aminocaproyl]phosphatidylethanolamine. Quantification of the amount of cell-associated NBD fluorescence using flow cytometry indicated a significant decrease in the absence of Drs2p, but this decrease was not restricted to the aminophospholipids (phosphatidylserine and phosphatidylethanolamine) and was dependent on culture conditions. Furthermore, the absence of Drs2p had no effect on the amount of endogenous PE exposed to the outer leaflet of the plasma membrane as detected by labeling with trinitrobenzene sulfonic acid. The steady state pool of Drs2p, which was shown to reside predominantly in the plasma membrane, increased upon shift to low temperature or exposure to various divalent cations (Mn 2؉ , Co 2؉ , Ni 2؉ , and Zn 2؉ but not Ca 2؉ or Mg 2؉ ), conditions that also inhibited the growth of a drs2 null mutant. The data presented here call into question the identification of Drs2p as the exclusive or major aminophospholipid translocase in yeast plasma membranes ( Most biological membranes appear to possess a nonrandom distribution of phospholipids between the two leaflets of the bilayer (1). An asymmetric organization of phospholipids, in which phosphatidylethanolamine (PE) 1 and phosphatidyl-serine (PS) are enriched in the inner leaflet facing the cytoplasm, and phosphatidylcholine (PC), sphingomyelin, and glycolipids are predominantly located on the outer leaflet, has been well documented for the plasma membranes of numerous cell types (2,3). The loss of this asymmetric distribution and the resulting appearance of PS at the cell surface triggers a variety of intercellular communication and signaling processes, such as platelet activation (4), clearance of senescent red cells (5), and phagocytosis of apoptotic cells (6,7). However, the establishment and regulation of this asymmetric distribution as well as its physiological function in single cells are poorly understood. It is generally thought that the transbilayer movement ("flip-flop") of phospholipids is mediated by ATP-dependent flippases, and several proteins with flippase activity have been identified in mammalian cells. The ABC transporters, human MDR1 and MDR3 (8,9), mouse mdr2 (10), and yeast Pdr5p and Yor1p (11) have been shown to exhibit outwarddirected phospholipid flippase activity.
A recent report on the cloning of a flippase from bovine chromaffin granules has implicated a novel subgroup within the P-type ATPase family in the inward-directed transport of aminophospholipids (12). Most P-type ATPases, represented by a group of 18 genes in the genome of the yeast Saccharomyces cerevisiae (13), are biochemically well characterized and known to function in the transport of mono-or divalent cations (14). The Na ϩ /K ϩ ATPases, various Ca 2ϩ ATPases of animal cells as well as the H ϩ -ATPases of fungi and plants belong to this family of ion transporters, which share a characteristic set of conserved regions and a similar transmembrane topology (14). The members of the new subgroup differ from the ion-transporting ATPases in several amino acids within transmembrane segments critically involved in ion translocation. Apart from the bovine cDNA, the yeast DRS2 gene and four related yeast genes (13) as well as sequences from Plasmodium falciparum and Caenorhabditis elegans appear to carry these changes (12) whereby negatively charged residues have been replaced by bulky, hydrophobic groups. The observation of a defect in fluorescent-labeled PS (P-C 6 -NBD-PS) internalization in a drs2 mutant at low temperature has been interpreted as evidence for the biochemical function of this group of enzymes as aminophospholipid translocases (12).
In this report, we examined the potential role of Drs2p in phospholipid transport using fluorescence microscopy, flow cytometry, and TNBS labeling. The deletion of DRS2 had no effect on the uptake or distribution of fluorescent-labeled PS (P-C 6 -NBD-PS) or PE (M-C 6 -NBD-PE) detected by fluorescence microscopy. Quantification of the amount of cell-associated NBD fluorescence indicated a significant decrease in the absence of Drs2p, but this decrease was not exclusive to the aminophospholipids (PE and PS) and was dependent on culture conditions. The absence of Drs2p had no effect on the amount of endogenous PE exposed to the outer leaflet of the plasma membrane as detected by labeling with TNBS. The steady state pool of Drs2p, which was shown to reside predominantly in the plasma membrane, increased upon shift to low temperature or exposure to various heavy metal cations, conditions that inhibited the growth of a drs2 null mutant. The data presented here call into question the identification of Drs2p as the exclusive or major aminophospholipid translocase in yeast plasma membranes (12).
Construction of drs2-1::URA3 and HA::DRS2 Alleles-To construct drs2-1::URA3, the 5.9-kb EcoRI fragment carrying the entire DRS2 gene was isolated from a DRS2-containing plasmid (a gift from J. Woolford) and subcloned into pUC19 (18). Subsequently, the 3.5-kb BstEII-HpaI fragment within the DRS2 ORF was replaced with URA3 sequences by inserting the 1.1-kb HindIII fragment of URA3 via bluntend ligation. DNA from the resulting plasmid was digested with EcoRI and used to transform YR98. The replacement of the DRS2 gene by drs2-1::URA3 DNA was verified in one of the Ura ϩ transformants (YR884) by Southern blot analysis.
To obtain the HA::DRS2 allele, a 27-base pair insertion of HAencoding DNA (19) was introduced into the 1.1-kb EcoRI-XhoI fragment of DRS2 by joining the polymerase chain reaction products of two independent polymerase chain reaction reactions on a DRS2 template via recombinant polymerase chain reaction (20). One reaction utilized primers A (5Ј-ggcggccgGAATTCAGCCAAGAGACGTAAG, capital letters indicate bases pairing with the DRS2 template; the EcoRI site is in italics) and B (5Ј-AGCGTAGTCTGGGACGTCGTATGGGTAATTCATG-GTAAAATCAGGGAATGAAAGAAC, underlined letters correspond to bases encoding the 9-amino acid HA epitope); the other reaction used primers C (5Ј-TACCCATACGACGTCCCAGACTACGCTGACGACAGA-GAAACCCCCCCAAAGAGG) and D (5Ј-CTTATTCCTCGAGTCTAG-ATA, XhoI site). A mixture of the two reaction products together with primers A and D was used to amplify a recombinant EcoRI-XhoI DRS2 fragment, which carried the HA DNA inserted between the second and third codon of the DRS2 open reading frame. To obtain a full-length HA::DRS2 allele, the recombinant fragment was digested with CelII and XhoI and subcloned to replace the CelI-XhoI fragment within the 5.9 kb of wild-type DRS2 DNA (EcoRI-EcoRI) present in pUC19. For chromosomal integration of HA::DRS2, DNA from the resulting plasmid was cut with EcoRI and transformed into drs2-1::URA3 cells together with pRS425, a LEU2-containing plasmid. In one of the Leu ϩ transformants (YR886) able to grow at 23°C like wild type, expression of HA-Drs2p was verified by Western blot analysis.
Vesicle Preparation-P-C 6 -NBD-PS, M-C 6 -NBD-PE, M-C 6 -NBD-PC, DOPC and N-Rh-DOPE were from Avanti Polar Lipids Inc. (Alabaster, AL). Phospholipids were stored at Ϫ20°C, periodically monitored for purity by thin-layer chromatography, and repurified when necessary. Phospholipid concentrations were determined by a lipid phosphorus assay (21). To prepare vesicles, lipids were first mixed in desired proportions, and the chloroform solvent was removed by evaporation under nitrogen followed by overnight vacuum desiccation. Desiccated lipids were solubilized in SDC medium, and the mixture was passed eight times through a Lipex Extruder (Lipex Biomembranes Inc, Vancouver, BC, Canada) equipped with 0.1-mm filters to produce vesicles. Total lipid concentration in the stock vesicle preparation was 1 mM. Proportions were 40 mol % NBD-phospholipid, 2 mol % N-Rh-DOPE, 58 mol % DOPC.
Internalization of NBD-phospholipids into Yeast Cells-Cells were grown overnight in SDC at 30°C, diluted, and allowed to grow to an A 600 of 0.2. Donor vesicles containing 40% P-C 6 -NBD-PS, M-C 6 -NBD-PE, or M-C 6 -NBD-PC, 58% DOPC, and 2% N-Rh-DOPE (50 M total final lipid concentration) were added to the cells and incubated for 30 min. Cells were washed 3 times with ice-cold SDC/NaN 3 before analysis by fluorescence microscopy and flow cytometry.
Fluorescence Microscopy-Fluorescence microscopy was performed on a Zeiss Axiovert microscope equipped with barrier filters that allowed no detectable crossover of NBD and rhodamine fluorescence. The fluorescence image was enhanced with a VE1000-SIT image-intensifying camera (DAGE-MTI Inc., Michigan City, IN), digitized, and stored. Image manipulation and editing were performed with Metamorph software (Universal Imaging Corp., West Chester, PA).
Flow Cytometry-Flow cytometric analysis of the NBD-phospholipidlabeled cells was performed with a FACScan cytometer (Becton-Dickinson Immunocytochemistry, San Jose, CA) equipped with an argon laser operating at 488 nm. Ten l of a 50 mg/ml stock solution of propidium iodide was added to approximately 4 ϫ 10 5 cells in 200 l of SDC/NaN 3 immediately before dilution (ϳ3ϫ) and flow cytometric analysis. Ten thousand cells were analyzed without gating during acquisition. Analysis was performed with Lysis II (Becton-Dickinson Immunocytochemistry Systems) software. A dot plot of forward scatter versus the red fluorescence channel (propidium iodide) was used to set a gate that excluded dead cells from the analysis. The remaining live cells were plotted on a histogram of green fluorescence (NBD-phospholipid) and analyzed to obtain the average and standard deviation of the distribution of NBD fluorescence per cell.
TNBS Labeling-Cells were grown for 18 h at 30°C to mid-logarithmic phase (A 600 ϭ 0.4 -0.7) in SDC containing 500 Ci of [ 32 P]KH 2 PO 4 (specific activity 1 Ci/mmol; NEN Life Science Products). Plasma membrane, outer leaflet PE was labeled with TNBS as described previously (22). Cells were harvested by centrifugation, washed twice in ice-cold 40 mM NaCl, 120 mM NaHCO 3 , pH 8.4, resuspended in the same buffer containing 5 mM TNBS (Sigma), and immediately placed on ice for 1 h with periodic vortex mixing. After TNBS labeling, cells were washed by centrifugation three times in fresh buffer, pelleted, and disrupted by vortexing with glass beads. Cellular lipids were extracted with chloroform/methanol (2:1) and separated by two-dimensional TLC (1 st solvent: chloroform/methanol/ammonium hydroxide (65:35:5); 2 nd solvent: chloroform/methanol/acetone/acetic acid/water (50:10:20:10:5)). Spots were identified by comparison with known standards. The percentages of PE, PS, trinitrophenylnucleotide-PE and trinitrophenylnucleotide-PS were quantified by phosphorimaging with a PhosphorImager SI scanning instrument (Molecular Dynamics, Sunnyvale, CA). The percent viability was determined by counting the number of cells labeled after dilution into 0.02% methylene blue with a hemocytometer and did not differ significantly between the two strains. The data are presented in the text as the average percentage of trinitrophenylnucleotide-PE or -PS relative to total cellular PE or PS ϮS.D. for four trials.

RESULTS
Construction of the drs2-1::URA3 Null Allele-Using a drs2::TRP1 mutant allele, wherein the 1.4-kb BglII-NcoI fragment corresponding to the segment from positions 529 to 1201 within the DRS2 open reading frame (1355 amino acids) was replaced by the yeast TRP1 gene (see Fig. 1), Ripmaster et al. (23) show that DRS2 is required for mitotic growth at 23°C or below. Because these authors also reported that a 2.2-kb EcoRI-BglII fragment encoding the first 528 amino acids at the N terminus of Drs2p was still able to complement a coldsensitive drs2 mutant, we constructed a new drs2 mutant allele  (16,19) in HA::DRS2 is located between the second and third codon of the DRS2 coding sequence is indicated by an asterisk. essentially lacking the entire DRS2 coding sequence. In this drs2-1::URA3 null allele, the 3.5-kb BstEII-HpaI fragment encoding amino acids 33 to 1206 of the DRS2 open reading frame was substituted by the yeast URA3 gene. Transformation of a haploid wild-type strain (YR98) with drs2-1::URA3 DNA produced the congenic drs2-1::URA3 strain YR884 used in all subsequent studies. Growth of YR884 was examined on different solid media (yeast extract/peptone/glucose (YPD)), synthetic complete and minimal medium) at 23, 30, and 37°C. On all media tested, this drs2-1::URA3 strain grew well at 37 and 30°C but was unable to grow at 23°C (data not shown), as reported for the drs2::TRP1 mutant (23).
Loss of DRS2 Does Not Alter the Internalization and Distribution of NBD-labeled Aminophospholipids Detected by Fluorescence Microscopy-The internalization and distribution of the NBD-labeled aminophospholipids, P-C 6 -NBD-PS and M-C 6 -NBD-PE, were observed by fluorescence microscopy in the drs2-1::URA3 strain and its isogenic DRS2 parent (Fig. 2). No significant differences were detected between the two strains at 30°C (permissive for growth) or at 23°C (nonpermissive for growth). In previous experiments, it was concluded that M-C 6 -NBD-PE was internalized exclusively by inwarddirected transport across the plasma membrane (flip), resulting in its distribution to the nuclear envelope, endoplasmic reticulum, and mitochondria (24). Internalized M-C 6 -NBD-PE was not degraded intracellularly, but was readily transported outward across the plasma membrane (flop), where it was degraded by periplasmic phospholipases (24). The similar pattern of fluorescence distribution observed for P-C 6 -NBD-PS and M-C 6 -NBD-PE (Fig. 2) suggests that both of these aminophospholipids are internalized and distributed by similar mechanisms in both the drs2-1::URA3 and DRS2 strains.
To address the possibility that P-C 6 -NBD-PS was internalized by endocytosis, Tang et al. (12) labeled drs2⌬ cells with high concentrations of probe solubilized in Me 2 SO incubated on ice to inhibit endocytosis (12). Using our standard protocol to label cells on ice with P-C 6 -NBD-PS and M-C 6 -NBD-PE incorporated into liposomes resulted in no detectable fluorescence internalization using the sensitive SIT camera to capture images on the fluorescence microscope. We therefore followed the Tang et al. protocol for labeling cells on ice using high concentrations of Me 2 SO-solubilized NBD-lipids. Following this labeling protocol, no P-C 6 -NBD-PS fluorescence could be detected. However, a very low level of M-C 6 -NBD-PE fluorescence was detected, but no differences were observed between the drs2-1::URA3 and DRS2 strains. Although detectable with the SIT camera, the fluorescence was too faint and diffuse to produce publishable images.
Thus, for all labeling conditions in which detectable amounts of NBD-aminophospholipids were obtained, the loss of Drs2p had no effect on their internalization and distribution. These observations are inconsistent with the previous study in which P-C 6 -NBD-PS internalization was abolished in a drs2 mutant strain (12). The previous conclusion by Tang et al. (12) about the function of Drs2p was based on "back exchange" experiments and was not confirmed by direct observation of internalization and distribution by fluorescence microscopy. In the back exchange measurement, inward-directed transport (flip) is inferred from the amount of NBD-phospholipid that cannot be extracted from the surface of labeled cells by incubation with bovine serum albumin. This technique has been used successfully for many years to assay NBD-phospholipid transport in blood cells and reconstituted vesicles (1,25). However, the report by Tang et al. (12) was the first use of back exchange to measure flip in yeast, and in the absence of proper controls, differences in the amount of NBD-phospholipid aggregates sticking to the cell wall or trapped in the periplasm or differences in the rate of NBD-phospholipid hydrolysis by periplasmic phospholipases may have been misinterpreted as differences in inward translocation. Direct observation of NBDphospholipid internalization by fluorescence microscopy is not subject to these artifacts. One of the advantages of labeling cells with liposomes containing trace amounts of N-Rh-DOPC is that the rhodamine fluorescence can be used to determine the extent of cell-associated NBD fluorescence resulting from stuck vesicles (26,24). This is not possible when cells are labeled with Me 2 SO-solubilized NBD-phospholipids.
Quantification of NBD-labeled Phospholipid Accumulation by Flow Cytometry-To make a more quantitative evaluation of the results obtained by fluorescence microscopy, cells were labeled with either P-C 6 -NBD-PS, M-C 6 -NBD-PE, or M-C 6 -NBD-PC, and the average cell-associated NBD fluorescence per cell was obtained by flow cytometry. Fluorescent lipid accumulation was measured for drs2-1::URA3 and DRS2 strains grown at 30°C in two different media (YPAD and SDC). Accumulation of the three NBD-phospholipids was compared at 30°a nd 23°C, the nonpermissive growth temperature for drs2-1::URA3. The ratio of the cell-associated fluorescence in drs2-1::URA3 to that of DRS2 is presented in Table I for the two temperatures and growth media. Similar results were ob-FIG. 2. Internalization and distribution of P-C 6 -NBD-PS and M-C 6 -NBD-PE in drs2-1::URA3 and DRS2 strains. Cells were grown to early log phase in SDC, washed three times in SDC, and incubated for 30 min at 23 or 30°C with vesicles containing either P-C 6 -NBD-PS or M-C 6 -NBD-PE and N-Rh-DOPC and DOPC (40:2:58; molar ratio). Cells were washed three times in icecold SCN 3 before imaging by fluorescence microscopy as described under "Experimental Procedures." DIC, differentiol interference contrast.
tained for the two strains and growth media at 30 and 23°C. However, the results differed dramatically depending on the growth media. For cells grown in YPAD, the NBD fluorescence in the null strain versus its parent was decreased from ϳ30 and 50% for the two NBD-aminophospholipids as well as the choline lipid, M-C 6 -NBD-PC. On the other hand, for cells grown in SDC, the null strains actually accumulated ϳ16% more P-C 6 -NBD-PS but ϳ30% less M-C 6 -NBD-PE. The reduction in P-C 6 -NBD-PS and M-C 6 -NBD-PE internalization observed for cells grown in YPAD is consistent with the interpretation that Drs2p is an aminophospholipid translocase responsible for 30 to 50% that of the internalization measured in the parent strain. However, given the similar reduction in M-C 6 -NBD-PC internalization, one would have to surmise that Drs2p was not functionally homologous to its mammalian counterpart in its ability to discriminate between amino and choline head groups. On the other hand, the observation that P-C 6 -NBD-PS internalization is actually increased in cells lacking Drs2p grown in SDC does not support the conclusion that Drs2p is an aminophospholipid translocase. It is conceivable that drs2-1::URA3 cells grown in SDC up-regulate a functional homologue that overcompensates for P-C 6 -NBD-PS but not for M-C 6 -NBD-PE internalization. A more likely interpretation is that the absence of Drs2p indirectly alters the NBD-phospholipid internalization.
The average cell-associated fluorescence resulting from identical labeling procedures is highly dependent on the strain and the stage of growth. In unpublished experiments, 2 M-C 6 -NBD-PE accumulation varied as much as 10-fold between different wild-type laboratory strains. Furthermore, its accumulation was dramatically reduced in cells allowed to grow to mid-log (optical density Ͼ1.0). 3 Thus, isogenic backgrounds and identical growth conditions are essential for meaningful comparisons between strains to be made. The existence of unidentified regulatory mechanisms that control the extent of NBDphospholipid internalization provides a means by which the loss of DRS2 could indirectly decrease NBD-lipid internalization.

Loss of Drs2p Does Not Alter the Amount of Endogenous Aminophospholipid in the Outer Leaflet of the Plasma Membrane-
The amount of endogenous PS and PE in the outer leaflet of the plasma membrane of drs2-1::URA3 and DRS2 strains was measured by TNBS labeling. No significant differences were observed in the percent of total PS or PE exposed to the outer leaflet between the two strains. Very little (Ͻ0.5%) of the total cellular PS was exposed to the outer leaflet, making it difficult to detect; however, no measurable differences were observed between the two strains. A readily detectable and reproducible amount of PE was labeled by TNBS. The percentage of total cellular PE exposed to the outer leaflet of the plasma membrane was 1.4 Ϯ 0.3 for the drs2-1::URA3 strain and 1.6 Ϯ 0.4 for the isogenic DRS2 parent strain. Thus, the loss of Drs2p had no significant effect on the amount of PE or PS residing in the outer leaflet of the plasma membrane.
Drs2p Localizes to the Plasma Membrane-To facilitate detection of the Drs2p protein, we constructed a derivative of the DRS2 gene, HA::DRS2, harboring a 27-base pair sequence encoding the nine amino acid HA epitope (16,19) inserted in-frame after the second DRS2 codon (see Fig. 1). Haploid strain YR886 containing HA::DRS2 at the chromosomal DRS2 locus grew indistinguishable from DRS2 strains, indicating that the expressed HA-Drs2p protein was fully functional (data not shown). To determine the subcellular localization of HA-Drs2p, extracts from strain YR886 (HA::DRS2) were fractionated by sucrose gradient centrifugation. All fractions collected from the gradients were tested by SDS-polyacrylamide gel electrophoresis and Western blotting for the presence of marker proteins specific for plasma membrane (H ϩ -ATPase Pma1 (27)) and endoplasmic reticulum (dolichol phosphate mannose synthase Dpm1 (28 -30)). Golgi and vacuolar membranes were identified by monitoring activities of GDPase (31) and ␣-mannosidase, Ams1 (32), respectively. As demonstrated by the data shown in Fig. 3, the bulk of HA-Drs2p co-fractionated with the plasma membrane ATPase, well separated from endoplasmic reticulum membranes and the bulk of GDPase or ␣-mannosidase activity. Consistent with this observations, cells expressing HA-Drs2p from the chromosomal locus exhibited in indirect immunofluorescence microscopy a ring-shaped rim staining pattern, but unfortunately the signal was very low (data not shown). Attempts to increase expression of HA-Drs2p from a strain relative to isogenic parent Fluorescence accumulation was measured by flow cytometry. A fluorescence intensity histogram was plotted for ϳ10,000 live cells, and the means and S.D. were calculated for each strain under the appropriate conditions. The percent average accumulation for the ⌬drs2 strain relative to its isogenic parent is presented ϮS.D. The number of independent experiments (n) is in parenthesis. For those experiments with an n of 2, the mean Ϯ the range is presented. NBD a Cells were grown to early log phase in either YPAD or SDC at 30°C. b Cells were labeled by incubation with donor vesicles containing the appropriate NBD-phospholipid for 30 min at the stated temperature before washing 3 times with SDC/NaN 3 .

FIG. 3. Fractionation of HA-Drs2p on sucrose gradients.
Whole cell extracts of the HA::DRS2 strain YR886 were fractionated by density centrifugation as described (38). Aliquots of the gradient fractions were separated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting with anti-HA antibodies, anti-Pma1 antibodies, and anti-Dpm1 antibodies. The sizes of marker proteins are given in kDa; the fraction numbers are indicated. Activities for ␣-mannosidase (Ams1) (32) and guanosine diphosphatase (31) were determined as described and are given in arbitrary units. Density (% sucrose, w/w) and protein concentration (arbitrary units) are plotted against the fraction number.
multicopy plasmid yielded a more pronounced rim staining but, in addition, produced a prominent perinuclear staining pattern, suggesting that the bulk of HA-Drs2p remained in the endoplasmic reticulum under these conditions (Fig. 4). Taken together, our data strongly argue for a steady-state localization of Drs2p in the plasma membrane but indicate that Drs2p could also exert functions in membranes of secretory organelles along the pathway to the plasma membrane.
The Steady-state Level of Drs2p Is Dependent on Growth Phase and Temperature-In monitoring expression of HA-Drs2p in HA::DRS2 cells growing in YPD medium at 30°C, we found a substantial alteration of the steady-state level of HA-Drs2p with the growth phase of the culture. Fig. 5A (top panel) shows a Western blot analysis of crude membranes prepared from HA::DRS2 cells at various stages of growth (A 600 of 0.6, 1.1, and 2.0). HA-Drs2p was fairly abundant during early and mid logarithmic growth but essentially disappeared as cells entered stationary phase. In contrast, the steady-state level of the plasma membrane H ϩ -ATPase (Pma1p), which like Drs2p, is a member of the P-type ATPase family, did not show such a dramatic decrease, although Pma1p appeared somewhat reduced in cells approaching stationary phase (Fig. 5A, bottom  panel). However, as demonstrated in Fig. 5B, cells entering stationary phase while growing at 23°C retained a substantial amount of HA-Drs2p. This finding was consistent with the observed requirement of DRS2 for growth at this temperature (23). Taken together, these data suggested an important function for Drs2p, particularly during early growth phases at 30°C and during growth at lower temperatures.
The Steady-state Level of Drs2p Increases upon Treatment with Heavy Metals-In the course of our genetic analysis, we discovered a hypersensitivity of drs2-1::URA3 mutants toward various heavy metals. As displayed in Fig. 6, serial dilutions of a drs2-1::URA3 culture were spotted onto solid minimal media containing the indicated concentrations of manganese (II), cobalt (II), nickel (II), or zinc (II) chloride. Evidently, the drs2-1::URA3 strain was particularly sensitive to Zn 2ϩ and Co 2ϩ , and the hypersensitivity toward Mn 2ϩ and Ni 2ϩ was less pronounced. This growth inhibition by divalent cations was apparently restricted to transition elements, because drs2-1::URA3 grew like wild type in the presence of high concentrations of Mg 2ϩ and Ca 2ϩ (see Fig. 6).
We also examined the steady-state levels of HA-Drs2p accumulating in cells during growth in the presence of various metal cations. To this end, HA::DRS2 cells were inoculated into YPD media to which the indicated amounts of metal chlorides have been added. At an A 600 of 0.6, 1.1, and 2.0, aliquots were withdrawn and analyzed for the presence of HA-Drs2p in crude membrane fractions. As seen in Fig. 7, all cations inhibitory to growth, i.e. Co 2ϩ , Ni 2ϩ , Mn 2ϩ , and Zn 2ϩ , also led to a pronounced accumulation of HA-Drs2p relative to the control culture. In contrast, the presence of Mg 2ϩ , Ca 2ϩ , or monovalent cations had only slight effects on the steady-state level of HA-Drs2p. Both observations, the hypersensitivity of drs2-1::URA3 cells to heavy metals and the intracellular accumulation of HA-Drs2p triggered by the same divalent cations, indicated that cells required Drs2p function to effectively endure the toxic effects of these transition elements. DISCUSSION The current knowledge about the mechanisms leading to internalization of phospholipids into the yeast S. cerevisiae stems primarily from studies utilizing phospholipid molecules carrying one short acyl chain labeled with a fluorescent NBD group. Digital, video-enhanced fluorescence microscopy and spectrofluorometry have shown that at least two distinct pathways for phospholipid internalization exist: 1) transport by endocytosis to the vacuole, which partially accounts for the uptake of the NBD-labeled PC analog, M-C 6 -NBD-PC (26) and 2) transport by a nonendocytic pathway to the nuclear envelope and mitochondria. This route accounts for the remainder of the M-C 6 -NBD-PC internalization, whereas the NBD-labeled analogs of the aminophospholipids, M-C6-NBD-PE (24) or P-C6-NBD-PS (33), appear to be exclusively internalized via this pathway. Based on the inhibition of M-C6-NBD-PE uptake by treatment with NEM at low temperature, this pathway is thought to involve a protein-mediated translocation of phospholipids from the outer to the inner leaflet of the plasma membrane (24). Because these NBD-labeled aminophospholipids are more water-soluble than their endogenous counterparts (34), they most likely spontaneously redistribute between the inner leaflet of the plasma membrane and membranes of other intracellular organelles. Because cells are also able to translocate M-C6-NBD-PE outward across the plasma membrane (24), the extent of intracellular accumulation depends on the steadystate distribution of the NBD-labeled phospholipid established between the inner and outer leaflets of the plasma membrane by the coordinate regulation of the influx and efflux pathways. As a result, the extent of intracellular accumulation can be considered to be an amplification of the steady-state amount of the NBD-labeled aminophospholipid residing in the inner leaflet that is established by the regulation of the influx and efflux pathways. This view is corroborated by the analysis of pdr1-11 and pdr3-11 mutant strains, in which the net influx of M-C 6 -NBD-PE is reduced from 50-to 100-fold. These strains also exhibited a three to four-fold increase in the amount of endogenous PE residing in the outer leaflet of the plasma membrane (24). In this case, a decrease in the intracellular accumulation of M-C 6 -NBD-PE correlated with an increase in the amount of endogenous PE residing in the outer leaflet of the plasma membrane.
Given the identification of DRS2 as the exclusive or at least major functional aminophospholipid transporter in the plasma membrane of S. cerevisiae (12), we compared the intracellular accumulation of P-C 6 -NBD-PS and M-C 6 -NBD-PE as well as the percent of TNBS-labeled PS and PE in the outer leaflet of the plasma membrane between a drs2-1::URA3 mutant strain and its isogenic DRS2 parent. The deletion of the DRS2 gene had no effect on the internalization and distribution of P-C 6 -NBD-PS and M-C 6 -NBD-PE detected by fluorescence imaging, nor did the deletion have a measurable effect on the distribution of endogenous PS and PE across the plasma membrane. Quantitation of the extent of accumulation by a large population of cells (ϳ10,000 live cells) indicated that the deletion of the DRS2 gene resulted in a significant decrease in P-C 6 -NBD-PS, M-C 6 -NBD-PE, and M-C 6 -NBD-PC accumulation following growth in YPAD at 30 and 23°C. However, the decrease of P-C 6 -NBD-PS accumulation was completely reversed in cells grown in SDC. The mean accumulation of M-C 6 -NBD-PE was decreased in drs2-1::URA3 cells grown in both media. Thus, the observed decrease in NBD-phospholipid internalization in the drs2 null strain was dependent on the growth media and was not exclusive to the aminophospholipids.
These data definitively exclude the possibility that Drs2p is the exclusive or major aminophospholipid translocase in the plasma membrane of S. cerevisiae. However, these negative results do not exclude the possibility that Drs2p functions as a phospholipid translocase in intracellular membranes or is one of several plasma membrane aminophospholipid translocases. In the latter case, the partial loss of aminophospholipid internalization in the drs2 strain would be compensated by additional transporters under the appropriate growth conditions. Interestingly, DRS2 and four yeast homologs form a distinct subgroup, cluster II, within the P-type ATPase family (13). These four homologs have no known function and are potential candidates to code for additional aminophospholipid transporters. However, because of the complex nature of the regulation of aminophospholipid transport across the plasma membrane and its dependence on growth conditions, one cannot exclude from the data presented that the observed alterations in P-C 6 -NBD-PS and M-C 6 -NBD-PE accumulation in the drs2-1::URA3 null strain are secondary to other effects on growth or some other cellular function.
At present, the mechanisms leading to the observed hypersensitivity of a drs2 null mutant toward some heavy metal cations, in particular Zn 2ϩ , Co 2ϩ , Mn 2ϩ , and Ni 2ϩ , are not known. In our view, these phenotypes could either reflect a yet undiscovered role of Drs2p as a transporter of divalent cations or are indeed a consequence of the proposed function of Drs2p in aminophospholipid translocation. Two models, not necessarily exclusive, could provide an explanation for a link between aminophospholipid translocation and cation sensitivity. (i) An altered asymmetric distribution of lipids within some cellular membranes, presumably resulting from the loss of Drs2p, could compromise the activity of cation transporters embedded in the affected membranes. Given a prominent localization of Drs2p in the plasma membrane and the numerous exocytic and endocytic pathways leading to and originating from this mem- brane, it seems possible that Drs2p function might impinge upon Golgi and vacuolar membranes, which both harbor numerous transport proteins engaged in the sequestration of divalent cations within these organelles. (ii) Alternatively, an altered lipid distribution, i.e. a reduction in the amount of negatively charged PS in the cytoplasmic leaflets of some intracellular membranes, might directly result in altered cation binding properties of these membranes, thereby affecting some cation-dependent steps in membrane fusion reactions. Noteworthy, in vitro studies have shown that Zn 2ϩ ions are more effective than Ca 2ϩ to induce fusion of phospholipid vesicles with a low content of PS (35), and the depletion of Zn 2ϩ blocks endosome fusion in a cell-free system (36).
It should be noted that the loss of Drs2p function also impairs ribosome biogenesis. The DRS2 gene was first discovered in a search for mutants with an altered ratio of free 40 to 60 S ribosomal subunits or qualitative changes in polyribosome profiles. The drs2 mutant isolated in this screen processes the 20 S precursor of the mature 18 S rRNA slowly and is deficient in 40 S ribosomal subunits (23). It has been demonstrated that a block in the secretory pathway, which can readily be introduced at different stages of the pathway through the use of conditional alleles in various SEC genes absolutely required for secretion, leads to the rapid shut-down of ribosome biogenesis. Thus, the continuous functioning of the secretory pathway appears to be a prerequisite for the biogenesis of ribosomes (37). Evidently, the proposed role for Drs2p as a flippase affecting lipid distribution in vesicles of the late secretory pathway or at the plasma membrane would provide an intriguing explanation for the ribosome-related defects observed in the drs2 mutant (23).
Finally, we would like to emphasize that the previous demonstration of the abolishment of PS internalization in a yeast drs2 mutant (12) provided the sole functional data for the assignment of phospholipid translocase activity to the Drs2p subfamily within the class of P2 ATPases (13). Our inability to confirm a role of Drs2p in phospholipid translocation underscores the need for future experiments to carefully reevaluate this functional assignment.