ATPase and Multidrug Transport Activities of the Overexpressed Yeast ABC Protein Yor1p*

The Saccharomyces cerevisiae genome encodes 15 full-size ATPbinding cassette transporters (ABC), of whichPDR5, SNQ2, and YOR1 are known to be regulated by the transcription factors Pdr1p and Pdr3p (pleiotropic drug resistance). We have identified two new ABC transporter-encoding genes,PDR10 and PDR15, which were up-regulated by thePDR1–3 mutation. These genes, as well as four other ABC transporter-encoding genes, were deleted in order to study the properties of Yor1p. The PDR1–3 gain-of-function mutant was then used to overproduce Yor1p up to 10% of the total plasma membrane proteins. Overexpressed Yor1p was photolabeled by [γ-32P]2′,3′-O-(2,4,6-trinitrophenyl)-8-azido-ATP (K 0.5 = 45 μm) and inhibited by ATP (K D = 0.3 mm) in plasma membranes. Solubilization and partial purification on sucrose gradient allowed to detect significant Yor1p ATP hydrolysis activity (∼100 nmol of Pi·min−1·mg−1). This activity was phospholipid-dependent and sensitive to low concentrations of vanadate (I50 = 0.3 μm) and oligomycin (I50 = 8.5 μg/ml). In vivo, we observed a correlation between the amount of Yor1p in the plasma membrane and the level of resistance to oligomycin. We also demonstrated that Yor1p drives an energy-dependent, proton uncoupler-insensitive, cellular extrusion of rhodamine B. Furthermore, cells lacking both Yor1p and Pdr5p (but not Snq2p) showed increased accumulation of the fluorescent derivative of 1-myristoyl-2-[6-(NBD)aminocaproyl]phosphatidylethanolamine. Despite their different topologies, both Yor1p and Pdr5p mediated the ATP-dependent translocation of similar drugs and phospholipids across the yeast cell membrane. Both ABC transporters exhibit ATP hydrolysis in vitro, but Pdr5p ATPase activity is about 15 times higher than that of Yor1p, which may indicate mechanistic or regulatory differences between the two enzymes.

The yeast YOR1 1 gene confers oligomycin resistance on overexpression in a 2-m plasmid (1). Its nucleotide sequence reveals an ORF of 1477 amino acids encoding an ABC protein highly homologous to mammalian transporters such as the multidrug resistance-conferring enzyme MRP (BLAST (see Ref. 2) sequence homology score: p ϭ e Ϫ228 ), the organic anion transporter cMOAT (p ϭ e Ϫ216 ), the sulfonylurea receptor (p ϭ e Ϫ164 ), and the cystic fibrosis transmembrane conductance regulator CFTR (p ϭ e Ϫ132 ). Yor1p is a "full-size" ABC transporter with the topology (TM-NBF) 2 (3,4). It consists of two homologous halves, with each containing a putative ATP-binding domain (NBF) and a transmembrane domain of six membrane spans (TM). Cui et al. (5) showed that Yor1p confers resistance to a series of drugs including reveromycin A and suggested that Yor1p may be involved in the cellular efflux of organic anions including the fluorescent dye rhodamine B. They also showed that incubation with reveromycin A increases the YOR1 mRNA level. The transcription of YOR1 is controlled by the homologous pair of transcription factors Pdr1p/Pdr3p. The level of YOR1 transcription is decreased by the deletion of either PDR1 or PDR3 and increased in the presence of the gain-of-function PDR1 alleles (1).
In this paper, we have investigated the transport activity of Yor1p. Building on previous studies, which indicated that the (TM-NBF) 2 -type Yor1p, together with the (NBF-TM) 2 -type Pdr5p and Snq2p ABC transporters, are overexpressed in the PDR1-3 mutant plasma membrane (6 -8), the PDR1-3 mutant has been used as a tool that enhances the Yor1p protein level. As another investigative tool, we constructed a set of isogenic strains, in the PDR1-3 mutant, with multiple deletions of homologous ABC genes since, in situations where two or more proteins located in the same subcellular compartment share a common substrate, a clear phenotype is only seen when all the corresponding genes are deleted, as illustrated by the work of Mahé et al. (9), who showed that Pdr5p and Snq2p have an overlapping transport capacity for steroids. We deleted the yeast ABC transporter-encoding genes known or suspected to * This work was supported by grants from the Service des Affaires Scientifiques, Techniques et Culturelles, Pôles d'Attraction Interuniversitaires, the Fonds National de la Recherche Scientifique, and the Foundation for Research Development of South Africa, and by NATO Collaborative Exchange Research Grant CRG940493 and National Institutes of Health Grant GM52410 (to J. W. N.) and a National Institutes of Health minority predoctoral fellowship (to A. M. G.). 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.
ʈ To whom correspondence and reprint requests should be addressed. Tel.: 32-10-473614; Fax: 32-10-473872; E-mail: goffeau@fysa.ucl.ac.be. be controlled by the transcription factors Pdr1p and Pdr3p. The YCF1 gene, which encodes a glutathione S-conjugate pump (10), was also deleted. The multiply deleted mutants have allowed the demonstration that Yor1p and Pdr5p share several substrates, which include fluorescent phosphatidylethanolamine, rhodamine B, and oligomycin, even though previous studies had concluded that Pdr5p was not involved in oligomycin resistance (11,12). The pumping of phospholipids is in line with reports of "flippase" activity with several human and mouse ABC transporters (13)(14)(15)(16). It is also in agreement with the defective phospholipid accumulation of two new mutant yeast alleles, PDR1-11 and pdr3-11 (17). Construction of a stronger Yor1p-overexpressing strain allowed us to detect vanadate-and oligomycin-sensitive ATPase activity associated with Yor1p while no UTPase activity was detectable. Despite similar nucleotide binding specificities and transport capacities for Yor1p and Pdr5p, the Yor1p enzyme showed 15 times lower levels of ATP hydrolysis rate than Pdr5p.
Yeast Strains-The Saccharomyces cerevisiae strains used in this study are listed in Table I. Multiple deletions were performed sequentially in the US50 -18C PDR1-3 strain by repeated use of the hisG-URA3-hisG cassette followed by selection of the ura3 auxotrophic marker with 5-fluoroorotic acid (20). The plasmids for the deletion of PDR5 (12), SNQ2, and YOR1 (1) genes were kindly provided by W. S. Moye-Rowley (Department of Physiology and Biophysics, University of Iowa, Iowa City, IA). For the deletion of PDR10, PDR11, PDR15, PDR3, and YCF1 genes, we amplified fragments of the promoter and the ORF 3Ј-end of each gene (Table II). The gene promoters were cloned into the EcoRI/BamHI sites of pSK, the ORF ends were cloned into the BamHI/ XbaI sites (except for the PDR11 gene; see Table II), and the BamHI/ BglII hisG-URA3-hisG cassette was cloned into the BamHI site.
Linearized fragments of the plasmids were used to transform the yeast for deletion of PDR5 (BamHI, SalI), SNQ2 (SstI, SalI), YOR1 (SacII/BamHI), PDR10 (EcoRI/NotI), PDR11 (KpnI/NotI), PDR15 (EcoRI/NotI), YCF1 (EcoRI/NotI), and PDR3 (EcoRI/NotI). The deletion of PDR5, SNQ2, and YOR1 genes was checked by Southern blotting analysis. Deletion of the other genes was checked by PCR screening: yeast cells from a 1.5-ml overnight rich glucose medium (YD: 2% yeast extract, 5.8% glucose) culture were washed, resuspended in 200 l of 10 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 1 mM EDTA and broken by vortexing for 2 min with glass beads and 300 l of phenol-chloroform (50:50). After a second phenol-chloroform extraction and a washing with 1 volume of ether, 2 l of cell extract were used for the PCR analysis. Recovery of the ura3 marker was monitored by plating onto 5-fluoroorotic acid 10 7 cells of an overnight YD culture. The resistant cells were screened by PCR as described above. The deletions of PDR5, SNQ2, and YOR1 were monitored by Southern analysis.
The SUPERYOR strain was constructed as follows: 1129 bp of the PDR5 promoter (from position Ϫ1141 to Ϫ12) and 505 bp of the PDR5 ORF end (from ϩ4026 to ϩ4531) were amplified by PCR using the following primer sequences: 5Ј-CCATCGATGGTCCGTCATATACG and 5Ј-CCCCCGGGTCTTTCGAACGAGCG (promoter) and Ј-CCCCCGGG-TCTGCTTGTCATTTC and 5Ј-GCTCTAGACTTGGAGAGTTTACC (ORF end) to which restriction sites had been added (underlined sequences). The PCR fragments were cut by either ClaI and SmaI (PDR5 promoter) or XbaI and SmaI (PDR5 end) and cloned into ClaI-XbaI-cleaved pSK (pSK::PDR5PROM-PDR5STOP). The YOR1 DNA was prepared as follows. the pEGH452 cosmid (kind gift of Hervé Tettelin) containing 35.4 kb of S. cerevisiae chromosome VII was cut by MluI. The 5.7-kb fragment, containing the YOR1 ORF, was blunted with Klenow Transcriptional Activity of the Yeast ABC Genes-Transcriptional activity of the yeast ABC genes was measured after transformation of the FY1679 -28C/EC (pdr1⌬) strain with 2 centromeric plasmids. The first plasmid carries either no allele (pRS315), or the wild type (pRS315::PDR1), or the mutant allele (pRS315::PDR1-3) of the PDR1 gene as described previously (21). The second plasmid (pSEYC102) bears a translational fusion of the ABC gene promoters and the Escherichia coli lacZ gene. The PDR5-lacZ fusion (12), the SNQ2-lacZ fusion (8), and the YOR1-lacZ construct (1) have been described previously. For the construction of the PDR11-lacZ fusion, 687 bp of the PDR11 gene, including 633 bp of the promoter, were amplified by PCR using the primers 5Ј-CGGGATCCCGATCAAAGGTGACTGAAGC and 5Ј-GGAATTCCACTTTGACGCCCCTTTATGC (restriction sites are underlined). The PCR-amplified promoter was cloned into a BamHI-EcoRIcleaved pSEYC102 vector. The same strategy was used to clone 680 bp of the PDR10 promoter using the primer sequences: 5Ј-GGAATTCCT-GCCTGACTTACAGATAC and 5Ј-CGGGATCCCACATCCTAACAAC-TATG and 952 bp of the PDR15 gene, including 897 bp of the promoter, with the primer sequences: 5Ј-GGAATTCCGCCCAGCCTTTTATACCT and 5Ј-CGGGATCCCTTGAGCTCGAGCTCCG. ␤-Galactosidase activity was measured as described by Sambrook et al. (22) on cell extracts from yeast grown in synthetic complete glucose medium (SDC: 0.7% yeast nitrogen base, 2% glucose, complete amino acid supplements) lacking uracil and leucine.
Complementation of Deleted yor1 Strains by the YOR1 Gene on a Centromeric Plasmid-The pEGH452 cosmid was cut by MluI. The 5.7-kb restriction fragment containing YOR1, 0.4 kb of the gene promoter including the Pdr1p/Pdr3p binding site, and 0.9 kb downstream of the STOP codon was blunted with Klenow, cloned into the SmaI site of pRS316 (pRS316::YOR1) and used to transform the AD13 and AD1234568 strains for testing complementation of oligomycin and rhodamine B resistance on plates.
Isolation of Plasma Membranes-Plasma membranes were isolated from the particulate fraction pelleted at 15,000 ϫ g for 40 min after selective precipitation of mitochondria at pH 5.2 (23).
Solubilization of Plasma Membrane Proteins and Centrifugation on a Continuous Sucrose Gradient-AD1234567 and SUPERYOR plasma membrane proteins (5 mg/ml) were solubilized with 0.2% (w/v) n-dodecyl-␤-D-maltoside in the presence of 8 mM Tris-HCl, pH 7.5 (7) and the solubilized proteins (7 mg) were separated on a 33.3-ml linear sucrose gradient as described previously (8).
Nucleoside Triphosphatase Assays-Nucleotide hydrolysis of plasma membrane-enriched fractions was measured by incubation at 35°C in a final volume of 100 l containing 6 mM NTP, 7 mM MgSO 4 , 10 mM NaN 3 , 50 mM MES, 50 mM MOPS, and 50 mM Tris, adjusted to the right pH with either HCl or NaOH. In the sucrose gradient fractions, the NTP hydrolysis was measured in the presence of 6 mM NTP, 7 mM MgCl 2 , 50 mM MES, 50 mM MOPS, 50 mM Tris (pH adjusted with either HCl or NaOH) and 150 g/ml asolectin. Assays were carried out as described previously (7).
Drug Resistance Assays-The strains were tested for drug resistance on solid medium containing 1% yeast extract (Difco) and either 2% glucose and rhodamine B, or 4% glycerol and oligomycin dissolved in ethanol. Drug resistance assays after yeast transformation with the YOR1 gene-containing plasmid were performed on solid synthetic media containing either 2% glucose or 4% glycerol plus 0.7% yeast nitrogen base supplemented with amino acids lacking uracil. A 32-well replicator was used for plating and drug resistance was scored after 3-4 days at 30°C as described previously (24,25).
Rhodamine B Fluorescence Measurements in Intact Cells-Four-ml YD cultures were inoculated with ϳ50 ϫ 10 6 cells from an overnight preculture and incubated for 3 h at 30°C. Culture aliquots of 750 l (ϳ40 ϫ 10 6 cells/ml) were washed three times with buffer A (50 mM Hepes-NaOH, pH 7.0), resuspended in 2 ml of buffer A containing 100 g/ml rhodamine B and either 5 mM D-glucose or 5 mM 2-deoxy-Dglucose and incubated for 2 h at 28-30°C. A 1.5-ml sample was pelleted and washed three times with buffer A. The cell pellet resuspended in 800 l of water was maintained on ice until cell fluorescence was measured using an SLM Aminco 48000 S spectrofluorimeter. The excitation wavelength was 555 nm (slit of 4 nm), and the emission wavelength was 575 nm (slit of 4 nm).
In rhodamine B extrusion experiments, cells from 3.5 ml of YD culture (ϳ40 ϫ 10 6 cells/ml) were washed three times with buffer A, incubated in 2 ml of buffer A containing 100 g/ml rhodamine B and 5 mM 2-deoxy-D-glucose for 2 h at 28-30°C and then washed three times with buffer A and resuspended in 1.5 ml of buffer A. Rhodamine B extrusion was measured in response to either 10 mM D-glucose, 4% ethanol, or no carbon source (control). At the indicated times, the fluorescence of 300 l of both the cell-free supernatant and the cell pellet were measured.
Yeast Cell Labeling with NBD-phosphatidylethanolamine-Lipid vesicles including 50 M total lipids comprising M-C 6 -NBD-PE (40 mol%), dioleoylphosphatidylcholine (58 mol%), and N-rhodamine-dioleoylphosphatidylethanolamine (2 mol%) were prepared as described previously (17). Phospholipid concentrations were determined by the lipid phosphorus assay (26). For internalization of M-C 6 -NBD-PE, yeast cells were grown overnight in SDC at 30°C, diluted, and allowed to grow to an A 600 of 0.2-0.3. Donor vesicles containing the fluorescent lipids were added to the yeast cells and incubated for 30 min at 37°C. Cells were washed three times with ice-cold SCNaN 3 (SDC lacking glucose but containing 2% sorbitol and 20 mM sodium azide) prior to 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-intensify- Flow Cytometry-Flow cytometric analysis of the M-C 6 -NBD-PE labeled 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 g/ml stock solution of propidium iodide was added to approximately 4 ϫ 10 5 cells in 200 l of SCNaN 3 immediately prior to dilution (ϳ3 times) 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 with the green fluorescence (M-C 6 -NBD-PE) plotted on a log scale, and the mean (F cell ) and standard deviation of the fluorescence intensity of the live cells calculated.
Other Methods-The protein content was measured as described by Lowry et al. (27) with bovine serum albumin as the standard. The protein samples were electrophoresed on SDS-polyacrylamide gel according to Laemmli (28) and stained with either Coomassie Blue or silver. The yeast cells were transformed as described by Kuo and Campbell (29), and bacteria (DH5␣ strain) transformation was performed by electroporation using a Bio-Rad Gene Pulser apparatus, following the manufacturer's instructions.

RESULTS
Overexpression of Yor1p and Genetic Purification-The low level of expression of YOR1 in wild-type yeast (1) precludes characterization of its properties. Traditional overexpression of plasma membrane proteins using strong promoters and multicopy vectors often causes mistargeting and stimulates accumulation of intracellular membranes (30). These problems have been overcome in a new approach, which has allowed dramatically enhanced overexpression of Pdr5p and Snq2p ABC transporters in the yeast plasma membrane (6 -8). The method takes advantage of point mutations in the transcription factorencoding genes PDR1/PDR3, which activate the transcription of their target genes. The target gene promoters contain typical binding sequences called PDREs (for PDR response elements) which correspond to the 5Ј-TCCG(C/T)GGA-3Ј consensus sequence (12,31,32) (Table III). An inherent problem to this approach is the simultaneous overexpression of several Pdr1pregulated proteins, including other ABC transporters. It was therefore necessary to identify all potentially interfering proteins and eliminate them by gene deletion. Systematic sequencing of the yeast genome has revealed new ORFs, which encode a total of 15 full-size ABC transporters (3) including the PDR10 (33), PDR15 (34), and PDR11 (35) genes whose promoters display at least one putative PDRE. Table III shows the transcription activity mediated by the PDR5, YOR1, SNQ2, PDR10, PDR15, and PDR11 promoters in the presence of PDR1 wild type-, null, or mutated PDR1-3 allele as measured by the ␤-galactosidase activity of fusion constructs. The PDR11 promoter-mediated transcription activity was weak and not significantly affected by Pdr1p, possibly because the 5Ј-TCCG-CAGA-3Ј sequence in the promoter was insufficient for Pdr1p/ Pdr3p recognition. The PDR1-3 mutation slightly increased the efficiency of the PDR15 gene promoter despite the presence of a perfect Pdr1p/Pdr3p-binding site. In contrast, the PDR1-3 allele increased the efficiency of the PDR10 promoter 11-fold. Note, however, that the putative PDREs of the PDR10, PDR15 and PDR11 gene promoters have yet to be verified experimentally. In the presence of the wild-type allele of PDR3, the PDR1-3 mutation increased expression of ␤-galactosidase 25 times for the YOR1 promoter, 17 times for the PDR5 promoter, and 5 times for the SNQ2 promoter.
Table III also shows that the number of putative PDREs per promoter did not correlate with the extent of the induction of ␤-galactosidase activity mediated by Pdr1p. Although the PDR5 and SNQ2 promoters each have three PDREs (12,31,32), PDR5-lacZ expression was enhanced 5 times by the wildtype Pdr1p while the expression of the SNQ2-lacZ fusion was increased only 1.4 times upon the addition of the wild-type PDR1 allele. Under the same conditions, the induction of the other translational lacZ fusions showing 1 or 2 putative PDREs varied between 1.4 (YOR1-lacZ) and 2.2 (PDR10-lacZ). A further important point is that the ␤-galactosidase activities of PDR5-, SNQ2-, and YOR1-lacZ fusions reported here in the presence of wild type Pdr3p are quite different to when Pdr3p is absent (21).
In agreement with the results of Table III, Fig. 1 shows that the PDR1-3 mutant plasma membrane (compared with that from either wild type or pdr1⌬ strains) dramatically overexpressed a 160-kDa Coomassie Blue-stained band, which mainly comprised Pdr5p, Snq2p, and Yor1p. In order to free Yor1p from membrane contaminants of similar size and possibly overlapping function, we deleted PDR5, SNQ2, PDR10, and PDR15 yeast ABC transporter-encoding genes whose expression was also induced by the PDR1-3 mutation. The PDR11 gene whose product is a constituent of the 160-kDa overexpressed band (8) was also deleted. YCF1 was deleted because this gene is involved in drug resistance (36,37).
Yor1p Binds TNP-8-azido-ATP and Shows ATPase Activity-Using the AD234567 multiply deleted strain overexpressing Yor1p, we started to investigate the Yor1p potential NTP hydrolysis activity. Pdr5p has an NTPase activity, which was distinguished from that of the H ϩ -pump Pma1p by its a The conserved residues of Pdr1/3p binding motifs found in the promoter regions of the genes are shown in boldface type, and the palindromic sequences are underlined.
b The measurements were done in the FY1679/EC (pdr1⌬) strain as described under "Experimental Procedures." The ␤-galactosidase activities, measured in four independent yeast transformants, are expressed with the standard errors.
broader nucleotide specificity and pH dependence (7,8). With similar protein levels in the plasma membrane, the UTPase activity of Pdr5p was up to 10-fold higher than that of Pma1p (Fig. 2), which is the major ATPase in the yeast plasma membrane. The situation with Yor1p-enriched plasma membrane (AD234567) was very different since no significant difference in the NTPase activity was detected compared with the Yor1pdepleted strain (AD1234567). The same results were obtained with the AD2345678/AD12345678 strains. However, in the SU-PERYOR strain obtained by fusion of the PDR5 promoter to the YOR1 ORF, a survey of pH from 5.0 to 8.5 revealed that some very low ATPase activity could be associated with SuperYor1p above pH 7.0 (data not shown).
Partial purification of SuperYor1p by centrifugation on sucrose gradient, separated it from H ϩ -ATPase (Fig. 3A). This allowed us to detect ATPase activity associated with the SuperYor1p-enriched fractions (fractions 6 -12) after subtraction of the contaminating ATPase activity (ϳ75 nmol of P i ⅐min Ϫ1 ⅐mg Ϫ1 ) measured in the corresponding fractions of the AD1234567 Yor1p-depleted strain (Fig. 3B). Detection of the ATPase activity required the presence of 150 g/ml asolectin, which was found to be optimal concentration (data not shown). By scanning the SDS-polyacrylamide gel, we found good correlation between the SuperYor1p band intensity and its ATPase activity (Fig. 3B). The enzyme was able to hydrolyze ATP from pH 7.0 to 8.5 with similar efficiency (Fig. 3C). ATPase activity was strongly sensitive to vanadate (I 50 ϳ 0.3 M) and inhibited by oligomycin concentrations above 3 g/ml (Fig. 3, D and E).
Overexpressed Yor1p Confers Increased Resistance to Oligomycin in Vivo-Overexpressed Yor1p has been reported to confer resistance to oligomycin (1). The analysis of the multiple deletions shows that oligomycin resistance of the PDR1-3 mutant strain is also dependent on the presence of Pdr5p. Unlike AD1 (yor1⌬) and AD12 (yor1⌬snq2⌬) strains, the AD13 (yor1⌬pdr5⌬) strain does not grow on plates containing 0.25 g/ml oligomycin (Fig. 5). Previous studies showed unmodified oligomycin sensitivity in single PDR5 deletants (11, 12) but Kolaczkowski et al. (43) showed that oligomycin is a competitive inhibitor of Pdr5p-mediated transport of rhodamine 6G. As also shown by Fig. 5, Pdr1p more strongly influences oligomy-cin resistance than its homolog Pdr3p (compare the FY1679/EC (pdr1⌬) and FY1679/TD (pdr3⌬) strains). This is in agreement with previous observations (12) even though no difference in oligomycin resistance between pdr1⌬ and pdr3⌬ strains was observed by Delaveau et al. (44). As also shown by Fig. 2, oligomycin resistance was increased 40 times in the strain AD234567 which overexpresses Yor1p and was further increased by a factor of more than 8 in the SUPERYOR strain. These data indicate that, at the cell level, the drug transport properties of Yor1p are conserved during overexpression.
Overexpressed Yor1p Confers Resistance to Rhodamine B-Cui et al. (5) recently found that deletion of YOR1 increases the cellular content of rhodamine B (which is negatively charged, unlike the Pdr5p substrate rhodamine 6G; see Ref. 43). Analysis of the multiply deleted mutants revealed that rhodamine B resistance is mediated by both Yor1p and Pdr5p. The deletion of PDR5 did not allow cell growth in the presence of 500 g/ml rhodamine B and deletion of both PDR5 and YOR1 (AD13) increased drug sensitivity so that the cells fail to grow in 250 g/ml rhodamine B (Fig. 5). Finally, the combined deletion of the YOR1, SNQ2, PDR5, PDR10, PDR11, PDR15, and YCF1 genes (strains AD1234568 and AD12345678) further reduced the resistance to rhodamine B since growth was diminished at 50 g/ml rhodamine B and abolished at 100 g/ml rhodamine B. Pdr1p affects rhodamine B and oligomycin resistance more drastically than Pdr3p (FY1679/EC (pdr1⌬) compared with FY1679/TD (pdr3⌬)). Thus, Pdr1p and Pdr3p me- Solubilized plasma membrane proteins (7 mg) from strains SUPERYOR (yor1⌬, snq2⌬, ⌬pdr5::PDR5PROM-YOR1- PDR5  STOP, pdr10⌬, pdr11⌬, ycf1⌬, pdr3⌬) and AD1234567 (yor1⌬, snq2⌬, pdr5⌬,  pdr10⌬, pdr11⌬, ycf1⌬, pdr3⌬) (data not shown) were separated by centrifugation through a continuous sucrose gradient. Fifty l of fractions 4 -15 were analyzed by SDS-PAGE and stained with silver (A). The polyacrylamide gel was scanned and the intensity of the SuperYor1p band was estimated using the Image Master 1D program (Amersham Pharmacia Biotech) (B, closed triangles). Ten l of fractions 4 -12 (SUPERYOR gradient) were assayed for ATPase activity at pH 7.5 as described under "Experimental Procedures" and, each time, the contaminating ATPase activity measured in the corresponding fractions of the deleted strain (AD1234567) was subtracted (B, open circles). The activities were calculated for three independent gradients of each strain. Similarly, the effects of pH (C), vanadate at pH 7.5 (D), and oligomycin at pH 7.5 (E) on SuperYor1p ATPase activity were estimated. diate rhodamine B and oligomycin resistance through both Pdr5p and Yor1p. Notice that Yor1p and Pdr5p share other common substrates, including the fungicide miconazole (data not shown).
The involvement of Yor1p in rhodamine B and oligomycin resistance of pdr5⌬ strains was confirmed by complementation of the AD13 (yor1⌬pdr5⌬) and the AD1234568 (yor1⌬, snq2⌬,  pdr5⌬, pdr10⌬, pdr11⌬, ycf1⌬, pdr15⌬) strains with the YOR1 gene. Transformation of both strains with the YOR1 gene on centromeric plasmid allowed recovery of growth on both 0.5 g/ml oligomycin and 250 g/ml rhodamine B as shown for the AD3 and AD234568 strains (data not shown).
The Efflux of Rhodamine B in Yeast Cells Overexpressing Yor1p Is Energy-dependent-We therefore used the Yor1poverexpressing strain AD2345678 (snq2⌬, pdr5⌬, pdr10⌬,  pdr11⌬, ycf1⌬, pdr3⌬, pdr15⌬) and its isogenic control, deleted in YOR1, AD12345678 (yor1⌬, snq2⌬, pdr5⌬, pdr10⌬, pdr11⌬,  ycf1⌬, pdr3⌬, pdr15⌬) to demonstrate the involvement of Yor1p in the rhodamine B cell content. In the yor1⌬ strain, rhodamine B accumulation was slightly lowered (82%) by deoxyglucose compared with glucose (100%). However, in the presence of overexpressed Yor1p, rhodamine B accumulation was much higher (58%) in energy-starved cells (deoxyglucose), while glucose caused a drastic reduction (to 8%) in the cellular rhodamine B content (data not shown). Fig. 6A shows the Yor1pmediated energy-dependent extrusion of rhodamine B from pre-loaded cells incubated with deoxyglucose, which depletes the intracellular ATP (45). The addition of glucose caused rapid extrusion of rhodamine B from the Yor1p-expressing cells, while no glucose-dependent effect was observed in the yor1⌬ cells. Fig. 6B shows the difference in the supernatant fluorescence (F glucose Ϫ F control ) in the same experiment and establishes the Yor1p energy-dependent extrusion of rhodamine B out of the cell. Addition of 30 M protonophore FCCP in the presence of the respiratory substrate ethanol as the sole energy source completely abolished rhodamine B transport. while, in the presence of glucose, FCCP only slightly affected Yor1pmediated rhodamine B transport (Fig. 6C). This may be explained either by the partial involvement of oxidative phosphorylation in the ATP formation of glucose-grown cells and/or by a possibly higher use of cellular ATP by the Pma1p H ϩ -ATPase under these conditions. In the absence of an energy source, the application of a pH gradient of 2 units (pH 7.0 inside the cell and pH 5.0 outside the cell) did not cause rhodamine B extrusion (Fig. 6C).
Yor1p and Pdr5p Are Involved in M-C 6 -NBD-phosphatidylethanolamine Accumulation in Vivo-Several mammalian ABC transporters translocate phospholipids (13)(14)(15)(16). A "flipflop" of hydrophobic drugs from the inner leaflet to the outer one has been proposed as part of the mechanism of drug transport (49). We have tested whether overexpressed Yor1p was involved in the transport of a fluorescent phospholipid analog, M-C 6 -NBD-PE. Cell fluorescence was measured by flow cytometry after the incubation of yeast cells with M-C 6 -NBD-PE (Fig.  7A) and analyzed by fluorescence microscopy (Fig. 7B). The average accumulation of M-C 6 -NBD-PE in the PDR1-3 mutant strain (US50 -18C) was about 13% of the PDR1 parent strains (IL125-2B and 2229 -5C). In similar experiments, another mutant, PDR1-11, accumulated 1-2% of its isogenic parent. 2 A strain in which the PDR1 gene was deleted (D1-3/3) accumulated about 70% more M-C 6 -NBD-PE than the PDR1 parent strains (Fig. 7A). It therefore seems likely that Pdr1p activates the expression of genes encoding proteins that decrease the steady-state accumulation of M-C 6 -NBD-PE by either increasing its efflux or decreasing its influx.
Thus, to determine if any of the seven ABC transporters included in this study were responsible for the efflux of M-C 6 -NBD-PE, we analyzed the multiply deleted strains. Based on the average M-C 6 -NBD-PE fluorescence intensity for 10,000 cells, the PDR1-3 strain accumulated only 8% of the pdr1⌬ strain. Deletion of YOR1 (AD1) or PDR5 (AD3) in the PDR1-3 strain resulted in an increase to 13% M-C 6 -NBD-PE accumulation. The fluorescence intensity of the doubly deleted strain yor1⌬ pdr5⌬ (AD13), was increased to 36% of the pdr1⌬ strain (60% of the PDR1 wild-type strain), a value slightly higher than if the two effects were additive. Single deletions of SNQ2 (AD2), PDR10 (AD4), PDR11 (AD5), YCF1 (AD6), or PDR15 (AD8) had no effect on M-C 6 -NBD-PE accumulation. Thus, of the ABC transporters tested, only Pdr5p and Yor1p appeared to transport M-C 6 -NBD-PE. DISCUSSION This study provides some new information on the control of YOR1 and other yeast ABC genes by the transcription factors Pdr1p and Pdr3p. The expression of lacZ gene fusions with PDR10, PDR15, and PDR11 gene promoters containing putative PDREs reveals that PDR10 may be a target of Pdr1p/ Pdr3p transcription factors. However, its level of expression was very low compared with that of PDR5, SNQ2, and YOR1, whose transcription was greatly enhanced by the PDR1-3 mutation. Transcription of the PDR15 gene, despite the presence of one perfect Pdr1p/Pdr3p-binding site (PDRE) in its promoter, was only increased 2 fold by the PDR1-3 mutation. The transcription level of the PDR11 gene was not modified upon addition of either the wild type PDR1 or the mutated PDR1-3 allele, possibly because the PDRE sequence in its promoter is not sufficient for Pdr1p/Pdr3p binding (TCCGCAGA instead of TCCG(T/C)GGA).
In the PDR1-3 gain-of-function mutant, the PDR5 gene promoter gave the highest absolute ␤-galactosidase activity among the fusion products with five different Pdr1p-regulated gene promoters. This is consistent with the SDS-PAGE analysis of plasma membrane-enriched fractions of multiply deleted strains. The latter analysis also confirmed that Pdr5p, Snq2p, and Yor1p ABC transporters are the major components of the 160-kDa overexpressed protein band in the PDR1-3 mutant.
Characterization of Yor1p, a minor protein of the plasma membrane, required both the overexpression of the Yor1p protein and also the deletion of related ABC transporters, in particular those suspected to be activated by the PDR1-3 mutation. We were able to delete up to 7 full-size yeast ABC The bars represent the standard deviations. According to unpaired Student's t test, values are statistically different from value for US50 -18C strain with p Ͻ 0.01 ( §); with p Ͻ 0.05 (*); values are not significantly different from value for US50 -18C strain since p Ͼ 0.1 (£). B, fluorescence microscopy: 5% means that a neutral density filter was used to reduce the excitatory light by 95%. In the second column (100%), the filter was removed. The last column shows the differential interference contrast optics. transporters without impairing growth.
Membrane-bound Yor1p was labeled with TNP-8-azido-ATP in vitro in a saturable manner and the labeling was inhibited by ATP. Interestingly, the K 0.5 for TNP-8-azido-ATP that we found for Yor1p (about 45 M) is very close to the value found by Liu and Sharom for TNP-ATP binding to human P-glycoprotein (43 M) (50). The nucleotide binding properties of Pdr5p and Yor1p appeared very comparable.
Investigation of the drug transport properties of intact cells overexpressing Yor1p showed that the Yor1p enzyme was active in vivo and required energy. The oligomycin resistance was increased 40 times in the strain overexpressing Yor1p and further increased by about 8-fold in the SUPERYOR strain, indicating that overexpression of Yor1p does not change its oligomycin transport capability in vivo. Absence of Yor1p or Pdr5p caused increased sensitivity to both oligomycin and rhodamine B. These effects were additive when both proteins were missing. A requirement for energy was demonstrated by the strong efflux of rhodamine B from Yor1p-enriched energystarved cells on glucose addition. The energy required for this process appears to be provided by ATP rather than ⌬pH. Finally, absence of either Yor1p or Pdr5p resulted in increased accumulation of a fluorescent phosphatidylethanolamine. Again, the effect was more pronounced when both ABC transporters were deleted, indicating that the transporters may act independently. Conversely, none of the 5 other ABC transporters, including the overexpressed Snq2p transporter, exhibited this activity.
In plasma membranes from the AD234567 Yor1p-expressing strain, it was difficult to measure ATPase activity distinct from that of Pma1p. Enhancing the level of Yor1p by fusing the YOR1 ORF to the PDR5 promoter in the PDR1-3 mutant (SUPERYOR strain) allowed us, however, to detect ATPase activity in solubilized and partially purified SuperYor1p fractions. The SuperYor1p activity of ϳ100 nmol of P i ⅐min Ϫ1 ⅐mg Ϫ1 was sensitive to both vanadate and oligomycin, establishing that we are not dealing with either P-type or F-type ATPase contaminants. The use of the PDR1-3 gain-of-function allele of PDR1 has already allowed characterization of the Pdr5p and Snq2p NTPase activities (7,8). Surprisingly, Pdr5p, Snq2p, and (Super)Yor1p NTPase activities show distinct characteristics; Pdr5p hydrolyzes all Mg-NTPs over a broad pH range, whereas Snq2p and Yor1p hydrolyze ATP preferentially. The pH profile of Yor1p ATPase activity is also very broad (with an optimum at pH 7.5), while it is much sharper for Snq2p (pH 6.3). Only Pdr5p and Yor1p ATPase activities are oligomycinsensitive, while vanadate was shown to inhibit all three enzyme ATPase activities. Finally, taking into account the relative amount of each transporter in the plasma membrane, one can estimate that the Yor1p ATPase activity is more or less 15 times lower than that of Pdr5p or Snq2p.
As the pumping capacity (and specificity) of the Yor1p and Pdr5p transporters appears similar in vivo, the low ATPase activity of Yor1p (ϳ0.1 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 ) compared with that of Pdr5p (ϳ1.5 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 ) in similar conditions may indicate that the purified Yor1p transporter is a more highly coupled pump than Pdr5p. The high ATPase activity and apparently low pumping capacity of ABC drug transporters (51-54) has been a puzzling feature. For instance, Pdr5p ATPase activity is very high in the apparent absence of drugs or other substrates, slightly stimulated by substrates, and shows a broad nucleotide specificity despite an in vivo requirement for ATP for transport (43). The poor NTPase activity of Yor1p might be due to the fact that an activation factor required for ATPase activity is lost during the plasma membrane preparation or that some phospholipids block the Yor1p ATPase activity when tested in membranes. Another possibility is that Yor1p, like its ortholog CFTR, has low ATPase activity (ϳ0.05 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 ) that is sufficient for the control of channel gating (55). Very recently, the close human ortholog of Yor1p, MRP, was purified and shown to hydrolyze ATP at a rate of ϳ0.3 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 in the presence of 6 mM Mg-ATP (56). If we consider that, in our preparations, the "purified" SuperYor1p amounts to 20% of the total proteins, one can estimate that SuperYor1p may exhibit an ATP hydrolysis rate of ϳ0.5 mol of P i ⅐min Ϫ1 ⅐mg SuperYor1p Ϫ1 .
The mediation of phospholipid efflux by yeast Pdr5p and Yor1p is consistent with recent reports that identified other ABC transporters as phospholipid transporters or flippases. These include mouse mdr1 (16) and mdr2 (13,14) and human MDR1 (16) and MDR3 (15,16). Phospholipid transport by MDR1 and mdr1 is not head-group or glycerol backbone-specific, whereas the MDR3 P-glycoprotein transports phosphatidylcholine exclusively. The specificity of Yor1p and Pdr5p for phospholipids other than phosphatidylethanolamine remains to be seen, but it could be different for the two pumps. It seems possible that the high steady state of ATP hydrolysis by Pdr5p and the low ATPase activity of Yor1p may be related to activation of the former with certain yeast phospholipids, which may not be substrates or even inhibitors of the latter. Anyway, it is intriguing that Pdr5p and Yor1p share similar phospholipid translocation properties which were not observed for Snq2p.
Finally, we wish to point out that, taking advantage of the strong PDR5 promoter associated with the gain-of-function PDR1-3 mutation, we have developed an important new tool for overexpression of ABC transporters in the yeast plasma membrane. This system allows the overexpression of functional yeast Yor1p at a level that represents more than 10% of total plasma membrane proteins. For comparison, overexpression of the human MDR1 from an high copy number expression vector in yeast by Mao and Scarborough (57) yielded protein amounts to only 0.4% of the total yeast membrane proteins. We anticipate that the PDR system, which dramatically overexpresses the functional yeast Yor1p in the plasma membrane without associated intracellular trafficking problems, and the use of the PDR5 promoter in strains deleted in the majority of the fullsize ABC transporter-encoding genes provides a prototype for high level expression of orthologous ABC transporters of medical interest.