PDR16 and PDR17, two homologous genes of Saccharomyces cerevisiae, affect lipid biosynthesis and resistance to multiple drugs.

The Saccharomyces cerevisiae open reading frame YNL231C was recently found to be controlled by the multiple drug resistance regulator Pdr1p. Here we characterize YNL231C (PDR16) and its homologue YNL264C (PDR17). Deletion of PDR16 resulted in hypersensitivity of yeast to azole inhibitors of ergosterol biosynthesis. While no increase in drug sensitivity was found upon deletion of PDR17 alone, a Deltapdr16,Deltapdr17 double mutant was hypersensitive to a broad range of drugs. Both mutations caused significant changes of the lipid composition of plasma membrane and total cell extracts. Deletion of PDR16 had pronounced effects on the sterol composition, whereas PDR17 deletion mainly affected the phospholipid composition. Thus, Pdr16p and Pdr17p may regulate yeast lipid synthesis like their distant homologue, Sec14p. The azole sensitivity of the PDR16-deleted strain may be the result of imbalanced ergosterol synthesis. Impaired plasma membrane barrier function resulting from a change in the lipid composition appears to cause the increased drug sensitivity of the double mutant strain Deltapdr16,Deltapdr17. The uptake rate of rhodamine-6-G into de-energized cells was shown to be almost 2-fold increased in a Deltapdr16,Deltapdr17 strain as compared with wild-type and Deltapdr5 strains. Collectively, our results indicate that PDR16 and PDR17 control levels of various lipids in various compartments of the cell and thereby provide a mechanism for multidrug resistance unrecognized so far.

The Saccharomyces cerevisiae open reading frame YNL231C was recently found to be controlled by the multiple drug resistance regulator Pdr1p. Here we characterize YNL231C (PDR16) and its homologue YNL264C (PDR17). Deletion of PDR16 resulted in hypersensitivity of yeast to azole inhibitors of ergosterol biosynthesis. While no increase in drug sensitivity was found upon deletion of PDR17 alone, a ⌬pdr16,⌬pdr17 double mutant was hypersensitive to a broad range of drugs. Both mutations caused significant changes of the lipid composition of plasma membrane and total cell extracts. Deletion of PDR16 had pronounced effects on the sterol composition, whereas PDR17 deletion mainly affected the phospholipid composition. Thus, Pdr16p and Pdr17p may regulate yeast lipid synthesis like their distant homologue, Sec14p. The azole sensitivity of the PDR16deleted strain may be the result of imbalanced ergosterol synthesis. Impaired plasma membrane barrier function resulting from a change in the lipid composition appears to cause the increased drug sensitivity of the double mutant strain ⌬pdr16,⌬pdr17. The uptake rate of rhodamine-6-G into de-energized cells was shown to be almost 2-fold increased in a ⌬pdr16,⌬pdr17 strain as compared with wild-type and ⌬pdr5 strains. Collectively, our results indicate that PDR16 and PDR17 control levels of various lipids in various compartments of the cell and thereby provide a mechanism for multidrug resistance unrecognized so far.
The yeast Saccharomyces cerevisiae has, like many other organisms, the ability to acquire multiple drug resistance, i.e. become less sensitive to a broad range of chemically and func-tionally unrelated cytotoxic compounds (1,2). In yeast this phenomenon can be provoked by a regulatory disorder, namely a mutation in the transcription factors Pdr1p or Pdr3p. Pdr1p and Pdr3p are homologous Zn 2 Cys 6 DNA-binding proteins which control the expression of drug efflux pump-encoding genes (3,4). Gain-of-function mutations in the PDR1 or PDR3 genes may result in increased production of these efflux pumps, leading to drug resistance (5,6). Neither Pdr1p or Pdr3p nor the drug efflux pumps which they regulate are required for growth of yeast in the absence of drugs. It is not known whether the true physiological function of these drug resistance determinants is to protect the cell from external toxic compounds or whether they may play other roles. In order to get more insight into the physiological role of Pdr1p, we recently screened for target genes regulated by this transcription factor. This screening resulted in the identification of a broad range of novel Pdr1p target genes, one of which was the open reading frame with the systematic name YNL231C (PDR16). Expression of the PDR16 gene is five times higher in strains carrying pdr1-3, a strong constitutive allele of PDR1, as compared with isogenic wild-type strains or strains deleted for PDR1. 1 The PDR16 gene encodes for a protein of 351 amino acids. This protein is 49% identical and 75% similar to the product of the YNL264C (PDR17) gene of S. cerevisiae. Neither PDR16 nor PDR17 has been functionally characterized. The Pdr16p is also 23% identical and 54% similar to the product of the S. cerevisiae SEC14 gene. This homology is spread throughout the protein sequence. However, three sequence blocks which are strongly conserved among the SEC14 proteins from different yeasts (around amino acid positions 55-60, 205-210, and 235-240) (7) are also the most conserved areas between these proteins and PDR16 and PDR17.
Sec14p was initially identified as a PtdIns 2 transfer protein (PITP) which can perform transfer of phospholipids between membranes in vitro (8,9). Subsequently, it was shown that Sec14p/PITP is required for transport of proteins through the Golgi complex (10). It has been proposed that in vivo Sec14p senses the levels of PtdIns and PtdCho in the Golgi complex and exerts negative feedback on PtdCho synthesis through the Kennedy pathway (11). More recently, it was suggested that Sec14p/PITP may also regulate formation of secretory vesicles from the Golgi by stimulating the turnover of phospholipids (12,13). The involvement of PITP in signal transduction of higher eukaryotes has been discussed (14 -16).
In the present work, we analyze the role of the distant yeast SEC14 homologues, PDR16 and PDR17, in drug resistance and lipid biosynthesis/sorting in a wild-type and a pdr1-3 background. We find that deletion of PDR16 leads to a strongly increased sensitivity to azole antifungals. Deletion of both PDR16 and PDR17 leads to reduced resistance to a broad range of drugs. We, furthermore, show that the mutations affect the phospholipid and sterol composition of the plasma membrane, and that they change the total yeast lipid composition. We propose that the azole sensitivity of the ⌬pdr16 single mutant is mainly due to impaired sterol synthesis and that the broad increase in drug sensitivity of the double mutant is a result of a more general change in plasma membrane composition. The possible roles of Pdr16p and Pdr17p in lipid biosynthesis and sorting are discussed.
A construct for the disruption of the YOR1 gene (pDK30) (21) was kindly supplied by Dr. W. Scott Moye-Rowley. All deletions of the PDR16 gene and the deletions of the PDR17 gene in US50 -18c and its derivatives were constructed using the procedure described by Alani et al. (22). A DNA fragment corresponding to the 5Ј-flanking region upstream of the gene was generated by PCR and cloned into pSKϩ (Stratagene). For the PDR16 gene, the primers for the PCR were 5Ј-GCACGAATTCTCAAAGACGGCGGATTCA-3Ј and 5Ј-AACCGGATC-CCCTGGGTCTTCTGGAGCCC-3Ј, while for the PDR17 gene, the primers were 5Ј-ACCGAATTCTGATTGAAGAGATCAAAGA-3Ј and 5Ј-TAA-GGATCCGGCAGGAGGGTCCAA-3Ј. Subsequently, a 3Ј-flanking fragment, encompassing the 3Ј end of the gene, was generated by PCR. For the PDR16 gene, the PCR primers were 5Ј-TTTGGATCCTTGGTTAG-CATGG-3Ј and 5Ј-TTGGAGCTCAGTGCATATAGACGCG-3Ј, while for PDR17, they were: 5Ј-ATGGGATCCTTGGAGGCATTGTCGGA-3Ј and 5Ј-CGGGAGCTCGATTGATTAGCTGGAAC-3Ј. This 3Ј-flanking PCR product was subcloned into the plasmid that already contained the 5Ј-upstream fragment. The resulting plasmids were termed pBVH742 for PDR16 and pBVH1068 for PDR17. In a final subcloning step, a 3.8-kilobase BamHI-BglII fragment containing a hisG-URA3-hisG cassette (22) was inserted into the BamHI site between the 5Ј-and 3Јflanking regions of the gene, resulting in the plasmids pBVH778 (PDR16) and pBVH1127 (PDR17). Each of the resulting plasmids was treated with EcoRI and SacI to generate a linear fragment consisting of the 5Ј-and 3Ј-flanking regions of the gene interrupted by hisG-URA3-hisG. The fragment was transformed into yeast and cells that had become Ura ϩ due to chromosomal replacement of wild-type gene by the linear fragment were selected on SC plates lacking uracil. Subsequently, the URA3 gene was "looped-out" by growth on non-selective media, allowing recombination between the hisG sequences. Cells in which a loop-out event had occurred were selected by plating on media containing 5-fluoroorotic acid (22).
Deletions of PDR17 in strain FY1679 -28C and its derivatives were generated by one-step gene replacement as described by Winston et al. (23). A 1.15-kilobase BamHI fragment from plasmid YDp-H, containing the HIS3 gene, was subcloned into pBVH1068, resulting in plasmid pBVH1451. The pBVH1451 plasmid was subsequently treated with EcoRI and SacI to generate a linear fragment consisting of the 5Ј-and 3Ј-flanking regions of the PDR17 gene interrupted by HIS3. This fragment was transformed into yeast, and cells that had become His ϩ due to chromosomal replacement of the wild-type PDR17 gene by the linear fragment were selected on SC plates lacking histidine.
All deletions were verified by PCR amplification of a chromosomal region encompassing the deletion, followed by analysis of the PCR product by agarose gel electrophoresis. Furthermore, the absence of an mRNA signal from PDR16 in the US50 -18c derivative deleted for this gene was verified by Northern blotting, using a fragment of the gene as a probe.
Other Plasmid Constructions-Centromeric plasmids containing the PDR16 and PDR17 genes were obtained by "gap repair." An EcoRI-SacI fragment consisting of the 5Ј-and 3Ј-flanking regions of the PDR16 gene was subcloned from pBVH742 into the URA3 centromere plasmid pRS316 (24), resulting in plasmid pBVH1222. Similarly, a ClaI-SacI fragment consisting of the 5Ј-and 3Ј-flanking regions of the PDR17 gene was subcloned from pBVH1068 into pRS316, giving pBVH1378. The pBVH1222 and pBVH1378 plasmids were linearized using BamHI and transformed into yeast strain US50 -18c. Cells in which recombination between the linearized plasmid and the chromosomal locus had generated a circular centromeric plasmid containing the entire gene were selected on media lacking uracil. Plasmids were isolated from yeast, transformed into E. coli, and restriction analysis of plasmid preparations was performed in order to verify the presence of the gene. A multicopy plasmid containing PDR16 was generated by insertion of a SalI-SacI fragment from the single-copy plasmid containing the gene into pRS426 (25).
Growth Media and Drug Resistance Assays-E. coli was grown in standard Luria broth medium (17). Yeast was grown on standard rich glucose (YPD) or glycerol (YPG) media, or on SC medium lacking appropriate amino acids for plasmid maintenance (26). For drug resistance assays on solid media, drugs were added to the media immediately prior to pouring. The drug concentrations tested were the following: Growth tests on non-fermentable carbon sources, at high pH, osmolarity and temperature were performed as described by Rank et al. (27), i.e. on solid media containing 0.1% yeast extract, 0.5% bacto-peptone, and 100 mM potassium phosphate, pH 7.1, supplemented with 1% glucose (with or without 0.5 M potassium chloride), glycerol or ethanol. Plates were incubated at 30 or 37°C.
Rhodamine-6-G Uptake Assay-Approximately 10 7 cells from an overnight culture were inoculated in 20 ml of YPD and grown for 3 h at 30°C. About 5.6 ϫ 10 8 cells were pelleted and washed three times with buffer A (50 mM HEPES/NaOH, pH 7.0). The cells were subsequently resuspended in 4 ml of de-energization buffer (1 M antimycin A, 5 mM 2-deoxy-D-glucose in buffer A), incubated for 2 h 30 min at 30°C, and transferred to a water bath at 20°C. A 200-l aliquot of the de-energized cell suspension was pelleted, washed once with 200 l of cold buffer A, and resuspended in 2 ml of the same cold buffer. Cell fluorescence background was measured using an SLM Aminco 48000 S spectrofluorimeter. The excitation wavelength was 529 nm (4 nm slit), and the emission wavelength was 553 nm (4 nm slit). Rhodamine-6-G was added to the remaining cell suspension to a final concentration of 5 mM, and 200-l aliquots were taken every 5 min up to 1 h. The cells of each aliquot were pelleted, washed, and resuspended as described above and immediately subjected to fluorescence measurements. The cell fluorescence background values after 0-and 1-h incubation were averaged, and this average was subtracted from each fluorescence value.
Isolation of Yeast Subcellular Fractions-Cells were grown on YPD medium containing 2% glucose, 2% peptone (Oxoid), and 1% yeast extract (Oxoid) under aerobic conditions at 30°C. Growth of yeast cells was monitored by measuring the OD (600 nm).
Highly purified plasma membrane was isolated as follows. The pellet of crude plasma membrane (28) was suspended in 5 mM Mes, 0.2 mM EDTA, pH 6.0, with 10 strokes in a loose-fitting Dounce homogenizer and layered on top of a sucrose density gradient made of 10 ml of 38% (w/w), 10 ml of 43% (w/w), and 10 ml of 53% (w/w) sucrose in 5 mM Mes, 0.2 mM EDTA, pH 6.0. Centrifugation was carried out at 100,000 ϫ g for 2.5 h in an SW-28 rotor (Beckman). The highly purified plasma membrane was withdrawn from the 43/53% sucrose interface, the suspension was diluted 3-fold with 10 mM Tris-HCl, pH 7.4, and the plasma membrane was sedimented at 48,000 ϫ g for 20 min in an SS-34 rotor (Sorvall). The plasma membrane pellet was suspended in 10 mM Tris-HCl, pH 7.4, using a loose-fitting Dounce homogenizer and stored at Ϫ70°C.
Yeast spheroplasts and mitochondria were isolated by published procedures (29). Microsomal fractions were prepared from the postmitochondrial supernatant that had been cleared of small mitochondria by centrifugation for 30 min at 20,000 ϫ g in an SS-34 rotor (Sorvall). The resulting supernatant was subjected to successive steps of differential centrifugation at 30,000, 40,000, and 100,000 ϫ g (30). The 100,000 ϫ g supernatant contains the cytosolic proteins. Protein content and quality of the preparations as well as cross-contamination with other organelle membranes were assessed as described previously (30 -33).
Lipid Analysis of Whole Cell Extracts and Subcellular Fractions-Cells were homogenized for 3 min with CO 2 cooling in the presence of glass beads using a Merckenschlager Homogenizer. Lipids of whole yeast cells and organelle preparations were extracted by the procedure of Folch et al. (34). Analysis of individual phospholipids and neutral lipids was carried out by published procedures (35,36). Alkaline hydrolysis of lipid extracts was carried out as described elsewhere (37). Individual sterols were analyzed by gas-liquid chromatography (GLC) on an HP 5890 Series II Plus GC equipped with electronic pressure control and an HP chemstation software package. An HP 5972 mass selective detector and authentic standards were used for identification of sterols. Injector and interface were kept at 250 and 300°C, respectively. GLC/MS analysis was performed on a capillary column, HP-5MS 30 m ϫ 0.25 mm ϫ 0.25-m film thickness, programmed from 150°C to 320°C at 20°C/min after a 2-min hold at 150°C. Finally, the column was kept at 320°C for 10 min. All analyses were carried out in the constant flow mode. Helium was used as carrier gas with a linear velocity of 34.1 cm/s. One-l aliquots of the samples were injected with an HP 7673 autosampler in splitless mode. Electron impact ionization with 70 eV ionization energy was used for mass spectrometry. Data were collected by scanning from 150 to 600 atomic mass units at 1.6 scans/s. Alternatively, GLC was performed on an HP 5890 equipped with a flame ionization detector (FID) operated at 320°C using a capillary column (HP5, 30 m ϫ 0.32 mm ϫ 0.25-m film thickness). After a 1-min hold at 50°C the temperature was increased to 310°C at 10°C/min. The final temperature was held for 10 min. Nitrogen was used as carrier gas and 1-l aliquots of samples were injected cool on column. Relative retention times of sterols were similar as described previously (38 -40).
Determination of Lipid Transfer Activity-Lipid transfer activity of cytosolic fractions and peripheral organelle membrane proteins was measured according to Ceolotto et al. (41). Integral and peripheral membrane proteins were separated by treatment of organelle membranes with 0.25 M KCl for 20 min on ice. Insoluble membrane components were sedimented by centrifugation at 100,000 ϫ g for 1 h, whereas solubilized proteins were recovered in the supernatant.
The rate of protein-catalyzed transfer of fluorescently labeled phospholipids from small unilamellar donor vesicles to unlabeled unilamellar acceptor membranes was measured as described previously (9, 40 -43). The phospholipid transfer activity was measured using a Shimadzu RF-5301 spectrofluorimeter. The excitation wavelength was set at 342 nm (1.5 nm slit) and the emission wavelength was set at 380 nm (3 nm slit). The assay was performed for 7 min with measurements taken every 0.5 s. Fluorescently labeled PtdCho (44), PtdIns (45), PtdSer (46), and N-trinitrophenyl phosphatidylethanolamine (47) were synthesized by published procedures.
Anisotropy Measurement-Fluidity of the plasma membrane was determined in vitro by measuring the fluorescence anisotropy of TMA-DPH. Samples containing 100 g of membrane protein were incubated with 2.7 nmol of TMA-DPH for 30 min at 30°C. Fluorescence measurements were carried out using a Shimadzu RF 5301 spectrofluorimeter as described previously (48).

Drug Sensitivity of Strains Deleted for PDR16 and/or Its
Homologue PDR17-We recently identified PDR16 (YNL231C) as one of several novel genes controlled by the yeast multiple drug resistance regulator Pdr1p. 1 To investigate whether PDR16 like other Pdr1p targets is involved in multiple drug resistance, we deleted the gene and studied the effects on drug sensitivity. In order to maximize the possible effects, the deletion was made in a pdr1-3 genetic background (strain US50 -18c) which led to overexpression of Pdr16p.
Deletion of PDR16 had no effect on yeast growth in the absence of drugs. However, the PDR16-deleted strain (⌬pdr16) exhibited a strongly increased sensitivity to miconazole and ketoconazole as compared with the parental strain US50 -18c (Table I). For miconazole, the sensitivity of ⌬pdr16 was increased approximately 20-fold over the control: while the minimal inhibitory concentration was 2 g/ml for the parental strain, it was only 0.1 g/ml for the ⌬pdr16 strain. Sensitivity to ketoconazole increased about 10 times. Similar results were obtained with itraconazole (data not shown). The ⌬pdr16 strain was also slightly more sensitive to nystatin than the parental strain. No significant changes in sensitivity to any of the other drugs tested were observed (Table I).
To verify whether the drug sensitivity phenotype was indeed due to loss of PDR16 gene function, we introduced a single-copy plasmid carrying the intact PDR16 gene in the ⌬pdr16 mutant. The resulting transformant had a level of miconazole resistance identical to that of the US50 -18c parental strain, indicating that the mutant phenotype was indeed due to loss of PDR16 function (data not shown).
The PDR16 gene has a close homologue in S. cerevisiae termed PDR17 (YNL264C). To test whether there is a functional relationship between these two genes, we generated an isogenic strain deleted for PDR17, and a double mutant deleted for both PDR16 and PDR17. The single PDR17-deleted strain (⌬pdr17) did not exhibit a growth defect as compared with US50 -18c. Moreover, the ⌬pdr17 strain did not show increased drug sensitivity, except for a minor increase in sensitivity to 4-nitroquinoline-N-oxide (Table I). The growth rate of the double mutant strain ⌬pdr16,⌬pdr17, on the other hand, was slightly decreased on rich media as compared with the parental and the single mutant strains (data not shown). Growth of the various strains was also tested on non-fermentable carbon sources, at high pH, osmolarity, and temperature. While most of these adverse growth conditions did not differentially affect the growth of the parental and mutant strains, growth of the double-deleted strain ⌬pdr16,⌬pdr17 was severely reduced as compared with the parental and the single-deleted ⌬pdr16 and ⌬pdr17 strains at 37°C on potassium phosphate-buffered pH 7 plates containing 0.5 M potassium chloride (data not shown). Furthermore, the ⌬pdr16,⌬pdr17 strain was even more sensitive to the azole antifungals miconazole and ketoconazole than the ⌬pdr16 strain (Table I). Moreover, the double mutant also displayed increased sensitivities to cycloheximide, rhodamine-6-G, oligomycin, 4-nitroquinoline-N-oxide, antimycin A, and PDR16 and PDR17 of Yeast crystal violet (Table I). The increase in sensitivity as compared with the ⌬pdr16 single-deleted strain was about 2-4-fold for most drugs; only sensitivities for ethidium bromide and nystatin were not increased. The increased drug sensitivity phenotypes of the ⌬pdr16,⌬pdr17 strain were indeed due to loss of PDR17 function, because introducing a single-copy plasmid carrying PDR17 restored the miconazole and cycloheximide resistance of the ⌬pdr16,⌬pdr17 strain to the levels of the single PDR16-deleted strain.
The PDR16-and PDR17-related phenotypes that we described thus far were observed in the US50 -18c genetic background. US50 -18c is highly drug resistant due to the pdr1-3 mutation which results in a strong overexpression of Pdr1pregulated drug efflux pumps such as Pdr5p, Snq2p, and Yor1p (5,49). In order to test whether the PDR16-and PDR17-related phenotypes were specific for this particular genetic background, or whether they also occurred in an otherwise wildtype context, we deleted these two genes in the wild-type strain FY1679 -28c and studied drug sensitivity phenotypes.
As expected, due to lack of overexpression of the drug efflux pumps, the FY1679 -28c strain was generally much more sensitive to most drugs than the US50 -18c strain. As can be seen from Table II, however, the effects of deletion of PDR16 and/or PDR17 in wild-type were roughly the same as in US50 -18c. The sensitivity to miconazole and ketoconazole was increased about 10 -20-fold upon deletion of PDR16, and most drug sensitivities were increased 2-5-fold upon additional deletion of PDR17. Thus, the PDR16/PDR17-dependent drug sensitivity phenotypes are not specific for US50 -18c, but can also be observed in a wild-type genetic background. The lower resistance of FY1679 -28c toward crystal violet and rhodamine-6-G allowed detection of a slightly increased sensitivity to these drugs in the ⌬pdr16 strain as compared with the parental strain (Table II).
Deletions of PDR16 and PDR17 Rather Affect Drug Uptake Than Drug Efflux-Two yeast drug efflux pumps known to mediate resistance to azoles are Pdr5p and Yor1p (50). In order to investigate whether the effects of PDR16 on azole resistance were due to reduced Pdr5p and Yor1p function, we constructed strains deleted for PDR16 as well as for PDR5 and/or YOR1, and investigated their drug sensitivities. As can be seen in Table III, a triple mutant ⌬pdr16, ⌬pdr5, ⌬yor1 is more sensitive against some drugs, e.g. ketoconazole and miconazole, than a ⌬pdr5,⌬yor1 strain, indicating that at least part of the effect of the PDR16 gene on drug resistance is independent of Pdr5p and Yor1p function. Furthermore, Table III shows a comparison of the drug sensitivities of strains deleted for PDR5 and/or YOR1 to those of the ⌬pdr16,⌬pdr17 strain. The ⌬pdr16, ⌬pdr17 strain is more resistant to cycloheximide and rhodamine-6-G, two typical substrates for Pdr5p, than a ⌬pdr5 strain, and more resistant to oligomycin, a typical Yor1p substrate, than a ⌬yor1 strain. This strongly suggests that Pdr5p and Yor1p are at least partially active in the ⌬pdr16,⌬pdr17 strain, and that deletion of PDR16 and PDR17 does not lead to loss of function of these drug efflux pumps.
To test whether a difference in passive drug transport was the reason for the reduced drug resistance of the PDR16-or PDR17-deleted strains, we investigated drug uptake into cells in which active transport was blocked by energy depletion. As a probe for drug uptake we used rhodamine-6-G, a toxic pink colored fluorescent dye to which ⌬pdr16,⌬pdr17 strain is more sensitive than the wild-type or PDR16 or PDR17 single-deleted strains (Table II), but less sensitive than ⌬pdr5 (Table III).
Exponentially growing cells of wild-type, ⌬pdr5, and ⌬pdr16,⌬pdr17 strains were depleted for energy by incubation with 2-deoxy-D-glucose and antimycin A. After 2.5 h at 30°C, rhodamine-6-G was added and its cellular uptake was followed (Fig. 1). The wild-type and the ⌬pdr5 strain showed similar rates of rhodamine-6-G uptake indicating that Pdr5p, a strong rhodamine-6-G pump, was not active under these conditions and energy depletion was complete. Under the same conditions, the ⌬pdr16,⌬pdr17 strain showed an almost 2-fold higher rate of rhodamine-6-G uptake. These data indicate that the increased rhodamine-6-G sensitivity of the ⌬pdr16,⌬pdr17 is at least partially due to an increased passive drug uptake into these cells.
Plasma Membrane Lipid Composition of Strains Deleted for PDR16 and/or PDR17-The fact that ⌬pdr16,⌬pdr17 mutations appear to affect the uptake of drugs into yeast led us to investigate some properties of the plasma membrane of the respective mutants. Homologies of the PDR16 and PDR17 gene  products to Sec14p also suggested that perhaps lipid synthesis and/or sorting might be controlled by these genes. Tables IV and V show that deletion of PDR16 and PDR17 in the FY1679 -28c and US50 -18c background caused several alterations of the plasma membrane lipid composition. Whereas the amount of total phospholipids was similar in all strains tested and the pdr1-3 mutation did not have marked effects, the pattern of individual phospholipids was significantly changed in the plasma membranes of ⌬pdr16, ⌬pdr17, and the double mutant (Table IV). In plasma membrane preparations of ⌬pdr16,⌬pdr17 strains concentrations of PtdCho and PtdIns were markedly increased as compared with the control strains FY1679 -28c and US50 -18c, whereas the level of PtdEtn was dramatically reduced. Furthermore, the amount of PtdSer was increased in the plasma membrane of the double mutant in wild-type, but remained constant in the pdr1-3 background. Changes in the PtdCho and PtdIns levels appear to be a cumulative effect of both ⌬pdr16 and ⌬pdr17, whereas alterations in PtdEtn and PtdSer seem to be more clearly expressed in the ⌬pdr17 strain. It is noteworthy that the ⌬pdr17 deletion causes a major increase in the amount of negatively charged phospholipids, PtdSer and PtdIns, in the plasma membrane. This fact may influence surface properties and/or function of membrane bound proteins.
The total sterol content of the plasma membrane (Table V) was significantly reduced by the ⌬pdr16,⌬pdr17 mutations in the background of the FY1679 -28c strain. This effect was not seen in US50 -18c which bears a pdr1-3 mutation. In both backgrounds, however, ergosterol precursors were observed in the plasma membrane of the double mutant. Especially the unusual presence of 4,4-dimethylzymosterol and lanosterol in the plasma membrane deserves our attention. It is well known that Erg11p, the cytochrome P450-dependent lanosterol 14␣demethylase which uses lanosterol as a substrate, is a most sensitive target to azole inhibitors (for a review, see Ref. 51). A possible effect on enzymes of ergosterol biosynthesis of the ⌬pdr16,⌬pdr17 mutations might cause increased sensitivity to azoles as shown in this study (see Tables I-III). As an alternative, the PDR16 and PDR17 deletions might cause mistargeting of the sterol precursor which is normally found at significant amounts only in microsomal membranes and, in the form of fatty acyl esters, in lipid particles (52).
Despite the marked changes of the plasma membrane lipid composition caused by the ⌬pdr16,⌬pdr17 mutations the bulk fluidity of the plasma membrane appears to be largely preserved. Measurement of anisotropy using the fluorescent marker TMA-DPH as a probe revealed that membrane fluidity was not changed (data not shown). Thus, the mutant cells obviously compensate in that respect almost perfectly for the above mentioned alterations.
Total Lipid Composition of Strains Deleted for PDR16 and/or PDR17-To elucidate the possible role of PDR16 and PDR17 in maintaining a distinct lipid composition of yeast FIG. 1. Rhodamine-6-G uptake in de-energized cells. Cells were depleted for energy as indicated under "Materials and Methods." The cellular uptake of rhodamine-6-G was subsequently followed by determination of the cell-associated fluorescence. The strains used for this experiment were FY1679 -28c (wild type) (f), FY ⌬pdr5 (q), and FY⌬pdr16, ⌬pdr17 (). Each time point represents the average of several independent experiments (three experiments for the ⌬pdr5 strain, five for the wild-type, and six for the ⌬pdr16, ⌬pdr17 strain). The variation indicated for each time point is the standard error of the mean. There was no difference in survival of the strains. PDR16 and PDR17 of Yeast plasma membrane by either regulating biosynthesis or sorting of various lipids we compared the lipid composition of plasma membrane preparations to that of total membranes. Both in wild-type and in the pdr1-3 background the phospholipid composition of total cellular membranes was changed significantly upon deleting PDR16 and PDR17 (Table VI). While the PtdCho content was increased, the PtdEtn content was decreased in ⌬pdr16,⌬pdr17 compared with the control strains. The level of PtdSer in total membranes was hardly affected by deleting the two genes, and only a minor decrease of PtdIns was observed. Thus, apart from PtdIns, the changes in phospholipid composition of total membranes corresponded very well to the alterations of plasma membrane phospholipid composition (see Tables IV and VI).
Deletion of PDR16 and PDR17 also changed the sterol pattern of total cells in the pdr1-3 and wild-type background (Table VII). Especially the accumulation of 4,4-dimethylzymosterol and lanosterol in the double mutants can be correlated with the appearance of these two ergosterol precursors in the plasma membrane of ⌬pdr16,⌬pdr17 strains (see Tables V and  VII). The increased total sterol content in US50 -18c ⌬pdr16,⌬pdr17 relative to US50 -18c may be attributed to the slower growth of the former strain that is accompanied by accumulation of steryl esters.
Lipid Transfer Activities of PDR16 and PDR17 Gene Products in Vitro-Since the PDR16 and PDR17 genes are distant homologues of the SEC14 gene which encodes the yeast PITP, it was tempting to speculate that Sec14p and the PDR16 and PDR17 gene products may have similar functions. Therefore we investigated whether the levels of PtdIns, PtdCho, and PtdSer transfer activities in isolated subcellular fractions differed between the various strains used in this study. Cytosolic Sec14p/PITP (9) and membrane-bound lipid transfer proteins (41) prefer PtdIns as a substrate (Fig. 2). In the US50 -18c background the lipid transfer activity of the cytosol was only affected to a minor extent by mutations of the PDR16 and PDR17 genes. In contrast, the effect of these mutations on the membrane-bound lipid transfer activity was pronounced. Especially in microsomal fractions (Fig. 2) of ⌬pdr16, ⌬pdr17 strains the PtdIns transfer activities were largely reduced as compared with the control strain US50 -18c. Reduced transfer activities were also observed for microsomal fractions of both single mutants. Effects of the ⌬pdr16 and ⌬pdr17 single mutations on PtdIns transfer activity were not additive, which might be an indication that the products of PDR16 and PDR17 somehow interact and deletion of one gene reduces the transfer activity conferred by the other. However, absolute rates of phospholipid transfer activities can only be measured with a mean standard deviation of Ϯ15-30% and should be therefore interpreted with caution. The levels of transfer activity in mitochondrial and plasma membrane fractions were low, and no significant difference was observed between any of the strains (data not shown). In subcellular fractions of FY1679 -28c, the level of PtdIns transfer activity was similar to the ⌬pdr16, ⌬pdr17 double mutant in the pdr1-3 background, and deletion of PDR16 and PDR17 did not further reduce the transfer activities (data not shown). In summary, these data suggest that the increase of the in vitro transfer activities due to the  pdr1-3 mutation is the result of enhanced activities of the PDR16 and PDR17 gene products. PtdCho and PtdSer transfer activities, however, were generally much lower than PtdIns transfer activity in all fractions tested. Furthermore, PtdCho and PtdSer transfer activities were independent of the intactness of PDR16 and/or PDR17.

DISCUSSION
Pdr1p has been known for several years as a transcriptional regulator controlling yeast multiple drug resistance (for review, see Ref. 2). Pdr1p together with its homologue Pdr3p regulates the expression of the drug efflux pumps Pdr5p, Snq2p, Yor1p, Pdr10p, and Pdr15p (18,53). Recently, it was found that Pdr1p and Pdr3p control transport of phospholipids across the plasma membrane via Pdr5p and Yor1p (18,54), raising the possibility that such transport is a physiological function of the network of PDR proteins. Furthermore, Pdr1p and Pdr3p regulate the expression of two hexose transporterencoding genes (6). Thus, Pdr1p controls plasma membrane function by regulating the level of various active transport systems. The present work suggests that Pdr1p may also affect structure and function of the plasma membrane and drug resistance of the yeast through another mechanism, namely by controlling the expression of the PDR16 gene.
The PDR16 gene is specifically required for resistance of yeast cells to miconazole and ketoconazole (see Tables I and II). These two drugs have a similar mode-of-action: they affect ergosterol biosynthesis at the level of the ERG11 gene product, the cytochrome P450-dependent lanosterol 14␣-demethylase. One possible explanation for the increased azole sensitivity of the ⌬pdr16 strain is that the intracellular levels of azoles are increased due to a change in structure and/or function of the plasma membrane of the mutant strain. However, although plasma membrane composition was found to be altered in the mutant, deletion of PDR16 alone had little effect on resistance of yeast cells to other drugs. Thus, it is not likely that the permeability of the plasma membrane has changed severely in the single mutant. The accumulation of precursor sterols in the plasma membrane of the ⌬pdr16 mutant in the absence of azoles rather suggests that the activity of enzymes that play a role in ergosterol biosynthesis is affected in this strain, making it more sensitive to inhibition by azoles. The finding that the level of precursor sterols is not only elevated in the plasma membrane, but also in total cell extracts of the double mutant ⌬pdr16,⌬pdr17 supports that hypothesis. Drug resistance phenotypes due to deletion of PDR17 were only seen in the absence of PDR16. The double-deleted strain ⌬pdr16,⌬pdr17 exhibits a broad drug sensitivity spectrum, although the most dramatic effects were observed with miconazole and ketoconazole. The increased sensitivity to a broad range of drugs, including mutagens, inhibitors of protein synthesis and mitochondrial energy production, rather points toward a general change in intracellular drug concentrations due to the mutations than to an impairment of all the individual drug target functions. Most drugs for which increased sensitivity was found are hydrophobic and are believed to enter the cell by passive diffusion. However, the yeast plasma membrane contains a number of protein pumps which can extrude drugs from the cell in an ATP-dependent manner. Thus, a change in intracellular drug concentrations could be due to increased passive uptake of the drugs through the plasma membrane or to reduced active, protein-mediated drug efflux out of the yeast. The ⌬pdr16,⌬pdr17 strain was more resistant to cycloheximide and rhodamine-6-G than a ⌬pdr5 strain, and more resistant to oligomycin than a ⌬yor1 strain, indicating that Pdr5p and Yor1p are active in the ⌬pdr16,⌬pdr17 strain. Moreover, deletion of PDR16 further reduced drug resistance of a ⌬pdr5,⌬yor1 strain, indicating that at least part of the reduced resistance is independent of Pdr5p and Yor1p. Thus, there is no indication that active protein-mediated drug efflux is reduced in strains deleted for PDR16 and/or PDR17. We cannot exclude that Pdr5p and Yor1p function is partially affected, or that other yet unidentified drug efflux systems are less active in the ⌬pdr16,⌬pdr17 strain. The observation, however, that energydepleted ⌬pdr16,⌬pdr17 cells exhibit an increased rate of rhodamine-6-G uptake as compared with wild-type and ⌬pdr5 cells suggests, that at least part of the difference in drug sensitivity is due to a difference in passive transport (see Fig.  1). Thus, the increased sensitivity of the double mutant strain ⌬pdr16,⌬pdr17 is at least partially, and perhaps entirely, due to an increased passive uptake of drugs into the cell.
Increase in passive uptake of drugs into the cell might be explained by the changed lipid composition of the plasma membrane of the mutants. Both PDR16 and PDR17 appear to affect the lipid composition of the plasma membrane, although in a different manner. Whereas deletion of the PDR16 gene mostly affects the sterol composition (Table V) deletion of PDR17 rather alters the phospholipid composition of the plasma membrane (Table IV). Most strikingly, the ratio of the negatively charged phospholipids, PtdIns and PtdSer, to the uncharged phospholipids, PtdCho and PtdEtn, is dramatically increased  in the ⌬pdr17 deletion strain. To distinguish between the influence of both mutations on lipid synthesis and transport to the plasma membrane, the bulk membrane lipid composition was compared with that of the plasma membrane. The changes in plasma membrane lipid composition reflect to a large extent those of total membranes. Thus, mutation of PDR16 and PDR17 rather appear to influence synthesis than transport of lipids.
In rich medium the changes of the lipid composition in the plasma membrane in the double mutant do not cause a severe growth defect. Local replacement of certain lipids appears to compensate for deficiencies caused by the ⌬pdr16,⌬pdr17 mutations. Since the level of total membrane PtdCho in the ⌬pdr16,⌬pdr17 strain is also increased it is most likely that the mutations cause alterations in the biosynthesis of PtdCho, probably by regulating the pathway in an as yet unknown way. The effect of PDR16 on sterol biosynthesis could be direct by probing local sterol concentrations and influencing the activity of ergosterol synthesizing enzymes, or by modulating the local availability of sterol precursors. Alternatively, the effect of PDR16 could be indirect by changing phospholipid levels in the endoplasmic reticulum in such a way that enzymes involved in sterol synthesis function less well. This view is supported by the finding that the levels of ergosterol precursors are much higher in the endoplasmic reticulum of the double mutant strain ⌬pdr16,⌬pdr17 as compared with the corresponding wild-type strain FY1679 -28c. 3 Both the PDR16 and PDR17 genes exhibit homology to SEC14. In contrast to the sec14 mutation, however, deletions of PDR16 and PDR17 are not lethal. Furthermore, the cellular level of Sec14p in wild-type and pdr1-3 background is the same independent of the intactness of PDR16 and/or PDR17 3 indicating that the expression of Sec14p is not regulated at the same level as that of the PDR16 gene product. Sec14p is a regulator of PtdCho synthesis through the Kennedy pathway (11,55) or PtdCho turnover (12,13). The imbalance of the PtdCho level in the Golgi caused by SEC14 dysfunction was shown to negatively influence the formation of Golgi-to-plasma membrane secretory vesicles. Further investigation will be needed to demonstrate whether Pdr16p and Pdr17p have functions similar to Sec14p, i.e. modulation of lipid levels in subcellular compartments.