Casein Kinase I-dependent Phosphorylation and Stability of the Yeast Multidrug Transporter Pdr5p*

The pleiotropic drug resistance protein, Pdr5p, is an ATP-binding cassette transporter of the plasma membrane ofSaccharomyces cerevisiae. Overexpression of Pdr5p results in increased cell resistance to a variety of cytotoxic compounds, a phenotype reminiscent of the multiple drug resistance seen in tumor cells. Pdr5p and two other yeast ATP-binding cassette transporters, Snq2p and Yor1p, were found to be phosphorylated on serine residuesin vitro. Mutations in the plasma membrane-bound casein kinase I isoforms, Yck1p and Yck2p, abolished Pdr5p phosphorylation and modified the multiple drug resistance profile. We showed Pdr5p to be ubiquitylated when overexpressed. However, instability of Pdr5p was only seen in Yck1p- and Yck2p-deficient strains, in which it was degraded in the vacuole via a Pep4p-dependent mechanism. Our results suggest that casein kinase I activity is required for membrane trafficking of Pdr5p to the cell surface. In the absence of functional Yck1p and Yck2p, Pdr5p is transported to the vacuole for degradation.

The pleiotropic drug resistance protein, Pdr5p, is an ATP-binding cassette transporter of the plasma membrane of Saccharomyces cerevisiae. Overexpression of Pdr5p results in increased cell resistance to a variety of cytotoxic compounds, a phenotype reminiscent of the multiple drug resistance seen in tumor cells. Pdr5p and two other yeast ATP-binding cassette transporters, Snq2p and Yor1p, were found to be phosphorylated on serine residues in vitro. Mutations in the plasma membrane-bound casein kinase I isoforms, Yck1p and Yck2p, abolished Pdr5p phosphorylation and modified the multiple drug resistance profile. We showed Pdr5p to be ubiquitylated when overexpressed. However, instability of Pdr5p was only seen in Yck1p-and Yck2p-deficient strains, in which it was degraded in the vacuole via a Pep4p-dependent mechanism. Our results suggest that casein kinase I activity is required for membrane trafficking of Pdr5p to the cell surface. In the absence of functional Yck1p and Yck2p, Pdr5p is transported to the vacuole for degradation.
Overexpression of human P-glycoprotein MDR1 is responsible for the phenomenon of multiple drug resistance in various resistant cell lines and tumors. P-glycoprotein is an integral plasma membrane protein that acts as an ATP-dependent efflux pump to extrude seemingly unrelated hydrophobic compounds from the cell interior, thus conferring resistance to a large variety of drugs (1,2). It belongs to the ATP-binding cassette (ABC) 1 transporter superfamily and consists of a tandem repeat of transmembrane domains and conserved nucleotide-binding motifs, connected by a central linker region (1,3,4).
The linker region contains four serine residues (Ser 661 , Ser 667 , Ser 671 , and Ser 683 ), which are targets for in vitro phosphorylation by protein kinase C and/or protein kinase A (5-7). Site-directed mutagenesis of the phosphorylatable serine resi-dues has shown that phosphorylation does not play an essential role in the multidrug resistance phenotype (8) and that phosphorylation of the linker region does not alter the maximum ATPase activity or its dependence on the ATP concentration. However, phosphorylation modulates interaction with certain drugs by increasing their apparent affinity at low concentrations (9).
The yeast Saccharomyces cerevisiae contains 29 -31 ABC proteins (10,11). The closest homolog of human P-glycoprotein is the a-pheromone transporter, Ste6p (12,13). In the same family of multidrug resistance-associated proteins is found Yor1p, the overexpression of which leads to increased growth resistance to oligomycin (14) and organic ions (15). Two other yeast ABC transporters, Pdr5p (16,17) and Snq2p (18), are considered as functional counterparts of MDR1 and MRP1, although they are classified in a third ABC family in regard to their limited sequence identity and an inverted topology of the nucleotide-binding folds relative to the transmembrane domains. Elevated levels of Pdr5p and Snq2p lead to growth resistance to a distinct subset of structurally unrelated xenobiotics (16 -21). Gene expression of both transporters is regulated by the transcription factors, Pdr1p and Pdr3p (16,22,23). The gain-of-function pdr1-3 allele has been identified on the basis of its ability to increase cycloheximide resistance as a result of Pdr5p overexpression (24).
Ste6p (25,26), Pdr5p (27,28), and Yor1p (29) have been reported to be short lived proteins as a result of their modification by ubiquitin. Ubiquitination of plasma membrane proteins acts as a signal to initiate their endocytosis and subsequent Pep4p-dependent degradation in the vacuole (30). Casein kinase I-mediated phosphorylation is required for efficient constitutive endocytosis of the yeast mating pheromone receptors, Ste3p (31) and Ste2p (32), and the uracil permease, Fur4p (33), but it is still unknown whether casein kinase I phosphorylates these proteins. The replacement of serine residues by alanine in a PEST-like sequence of Fur4p (33) and in a sequence surrounding a ubiquitination site of Ste2p (34) decreases the level of phosphorylation and ubiquitination of each protein.
In this study, we show that Pdr5p, like Yor1p and Snq2p, is subjected to phosphorylation in vitro. This phosphorylation is abolished by mutations in the YCK1 and YCK2 genes encoding plasma membrane-bound casein kinase I isoforms. Loss of Yck1p and Yck2p function modulates the multiple drug resistance phenotype of Pdr5p. Phosphorylation probably protects Pdr5p against vacuolar degradation, as suggested by the abnormal instability of Pdr5p in the casein kinase I-deficient strains.

MATERIALS AND METHODS
Strains and Media-The S. cerevisiae strains used are listed in Table  I. The casein kinase I mutants, LRB266 (yck1⌬), LRB343 (yck2⌬), and LRB346 (yck1⌬ yck2 ts ), are isogenic to the parental strain, LRB341 (35). The pep4::Kan R null allele (see below) was transformed into the LRB346 strain, using the lithium acetate method (36), to give the ROY1 strain. Geneticin-resistant transformants were isolated on YPD containing 200 g/ml of Geneticin (Life Technologies, Inc.). Correct integration of the Kan R marker at the PEP4 target site was confirmed by PCR amplification of genomic DNA prepared from the Geneticin-resistant transformants (37). The yck1⌬ (ADK1) and yck1⌬ yck2 ts (ADK12) mutant strains were derived from AD124567 as follows. The 4.5-kilobase pair EcoRI/XhoI fragment of the pAD-⌬YCK1 plasmid carrying the yck1⌬::hisG::URA3::hisG null allele (see below) was used to replace the chromosomal YCK1 gene of the AD124567 strain. Yeast transformants were grown overnight in rich medium and plated onto 5-fluoroorotic acid-containing medium to allow the isolation of ADK1 (yck1⌬::hisG) from the Ura Ϫ clones. The pL2.330-URA3 plasmid carrying the yck2 ts ::URA3 construct (see below) was then cut with SacI and XbaI and introduced into the ADK1 strain to replace the YCK2 gene with the yck2 ts allele, yielding the ADK12 strain. Correct gene disruption was checked by PCR as described above.
For multiple drug resistance assays, cultures grown overnight at 23°C in YD medium were spotted onto YD medium supplemented with 1 or 2 g/ml cycloheximide; 500 or 750 g/ml fluconazole; 15 or 20 g/ml ketoconazole; or 0.1, 2, or 3 g/ml miconazole using a 32-well replicator, and then the plates were incubated at 23°C for 3 days. For determination of cycloheximide minimal inhibitory concentrations (MIC), cultures were grown in minimal SD medium at 23°C to a density of 1-2 ϫ 10 7 cells/ml. Five l of undiluted and 100-fold diluted samples were spotted onto YD or SD medium supplemented with increasing concentrations of cycloheximide, and the plates were incubated at 23, 28, or 37°C for 3 days.
Isolation of Plasma Membranes and Nucleoside Triphosphatase Assays-Plasma membrane-enriched fractions were prepared from cells grown to late exponential phase as described previously (40). Protein concentrations were determined by the Lowry method (41), using bovine serum albumin as the standard. Nucleoside triphosphatase activity was assayed at 35°C in 100 l of 6 mM NTP (ATP or UTP), 7 mM MgSO 4 , 10 mM NaN 3 , 50 mM Mes/Mops/Tris adjusted to the indicated pH with either HCl or NaOH (42).
Rhodamine B Fluorescence Measurements in Intact Cells-Rhodamine B accumulation and extrusion experiments were carried out as described previously (43).
In Vitro Phosphorylation of Plasma Membranes-Plasma membrane-enriched fractions containing 50 g of protein were incubated for 15 min at 35°C in 50 l of 6 mM MgCl 2 , 10 mM NaN 3 , 10 M [␥-32 P]ATP (300 Ci/mmol), 17 mM Mes/Mops/Tris, at the indicated pH. The phosphorylation reaction was stopped by the addition of 25 l of 3-fold concentrated Laemmli sample buffer (240 mM Tris, pH 6.8, 30% glycerol, 6% SDS, 3% dithiothreitol, 0.015% bromphenol blue) and analyzed by electrophoresis on a 7% SDS-polyacrylamide gel (PAGE), and then the gel was dried and subjected to autoradiography at Ϫ80°C. In vitro phosphorylated plasma membranes (150 g of protein) from the US50 -18C and AD3 strains were subjected to SDS-PAGE. Gel bands corresponding to 32 P-labeled ABC transporters (M r 160,000) and Pma1p (M r 100,000) were cut into small pieces, and the proteins were eluted onto a dialysis membrane by electrophoresis for 20 h at 121 V in 25 mM Tris, 192 mM glycine, 0.1% (w/v) SDS, pH 8.3 (HCl). The eluted proteins were hydrolyzed in 800 l of 6 N HCl for 1.5 h at 110°C, and then the hydrolysate was dried and redissolved in 3 l of chromatography buffer (pyridine/AcOH/water, 10:100:1890) containing 1 mg/ml phosphoserine, phosphothreonine, and phosphotyrosine. The amino acid residues were separated by thin layer isoelectric focusing for 45 min at 750 V on cellulose plates, using a double-chamber apparatus (Desaga, Germany). Then the positions of the radioactive residues and amino acid standards were detected by autoradiography at Ϫ80°C and spraying with 0.2% ninhydrin in acetone, respectively (44).
Immunoprecipitation and Immunoblotting of Plasma Membranes-Plasma membranes were incubated for 15 min at 56°C in 200 l of solubilization buffer (50 mM Hepes, pH 7.5, 5 mM EDTA, pH 8.0, 0.25% SDS). After centrifugation at 14,000 rpm for 10 min in an Eppendorf microcentrifuge (4°C), the solubilized membrane proteins were diluted 2.5 times with immunoprecipitation buffer (50 mM Hepes pH 7.5, 5 mM EDTA, pH 8.0, 150 mM NaCl, 1% Triton X-100) containing 10 l of polyclonal anti-Pdr5p antibodies 2 that had been affinity-purified on polyvinylidene difluoride blots of Pdr5p protein separated by SDS-PAGE as described previously (45). After overnight incubation at 4°C on an end-over-end stirrer, 20 l of a 50% (v/v) suspension of protein A-Sepharose beads (Amersham Pharmacia Biotech) were added, and the mixture was incubated at 4°C for 1 h. The immune complexes bound to the protein A-Sepharose beads were washed four times with immunoprecipitation buffer containing 0.1% SDS, resuspended in 15 l of 3-fold concentrated Laemmli sample buffer, and heated to 56°C for 15 min.
After centrifugation at 14,000 rpm for 1 min at room temperature in an Eppendorf centrifuge, plasma membrane and immunoprecipitate fractions were subjected to SDS-PAGE (8% polyacrylamide gel), transferred to nitrocellulose membranes (Schleicher & Schuell), and probed with purified polyclonal anti-Pdr5p or monoclonal mouse anti-ubiquitin (Zymed Laboratories Inc.) antibodies at respective dilutions of 1:2,000 and 1:200. Immunoreactive complexes were detected by chemiluminescence, using anti-rabbit or anti-mouse IgG antibodies conjugated to horseradish peroxidase (1:25,000 dilution; Roche Molecular Biochemicals). When required, the bound antibodies were removed according to the manufacturer's instructions, and the membrane was reprobed with polyclonal anti-Pdr5p or anti-Pma1p antibodies (1:10,000 dilution) (46) as described above.
Pulse-Chase Experiments-Cells were grown at 23°C overnight in minimal SD medium to an optical density at 600 nm (A 600 ) of 0.5-1.0. Ten ml of cells were harvested, resuspended in 400 l of SD and metabolically labeled by adding 215 Ci of Pro-Mix L-[ 35 S] labeling mix (Amersham Pharmacia Biotech) and shaking at 23°C for 15 min (pulse). After washing twice with SD medium containing 10 mM Lmethionine (chase medium), they were resuspended in 400 l of chase medium and incubated at 23 or 28°C to impose the yck1⌬ yck2 ts block. Time point aliquots were transferred to chilled tubes containing 400 l of lysis buffer (50 mM Hepes, pH 7.5, 5 mM EDTA, pH 8.0, 1 mM phenylmethylsulfonyl fluoride, and 2 g/ml each of chymostatin, leupeptin, antipain, and pepstatin) and 400 l of glass beads, and then the cells were lysed by vigorous vortexing (2 ϫ 1.5 min with a 1-min cooling interval). SDS was then added to a final concentration of 0.25%, and the mixture was again vortexed (2 ϫ 1 min). The lysates were clarified by centrifugation at 14,000 rpm for 10 min in an Eppendorf microcentrifuge (4°C) and diluted 2.5 times with immunoprecipitation buffer. Samples containing the same amount of trichloroacetic acid-insoluble labeled material were processed for immunoprecipitation, using 5 l of polyclonal anti-Pdr5p serum, and SDS-PAGE analysis as described above. The relative intensity of the Pdr5p bands at each time point was quantified by scanning densitometry of the autoradiograms, using Imagemaster1D software (Amersham Pharmacia Biotech) or by phosphor imaging, using the Molecular Imaging System (Bio-Rad).

In Vitro Phosphorylation of Yeast Plasma Membrane ABC
Transporters-The expression levels of several ABC transporters, including Pdr5p, Snq2p, and Yor1p, are abnormally high in pdr1-3 mutant strains (16,42,43,47). Plasma membraneenriched fractions from the pdr1-3 strain, US50 -18C, were incubated with [␥-32 P]ATP and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography (Fig. 1A). Two major bands with molecular weights of 160,000 and 100,000 were found to be phosphorylated in vitro. The 100-kDa polypeptide was the plasma membrane H ϩ -ATPase, Pma1p, previously shown to be phosphorylated by casein kinase I (48), whereas the 160-kDa band consisted of Pdr5p, Snq2p, and Yor1p, as the signal intensity was drastically decreased in the PDR1 wildtype IL125-2B strain, which does not overproduce these three ABC proteins (data not shown), and successive deletions of PDR5 (strain AD3), SNQ2 (strain AD23), and YOR1 (strain AD123) in the US50 -18C strain resulted in a decreasing intensity of the 160-kDa signal (Fig. 1A).
The contribution of Yor1p to the degree of phosphorylation of the 160-kDa band was rather low, most likely reflecting differences in the relative abundance of Pdr5p, Snq2p, and Yor1p. We then examined the phosphorylation of plasma membranes prepared from the SUPERYOR strain (43), which contains multiple deletions of the ABC transporter genes, including SNQ2 and PDR5, and expresses YOR1 from the PDR5 promoter (P PDR5 ::YOR1). The phosphorylation level of Yor1p was increased in the SUPERYOR strain compared with the YOR1 wild-type AD234567 strain, and the labeled band was not detected in membranes from the yor1⌬ derivative, AD1234567 (Fig. 1B).
Taken together, these results indicate that Pdr5p, Snq2p, and Yor1p are subject to phosphorylation in vitro. It is unlikely that the pdr1-3 mutation affects plasma membrane-bound kinase activity, since the level of Pma1p phosphorylation in the PDR1 wild type and pdr1-3 mutant strains is the same (data not shown).
Casein Kinase I-mediated Phosphorylation of Major Plasma Membrane ABC Transporters-The 160-kDa band phosphorylated in the membranes of the PDR5 wild-type (US50 -18C) and pdr5⌬ mutant (AD3) strains was eluted from polyacrylamide gels and subjected to acid hydrolysis. After electrophoresis on thin layer cellulose plates, the phosphorylated residues were identified as serine (Fig. 1C). In a separate control experiment, in which the 100-kDa band was treated similarly, only phosphoserine residues were found, confirming the observation that Pma1p is phosphorylated on serine residues (44,49).
Digestion with trypsin of the 160-kDa band from the pdr1-3 DRI9-T8 membranes and separation of the resulting peptides by reverse phase high performance liquid chromatography (48) revealed the presence of at least four labeled fragments. Microsequencing of one peptide gave the sequence Gln 412 -Thr-Thr-Ala-Asp-Phe-Leu-Thr-Ser-Val 421 of Pdr5p, suggesting that Ser 420 could be a target for in vitro phosphorylation. The sequence of the remaining peptides could not be determined because of low amount or incomplete purification.
Phosphorylation of the 160-kDa band was higher at pH 7.6 than at pH 6.0 ( Fig. 2A). Interestingly, the signal intensity decreased upon the addition of Zn 2ϩ ions, with complete inhibition of phosphorylation being seen at 0.6 mM (Fig. 2B). No inhibition of phosphorylation was seen in the presence of 0.5 mM EGTA or 1 M staurosporine, an inhibitor of protein kinase C (data not shown). However, there is no evidence suggesting the presence of protein kinase C activity in yeast plasma membranes (50,51).
Phosphorylation of Pdr5p Is Impaired in Casein Kinase I-deficient Strains-Plasma membrane-associated casein kinase I activity is encoded by the YCK1 and YCK2 genes (52,53). The absence of either gene has no discernible effect on cell growth, but the loss of both genes is lethal (52). To assess the involvement of casein kinase I in Pdr5p phosphorylation, the yck1⌬ deletion and the yck2 ts temperature-sensitive allele were introduced into the Pdr5p-overexpressing AD124567 strain to give ADK12. This strain also contains deletions of numerous ABC genes, including SNQ2 and YOR1 (see Table I). The yck1⌬ and yck2 ts double mutation conferred growth temperature sensitiv-  snq2⌬ pdr5⌬). B, pdr1-3 strains AD1234567 (yor1⌬), AD234567 (YOR1), and SUPERYOR (P PDR5 ::YOR1). C, plasma membranes (150 g of proteins) from the US50 -18C and AD3 strains were phosphorylated in vitro and separated by SDS-PAGE. The 160-and 100-kDa 32 P-labeled protein bands were subjected to acid hydrolysis, and then the amino acids were separated on a cellulose thin layer chromatography plate. The positions of phosphoserine, phosphothreonine, phosphotyrosine, and free radioactive phosphate (P i ) are indicated. The comparison with amino acid standards shows that only serine residues are phosphorylated in the samples. Separation of the amino acids from the AD3 160-kDa band was less efficient because of uneven migration throughout the thin layer cellulose plate (smiling effect).

FIG. 2. Characteristics of plasma-membrane kinase activity.
A, pH dependence of the phosphorylation of the plasma membrane 160-kDa ABC band from the pdr1-3 DRI9-T8 strain. B, inhibition by Zn 2ϩ ions. Sensitivity to Zn 2ϩ was assayed in the phosphorylation mix containing the indicated ZnCl 2 concentration. Band intensity was quantitated by scanning densitometry with the Image Master 1D (Amersham Pharmacia Biotech) program and is represented as a percentage of the maximal level. ity on ADK12 cells at 30°C (restrictive temperature) and reduced cell growth at 23°C (semipermissive temperature). Note that the restrictive temperature for the ADK12 cells (30°C) is lower than the temperature (37°C) originally described for the yck1⌬ yck2 ts LRB346 strain (35). The ADK1 strain is similar to ADK12, except for the presence of the YCK2 wild-type gene.
Cells of the AD124567 (YCK1 YCK2), ADK1 (yck1⌬ YCK2), and ADK12 (yck1⌬ yck2 ts ) strains were grown overnight at 23°C and then shifted to 30°C for 2 h before preparation of plasma membranes. Each membrane fraction was incubated with [␥-32 P]ATP for 15 min at 35°C, and then the proteins were separated by SDS-PAGE. As shown in Fig. 3, less Pdr5p phosphorylation was seen with ADK1 membranes compared with AD124567 membranes. Pdr5p phosphorylation was blocked in the casein kinase I-deficient ADK12 strain grown at the restrictive temperature. The lowest molecular weight polypeptide of unknown identity was used as a loading control, since its phosphorylation was not mediated by casein kinase I. The yck1⌬ yck2 ts double mutation completely abolished phosphorylation of Pma1p, as previously published (48).
Loss of Casein Kinase I Activity Affects Pdr5p-dependent Multiple Drug Resistance-In contrast to AD124567 (pdr1-3 YCK1 YCK2) cells, ADK12 (pdr1-3 yck1⌬ yck2 ts ) cells failed to grow on solid media supplemented with 2 g/ml cycloheximide, 20 g/ml ketoconazole, or 2 g/ml miconazole (Fig. 4A). The resistance of the ADK12 strain to fluconazole was no lower than that of AD124567; in fact, it grew slightly better than AD124567 in the presence of 750 g/ml fluconazole. No changes in sensitivity to rhodamine B or rhodamine 6G were detected by the plate assay, using rich media containing 2% glucose and glycerol, respectively (data not shown). However, comparison of relative growth curves revealed that the resistance of the ADK12 strain to rhodamine 6G at concentrations up to 500 g/ml was increased in comparison with that of the AD124567 strain (Fig. 4B). The growth of the pdr5⌬ strain, AD1234567, was strongly inhibited by the lowest concentration of drug used (1 g/ml cycloheximide, 500 g/ml fluconazole, 15 g/ml ketoconazole, 0.1 g/ml miconazole, or 50 g/ml rhodamine 6G) ( Fig. 4A and data not shown), confirming that the observed multiple drug resistance phenotype is dependent on Pdr5p. The multiple drug resistance of ADK1 (pdr1-3 yck1⌬ YCK2) cells was less affected than that of the ADK12 cells. For instance, miconazole toxicity was only seen at 3 g/ml compared with 2 g/ml for ADK12 cells.
The finding that Pdr5p-mediated cycloheximide resistance was altered by mutations in YCK1 and YCK2 was reproduced with a different set of casein kinase I-deficient strains harboring a high copy number PDR5 plasmid, pKMR3.3 (20). At the permissive and semipermissive temperatures of 23 and 28°C, respectively, the cycloheximide resistance of LRB346 (yck1⌬ yck2 ts ) cells containing pKMR3.3 was decreased 2-fold compared with that of the parental LRB341 (YCK1 YCK2) strain containing pMKR3.3 (Table II). The LRB346 strain failed to grow at 37°C (restrictive temperature). The MIC was defined as the lowest concentration of cycloheximide (g/ml in the medium) at which inhibition of cell growth was seen in comparison with on a control plate lacking the drug. At 28°C, the MIC for LRB346 cells was only 0.75 g/ml, while it was 1.5 g/ml for LRB341 cells. The LRB343 (yck2⌬) strain containing pMKR3.3 showed an intermediate resistance phenotype with a MIC of 1 g/ml. Loss of functional Yck1p and Yck2p also affects the MIC for cycloheximide in cells that only express chromosomal PDR5 (Table II). However, the cycloheximide resistance of these cells might be mediated by genes other than PDR5 (54).
The UTPase Activity of Pdr5p Is Not Regulated by Casein Kinase I Phosphorylation-Like human P-glycoprotein, Pdr5p

FIG. 3. Phosphorylation of Pdr5p is impaired in plasma membranes of casein kinase I-deficient strains.
Plasma membraneenriched fractions were prepared from the Pdr5p-overexpressing strains AD124567 (YCK1 YCK2), ADK1 (yck1⌬ YCK2), and ADK12 (yck1⌬ yck2 ts ), which were incubated for 2 h at the restrictive temperature. These membrane fractions were assayed for phosphorylation as described in the legend to Fig. 1. shows nucleoside triphosphatase activity (42). Plasma membrane-enriched fractions from the AD124567 (YCK1 YCK2) and ADK12 (yck1⌬ yck2 ts ) strains were assayed for UTPase activity at pH 6.3 to determine whether this activity is dependent on functional Yck1p and Yck2p.
In double-reciprocal plots where the MgUTP concentration was varied from 0.1 to 8 mM, the UTPase activity of AD124567 membranes showed an apparent K m for MgUTP of 1.3 Ϯ 0.5 mM and an estimated V max of 1.1 Ϯ 0.3 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 . ADK12 membranes gave a K m value of 1.1 Ϯ 0.2 mM and a V max value of 1.0 Ϯ 0.2 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 . Thus, the UTPase activity of Pdr5p is not significantly affected by the loss of functional Yck1p and Yck2p. In comparison, the UTPase activity of pdr1-3 pdr5⌬ membranes is 0.4 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 . The remaining activity is associated with Snq2p, since it decreases to 0.1 mol of P i ⅐min Ϫ1 ⅐mg Ϫ1 in pdr1⌬ mutants (42).
The Drug Transport Activity of Pdr5p Is Not Altered in Casein Kinase I-deficient Strains-Since the UTPase activity of Pdr5p was unchanged in casein kinase I-deficient strains, we tested whether the drug transport activity could be altered. For this purpose, we used rhodamine B as a fluorescent substrate of Pdr5p (43).
Cells of the AD124567 (YCK1 YCK2), AD1234567 (a pdr5⌬ derivative of AD124567), ADK1 (yck1⌬), and ADK12 (yck1⌬ yck2 ts ) strains were grown at 23°C (semipermissive temperature) and then loaded with rhodamine B at 30°C (restrictive temperature), under conditions of energy depletion by incubation with 2-deoxy-D-glucose (Fig. 5A). No significant differences in rhodamine B accumulation were seen between the YCK1 YCK2 wild type and yck1⌬ yck2 ts mutant strains. The addition of glucose caused a drastic reduction (to 7%) in the cellular rhodamine B content of all strains, except the pdr5⌬ null mutant, confirming that rhodamine B extrusion is mediated by Pdr5p. No significant differences in drug accumulation were seen between the YCK1 YCK2 wild-type and yck1⌬ yck2 ts mutant strains incubated with either 2-deoxy-D-glucose or glucose. It is therefore unlikely that casein kinase I activity regulates another putative passive drug transport mechanism.
We also measured the Pdr5p-mediated energy-dependent efflux of rhodamine B (Fig. 5B). The cells were first loaded with rhodamine B in the presence of 2-deoxy-D-glucose, and then glucose was added, causing rapid extrusion of rhodamine B from Pdr5p-expressing cells. This glucose-dependent effect was consistently not seen in the pdr5⌬ null mutant. Comparison of the efflux rates seen with YCK1 YCK2 and yck1⌬ yck2 ts cells at 30°C revealed no significant differences.
Decreased Pdr5p Levels in Plasma Membranes from Casein Kinase-deficient Strains-We isolated plasma membrane-enriched fractions from LRB341 (YCK1 YCK2), LRB343 (YCK1 yck2⌬), and LRB346 (yck1⌬ yck2 ts ) cells containing the high copy number PDR5 plasmid (Fig. 6). The LRB341 and LRB343 cells were grown at 23°C, while the LRB346 cells were either grown at 23°C and then shifted to 32 and 37°C for 3 h to heat-inactivate casein kinase I, or left at 23°C. The amount of Pdr5p in each membrane fraction was analyzed by immunoblotting, using affinity-purified antibodies against Pdr5p or Pma1p; immunological detection of Pma1p served as an internal standard for equal protein loading. As shown in Fig. 6, the steady-state level of Pdr5p was affected by the yck2⌬ deletion (LRB343) and yck1⌬ yck2 ts double mutation (LRB346), decreasing to about 70% of the wild-type level in LRB346 cells grown at the restrictive temperature. Similar results were obtained with the ADK12 strain (data not shown).
Protein Stability of Pdr5p Requires Casein Kinase I Activity-The reduced steady-state levels of Pdr5p in yck1⌬ yck2 ts cells suggest that Pdr5p in these cells might be less stable than in the wild-type strain. To directly test this possibility, pulsechase experiments were performed on YCK1 YCK2 (LRB341) and yck1⌬ yck2 ts (LRB346) cells containing the high copy PDR5 plasmid (Fig. 7).
In striking contrast to early reports showing degradation of Pdr5p with a half-life of 60 -90 min (27,28), we found Pdr5p to be stable in the LRB341 cells. The fraction of Pdr5p remaining 180 min after chase initiation was approximately 80 -90%, the degradation occurring within the first hour. The same result (not shown) was obtained with other parental strains of either European (⌺1278B, IL125-2B) or North American (W303a, FY1679-28C) origin and expressing Pdr5p from a high copy number plasmid or the chromosomal gene. In the latter case, however, the Pdr5p signal was so weak that the results should be interpreted with great caution. In a separate control exper-  (PDR5 yck1⌬ YCK2), and ADK12 (PDR5 yck1⌬ yck2 ts ) were spotted onto YD plates containing no drug; 1 or 2 g/ml cycloheximide; 500 or 750 g/ml fluconazole; 15 or 20 g/ml ketoconazole; or 0.1, 2, or 3 g/ml miconazole, and cell growth was measured after a 72-h incubation at the semipermissive temperature of 23°C. B, overnight cultures (5 ϫ 10 5 cells) of the AD1234567 (circles), AD124567 (squares), and ADK12 (triangles) strains were diluted in 2.5 ml of YG medium (2% yeast extract, 4% glycerol) supplemented with increasing concentrations of rhodamine 6G. The number of cells was determined for each culture after a 24-h incubation at 23°C. Cell number values (represented as a percentage of maximal growth in the absence of rhodamine 6G) were plotted against the indicated drug concentration.
iment, when the turnover of chromosomally expressed Ste6p was examined by pulse-chase, we found the half-life to be 48 min (data not shown), comparable with the published value of 39 min (25).
In yck1⌬ yck2 ts (LRB346) cells incubated at a semipermissive temperature (28°C), the Pdr5p protein was stable for a chase period up to 2 h and was then quickly degraded (half-life of 60 min). The delay in degradation could be explained by the time required for Pdr5p to reach the vacuole (see below) from an internal compartment where it transiently accumulates. At 23°C, however, no degradation of Pdr5p could be seen. Similar results were seen with Pdr5p-overexpressing pdr1-3 strains (data not shown).
To determine whether Pdr5p was degraded in the vacuole, as previously shown (27,28), we constructed a yck1⌬ yck2 ts strain containing a deletion of the PEP4 gene (strain ROY1). PEP4 is required for maturation of vacuolar proteases, and its deletion results in the stabilization of proteins that are normally degraded in the vacuole. Pulse-chase analysis showed that Pdr5p instability in yck1⌬ yck2 ts cells grown at 28°C was suppressed by the pep4⌬ deletion in ROY1 cells (Fig. 7), consistent with vacuolar degradation of Pdr5p.
We failed to detect Pdr5p ubiquitination in PDR1 PDR5 wild-type (IL125-2B) plasma membranes (Fig. 8A). Western blot analysis of plasma membrane fractions indicated that the amount of Pdr5p in pdr1-3 US50 -18C cells was increased 10-fold in comparison with IL125-2B cells (Fig. 8B). Consequently, 10-fold more plasma membrane protein (150 g) from the IL125-2B strain was subjected to immunoprecipitation with anti-Pdr5p antibodies, followed by immunoblotting of the immunoprecipitates with anti-ubiquitin antibodies (Fig. 8C); no signal indicative of Pdr5p ubiquitination was seen, indicating that the lack of signal was not due to a low expression level. As shown in Fig. 8D, the amount of immunoprecipitated Pdr5p (IP) appeared to be almost the same as that originally present in 15 g of IL125-2B membranes (PM), i.e. 10-fold more proteins than in the US50 -18C membranes (1.5 g). We also analyzed the ubiquitination level of Pdr5p in membranes of LRB341 (YCK1 YCK2) and LRB346 (yck1⌬ yck2 ts ) cells containing the high copy PDR5 plasmid. Pdr5p ubiquitination was detectable in both strains grown at 23 or 32°C, suggesting that Pdr5p ubiquitination does not require functional Yck1p and Yck2p (data not shown).

DISCUSSION
In this study, we demonstrated that the yeast ABC transporter, Pdr5p, was phosphorylated by casein kinase I in vitro. First, the major phosphorylation sites within Pdr5p were identified as serine residues. Second, Pdr5p phosphorylation was prevented either by 0.1 mM Zn 2ϩ , which inhibits casein kinase I activity (44), or by mutations in the YCK1 and YCK2 genes, which encode plasma membrane-bound isoforms of casein kinase I (52,53). One of the phosphorylatable residues, Ser 420 , was found within the sequence Thr 419 -Ser-Val-Thr-Ser-Pro 424 , which conforms with the casein kinase I phosphorylation site motif, Xaa-Ser(P)-Xaa-Xaa-Ser*-Xaa, in which the asterisk indicates the phosphorylated residue and (P) denotes a phosphorylated residue that acts as a substrate specificity determinant (55). Ser 420 and Ser 423 are located in the first nucleotide-binding domain, between the Walker B motif and the first transmembrane span. In comparison with human P-glycoprotein, this region would be analogous to the linker region that contains the phosphorylated residues Ser 661 , Ser 667 , and Ser 671 (5)(6)(7).
Inactivation of plasma membrane-associated casein kinase I caused a discernible change in the multiple drug resistance profile of wild type and Pdr5p-overexpressing strains. Interestingly, resistance to cycloheximide and antifugal azoles, such as miconazole and ketoconazole, was reduced, while the reverse effect was seen for fluconazole and rhodamine 6G. Such a change in drug specificity has been reported for mutations within the last extracellular loop of Pdr5p (56). The multiple drug resistance profile of Pdr5p could be modified by phosphorylation of serine residues within the first nucleotide binding, which regulates its interaction with the drug-binding site(s) in or near the transmembrane region, through the introduction of negatively charged phosphate groups.
We found that the steady-state levels of overexpressed Pdr5p were slightly decreased in plasma membranes from cells lacking functional Yck1p and Yck2p. We also found that the reduced levels of Pdr5p in casein kinase I-defective cells resulted from protein instability and subsequent degradation in the vacuole at 28°C. Pdr5p has been reported to be a short lived protein with a half-life of 60 -90 min (27,28). In striking con- , and AD1234567 (a pdr5⌬ derivative of AD124567) were processed for immunoprecipitation, using anti-Pdr5p antibodies. The Pdr5p immunoprecipitates (IP) were separated by SDS-PAGE and analyzed by immunoblotting with monoclonal mouse antiubiquitin antibodies. B, Western blot analysis of plasma membrane fractions (1.5 g of protein) using anti-Pdr5p antibodies. C and D, IL125-2B plasma membranes (15 g of protein) and Pdr5p immunoprecipitates from 150 g of IL125-2B membrane proteins were separated by SDS-PAGE and analyzed on a nitrocellulose membrane probed with anti-ubiquitin antibodies (C). After removal of the anti-ubiquitin antibodies, the same membrane was probed with anti-Pdr5p antibodies (D). The two contaminant bands in Pdr5p immunoprecipitates correspond to rabbit IgG reacting with secondary anti-rabbit IgG antibodies. trast to these early reports, but in accordance with a recent study (57), we found that, in YCK1 YCK2 wild-type cells, 80 -90% of the initial Pdr5p was still present 90 -180 min after the chase. This discrepancy could result from subtle differences in the conditions and/or antibodies (directed against the N-terminal or central part of Pdr5p) used by the two laboratories. The stability of Pdr5p does not depend on expression levels, since similar results were obtained with chromosomal or high copy PDR5. On the other hand, only overexpressed Pdr5p was found to be ubiquitinated, as described previously (28). The absence of multiple ubiquitin-protein conjugates indicates that Pdr5p does not carry polyubiquitin chains. It is possible that the cell may monoubiquitinate, rather than polyubiquitinate, Pdr5p as a protection mechanism to avoid recognition by the endocytic machinery (58). Under physiological (high temperature, presence of drugs) or artificial (overexpression of ubiquitin from the CUP1 promoter, utilization of a hemagglutinin-tagged protein) conditions, Pdr5p could become polyubiquitinated and, thus, targeted for vacuolar degradation. We found that the ubiquitination level of Pdr5p was not affected by loss of casein kinase I activity, consistent with degradation of the ABC transporter in yck1⌬ yck2 ts mutant cells at semipermissive temperature. It is therefore unlikely that the ubiquitination of Pdr5p requires phosphorylation by Yck1p and Yck2p kinases. Casein kinase I activity is required for vesicle transport at the plasma membrane, a process that involves a newly discovered clathrin adapter-related complex (31). It is reasonable to suppose that the principal role of Pdr5p phosphorylation is to ensure that the ABC transporter is properly delivered to the cell surface.
Phosphorylation could act as a sorting signal, allowing internalized Pdr5p to return to the plasma membrane. Recent evidence suggests the existence of a recycling endosomal compartment (59, 60), but it is still unknown whether transport from the endosome to the cell surface is direct or indirect, through the Golgi apparatus. The post-Golgi prevacuolar compartment, which is defined by class E VPS gene products, constitutes a point of intersection between the endocytic and vacuole biogenesis pathways (61,62). Loss of Yck1p and Yck2p function could also block transport of newly synthesized Pdr5p in late Golgi. Consequently, the mislocalized molecules would be targeted to vacuolar degradation, probably through the prevacuolar compartment. Further studies, including mutational analysis of the serine residues within the Pdr5p sequence that are phosphorylated in vivo and subcellular localization of the Pdr5p variants, are needed to determine the exact role of the casein kinase I-mediated phosphorylation of Pdr5p in membrane trafficking and vacuolar degradation.