Functional Expression of the Multidrug Resistance-associated Protein in the Yeast Saccharomyces cerevisiae *

The multidrug resistance-associated protein (MRP) is a member of the ATP binding cassette superfamily of transporters which includes the mammalian P-glyco- proteins (P-gp) family. In order to facilitate the biochemical and genetic analyses of MRP, we have ex- pressed human MRP in the yeast Saccharomyces cerevisiae and have compared its functional properties to those of the mouse Mdr3 P-gp isoform. Expression of both MRP and Mdr3 in the anthracycline hypersensitive mutant VASY2563 restored cellular resistance to Adriamycin in this mutant. MRP and Mdr3 expression pro- duced pleiotropic effects on drug resistance in this mutant, as corresponding VASY2563 transformants also acquired resistance to the anti-fungal agent FK506 and to the K (cid:49) /H (cid:49) ionophore valinomycin. The appearance of increased cellular resistance to the toxic effect of Adria- mycin (ADM) in MRP and Mdr3 transformants was concomitant with a reduced intracellular accumulation of [ 14 C]ADM in spheroplasts prepared from these cells. Moreover, MRP and Mdr3, but not control spheroplasts, could mediate a time-dependent reduction in the overall cell-associated [ 14 C]ADM from preloaded cells, suggesting the presence 2% glucose (2 ml, in a 6-ml push-cap tube). Equal aliquots were plated in parallel on SD-Ura plates (number of total haploid JPY201 cells) and on MM plates (number of JPY201 diploid cells), and the ratio of the two numbers defined the mating frequency. Analytical Methods— Immunofluorescence analysis of yeast transformants expressing either the MRP or the Mdr3 proteins was performed according to a standard protocol (32). The anti-MRP antiserum MRP-2 (15), and the anti-P-gp antiserum ES4 (33), were used for the detection of the two proteins.

brane protein of approximately 170 kDa which binds drugs, binds and hydrolyzes ATP, and that functions as a drug transport system to reduce the intracellular accumulation of structurally unrelated drugs in resistant cells (2). The emergence of MDR in the absence of P-gp expression in a number of cell lines has been associated with the overexpression of another membrane protein, the multidrug resistance-associated protein or MRP (3). MRP was initially discovered in, and its cDNA cloned from, the H69AR small cell lung carcinoma cell line selected for increasing levels of resistance to the anthracycline Adriamycin (ADM) (3,4). The protein was subsequently found to be overexpressed in several other multidrug-resistant cell lines (5)(6)(7)(8). Transfection experiments in drug-sensitive cells have clearly established a direct causal relationship between MRP overexpression and pleiotropic resistance to drugs, but have also shown that the MDR phenotype conveyed by MRP, although similar, is distinct from that encoded by P-gp (9 -11). Drug accumulation and efflux studies in MRP-transfectants have identified an ATP-dependent decrease in drug accumulation associated with a concomitant increase in drug efflux from preloaded cells (10,11). Studies in inside-out membrane vesicles have indicated that MRP can transport in an ATP-dependent fashion leukotriene C 4 (12,13), and several other types of glutathione S-conjugates (12)(13)(14). This transport can be inhibited by a leukotriene D 4 receptor antagonist, and also by relatively high concentrations of cyclosporin A and PSC833, the latter two being known modulators of P-glycoprotein-mediated MDR. These findings have led to the suggestion that MRP may be the multiple organic anion transporter and may therefore transport drug molecules as GSH conjugates (13).
Predicted amino acid sequence analysis of MRP identifies an integral membrane protein (minimal size, 170 kDa), with two putative nucleotide binding domains and two large highly hydrophobic, probably membrane-associated regions (3,7). It has been confirmed biochemically that MRP is an integral membrane glycophosphoprotein capable of binding ATP (15). These characteristics and its predicted primary amino acid sequence place it in the ATP binding cassette superfamily of membrane transporters, known as the ABC superfamily (16). In prokaryotes, these transporters form the large group of bacterial periplasmic permeases or traffic ATPases, that import or export at the expense of ATP, a large number of structurally unrelated substrates (19). In eukaryotes, the best characterized ABC transporters include the mating pheromone a transporter Ste6 of the yeast Saccharomyces cerevisiae (20), the pfmdr1 gene of Plasmodium falciparum associated with chloroquine resistance in the malarial parasite (21), the CFTR chloride channel in which mutations cause cystic fibrosis in humans (18), the TAP1 and TAP2 peptide pumps participating in antigen presentation in lymphocytes (22), and the family of mammalian P-gps (1,2). A number of eukaryotic proteins of this group, including P-gps (2) and CFTR (18), appear to be formed by two relatively similar halves each encoding six putative transmembrane domains and one nucleotide binding site (17).
We have previously used the yeast S. cerevisiae as to characterize the mechanism of action of the three murine P-gp isoforms (23)(24)(25)(26)(27). The Mdr1 and Mdr3 P-gp isoforms expressed in yeast bind drug analogs and confer cellular resistance to known MDR drugs (e.g. the anti-fungal macrolite peptide FK506) by reducing their intracellular accumulation (23,24). The structural similarity between P-gps and the endogenous yeast Ste6 peptide pheromone transporter translates into functional homology, since the mouse Mdr3 P-gp isoform can complement the biological activity of Ste6 and partially restore mating in a sterile ste6 null mutant (23). Studies of P-gp expressed in the membrane of secretory vesicles from the yeast mutant sec6-4 have shown that P-gp uses direct ATP hydrolysis to transport drug molecules across the membrane, this being independent of an intact membrane electrochemical gradient (⌬, ⌬pH) (26). Finally, using the same expression system, we have been able to show that the normal physiological role of the liver-specific P-gp isoform Mdr2 is an ATP-dependent phosphatidylcholine flippase that translocates lipid molecules from the inner to the outer leaflet of the membrane lipid bilayer (27).
The exact mechanism by which MRP confers multidrug resistance and the normal physiological function of this protein are not understood. To facilitate the functional dissection of this protein in a system that would permit biochemical and genetic analyses, we have expressed MRP in normal and mutant strains of yeast. We have compared in this system the phenotypic characteristics acquired by MRP transformants to those observed in the same hosts expressing the mouse Mdr3 P-gp isoform, with respect to cellular resistance to known cytotoxic drugs, accumulation, and release of radiolabeled ADM and complementation of a null mutation at the endogenous Ste6 gene.

EXPERIMENTAL PROCEDURES
Materials-Zymolyase-100T was purchased from ICN Biomedicals; concanavalin A and Ficoll 400 were obtained from Pharmacia Biotech Inc. [14-14 C]Doxorubicin hydrochloride (generic name; [ 14 C]Adriamycin: trade name) (53 mCi/mmol; 1.96 GBq/mmol) was purchased from Amersham Corp. Protease inhibitors, creatine kinase and phosphocreatine, were obtained from Boehringer Mannheim. All other chemicals and biochemicals were of high quality reagent grade.
Yeast Strains, Plasmids, Transformation, and Culture Conditions-The S. cerevisiae strain JPY201 (MATa, ste6⌬::his3) and VASY2563 (MAT␣, his4 -15, lys9, ura3-52, erg6⌬, leu2::his G, rad52::LEU2) were used throughout this study. For high level protein expression the plasmid vector pVT101-U (pVT) was used to transform both strains (28). A full-length cDNA for the human MRP gene (9) was introduced in the PvuII site of pVT 101-U as a SacII/KpnI fragment repaired with T4 DNA polymerase, using standard cloning techniques, to create plasmid pVT-MRP. The pVT-mdr3S and pVT-mdr3F constructs containing wild type and mutant mdr3 cDNA inserts, respectively, have been described elsewhere (23). The MRP-GFP fusion protein was constructed by introducing the green fluorescent protein (GFP) (29) cDNA (Columbia Innovation Enterprise, New York, NY) as a full-length fragment (repaired with T4 DNA polymerase) into the unique NaeI site of the MRP cDNA at position 4780, which is positioned 9 nucleotides upstream of the stop codon. DNA transformation of JPY201 cells was performed by the lithium acetate method of Ito et al. (30). The transformation of the VASY2563 strain required the following modifications (31) of the standard protocol (30). Cultures were grown to midlogarithmic phase in YPD (1% yeast extract, 2% Bactopeptone, 2% glucose), diluted to A 600 ϭ 0.8, and the cells were further incubated for 3 h at 30°C with constant shaking. Aliquots (250 ml) were harvested, washed once in lithium acetate buffer (LIAT buffer: 100 mM lithium acetate, 10 mM Tris-HCl, pH 8.0, 1 mM EDTA) and resuspended in a final volume of 1 ml of LIAT buffer. In a 1.5-ml Microfuge tube, 10 -20 g of plasmid DNA of the various constructs in 10 l of LIAT buffer was mixed with 40 g (in 10 l) of freshly prepared total yeast RNA (32) and 150 l of the yeast cell suspension. After incubation for 2 h at 30°C, the mixture was suspended in 0.7 ml of 36% polyethylene glycol, 100 mM lithium acetate, 10 mM Tris-HCl, pH 8.0, and incubated for an additional 2 h at 30°C. The samples were then heat-shocked for 15 min at 42°C, the cells pelleted, washed once in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, and plated on synthetic medium lacking uracil (SD-Ura). Five to ten transformants were picked, pooled, grown as a mass culture in SD-Ura medium, followed by freezing at Ϫ80°C in SD-Ura medium supplemented with 30% glycerol. All subsequent cultures were always started from these stock samples and grown either in SD-Ura or YCG medium (0.75% yeast nitrogen base without amino acids, 0.35% bactocasamino acids, 2% glucose) as indicated.
Growth Inhibition Assays-Growth inhibition with cytotoxic drugs was determined in JPY201 transformants in a liquid assay exactly as described previously (24). In the case of VASY2563 transformants, the protocol was modified as follows. Cells grown in YCG medium to midlogarithmic cultures (A 600 ϭ 1.5-2) were diluted in the same medium to exactly A 600 ϭ 0.1. Subsequently, for the liquid growth assay, 50-l aliquots were added to 0.95 ml of YCG medium supplemented with the appropriate drugs at the indicated final concentrations. The samples were grown under constant agitation at 30°C, and time-dependent growth was determined by pipetting 50-l aliquots in 96-well microtiter plates and measuring the optical density (A 595 ) in a standard ELISA plate reader (24). Growth of the various transformants in cytotoxic drugs was calculated as relative growth of the same transformants grown in drug-free medium and is expressed as percentages. For the plating assay, 20-l aliquots were removed after 4-h incubation periods of the various transformants in drug-containing medium, followed by plating onto YCG agar plates and incubation 48 h at 30°C.
ADM Accumulation in Spheroplasts-JPY201 and VASY2563 cells were grown in YPD, YCG, or LA medium to midlogarithmic phase (A 600 ϭ 1.5-2), harvested, and converted to spheroplasts by Zymolyase treatment, as described (24). The final spheroplast pellets were resuspended in SM1 buffer (1.2 M sorbitol, 2 mM EDTA, 20 mM HEPES-KOH, pH 7.0) to A 600 ϭ 5. For ADM accumulation measurements, aliquots of spheroplasts were diluted 20-fold with their original growth medium (at O°C) and supplemented with 1 M ADM, including 5 nCi/ml [ 14 C]ADM. Drug accumulation was initiated by placing the samples into a 30°C water bath. At 5-min intervals, four 0.5-ml aliquots were removed and applied to a 1.5-ml Eppendorf tube containing 100 l of an oil suspension (consisting of a mixture of mineral and silicon oil, 1:4) overlaid with 0.25 ml of SM2 buffer (containing 0.6 M sorbitol, 2 mM EDTA, 0.5 mg/ml BSA, 20 mM HEPES-KOH, pH ϭ 7.0). After a 15-s spin in a Microfuge, the water and the oil phases were carefully removed with a pipette and the collected spheroplast pellets were dissolved in 0.5 ml of 5% SDS. Cell-associated radioactivity was determined by liquid scintillation counting.
Quantitative Mass Mating Assay-Quantitative mass mating was performed as described previously (23), but with the following minor modifications. Briefly, JPY201 transformants were grown at 30°C in YCG medium to midlogarithmic cultures (A 600 ϭ 1.5-2) and diluted with the same medium to A 600 ϭ 0.6. After an additional growth period of 3 h at 30°C, 0.75-ml aliquots were mixed with 0.25 ml of a midlogarithmic phase culture (A 600 ϭ 1.5) of the DC17 MAT␣ tester strain. One-third of each suspension was then concentrated onto a glass fiber filter (Whatman, GF/C filters), transferred to a YPD plate, and incubated for 3 h at 30°C to allow mating. The cells were then removed from the filters by vigorous agitation in MM medium containing 0.68% yeast nitrogen base without amino acids and 2% glucose (2 ml, in a 6-ml push-cap tube). Equal aliquots were plated in parallel on SD-Ura plates (number of total haploid JPY201 cells) and on MM plates (number of JPY201 diploid cells), and the ratio of the two numbers defined the mating frequency.
Analytical Methods-Immunofluorescence analysis of yeast transformants expressing either the MRP or the Mdr3 proteins was performed according to a standard protocol (32). The anti-MRP antiserum MRP-2 (15), and the anti-P-gp antiserum ES4 (33), were used for the detection of the two proteins.

RESULTS
The aim of this study was to use yeast as an experimental system to express and functionally characterize the MRP.
Yeast expression plasmids pVT containing full-length cDNA clones for human MRP and mouse wild type mdr3 (used as control) were transformed into S. cerevisiae strain JPY201(MATa, ste6⌬::his3). MRP protein expression was monitored by immunofluorescence on permeabilized spheroplasts using a polyclonal antiserum raised against a MRP peptide (15). As shown in Fig. 1A (panel 1), a strong fluorescent signal was detected in JPY201 cells transformed with the pVT-MRP plasmid, which was absent in control cells transformed with pVT (panel 2). Likewise, a strong fluorescent signal was produced by the polyclonal anti-P-gp antiserum ES4 in JPY201 cells transformed with the pVT-mdr3 plasmid (panel 3). While in Mdr3 cells, the fluorescent signal had a ring-like pattern, seemingly associated with cell membrane expression of the protein, the anti-MRP antibody generated a strong additional punctate fluorescence in the cytoplasm, suggesting the presence of MRP protein associated with intracellular organelles.
Although expression of wild type and mutant Mdr3 proteins in yeast plasma membranes can be readily demonstrated by immunoblotting and immunoprecipitation (23,24), similar attempts using a number of well characterized anti-MRP monoclonal or polyclonal antibodies (15) failed to detect the presence of full-length MRP polypeptides (data not shown). These results suggest a high turnover and short half-life of mammalian MRP in yeast cells, resulting in low steady state levels of intact protein. To determine if full-length MRP was indeed synthesized in our yeast strain, we introduced a DNA fragment coding for the naturally fluorescent GFP (29) in-frame at the carboxyl terminus of MRP and determined if a full-length fluorescent MRP-GFP protein could be synthesized in yeast. As shown in panel 4, expression of this MRP-GFP construct in JPY201 cells resulted in the production of a naturally fluorescent polypeptide (ring-like signal), which was also recognized by the anti-MRP antiserum (panel 5). These results suggest that a fulllength recombinant MRP-GFP protein was indeed synthesized in these yeast cells.
We then determined if MRP expression in yeast JPY201 cells could alter sensitivity to cytotoxic drugs. We tested the sensitivity of yeast cells transformed with pVT plasmids carrying full-length cDNAs for wild type mdr3, wild type MRP, or the MRP-GFP chimera to the anti-fungal macrolite FK506 (at 50 g/ml), a known P-gp substrate (24). In these experiments, we included as an additional control a mutant mdr3 cDNA carrying a single Ser to Phe substitution at position 939 (Mdr3F) which shows drastically reduced biological activity in mammalian and yeast cells (34). Control experiments in Fig. 1B (left panel) indicate that growth of Mdr3F cells was completely inhibited by FK506, while growth of Mdr3S cells was unaffected, reaching cellular densities at 24 h similar to those of Mdr3S and Mdr3F cells grown in the absence of drug. Likewise, cells expressing either wild type MRP or the MRP-GFP fusion protein showed a significant degree of resistance to FK506, showing only a small growth retardation (lag phase), but reaching cell densities at 20 h similar to those observed in the absence of drug (Fig. 1B, right panel). These results strongly suggest that MRP overexpression in yeast is sufficient to confer cellular resistance to the anti-fungal drug and Mdr3 substrate FK506, suggesting functional MRP expression in these cells.
A major goal of our study was first, to determine if MRP can directly confer resistance to ADM in yeast (a drug to which MRP confers resistance in mammalian cells), and second, to gain insight into the underlying mechanism of resistance. Preliminary experiments in JPY201 and other commonly used S. cerevisiae laboratory strains indicated that yeast cells are intrinsically resistant to the cytotoxic effect of ADM (and other MDR drugs), precluding the analysis of MRP or Mdr3 activity against ADM in these strains (data not shown). To circumvent this limitation, we took advantage of the unique phenotype of the mutant S. cerevisiae strain VASY2563 (MAT␣, his4 -15, lys9, ura3-52, erg6⌬, leu2::his G, rad52::LEU2). This strain carries a null allele at the ERG6 locus with pleiotropic effects on membrane permeability and fluidity (31), including hypersensitivity to anthracyclines such as ADM. In addition, this strain carries a mutation at the rad52 locus, which is phenotypically expressed as a defect in DNA repair. We tested in a growth inhibition assay in liquid culture the capacity of MRP and wild type Mdr3 (Mdr3S) to increase cellular resistance to ADM in this strain. Control, MRP-, Mdr3S-, and Mdr3F-expressing VASY2563 cells at stationary phase were diluted to 5 ϫ 10 4 cells/ml, further incubated 1 h at 30°C, and then supplemented with ADM (1, 2, and 3 g/ml). After 48 h of incubation at 30°C in the dark (ADM is light-sensitive), the extent of cell growth was measured by optical density (at A 595 ) and was expressed as a percentage of growth observed for the same clone in drug-free medium. In these experiments, we also tested the sensitivity of the various transformants to FK506 and the K ϩ /H ϩ ionophore VAL (both present at 50 g/ml), two known Mdr3S substrates in yeast (24). As shown in Fig. 2, ADM (at all concentrations tested), FK506 and VAL completely inhibited growth of control pVT and Mdr3F VASY2563 transformants. By contrast, Mdr3S-and MRP-expressing VASY2563 cells were resistant to all concentrations of ADM, reaching cell densities after 48 h similar to those observed for control cultures in drug-free medium. Likewise, both Mdr3S and MRP VASY2563 transformants displayed significant resistance to growth inhibition by FK506 and VAL (Fig. 2), in agreement with results obtained with Mdr3 and MRP JPY201 transformants (Fig. 1B). To verify that the results shown in Fig. 2 truly reflected cellular resistance of the MRP and Mdr3S VASY2563 transformants, culture aliquots were sampled after 4 h of drug incubation and plated onto YPD agar (free of drug) to determine the number of live yeast cells (Fig. 3). Upon quantitation, only a small fraction of the control (8.5% Ϯ 2.5%) and of Mdr3Fexpressing cells (6% Ϯ 1.5%) were found to survive 4 h of incubation with all ADM concentrations tested. In contrast, MRP and Mdr3S transformants were not affected by a 4-h drug exposure, even at the highest ADM concentration. On the other hand, incubation of MRP but not Mdr3S transformants in FK506 (70.5% Ϯ 3.5% survival) or valinomycin (43% Ϯ 4% survival) resulted in a decreased fraction of surviving cells (data not shown). Together, these results suggest that MRP and Mdr3S are functionally expressed in VASY2563 cells and confer resistance to at least three known MDR drugs. Since we have shown in yeast spheroplasts that Mdr3S confers resistance to FK506 by reducing its intracellular accumulation (24), our results suggested that MRP may confer drug resistance in yeast by a similar drug transport process.
We next used [ 14 C]ADM to test the possibility that cellular resistance to this drug in MRP and Mdr3S VASY2563 transformants may be due to a reduced intracellular accumulation. The low specific activity of radiolabeled ADM, together with the high nonspecific binding of this drug to intact yeast cells and spheroplasts required to determine first optimal experimental conditions for ADM accumulation in control yeast cells. Spheroplasts from JPY201 and VASY2563 cells grown in either YPD or YCG medium were found to accumulate [ 14 C]ADM poorly, with total cell-associated radioactivity reaching only a 2-3-fold increase above background after a 2-h incubation period (data not shown). On the other hand, JPY201 spheroplasts derived from cells grown in LA medium (which requires robust mitochondrial respiration) were found to accumulate at least 20 times more ADM (8.25 pmol/10 7 cells prior compared with 247.5 pmol/10 7 cells after a 2-h incubation at 30°C) than spheroplasts prepared in either YPD medium (9.25 pmol/10 7 cells prior compared with 57.5 pmol/10 7 cells after a 2-h incubation at 30°C) or YCG medium (11.5 pmol/10 7 cells prior compared with 62 pmol/10 7 cells after a 2-h incubation at 30°C).
Therefore, we used such preparations of JPY201 spheroplasts to monitor the effect of MRP and Mdr3S expression on cellular accumulation of [ 14 C]ADM. Spheroplasts were diluted in LA medium containing [ 14 C]ADM, placed at 30°C, and at 5-min intervals the amount of cell-associated ADM was determined. During the first 15 min of incubation, ADM accumulation in control, MRP, and Mdr3S spheroplasts was rapid and followed identical kinetics in the three groups (Fig. 4). ADM accumulation continued at the same rate in control pVT spheroplasts to reach a maximum of approximately 200 pmol/ 10 7 cells after 45 min of incubation. By contrast, the rate of ADM accumulation in both MRP and Mdr3S spheroplasts started to decrease significantly after the initial 15-min incubation period, resulting in a lower level of steady state ADM accumulation in these cells compared with pVT controls at 45 min (110 -125 pmol/10 7 cells; Fig. 4). These results suggest that Mdr3S and MRP are functionally expressed in yeast and may mediate cellular resistance and reduced ADM accumulation through an active drug transport process.
We initiated experiments to determine if MRP could actively decrease intracellular drug pools through an active drug efflux mechanism (24). In these experiments, we loaded pVT control, MRP and Mdr3S spheroplasts with [ 14 C]ADM for 25 min at 30°C under normal conditions, at which time we attempted to deplete the medium of free drug molecules by adding excess amounts of BSA. This was done with the intent of measuring potential drug efflux from ADM-preloaded cells, as previously performed by others using DNA (11). The analysis of the fluorescence profile of a 1 M solution of ADM in the presence and absence of a 20 molar excess of BSA shown in Fig. 5A indeed demonstrated complete quenching of fluorescence associated with the free drug molecules by BSA (Fig. 5A). Addition of BSA to pVT control spheroplasts after the initial 25-min exposure to ADM caused a small decrease of the rate of drug accumulation, with a 5-10% reduction in the steady state level of ADM (compare Figs. 4 and 5B). The addition of BSA to MRP and Mdr3S spheroplasts caused a small reduction in the rate of ADM accumulation in these cells, which was followed 10 min later by an actual decrease in the total amount of spheroplasts associated radioactivity. This decrease resulted in a reduction of total ADM accumulation of 20 -25% from peak levels attained in the presence of BSA (Fig. 5B, compare 45-and 90-min time points) and 30 -35% from peak levels reached in these cells in the absence of BSA (compare 90-min time points from Figs. 4 and 5B). Such a decrease in cell associated ADM from steady state levels was not seen in pVT control transformants (Fig. 5B). These results are compatible with the proposition that MRP and Mdr3S can reduce the intracellular ADM concentration from preloaded cells, possibly through an active transport (efflux) mechanism. In this system, the presence of extracellular BSA probably contributes to 1) reduce the total amount of free drug available for uptake, and 2) possibly forming complexes with ADM molecules extruded from preloaded cells.
The similarities detected between MRP and Mdr3S yeast transformants in drug cytotoxicity and drug transport experiments prompted us to test the possibility that MRP shares another functional characteristic of Mdr3S in yeast, the ability to complement the mating defect of a ste6 null mutant. Ste6 is an ABC transporter which shares some structural similarity and limited sequence homology with both Mdr3 and MRP and which transports the a mating pheromone in S. cerevisiae. We have previously shown that expression of wild type (Mdr3S), but not mutant, forms of Mdr3 (e.g. Mdr3F) in a deletion ste6 JPY201 mutant partially restores the ability of these cells to transport a factor and mate to MAT␣ cells (23). Therefore, MRP and Mdr3S transformants together with control pVT and pVT-Ste6 transformants in JPY201 cells were analyzed for their ability to form diploids upon mating with a cell of the opposite mating type (DC17, MAT␣) in a quantitative mating assay (23). Results shown in Fig. 6 indicate that expression of MRP in otherwise sterile JPY201 cells caused an 800-fold increase in the ability of these cells to mate and form diploids on selective medium. Although the mating frequencies measured in MRP transformants (1.9 ϫ 10 Ϫ2 ) were only 13% of those detected in yeast cells expressing wild type Ste6 (23), they were in the same range and sometimes superior to those measured in cells expressing Mdr3S (1.9 ϫ 10 Ϫ3 ) (Fig. 6). Similarly, the MRP-GFP fusion protein also complemented ste6 with efficiency 80% of wild type MRP (data not shown). These findings establish that MRP can, like Mdr3S, mediate the export of the yeast mating pheromone a and restore mating in a sterile mutant.  6. Complementation of the ste6 mutation by MRP and Mdr3S. JPY201 cells transformed with either control pVT plasmid, or with the same plasmid carrying full-length cDNAs for mdr3S, Ste6, or MRP, were allowed to mate with a haploid MAT-␣ tester stock (DC17), and the formation of diploid cells on selective medium was monitored as described under "Experimental Procedures." A depicts the growth of diploid cells on MM medium agar plates after 48-h incubation at 30°C. B, data of three independent experiments are expressed as the average relative mating frequencies, calculated from the number of haploid cells introduced in the mating reaction (23). DISCUSSION MRP was originally identified in the P-gp-negative drugresistant small-cell lung carcinoma cell line H69AR, in which the gene is both amplified and overexpressed (3,4). The specific drug resistance profile of this cell line together with that characteristic of MRP transfectants (10) indicate that MRP overexpression also confers cellular resistance to a large group of structurally distinct drugs such as vincristine, VP-16, daunorubicin, and colchicine, all of which are known MDR drugs and P-gp substrates. Also analogous to P-gp isoforms, MRP seems to be expressed at the cell surface where it functions as an active, ATP-dependent drug transporter (10,11,15), although it may also be active in subcellular membranous compartments. Distinguishing features of MRP, however, are that it can confer resistance to heavy metal oxyanions (10) and can efficiently transport anionic amphiphiles such as cysteinyl leukotrienes and glutathione S-conjugates (12,14). This suggests that MRP may have a broader substrate specificity than P-gp, being able to recognize and transport anionic (leukotrienes) and possibly conjugates or metabolites of cationic amphiphiles (MDR drugs). Although attractive, this model is difficult to reconcile with the observations that 1) MRP overexpression does not confer cellular resistance to cis-platinum (9), a drug detoxified after modification via glutathione conjugation, and 2) ADM, a drug to which MRP confers resistance to (9), is metabolized to a C-13-OH, and is not known to be biotransformed to conjugates (35). In the present study, we have used yeast as a heterologous expression system to initiate the characterization of the mechanism of MRP action.
Immunofluorescence was used to monitor MRP protein expression in yeast. Anti-MRP antibodies produced in MRP transformants a fluorescent signal which displayed a ring-like pattern typical of cell membrane staining, but we also observed a strong cytoplasmic punctate staining pattern, suggesting that a portion of the overexpressed protein may be associated with intracellular organelles. In parallel analyses of membrane-enriched fractions of MRP transformants by immunoblotting, we were unable to identify a polypeptide of molecular weight compatible with that predicted from MRP amino acid sequence analysis (in the range of 170 kDa). Rather, specific 95-100-kDa polypeptides were always detected by the anti-MRP antibodies in cell fractions (data not shown), suggesting that the majority of the MRP protein produced in yeast was either truncated or rapidly degraded during cell growth or in the isolation procedure. Degradation due to incomplete synthesis or aberrant processing of heterologous mammalian membrane proteins by the yeast machinery may be linked to the absence of important signals missing from the primary amino acid sequence of the protein (for sorting, targeting) and/or to very active proteolysis in these cells. Finally, the standard plasma membrane isolation protocol used here includes as a first step, the conversion of intact cells to spheroplasts by enzymatic digestion of the cell wall with Zymolyase. This enzymatic step is critical and can be significant cause of degradation, as not only exogenous but also endogenous yeast membrane proteins (such as the PMA1 H ϩ -ATPase) may be very sensitive to proteases frequently contaminating the commercial preparations of this enzyme. To unequivocally demonstrate that MRP was indeed produced in yeast as a full-length protein, we constructed a chimeric MRP cDNA encoding at its 3Ј end the GFP protein (29). The staining pattern of the expressed MRP-GFP fusion protein detected by fluorescence analysis suggested that the MRP-GFP chimera was associated with the plasma membrane. In addition, we observed that both MRPand MRP-GFP-expressing yeast cells showed cellular resistance toward the P-gp substrate and anti-fungal drug FK506, but could also restore mating in an otherwise sterile ste6 mutant. These results strongly suggest that at least a fraction of the MRP protein in yeast is intact, expressed properly folded at the plasma membrane, and functional.
Since ADM is the drug that had been used in the initial selection of the H69AR cell line, we focused on this cytotoxic drug as a tool to further characterize MRP in yeast. The use of ADM as a cytotoxic drug and a pharmacological transport substrate turned out to be problematic in yeast, for various reasons. First, most common laboratory strains of S. cerevisiae are resistant to extremely high concentrations of this drug. Second, ADM is photolabile and quite unstable at room temperature, as ADM solutions of less than 5 M are known to loose 95% of their activity in less than 1 h at 20°C (36). Third, ADM binds nonspecifically to plastic and other surfaces and can also form aggregates with itself or other molecules at concentrations higher than 1 M (36). Despite these limitations, we found that MRP expression in the mutant S. cerevisiae strain VASY2563, which is hypersensitive to anthracyclines and other lipophilic compounds, dramatically increased resistance to ADM. We noticed that MRP expression in these cells also conferred cellular resistance to two P-gp substrates FK506 and to the K ϩ /H ϩ ionophore VAL (24). Overall, MRP behaved in a manner very similar to P-gp and conferred cellular resistance to structurally unrelated drugs.
Since we had previously established that increased resistance to drugs such as FK506 in yeast Mdr3 transformants is linked to a decreased drug accumulation and increased drug efflux (24), and since studies in mammalian MRP transfectants also indicate that overexpression of MRP is associated with decreased drug accumulation and increased efflux (10, 11), we used [ 14 C]ADM to monitor and compare the kinetics of drug accumulation in MRP and Mdr3 transformants. Results in Fig.  4 indeed showed that spheroplasts from MRP and Mdr3 transformants accumulated significantly less ADM (steady state level after 60 min) than spheroplasts from control pVT yeast cells. We attempted to create ADM efflux conditions from preloaded spheroplasts in order to determine if the reduced accumulation seen in MRP and Mdr3 transformants was indeed associated with increased drug release from these cells. This was done by adding large excess of BSA extracellularly after a 25-min drug preloading period. The rationale for using BSA in such a manner is that this protein binds ADM nonspecifically, which quenches its natural fluorescence (Fig. 5A), but, more importantly, reduces the availability of the free form of the drug. We speculated that this may allow monitoring drug release into the medium from preloaded cells, as released drug molecules would bind nonspecifically BSA causing an overall "displacement" of the drug chemical equilibrium toward the outside (reduction of total cell-associated radioactivity). Results shown in Fig. 5B indicated that actual drug efflux may be ongoing in spheroplasts from MRP and Mdr3 transformants. Indeed, while control spheroplasts continued to accumulate ADM (although at a slower rate) after the addition of BSA, a gradual reduction of ADM followed by an actual decrease in the amount of total cell-associated radioactivity was detected in MRP and Mdr3 spheroplasts (30 -35% reduction in steady state accumulation). Taken together, the results presented in Figs. 4 and 5 clearly indicated that MRP can reduce intracellular accumulation of ADM, and this reduction may be associated with an increase drug release from these cells.
The fact that overexpression of MRP and P-gp (Mdr3) in yeast cells confers cellular resistance to three structurally unrelated drugs (ADM, FK506, and VAL), and is associated with reduced drug accumulation, suggests that both proteins probably act by very similar mechanisms. In addition, the fact that one of these three cytotoxic drugs is known to act at the level of the cell membrane without binding to an intracellular target clearly suggests that MRP and P-gp can recognize and act on intact drug molecules that are not otherwise modified. The data presented here, together with the previously demonstrated ability of MRP to transport leukotrienes and glutathione S-conjugates (12)(13)(14), are in agreement with the proposal that MRP may be capable of acting on both anionic and cationic amphiphiles. Therefore, although both proteins may act according to a common mechanistic basis, their substrate specificity, although overlapping, is clearly distinct. We have demonstrated that MRP can complement a null mutation at the Ste6 locus and restore a mating factor transport and mating in an otherwise sterile yeast mutant (Fig. 6). These experiments suggest that structural and/or functional determinants required for the transport of this pheromone by MRP, Mdr3, and Ste6 proteins have been evolutionarily conserved in these three very distant members of the ABC superfamily. The nature of these common determinants is unknown and can only be speculated upon. The a mating pheromone is a dodecapeptide, which is modified on its terminal cysteine by methylation and by the addition of a fernosyl lipid moiety, and both modifications of the peptide backbone are absolutely required for export. Since we have previously established that one of the members of the mouse P-gp family (the liver-specific Mdr2 isoform) is a lipid flippase capable of actively translocating phosphatidylcholine molecules from the inner to the outer leaflet of the cell membrane (27), it is tempting to speculate that some aspect of lipid transport by Mdr2 may be preserved and be important for the translocation of the lipidated a pheromone by MRP, Mdr3, and Ste6 in yeast. Although highly speculative, such a proposition can now be tested experimentally in yeast.
The yeast proved to be extremely useful as an experimental system for the study and characterization of eukaryotic ABC transporters such as MRP. Functional expression of this protein in yeast will permit detailed pharmacological and biochemical analyses of the mechanism of transport using well defined secretory vesicles from sec mutants (26), where the parameters of transport can be experimentally modulated by the use of specific inhibitors or with the combined use of additional mutant yeast genetic background.