INTRODUCTION

Substantial progress in treating human malignancies has been achieved through systemic administration of chemotherapeutic agents. Several malignant diseases are now curable with chemotherapy, and the treatment of others is slowly advancing. Although promising new antineoplastic agents have been identified, clinical efficacy is frequently limited by inherent multidrug resistance of neoplastic cells. Multidrug resistance (MDR)¹ refers to a phenotype of increased resistance to a wide range of structurally and functionally distinct cytotoxic compounds and is often attributable to overexpression of the plasma membrane ATPase, P-glycoprotein (Pgp) (1).

Pgp is an evolutionarily conserved protein composed of structurally similar halves each consisting of an ATP binding site and six membrane-spanning domains (2). Pgp homologs are highly conserved from bacteria to mammals (3). For example, the yeast pheromone transporter STE6 and the human MDR1 gene product are 57% identical at the amino acid level (4, 5). The mammalian Pgp isolated from multidrug-resistant cells is an energy-dependent efflux pump that reduces the intracellular concentration of many small hydrophobic molecules (1). Three murine and two human genes encoding P-glycoproteins have been cloned. The murine mdr1 (mdr1b) and mdr3 (mdr1a) genes encode Pgps that confer multidrug resistance (6). In addition, the murine mdr3 gene product comprises part of the functional blood-brain barrier whereas the mdr2 gene encodes a phosphatidylcholine translocase important in bile transport (7, 8). In humans, overexpression of the MDR1 gene is responsible for multidrug resistance (1).

In vitro, multidrug resistance can be partially overcome by simultaneous treatment with certain noncytotoxic drugs. Like their cytotoxic counterparts, these drugs are relatively small and hydrophobic but share little structural similarity. The immunosuppressant cyclosporin A (CsA) is one such MDR reversing drug (9, 10, 11), and two additional immunosuppressants, FK506 and rapamycin, also partially overcome multidrug resistance (12, 13, 14, 15, 16). These observations raise the possibility that the targets required for immunosuppression might regulate Pgp function.

CsA, FK506, and rapamycin suppress the immune system by inhibiting T-cell activation (17, 18). All three compounds function in complex with cytoplasmic proteins known as immunophilins; CsA binds to cyclophilin A and FK506 and rapamycin bind to FKBP12. The immunophilins are widely distributed, abundant enzymes that catalyze cis-trans-peptidyl-prolyl isomerization, a rate-limiting step in protein folding. Drug binding inhibits this activity but is not sufficient for immunosuppression. Rather, the cyclophilin A-CsA and FKBP12-FK506 complexes inhibit the protein phosphatase calcineurin, a component of the T-cell receptor signaling cascade (19), reviewed in Ref. 20. The targets of the FKBP12-rapamycin complex are the TOR proteins, kinase homologs that regulate cell proliferation in response to growth factors (21, 22, 23, 24, 25).

In contrast to their established mechanism of action in T-cells as immunophilin-drug complexes, CsA, FK506, and rapamycin are thought to reverse MDR by directly inhibiting Pgp function. FK506 is a known Pgp substrate (26, 27, 28), and CsA binds to Pgp, competes with Vinca alkaloids for Pgp, and inhibits verapamil-stimulated Pgp ATPase activity (28, 29, 30). Drug analog studies have suggested that the ability of CsA and FK506 to inhibit calcineurin and Pgp are distinct activities (9, 10, 11, 16, 31, 32). We wished to test directly whether the immunophilins or calcineurins modulate Pgp functions in vivo.

The yeast Saccharomyces cerevisiae is a useful model system for the study of Pgp function. Expression of murine mdr3 in yeast restores a-factor pheromone export and complements the mating defect of mutants lacking the STE6 Pgp homolog (33). Murine mdr3 also confers FK506 resistance (27, 34). Similarly, the yeast ste6 mutation is complemented by the Pgp homolog of Plasmodium falciparum encoded by the pfmdr1 gene (35). Human MDR1 expression in yeast conferred resistance to the ionophore valinomycin (36). Importantly, the targets of CsA, FK506, and rapamycin (cyclophilin A, FKBP12, calcineurin, and the TOR proteins) are highly conserved from yeast to man, and the mechanism of drug action is essentially identical (reviewed in Ref. 20).

Although yeast has served as a valuable model system to analyze MDR function, studies have been hampered by the inherent resistance of yeast cells to chemotherapeutic agents. In this report, we demonstrate that an erg6 mutation, which blocks synthesis of the yeast membrane sterol ergosterol (37), permeabilizes the cell and renders S. cerevisiae sensitive to anthracyclines and to dactinomycin, clinically useful cytotoxic drugs that are Pgp substrates. We find that expression of the murine mdr3 Pgp in erg6 mutant cells confers dactinomycin resistance. Moreover, we demonstrate that murine mdr3 expression renders a CsA-FK506-sensitive yeast strain CsA- and FK506-resistant.

By a genetic approach, we analyze the effects of the targets of the MDR reversing immunosuppressants CsA, FK506, and rapamycin on the function of murine Pgp in yeast. While cyclophilin A may modulate Pgp function, it is not essential for murine mdr3 function. Furthermore, the target of cyclophilin-CsA and FKBP12-FK506, the phosphatase calcineurin, is dispensable for MDR function in yeast. In contrast, FKBP12 is required for mdr3-mediated resistance to dactinomycin and CsA in yeast. Our findings suggest CsA and FK506 could reverse multidrug resistance by directly inhibiting Pgp but that FK506 and rapamycin likely also interfere with Pgp function by inhibiting FKBP12-Pgp interaction. Our studies reveal a novel role for the FKBP12 prolyl isomerase in Pgp function that may be similar to FKBP12-dependent function of other large membrane proteins, including the inositol 1,4,5-trisphosphate and ryanodine receptor Ca²⁺ channels (38, 39).

MATERIALS AND METHODS

Dactinomycin (Merck Sharp Dohme), doxorubicin (Adria), and cyclosporin A (Sandoz) were obtained commercially. Rapamycin was provided by the Drug Synthesis and Chemistry Branch of the National Cancer Institute and FK506 by Fujisawa Pharmaceuticals. YPD medium and minimal selective media were prepared as described (40). Drug-containing media were prepared by adding a sterile drug stock solution to autoclaved media.

Yeast strains are listed in Table I. Yeast strain R757 and the isogenic erg6 derivative R2563 were provided by Rick Gaber (Northwestern University). These strains were converted to leucine auxotrophy by disrupting the LEU2 gene with leu2::hisG-URA3-hisG as described (41). The leu2::hisG-URA3-hisG allele was converted to leu2::hisG by selection on 5-FOA medium. The genes encoding cyclophilin A (CPR1) and the calcineurin B regulatory subunit (CNB1) were disrupted as described (42, 43, 44, 45) to generate strains CHY2563-4 and CHY2563-2, respectively. An FKBP12 null allele was selected by plating yeast on YPD medium containing 0.1 µg/ml rapamycin and incubating for 5 days. Rapamycin-resistant isolates were tested for cross-resistance to 1.0 µg/ml FK506 plus 100 mM LiCl (44). Protein extracts from drug-resistant isolates were analyzed by Western blot with anti-FKBP12 antiserum (25) to confirm the absence of FKBP12. The resulting fpr1 strain was designated CHY2563-1. RAD52 was disrupted as described (46), and the erg6 rad52::LEU2 strain was designated VASY2563. Yeast strain CHY251 contains a vph6::TRP1 deletion (47). Isogenic cpr1::LEU2 (CHY515) and fpr1::ADE2 (CHY516) derivatives were produced as described (42, 43, 44).

Table I. Yeast strains Strain Genotype Ref. JK93d MAT his4 HML leu2-3,112 rme1 trp1 ura3-52 2 MH250-2C MAT cpr1::LEU2 his4 HML leu2-3,112 rme1 trp1 ura3-52 8 JHY3-3D MAT fpr1::URA3 his4 HML leu2-3,112 rme1 trp1 ura3-52 8 TB85 MAT cnb1::LEU2 his4 HML leu2-3,112 rme1 trp1 ura3-52 8 R757 MAT his4-15 lys9 ura3-52 7 R2563 MAT erg6 his4-15 lys9 ura3-52 7 R2563leu2 MAT erg6 his4-15 leu2::hisG lys9 ura3-52 This study VASY2563 MAT rad52::LEU2 erg6 his4-15 leu2::hisG lys9 ura3-52 This study CHY2563-1 MAT fpr1 erg6 his4-15 leu2::hisG lys9 ura3-52 This study CHY2563-2 MAT cnb11::LEU2 erg6 his4-15 leu2::hisG lys9 ura3-52 This study CHY2563-4 MAT cpr1::LEU2 erg6 his4-15 leu2::hisG lys9 ura3-52 This study CHY251 MAT his3200 leu2-3,112 trp1101 ura3-52 vph6::TRP1 7 CHY515 MAT his3200 leu2-3,112 trp1101 ura3-52 vph6::TRP1 cpr1::LEU2 This study CHY516 MAT his3200 leu2-3,112 trp1101 ura3-52 vph6::TRP1 fpr1::ADE2 This study DC17 MAT his1 3 SM2375 MAT can1 cyh2 his6 met1 sst2-1 4

Yeast plasmid pVTMDR3S encoding the full-length murine mdr3 cDNA, and the expression plasmid pVT, were generously provided by Martine Raymond and Philip Gros (Clinical Research Institute of Montreal and McGill University). These are multi-copy 2-µm plasmids containing the constitutive yeast alcohol dehydrogenase promoter and have been previously described (27, 33, 48, 49). Plasmid pFP101 was produced by inserting the 2-kilobase EcoRI fragment containing the FPR1 gene (42) into the EcoRI site of the CEN LEU2 plasmid YCplac111 (50). Plasmid pFP102 was constructed by replacing the internal KpnI/HpaI fragment in the FPR1 gene of plasmid pYJH26 with the F43Y FPR1 mutant sequence from plasmid pSM14-3 (51) and subcloning the resulting 2-kilobase EcoRI fragment into YCplac111, as above. erg6 strains were transformed as described (37).

The sensitivity of the erg6 yeast strain was first determined by incubation on solid YPD medium containing concentrations of dactinomycin or doxorubicin ranging from 0 to 100 µg/ml. Growth was scored in comparison to the isogenic ERG6 parental strain following incubation at 30 °C for 72 h. Quantitative drug sensitivity was determined as follows. Transformed erg6 yeast strains were incubated at 30 °C for 18 h in selective medium and then diluted to an A₆₀₀ of 0.8 with selective medium. 20 µl of yeast culture was added to a microtiter well containing 80 µl of selective medium with or without dactinomycin. Cells were incubated for 16 h and 10-fold serially diluted in sterile water, and 10 µl of each dilution was spotted to YPD medium and incubated at 30 °C for 72 h. All colonies visible without magnification were scored as a viable inoculating cell. A survival index was calculated as the ratio of viable cells in drug-containing medium versus medium without drug. Relative drug resistance conferred by Pgp expression was calculated as the ratio of the survival index of an mdr3-expressing yeast strain to the survival index of the isogenic strain with the control plasmid.

Total cell lysates were prepared from cells growing exponentially in selective medium. After harvesting by centrifugation, cells were washed in water and resuspended in lysis buffer (20 mM Tris, pH 7.4, 100 mM KCl, 0.1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 100,000 units/ml Trasylol) and acid-washed glass beads. Cells were lysed by vigorous mechanical shaking in a bead beater. Unlysed cells were removed by centrifugation (1000 × g, 5 min). Total cell extracts (60 µg of protein) were mixed with Laemmli sample buffer, heated, and subjected to SDS-8% polyacrylamide gels. Proteins were transferred to nitrocellulose membranes, and Pgp was detected with the murine monoclonal anti-Pgp antibody C219 (Signet Laboratories) diluted to 1 µg/ml. Cyclophilin A, FKBP12, and calcineurin B were detected with rabbit polyclonal antisera diluted 1:1000 as described previously (25, 52, 53). Immune complexes were detected with goat anti-mouse or goat anti-rabbit immunoglobulin coupled to horseradish peroxidase and the enhanced chemiluminescence system (Amersham Corp.). Fractionation into S100 soluble and P100 membrane-pellet fractions and indirect immunofluorescence were performed as described (25).

The yeast mating pheromone a-factor was assayed as described using the a-factor supersensitive sst2 mutant strain SM2375 (Table I) (54). Mating ability was assayed by replica plating patches of strains grown on YPD to a MAT tester lawn of strain DC17 and, following 24 h growth at 30 °C on YPD, replica plating to minimal media. For quantitative matings, ~10⁷ cells of each mating partner were mixed in 2 ml of YPD, incubated with gentle agitation for 4 h at 30 °C, pelleted, washed, resuspended, and serially diluted in water, and portions were spotted to YPD and minimal medium to determine the fraction of prototrophic diploids/total number of cells.

RESULTS

We wished to test the effects of heterologous expression of murine Pgp on the toxic effects of clinically relevant chemotherapeutic compounds in the yeast S. cerevisiae. Unfortunately, wild-type yeast strains are highly resistant to antineoplastic drugs, possibly as a consequence of low drug permeability relative to mammalian cells (55). In S. cerevisiae, mutations in the ERG6 gene block sterol methylation at the C-24 position, the final step in the biosynthesis of the fungal membrane sterol ergosterol. erg6 mutant strains exhibit normal vegetative growth, but their membrane composition and function are altered compared with wild-type strains (37). Importantly, erg6 mutant strains are hypersensitive to many compounds, including cycloheximide, camptothecin, and brefeldin A (37, 56, 57). We found that the growth of an erg6 mutant strain (R2563, Table I) is inhibited by the chemotherapeutic agents dactinomycin and doxorubicin with a minimum inhibitory concentration of ~25 µg/ml; in contrast, growth of the isogenic ERG6 parental strain (R757) was unaffected at concentrations up to 100 µg/ml (data not shown).

The rad52 mutation results in a defect in DNA repair and has been previously shown to impart sensitivity to many toxins whose target is DNA (58). We therefore introduced a rad52 mutation into the erg6 mutant strain. The resulting erg6 rad52 double mutant strain (VASY2563, Table I) is highly sensitive to dactinomycin and doxorubicin, even at concentrations as low as 1 µg/ml (data not shown). Because the rad52 mutation also prevents homologous recombination and thus precludes introduction of additional gene disruptions, in most cases we have employed the erg6 mutant RAD52 wild-type strain R2563 to test the effects of murine mdr3 gene expression on chemotherapy resistance.

Yeast plasmid pVTMDR3S is a 2-µm high-copy number self-replicating plasmid in which the cDNA for the mouse mdr3 gene is expressed from a strong constitutive promoter derived from the alcohol dehydrogenase gene; previous studies have established that this plasmid efficiently expresses mdr3 in yeast (27, 33, 48, 49). The erg6 mutant strain R2563 was transformed with pVTMDR3S and with the control plasmid pVT, and the resulting transformed strains were tested for dactinomycin resistance. Cell viability was assessed after a 16-h incubation in the presence or absence of dactinomycin (Fig. 1A). Yeast transformed with the murine mdr3-containing plasmid exhibited a 6-fold increase in the number of viable cells when compared with control transformants (Fig. 1B). Similarly, the murine mdr3 gene, but not a control plasmid, rendered the erg6 rad52 mutant strain dactinomycin-resistant (data not shown). Drug resistance was attributable to the murine mdr3 gene and not to an extraneous genomic mutation, because cells that had lost the plasmid-borne mdr3 and URA3 genes by selection on 5-FOA medium were once again dactinomycin-sensitive (data not shown). A previous study has suggested that ergosterol inhibits the human MDR1 Pgp (59), and thus the erg6 mutation might contribute to both increased mdr3 function and enhanced cell permeability.

Western blot analysis with the Pgp-specific monoclonal antibody C219 confirmed that cells containing the murine mdr3 gene expressed Pgp, whereas cells containing the control plasmid did not (Fig. 2, lanes 1 and 2). Western blot analysis of S100 supernatant and P100 membrane-pellet fractions resulting from a 100,000 × g centrifugation revealed that murine mdr3 was present exclusively in the membrane-pellet fraction when expressed in yeast (Fig. 3). In addition, by indirect immunofluorescence murine mdr3 was expressed on the plasma membrane (data not shown). Collectively, these findings demonstrate that murine mdr3 is appropriately localized and confers dactinomycin resistance in yeast, thus establishing the utility of the erg6 and erg6 rad52 mutant strains for studies on MDR function.

It has previously been demonstrated that murine mdr3 expression confers resistance to the calcineurin-independent antifungal action of FK506 and that FK506 is a substrate for P-glycoprotein (26, 27). In contrast, although CsA is a known MDR reversing drug that competitively inhibits binding of other substrates to Pgp, it has been controversial whether or not CsA is itself a Pgp substrate. To address this issue, we employed a vph6 mutant yeast strain in which we recently demonstrated the target of CsA, calcineurin, is essential for viability (47). The vph6 mutant yeast strain CHY251 was transformed with the mdr3 expression plasmid pVTMDR3S and with the control plasmid pVT, and the resulting transformed strains were tested for CsA resistance. As shown in Fig. 4, expression of murine mdr3, but not of the control plasmid, effectively rendered the vph6 mutant strain resistant to both CsA and to FK506. This finding demonstrates that CsA is a P-glycoprotein substrate and that murine mdr3 effectively exports both CsA and FK506 from yeast cells.

The immunosuppressants CsA, FK506, and rapamycin inhibit Pgp function in mammalian cells. CsA also inhibits the peptidyl-prolyl isomerase activity of cyclophilin A, whereas FK506 and rapamycin inhibit that of FKBP12. We therefore tested whether these immunosuppressants reverse Pgp-mediated multidrug resistance by inhibiting prolyl isomerase activity required for Pgp function. A cyclophilin A disruption mutation was introduced into the erg6 and vph6 mutant strains (see ``Materials and Methods''), and the resulting erg6 cyclophilin A and vph6 cyclophilin A mutant strains were transformed with the murine mdr3 gene or the control plasmid. Western blot analysis confirmed that the cyclophilin A mutant strains expressed murine mdr3 at a level comparable to wild-type cells (Fig. 2, lanes 3 and 4). Furthermore, Pgp is detected solely in the membrane fraction (Fig. 3), and immunolocalization studies reveal patchy plasma membrane staining in cyclophilin A wild-type and mutant cells (data not shown).

When exposed to dactinomycin, the relative resistance of the mdr3-expressing cyclophilin A mutant strain was 60-fold with respect to isogenic cells containing the control plasmid alone (Fig. 1). This increase in the relative dactinomycin resistance of an mdr3-expressing erg6 cyclophilin A mutant yeast strain compared with cyclophilin A wild-type cells suggests that cyclophilin A could normally inhibit murine Pgp function in yeast. In contrast, the Pgp-mediated FK506 resistance of a vph6 cyclophilin A mutant yeast strain was similar to the isogenic cyclophilin A wild-type vph6 mutant yeast strain (Fig. 4). We conclude that while cyclophilin A may modulate Pgp function in an erg6 mutant strain, Pgp-mediated drug resistance does not require cyclophilin A in yeast.

We next tested whether the FK506-rapamycin target protein FKBP12 is required for Pgp-mediated drug resistance in yeast. For this purpose, erg6 FKBP12 mutant and vph6 FKBP12 mutant strains were constructed (see ``Materials and Methods'') and transformed with control plasmid pVT and the pVTMDR3S mdr3 expression plasmid. Immunoblot demonstrated that murine mdr3 was expressed at similar levels in the FKBP12 wild-type and FKBP12 mutant erg6 yeast strains (Fig. 2) and that the proteins were localized to the membrane-pellet fraction (Fig. 3). Indirect immunofluorescence microscopy showed discrete patchy staining of only the plasma membrane in both wild-type and FKBP12 mutant yeast strains (data not shown).

Exposure of mdr3-expressing erg6 FKBP12 mutant strains to dactinomycin resulted in a relative resistance of only 2-fold (Fig. 1), a value considerably less than either wild-type erg6 strains, the erg6 cyclophilin A mutant, or the erg6 calcineurin mutant strains. Pgp-mediated resistance to CsA was then tested in a CsA-sensitive vph6 mutant strain background. While mdr3 expression in the wild-type vph6 strain conferred CsA resistance, mdr3 expression in the vph6 FKBP12 mutant strain failed to confer CsA resistance (Fig. 4). Expression of the yeast FPR1 gene encoding FKBP12 from a centromeric plasmid complemented this defect, restoring the ability of mdr3 to confer CsA resistance in the vph6 fpr1 mutant strain (Fig. 5). Taken together, these findings reveal that FKBP12 is required for the ability of murine mdr3 to confer resistance to two structurally dissimilar compounds in yeast.

We next tested whether it was the peptidyl-prolyl isomerase activity of FKBP12 that was required for mdr3 function in yeast. A single amino acid substitution in the active site of human FKBP12 (F36Y) has been described that reduces prolyl isomerase activity to ~5% (60). We introduced the corresponding mutation in yeast FKBP12 (F43Y) and find that the mutant protein has ~5% prolyl isomerase activity but is expressed at about one-fifth the level of wild-type FKBP12.² Thus, we estimate that cells expressing the F43Y FKBP12 mutant protein contain ~1% the wild-type level of FKBP12 peptidyl-prolyl isomerase activity. When a yeast centromeric plasmid expressing the F43Y FKBP12 mutant protein was introduced into the mdr3-expressing vph6 fpr1 mutant yeast strain, expression of the mutant FKBP12 protein fully restored the ability of murine mdr3 to confer CsA resistance (Fig. 5). This observation suggests that FKBP12 prolyl isomerase activity is not required for FKBP12-dependent mdr3 function in yeast.

The protein phosphatase calcineurin is inhibited by the immunophilin-immunosuppressant complexes cyclophilin A-CsA and FKBP12-FK506 in vertebrate lymphocytes and in yeast (19, 44, 45, 61, 62). Phosphorylation of Pgp by protein kinase C appears to enhance Pgp activity (63, 64, 65, 66); however, some known multidrug resistance reversal agents may promote Pgp phosphorylation (67). We have taken a mutational approach to assess the effects of calcineurin phosphatase activity on Pgp function. In S. cerevisiae, the regulatory subunit of calcineurin is encoded by one gene, CNB1, and disruption of the CNB1 gene results in a complete loss of calcineurin activity (43, 44, 68). An erg6 cnb1 calcineurin mutant strain was constructed and used to test murine mdr3-mediated drug resistance. Western blot confirmed that the calcineurin mutant strain expressed murine mdr3 to the same extent as isogenic wild-type cells (Fig. 2, compare lanes 7 and 8 to 1 and 2) and that this protein was properly localized in the P100 membrane fraction (Fig. 3). Loss of calcineurin activity had no significant effect on dactinomycin resistance conferred by mouse mdr3 expression. The relative resistance of the mdr3-expressing cnb1 mutant strain was 10.7-fold compared with 6-fold for the wild-type strain (Fig. 1, A and B). We conclude that calcineurin is not required for murine mdr3 function in yeast.

The yeast pheromone a-factor is a lipid-modified peptide that does not exit the cell through the standard secretory pathway. Instead, a yeast P-glycoprotein homolog, STE6, transports a-factor across the plasma membrane (4, 5, 54). We sought to determine whether FKBP12, cyclophilin A, or calcineurin are required for STE6 function. Quantitative a-factor export assays were performed (54) on wild-type cells, and cyclophilin A, FKBP12, and calcineurin null mutant strains. No differences in the titers of exported a-factor were observed (Fig. 6). Similarly, isogenic MATa wild-type, fpr1, cpr1, and cnb1 mutant cells mated with similar efficiencies (an independent measure of a-factor export), as detected by either a patch mating test (Fig. 7) or quantitative mating assays in which the relative mating efficiencies were wild-type (100%), cpr1 (103%), fpr1 (115%), and cnb1 (44%). Thus, cyclophilin A, FKBP12, and calcineurin are not required for function of the yeast Pgp family member, STE6. We conclude that FKBP12 is required for function of some Pgp family members but not others.

DISCUSSION

Recent investigation has focused on Pgp because of its role in chemotherapy resistance (1). While several promising Pgp inhibitors have been identified, including verapamil and trifluoperazine, their clinical utility is compromised by high drug concentrations required to attenuate MDR. The immunosuppressants CsA, FK506, and rapamycin represent potentially useful adjuncts to cytotoxic chemotherapy because of their ability to overcome multidrug resistance in vitro. However, the mechanism by which these agents inhibit Pgp function has not been completely established.

Several models can be envisioned to explain the MDR reversal effects of the immunosuppressants. The most prevalent view is that CsA, FK506, and rapamycin directly inhibit Pgp and that their immunosuppressive and immunophilin-binding properties are unrelated to MDR reversal. FK506 binds to the Pgp substrate binding site and is transported in an ATP-dependent process in both yeast and mammals (26, 27, 28). While CsA also binds the Pgp substrate site and inhibits verapamil transport, the data are conflicting regarding its ability to be transported across a membrane (26, 28). Rao and Scarborough (28) found that CsA inhibited verapamil-stimulated P-glycoprotein ATPase activity of baculovirus-expressed human mdr1, but CsA alone failed to stimulate ATPase activity, suggesting CsA is not a substrate. In contrast, our finding that murine mdr3 confers CsA- and FK506 resistance in yeast (Fig. 4) unequivocally demonstrates that CsA is a substrate for the murine mdr3 P-glycoprotein. CsA might not be a substrate for human mdr1, or a slow pumping rate sufficient to confer cellular CsA resistance might not lead to detectable stimulation of ATPase activity in vitro. As Pgp substrates, the immunosuppressants could competitively inhibit cytotoxic drug binding or transport by Pgp, and our results, at least in part, support this view.

A second model is that calcineurin positively regulates Pgp function, and CsA and FK506 reverse multidrug resistance by inhibiting calcineurin. Calcineurin activates at least one energy-dependent plasma membrane pump, the Na⁺/K⁺ ATPase of renal tubule cells (69). However, our finding that calcineurin is not required for mdr3 or STE6 function in yeast conclusively demonstrates that calcineurin is not required for either Pgp to function. Several additional findings support this view. First, several nonimmunosuppressive CsA and FK506 analogs, at least some of which do not inhibit calcineurin, still reverse MDR (10, 11, 16, 31, 32). Second, phosphorylation is known to stimulate Pgp activity (66, 70, 71, 72), and thus calcineurin inhibition should increase, rather than decrease, Pgp activity. Finally, rapamycin inhibits MDR but does not inhibit calcineurin.

A third model is that the immunophilins cyclophilin A and FKBP12 are required for mdr function and that CsA, FK506, and rapamycin interfere with immunophilin-dependent Pgp function. Our results demonstrate that cyclophilin A is not required for Pgp activity in yeast. Our finding that CsA is an mdr3 substrate in yeast suggests that CsA might act directly on Pgp. However, because yeast cells express six other cyclophilins, most of which are highly conserved in mammals, we cannot exclude that all or some of the CsA effect is exerted through one or more of these CsA binding proteins.

In contrast, in two different assays of mdr3 function in yeast, we find that Pgp-mediated drug resistance is severely compromised in yeast mutants lacking FKBP12. Expression of wild-type FKBP12 complements the defect and restores Pgp function. We find no difference in the level of mdr3 expression in wild-type versus FKBP12 mutant yeast strains, and the mdr3 protein appears to be properly localized, as assessed by Western blot of S100/P100 cellular fractions and indirect immunofluorescence. FKBP12 may therefore be required for the proper function of mature mdr3 in the plasma membrane. We have thus far been unable to detect a direct mdr3-FKBP12 complex by either FKBP12 affinity chromatography or by co-immunoprecipitation (data not shown). We conclude that FKBP12 is required for proper function of the murine mdr3 Pgp in yeast but that this interaction is either indirect or too transient to be detected by these methods.

Expression of either wild-type FKBP12 or of an FKBP12 mutant protein with greatly reduced prolyl isomerase activity restored mdr3-mediated drug resistance in yeast. This finding suggests that the prolyl isomerase activity of FKBP12 is not required for FKBP12-dependent mdr3 function. Similar studies have revealed that FKBP12-dependent function of the ryanodine receptor Ca²⁺ channel is also independent of FKBP12 prolyl isomerase activity (60). One possible model is that the hydrophobic drug binding/active site of FKBP12 is required for some FKBP12 functions in which the enzymatic activity of this binding site is dispensable.

Although many immunophilin functions remain to be discovered, an emerging common function is a role in processing and regulating large membrane proteins. For example, the Drosophila ninaA cyclophilin is required for proper trafficking of rhodopsins in the eye (73, 74). FKBP12 is also a regulatory subunit of both the ryanodine receptor and the inositol 1,4,5-trisphosphate receptor, both large, multimeric, integral membrane Ca²⁺ channels (38, 39, 60). Likewise, Pgp, another large transmembrane protein, may be folded or regulated by immunophilins. Pgp is known to associate with the molecular chaperones calnexin and Hsc70 (75). Binding of the immunosuppressants might interfere with immunophilin-dependent Pgp activity. Our findings clearly support this model for FKBP12 and murine mdr3 expressed in yeast. This model is also consistent with the finding that CsA, FK506, and rapamycin concentrations required to reverse MDR are significantly higher than those required for immunosuppression and may suffice to inhibit the abundant immunophilins (9, 12, 14). Furthermore, although some nonimmunosuppressive CsA and FK506 analogs reverse MDR (10, 11, 16), the ability of these analogs to bind and inhibit immunophilins has not yet been reported and might contribute to mdr reversing activity.

The heterologous expression of mammalian Pgp in yeast has been previously well described (27, 33, 34, 36, 48, 49). In this report we exploit the ability to genetically manipulate S. cerevisiae to identify an erg6 mutant yeast strain as sensitive to several additional cytotoxic drugs including the clinically relevant Pgp substrates, dactinomycin and doxorubicin, and show that mdr3 expression renders the erg6 mutant strain drug-resistant. Our findings extend the utility of the yeast heterologous expression system to address many aspects of Pgp function and regulation. In fact, this system was recently used to demonstrate that an MDR-related protein often mutated in human lung cancers, MRP, similarly confers drug resistance when expressed in the erg6 mutant yeast strains described here (76).

We employed this model system to demonstrate that CsA is a substrate for P-glycoprotein and to specifically address, and exclude, obligate roles for calcineurin and cyclophilin A in Pgp function. Our findings suggest that CsA may act directly on Pgp. In contrast, we found that FKBP12 is required for Pgp function. This finding reveals a novel role for FKBP12 in regulating the mdr membrane protein pump and suggests that the ability of FK506 and rapamycin to reverse MDR may result from disruption of FKBP12-dependent Pgp activity. In conclusion, we suggest that erg6 mutant strains should prove useful for a broad range of studies aimed at defining the molecular mechanisms of drug action by genetic approaches in yeast.