INTRODUCTION

The development of drug resistance in tumor cells is a major obstacle to clinical response in cancer chemotherapy. A well characterized cellular phenotype termed multidrug resistance (MDR)¹ is mediated by the multidrug transporter P-glycoprotein (P-gp), which functions as an ATP-dependent drug efflux pump of broad substrate specificity (1-3). The mdr1 gene, which encodes P-gp, is expressed in some normal and malignant tissues and has been associated with a poor prognosis in several types of cancer (4-9). The reversal of MDR by inhibitors or modulators of P-gp may improve the outcome of cancer chemotherapy (4, 9-15). Cyclosporine, its analogue PSC 833 (PSC; Refs. 13 and 14), verapamil, and other MDR modulators have been shown to increase cellular drug accumulation and reverse MDR through competitive binding to P-gp (reviewed in Refs. 2, 4, and 9). Current clinical trials using antitumor agents combined with MDR modulators to circumvent MDR have raised the issue of whether the suppression of P-gp function might result in the selection of alternative mechanisms that could confer resistance to multiple agents. These mechanisms include changes in topoisomerase (Topo) II or II (reviewed in Refs. 16-17 and 18) and increased expression of the gene for the multidrug resistance-associated protein, mrp (19, 20), or the p110 major vault protein, LRP-56 (21).

In this study, we utilized co-selection to investigate the mechanisms conferring resistance to both doxorubicin (DOX) and PSC. The parental cells expressed high levels of P-gp. We hypothesized that the suppression of P-gp function might favor the emergence of an altered form of P-gp or an alternative mechanism of resistance (22, 23). Our data reveal the expression of a novel mutant P-gp in variant cells with an altered spectrum of cross-resistance to cytotoxins and resistance to modulation by cyclosporins. The resistant cells and mutant P-gp were characterized in terms of their patterns of drug resistance, modulation by inhibitors of P-gp, and transport of drugs and their P-gp expression, gene sequence, and transfection of the mutant mdr1.

EXPERIMENTAL PROCEDURES

Doxorubicin was obtained from Adria Laboratories (Columbus, OH), etoposide from Bristol-Myers (Evansville, IN), and vinblastine from Eli Lilly and Co. (Indianapolis, IN). PSC was provided by Sandoz Pharmaceutical Corporation (Basel, Switzerland). Rhodamine 123 (Rh-123) was purchased from Molecular Probes (Eugene, OR). All other anti-cancer agents were obtained from the NCI (National Institutes of Health), and all other chemicals were from Sigma.

Details of the development and characterization of the human cell line MES-SA and its MDR variant MES-SA/Dx5 (Dx5), which were derived from sarcoma elements of a uterine mixed Mullerian tumor, have previously been described (24, 25). MES-SA/DxP5002 (DxP) cells were derived from Dx5.05 cells (Dx5 cells selected and maintained at a DOX concentration of 0.5 µM) by stepwise selection in the presence of increasing DOX concentrations (from 40 nM to 0.5 µM) and a constant PSC concentration at 2 µM over a 1-year period.

Metaphase chromosome preparations were examined for the presence and structure of chromosome 7, where the human mdr1 gene is normally located, by karyotyping and in situ hybridization with a chromosome 7-specific probe (Oncor, Inc., Gaithersburg, MD). The hybridized chromosome was visualized by the method of Sasai et al. (26) using fluorescence microscopy.

Growth inhibition and doubling times were evaluated by the MTT colorimetric assay in triplicate or quadruplicate as described previously (22, 23). The modulation of resistance to DOX, paclitaxel, vinblastine, and etoposide was also determined by MTT assays in the presence of the MDR modulator PSC (2 µM) or verapamil 6 µM. These concentrations of the modulators are noncytotoxic and completely reverse resistance to DOX in our cellular MDR models. The modulation ratio was obtained by comparing IC₅₀ values in the presence and absence of modulators (22).

Intracellular drug accumulation was quantified using radiolabeled drugs as described previously (23). All values were normalized to protein content.

The flow cytometric analysis of Rh-123 retention and P-gp expression were determined by a dual laser flow cytometer (FACS-II^TM; Becton-Dickinson Corp., Mountain View, CA). Double labeling was performed with Rh-123 and the monoclonal antibody UIC2 Immunotech, Inc. (Westbrook, ME).

The oligonucleotides used as amplimers in this study were synthesized by Operon Technologies (Alameda, CA) and Ana-Gen Inc. (Palo Alto, CA). The sets of amplimers specific for Topo II and II protein and mrp have been previously documented (22, 23). A panel of mdr1 primers for mRNA sequence amplification was designed according to the published sequence (27) and is summarized in Table I. 28 S ribosomal cDNA was used as an endogenous control for PCR (22, 23).

Table I. Oligonucleotides used in amplification and detection of mdr1 cDNA Amplimersa Location Sequence Humdr1-Fb 181-199 5-ATGGATCCGGCCGGGAGCAGTCATCTG-3 Humdr1-F 421-440 5-ATGGATCCCGGGATGGATCTTGAAGGGG-3 Humdr1-Rc 476-495 5-CTGAATTCCACTTTTATTGTTCAGTTT-3 Humdr1-F 714-735 5-ATGGATCCATATCAATGATACAGGGTTCTT-3 Humdr1-R 822-841 5-CTGAATTCTGCCAGGCACCAAAATGAAA-3 Humdr1-F 951-971 5-ATGGATCCAGATGATGTCTCTAAGATTA-3 Humdr1-F 1194-1213 5-ATGGATCCTCTTGGCAGCAATTAGAACT-3 Humdr1-R 1238-1257 5-CTGAATTCTTGTACCTTTCAAGTTCTTT-3 Humdr1-F 1460-1477 5-ATGGATCCGGACAGGCATCTCCAAGC-3 Humdr1-R 1501-1519 5-CTGAATTCGATTTCATAAGCTGCTCCT-3 Humdr1-F 1702-1721 5-ATGGATCCTGGAAACAGTGGCTGTGGGA-3 Humdr1-R 1703-1721 5-CTGAATTCCTCCCACAGCCACTGTTTCC-3 Humdr1-R 2025-2043 5-CTGAATTCGCGATCCTCTGCTTCTGC-3 Humdr1-R 2226-2247 5-CTGAATTCTCCACAATGACTCCATCATC-3 Linker-F 2228-2247 5-CGATGATGGAGTCATTGTGGA-3 Humdr1-R 2433-2450 5-CTGAATTCGTGATCCACGGACACTCC-3 Linker-R 2593-2614 5-TATTGCAAATGCTGGTTGCAGG-3 Humdr1-F 2593-2614 5-ATGGATCCCTGCAACCAGCATTTGCAATA-3 Humdr1-F 2797-2814 5-ATGGATCCGGTTTTCCGATCCATGCT-3 CP482-F 3020-3039 5-CCCATCATTGCAATAGCAGG-3 CP483-R 3168-3187 5-AGCATACATATGTTCAAACT-3 Humdr1-F 3105-3124 5-ATGGATCCGGAAGATCGCTACTGAAGCA-3 Humdr1-F 3395-3415 5-ATGGATCCGTCAGTTCATTTGCTCCTGAC-3 Humdr1-R 3402-3421 5-CTGAATTCGGCATAGTCAGGAGCAAATG-3 Humdr1-F 3705-3724 5-ATGGATCCTGCTGCTTGATGGCAAAGAA-3 Humdr1-R 3768-3788 5-CTGAATTCGGATGGGCTCCTGGGACACGA-3 Humdr1-R 4043-4062 5-CTGAATTCACCTTTTCACTTTCTGTATC-3 Humdr1-R 4250-4269 5-CTGAATTCGTTCACTGGCGCTTTGTTCC-3 Humdr1-R 4380-4398 5-ATGGATCCCTCTGAACTTGACTGAGG-3 Humdr1-R 4548-4570 5-CTCTCGAGCAAGGCAGTCAGTTACAGTCCA-3 a  Amplimers were designed according to a previously published sequence (26). The boldface bases in amplimers represent the restriction sites of BamHI, EcoRI, or XhoI for facilitating subcloning. b  F, forward amplimers. c  R, reverse amplimers.

The isolation of total RNA and procedures of RT-PCR were performed as previously reported (22, 23). PCR conditions for each pair of primers were determined by evaluating the linearity of the generation of PCR products with serial dilutions of cDNA (5, 25, 50, and 100 ng per reaction). Reactions were determined to be in the linear range of amplification by comparison of at least two dilutions of each sample and two cycle end points. All PCR experiments were performed on RNAs from several different preparations.

mdr1 PCR products were subjected to heteroduplex analysis using the mutation detection enhancement system (J.T. Baker Inc.). An 8-12% MDE^TM Gel (J.T. Baker) was used and stained with ethidium bromide, and the desired products were located on an ultraviolet transilluminator and photographed. The results were further assessed by DNA sequencing of the PCR products (Amersham Life Science Inc.). Sequences were obtained from several preparations of cDNA, and these were compared with the parental cell line, Dx5 and the published human mdr1 sequence (27).

In order to identify the presence of the mutation in genomic DNA of DxP cells, we isolated genomic DNA from MES-SA, Dx5, and DxP cells. The PCR products were obtained by amplifying the target region using genomic specific primers located in introns 9 and 10, which are adjacent to exon 10 of mdr1: forward primer, 5-ATGGATCCTCTTCACATTCCTCAGGTAT-3; reverse primer, 5-CTCTCGAGGGCCAACTCAGACTTACATTAT-3 (27). Heteroduplex analysis was performed, and heteroduplex bands were subjected to PCR reamplification. The bands were purified from 8% polyacrylamide gel and subcloned into the pGEM-T vector (Promega). Individual clones were selected for sequencing.

Western Blotting of P-gp and Topo II

Western blotting with chemiluminescent detection (ECL kit, Amersham Corp.) was used for the analysis of P-gp and Topo II proteins. Total cell lysates from the exponentially growing cells were used for P-gp immunoblotting (22, 23), and a nuclear protein extract, prepared by the methods previously described (23), was utilized in the Topo II assay.

Immunohistochemical analyses were performed using the monoclonal antibodies MRK16 and C219 for P-gp and the antibody LRP-56 for the p110 major vault protein (Caltag, S. San Francisco, CA) (20, 22, 23).

Plasma membranes for photoaffinity labeling were prepared as described (28), followed by assays of displacement of the P-gp probes [¹²⁵I]iodoarylazidoprazosin and [³H]azidopine (29). MES-SA, Dx5, and DxP cells (100 µg of protein) were incubated with [¹²⁵I]iodoarylazidoprazosin (81.4 TBq/mmol; DuPont NEN) and [³H]Azidopine (Amersham) in the presence or absence of modulators in a 10 mM Tris-HCl buffer (pH 7.4) at a final volume of 50 µl. These preparations were incubated at 25 °C for 1 h in the dark, followed by a 20-min exposure to a 254-nm UV source (UVP, Inc., San Gabriel, CA). 50 µl of loading buffer was added, and proteins were separated on an 8% SDS-polyacrylamide gel, dried, and analyzed by autoradiography.

Subcloning of cells was performed by limiting dilution. Both resistant and revertant clones were analyzed for the MDR phenotype, drug accumulation, and RT-PCR or heteroduplex as described above.

A cDNA library was constructed from DxP cells in gt10 phage, and mutant mdr1 clones were identified by hybridization, digested with NotI, and subcloned into the expression vector NotI site of pcDNA3 (Invitrogen). The plasmid pcDMDR1.4, which codes for wild-type P-gp was obtained by subcloning a BamHI-XhoI fragment of pMDR2000XS containing mdr1 sequence from position 140 to position 4240 into pcDNA3 multiple cloning sites. An additional mutant mdr1 plasmid, pcDMDR3.5, was derived by replacing a 211-base pair NsiI-SfuI fragment of mdr1 from pMDR2000XS with a PCR fragment from DxP cells spanning the position Phe³³⁵ deletion. Transfection of drug-sensitive MES-SA cells and K562 cells was performed by electroporation (30) and Lipofectin-mediated DNA transfer (Life Technologies, Inc.). Recipient cells (2 × 10⁷), were transfected with 10 µg of plasmid. In a transient transfection assay, the MDR phenotype was tested after 48 h of incubation. Stably transfected cells were obtained by selection with G418 (800 µg/ml) for 14 days.

RESULTS

Stepwise selection of the MDR human sarcoma cell line Dx5 in the presence of increasing DOX concentrations and constant exposure to PSC (2 µM) resulted, over a 1-year period, in the stably DOX- and PSC-resistant cell line, DxP. Karyotypic analysis revealed 47 or 48 chromosomes in both the Dx5 and DxP cells with a similar G-banding pattern of chromosome 7 (data not shown). Fluorescence in situ hybridization demonstrated two copies of chromosome 7 in both Dx5 and DxP cells, whereas parental, drug-sensitive MES-SA cells have 45 chromosomes and one chromosome 7 (24, 25).

The drug resistance phenotype of DxP cells is shown in Table II and Fig. 1. Their level of resistance to the anthracycline DOX is similar to that of the parental Dx5 cells, with a slightly decreased cross-resistance to daunorubicin. The most notable alterations were found in levels of resistance to dactinomycin and Vinca alkaloids. Resistance to vinblastine decreased 17-fold, to vincristine 25-fold, and to vinorelbine 9-fold (to a level similar to that of drug-sensitive MES-SA cells). Resistance to amsacrine decreased 13-fold, and sensitivity to dactinomycin was completely restored. This cell line maintained high levels of resistance to colchicine and paclitaxel (Taxol®), and moderately decreased resistance to podophyllotoxins (etoposide and teniposide). There was no significant difference in cellular generation time (22 h) compared with parental Dx5 cells.

Table II. The multidrug resistance phenotype of Dx5 and DxP cells Cytotoxin -Fold Resistancea MES-SA Dx5 DxP Dx5/DxPb Anthracyclines   Daunorubicin 1  (20)c 75 25 3   Doxorubicin 1  (40) 80 77 1 Anti-microtubule   Colchicine 1  (1) 79 75 1   Paclitaxel 1  (1) 175 100 2   Vinblastine 1  (4) 243 14 17   Vincristine 1  (2) 199 8 25   Vinorelbine 1  (60) 17 2 9 Epipodophyllotoxins   Etoposide 1  (500) 42 17 2   Teniposide 1  (100) 135 53 3 Others   Amsacrine 1  (100) 26 2 13   Dactinomycin 1  (0.3) 183 1 183 a  -Fold resistance was determined as described under "Experimental Procedures." Values are the means of at least three to six independent experiments. b  -Fold resistance difference in DxP cells compared with parental Dx5 cells. c  Values in parenthesis represent the IC50 values (nM of drug-sensitive MES-SA cells.

Modulation of resistance to several MDR-related cytotoxins by PSC (2 µM), cyclosporine (3 µM), and verapamil (6 µM) was examined by the MTT assay. PSC, the most potent of the modulators used, completely restored the sensitivity of the highly multidrug-resistant cell line Dx5 to DOX, vinblastine, paclitaxel, vincristine, and colchicine, relative to the drug-sensitive MES-SA cells. In contrast, DxP cells were almost completely resistant to the modulating effects of PSC, substantially resistant to cyclosporine, and somewhat less resistant to modulation by verapamil (Table III, Fig. 1).

Table III. Modulation of multidrug resistance in Dx5 and DxP cells Cytotoxin -Fold resistancea MES-SA Dx5 DxP Doxorubicin 1 80 77   +2 µM PSC 1 1 53   +3 µM Cyclosporine 0.5 2 30   +6 µM Verapamil 0.5 4 18 Paclitaxel 1 175 100   +2 µM PSC 0.3-1b 1 5   +3 µM Cyclosporine 0.5-1 13 38   +6 µM Verapamil 0.3-1 25 44 Vinblastine 1 243 14   +2 µM PSC 0.3-1 1 12   +3 µM Cyclosporine 0.5-1 6 7   +6 µM Verapamil 0.4-1 4 3 Vincristine 1 199 8   +2 µM PSC 0.7 1 7   +6 µM Verapamil 0.3-1 7 3 Colchicine 1 79 75   +2 µM PSC 1 1 70   +3 µM Cyclosporine 1 6 30   +6 µM Verapamil 1 12 30 Etoposide 1 42 18   +2 µM PSC 1 1 18 a  Fold resistance was determined as described under "Experimental Procedures." b  Increased sensitivity to cytotoxins was observed occasionally in MES-SA cells treated with MDR modulators, although these cells do not express mdr1.

Analysis of mdr1, Topo II and II, and mrp Expression

Total RNA was extracted from isolated DxP clones and analyzed by RT-PCR for the presence of mdr1, Topo II, Topo II, and mrp transcripts (Fig. 2). In comparison with the parental Dx5 cells, DxP displayed levels of mdr1 transcripts similar to the parental Dx5 cells. The expression of mrp was compared with MES-SA and Dx5 cells and with normal human lung tissue as a positive control. A similar level of expression of mrp was observed in human lung tissue, MES-SA sarcoma cells (data not shown), and Dx5 and DxP cells. While no significant difference was seen in Topo II transcripts, DxP cells manifested a 2-fold elevation of Topo II expression relative to Dx5 cells (Fig. 2).

Analysis of mdr1, mrp, Topo II, and Topo II gene expression in DxP cells.

P-gp expression and function were analyzed by RT-PCR, immunoblotting with the C219 antibody, and flow cytometry with the antibody UIC2 (Figs. 3 and 4). Dx5 and DxP cells displayed similarly high levels of expression of P-gp. Immunohistochemical staining of cells with the MRK16 and C219 antibodies confirmed that the expression of P-gp on DxP cells was predominantly localized to the plasma membrane and similar in amount to that observed in Dx5 cells (data not shown).

Topo II and p110 Expression

Expression of Topo II isoforms by immunoblotting revealed a slight increase in DxP compared with Dx5 cells. Neither Dx5 nor DxP cells had detectable expression of the p110 major vault protein, although the drug-sensitive MES-SA cells from which Dx5 cells were originally derived are weakly positive for p110 (data not shown).

The cellular accumulation of the fluorescent P-gp substrate Rh-123 was markedly decreased in Dx5 cells, as expected for MDR cells (Fig. 4). In contrast, DxP cells (although expressing abundant P-gp by UIC2 staining) manifested Rh-123 retention similar to the MDR-negative MES-SA cell line. PSC (2 µM) completely restored Rh-123 accumulation in the Dx5 cell line, to levels similar to MES-SA and DxP cells (Fig. 4).

Intracellular accumulations of [³H]daunorubicin, [³H]vinblastine, and [³H]cyclosporine were measured to compare the function of P-gp in Dx5 and DxP cells. Both Dx5 and DxP cells displayed similar decreases in [³H]daunorubicin accumulation, relative to the MES-SA cell line, while the accumulation of [³H]vinblastine was approximately 3-fold higher in DxP than Dx5 cells (Fig. 5, A and B). The decreased drug accumulation in DxP cells was not modulated by PSC, whereas complete modulation was displayed by Dx5 cells (Fig. 5). Uptake of [³H]cyclosporine was examined to assess the ability of P-gp to transport the cyclosporins. The intracellular concentration of cyclosporine in DxP cells was equal to that in MES-SA cells and 2-fold higher than in Dx5 cells (Fig. 6). The decreased accumulation of cyclosporine in Dx5 cells was completely modulated by PSC to the levels achieved in DxP and MES-SA cells (data not shown).

DxP cells displayed enhanced photoaffinity binding of P-gp by [¹²⁵I]iodoarylazidoprazosin in the presence of 0.1 and 10 µM PSC and vinblastine, in contrast to Dx5 cells in which the photoaffinity labeling was effectively competed by PSC and vinblastine (Fig. 7). The higher concentration (100 µM) of PSC or vinblastine abolished detectable P-gp labeling by azidoprazosin in both cell lines. DxP cells were also resistant to the displacement by PSC or vinblastine of [³H]azidopine photoaffinity labeling (Fig. 8, A and B). Verapamil was moderately active in both DxP and Dx5 cells in displacing [³H]azidopine.

RT-PCR using primer sets spanning the P-gp coding sequences (Table I) confirmed that the levels of expression of mdr1 were similar and that the PCR products showed no differences in size when Dx5 and DxP cells were compared. DNA heteroduplex analysis revealed the formation of a heteroduplex with primers spanning nucleotides 1194-1519 of mdr1 cDNA, suggesting a sequence difference in transmembrane region 6 (TM6), Fig. 9A. Sequencing of this PCR product identified a deletion of base pairs 1427-1429 in this region, which encode the amino acid phenylalanine at position 335 of P-gp (Phe³³⁵) (Fig. 9, B and C). The PCR and sequencing results were reproduced in four different preparations of cDNA from DxP cells. The deletion of Phe³³⁵ is the only functional mutation in DxP cells compared with Dx5 and the published human mdr1 sequence (27). There are other changes from the published mdr1 sequence that do not alter the P-gp amino acid sequence in DxP cells: from TCT (Ser) to TCC (Ser) in nucleotide 964 and from ATC (Ile) to ATT (Ile) in nucleotide 3859. These may be polymorphisms of P-gp, since both DxP and Dx5 have these same substitutions. Amino acid 185 was identified as Gly in both Dx5 and DxP cells (data not shown).

In order to identify the presence of the mutation in genomic DNA of DxP cells, we amplified genomic DNA using genomic specific primers of mdr1. Heteroduplex analysis was performed and revealed that a heteroduplex band existed in DxP but not Dx5 cells. The TTC deletion at codon 335 was also verified by sequencing of the genomic PCR product. Thus, the one-codon deletion was confirmed in DxP cells (Fig. 10).

Single clones were obtained to analyze the genetic heterogeneity in DxP cells. As shown in Table IV, 11 isolated clones derived from DxP cells showed a similar phenotype including resistance to DOX, resistance to modulation by PSC, and a lower degree of resistance to vinblastine relative to parental Dx5 cells. All tested clones expressed the mutant mdr1.

Table IV. Drug resistance phenotypes of subclones of DxP cells Cells DOXa DOX + PSC VBLb VBL + PSC mdr1c Controls   MES-SA 1 1 1 0.5 Negative   Dx5 80 1 243 1 WT   DxP 77 53 14 12 Mut Subclonesd   Dx/PAS 163 51 15 5 Mut   Dx/PBS 67 13 3 1 NDe   Dx/PCS 50 23 8 3 Mut   Dx/PDS 150 100 20 6 Mut   Dx/PES 96 100 3 1 Mut   Dx/PFS 150 75 21 8 ND   Dx/PGS 7 7 2 1 ND   Dx/PHS 83 37 ND ND ND   Dx/PIS 13 3 2 1 Mut   Dx/PJS 175 75 19 10 ND   Dx/PMS 100 50 5 2 Mut a  The numbers represent -fold resistance relative to control, drug-sensitive MES-SA cells as determined by the MTT assay. b  VBL, vinblastine. c  Identification of mdr1 gene expression by RT-PCR and DNA heteroduplex assay. WT, wild-type mdr1; Mut, mutant mdr1 mRNA (deletion of the codon for amino acid 335). d  Subclones of DxP were obtained by limiting dilution of cell populations in 96-well plates. The clones were maintained in drug-free conditions over 2 months and were tested for their drug resistance phenotypes. e  ND, not determined.

Plasmids containing wild-type and mutant mdr1 cDNAs were transfected into drug-sensitive MES-SA cells. Expression of the mutant mdr1 was confirmed in the appropriate transfectants by DNA heteroduplex analysis. Both wild-type and mutant mdr1 conferred resistance to doxorubicin, paclitaxel, etoposide, and colchicine. Representative data demonstrating resistance to doxorubicin are shown in Fig. 11A. Exposure of transfectants to both doxorubicin and PSC revealed that only the cells transfected with the mutant mdr1 bearing the Phe³³⁵ deletion survived, Fig. 11B.

DISCUSSION

The development of resistance to anticancer drugs is a major impediment to successful chemotherapy, and it is often mediated by the membrane-bound drug-efflux pump, P-gp (1-4). Substances that inhibit P-gp function and reverse the resistance phenotype in vitro, termed MDR modulators, have been developed with the intention of administering them in conjunction with MDR-related cytotoxins (9-15). This experiment was designed to examine the resistance mechanisms that arise in MDR cells during multistep selection with DOX, an MDR substrate, and PSC, an effective modulator. A similar selection in non-MDR cells suppressed the activation of mdr1 and resulted in the emergence of mutants expressing alternative mechanisms of resistance, notably decreased expression of Topo II (23). Under the conditions of drug exposure in our present experiment, DxP cells displayed cross-resistance to several MDR-related drugs including anthracyclines (DOX and daunorubicin), epipodophyllotoxins (etoposide and teniposide), colchicine, and paclitaxel. However, DxP differed from the parental Dx5 cells in their decreased resistance to Vinca alkaloids and lack of cross-resistance to dactinomycin (Table II, Fig. 1). Most notably, the MDR phenotype was not modulated by treatment with the P-gp inhibitor PSC (Table III, Fig. 1).

Although overexpression of mdr1 is the best characterized mechanism of pleiotropic drug resistance, other mechanisms have been identified. Decreased expression or altered structure of Topo II has been observed in many models of resistance to epipodophyllotoxins, mitoxantrone, and anthracyclines such as DOX (16-18, 23). A membrane ATPase of 190 kDa, distinct from P-gp, has been termed the multidrug resistance-associated protein, encoded by the mrp gene, which has been cloned and sequenced in a DOX-selected lung cancer cell line (19). Under our experimental conditions, overexpression of the mrp mRNA or significant changes in Topo II and II were not observed. DxP cells, like the parental Dx5 cells, were not found to express the p110 major vault protein, recognized by the LRP-56 antibody and associated with doxorubicin resistance in some cell models (21). The apparent lack of an alternative resistance mechanism in DxP cells, the residual high expression of P-gp, and the pleiotropic nature of the resistance suggested that a mutant or modified P-gp with decreased affinity for cyclosporins was responsible for the phenotype of these cells.

The altered phenotype of DxP cells correlated very well with an altered functional activity of P-gp assessed by the cellular uptake of daunorubicin, vinblastine, and cyclosporine. DxP cells displayed a significant decrease in daunorubicin accumulation, which was insensitive to modulation by PSC (Fig. 5A). These cells also exhibited increased vinblastine accumulation compared with Dx5 cells, although the level of this drug was not as high as in drug-sensitive MES-SA cells and was not further increased by PSC (Fig. 5A). The accumulation of cyclosporine in DxP cells was equivalent to that of drug-sensitive MES-SA cells, which do not express P-gp, strongly suggesting an altered affinity of the multidrug transporter for cyclosporins and consistent with the data that cyclosporine and its analogue PSC did not modulate the MDR phenotype of DxP cells (Fig. 6).

As previously reported, compartmentalization or redistribution of P-gp leading to redistribution of cytotoxins may result in resistance to modulation (31). Our immunohistochemical experiments localized P-gp to the cell membrane in both Dx5 and DxP. Furthermore, the same amount of P-gp expression and the existence of equivalent daunorubicin accumulation defects in the two cell lines demonstrated that P-gp in DxP cells was capable of transporting some substrates as well as the P-gp in Dx5 cells (Figs. 3, 4, and 5A). The P-gps from the two cell lines have a similar electrophoretic mobility (Fig. 3). Thus, a redistribution or marked structural change in the P-gp expressed in DxP cells was not evident.

Our results identified a novel mutation, consisting of a single codon deletion (Phe³³⁵) in the TM6 region of P-gp in DxP cells. The drug resistance phenotype we observed is similar, in some respects, to that described in a previously published report of the functional consequences of a substitution for the phenylalanine at codon 335 by site-directed mutagenesis, which resulted in decreased dactinomycin resistance (32). The expression of this mutant P-gp in DxP cells is probably the result of selection of a spontaneously arising mutant or selective allelic expression of an mdr1 gene that confers a PSC-resistant MDR phenotype. Karyotyping and fluorescent in situ hybridization analysis of chromosome 7 revealed no obvious alterations in chromosome structure in DxP cells.

The deletion of codon 335 as a mechanism of resistance to modulation of MDR by cyclosporins may be an extremely rare event. We have not found a similar mutation in 13 other mutants separately derived from MES-SA cells that were co-selected by DOX and PSC, with a mutation rate of 2.5 × 10⁷ per cell generation (23). Those mutants were resistant to DOX and PSC on the basis of a down-regulation of Topo II expression. The MES-SA cells used in that experiment do not express mdr1, whereas the MDR variant Dx5 cells used to select the DxP variant with DOX and PSC actively transcribed mdr1, a factor that may have contributed to the selection of a mutant P-gp rather than to an alternative mechanism of resistance such as altered Topo II expression.

Several point mutations or polymorphisms of mdr1 have been identified in different species including hamsters, mice, and humans (reviewed in Ref. 33; Refs. 34-38). Some of these have been associated with an altered phenotype, such as the substitution from Gly¹⁸⁵ to Val¹⁸⁵ in human P-gp, which confers preferential resistance to colchicine (36, 37). The colchicine selection of KB cells in this case may have favored overexpression of one allele of mdr1 and suggests that codon 185 in human P-gp may be subjected to DNA polymorphism. Both Dx5 and DxP cells express the Gly¹⁸⁵. The impact of structural alterations of P-gp on modulation of MDR by inhibitors is poorly understood (9, 32, 33, 38).

The novel mutation we have identified in this study (deletion of Phe³³⁵) provides insight into the relationship between P-gp structure and modulation of MDR by cyclosporins. Our data support the theory that a specific ligand-receptor mechanism is involved in P-gp-mediated MDR. PSC and cyclosporine are known to bind to P-gp, and cyclosporine is a transport substrate for P-gp. Phenylalanine and other aromatic residues are preferentially located at the cytoplasmic boundaries, where they are thought to position the transmembrane segments. Several such residues, including Phe³³⁵, Phe⁷⁷⁷, and Phe⁹⁷⁸, are thought to be of functional importance in P-gp (32). Substitution of a nonaromatic residue for Phe³³⁵ substantially impairs the ability of the transporter to confer resistance to vinblastine and dactinomycin, while the ability to confer resistance to DOX and colchicine is preserved (32).

DxP cells showed enhanced iodoarylazidoprazosin labeling in the presence of both PSC and vinblastine and decreased ability of either PSC or vinblastine to replace the P-gp probe (Fig. 7). Photoaffinity experiments have localized azidopine binding domains in P-gp to TM5-6 or to TM6 and TM12 (reviewed in Ref. 33). Previous experiments have demonstrated that the two halves of P-gp come together to form a single site for drug binding, involving the TM5-6 and TM11-12 regions (33). Additional data suggest that azidopine and vinblastine may not bind to the same site (33, 39) and that conformational or allosteric effects could be responsible for the inhibition of labeling by azidopine in the presence of vinblastine. The collateral increased affinity to iodoarylazidoprazosin labeling in DxP cells may be due to an allosteric effect of the deletion of Phe³³⁵. The deletion of Phe³³⁵ in the P-gp expressed by DxP cells resulted in loss of the capacity to bind or transport cyclosporine, PSC, and vinblastine. Our results suggest that cyclosporine, PSC, vinblastine, Rh-123, and dactinomycin share at least one binding domain on P-gp (Tables II and III, Figs. 4, 5, 6, 7). These results indicate that this residue plays an important role in the interaction of P-gp with cyclosporine and PSC.

Our results do not completely rule out the existence of other modifying factors that may directly or indirectly affect the multidrug resistance phenotype in DxP cells, although no obvious alternative mechanism was observed. Several major known alternative mechanisms of resistance to DOX were not different between Dx5 and DxP cells. Moreover, the clonal analysis of the DxP cell population supported the association of the mutant P-gp expression with resistance to modulation by PSC and decreased cross-resistance to vinblastine in every subclone tested (Table IV). Finally, transfection of the mutant mdr1 gene conferred a drug resistance phenotype that was resistant to modulation by PSC (Fig. 11).

In summary, our data reveal a functionally important mutation of P-gp arising from co-selection of mdr1-positive cells with DOX and PSC. The resistance phenotype of the resulting DxP cell line may be attributed to the deletion of Phe³³⁵ from P-gp. Our data suggest that the phenylalanine residue at codon 335 may be important in the binding and transport of cyclosporins by P-gp and to their ability to modulate MDR. In addition, this mutation results in decreased resistance to Vinca alkaloids, lack of cross-resistance to dactinomycin, and markedly decreased ability to transport Rh-123.