Topology Mapping of the Amino-terminal Half of Multidrug Resistance-associated Protein by Epitope Insertion and Immunofluorescence*

The multidrug resistance-associated protein (MRP) is an integral membrane protein that causes multidrug resistance when overexpressed in mammalian cells. Within the ATP-binding cassette superfamily, MRP belongs to a subgroup of structurally and functionally related proteins that includes the yeast cadmium factor 1 and yeast oligomycin resistance I proteins, and the mammalian sulfonylurea receptors SUR1 and SUR2. Hydropathy analysis of these proteins predicts a unique membrane-associated region at the amino terminus followed by a structural unit composed of 12 transmembrane (TM) domains and two nucleotide-binding domains that is characteristic of eukaryotic ATP-binding cassette transporters. The topology of the membrane-associated regions of MRP remains largely unknown and was investigated. Small hemagglutinin epitopes (YPYDVPDYAS) were inserted in predicted hydrophilic segments of the membrane-associated regions from the amino-terminal half of MRP and these proteins were expressed in HeLa cells, and tested for their capacity to confer etoposide resistance. The polarity of the inserted tags with respect to plasma membrane was then deduced by immunofluorescence in intact and permeabilized cells. Insertion of epitopes at positions 4, 163, 271, 574, and 653 produced functional proteins while insertions at positions 127, 417, 461, and 529 abrogated the capacity of MRP to confer drug resistance. Epitopes inserted at positions 4, 163, and 574 were localized extracellularly, whereas those inserted at positions 271 and 653 revealed an intracellular location. Although a single epitope inserted at position 461 was compatible with MRP function, it was inaccessible to the anti-epitope antibody and two copies of the tag at that site abrogated MRP function. These results indicate that the amino terminus of MRP is extracellular, while the linker segment joining the first and second membrane-associated regions is intracellular as is the first nucleotide-binding domain. Our findings are therefore consistent with a topological model of MRP that contains 5 TM segments in the first membrane-associated region and 6 TM segments in the second membrane region.

The emergence of cellular resistance to structurally unre-lated cytotoxic drugs (multidrug resistance) is a major limitation to the chemotherapeutic treatment of many types of human tumors (1). In vitro, multidrug resistance is caused by the overexpression of members of the P-glycoprotein and the multidrug resistance-associated protein (MRP) 1 gene families (2,3). P-glycoprotein and MRP are both integral membrane proteins (4,5) that have been proposed to act as membrane pumps with broad substrate specificity (6 -9).
MRP is a small gene family composed of up to five members. MRP1, which was initially cloned from Adriamycin-resistant H69AR small cell lung carcinoma cell line (10), was shown to confer resistance to cytotoxic drugs such as vinca alkaloids and epipodophyllotoxins in transfection studies (11). Studies in inside-out membrane vesicles suggest that MRP1 transports glutathione conjugates (12), most notably, cysteinyl leukotrienes (13,14), anionic conjugates of bile salts and steroid hormones (15), and glutathione conjugates of chemotherapeutic agents (9). The transport of leukotriene C 4 by MRP can be inhibited by MK571, a leukotriene D 4 receptor antagonist (13,16). MRP2 (cMOAT) is expressed predominantly on the canalicular membrane of hepatocytes and is responsible for the secretion of amphiphilic anionic conjugates from hepatocytes into the bile. TR Ϫ mutant rats characterized by a liver defect in biliary transport of organic anions show a single nucleotide deletion in the rMrp2 gene (17,18). Likewise, a mutation in human MRP2 has been detected in Dubin-Johnson syndrome, a pathology characterized by a defect in hepatic multispecific organic anion transport (19,20).
Both MRP and P-glycoprotein belong to a larger gene family known as the ATP-binding cassette (ABC) superfamily of transport proteins which has been highly conserved in the evolution of eukaryotes and prokaryotes (21). This group of transporters is characterized by the presence of two membrane-associated (TM) units, and two highly similar nucleotide-binding domains (NBD) with characteristic Walker consensus motifs (22). These structural units can be assembled from independent peptides exemplified by the bacterial His and Mal transporters (21), as TM-NBD heterodimers characteristic of TAP1/TAP2 (23)(24)(25), or they can be fused in the same polypeptide as seen in P-glycoprotein and CFTR (21). Within this gene family, MRP belongs to a subgroup of structurally and functionally distinct proteins that include the yeast cadmium resistance factor, yeast cadmium factor 1 (YCF1) (43% amino acid sequence identity; 26), the yeast oligomycin resist-ance I protein YORI (33% identity) that mediates oligomycin resistance (27), the Leishmania tarentolae P-glycoprotein A (32% identity) associated with resistance to arsenicals and antimony (28,29), and the mammalian sulfonylurea receptors SUR1 and SUR2 (rSUR1: 29% identity) (30,31).
In contrast to P-glycoprotein (32), CFTR (33) and Pgh-1 (34), which are formed by two sequence homologous halves encoding 6 putative TM domains and one NBD (35)(36)(37)(38), hydropathy analysis of the MRP group identifies a unique additional membrane-associated region at the amino terminus of the protein (39,40). Several topological models have been proposed based on computer-assisted analysis of the primary amino acid sequence of MRP and of other members of this subgroup of ABC transporters (39,40). These analyses suggest 3 major membrane-associated regions in MRP. The first one (residues 1-190) may contain 4, 5 (41), or 6 TM segments (42), while the second and third regions would each contain 6 TM domains as in P-glycoprotein and CFTR (41,42). However, direct topological data remains limited for MRP. Indeed, such data is limited to observations made by epitope mapping (43)(44)(45) and accessibility to proteolytic cleavage in intact cells and isolated membranes (41), and suggests that residues 192-360 (L-1), 918 -924 (L-2), and 1294 -1430, and/or 1497-1531 (NBD-2) of MRP are intracellular.
We have used a biochemical approach to initiate topology mapping of the membrane-associated regions of MRP. This method is based on the insertion of a 10-amino acid antigenic peptide (HA epitope from influenza virus) in predicted hydrophilic peaks from the TM regions of MRP (37). These mutant proteins are then expressed in HeLa cells to test their capacity to confer drug resistance, and the polarity of the epitope tag with respect to the plasma membrane (intra-versus extracellular) is then determined by immunofluorescence in intact or permeabilized cells using a monoclonal antibody directed against the epitope tag. Here, we report on the topology mapping of the two membrane-associated regions from the aminoterminal half of MRP, including the unique domain characteristic of the MRP class of ABC transporters. Site-directed Mutagenesis-HA epitopes were inserted at discrete locations in MRP as 10-amino acid peptides (YPYDVPDYAS) by sitedirected mutagenesis of the cDNA, as described previously (36). In addition, the mutagenic oligonucleotides were designed to contain a unique NheI site to facilitate the insertion of additional tags. The positions of epitope insertions in the amino acid sequence of the protein are listed in Fig. 1, and the oligonucleotides used for mutagenesis are shown in Table I. Briefly, a full-length cDNA for human MRP cloned into the plasmid vector pBluescript (pBl/MRP) was digested with SacI and SphI, and a 2.7-kilobase restriction fragment (polylinker to position 2695) that corresponds to the 5Ј-half of the MRP cDNA was cloned into the corresponding sites of M13mp18 (M13/MRP). To facilitate subsequent cloning procedures, two silent mutations generating novel SacII (position 12; mutagenic oligonucleotide 3Ј-CGTACCGCGAGGCGC-CGAAGACGTC-5Ј) and AflII (position 1546; mutagenic oligonucleotide 3Ј-CTAGTTTCACGAATTCGAAATACGGAC-5Ј) restriction sites were introduced in the MRP cDNA. For this, mutagenic oligonucleotides were annealed to single-stranded M13/MRP DNA template and mutagenesis carried out using a commercially available in vitro kit. The 2.7-kilobase SacI to SphI fragment containing novel SacII and AflII sites was reinserted into pBl/MRP prior to cloning of the full-length modified MRP cDNA (as a SacI to KpnI fragment) in the mammalian expression vector pCB6 (46). Insertional mutagenesis in the aminoterminal half of MRP (constructs 2-5, 6a, 7, and 8a) was carried out by in vitro mutagenesis using single strand M13/MRP DNA template and the oligos listed in Table I. The integrity of the mutagenized MRP cDNA inserts was verified by nucleotide sequencing (47), and the modified MRP cDNA inserts were subcloned either as SacII/BamHI (constructs 2, 3, and 4), BamHI/AflII (constructs 5 and 6a), or AflII/Eco NI (constructs 7 and 8a) fragments into pCB6/MRP. In the case of mutants 1 and 9, the HA epitope was introduced by cloning a short doublestranded segment composed of two complementary oligonucleotides into pre-existing restriction sites, as described previously (37). HA epitopes were inserted in this way at nucleotide positions 12 (SacII site) and 1959 (Eco NI site) using oligos with cohesive ends compatible with the respective cloning site. A similar cloning strategy was used to insert a second consecutive HA epitope in constructs 6b and 8b, using the unique NheI site present in the first inserted HA epitope.

Materials-Genetycin
Cell Culture-Drug-sensitive human HeLa cells (ATCC) were grown in Dulbecco's modified Eagle's medium (high glucose) supplemented with 10% fetal calf serum, 2 mM glutamine, penicillin (50 units/ml), and streptomycin (50 g/ml). pCB6/MRP constructs were transfected into HeLa cells by calcium phosphate co-precipitation, as we have previously described (36). After 2 days, cells were subcultured at a 1:5 dilution and stable transfectants were selected in medium containing G418 (600 g/ml). Mass populations of G418 R transfectants were harvested after 9 days and subcultured in medium containing the VP-16 (final concentration 250 ng/ml) to select mass populations of drug-resistant cell clones overexpressing the tagged MRP proteins. VP-16 R populations were harvested 1-2 weeks later, expanded in culture, and frozen at Ϫ80°C in 90% serum and 10% dimethyl sulfoxide.
Membrane Preparation and Western Blotting-Crude membrane fractions from transfected HeLa cells were isolated as described previously (48). Briefly, cells were grown to 70% confluency and harvested in cold PBS (phosphate-buffered saline) containing sodium citrate. The cell pellet was homogenized (Dounce homogenizer, 25 strokes) in buffer containing 1 mM MgCl 2 and 10 mM Tris, pH 7.0, supplemented with protease inhibitors leupeptin, 2 g/ml, aprotinin, 2 g/ml, and pepstatin, 1 g/ml. Unbroken cells and nuclei were removed by centrifugation (400 g/10 min) and a crude membrane fraction was prepared by centrifugation of the supernatant fraction at 100,000 ϫ g for 60 min. The protein concentration in the crude membrane fraction was determined by the method of Bradford using a commercially available reagent (Bio-Rad). For immunodetection of recombinant MRP, 25 g of protein was resolved on a SDS-containing 6.5% polyacrylamide gel and transferred by electroblotting to nitrocellulose membranes. The immunoblots were incubated with the MRP-specific monoclonal antibody QCRL-1 at a dilution of 1:150, or the monoclonal anti-HA antibody 16B12 at a 1:3000 dilution, and specific immune complexes were revealed using a horseradish peroxidase-conjugated sheep anti-mouse antibody (1:10000) (Amersham).
Cytotoxicity Assay-Drug cytotoxicity assays were performed using sulforhodamine B to stain cellular proteins, as described previously (49). For this, 6 ϫ 10 3 cells from VP-16 R mass populations expressing independently tagged MRP proteins and drug-sensitive HeLa control cells were seeded in 96-well titer plates with Dulbecco's modified Eagle's medium containing increasing concentrations of ADM, ACTD, VCR, or VP-16. The cells were incubated at 37°C for 96 h, fixed in 17% trichloroacetic acid in PBS, and cellular protein was stained for 10 min at room temperature with 0.4% sulforhodamine B in a 1% acetic acid solution. The plates were washed with water, dried, and the stain dissolved in 0.2 ml of 10 mM Tris (pH 9). Quantification of sulforhodamine B was done using an automated enzyme-linked immunosorbent assay plate reader (Bio-Rad Model 450) at a wavelength of 490 nm. The relative plating efficiency of each clone was determined by dividing the absorbance observed at a given drug concentration by the absorbance detected in the same clone in the absence of drug.
Immunofluorescence-Localization of the epitope was performed by immunofluorescence on 3 ϫ 10 4 cells from VP-16 R mass populations of individually tagged MRP transfectants that had been grown on glass coverslips for 2 days. Drug-sensitive HeLa cells were used as a negative control. Nonpermeabilized HeLa cells were first incubated with the monoclonal antibody 16B12 (1:250) in PBS containing 5% goat serum and 1% bovine serum albumin for 1 h at 4°C. The cells were then fixed in 4% paraformaldehyde in PBS, and permeabilized and blocked with PBS, containing 0.05% Nonidet-P40, 5% goat serum, and 1% bovine serum albumin at room temperature for 15 min. A secondary antibody (Rhodamine-conjugated goat anti-mouse IgG or cyanine-conjugated goat anti-mouse IgG) was applied in the same buffer (1:200 or 1:800 dilutions, respectively). In experiments with permeabilized cells, the cells were fixed and permeabilized as described above, and blocked with PBS, containing 0.05% Nonidet P-40, 5% goat serum, and 1% bovine serum albumin. The cells were then incubated with the primary antibody (1:1000 dilution) for 1 h at room temperature prior to exposure to the secondary antibody (1:200 or 1:800 dilutions). Immunofluorescence microscopy was performed using standard epifluorescence optics (Nikon).

Construction and Expression of Epitope-tagged Mutant
MRPs-Combined hydropathy analysis of MRP and other ABC transporters suggests a protein with two hydrophilic ATPbinding domains (amino acids ϳ615-840 and ϳ1260 -1532), two membrane-associated regions in the amino-terminal half of the protein (amino acids ϳ1-190 and ϳ310 -615), and one membrane-associated region in the COOH-terminal half of the protein delineated by residues ϳ970 and 1260 (Fig. 1A). In addition, subjecting the MRP sequence to different hydropathy algorithms, hydrophobic moment analyses, or TMAP and TO-PRED analyses suggested different numbers and positions of TM domains in the three highly hydrophobic regions of MRP (data not shown). To resolve these discrepancies, we have analyzed the membrane topology of the two predicted membraneassociated regions in the amino-terminal half of MRP (residues 4 -653) by the epitope insertion method that we previously used to determine the membrane topology of P-glycoprotein (37). This involved insertion of the antigenic epitope tag YPY-DVPDYAS (HA) in some of the most hydrophilic peaks of these membrane-associated regions (Fig. 1A). Single epitopes were inserted at amino acid positions 4 (construct 1), 127 (construct 2), 163 (construct 3), 271 (construct 4), 417 (construct 5), 461 (construct 6a), 529 (construct 7), 574 (construct 8a), and 653 (construct 9) (Fig. 1, B and C). Since the single epitopes inserted at positions 461 and 574 were found to be poorly acces-sible to the anti-epitope antibody (see below; Fig. 5), two copies of the HA epitope were inserted at these positions (construct 6b and 8b, respectively). The wild-type MRP and mutant HAtagged MRP cDNAs (numbered 1-9) were cloned into the mammalian expression vector pCB6 and transfected into HeLa cells. Stable transfectants were selected in G418 and mass populations of G418 R colonies were further selected in VP-16 (250 ng/ml). Drug-resistant colonies emerged within 1 to 2 weeks of selection when cells had been transfected with the wild-type or mutant MRP 1, 3, 4, 6a, 8a, 8b, and 9 constructs. However, cells transfected with the pCB6 vector alone or constructs 2, 5, 6b, and 7 gave no drug-resistant colonies even after 4 weeks of selection. These results indicate that HA epitope insertions at positions 4, 163, 271, 461, 574, and 653 or two HA epitopes at position 574 were compatible with MRP protein function. On the other hand, insertion of HA epitopes at positions 127, 417, and 529 and the presence of two HA epitopes at position 461 caused an apparent loss of MRP function. Epitope tags insertion at these sites may therefore interfere with protein folding or targeting, or affect structural and functional domains that are essential for the drug transport activity of the protein.
Expression of the mutant MRP proteins in HeLa cells was analyzed by Western blotting (Fig. 2, A and B). Crude membrane fractions prepared from control, nontransfected HeLa cells, and VP-16 R mass populations corresponding to wild-type MRP and constructs 1, 3, 4, 6a, 8a, 8b, and 9 were separated on a 6.5% SDS-polyacrylamide gel and analyzed by immunoblotting using the anti-MRP antibody QCRL-1 directed against an epitope located in the linker region connecting the first NBD to the COOH-terminal half of the protein (44, 45) ( Fig. 2A). Specific immunoreactive bands of ϳ180 kDa were observed in membrane fractions from all VP-16 R MRP transfectants that was absent in membranes from control HeLa cells ( Fig. 2A). Immunoblotting of the same membrane preparations was also carried out using the anti-HA epitope antibody 16B12 (Fig. 2B). As expected, a similar immunoreactive band at ϳ180 kDa was detected by this antibody in membrane fractions from all VP-16 R mass populations expressing epitope-tagged MRPs and was absent in membranes from control HeLa cells or VP-16 R cells transfected with wild-type MRP. Whereas cells transfected with constructs 1, 3, 4, 8b, and 9 expressed similar amounts of proteins, expression was lower in cells transfected with constructs 6a and 8a. The membrane fractions from the transfectants showed a second MRP species of slightly faster electrophoretic mobility (ϳ160 kDa) which was immunoreactive with both antibodies. This species may represent a less mature, partially glycosylated form of MRP. Taken together, these results show that the biologically active HA-tagged MRP proteins encoded by the constructs 1, 3, 4, 6a, 8a, 8b, and 9 are expressed in the membrane fraction of these transfectants.
Drug Resistance Profiles of Mutant Proteins 1, 3, 4, 6a, 8a, 8b, and 9 -To determine whether insertion of HA epitopes may have more subtle effects on MRP function, the drug survival characteristics of VP-16 R mass populations expressing constructs 1, 3, 4, 6a, 8a, 8b, and 9 were established for known MRP substrates. The cells were plated in increasing concentrations of ADM, ACTD, VP-16, and VCR (Fig. 3), and the drug concentration required to reduce the plating efficiency of each mass population by 50% (IC 50 ) was calculated. All mass populations of cells expressing these epitope-tagged proteins displayed similar resistance levels to ADM (fold resistance: 5-9 ϫ), ACTD (fold resistance; 4 -7 ϫ), and VP-16 (fold resistance: 9 -12 ϫ), and these were comparable to resistance levels measured in transfectants expressing wild-type MRP. A subtle difference in resistance profiles was observed for VCR, where cells expressing the mutant protein 6a were less resistant (fold resistance: 3 ϫ) than cells expressing wild type MRP (fold resistance: 8 ϫ), and cells expressing mutants 1, 3, 4, 8a, 8b, and 9 displayed a slightly higher resistance to the drug (fold resistance: 14 -17 ϫ). Nevertheless, the drug resistance profiles encoded by MRP mutants 1, 3, 4, 6a, 8a, 8b, and 9 were similar to that of wild-type MRP, indicating that epitope insertions at positions 4, 163, 271, 461, 574, and 653 did not drastically alter MRP protein function.
Localization of the Epitope Tags-To determine the polarity of the epitopes with respect to the plasma membrane, immunofluorescence was performed using the anti-HA monoclonal antibody 16B12 on VP-16 R transfectants expressing individual HA-tagged MRP proteins. Immunofluorescence was carried out using intact cells, or cells that had been permeabilized with Nonidet P-40, to detect extracellular and intracellular tags, respectively (Figs. 4 and 5). No significant cell-associated fluorescence could be detected with the antibody in intact or permeabilized control HeLa cells or in cells expressing the wild-type MRP protein (Fig. 4, A-C). On the other hand, cells that expressed mutant proteins 1 (amino acid position 4), 3 (amino acid position 163), and 8b (amino acid position 574) gave a strong fluorescent signal in both intact and permeabilized cells (Fig. 4, D-I and M-O). This unambiguously demonstrates the extracellular location of the HA epitope in these MRP variants. In contrast, mass populations of cells expressing mutant proteins 4 (position 271) and 9 (position 653) showed bright fluorescence only in permeabilized cells (Fig. 4, L and R), but not in intact cells (Fig. 4, J, K, P, and Q), suggesting an intracellular location of the HA epitope in these MRP variants. Transfectants expressing constructs 4 and 9 also showed perinuclear staining in permeabilized cells in addition to the plasma membrane staining (Fig. 4, L and R). These results suggest that only a fraction of the recombinant MRP protein is processed to the plasma membrane in these cells, and a portion remains associated with intracellular membranous compartments such as the Golgi or the endoplasmic reticulum. On the other hand, an unusually bright signal was observed in intact cells expressing mutant protein 3 (Fig. 4, G and H), suggesting that position 163 is in a portion of the MRP protein that is easily accessible to the antibody.
Transfectants expressing functional mutants 6a and 8a, which contain single HA epitopes inserted at positions 461 or 574, were either weakly immunoreactive (construct 8a; Fig. 5, J, K, and L) or did not react at all (construct 6a; Fig. 5, G, H, and I) with the anti-HA antibody. The reason for the apparent poor accessibility of these epitopes remains unclear but suggests that they are in close proximity to or within membraneembedded TM segments. Addition of a second consecutive HA epitope at position 574 (mutant 8b) allowed the unambiguous localization of this region of MRP to the extracellular face of the membrane (Fig. 4, M and N). However, the fluorescent signal observed in 8b transfectants was still modest compared with that seen with other extracellular epitopes, in agreement with the proposition that the antibody has limited access to the epitope inserted at this residue/segment. A similar double tagging approach at position 461 (construct 6b) appeared to abro- mutant proteins 1, 3, 4, 6a, 8a, 8b, and 9. Control drug-sensitive HeLa cells (ϫ, stippled line) and mass populations corresponding to either wild type MRP (q, dashed line) or MRP constructs 1 (Ⅺ), 3 (⌬), 4 (ࡗ), 6a (E), 8a (OE), 8b (छ), and 9 (f) were plated in increasing concentrations of ADM, ACTD, VCR, or VP-16 and incubated for 96 h. Drug cytotoxicity was measured using a sulforhodamine B staining procedure ("Experimental Procedures"). The relative plating efficiency of each cell population was calculated by dividing the absorbance measured at a given drug concentration by the value obtained for the same clone in the absence of drug, and was expressed as a percentage (%). Each point represents the average of at least two independent experiments performed in duplicate. gate function and precluded localization of this tag by immunofluorescence. Finally, when a different secondary antibody (Cy3 TM ) was used ( Fig. 4 and 5, left panels), slightly enhanced fluorescence was observed in all transfectants except for construct 6a where no fluorescent signal could be detected. A comparison of the hydropathy profiles of these transporters with that of other ABC transporters such as P-glycoprotein, CFTR, and STE-6 (39, 40, 50) suggested novel topological features for the MRP class. In addition to the structural unit of P-glycoprotein which consists of two membrane-associated regions (Fig.  1A, MAR2 and MAR3) and two NBD (Fig. 1A, NBD1 and NBD2) connected by a hydrophobic linker (L2), the MRP subgroup also includes a novel amino-terminal hydrophobic membrane-associated region (Fig. 1A, MAR1) attached to the rest of the protein by a hydrophilic linker (Fig. 1A, L1) (39, 40). This novel segment and associated linker are approximately 300 residues in length and account for the difference in size between this group of proteins and that of P-glycoprotein, Pgh-1, and STE-6 (1276 to 1300 residues). Although few direct topological measurements are available for MRP, the cytoplasmic location of the L2 region has been verified using an antibody directed against a peptide epitope comprising residues 918 -924, which is accessible only in permeabilized cells (44,45) and by the detection of a trypsin cleavage site in isolated membranes (41,45). The intracellular location of NBD2 has been verified by epitope mapping in permeabilized cells using antibodies directed against synthetic peptides corresponding to positions 1294 -1430 and/or 1497-1531 (43). Furthermore, the observations that an antibody directed against a peptide overlapping positions 192-360 recognizes an intracellular determinant (43) and that the residues 191-320 contain a chymotrypsin site that is not accessible in intact cells (41) suggests that at least part of the L1 region is intracellular. The paucity of more refined biochemical data has precluded verification of the various topological models which have been derived from alignment of multiple sequences and their corresponding hydropathy profiles. In particular, the membrane organization of the amino-terminal MAR1 region unique to the MRP group (Fig.  1A) remains unknown and has alternatively been proposed to cross the plasma membrane 4, 5 (Fig. 5, C and D; Ref. 41), or 6 times ( Fig. 5B; Ref. 42), resulting in both intra-and extracellular positioning of the amino terminus of the protein (Fig. 5,  B-D).

FIG. 3. Drug survival characteristics of control HeLa cells and drug-resistant MRP transfectants expressing wild type (WT) or
Although several techniques have been used to deduce topology of individual TM domains in integral membrane proteins (35)(36)(37)(38)(51)(52)(53), we have opted for an epitope insertion approach for the following reasons: the results are based on the analysis of full-length, biologically active protein rather than on truncated proteins, and the protein is analyzed for topology in whole mammalian cells which provide a normal physiological background where proper post-translational modification can take place. We have previously used this approach to determine the topology of individual TM domains of P-glycoprotein (36,37). Epitope tags from viral HA protein were in- FIG. 4. Detection of epitope-tagged MRP proteins by immunofluorescence. HeLa cells transfectants stably expressing either wild type MRP (WT) mutant MRPs 1, 3, 4, 8b, and 9 were exposed to the mouse monoclonal anti-HA epitope antibody 16B12 without pretreatment (intact cells, left and middle columns) or with pretreatment with 0.05% Nonidet P-40 (permeabilized cells, right column). Cells were then incubated with a secondary goat anti-mouse antibody conjugated to rhodamine (TRITC) (middle and right column) or a goat anti-mouse antibody conjugated to Cy3 TM (left column), and the cells were photographed using a fluorescence microscope. The HA epitopes in constructs 1 (D-F), 3 (G-I), and 8b (M-O) were detected in non-permeablilzed and permeabilized cells, while the HA epitopes in constructs 4 (J-L) and 9 (P-R) were only detectable in permeabilized cells. For photography, all exposure times were identical (A-F and I-R), except in the case of construct 3 (G and H) which was a shorter (two-thirds of the previous exposure time).

FIG. 5. Detection of epitope-tagged MRP proteins by immunofluorescence.
HeLa cells transfectants stably expressing either wild type MRP (WT) mutant MRPs 1, 6a, and 8a were exposed to the mouse monoclonal anti-HA epitope antibody 16B12 as described in the legend of Fig. 4, and the cells were photographed using a fluorescence microscope. The HA epitope in construct 6a could not be detected (G-I), while the staining of the HA epitope in construct 8a was weak (J-L). For photography, all exposure times were identical. serted in discrete regions of MAR1 and MAR2 that were predicted by hydropathy analyses to be hydrophilic and, therefore, most likely located in extracellular or intracellular loops linking individual TM domains. We also introduced HA tags at the amino terminus and in the predicted NBD1. The modified proteins were expressed in mammalian cells so that any deleterious effect of epitope insertion on MRP function could be monitored, and the accessibility of the epitope to the anti-HA tag antibody 16B12 was determined in intact and permeabilized cells to establish its intra-or extracellular location. Western blotting studies identified two MRP isoforms in all MRP transfectants, of approximately 160 and 180 kDa (Fig. 2, A and  B). The nature of these isoforms remain unknown, but may correspond to immature and mature isoforms of the protein, modified by different extent of N-linked glycosylation. A similar situation has been previously observed in transfected Chinese hamster ovary cells expressing HA-tagged P-glycoproteins (36,37). We did not observe any correlation between the relative abundance of either of these forms and the degree of drug resistance of the MRP transfectants (Fig. 3), in agreement with previous studies that have suggested that N-linked glycosylation does not drastically alter the activity of MRP (41,42). We have localized HA epitopes inserted at positions 4, 163, and 574 to the extracellular side of the membrane (Fig. 4). HA epitopes inserted at positions 271 and 653 were detected only in permeabilized cells and were therefore assigned an intracellular location (Fig. 4). While our experimental protocol allows unambiguous assignment of extracellular epitopes detected in intact cells, the assignment of intracellular HA epitopes is tentative since MRP expression in these cells was not restricted to the plasma membrane. For instance, cells expressing constructs 4 (position 271) and 9 (position 653) showed intracellular staining in addition to plasma membrane staining, suggesting that only a portion of the MRP variant expressed in these cells was targeted to the plasma membrane. Therefore, we cannot exclude the possibility that HA epitope insertion in constructs 4 and 9 may have altered protein maturation or targeting that is manifested by staining of the Golgi apparatus and/or the endoplasmic reticulum. However, the observation that constructs 4 and 9 could convey drug resistance suggests that enough of the mature, correctly folded MRP protein reaches the plasma membrane to convey the drug resistance phenotype. Taken with the lack of fluorescent signals in intact cells expressing constructs 4 and 9, these results clearly suggest an intracellular location for these tags. In addition, the intracellular localization of the HA tags at positions 271 (construct 4) and 653 (construct 9) is in agreement with previously published reports placing both the L1 region (position 192-310) and the NBD1 segment on the intracellular side of the membrane (41,43,45). Together, these results indicate that positions 271 and 653 are located on the intracellular face of the membrane.
The results of our epitope mapping experiments help discriminate between several of the current structural models for MRP proposed by hydropathy profiling. Our results do not support the early model proposed by Cole et al. (10) that was based on the analysis of a single MRP sequence (Fig. 6A) and which would position residues 4 and 163 intracellularly, and residue 271 extracellularly. Likewise, our data do not support  (42); panels C and D, 10 and 11 TM domain models based on trypsin and chymotrypsin digestion studies (41). Panels D and E, topological models that are consistent with the mapping of inserted HA epitopes reported in this study. the model of Loe et al. (42) (Fig. 6B) which predicts intracellular locations for positions 4 and 163. Importantly, we have unambiguously established that the amino terminus of MRP (position 4) is extracellular (Fig. 4). This observation, together with the intracellular positioning of the L1 and NBD1 regions by us (Fig. 4) and others (41,43,45), strongly support an odd number of TM domains for the amino-terminal half of MRP (MAR1 and MAR2 combined). This data is clearly incompatible with structural models predicting 8 ( Fig. 6A; Ref. 10), 10 ( Fig.  6C; Ref. 41), or 12 TM domains ( Fig. 6B; Ref. 42) for the amino-terminal half of MRP. However, our data do favor a topological model consisting of either 9 (Fig. 6E) or 11 TM domains ( Fig. 6D; Ref. 41) in this region. Indeed, after combining the epitope mapping data presented here with the hydropathy analysis of MRP (Fig. 1B) and other members of this group of ABC transporters (39,40), we strongly favor a model with 11 TM domains. In this model, the amino terminus is extracellular, and is followed, in order, by the MAR1 segment consisting of 5 TM segments, the intracellular hydrophilic L1 linker, the MAR2 segment consisting of 6 TM domains, and the intracellular NBD1 (Fig. 6D). The MAR1 region is unique to the MRP group of ABC transporters and multiple sequence and hydropathy alignments indicate that its secondary structure has been conserved in the cMOAT, yeast cadmium factor 1, and SUR proteins (39,40). Although hydropathy analyses of this structural unit in these transporters suggests 5 or 6 TM segments varying by the strength of the sixth hydrophobic peak (39,40), we favor a model of 5 TM domains, as it is most compatible with the extracellular location of the amino terminus and intracellular location of the L1 region (Fig. 6D). However, further analyses with additional epitope insertions must be carried out to unambiguously establish the number and position of individual TM domains in this segment. Our epitope mapping data are in agreement with the previously proposed 6 TM domains model for the MAR2 region of MRP (41,42) and would therefore resemble the established topology of the corresponding regions of P-glycoprotein and CFTR (35)(36)(37)(38). Although the MAR2 region (Fig. 1B) shows only 5 highly hydrophobic segments, it is likely that 6 TM domains are encoded by this segment as the hydrophobic peak delineated by positions 440 and 485 is broad enough to encode 2 TM domains of approximately 20 residues each (Fig. 1B). Furthermore, we note that insertion of an HA tag at position 461 is tolerated for MRP function, indirectly suggesting that this region may cross the membrane twice, since it is unlikely that the insertion of an HA tag within a TM segment would occur without any structural and functional consequences on MRP protein activity. Additional experiments will be required to establish the exact topology of this region.
Insertion of single HA epitopes at positions 127, 417, and 529 abrogated MRP protein activity. According to the MRP topological model shown in Fig. 5D, these residues would be located in intracellular loops. In contrast, insertion of HA tags at positions 4, 163, and 574 in extracellular loops, and the single epitope insertion at residue 461 which is probably also extracellular (Fig. 6, D and E), did not affect protein function. We have previously noted in similar studies of P-glycoprotein that epitope insertions in extracellular loops are much more likely to preserve function than those occurring in intracellular loops (37). Intracellular loops are usually more conserved between different isoforms of the protein than extracellular loops, and the integrity of the former may be more important for proper protein folding, processing, or function. This seems to be true for MRP as well. On the other hand, although single epitope insertions at positions 461 and 574 were compatible with function, they were either not (construct 6a) or only poorly accessible (construct 8a) to the anti-HA antibody (Fig. 5). This could be overcome by insertion of a second HA epitope at position 574 to enable localization by immunofluorescence, but insertion of a second tag at position 461 abrogated MRP function. These findings suggest that the extracellular loop containing position 574 may be fairly short, or that it may be tightly packed within the TM domains, being difficult to access for the antibody in both cases.