Identification of domains participating in the substrate specificity and subcellular localization of the multidrug resistance proteins MRP1 and MRP2.

The human multidrug resistance protein MRP1 and its homolog, MRP2, are both thought to be involved in cancer drug resistance and the transport of a wide variety of organic anions, including the cysteinyl leukotriene C4 (LTC4) (Km = 0.1 and 1 microm). To determine which domain of these proteins is associated with substrate specificity and subcellular localization, we constructed various chimeric MRP1/MRP2 molecules and expressed them in polarized mammalian LLC-PK1 cells. We examined the kinetic properties of each chimeric protein by measuring LTC4 and methotrexate transport in inside-out membrane vesicles, sensitivity to an anticancer agent, etoposide, and subcellular localization by indirect immunofluorescence methods. The following results were determined in these studies: (i) when the NH2-proximal 108 amino acids of MRP2, including transmembrane (TM) helices 1-3, were exchanged with the corresponding region of MRP1, Km(LTC4) values of the chimera decreased approximately 4-fold and Km(methotrexate) values increased approximately 5-fold relative to those of wild-type MRP2 and MRP1, respectively, whereas resistance to etoposide increased approximately 3-fold; (ii) when the NH2-proximal region up to TM9 of MRP2 was exchanged with the corresponding region of MRP1, a further increase in etoposide resistance was observed, and subcellular localization moved from the apical to the lateral membrane; (iii) when two-thirds of MRP2 at the NH2 terminus were exchanged with the corresponding MRP1 region, the chimeric protein transported LTC4 with an efficiency comparable with that achieved by the wild-type MRP1; and (iv) exchange of the COOH-terminal 51 amino acids between MRP1 and MRP2 did not affect the localization of either of the proteins. These results provide a strong framework for further studies aimed at determining the precise domains of MRP1 and MRP2 with affinity for LTC4 and anticancer agents.

framework for further studies aimed at determining the precise domains of MRP1 and MRP2 with affinity for LTC 4

and anticancer agents.
Two representative genes for the ATP binding cassette (ABC) 1 transporter superfamily proteins, P-gp/MDR1 and MRP1, mediate acquisition of a multidrug resistance phenotype through altered membrane transport of various anticancer agents in tumor cells (1,2). MRP1 confers resistance to a number of relatively hydrophobic natural product drugs including certain anthracyclines, epipodophyllotoxins, methotrexate, and vinca alkaloids (3)(4)(5)(6)(7). However, unlike P-gp, MRP1 can also transport a wide range of relatively hydrophilic anionic compounds including potential physiological substrates such as LTC 4 and E 2 17␤G (8 -15). Topology studies of MRP1 have demonstrated that MRP1 and P-gp share a similar core structure consisting of two membrane-spanning domains (MSD2 and MSD3) and two nucleotide-binding domains (NBD1 and NBD2), referred to as the MDR-like core (16). The primary distinguishing characteristic of MRP1 and its related proteins, MRP2, -3, -6, and -7, is an additional NH 2 -terminal region forming a membrane-spanning domain (MSD1) with five transmembrane (TM) helices; however, the function of MSD1 remains to be clarified (1, 16 -19). This region is linked to the MDR-like core by a cytoplasmic loop (CL) of ϳ130 amino acids, which is sometimes referred to as CL3 (17).
The MRP1 homolog, MRP2/cMOAT, shows only a 49% amino acid identity to human MRP1; however, its secondary structure is similar to that of MRP1 (19,20). Mutations in MRP2 have been identified in patients with Dubin-Johnson syndrome (19,(21)(22)(23), indicating an important role for MRP2 in the export of bilirubin into hepatic bile ducts. In addition, MRP2 is involved in the transmembrane transport of some anticancer agents and confers a drug resistance pattern that is similar but not identical to that of MRP1. Resistance to vincristine and etoposide is markedly enhanced in MRP1 cDNA-transfected cells and is diminished in mrp1 Ϫ/Ϫ cells (5,11,24). In contrast, our previous study showed MRP2 cDNA-transfected cells display resistance to vincristine but only low level resistance to etoposide (25,26). Furthermore, MRP2 transported the ␤-O-glucuronide * This work was supported by grants-in-aid for Cancer Research, Scientific Research on Priority Areas (C) "Medical Genome Science," and Scientific Research on Priority Areas (B) "Biological Transport Nano-Machine: Structure, Function, and Regulation" from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. This work was also supported by grants from the Second Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health, Labor, and Welfare and Core Research for Evolutional Science and Technology of the Japan Science and Technology Corporation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom all correspondence should be addressed: Dept. of Medical Biochemistry, Graduate School of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Fukuoka 812-8582, Japan. Tel.: 81-92-642-6100; Fax: 81-92-642-6203; E-mail: wada@biochem1.med.kyushu-u.ac.jp. conjugate of the tobacco-specific carcinogen, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol, with higher efficiency than MRP1, and GSH inhibited rather than stimulated uptake of this conjugate (27). Thus, MRP1 and MRP2 share several substrates in common, but the relative affinities of the two proteins for some of these substrates appear to vary markedly, and the substrate binding pockets of the two proteins must have some significant differences (28,29). Although recent studies have identified regions and specific amino acids of MRP1 that are important for substrate binding (30 -35), the structural basis for the differences between MRP1 and MRP2 as regards substrate affinities and specificities remains unknown.
The most well characterized substrate of MRP1 is the cysteinyl leukotriene, leukotriene C 4 (LTC 4 ). Knock-out mice lacking mrp1 show an impaired response to a leukotriene-mediated inflammatory stimulus (36). Other studies have shown that the mrp1-mediated efflux of LTC 4 is involved in regulating dendritic cell migration to lymph nodes (36,37). At present, LTC 4 is the substrate with the highest affinity to MRP1 and mrp1 (K m ϳ100 nM). MRP2 also shows relatively high affinity for this substrate; however, the K m (LTC 4 ) of MRP2 is ϳ10-fold higher than for MRP1 (38,39).
Many MRP1 structure-function studies have been based on LTC 4 transport activity. Gao et al. (30) suggested that the CL3 region between amino acids 204 and 281 is essential for LTC 4 transport activity. A recent study indicated that the deletion of all transmembrane helices of MSD1 (MRP1-(204 -1531)) had no effect on MRP1-mediated LTC 4 transport (40), suggesting that MSD1 may not participate in LTC 4 transport. However, an LTC 4 photolabeling study revealed that the 281 NH 2 -proximal amino acids of MRP1 were essential in the binding of LTC 4 to the NH 2 -proximal half of the protein, and CL3 contained a region that is critical for the correct folding of MRP1 (33). Cys 7 (TM1) mutation of MRP1 changed the NH 2 -terminal conformation and LTC 4 transport activity (35). In addition, TM6 (MSD2) may contribute to LTC 4 transport activity (34). Taken together, these data demonstrate that several regions of MRP1 play an important role in the recognition and/or transport of LTC 4 . In contrast to MRP1, little is known about regions of specific amino acids responsible for MRP2 substrate specificity and affinity.
Despite the close structural similarity between MRP1 and MRP2 proteins, their subcellular localization differs. MRP1 is present in the basolateral membrane, whereas MRP2 is localized to the apical membrane of polarized cells (41)(42)(43). Comparison of the primary structure of MRP2 with that of MRP1 revealed a 7-amino acid extension at its COOH terminus, the last 3 amino acids (TKF) of which comprise a PDZ-interacting motif. It has been reported that interaction of this motif with a PDZ domain-containing protein is important for the apical localization of membrane proteins (44,45). In particular, the PDZ-interacting motif is a determinant for the apical localization of CFTR (46). Furthermore, MRP2 was isolated by a yeast two-hybrid system screen using a PDZ domain-containing protein as the bait (47). These results suggest that the COOH terminus, particularly the PDZ-interacting motif, might be important for the apical sorting of MRP2. However, the experimental results that have been obtained by examining the role of this motif in apical targeting remain controversial. Harris et al. (48) showed that the deletion of the TKF motif resulted in the basolateral targeting of MRP2. Using an NH 2 -terminal fusion of MRP2 to green fluorescent protein in transiently transfected cells, Nies et al. (49), by contrast, showed that deletion of the COOH-terminal 11 amino acids, including the PDZ-interacting motif, did not abolish apical targeting but that deletion of 15 or more amino acids did have this effect in HepG2 cells. On the other hand, NH 2 -terminal MSD1 and/or CL3 are thought to be necessary for membrane routing, as shown by deletion experiments involving these regions in MRP1 and MRP2 (45,50). It appears likely that this region of the MRP2 protein is involved in the integration into the membrane of the endoplasmic reticulum, intracellular trafficking, and/or recycling by endocytosis. However, additional signals responsible for the differential targeting of the multidrug resistance proteins to particular membrane domains have not been determined.
To identify the regions involved in determining substrate specificity, substrate affinity, and the subcellular localization of MRP1 and MRP2, we created a series of MRP1/MRP2 chimeras and tested them for their ability to transport LTC 4 , confer resistance to etoposide, and localize to a distinct membrane region. Cell Lines and Culture-LLC-PK1 (polarized pig kidney epithelial cells) were purchased from the Health Science Research Resources Bank (Osaka, Japan). These cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 60 g/ml kanamycin, as described previously (26).

Materials
Western Blot Analysis and Determination of Protein Levels in Transfected Cells-Immunoblotting was performed as described previously (26). In brief, aliquots of membrane vesicles (40 g) were electrophoresed on 7.5% SDS-polyacrylamide gels and transferred onto Immobilon-P membranes. The MRP1 proteins were identified using the monoclonal antibody (mAb) MRPm6, which reacts with an epitope close to the COOH terminus of MRP1 (amino acids 1511-1520) (51). The MRP2 proteins were detected with the mAb M 2 III-6, which reacts with an epitope close to the COOH terminus of MRP2 (amino acids 1339 -1541) (52). Membranes were incubated with the mAbs for 2 h at room temperature and then with horseradish peroxidase-linked secondary antibody for 40 min at room temperature. Membranes were developed by chemiluminescence following the ECL protocol (Amersham Biosciences). Relative levels of protein expression were estimated by densitometric analysis using an Image Gauge V3.41 software (Fuji film, Tokyo, Japan). The relative protein expression levels were calculated by dividing the densitometry value obtained for 40 g of total membrane protein from transfectants expressing the chimeric proteins by the densitometry value obtained for comparable amounts of total membrane proteins from transfectants expressing wild-type MRP2 proteins. Each comparison was performed at least three times in independent experiments, and the mean values were used for normalization purposes.
The MRP1-(1-1480)/MRP2 vector was generated by PCR amplification of nucleotides 3881-4442 of MRP1 using a forward primer, 5Ј-GAG GCT CAA GGA GTA TTC AGA G-3Ј, and a reverse primer, 3Ј-CTG ACG TGG Cac tag tGG TAG-5Ј. The PCR product was digested with EcoRI and BclI, and the fragment was isolated. The MRP2 cDNA clone was digested with BclI and NotI, leaving nucleotides 4413-4638 of MRP2 attached to the vector pT7 Blue-3. The digested vector and attached insert were then ligated to the EcoRI-BclI PCR product. The pJ3⍀-MRP1 construct was digested with NotI and EcoRI, yielding a fragment composed of nucleotides 1-3880 of MRP1 with a portion of the vector polylinker at its 5Ј end. This 3.9-kb NotI-EcoRI MRP fragment was ligated to the EcoRI-digested construct. The insert was excised using NotI and transferred into a NotI-digested pCIneo expression vector to give the construct pCIneo-MRP1-(1-1480)/MRP2. The integrity of the hybrid constructs was confirmed by restriction analysis and by sequencing full-length constructs.
Transient and Stable Transfection with Various Chimeric MRP1/ MRP2 Expression Vectors-For transient transfections, wild-type and chimeric pCIneo-MRP1/MRP2 expression vectors were transfected into LLC-PK1 cells. Cells were seeded into two-well culture slides (Falcon) 48 -72 h prior to transfection, and DNA (2 g/well) was added using LipofectAMINE 2000 (Invitrogen), according to the manufacturer's instructions. After incubation for 48 h, cells were washed with fresh medium. After 72-96 h, the LLC-PK1 cells were tested. For stable transfections, exponentially growing LLC-PK1 cells in 24-well plates (1.5 ϫ 10 4 ) were washed with phosphate-buffered saline and placed in serum-free medium. Cells were incubated in the presence of 2 g of LipofectAMINE 2000 and 0.8 g of expression vector DNA for 24 h and washed with fresh medium. Cells were then incubated in selection medium containing 550 g/ml G418 for 3-4 weeks. Stable transfectants were selected from the G418-resistant transfectants. We also isolated a G418-resistant mock transfectant, LLC-PK1/Vec, produced by transfection of the vector alone.
Chemosensitivity Testing by Colony Formation Assay-Cell survival was determined by plating 3 ϫ 10 2 cells from LLC-PK1 cells stably transfected with chimera MRP1/MRP2 cDNAs in 35-mm dishes in the absence of drug treatment, which was added 24 h later. After incubation for 7 days at 37°C, the colonies were counted following the Giemsa staining procedure (25,26). All drugs were freshly prepared in physiologic saline or dimethyl sulfoxide. Equivalent volumes of saline or dimethyl sulfoxide were added in all control experiments. The 50% lethal dose (IC 50 ) for each cell line was determined from the doseresponse curves. IC 50 values and S.D. values were obtained from the best fit of the data to a sigmoidal curve using GraphPad Prism™ software. Relative resistance was obtained by dividing the IC 50 of cells transfected with vectors encoding either wild-type or MRP1/MRP2 hybrid proteins by the IC 50 of cells transfected with the pCIneo vector alone.
Immunofluorescence Study-The localization patterns of the immunoreactive MRP1/MRP2 chimeras in transiently transfected LLC-PK1 cells were determined by indirect fluorescent immunostaining. Cells were examined with an MRC-1024 confocal laser-scanning microscope (Bio-Rad) equipped with a ϫ 100 objective. Cells were fixed by incubation in paraformaldehyde for 20 min. Labeling was performed using MRP1 (MRP m6) and MRP2 (M 2 III-6) mAb, followed by fluorescein isothiocyanate-conjugated mouse IgG, as described previously (23). For each transient transfection, at least 30 transfected (as observed by fluorescein isothiocyanate fluorescence) cells were counted on a confocal laser-scanning microscope.
Membrane Vesicle Preparation-All steps were performed at 0 -4°C. Stably transfected LLC-PK1 cell monolayers were washed and scraped into phosphate-buffered saline and washed by centrifugation (4000 ϫ g for 10 min). The cell pellet was stored at Ϫ80°C until required. Membrane vesicles were prepared from defrosted LLC-PK1 cells according to a recently described method (23). Vesicles were frozen in liquid nitrogen and stored at Ϫ80°C until required for use. Protein concentrations were determined by the Lowry method, using bovine serum albumin as a standard (53). The orientation of membrane vesicles was determined by examining the nucleotide pyrophosphatase activity in the presence and absence of 1% Triton X-100 with p-nitrophenyl-thymidine 5-monophosphate as the substrate (54), and it was determined that 49% of LLC-PK1 vesicles were inside-out.
Membrane Vesicle Transport Studies-ATP-dependent transport of LTC 4 or MTX into the membrane vesicles was measured by filtration, essentially as described by Ishikawa and Ali-Osman (55). The standard incubation medium contained membrane vesicles (60 g of protein),

Construction of Chimeric MRP1/MRP2 Proteins and Their
Expression in Polarized Kidney Cells-To identify regions of MRP1 and MRP2 proteins responsible for their respective substrate specificities, we generated a series of chimeric MRP1/ MRP2 molecules (Fig. 1). Initially, the NH 2 -terminal 839 amino acids of MRP2 were replaced with the corresponding MRP1 sequence (MRP1-(1-846)/MRP2). In this case, because the cytoplasmic region linking the NBD1 to the MSD3 (linker region) is not required for the LTC 4 transport activity of MRP1 (30) . Because a single amino acid substitution has been shown to be capable of inducing instability of proteins through a protein quality control system (23,56), attempts were made, using the design of the chimera constructs, to prevent structural distortion leading to degradation. Specifically, we chose nonidentical amino acids as the sites for combining MRP1 and MRP2, because identical and conserved amino acid residues could be arranged to form a functional unit, and reorganization of them might disrupt proper protein configurations.
As expected, all chimera proteins expressed in the LLC-PK1 cell line exhibited the same electrophoretic mobility as wildtype MRP2. Fig. 2 shows an immunoblot of the chimeric MRP1/ MRP2 variants expressed in LLC-PK1 cells by an anti-MRP2 mAb. For these chimeric molecules, we obtained expression of proteins of similar size to those of the wild-type MRP2 (Fig. 2), suggesting that these variants might be processed to mature form with a molecular mass of 190 -200 kDa. On the other hand, the expression levels of chimeric proteins were low compared with those of wild-type MRP2, except in the case of MRP1-(1-239)/MRP2 and MRP1-(1-846)/MRP2. Reproducible results were obtained from the analysis of at least three independent transfected clones.
Cellular Localization of Chimeric MRP1/MRP2 Proteins-We next examined the membrane localization of various chimeric proteins in these polarized LLC-PK1 cells by indirect immunofluorescence with the respective specific antibodies (Fig. 3). We took advantage of the larger amount of proteins expressed and stronger signals in transiently transfected cells than were observed in the case of the stable transfectants. All of the transiently expressed chimeric proteins exhibited the same electrophoretic mobility as those of the wild type, as was the case with the stable transfection (data not shown). Antibodies against MRP1 (MRPm6) and MRP2 (M 2 III-6) were used for the immunofluorescence study, performed in order to identify each ABC transporter. Wild-type MRP1 and MRP2 were localized in the lateral membrane and the apical membrane of LLC-PK1 cells, respectively, as expected (Fig. 3, A and B). MRP1-(1-116)/MRP2 and MRP1-(1-239)/MRP2 proteins were predominantly localized in the apical membrane of LLC-PK1 cells, although some punctuated staining fluorescence of MRP2 in the cytoplasm was observed (Fig. 3, C and D). In contrast, MRP1-(1-480)/MRP2 was mainly present in the lateral membrane (Fig. 3E). MRP1-(1-555)/MRP2, MRP1-(1-846)/MRP2, and MRP1-(1-1480)/MRP2 were almost exclusively present in the lateral membrane, with some punctuated staining also present in intracellular compartments (Fig. 3, F-H). Taken together, these data indicate that the sorting information indispensable for the apical localization of MRP2 and/or the lateral localization of MRP1 lies between amino acids 239 and 480 of MRP1 and/or MRP2. Furthermore, the reciprocal chimera, MRP2/MRP1-(1480 -1531), was localized in the apical membrane (Fig. 3I), suggesting that the COOH terminus 51 and 65 amino acids of the two proteins may be necessary but were exchangeable between MRP1 and MRP2 to achieve the proper membrane localization. Although the staining signals were weaker than those of the transient transfectants, similar results were obtained using the stable transfectants (data not shown).
The COOH-proximal Half, Including MSD3 and NBD2, Is Exchangeable between MRP1 and MRP2 for High Affinity (MRP1-type) Binding of LTC 4 -In order to identify regions that might participate in the determination of substrate specificity and affinity, we examined the kinetic character of each chimeric protein for LTC 4 transport in isolated LLC-PK1 cell inside-out membrane vesicles. In the control experiments, membranes from LLC-PK1 cells transfected with empty coding expression vector showed that ATP-dependent tracer uptake was negligible. In all of the experiments, the uptake of [ 3 H]LTC 4 by membrane vesicles in the presence of ATP fell as the medium osmolarity increased (0.2-0.8 M sucrose). For the characterization of the function of the wild-type and chimeric proteins, in each case the linear phase of the tracer uptake (5 min) was used. ATP-dependent tracer uptake was calculated by subtracting the values measured in the presence of AMP.
We found that both MRP1 and MRP2 efficiently transported LTC 4 , but the K m value was ϳ10-fold higher for MRP2 than for MRP1, as described previously (39,44). The K m for the LTC 4 uptake of MRP1 was 175.3 nM, whereas that of MRP2 was 1660 nM. The transport data for LTC 4 are shown in detail in Fig. 4, and the K m and V max values for the wild-type and chimeric proteins are summarized in Table I Table I). The reason for this decrease in V max values among the chimeric proteins (except for MRP1-(1-480)/MRP2) remains unclear. Low expression or punctuate localization of some chimeras may have affected transport efficiency, but that phenotype was not directly associated with lower V max values.
The Role of MSD1 and the COOH-proximal Half, Including MSD3 and NBD2 of MRP1 and MRP2, in LTC 4 Transport Can Be Generalized to MTX Transport-We next examined the generality of the role of MSD1 and COOH-proximal half in the transport of a substrate other than LTC 4 , namely MTX.
In contrast to the LTC 4 K m values, the K m value for MTX was ϳ14-fold higher in the case of MRP1 (3.52 mM) than in the case of MRP2 (0.25 mM), indicating a significantly more efficient MTX transport by MRP2 (Fig. 5, Table II MRP2 showed very similar K m (MTX, 3.02 mM) values compared with those of the wild-type MRP1, indicating that the COOH-proximal half that includes MSD3 and NBD2 is basically exchangeable between MRP1 and MRP2 in terms of MRP1-type low affinity binding of MTX. On the other hand, the examination of MRP1-(1-116)/MRP2 demonstrated that the K m value for MTX transport increased ϳ5-fold from 0.25 to 1.27 mM (Fig. 4, Table II). This result indicates the involvement of MSD1 of MRP2, at least the TM1-3 region, in the high affinity recognition of MTX.
Exchange of CL3 and TM6 -9 of MRP2 by the Corresponding MRP1 Region Altered Sensitivity to Etoposide-In order to obtain insight into regions affecting the specificity of transport for substrates other than LTC 4 , we next examined whether the MRP1/MRP2 chimeric proteins showed altered drug sensitivity to an anticancer agent, etoposide, by a colony formation assay. The results are summarized as relative resistance factors in Table III. The LLC/MRP1 cell population showed an increase in resistance to etoposide relative to the control transfectants (17-fold resistance), whereas the LLC/MRP2 cell population displayed only a 2-fold resistance to etoposide (Table III). Consequently, we took advantage of the apparent differences in the  4 uptake were determined as described in the legend to Fig. 4. The normalized V max values were obtained by adjusting determined V max values to compensate for differences in the relative levels of the wild type and chimeric proteins. The relative levels of protein in the various transfectants were estimated to be LLC MRP2 Fig. 2). ND, not determined.   H]MTX uptake by membrane vesicles prepared from transfected LLC-PK1 cells was measured at various MTX concentrations (0.002-6 mM) for 5 min at 37°C, as described under "Experimental Procedures." The kinetic parameters for MTX transport were determined by Michaelis-Menten analysis of the combined data using GraphPad Prism™ software and are shown in Table  I resistance-conferring properties of these two proteins in order to search for a region(s) involved in mediating etoposide resistance.
Cells expressing all of the various chimeric proteins displayed increased resistance to etoposide when compared with cells transfected with vector alone (Table III) (Table III). Thus, we concluded that the NH 2 -terminal 116 amino acids and amino acids 239 -480 of the MRP1 protein were involved in conferring higher levels of resistance to etoposide compared with those achieved by MRP2. DISCUSSION In order to identify regions determining the substrate specificity of ABC transporters, several studies have been performed using deletion constructs, amino acid substitution, and photoaffinity labeling. However, the regions involved in the establishment of substrate specificity have remained unclear, as have the various kinetic parameters shared among closely related family members such as MRP1 and MRP2. In this paper, we sought to identify the regions responsible for the 10-fold difference in affinity for LTC 4 observed between MRP1 and MRP2; we also aimed to account for (i) differences in the ability of the two proteins to confer resistance to etoposide and (ii) differences in their distinct subcellular localizations.
We generated chimeric MRP1/MRP2 cDNA constructs and successfully expressed them in transiently transfected LLC-PK1 cells. For all of these molecules, we obtained the expression of proteins similar in size to that of wild-type MRP2 (Fig.  2), and the proteins were correctly routed to the plasma membrane in the stable transfectants (Fig. 3). However, there were some differences in the plasma membrane expression of different MRP1/MRP2 chimeras. Previous studies using the chimeric proteins murine and human MRP1 (57) and MDR1 and MDR2 (58) have revealed that chimeric proteins were expressed in relatively low levels compared with the levels of wild-type protein; our results agreed in this regard with the previous findings. One possible explanation for this low level of expression may be the slow protein-folding process or an unstable protein structure.
We then used transiently transfected cells in experiments conducted in order to determine subcellular localization. All of the chimeras examined were capable of being enzymatically active. Hence, the sorting behaviors observed in these studies were probably not due to the misfolding of the chimeric proteins but instead are more likely to have reflected the presence of an active apical localization signal in the MRP1 and/or MRP2 sequences. A recent study reported that a COOH-terminal deletion of at least 15 amino acids prevents efficient delivery of the MRP2 protein to the apical membrane, because part of a motif required for apical sorting is lost in polarized HepG2 cells (49). Our results clearly indicate that amino acids 239 -480 of the MRP1 and/or the corresponding region of MRP2, including a part of CL3, are responsible for the distinct distribution. A recent study revealed that co-expression of a P-gplike core with an isolated L0 region (amino acids 204 -281) yielded routing to the lateral membrane in MDCKII cells; however, the P-gp-like core, when expressed alone, resulted in a loss of polarized distribution (40). Similarly, co-expression in MDCKII cells of the NH 2 -terminal fragments MSD1-L0 and MSD1 of the MRP2 protein, with the core fragments ⌬-MRP2 and L0-⌬-MRP2, respectively, restored the altered cytoplasmic expression of the core alone to the apical plasma membrane (50). Photolabeling studies using [ 3 H]LTC 4 have suggested that CL3 might contain a region that is critical for the correct folding of MRP1 (33). Such results, when taken together, suggest that this cytoplasmic region is not only necessary for membrane routing, but that it also may play an additional role in creating a polarized distribution signal by cooperating with MSD2. Our results demonstrated that the 51 and 45 COOH terminus amino acids in MRP1 and MRP2, respectively, are exchangeable between the proteins, without inducing an alteration in their respective localizations. These results furthermore suggest that although the COOH terminus may be necessary for membrane anchoring, it does not appear to be responsible for distinctive localization processes. Certain differences in the amino acid residues of MRP1 and MRP2 in the region between 239 and 480 could be responsible for the apical targeting of MRP2 and/or the basolateral targeting of MRP1, depending on the signals by which the proteins would be sorted or stably anchored into distinct post-Golgi carriers in the trans-Golgi network, leading to distinct membranes (59). Future studies will determine the specific residues of the region 239 -480 of the MRP1 protein and/or the corresponding parts of MRP2 that are responsible for proper localization.
We observed that the K m values for LTC 4 transport decreased sequentially as the putative active region of MRP2 was exchanged by the corresponding region of MRP1; this replacement proceeded sequentially from the NH 2 -terminal to the COOH-terminal direction, suggesting that multiple regions located throughout MRP1 and MRP2 might be involved in the determination of specific affinity for LTC 4 . This result is consistent with a previous report (40) indicating that the addition of MSD1 and CL3 of MRP1 to P-gp is not sufficient to achieve transport of the glutathion conjugate; it is also thought that the rest of the MRP1 molecule is also required for substrate recognition and transport. Koike et al. (60) reported that amino acid substitution of conserved tryptophan residues at amino acid 361 (TM7), 445 (TM8), 459 (TM9), and 553 (TM10) eliminated or selectively reduced transport activity of organic substrates including LTC 4 , also consistent with our results.
Particularly surprising was our observation that the exchange of the NH 2 -proximal 116 amino acids of MRP2 by the corresponding MRP1 region (MRP1-(1-116)/MRP2) dramatically increased affinity for LTC 4 , suggesting the involvement of MSD1, at least the TM1-3 sequence, in determining the high affinity of MRP1 for LTC 4 . In addition, MRP1-(1-116)/MRP2 reduced affinity for MTX, which was recognized with 14-fold higher affinity by MRP2 than by MRP1, such that the MSD1 (including TM1-3) region of MRP2 appears to contain the region necessary for high affinity recognition of MTX. This result provides further evidence for the putative role played by TM1-3 in the modulation of an affinity to the substrate. Gao et al. (30) reported that the deletion of amino acids 1-66 of MRP1 markedly decreased LTC 4 transport activity and that substitution of the first MSD of MRP1 (the region 1-228) with the comparable region of the MRP2 also markedly reduced LTC 4 transport activity; that result was consistent with the present findings. Also, previous study has demonstrated that Cys 7 in TM1 is critical for the MRP1 function through maintaining proper structure (35). Similarly, it was reported that substitution of conserved lysine and asparate residues at amino acids 332 and 336, respectively, in TM6 diminished the ability to transport LTC 4 , and GSH transport was reduced (34). Because the cysteine, lysine, and asparagine residues are conserved between MRP1 and MRP2, it is not likely that those cysteine, lysine, and asparagine would be responsible for the K m difference between the two proteins, although mutational analysis of the residues did alter the transport activity. Overall, these data suggest that MSD1 and the adjacent region of MRP1 contains several important residues forming a high affinity LTC 4 transport site. Alternatively, Bakos et al. (61) reported that a truncated MRP1 mutant lacking the entire MSD1 region but still containing CL3 (amino acids 204 -1531) behaved like wild-type MRP1 in terms of LTC 4 uptake. The reason for this discrepancy is unclear at present, but the following explanations should be considered: (i) MSD1 may carry out a regulatory function to achieve the proper affinity of MRP1 and MRP2 for the substrate, LTC 4 , in mammalian cells but not in insect cells; (ii) the chimera protein assumes an unusual configuration to enhance transport activity; or (iii) one function of MSD1 may be to participate in some manner in a dimer configuration. As regards the last possibility, it has been reported that MRP1 appears to be dimeric upon reconstitution into crystalline protein/lipid arrays (62). MSD1 may be functionally dispensable in a dimer configuration, provided the dimer complex were stable without MSD1, or the alteration of a portion of a monomer may functionally compensate by a particular region of an accompanying MRP1 counterpart in a homodimeric complex. Along these lines, it is of interest that two polypeptides, CL3 and the remaining distal region of MRP1, when expressed separately in insect cells, can assemble into a functional LTC 4 transporter (30,40). We also found in the present study that the COOH-proximal half including TM 12-17 was exchangeable between MRP1 and MRP2, leading to functional expression and high affinity for LTC 4 . Recent photolabeling studies using N-(hydrocinchnidin-8Ј-yl)-4-azido-2-hydroxybenzamide (63), iodoaryl azidorhodamine 123 (64), LTC 4 (33), agosterol A (65), and LY475776 (66) have suggested that the COOH-proximal half, especially TM16 and TM17, is the substrate binding site but that the NH 2proximal half is also required. Although the COOH-proximal half could be the site for substrate binding itself, the region is unlikely to be fully responsible for the major differences in LTC 4 affinity observed between MRP1 and MRP2, because our results clearly demonstrated that the COOH-proximal half was exchangeable between MRP1 and MRP2. In other words, our results revealing the exchangeability of the COOH-proximal half of MRP1 and MRP2 suggest that identical or conserved amino acids in this COOH-proximal half of the proteins could be necessary for binding and/or subsequent transport processes, whereas nonidentical or nonconserved amino acids are less likely to participate in generating the high affinity site for LTC 4 . It has been reported that basic residues in TM6, -9, -16, and -17 of MRP2 that are identical with MRP1 are important for the transport of glutathione-methylfluorescein (67), and a tryptophan residue in TM17 is important for the transport of E 2 17␤G by MRP1 and for the transport of MTX, LTC 4 , and E 2 17␤G by MRP2 (31,68); these results all support our finding. The exchangeability of the COOH-proximal half of MRP1 and MRP2 in the case of both LTC 4 and MTX transport is consistent with this hypothesis.
We demonstrated that the residues at positions 1-116 (TM1-3) and 239 -480 (MSD2) of MRP1 were important for the protein's ability to confer high levels of resistance to etoposide (Table III). Combined with the LTC 4 transport data showing that amino acids 1-116 and 480 -846 may play an important role in forming the high affinity LTC 4 transport site, these observations regarding etoposide suggest that the structural determinants in MRP1 necessary for recognition and transport differ for each substrate and exist in multiple sites of the protein. In general, the present results appear to be consistent with those of previous studies.
In conclusion, we have identified the substrate specificity domains and regions containing the localization signal of MRP1 and MRP2. In MRP1, several sites mediating the recognition and transport of the anionic physiological substrate LTC 4 were found to overlap with, but were not identical to, those for chemotherapeutic agents. It may in the future be possible to design agents capable of inhibiting the ability of MRP1 to confer resistance to some chemotherapeutic agents without interfering with its ability to transport anionic physiological substrates. Future studies will be necessary to investigate which regions of MRPs are critical for recognizing and transporting a variety of anionic physiological substrates and chemotherapeutic agents.