Localization of a Substrate Specificity Domain in the Multidrug Resistance Protein*

Multidrug resistance protein (MRP) confers resistance to a number of natural product chemotherapeutic agents. It is also a high affinity transporter of some physiological conjugated organic anions such as cysteinyl leukotriene C4 and the cholestatic estrogen, 17β-estradiol 17(β-d-glucuronide) (E217βG). We have shown that the murine orthologue of MRP (mrp), unlike the human protein, does not confer resistance to common anthracyclines and is a relatively poor transporter of E217βG. We have taken advantage of these functional differences to identify region(s) of MRP involved in mediating anthracycline resistance and E217βG transport by generating mrp/MRP hybrid proteins. All hybrid proteins conferred resistance to the Vinca alkaloid, vincristine, when transfected into human embryonic kidney cells. However, only those in which the COOH-terminal third of mrp had been replaced with the corresponding region of MRP-conferred resistance to the anthracyclines, doxorubicin, and epirubicin. Exchange of smaller segments of the COOH-terminal third of the mouse protein by replacement of either amino acids 959–1187 or 1188–1531 with those of MRP produced proteins capable of conferring some level of resistance to the anthracyclines tested. All hybrid proteins transported cysteinyl leukotriene C4 with similar efficiencies. In contrast, only those containing the COOH-terminal third of MRP transported E217βG with an efficiency comparable with that of the intact human protein. The results demonstrate that differences in primary structure of the highly conserved COOH-terminal third of mrp and MRP are important determinants of the inability of the murine protein to confer anthracycline resistance and its relatively poor ability to transport E217βG.

Multidrug resistance protein (MRP) 1 and P-glycoprotein (Pgp) are very distantly related members of the superfamily of ATP binding cassette transmembrane transporters (1)(2)(3). Primary structure similarity between the two proteins is confined mainly to their nucleotide binding domains, regions that are generally conserved among ATP binding cassette superfamily members, and phylogenetic analyses suggest that MRP and Pgp evolved from different ancestral proteins (4,5). Despite the lack of structural similarity, both proteins confer resistance to a similar but not identical spectrum of natural product chemotherapeutic agents, which includes the Vinca alkaloids, the anthracyclines, and the epipodophyllotoxins (6 -9). However, several lines of evidence suggest that MRP and Pgp confer resistance to these drugs by different mechanisms.
Using plasma membrane vesicles enriched in Pgp, it has been possible to demonstrate direct transport of a number of chemotherapeutic agents and to label the protein with photoaffinity analogs of some drugs to which it confers resistance (10 -12). More recently, purified Pgp reconstituted into a lipid environment has been shown to transport a number of chemotherapeutic agents when provided with a suitable energy source (13,14). In contrast, it has not been possible to demonstrate direct active transport of unmodified drugs by MRPenriched membrane vesicles under similar conditions (15)(16)(17)(18)(19), and reports to the contrary have been retracted (20,21). However, we have shown that MRP can actively transport the Vinca alkaloid, vincristine, and the potent mutagen, aflatoxin B 1 , in such a membrane vesicle system but only in the presence of physiological concentrations of glutathione (15,18,22). It has also been possible to demonstrate that MRP-dependent transport of unmodified vincristine is accompanied by co-transport of reduced glutathione (18).
We previously reported the cloning and in vitro pharmacological characterization of the highly conserved murine orthologue of MRP, mrp (30,31). These studies revealed that the amino acid sequences of MRP and mrp are 88% identical. Both proteins confer resistance to Vinca alkaloids and the epipodophyllotoxin VP-16, and both transport LTC 4 with similar ki-* This work was supported by a grant from the National Cancer Institute of Canada with funds from the Terry Fox Foundation. 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 1 The abbreviations used are: MRP, multidrug resistance protein; Pgp, P-glycoprotein; MSD, membrane-spanning domain; NBD, nucleotide binding domain; E 2 17␤G, 17␤-estradiol 17-(␤-D-glucuronide); LTC 4 , leukotriene C 4 ; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; BCECF, 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein; PCR, polymerase chain reaction; kb, kilobase; HEK, human embryonic kidney cells. netic parameters. However, despite the high degree of primary structure identity, mrp did not confer resistance to any of several anthracyclines tested (30,32,33). In addition, the ability of the murine protein to transport E 2 17␤G was relatively poor when compared with MRP (31).
In the present study, we have taken advantage of functional differences between the two proteins to search for a region(s) of MRP involved in mediating anthracycline resistance and/or transport of substrates such as E 2 17␤G. We have stably expressed several mrp/MRP hybrid proteins in human embryonic kidney (HEK 293) cells and shown that they all confer resistance to vincristine and transport LTC 4 with similar efficiency. However, only those proteins containing the COOH-terminal third of MRP conferred resistance to two anthracyclines tested, and only these proteins transported E 2 17␤G with an efficiency comparable with that of wild-type MRP. By exchanging segments within the COOH-terminal third of the protein, we identified amino acids 959 -1187 of MRP as a region particularly critical for mediating anthracycline resistance.   vector was generated by PCR amplification of nucleotides 2575-3630 of mrp using a 5Ј hybrid primer complementary to nucleotides 2565-2574 of MRP, which included an XhoI site followed by nucleotides 2575-2592 of mrp and a reverse primer complementary to nucleotides 3610 -3630 of mrp (nucleotides numbered relative to beginning of the coding region; the EMBL/GenBank accession numbers are L05628, AF017145, and AF022824 -AF022853 for MRP and AF022908 for mrp). The PCR product was digested with XhoI and SacI, and the fragment containing nucleotides 2570 -3554 was isolated. cDNA clone 16Spe, which contains nucleotides 1612-4910 of mrp, was digested with XhoI and SacI, leaving nucleotides 3554 -4910 of mrp attached to the vector pBluescript II (30). The digested vector and attached insert was then ligated to the XhoI-SacI PCR product. The resulting construct was digested with KpnI in the polylinker region of the vector 5Ј to the insert and with XhoI at the site introduced by PCR. The pCEBV7-MRP1 construct was digested with KpnI and XhoI to yield a fragment comprised of nucleotides 1-2560 of MRP with some of the vector polylinker at its 5Ј end (7). This 2.6-kb KpnI-XhoI MRP fragment was ligated to the KpnI-XhoI-digested construct. The insert was excised using KpnI and NotI and transferred into KpnI/NotI-digested pCEBV7 expression vector to give construct pCEBV7-mrp/MRP  .

Materials-Doxorubicin
The cDNA specifying mrp/MRP 959 -1531 was generated by PCR amplification using a 5Ј primer corresponding to nucleotides 1861-1880 of mrp and a 3Ј hybrid primer containing nucleotides 2875-2885 of MRP, which included a HindIII site followed by nucleotides 2847-2862 of mrp. The product was cloned into the EcoRV site of pBluescript II KSϩ. Digestion of this construct with XhoI yielded a 4-kb BsmI-XhoI fragment comprised of nucleotides 1947 to 2885 of mrp attached to the vector. This fragment was ligated to a 1.9-kb XhoI-BsmI fragment containing nucleotides 1-1946 of mrp from pCEBV7-mrp (30). The resulting insert was excised using HindIII, which cut in the polylinker region 5Ј to the insert and at the 3Ј end of the insert at the HindIII site introduced by PCR. This fragment was then ligated to an 11.5-kb HindIII fragment containing nucleotides 2875-4823 of MRP attached to the pCEBV7 expression vector to generate construct pCEBV7-mrp/ MRP 959 -1531 (7).
The vector encoding mrp/MRP 959 -1187 was generated by ligating a HindIII-EcoRI fragment encompassing nucleotides 2875-3880 of MRP into HindIII-EcoRI digested pBluescript II KSϩ. This construct was digested with StuI at nucleotide 3551 of the insert and with SpeI at a site in the polylinker region 3Ј to the insert to generate a 3.7-kb StuI-SpeI fragment containing the vector attached to nucleotides 2875-3554 of MRP. The 3.7-kb StuI-SpeI fragment was isolated and ligated to a StuI-SpeI fragment containing nucleotides 3554 -4910 of mrp. This construct was linearized by HindIII digestion, treated with calf intestinal phosphatase, and ligated to a HindIII fragment containing nucleotides 1-2875 of mrp isolated from pCEBV7-mrp/MRP 959 -1531 . The resulting insert was excised using EcoRV and NotI and ligated into pCEBV7 digested with PvuII and NotI to give construct pCEBV7-mrp/MRP 959 -1187 .
The vector encoding mrp/MRP 1188 -1531 was constructed by digesting mrp cDNA clone 41 (containing nucleotides 2809 -5881 of mrp) with StuI at nucleotide 3575 of the insert and with BamHI in the polylinker region 3Ј to the insert to generate a fragment containing nucleotides 2809 -3575 of mrp attached to pBluescript II SKϩ (30). The 3.8-kb StuI-BamHI fragment was then ligated to a StuI-BamH fragment containing nucleotides 3562-4823 of MRP isolated from pCEBV7-MRP1 (31). The resulting construct was then digested with DraIII. The fragment encompassing nucleotides 3218 -4823 of the insert attached to nucleotides 668 -230 of pBluescript II SKϩ was isolated and ligated to a DraIII fragment containing nucleotides 231-667 of pBluescript SKϩ attached to nucleotides 1-3218 of mrp obtained by digestion of fulllength mrp in pBluescript II SKϩ (30). The hybrid insert was excised by digestion with HindIII and XhoI and then re-ligated into HindIII-XhoIdigested pCEBV7 to generate the pCEBV7-mrp/MRP 1188 -1531 construct. Integrity of the hybrid constructs was confirmed by restriction analysis and by sequencing across cloning junctions and those portions of the constructs contributed by PCR products.
Cell Lines and Tissue Culture-Stable transfection of HEK 293 cells with the pCEBV7-MRP1 and pCEBV7-mrp constructs has been described previously (31). pCEBV7 vectors containing mrp/MRP hybrid cDNAs were used to stably transfect HEK 293 cells in an identical fashion (31). Subpopulations of cells expressing high levels of wild-type mrp or mrp/MRP hybrid proteins were obtained by limiting cell dilution.
Determination of Protein Levels in Transfected Cells-The levels of wild-type mrp or MRP as well as the hybrid proteins were determined by immunoblot and/or dot blot analysis of membrane protein fractions from transfected cells, as described previously (34 -36).Wild-type or hybrid proteins were detected with the monoclonal antibody, MRPr1, which recognizes a linear epitope of 10 amino acids (238 -247), 9 of which are identical in mrp (36). Antibody binding was detected with goat anti-rat IgG (Pierce) followed by enhanced chemiluminescence detection (NEN Life Science Products).
Chemosensitivity Testing-Drug resistance was determined using the microtiter plate MTT assay (30,31,37). Cells were seeded in 96-well plates (1 ϫ 10 4 cells/well), incubated at 37°C for 24 h before the addition of drug, and then incubated for a further 72 h before the addition of MTT (2 mg/ml). IC 50 values and standard deviations were obtained from the best fit of the data to a sigmoidal curve using Graph-Pad software. The significance of the difference between IC 50 values of control and mrp/MRP transfectants was determined using an unpaired Student's t test. Relative resistance was obtained by dividing the IC 50 of cells transfected with vectors encoding either wild-type or mrp/MRP hybrid proteins by the IC 50 of cells transfected with the pCEBV7 vector (HEK PC7 ) alone.
LTC 4 and E 2 17␤G Transport by Membrane Vesicles-The kinetic parameters of [ 3 H]LTC 4 transport by inside-out membrane vesicles were determined as described previously (15,31). Vesicles (2.5 g of membrane protein) were incubated at 23°C in transport buffer (50 mM Tris-HCl, 250 mM sucrose, 0.02% sodium azide, pH 7.4) containing AMP or ATP (4 mM), MgCl 2 (10 mM), and [ 3 H]LTC 4 (15-1000 nM) in a final volume of 25 l. Uptake was terminated after 30 s by rapid dilution of 20-l aliquots into 1 ml of ice-cold transport buffer and filtration under vacuum through glass fiber filters. Filters were washed and dried before determination of the filter bound radioactivity. All data were corrected for the amount of [ 3 H]LTC 4 , which remained bound to the filter in the absence of vesicle protein (usually less then 5% of the total radioactivity). Data were plotted as V o versus [S] to confirm that the concentration range selected was appropriate to observe both zero-order and first-order kinetics. Kinetic parameters (K m and V max ) for the transport of [ 3 H]LTC 4 were determined from regression analysis of the Lineweaver-Burk transformation of the net uptake data (ATP-dependent minus AMP-dependent uptake).
ATP-dependent uptake of [ 3 H]E 2 17␤G was measured in membrane vesicles prepared from the transfectants as described for LTC 4 with the following modifications. Reactions were carried out at 37°C in a volume of 90 l at a single concentration of [ 3 H]E 2 17␤G (400 nM; 120 nCi) and 20 g of membrane protein. Uptake was terminated at various times by removing aliquots (20 l) and samples processed as described above.

Expression of Mouse and Human MRP in HEK 293 Cells-
Previously, we demonstrated that stable transfection of HEK 293 cells with either mrp or MRP expression vectors conferred similar drug resistance profiles, with the notable exception that only the human protein increased resistance to several anthracyclines tested (31). However, the levels of vector-encoded protein were severalfold lower in the mrp transfectant populations used for the original study than in the MRP transfectants used for comparison. To eliminate the possibility that the lower levels of mrp were responsible for the inability to detect anthracycline resistance, we isolated higher-expressing subpopulations from the original HEK mrp transfectants by limiting cell dilution. The level of mrp in the HEK mrp1 subpopulation is approximately equivalent to that of MRP in the HEK-MRP transfectants (Fig. 1A). Both populations of cells showed a similar increase in resistance to vincristine relative to control transfectants (27-and 23-fold resistance for the HEK mrp1 and HEK MRP transfectants, respectively). The HEK MRP population also displayed 8-and 11-fold resistance to the anthracyclines, doxorubicin and epirubicin, respectively (Table I). In contrast, despite the higher levels of mrp in the HEK mrp1 transfectants, no significant increase in resistance to either anthracycline could be detected as reported previously (Table I) (31).
Generation of Hybrid mrp/MRP Molecules-To identify regions of the mouse and human proteins responsible for the differences in anthracycline resistance, we generated a series of hybrid mrp/MRP molecules (Fig. 2). Initially, we replaced either the NH 2 -terminal 857 amino acids (mrp/MRP 1-857 ) or COOH-terminal 574 amino acids of mrp (mrp/MRP 959 -1531 ) with the corresponding human sequence. In both cases, locations used to connect between the segments of the hybrid proteins were in the poorly conserved cytoplasmic region linking the NH 2 -proximal NBD to the COOH-proximal membranespanning domain (Fig. 2A). The locations were chosen because we have shown previously that they are in a part of the linker region that is not required for the LTC 4 transport activity of MRP (38). Populations of transfectants expressing mrp/MRP hybrids were subjected to limiting cell dilution to isolate subpopulations with relatively high levels of protein (Fig. 1, A and  B). Based on dot blot analyses of membrane proteins, the approximate relative levels of wild-type and hybrid proteins in the various populations of transfectants were HEK MRP (1.0), HEK mrp1 (1.0), HEK mrp/MRP (1-857) (1.0), HEK mrp/MRP(959 -1531A) (0.5), and HEK mrp/MRP(959 -1531B) (2.0) (Fig. 1B).
Resistance Profile of mrp/MRP  and mrp/MRP 959 -1531 Proteins-Cells expressing either mrp/MRP 1-857 or mrp/ MRP 959 -1531 displayed increased resistance to vincristine when compared with cells transfected with vector alone (Table  I). When normalized for differences in expression levels, the hybrid and wild-type human and murine proteins were similarly effective at conferring resistance to this drug (Table I). Cells transfected with mrp/MRP  showed no significant increase in resistance to either doxorubicin or epirubicin (1.2-1.3-fold), as observed with cells transfected with the wild-type murine protein (Table I). In contrast, membranes from both populations of cells transfected with mrp/MRP 959 -1531 showed a significant increase in resistance to the anthracyclines ranging from 2.2-to 4.2-fold (Table I). Thus we conclude that the murine and human proteins differ at locations in the COOHterminal 572 amino acids that are important for mediating anthracycline resistance.  (Table I). As observed with the other hybrid proteins, the levels of resistance correlated well with the expression of both proteins. Resistance to doxorubicin and epirubicin in cells expressing either protein was similar (1.7-2.3fold) despite the lower level of expression of mrp/MRP 959 -1187 . When normalized to the expression levels of wild-type mrp in the HEK mrp1 transfectants, the adjusted resistance factors for HEK mrp/MRP(959 -1187) and HEK mrp/MRP(1188 -1531) transfectants were 5-and 3.6-fold for doxorubicin and 6.2-and 2.4-fold for epirubicin, respectively.

Exchange of Regions within the COOH-terminal 572 Amino
Transport of LTC 4 and E 2 17␤G by Hybrid Proteins-We also examined the ability of the hybrid proteins to transport two well characterized potential physiological substrates, LTC4 and E 2 17␤G. We have previously shown that membrane vesicles prepared from HEK cells transfected with either mrp or MRP transport LTC 4 in an ATP-dependent manner with similar kinetic parameters (31). The ability of hybrid proteins to transport LTC 4 was evaluated by carrying out comparable experiments with vesicles prepared from appropriately transfected cells. Vesicles from HEK mrp1 and HEK MRP transfectants were also included in these analyses (Fig. 3). K m and V max values for the wild-type and hybrid murine proteins are summarized in Table II,  wild-type murine and human proteins (31). V max values for vesicles containing the hybrid proteins ranged from 65 pmol mg Ϫ1 min Ϫ1 to 215 pmol mg Ϫ1 min Ϫ1 for vesicles from HEK mrp/ MRP(959 -1187) or HEK mrp/MRP(1-857) transfectants, respectively. However, when normalized for differences in expression levels, the V max values for the hybrid and wild-type murine proteins

cells transfected with wild-type and hybrid murine and human MRPs
The resistance of HEK cells transfected with expression vectors encoding wild-type and hybrid murine and human proteins relative to that of cells transfected with empty vector were determined using the tetrazolium salt-based microtiter plate assay. In each experiment, cell viability was determined in quadruplicate at each drug concentration. Data were then analyzed as described under "Experimental Procedures." The relative resistance was obtained by dividing the IC 50 (47). In this model, the protein has 17 predicted transmembrane domains (TM), which are organized into 3 membrane-spanning domains (MSD1, MSD2, and MSD3). The position of the two nucleotide binding domains (NBD1 and NBD2) and the poorly conserved linker region, which connects the two halves of the protein is also indicated. B, shown is an illustration of the hybrid mrp/MRP molecules generated. The position(s) of mrp where the splice with the corresponding sequence of MRP occurs is indicated. The amino acid numbering refers to the segment of human MRP present in each construct.
In contrast to their similar abilities to transport LTC 4 , we found that mrp was a less efficient transporter of E 2 17␤G than the human protein (31). Because of differences in the levels of human and mouse proteins in the vesicle preparations used originally, we confirmed these observations with vesicles from the HEK mrp1 and HEK MRP transfectants, which express comparable levels of protein (Fig. 4A). The rate of uptake of E 2 17␤G determined during the first 2 min was at least 10-fold higher with vesicles containing MRP than a comparable vesicle preparation containing the murine protein. We also compared the ability of the hybrid proteins, mrp/MRP 1-857 and mrp/ MRP 959 -1531 to transport this substrate using vesicles prepared from the HEK mrp/MRP(1-857) and HEK mrp/MRP(959 -1531A) transfectants. Despite the approximate 2-fold lower level of expression in the HEK mrp/MRP(959 -1531A) transfectants, the rate of E 2 17␤G uptake by vesicles containing this hybrid protein was more than 10-fold higher than that obtained with the mrp/MRP 1-857 vesicle preparation (Fig. 4B). Thus, as in the case of anthracycline resistance, the results clearly implicate sequence variation in the COOH-proximal third of the mouse protein as the major cause of the relatively inefficient transport of E 2 17␤G. In an attempt to further localize the region(s) responsible, we also compared the uptake of E 2 17␤G by vesicles containing hybrids mrp/MRP 959 -1187 and mrp/MRP 1188 -1531 . Uptake by vesicles containing mrp/MRP 1188 -1531 (Fig. 4C) was readily detectable and occurred at an initial rate that was approximately 50% that obtained with vesicles containing mrp/ MRP 959 -1531 (Fig. 4B). The levels of these two hybrid proteins were comparable in the vesicle preparations used. The rate of uptake obtained with vesicles containing mrp/MRP 959 -1187 was approximately 3-fold lower than the rate obtained with mrp/ MRP 1188 -1531 (Fig. 4C). However, the level of expression of this hybrid protein was also approximately 2-fold lower than that of mrp/MRP 1188 -1531 . DISCUSSION Like Pgp, MRP confers resistance to a number of relatively hydrophobic natural product drugs including certain anthracyclines, epipodophyllotoxins, and Vinca alkaloids (6 -9). However, unlike Pgp, MRP can also transport a wide range of    relatively hydrophilic anionic compounds including potential physiological substrates such as LTC 4 and E 2 17␤G (15,17,19,(22)(23)(24)(25)(26)(27)39). The region(s) of MRP involved in substrate recognition and/or transport has not been identified. A number of substrates such as vincristine plus GSH, LTC 4 , and E 2 17␤G are able to compete with one another for ATP-dependent transport by MRP-enriched vesicles (15,18,24). This suggests that these structurally diverse compounds interact with common or mutually exclusive sites on the protein during either initial binding, or some subsequent step in the transport process.
We found that all mrp/MRP hybrids tested, when normalized for protein expression levels, conferred resistance to vincristine as effectively as the wild-type murine and human proteins. Thus, residues critical for vincristine binding and transport or for the formation of functionally important long range interactions between the two halves of the protein appear to be conserved in both mrp and MRP. In contrast, only replacement of the COOH-terminal third of the murine protein with that of MRP generated a molecule, mrp/MRP 959 -1531 , capable of conferring significant resistance to both epirubicin and doxorubicin (Table I). Hybrid mrp/MRP  , which contains the first and second MSDs and the NH 2 -proximal NBD of MRP, was able to confer low levels of resistance (1.4-fold) to epirubicin but not doxorubicin (Table I). Thus, although conserved regions in the NH 2 -terminal half of the protein may be involved in mediating resistance to this class of drugs, critical locations that have presumably diverged between the two proteins are present in the COOH-terminal third. This result was somewhat unexpected since this segment of the molecule is more highly conserved than the NH 2 -terminal two-thirds, and conservation of the sequence between amino acid 1185 and the COOH terminus is exceptionally high (94% identity). Despite the high level of amino acid identity in the COOH-terminal portion, exchange of this region in mrp/MRP 1188 -1531 generated a protein capable of conferring low but significant levels of resistance to both anthracyclines. Exchange of the more variable section of the COOH-terminal third of the protein in hybrid mrp/MRP 959 -1187 generated a molecule that was in relative terms 2-3 times as effective as mrp/MRP 1188 -1531 at conferring resistance to the two examples of this class of drugs that were tested. We noted previously that although both mrp and MRP transport LTC 4 with similar kinetic parameters, the human protein transports E 2 17␤G with a considerably greater V max (31). Consequently, we investigated whether the regions important for transport of E 2 17␤G co-localized with those involved in conferring anthracycline resistance. Using membrane vesicles containing similar amounts of mrp or MRP, we confirmed that the murine protein transports the estrogen glucuronide relatively poorly with an initial rate of uptake that was less than 10% that of MRP (Fig. 4A). As observed when examining the ability to confer anthracycline resistance, only exchange of the COOHterminal third of mrp for that of MRP in mrp/MRP 959 -1531 generated a protein capable of transporting E 2 17␤G at rates comparable with intact MRP when normalized for protein expression levels (Fig. 4, A and B). In contrast, hybrid mrp/MRP  was no more effective than the wild-type murine protein at transporting E 2 17␤G (Fig. 4, A and B). Thus we conclude that primary structure variation in the COOH-terminal third of mrp and MRP is a major cause of differences in the ability of mrp and MRP to confer both anthracycline resistance and to transport E 2 17␤G. Attempts to further localize critical segments of MRP by exchange of smaller regions resulted in hybrid proteins (mrp/MRP 959 -1187 and mrp/MRP 1188 -1531 ) that both conferred intermediate levels of anthracycline resistance and had increased E 2 17␤G transport activity. Interestingly, both may be equally effective at transporting E 2 17␤G when their expression levels are normalized whereas mrp/MRP 959 -1187 confers proportionately higher levels of anthracycline resistance. These results suggest that the differences in functional characteristics of mrp and MRP probably involve alterations in amino acid sequence at more than one location in the COOH-proximal region.
Previously, we proposed that MRP contains identical or mutually exclusive binding sites for cholestatic steroids and leukotriene conjugates based upon the following experimental evidence. 1) LTC 4 and E 2 17␤G are able to compete reciprocally for ATP-dependent transport by MRP-enriched vesicles; 2) monoclonal antibody QCRL-3, which recognizes a conformation-dependent epitope in the first NBD of MRP, inhibits the transport of both substrates; and 3) E 2 17␤G inhibits photoaffinity labeling of MRP by [ 3 H]LTC 4 in a concentration-dependent manner (15,24,40,47). The fact that wild-type and hybrid proteins transport LTC 4 , but not E 2 17␤G, with comparable efficiencies clearly excludes the possibility that these two hydrophilic conjugates are recognized by identical determinants on the protein even if their binding is mutually exclusive.
It has been proposed that Pgp recognizes and transports its hydrophobic substrates from within the plasma membrane (9). This is supported by experiments in which Pgp-overexpressing cell lines have been shown to efflux the hydrophobic esters of the fluorescent dyes calcein AM and 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein AM (BCECF AM) but not their free acids, which are formed following cleavage by cytosolic esterases (41). In similar studies, MRP was able to transport both free calcein and calcein AM as well as BCECF AM, suggesting that the protein can recognize substrates in the cytoplasm as well as the plasma membrane (42)(43)(44)(45)(46). Thus initial interactions with MRP/mrp may occur via two different routes in which drugs and other hydrophobic substrates bind in the lipid environment of the plasma membrane, whereas the primary interaction of hydrophilic organic anions may be with the cytoplasmic loops of the protein. If so, it may be possible to design agents capable of inhibiting the ability of MRP to confer resistance to some chemotherapeutic agents without interfering with its ability to transport anionic physiological substrates. With this objective in mind, we are currently defining the locations of amino acid residues within the COOH-proximal third of mrp that contribute to its inability to confer anthracycline resistance and/or its relatively poor ability to transport E 2 17␤G.