Functional Importance of Three Basic Residues Clustered at the Cytosolic Interface of Transmembrane Helix 15 in the Multidrug and Organic Anion Transporter MRP1 (ABCC1)*

The multidrug resistance protein 1 (MRP1) mediates drug and organic anion efflux across the plasma membrane. The 17 transmembrane (TM) helices of MRP1 are linked by extracellular and cytoplasmic (CL) loops of various lengths and two cytoplasmic nucleotide binding domains. In this study, three basic residues clustered at the predicted TM15/CL7 interface were investigated for their role in MRP1 expression and activity. Thus, Arg1138, Lys1141, and Arg1142 were replaced with residues of the same or opposite charge, expressed in human embryonic kidney cells, and the properties of the mutant proteins were assessed. Neither Glu nor Lys substitutions of Arg1138 and Arg1142 affected MRP1 expression; however, all four mutants showed a decrease in organic anion transport with a relatively greater decrease in leukotriene C4 and glutathione transport. These mutations also modulated MRP1 ATPase activity as reflected by a decreased vanadate-induced trapping of 8-azido-[32P]ADP. Mutation of Lys1141 to either Glu or Arg reduced MRP1 expression, and routing to the plasma membrane was impaired. However, only the Glu-substituted Lys1141 mutant showed a decrease in organic anion transport, and this was associated with decreased substrate binding and vanadate-induced trapping of 8-azido-ADP. These studies identified a cluster of basic amino acids likely at the TM15/CL7 interface as a region important for both MRP1 expression and activity and demonstrated that each of the three residues plays a distinct role in the substrate specificity and catalytic activity of the transporter.

The multidrug resistance protein 1 (MRP1) mediates drug and organic anion efflux across the plasma membrane. The 17 transmembrane (TM) helices of MRP1 are linked by extracellular and cytoplasmic (CL) loops of various lengths and two cytoplasmic nucleotide binding domains. In this study, three basic residues clustered at the predicted TM15/CL7 interface were investigated for their role in MRP1 expression and activity. Thus, Arg 1138 , Lys 1141 , and Arg 1142 were replaced with residues of the same or opposite charge, expressed in human embryonic kidney cells, and the properties of the mutant proteins were assessed. Neither Glu nor Lys substitutions of Arg 1138 and Arg 1142 affected MRP1 expression; however, all four mutants showed a decrease in organic anion transport with a relatively greater decrease in leukotriene C 4 and glutathione transport. These mutations also modulated MRP1 ATPase activity as reflected by a decreased vanadate-induced trapping of 8-azido-[ 32 P]ADP. Mutation of Lys 1141 to either Glu or Arg reduced MRP1 expression, and routing to the plasma membrane was impaired. However, only the Glu-substituted Lys 1141 mutant showed a decrease in organic anion transport, and this was associated with decreased substrate binding and vanadate-induced trapping of 8-azido-ADP. These studies identified a cluster of basic amino acids likely at the TM15/CL7 interface as a region important for both MRP1 expression and activity and demonstrated that each of the three residues plays a distinct role in the substrate specificity and catalytic activity of the transporter.
Resistance to a wide variety of compounds used in chemotherapy is often due to the presence of energy-dependent membrane transport proteins that extrude these drugs, reducing their intracellular concentration below that required for cytotoxicity. One such transporter is the 190-kDa multidrug resistance protein (MRP) 3 1 (ABCC1) that is associated with the multidrug resistance phenotype frequently encountered in tumor cells (1)(2)(3). In addition to natural product antineoplastic agents, MRP1 substrates include organic anions conjugated to glutathi-one (e.g. LTC 4 ) and to glucuronate (e.g. E 2 17␤G), as well as unconjugated organic anions such as GSH and methotrexate (3). MRP1 belongs to the "C" subfamily of the ABC superfamily of membrane proteins, the members of which have been divided into seven classes (from A to G). There are 13 ABCC subfamily members, four of which are involved in genetic disorders. Thus, mutations in MRP2 (ABCC2) are found in patients with Dubin-Johnson syndrome (4), mutations in MRP6 (ABCC6) are responsible for the connective tissue disorder pseudoxanthoma elasticum (5), mutations in the cystic fibrosis transmembrane conductance regulator (CFTR/ABCC7) cause cystic fibrosis (6), and mutations in the sulfonylurea receptor SUR1 (ABCC8) underlie a genetic disorder known as persistent hyperinsulinemic hypoglycemia of infancy (7). Thus, far, no disease is associated with mutations in MRP1 (ABCC1), although it is possible that polymorphisms in MRP1 and other ABCC drug transporter genes may affect individual responses to drugs and other chemicals (8).
Both sequence analyses and biochemical studies predict that the core structures of the ABCC proteins consist of 12 TM helices divided between two MSDs, each of which is followed by an NBD (3). MRP1, as well as MRPs 2, 3, 6, and 7, contain a third NH 2 -terminal MSD composed of five TMs. A significant number of residues in the TMs or in close proximity to the membrane/cytosol interface of MRP1 (in particular, TMs 4,6,7,8,10,11,14,16,and 17) have been demonstrated by studies employing site-directed mutagenesis to be important for binding and transport of MRP1 substrates, as well as expression of the protein at the plasma membrane (9 -19). In contrast, with few exceptions, little is known about whether the extracellular loops and CLs that connect the TMs contain functionally important amino acids (20 -22).
Given that many of the molecules transported by MRP1 are anionic in character, it is perhaps not surprising that we and others have identified several positively charged residues that are important for substrate binding and/or transport. Mutations of some basic residues affect MRP1 transport in a substrate-selective manner. For example, Lys 332 and to a lesser extent His 335 , both located in predicted TM6, have been demonstrated to be selectively critical for LTC 4 binding and transport (11,12). Furthermore, the naturally occurring substitution of Arg 433 by a Ser at the interface of CL4 and TM8 causes a selective decrease in LTC 4 and estrone sulfate transport efficiency but increases drug resistance (23). In contrast, mutations of other basic residues cause a more general decrease in MRP1 transport activity. This was the case for opposite charge substitutions of Lys 396 (TM7) and Arg 593 (TM11) in the second MSD (12) and for substitutions of Arg 1197 (TM16) and Arg 1249 (TM17) in the third MSD (15,24). Thus, basic residues play several distinct roles as determinants of MRP1 function and substrate specificity depending on their location in the transporter.
Examination of the primary structure of MRP1 shows that some of the basic residues in this transporter occur in clusters. One such cluster includes the relatively highly conserved Arg 1138 , Lys 1141 , and Arg 1142 residues that are predicted to be part of an ␣-helix that, depending on the topology prediction algorithm used, is located at the cytosolic membrane interface of TM15 and CL7 (that connects TM15 to TM16) or in CL7 itself (Fig. 1). In the present study, these basic residues were replaced with same charge and opposite charge amino acids, and the consequences on MRP1 expression and function were investigated.

EXPERIMENTAL PROCEDURES
Materials- [14,15,19, (25,26). 4 These basic residues were also localized in the atomic homology models of MRP1 derived by molecular dynamics simulations using the crystal structure of the bacterial lipid transporter MsbA as template (10). The program Antheprot was used to predict the secondary structure of this region and to project the studied residues along a helical wheel. Sequences homologous (MRP2-9, CFTR, SUR1, SUR2) and orthologous (mouse, rat, monkey, canine, bovine Mrp1) to the Ser 1136 -Val 1146 segment in human MRP1 were aligned using ClustalW.
Transfections of MRP1 Expression Vectors in Human Embryonic Kidney Cells-Wild-type and Arg 1138 , Lys 1141 , and Arg 1142 mutant pcDNA3.1(-)-MRP1 expression vectors were transfected into SV40-transformed human embryonic kidney cells (HEK293T) (9). Briefly, ϳ10 ϫ 10 6 cells were seeded in 150-mm plates and transfected 24 h later (at 50 -80% confluency) with plasmid DNA (16 g) using FuGENE TM 6 (Roche Diagnostics) according to the manufacturer's instructions. After 72 h, the HEK293T cells were harvested, and membrane vesicles were prepared (9,27). Untransfected cells were included in all experiments as a negative control. Levels of wild-type and mutant MRP1 proteins were determined by immunoblot analysis as described below.
Measurements of MRP1 Protein Levels in Transfected Cells-The expression levels of wild-type and mutant MRP1 proteins were determined by immunoblot analysis of membrane protein fractions from transfected cells essentially as described (28) using the human MRP1specific murine monoclonal antibody QCRL-1 (diluted 1:10,000) (29) and a chemiluminescence detection kit (Western Lightning chemiluminescence reagent Plus blotting substrate (PerkinElmer Life Sciences)). Relative levels of MRP1 expression were estimated by densitometric analysis using ImageJ software. To confirm equal loading of protein in the wells, membranes were stained with Amido Black.
Confocal Microscopy-HEK293T cells were seeded at 3 ϫ 10 5 cells/ well in a 6-well plate on coverslips coated with 0.1% gelatin in Dulbecco's modified Eagle's medium containing 7.5% fetal bovine serum. Twenty-four h later, cells were transfected with the MRP1 constructs (1 g of plasmid DNA) using FuGENE TM 6 according to the manufacturer's instructions. Forty-eight h later, the coverslips were washed with PBS, and cells were fixed with 4% paraformaldehyde, washed in PBS, and permeabilized by adding 0.2% Triton X-100 in PBS. Cells were blocked with 1% bovine serum albumin in PBS and then incubated with MRP1specific monoclonal antibody QCRL-3 (diluted 1:2,500 in 0.1% Triton X-100, 1% bovine serum albumin in PBS) for 60 min (28,30). The coverslips were then washed in PBS and then incubated for 30 min with Alexa Fluor 488 anti-mouse IgG (HϩL) (FabЈ) 2 fragment in 0.1% Triton X-100 in PBS containing 10 g ml Ϫ1 RNase A, washed again, incubated with 2 g ml Ϫ1 propidium iodide, and then placed on slides containing SlowFade antifade solution (Molecular Probes, Inc., Eugene, OR). Cells were examined using a Leica TCS SP2 MS multiphoton system confocal microscope (Leica Microsystems, Heidelberg, Germany) (22).

MRP1-mediated Transport of 3 H-labeled Substrates by Membrane
Vesicles-ATP-dependent uptake of 3 H-labeled substrates by membrane vesicles was measured using a modified rapid filtration method (31) that was adapted to a 96-well microtiter plate format (32). LTC 4 transport assays were performed at 23°C in a 50-l reaction containing 1.8 g of membrane vesicle protein, 50 nM [ 3 H]LTC 4 (20 nCi/reaction), 10 mM MgCl 2 , 4 mM ATP or 4 mM AMP, 250 mM sucrose, and 50 mM Tris-HCl, pH 7.5 (transport buffer), with an ATP regenerating system consisting of 100 g ml Ϫ1 creatine kinase and 10 mM creatine phosphate. MRP1-mediated uptake was stopped after 1 min by rapid dilution in ice-cold buffer, the reaction mixture was filtered through a PerkinElmer Life Sciences unifilter GF/B plate, and the plate was washed four times with 2 ml of ice-cold Tris/sucrose buffer. Tritium associated with the vesicles was counted on a Top Count NXT microplate scintillation and luminescence counter (PerkinElmer Life Sciences). All data were corrected for the amount of [ 3 H]LTC 4 that remained bound to the filter, which was usually Ͻ10% of the total radioactivity. ATP-dependent LTC 4 uptake was calculated by subtracting the uptake in the presence of AMP from the uptake measured in the presence of ATP. The uptake rate was calculated based on the total protein content of the membrane vesicles. Unless specified, all transport assays were carried out in triplicate, and results were expressed as means (ϮS.D.). Uptake of [ 3 H]E 2 17␤G was measured in a similar fashion except that membrane vesicles (1.8 g of protein) were incubated at 37°C in a total reaction volume of 50 l containing 400 nM [ 3 H]E 2 17␤G (40 nCi/reaction) and the components as described for [ 3 H]LTC 4 uptake. Apigeninstimulated [ 3 H]GSH uptake was also measured by rapid filtration with membrane vesicles (20 g of protein) incubated at 37°C for 20 min in a 60-l reaction volume with 100 M [ 3 H]GSH (120 nCi/reaction, 10 mM dithiothreitol) and in the presence of apigenin (30 M). To minimize GSH catabolism by the ␥-glutamyltranspeptidase during transport, membranes were preincubated with 0.5 mM acivicin for 10 min at 37°C prior to measuring [ 3 H]GSH uptake (27). [ 3 H]MTX uptake was carried out at 37°C for 20 min in a 50-l reaction volume containing membrane vesicles (5 g of protein), 100 M MTX (200 nCi/reaction), and other components as above (33). To study the effect of E 2 17␤G on LTC 4 uptake, the same conditions were used as for [ 3 H]LTC 4 uptake described above, except that membrane vesicles were preincubated with different concentrations of E 2 17␤G (0, 12, 24 and 48 M) and uptake was stopped after 1 min of incubation at 23°C.
Kinetic Analysis of [ 3 H]E 2 17␤G Transport-K m and V max values of ATP-dependent E 2 17␤G uptake by membrane vesicles (4 g) were determined by measuring uptake at eight different E 2 17␤G concentrations (0.25-25 M) for 1 min at 37°C in 50 l of transport buffer containing components as described above. Data were analyzed using GraphPad Prism TM software, and kinetic parameters were calculated by non-linear regression and Michaelis-Menten analyses.
Photolabeling of MRP1 by [ 3 H]LTC 4 -Wild-type and mutant MRP1 membrane proteins were photolabeled with [ 3 H]LTC 4 essentially as described (31). Briefly, vesicles prepared from HEK293T cells trans-fected with wild-type and mutant MRP1 cDNAs (50 g of protein in 50 l) were incubated with [ 3 H]LTC 4 (120 nCi; 200 nM) and 10 mM MgCl 2 at room temperature for 30 min and then frozen in liquid nitrogen. Samples were then alternately irradiated at 302 nm for 1 min and snapfrozen in liquid N 2 10 times. Radiolabeled proteins (50 g) were resolved by SDS-PAGE, and the gel was processed for fluorography. The gel was exposed to Bioflex MSI film (InterScience, Troy, NY) for 2 days at Ϫ70°C. Relative levels of photolabeling were estimated by densitometric analysis as before.
Photolabeling of MRP1 by 8-Azido-[␣ 32 P]ATP-Membrane vesicles (15 g of protein) were dispersed in 20 l of transport buffer (50 mM Tris-HCl, pH 7.4, 250 mM sucrose) containing 5 mM MgCl 2 and 5 M 8-azido-[␣ 32 P]ATP. After a 5-min incubation on ice, the samples were exposed in an open, flexible 96-well plate to UV light at 302 nm on ice for 8 min at a distance of 8 cm. The reactions were stopped by the addition of 0.5 ml of ice-cold Tris-EGTA buffer (50 mM Tris-HCl, pH 7.4, 0.1 mM EGTA, 5 mM MgCl 2 ), and the membranes were centrifuged at 15,000 rpm for 15 min at 4°C. The pellets were washed again and resuspended in 20 l of the same buffer. Membrane proteins were then solubilized in Laemmli buffer, subjected to SDS-PAGE, and exposed to film (14,34).
Orthovanadate  A, human MRP1 residues 1136 -1146 have been aligned with the corresponding sequences in its ABCC homologs and orthologs using ClustalW. Residues that are identical to Arg 1138 , Lys 1141 , and Arg 1142 in MRP1 homologs and orthologs are in boldface type, whereas underlined residues indicate charge conservation. SUR, sulfonylurea receptor; Hum, human; Mus, mouse; Mon, monkey; Can, canine; Tau, bovine. B, the location of Arg 1138 , Lys 1141 , and Arg 1142 relative to the membrane varies according to the membrane protein topology algorithm used, although all three residues are in an ␣-helical structure. Outlined in a box is the nearest TM ␣-helix to Arg 1138 , Lys 1141 , and Arg 1142 (in boldface) as predicted by the algorithm of Eisenberg et al. (25) (i); the HMMTOPv2 and TMHMMv1 algorithms (ii); and the SOSUI algorithm with a modified "insidepositive" rule (26) (iii). Shown in iv is the location of Arg 1138 , Lys 1141 , and Arg 1142 according to an atomic homology model generated by molecular dynamics simulations using the crystal structure of the bacterial lipid transporter MsbA as template (10). A ␣-helix present in the homology model iv is shown below. C, a helical wheel projection of the predicted ␣-helix in MRP1 containing Arg 1138 , Lys 1141 , and Arg 1142 . Charged residues are black, and polar residues are shaded gray. The three basic residues mutated in the present study are indicated by asterisks, and a diamond denotes Arg 1131 mutated in a previous study (15).
buffer. The samples were transferred to a 96-well plate, irradiated, subjected to SDS-PAGE, and exposed to film as before.

RESULTS
Sequence Alignments and Location of Arg 1138 , Lys 1141 , and Arg 1142 in Human MRP1-Sequence alignments with other members of the C branch of ABC proteins show that Arg 1138 , Lys 1141 , and Arg 1142 are strictly conserved in human MRP2-5 as well as in all MRP1 orthologs sequenced to date (Fig. 1A). Arg 1138 is poorly conserved among all other ABCC family members except MRP7 and the sulfonylurea receptors SUR1/ABCC8 and SUR2/ABCC9. In contrast, Lys 1141 and Arg 1142 or at least their charges are quite well conserved except in SUR1 and SUR2.
In the case of MRP1, the topologies predicted by various algorithms are quite diverse, and the precise location of the clustered basic residues relative to the membrane differs substantially. However, all five programs used concur that the three residues are located in an ␣-helix (Fig.  1B). Most, including the Eisenberg (25), HMMTOPv2, TMHMMv2, and TM-Pred algorithms, predict that this helix (and thus Arg 1138 , Lys 1141 , and Arg 1142 ) are all well within a CL of the transporter, whereas the SOSUI algorithm (26) 4 places these basic residues at the membrane cytosol interface of TM15 with the TM boundary at residue Arg 1138 . Our recent atomic homology models of MRP1, which have been validated by a number of biochemical studies, also place this cluster of basic residues at the membrane cytosol interface of TM15 with the boundary of the TM portion of the helix ending at Lys 1141 or Leu 1140 (10). A helical wheel projection of this region of MRP1 indicates that Arg 1138 , Lys 1141 , and Arg 1142 are all located within a 100°arc on the face of the ␣-helix (Fig. 1C) where, according to our models, they all point in toward the putative substrate translocation pathway of the transporter. In these models, the closest neighbors of Arg 1138 , Lys 1141 , and Arg 1142 are pre-dicted to be residues in TMs 12, 14, and 17 (10). Also along the same face of the helix is the more NH 2 -proximal Arg 1131 residue investigated previously (15).
Protein Expression and Membrane Localization of Arg 1138 , Lys 1141 , and Arg 1142 Mutants of MRP1-Relative expression levels of the mutant proteins in membrane vesicles prepared from transfected HEK293T cells were determined by immunoblot analysis (Fig. 2A). Densitometric analysis showed that the Arg 1138 and Arg 1142 mutants were expressed at levels comparable with or only slightly lower (60 -80%) than that of wild-type MRP1, indicating that these two residues do not adversely affect the expression levels of MRP1 to any significant extent. Both Lys 1141 mutants were consistently expressed at lower levels (40 -50%) than wild-type MRP1, suggesting that substitutions of this residue may affect expression and/or processing of the transporter (Fig. 2A). Subsequent confocal microscopy studies showed that like wild-type MRP1, the Arg 1138 and Arg 1142 mutants are predominantly localized to the plasma membrane (Fig. 2B), although a portion of the opposite charge mutant R1142E could sometimes be found in intracellular membranes as well. Although also localized predominantly at the plasma membrane, a greater proportion of the Lys 1141 MRP1 mutant proteins were found in intracellular membranes. As these mutants were consistently expressed at lower levels than wild-type MRP1, these observations suggested that the loss of Lys 1141 may disrupt folding and/or processing of MRP1 leading to its retention in intracellular membranes.
Vesicular Uptake of Organic Anions by Arg 1138 , Lys 1141 , and Arg 1142 Mutants of MRP1-The effect of the mutations on the ability of MRP1 to transport organic anion substrates was measured in inside-out membrane vesicles, and the results are shown in Fig. 3. In general, the mutants with opposite charge substitutions showed a greater decrease in overall transport activity than did the mutants with the same charge substitutions, with some minor exceptions.
The most significant decreases (Ն50% inhibition) in vesicular uptake activity of the mutants were observed for LTC 4 (Fig. 3A) and GSH (Fig.  3B). Almost no GSH or LTC 4 transport could be detected for either of the Arg 1138 mutants, and transport of these substrates by both Arg 1142 mutants was also very low (30 -40% of wild-type). Thus, the presence of a positive charge at positions 1138 and 1142 was not enough to retain LTC 4 and GSH transport activity. In contrast, although substitution of Lys 1141 with Glu decreased LTC 4 transport by ϳ70%, an Arg substitution at this position did not. GSH transport by both Lys 1141 mutants was decreased by ϳ40% (Fig. 3B). Thus, the presence of a positive charge at position 1141 seemed sufficient for LTC 4 transport, whereas the less bulky side chain of Lys itself appeared important for GSH transport.
With respect to E 2 17␤G transport (Fig. 3C), Glu substitution of either of the two Arg residues or the Lys residue produced at least a 50% decrease in E 2 17␤G uptake. A same charge substitution of Arg 1138 and Lys 1141 had no effect, whereas the E 2 17␤G transport activity of the R1142K mutant was reduced by ϳ35%. These findings suggested that for retention of wild-type MRP1 E 2 17␤G uptake levels, the positive charges at positions 1138 and 1141 are important rather than the size of the side chain, whereas at position 1142, the presence of an Arg residue, although not essential, is preferred. The profile of vesicular uptake of estrone sulfate by the six mutants was similar to that seen for E 2 17␤G (results not shown). Finally, the effects of the mutations on MTX transport were relatively minor and ranged from no decrease to a maximum decrease of 35% (R1138E) when compared with wild-type MRP1 (Fig. 3D). 4 -To ascertain whether the decrease in LTC 4 uptake by the R1138E/K, K1141E, and R1142E/K mutants was related to a decrease in substrate binding, membrane vesicles enriched for wild-type or mutant MRP1s were photolabeled with [ 3 H]LTC 4 . As shown in Fig. 4, LTC 4 labeling of R1138E and K1141E was decreased by 50% or more, after correcting for differences in protein expression levels, consistent with their decreased LTC 4 transport activity. On the other hand, LTC 4 labeling of the R1138K mutant and both Arg 1142 mutants was comparable with wild-type MRP1, although these mutants had very low LTC 4 transport activity (Ͻ10 -40% wild type). Therefore, reduced substrate binding appeared to explain, at least in part, the loss of LTC 4 transport for the R1138E and K1141E mutants but not the R1138K, R1142E, and R1142K mutants.

Photolabeling of Arg 1138 , Lys 1141 , and Arg 1142 Mutants of MRP1 with [ 3 H]LTC
Kinetic Analysis of E 2 17␤G Uptake-The impact of the Lys 1141 mutations on E 2 17␤G transport was further investigated by determining kinetic parameters of E 2 17␤G uptake. Kinetic parameters (V max and K m ) were obtained after plotting the data according to the Michaelis-Menten equation for wild-type MRP1 and the K1141E and K1141R mutants ( Table 1). Consistent with previous reports, wild-type MRP1 exhibited a K m and V max of 1.9 Ϯ 0.4 M and 604 Ϯ 36 pmol mg Ϫ1 min Ϫ1 , respectively, for E 2 17␤G (12,14). However, although the V max for the K1141E mutant was unchanged, its K m for E 2 17␤G transport was significantly increased by 4.4-fold (8.8 Ϯ 0.7 M) when compared with the K1141R mutant (2.0 Ϯ 0.3 M) and wild-type MRP1 (1.9 Ϯ 0.4 M). This indicated that a decreased uptake affinity for E 2 17␤G contributes to the reduced E 2 17␤G transport observed for this Lys 1141 mutant. This decrease in E 2 17␤G affinity is consistent with the reduced ability of the conjugated estrogen to inhibit LTC 4 uptake by K1141E (Fig. 5). The K m (E 2 17␤G) values of the Arg 1138 and Arg 1142 mutants were not significantly different from wild-type MRP1, and accordingly, they exhibited no differences in the E 2 17␤G-mediated inhibition of LTC 4 transport when compared with wild-type MRP1 (data not shown). Uptake values were normalized based on mutant MRP1 protein levels relative to wildtype MRP1 protein levels (according to Fig. 2A) and are expressed as the percentage of uptake relative to uptake by WT-MRP1. The results shown are the means (ϮS.D.) of triplicate determinations in a single experiment, except for GSH uptake, which was done in duplicate. Similar results were obtained in at least two additional experiments with independently prepared batches of vesicles from independent transfections.   Fig. 2A).

Photolabeling of MRP1 Mutants with 8-Azido-[␣ 32 P] ATP and Vanadate-dependent 8-Azido-[␣ 32 P]ADP Trapping-We next investigated whether the mutations affected nucleotide interactions with MRP1.
Photolabeling with 8-azido-[␣ 32 P]ATP under non-hydrolytic conditions (4°C) showed that labeling of all six mutants was comparable with that of wild-type MRP1, indicating that the mutations had no major effects on the ability of MRP1 to bind ATP (Fig. 6A). In contrast, when the catalytic activity of MRP1 was measured at 37°C by the orthovanadate-induced trapping of 8-azido-[␣ 32 P]ADP in the post-hydrolysis state (Fig. 6B), five of the six mutants (R1138E/K, R1142E/K, and K1141E) showed substantially reduced nucleotide trapping (ϳ40 -60% that of wild-type MRP1). These observations suggested that the ability of these mutants to hydrolyze ATP is impaired, thus disrupting the transport cycle and contributing to the observed decreases in transport activity. Consistent with this interpretation, the orthovanadate-induced azido-ADP trapping by the same charge K1141R mutant, which had the least affected transport activity of all the mutants, was comparable with wild-type MRP1 (ϳ90%). 4 Photolabeling of the Arg 1142 Mutants of Human MRP1-Since the LTC 4 photolabeling of the MRP1 Arg 1142 mutants was not significantly different from wild-type MRP1 (Fig. 4) despite the fact that the LTC 4 transport activity of these mutants was reduced by 60 -80% (Fig.  3A), [ 3 H]LTC 4 photolabeling was performed in the presence of ATP alone or a combination of ATP and orthovanadate (Fig. 7). As shown previously, the strong [ 3 H]LTC 4 labeling of wild-type MRP1 was almost completely eliminated by ATP alone and by ATP/ADP in combination with vanadate, consistent with the idea that ATP binding to MRP1 causes the transporter to switch from a high to a low affinity binding state for LTC 4 and that the low affinity state persists after ATP hydrolysis (35,36). In contrast, although [ 3 H]LTC 4 labeling of the R1142E and R1142K mutants was also eliminated by ATP alone, labeling in the orthovanadate-induced ADP trapped (low affinity) state was stronger for these two mutants than for wild-type MRP1. One possible explanation for this is that the ability of R1142E and R1142K to release LTC 4 after ATP hydrolysis might be impaired, which could contribute to their reduced transport activity. The pattern of [ 3 H]LTC 4 labeling of the Arg 1138 mutants in the presence and absence of ATP and orthovanadate was similar to that observed with wild-type MRP1 (results not shown).

DISCUSSION
In this study, we have functionally characterized conservative and non-conservative mutations of a trio of basic residues in an ␣-helical structure in the third MSD of MRP1 and shown that Arg 1138 , Lys 1141 , and Arg 1142 each contribute in their own way to the expression, substrate specificity, and/or catalytic activity of the transporter. The relatively high degree of conservation of Arg 1138 , Lys 1141 , and Arg 1142 among ABCC homologs and orthologs is consistent with their functional importance demonstrated here and further suggested that analogous residues in related ABCC proteins are likely to be sensitive to mutation as well. In accordance with this conclusion, Le Saux et al. (37) found affected individuals in a family with autosomal recessive pseudoxanthoma elasticum to be homozygous for a mutation in ABCC6 (MRP6) resulting in substitution of Arg 1114 (which corresponds to MRP1-Arg 1142 ) to Pro (37). In addition, a His substitution of Arg 1150 in MRP2 (also analogous to MRP1-Arg 1142 ) has been reported in a group of Moroccan Jewish individuals affected by Dubin-Johnson syndrome (38). Functional characterization of this mutant MRP2 protein in vitro showed that it was deficient in organic anion transport activity, similar to the results reported here with its MRP1 counterpart.
Basic residues are well known to play a critical role in the proper tertiary structure and membrane insertion of membrane proteins.  After removal of unincorporated nucleotide, the samples were photo-crosslinked. In both cases, samples were resolved by SDS-PAGE. The dried gel was exposed to film for 1-24 h at room temperature. The relative levels of photolabeling were determined by densitometric analysis of the films and are indicated by the numbers below the figures, expressed before and after correcting for differences in protein expression. Similar results were obtained in at least two additional independent experiments.
Indeed, positively charged amino acids are often preferentially located near the membrane interfaces where they can interact with the negatively charged head groups of the membrane phospholipids. In addition, basic residues are predominantly positioned at the cytoplasmic rather than the extracellular face of the membrane, according to the "positiveinside" rule (26,39). Basic residues in this position can influence the membrane topology and anchoring of the protein in the membrane, and in doing so, they can enhance protein stability. It seems somewhat surprising, therefore, that most algorithms for predicting membrane protein topology do not place Arg 1138 , Lys 1141 , and Arg 1142 at a membrane interface region of MRP1. The exception is the SOSUI algorithm, which localizes this cluster of basic residues to the interface region of CL7 and TM15 of MRP1. This localization is in agreement with a proposed MRP1 structure derived by atomic homology modeling using molecular dynamics simulations and supported by a number of biochemical studies (10). The predicted location of Arg 1138 , Lys 1141 , and Arg 1142 (and the previously characterized Arg 1131 (15)) within a 100 o arc of an ␣-helix extending from TM15 into the cytoplasm (Fig. 1C), together with the prediction by atomic homology models that all four residues face into the putative substrate translocation pathway formed by the second and third MSDs, is consistent with their importance in MRP1 function. However, the varied properties of the six mutants characterized here also make it clear that even if Arg 1138 , Lys 1141 , and Arg 1142 are assumed to be at the TM15/CL7 interface of MRP1, they have a more complex role than simply acting as membrane anchors.
When Arg 1138 was replaced by either Lys or Glu, a major and relatively selective decrease in GSH and LTC 4 transport activity was observed, whereas transport of E 2 17␤G and MTX was only moderately affected or not affected at all. The reduction in GSH and LTC 4 transport appeared largely attributable to a substantial decrease in substrate binding in the case of R1138E but not R1138K. However, the basal ATPase activity of both mutants as reflected by the ability of orthovanadate to induce trapping of 8-azido-ADP was substantially reduced. This suggested that the reduction in basal ATPase activity caused by the Arg 1138 mutations adversely affects the transport of some organic anions more than others.
In contrast to the Arg 1138 mutants, the expression levels of the same charge and opposite charge Lys 1141 mutants were reduced, and routing of these proteins to the plasma membrane was disrupted to a significant degree. Nevertheless, the vesicular transport activity of the K1141R mutant was similar to wild-type MRP1, as was its ability to be photolabeled with LTC 4 and its ATPase activity. In contrast, the transport activity of the K1141E mutant was markedly decreased. This decrease appeared largely attributable to decreased substrate binding, as reflected by a substantial reduction in [ 3 H]LTC 4 photolabeling and a 4.4-fold increase in its K m for E 2 17␤G. Its decreased affinity for E 2 17␤G was further corroborated by the reduced ability of this conjugated estrogen to inhibit the residual LTC 4 transport activity of the K1141E mutant. However, orthovanadate-induced trapping of azido-ADP was also reduced, suggesting that the ability of the K1141E mutant to hydrolyze ATP is impaired. Thus, despite the differences in their side chain volumes and pI, the bulkier Arg can readily substitute for Lys at position 1141. Taken together, these data suggested that a positive charge at position 1141 is needed not only for high affinity substrate binding but also for the efficient catalytic activity of MRP1.
As with the Arg 1138 mutants and in contrast to the Lys 1141 mutants, our observations with R1142K and R1142E indicated that it is necessary to preserve both the charge and the size of the cationic side chain at position 1142 for full MRP1 transport activity. This may reflect the capacity of Arg to form bidentate hydrogen bonds that Lys cannot. Thus, both the Arg 1142 mutants showed substantially reduced transport of LTC 4 , GSH, and E 2 17␤G. Similarly, orthovanadate-induced azido-ADP trapping by both Arg 1142 mutants was reduced. However, photolabeling of the Arg 1142 mutants by [ 3 H]LTC 4 was comparable with wildtype MRP1, indicating that, in contrast to the Arg 1138 mutants and the K1141E mutant, their substrate binding properties are still intact. Also, unlike the Arg 1138 and Lys 1141 mutants, our data suggested that the ability of the Arg 1142 mutants to switch from a high affinity substrate binding state to a low affinity state may be impaired.
It is interesting that despite some differences in their substrate specificities and expression levels, all three non-conservatively substituted Arg 1138 , Lys 1141 , Arg 1142 mutants showed significantly reduced levels of orthovanadate-induced trapping of azido-ADP, but only the Arg 1142 mutations affected the ability of MRP1 to switch from a high affinity substrate binding state to a low affinity state. That each of these three basic residues was sensitive to mutation is all the more notable given that the close proximity of so many same charge residues would presumably be quite energetically unfavorable. Replacement of these residues would almost certainly perturb the geometry of the ␣-helix in which they are located, and this in turn could disrupt the movements of the loop connecting this helix to the TM segments.
The increasing availability of reliable crystal structures of ABC proteins, although still mostly from prokaryotes, has provided atomic information about the structures of the NBDs, and to some degree, the arrangements of TM helices (40). The crystal structure of the vitamin B12 transporter BtuCD and a recent structure of the lipid transporter MsbA have elicited particular interest because they provide physical evidence of contact between cytoplasmic loops that link the TMs and the NBDs themselves and suggest a functional interaction between them (41,42). However, it is worth mentioning that although the basic residues described in the present study might interact with the NBDs, they are not located in a region of MRP1 that corresponds to the cytoplasmic L-loop described in the crystal structure of BtuCD to make contact with NBDs (41).
In MRP1, it seems conceivable that Arg 1138 , Lys 1141 , and Arg 1142 could have a role in maintaining the structure or motion of CL7 during the transport cycle that is important for its interaction with NBD2, where most of the ATPase activity of MRP1 resides (34). Thus, the mutations may disrupt the communication between the NBDs and TMs that is necessary for coupling the binding and hydrolysis of ATP by MRP1 with the various steps of the transport cycle. This would be consistent with the deleterious effects that mutations in the comparable loops of the human ABC transporters CFTR, ABCC2/MRP2, and ABCC6/MRP6 have in individuals with cystic fibrosis, Dubin-Johnson syndrome, and pseudoxanthoma elasticum, respectively (8). Further- more, a global reduction in transport activity has also been observed when either Tyr 1189 or Tyr 1190 further downstream in CL7 was mutated (43). Also relevant in this regard is our previous study that showed that Ala substitution of the highly conserved Pro 1150 found at the COOH terminus of the ␣-helix containing Arg 1138 , Lys 1141 , and Arg 1142 leads to a complex phenotype in which increased E 2 17␤G transport is associated with an increased apparent uptake affinity (K m ) for this substrate but that orthovanadate-induced trapping of azido-ADP is essentially not detectable (14).
In conclusion, we have identified a region not previously implicated in MRP1 function, where three basic residues (most likely at the TM15/ CL7 interface) play a key role in substrate specificity and ATPase activity of the transporter, as well as its expression and routing to the plasma membrane. The distinct phenotypes associated with each of the mutants are presumably related to the fact that each of the three basic residues forms its own unique set of interhelical and intrahelical bonding interactions with other amino acids in the transporter and possibly with phospholipid head groups in the membrane bilayer. Thus far, charged amino acids, the substitution of which adversely affects MRP1 function, have typically been found either in essential motifs in the NBDs or within a TM helix or close to the TM-cytosol interface, as is likely the case for Arg 1138 , Lys 1141 , and Arg 1142 . On the other hand, relatively little is known about the functional importance of charged residues in the loops that connect the TM helices, although several predicted to be in the extracellular loop connecting TM10 to TM11 have been shown to be insensitive to mutation (12). Other clusters of charged residues are located downstream of the three basic residues studied here, with the SOSUI algorithm and our atomic homology model of MRP1 placing them well within CL7. In view of the growing body of evidence supporting a critical role for CL7 in the signaling between the activity of the NBDs and substrate translocation across the membrane bilayer (14,43), mutational analyses of these amino acids are the subject of ongoing investigations.