Photolabeling of Human and Murine Multidrug Resistance Protein 1 with the High Affinity Inhibitor [125I]LY475776 and Azidophenacyl-[35S]Glutathione*

Multidrug resistance protein 1 (MRP1/ABCC1) is an ATP-dependent transporter of structurally diverse organic anion conjugates. The protein also actively transports a number of non-conjugated chemotherapeutic drugs and certain anionic conjugates by a presently poorly understood GSH-dependent mechanism. LY475776is a newly developed125I-labeled azido tricyclic isoxazole that binds toMRP1 with high affinity and specificity in a GSH-dependent manner. The compound has also been shown to photolabel a site in the COOH-proximal region of MRP1's third membrane spanning domain (MSD). It is presently not known where GSH interacts with the protein. Here, we demonstrate that the photactivateable GSH derivative azidophenacyl-GSH can substitute functionally for GSH in supporting the photolabeling of MRP1 by LY475776 and the transport of another GSH-dependent substrate, estrone 3-sulfate. In contrast to LY475776, azidophenacyl-[35S] photolabels both halves of the protein. Photolabeling of the COOH-proximal site can be markedly stimulated by low concentrations of estrone 3-sulfate, suggestive of cooperativity between the binding of these two compounds. We show that photolabeling of the COOH-proximal site by LY475776 and the labeling of both NH2- and COOH- proximal sites by azidophenacyl-GSH requires the cytoplasmic linker (CL3) region connecting the first and second MSDs of the protein, but not the first MSD itself. Although required for binding, CL3 is not photolabeled by azidophenacyl-GSH. Finally, we identify non-conserved amino acids in the third MSD that contribute to the high affinity with which LY475776 binds to MRP1.

MRP1 is an active transporter of an extremely diverse array of organic anion conjugates as well as a number of anionic non-conjugated compounds. Some of the most well characterized substrates, such as the cysteinyl leukotriene C 4 (LTC 4 ), are conjugated to glutathione (10 -12), but the protein also transports glucuronidated (13,14) and sulfated compounds (15,16). The range of MRP1 substrates includes additional nonconjugated neutral or cationic chemotherapeutic drugs as well as certain anionic conjugates that display a dependence on GSH for transport (14, 16 -19). In some cases GSH appears to be co-transported with these substrates (17). In others no increase in GSH transport has been detected (14,16). GSH alone can be transported by MRP1, but with low affinity and efficiency (14,16). The transport of GSH can also be stimulated by certain compounds such as verapamil in the absence of net transport of verapamil itself (20).
MRP1 and its related ABCC proteins, MRP 2, 3, 6, and 7, differ topologically from most eukaryotic ABC transporters. In addition to the two polytopic membrane spanning domains (MSD) typical of ABC transporters, these proteins contain a third NH 2 -terminal MSD (1,3,21). In the case of MRP1 and 2, this domain has been shown to be comprised of five TM helices with an extracellular NH 2 terminus (22)(23)(24)(25). The function of this additional domain is presently not known. However, it has been established that MRP1 retains the ability to transport LTC 4 even if this domain is removed (26,27). Whether this is also true of other substrates, particularly those that depend on GSH for their transport, remains to be established.
In most cases, MRP1 substrates compete reciprocally for transport regardless of the identity of the conjugate moiety and the lack of structural similarity of the parental compounds. Some substrates that display GSH dependence for transport also compete with those that do not, although as might be expected their inhibitory potency is increased in the presence of GSH (14,16,17). These observations suggest that structurally diverse substrates interact with partially shared sets of amino acids that collectively form a binding surface or pocket on the protein. Recent studies indicate that a number of amino acids in predicted TM helix 17 and at least one residue in TM helix 14 are important in determining substrate specificity (28 -30). Furthermore, it is possible to make compensatory mutations in these two helices, suggesting that they may be in spatial proximity to one another in the native protein (30).
As an alternative approach to identifying regions of the protein directly involved in substrate binding, MRP1 has been photolabeled by its known high affinity physiological substrate LTC 4 as well as by compounds that are less well characterized with respect to binding and transport (10,12,27,(31)(32)(33)(34). Although these compounds compete with LTC 4 for binding to MRP1, their affinities appear to be considerably lower, and they also interact with the distantly related P-glycoprotein. Two of the compounds, 125 I-labeled N-(hydrocinchonidin-8Ј-yl)-4-azido-2-hydroxybenzamide (IACI) and iodoaryl azido rhodamine123 (IAARh123), bind to the protein in a GSHindependent manner and label sites in both the NH 2 -and COOH-proximal halves of the protein (32,33,35) (27). Both iodinated compounds have been shown to cross-link preferentially to proteolytic peptides that contain either TM helices 10 and 11 or helices 16 and 17 (31). A third non-physiological compound, an azido derivative of the marine sponge polyhydroxylated sterol acetate, agosterol-A (AG-A), has also been used to photolabel MRP1 (34). In contrast to the other compounds, binding of AG-A is GSH dependent, and labeling is restricted to a site in the COOH-proximal half of the protein (34). On the basis of these results it has been suggested that the site labeled by AG-A has a high affinity for drugs and that GSH binds to a site presumed to be in the NH 2 -terminal half of the protein, possibly involving the cytoplasmic linker (CL3) region between MSD1 and MSD2 (34). However, direct binding studies with photoactivateable derivatives of GSH have not been reported.
In this study we have used various combinations of MRP1 fragments co-expressed in insect Sf21 cells together with human and murine hybrid and mutant proteins to investigate the binding of a new and highly specific inhibitor of MRP1. LY475776 is an iodinated azido tricyclic isoxazole that displays essentially complete dependence on GSH for binding to MRP1, and which has a higher affinity for the protein than LTC 4 2 (36, 37). Concurrent with these studies we have also examined the binding of azidophenacyl-GSH, which we show can effectively support transport of a GSH-dependent substrate such as estrone sulfate as well as the photolabeling of MRP1 by LY475776. We have recently shown by limited trypsinolysis of full-length MRP1 that the major site photolabeled by LY475776 resides in the COOH-proximal segment of MSD3 (36). Using baculovirus co-expressed fragments of MRP1, we demonstrate that both halves of the protein are required for photolabeling of the site in MSD3, although photolabeling can occur in the absence of MSD1 (amino acids 1-203). However, a region in the CL3 connecting MSD1 and MSD2 (amino acids 204 -280) is essential for photolabeling. In contrast to the photolabeling profile of LY475776, azidophenacyl-GSH labels sites in both halves of the protein. Labeling of the COOH-proximal site, but not the NH 2 -proximal site, markedly stimulated the GSH-dependent substrate estrone 3-sulfate. Although CL3 is required for photolabeling of MRP1 by both LY475776 and azidophenacyl-GSH, neither compound photolabels this region of the protein. Finally, we demonstrate that LY475776 has a higher affinity for MRP1 when compared with its murine ortholog mrp1 and identify non-conserved amino acids in MSD3 that contribute to the binding of this compound by the human protein.   ) was cloned into pFASTBAC and MRP1 half molecules (MRP1 1-932 , MRP1 932-1531 ), and fragments (MRP1 1-280 , MRP1 1-932⌬228 -280 , and MRP1 281-1531 ) were cloned into pFASTBAC DUAL expression vectors (Invitrogen), as described (38 -40). Recombinant bacmids and baculoviruses were generated as previously described (38). Following infection of Sf21 cells, membrane vesicles were prepared by nitrogen cavitation and purified by sucrose gradient centrifugation (12,41).
Quantitation of MRP Polypeptides and Membrane Vesicle Preparation-SDS-PAGE of membrane vesicle preparations was performed as described previously using 5-15% gradient gels (38,45). Following transfer of the membrane proteins to Immobilon-P membranes (Millipore), MRP1 polypeptides were detected using an enhanced chemiluminescence kit (PerkinElmer Life Sciences) and monoclonal antibodies QCRL-1 and MRPm6 for MRP1 and MRPr1, which recognize a common epitope in MRP1 and mrp1 (46,47).
Vesicle Transport of LTC 4 and Estrone 3-Sulfate-The ATP-dependent uptake of LTC 4 and estrone 3-sulfate was measured using a rapid filtration assay as described previously (12,16).
Synthesis of Azidophenacyl-[ 35 S]GSH-Azidophenacyl-[ 35 S]GSH was prepared as previously described (48). In brief, 125 Ci of [ 35 S]GSH was extracted with ethyl acetate to remove dithiothreitol before being added to a reaction mixture containing potassium phosphate buffer (50 mM, pH 7.4), 4-azidophenacylbromide (10 mM), GSH reductase (120 milliunits), and NADPH (1 mM). The reaction was allowed to proceed at 22°C for 1 h, and the products were separated by Silica G thin layer chromatography using 1-propanol/water/acetic acid (12:5:1, v/v). The area on the thin layer chromatography plate containing the radioactivity with an Rf corresponding to unlabeled azidophenacyl-GSH was scraped off and extracted with 400 l of water six times. The extract was concentrated under a nitrogen stream.

Photolabeling of MRP1 and Related Proteins with [ 125 I]LY475776 or Azidophenacyl-[ 35 S]GSH-Membrane vesicles (75 g of protein from
Sf21 cells or 50 g of protein from HEK cells in 35 l of 50 mM Tris⅐HCl, pH7.4, 250 mM sucrose) were incubated with [ 125 I]LY475776 (0.5 nM) at 37°C or azidophenacyl-[ 35 S]GSH (0.5Ci) at room temperature for 10 min and UV-irradiated for 5 min at 254 nm or 312 nm on ice, for LY475776 and or azidophenacyl-[ 35 S]GSH, respectively. Proteins were solubilized in Laemmli's buffer and analyzed on a 5-15% gradient gel by SDS-PAGE. The gel was then dried onto blotting paper and exposed to x-ray film at room temperature for detection of [ 125 I]LY475776. For the azidophenacyl-[ 35 S]GSH-labeled proteins, gels were treated with Amplify (Amersham Biosciences), dried, and exposed to film at Ϫ70°C. Exposure times were typically 1-3 days.

Photolabeling of Truncated and Co-expressed MRP1 Fragments with [ 125 I]LY475776
-The ability of [ 125 I]LY475776 to label MRP1 was confirmed using plasma membrane vesicles from Sf21 insect cells and stable transfectants of human embryonic kidney (HEK) cells expressing the full-length protein.
As observed previously using membranes from multidrug resistant H69AR cells, photolabeling was specific for a protein of the anticipated size of MRP1 and was detectable only in the presence of GSH (Fig. 1, A and C, respectively) (36). No labeling of a comparably sized protein was detected when membranes from cells infected with a control vector encoding ␤-glucuronidase (␤-gus) were used, confirming that the protein was indeed MRP1 (Fig. 1C). Our recent studies have shown that photolabeling of MRP1 with the high affinity substrate LTC 4 results in labeling of a site in the NH 2 -proximal half of MRP1 and weaker labeling of a site in the COOH-proximal half (27). We have also demonstrated that association of the two halves of the protein is a prerequisite for labeling of both sites. To determine whether this was also the case for [ 125 I]LY475776, we examined photolabeling of individually expressed and co-expressed half molecules of MRP1. As observed previously with [ 3 H]LTC 4 , no binding of [ 125 I]LY475776 was detected when either halfmolecule was expressed alone (Fig. 1B). In contrast to the results obtained with LTC 4 , when the two halves of the protein were co-expressed, labeling occurred predominantly in the COOH-proximal half of the protein in the region extending from amino acid 932 to 1531, consistent with previous trypsinolysis studies using H69AR membranes (36). With the coexpressed half-molecules, we also detected very weak labeling of the NH 2 -proximal portion (Fig. 1B). Although weak, this labeling was GSH dependent and was enhanced in the presence of S-methyl-GSH, which we have shown previously can substitute for GSH in supporting the transport of some MRP1 substrates (Fig. 1B, right panel) (17). Thus, the weak photolabeling of the NH 2 -half of the protein appears to be the result of a specific GSH-dependent interaction.
Previous labeling and transport studies have shown that MSD1 extending from amino acid 1 to 204 is not required for the binding and transport of [ 3 H]LTC 4 (26,27). Deletion of part of CL3 between amino acids 204 and 281 essentially abolishes photolabeling of the region containing MSD2 and NBD1, but its removal has less of an effect on labeling of the COOH-proximal site (27,39). Consequently, we examined the extent to which binding and photolabeling with [ 125 I]LY475776 was dependent on the presence of both MSD1 and CL3. MRP1 lacking amino acids 1-204 could be efficiently photolabeled (Fig. 1C, left panel). However, truncation to amino acid 280 essentially eliminated labeling, and the low level of labeling detected was unaffected by GSH (Fig. 1C). As with LTC 4 , co-expression of MRP1 281-1531 with a fragment comprised of the missing NH 2proximal amino acids restored strong GSH-dependent labeling of the larger fragment despite the fact that LY475776 primarily labels the COOH-proximal region of the protein (Fig. 1C, right panel).
Azidophenacyl-GSH Can Substitute Functionally for GSH in Enhancing the Transport of Estrone Sulfate by MRP1 and Photolabeling of the Protein by LY475776 -Previous studies have shown that some MRP1 substrates that display a dependence on GSH for transport, such as vincristine, reciprocally stimulate the transport of GSH (17). These studies have also shown that GSH and vincristine markedly and reciprocally increase the apparent affinity with which each interacts with the protein, suggesting that there is positive cooperativity between binding of the two substrates. In contrast, other substrates such as estrone sulfate and 4-(methylnitrosamino)-1-(3pyridyl)-1-butanol-O-glucuronide (NNAL) glucuronide, which are also transported by MRP1 in a GSH-dependent manner, appear not to have a reciprocal effect on GSH transport (14,16). To directly examine GSH binding, we synthesized the photoactivateable analog azidophenacyl-GSH that has been shown previously to bind to glutathione S-transferase and to MRP1 (49,50). To establish whether this analog like some other GSH derivatives could substitute functionally for the parent compound (14,17,36), we tested its ability to stimulate transport of estrone sulfate and to enhance MRP1 binding of LY475776. Azidophenacyl-GSH at 100 M stimulated estrone sulfate transport ϳ2.5-fold, a level of stimulation similar to that obtained previously with the same concentration of GSH ( Fig. 2A) (16). The derivative also stimulated photolabeling of full-length MRP1 expressed in Sf21 cells, despite the fact that its solubility limited the concentration used to 100 M ( Fig. 2A). Thus, azidophenacyl-GSH is able to substitute for its parent molecule, both with respect to enhancing estrone sulfate transport and LY475776 binding. We then examined the ability of azidophenacyl-[ 35 S]GSH to photolabel MRP1. As shown in the highest concentration of LY475776 tested (Fig. 3A). In addition, we were unable to detect any stimulation of [ 3 H]GSH transport by LY475776 (data not shown). In contrast, photolabeling of MRP1 by azidophenacyl-[ 35 S]GSH was modestly increased in the presence of the GSH-dependent substrate estrone sulfate (Fig. 3B).
Photolabeling of the NH 2 (Fig. 4A). Thus, the profile of labeling is similar to that obtained with LTC 4 (27). As observed with both LTC 4 and LY475776, truncation of the protein to amino acid 280 essentially eliminated photolabeling with azidophenacyl-[ 35 S]GSH, as did an internal deletion of CL3 from amino acids 228 -280 (Fig. 4B). Labeling of MRP1 281-1531 could be efficiently restored by co-expression with a fragment containing amino acids 1-280 (Fig. 4B). No labeling of this NH 2 -terminal fragment could be detected, either when it was expressed alone or when co-expressed with MRP1 281-1531 .
We also determined whether the stimulation of photolabeling of the intact protein by estrone sulfate affected both sites equivalently. The conjugated estrogen markedly stimulated labeling of the COOH-proximal half of the protein in a concentration-dependent fashion with relatively little change in the extent of labeling of the NH 2 -proximal site (Fig. 4C). The increase in photolabeling of the COOH-proximal site was readily observed at a concentration of 1 M estrone sulfate that approximates the K m for transport of this compound in the presence of GSH. No stimulation of binding to either site was detected when a comparable experiment was carried out with LY475776 in place of estrone sulfate (data not shown).
[ 125 I]LY475776 Photolabels MRP1 More Efficiently Than Its Murine Ortholog, mrp1-Although MRP1 and its murine or-tholog are highly conserved, they differ with respect to their ability to confer resistance to several anthracyclines and to transport some estrogen conjugates, such as E 2 17␤G and estrone sulfate (43,51). 3 We have shown that sequence variations in the COOH-proximal third of the two proteins play a major role in these functional differences (28,30,44). Because this region of the protein contains the site that is strongly labeled by LY475776, we compared the relative ability of LY475776 to inhibit transport of LTC 4 by MRP1 and murine mrp1 (Fig. 5A). These analyses revealed that LY475776 was a more potent inhibitor of transport by the human protein than its murine ortholog. In the presence of 1 mM GSH, the EC 50 of LY475776 was ϳ50 and 300 nM for MRP1 and mrp1, respectively.
To further define the structural basis of the difference in  inhibitory potency of LY475776 with the human and murine proteins, we carried out photolabeling studies with wild type MRP1 and mrp1, as well as several previously characterized human/murine hybrid proteins (44). Membrane vesicles were prepared from HEK cells expressing comparable levels of either full-length MRP1 or mrp1 and photolabeled with azido-[ 125 I]LY475776 in the absence or presence of 1 mM GSH. Consistent with the inhibition studies, these experiments revealed that labeling of mrp1 was much weaker than that of the human protein (Fig. 5B). To localize the region responsible for the difference in labeling intensity, we used four previously characterized human and murine hybrid proteins (44). Photolabeling of a hybrid containing the NH 2 -proximal two-thirds of MRP1 was indistinguishable from wild type mrp1. However, photolabeling of the complementary hybrid containing the amino acids 959 -1531 of MRP1 was clearly enhanced relative to the wild type murine protein (Fig. 6B). To further define the region responsible for increased photolabeling, we used two additional hybrids that contained either amino acids 959 -1187 or 1188 -1531 of MRP1. Although neither was photolabeled to the same extent as the hybrid containing the entire region of MRP1 from 959 -1531, labeling of the hybrid containing MRP1 959 -1187 was enhanced relative to wild type mrp1 (Fig. 5B). No enhancement was observed with the hybrid containing MRP1 1188 -1531. Qualitatively, these results are similar to those we obtained when examining anthracycline resistance, which indicated that residues specific to the human protein in the region between amino acids 959 and 1187 are important for resistance to this class of drugs (44).
Identification of Individual Amino Acids Involved in Binding of LY475776 -Within the region between amino acids 959 and 1187 of MRP1, we have identified one non-conserved amino acid that is essential for conferring anthracycline resistance (28). Conversion of Glu-1089 in MRP1 to Gln, the corresponding amino acid in mrp1, abolished the ability to confer anthracycline resistance, whereas the reciprocal mutation in mrp1 created a protein that could confer resistance to this class of drugs (28). Consequently, we examined the ability of these two mutant proteins to bind both azido-[ 125 I]LY475776 and azidophenacyl-[ 35 S]GSH. The extent of [ 125 I]LY475776 labeling of E1089Q MRP1 was decreased severalfold relative to that of wild type protein, whereas [ 125 I]LY475776 labeling of Q1086E mrp1 was slightly enhanced when compared with wild type mrp1 (Fig. 6A, left panel). These mutations neither affected the ability of MRP1 or mrp1 to transport LTC 4 nor affected labeling by azidophenacyl-[ 35 S]GSH (Fig. 6A, right panel) (28).
To further investigate the labeling characteristics of LY475776, we carried out photolabeling experiments with mrp1 and MRP1 mutant proteins in which we had exchanged corresponding non-conserved amino acids in TM17, Thr-1242 in MRP1, and Ala-1239 in mrp1. We have shown previously that Thr-1242 in MRP1 is important for its ability to transport E 2 17␤G and to confer resistance to various drugs, but mutation of this residue has little effect on LTC 4 transport (30,52). Conversion of Thr-1242 in MRP1 to Ala as in mrp1, significantly decreased the extent of [ 125 I]LY475776 labeling of the protein, and the reciprocal mutation slightly increased labeling of the murine protein (Fig 6B, left panel). In contrast, when the same preparations of HEK membranes were photolabeled with azidophenacyl-[ 35 S]GSH, the extent of labeling of T1242A MRP1 was higher than that of the wild type protein. However, the relative levels of labeling paralleled the expression of the wild type and mutant proteins, as detected by Western blotting analysis (Fig. 6B, data not shown). Thus, in contrast to the effect on photolabeling by LY475776, mutation of this residue affects neither LTC 4 transport nor the binding of azidophenacyl-[ 35 S]GSH. DISCUSSION Photolabeling studies of MRP1 using azido derivatives of IACI and IAARh123 followed by partial proteolysis have identified two preferential sites of cross-linking involving peptides containing TM helices 10 and 11 or TM 16 and 17 (31)(32)(33). Our previous photolabeling studies of MRP1 have shown that LTC 4 , like IACI and IAARh123, labels sites in MSD2 and MSD3 (27). In the case of LTC 4 , labeling of the site in the NH 2 -proximal half of the protein is stronger than the COOH-proximal site and is preferentially attenuated when the protein is trapped in a transition state with NBD2 occupied by an ADP:Vi:PO 4 complex (27). The photolabeling by LTC 4 is also highly dependent on the integrity of CL3 but not on the presence of MSD1.
In contrast to the results obtained with compounds that bind and photolabel MRP1 in a GSH-independent manner, the azido derivative of AG-A in the presence of GSH labels only the COOH-proximal half of the protein (34). AG-A competitively inhibits LTC 4 uptake by MRP1-enriched membrane vesicles with a K i of 30 M (34). Whether or not AG-A is actually transported by the protein has not been established. Based on the profile of photolabeling observed with the azido derivative of AG-A, it has been proposed that the binding of GSH to a site in the NH 2 -half of MRP1, possibly CL3, enhances the binding of a second substrate, such as AG-A, to a site in MSD3 (34).
LY475776 is structurally unrelated to AG-A but also binds to MRP1 in a GSH-dependent manner 2 (36). However, the EC 50 for inhibition of LTC 4 transport by MRP1 is ϳ50 nM, compared with a K i of 30 M for AG-A (53). LY475776 is also specific for MRP1 and does not photolabel the MRP1 homologues MRP2 and MRP3, or P-glycoprotein (P-gp 2,3 ). Recent proteolysis studies in which [ 125 I]LY475776 has been used to photolabel MRP1 expressed in the multidrug resistant human small cell lung cancer cell line, H69AR, have revealed predominant labeling of a tryptic peptide containing TM helices 16 and 17 in addition to two minor fragments thought to contain TM helices 12 and 13 and 14 and 15 (36). Consistent with this observation, photolabeling studies with co-expressed baculovirus-encoded MRP1 1-932 and MRP 932-1531 clearly indicate that the compound strongly and preferentially photolabels a site in the COOHproximal fragment in a GSH-dependent manner. These studies also establish that co-expression with the NH 2 -proximal portion of the protein is a prerequisite for any photolabeling to occur. Thus, the pattern of labeling is similar to that reported for azido AG-A (34). However, using the co-expressed fragments of MRP1, we also detected GSH and S-methyl-GSH-dependent photolabeling of MRP1  . Although very weak relative to the labeling of the COOH-proximal site, dependence on the presence of GSH or a non-reducing analog such as Smethyl-GSH suggests that the binding is specific. If so, it implies that the NH 2 -proximal site is not simply a site to which GSH alone binds and facilitates the binding of a GSH-dependent substrate to a site in MSD3, as proposed (34).
We have attempted to gain more information about the binding characteristics of GSH by using the photoactivateable derivative, azidophenacyl-[ 35 S]GSH. Azidophenacyl-GSH has been reported previously to photolabel intact MRP1 (50). However, its ability to substitute for GSH in stimulating transport of other MRP1 substrates had not been demonstrated. Previous studies have shown that a number of GSH derivatives including the alkyl derivatives, S-methyl and S-ethyl-GSH, and tripeptides lacking the cysteine sulfur atom can substitute for GSH in enhancing the transport of some substrates (14,17).
Derivatives with longer alkyl chains fail to stimulate transport of GSH-dependent substrates and may inhibit transport of some GSH-independent substrates. Consequently, prior to using azidophenacyl-GSH for photolabeling studies, we first ascertained whether it could substitute functionally for GSH. Our studies show that the compound stimulates the transport of estrone sulfate and that it can also substitute for GSH in enhancing photolabeling of MRP1 by LY475776. Thus, azidophenacyl-GSH retains the ability to stimulate binding or transport of at least these two compounds. The photolabeling of intact MRP1 by azidophenacyl-[ 35 S]GSH was also efficiently blocked by the glutathione-conjugated substrate, LTC 4 , indicating that the labeling observed was specific.
Our data demonstrate that the photolabeling profile of azidophenacyl-GSH is similar to that of LTC 4 and that the compound labels sites in both halves of the protein (27). Although labeling of the two locations by both compounds requires the presence of CL3, neither photolabel this region. Consequently, it remains to be determined whether CL3 interacts directly with either GSH or the GSH moiety of LTC 4, as proposed (36). CL3 may have a more general function in maintaining the architecture of the protein, because its removal abolishes trafficking of MRP1 in mammalian cells and eliminates binding and/or transport of GS-conjugates, GSH-dependent substrates and other conjugated substrates such as E 2 17␤G (27,34,39,54).
Despite the extreme GSH dependence of photolabeling of MRP1 by [ 125 I]LY475776 and its preferential binding to the COOH-proximal site, no reciprocal enhancement of photolabeling by azidophenacyl-[ 35 S]GSH was observed. In addition, we detected no stimulation of GSH transport by LY475776. These and other observations are consistent with the possibility that the interaction of GSH or a derivative with MRP1 is a prerequisite for binding of Ly465776. They also suggest that the binding of LY475776, unlike estrone 3-sulfate, does not cooperatively enhance the binding of GSH. Thus, LY475776 may bind with high affinity to a conformation of the protein induced by interaction of GSH with one or both sites photolabeled by azidophenacyl-GSH. The exceptionally high apparent affinity of LY475776 for MRP1 in the presence of GSH, and its lack of reciprocal stimulation of the binding of azidophenacyl-GSH, may indicate that it binds without inducing any further change in the conformation of the protein.
In contrast to LY475776, estrone sulfate can be transported by MRP1 in the absence of GSH, although relatively poorly, and thus alone can clearly bind to the protein (16). The presence of GSH decreases the K m and increases the V max for estrone sulfate transport 3-4-fold (16). Studies presented here using dual-expressed half-molecules indicate that photolabeling by azidophenacyl-[ 35 S]GSH, particularly of the COOHproximal site, is enhanced in the presence of estrone sulfate. We have shown previously that S-methyl-GSH preferentially inhibits LTC 4 labeling of the NH 2 -proximal site in MRP1, whereas low concentrations weakly stimulate labeling of the COOH-proximal site (27). In addition, low concentrations of estrone sulfate also stimulate LTC 4 transport (16). These observations suggest that initial binding of GSH or a substrate such as estrone sulfate to the NH 2 -proximal site facilitates binding to the COOH-proximal site. The fact that trapping of the protein in a transition state selectively decreases binding of LTC 4 to the NH 2 -proximal site is also consistent with this suggestion (27).
We have established that among MRP homologs LY475776 is highly specific for MRP1. 2,3 Studies described here reveal that the compound also binds with 5-6-fold higher affinity to MRP1 when compared with its highly conserved murine ortholog, and that labeling was enhanced in hybrid proteins containing all or part of MSD3 from the human protein. These findings are consistent with very recent partial trypsinolysis studies indicating that LY475776 preferentially photolabels a proteolytic fragment encompassing TM helices 16 and 17 (36). Weak labeling of two additional peptides was also detected in these studies. Based on their size, the fragments may contain TM helices 12 and 13 and 14 and 15. Studies with MRP1/mrp1 hybrid proteins suggest that the latter region, although not strongly photolabeled, is important for high affinity binding of LY475776. Previously, we have shown that a non-conserved amino acid in this region, Glu-1089, in predicted TM14 of MRP1 is essential for the ability of the human protein to confer anthracycline resistance (28). This residue, which is replaced by glutamine in mrp1, is also required for high affinity binding of LY475776. Comparative studies of MRP1 and murine mrp1 have also identified a second non-conserved residue in TM17, Thr-1242 in MRP1, that is required for efficient transport of E 2 17␤G (30). Mutation of Thr-1242 to Ala, as found at the corresponding position in murine mrp1, decreases the ability to transport E 2 17␤G and to confer drug resistance, but does not affect LTC 4 transport. This mutation in the human protein also decreases photolabeling by LY475776, but as observed with the mutation in TM14, has no effect on photolabeling by azidophenacyl-[ 35 S]GSH. Thus, binding of LY475776 shares a requirement for some residues that are involved in the binding and or transport of chemotherapeutic drugs and some conjugated steroids, but which appear not to be involved in the binding of the two GSH conjugates, LTC 4 and azidophenacyl-GSH. Whether there is a specific subset of residues involved in the binding of GSH-conjugates that is distinct from those that contribute to the binding of other conjugated and non-conjugated substrates is not yet known.