Multidrug Resistance Protein

Multidrug resistance protein (MRP) is a broad specificity, primary active transporter of organic anion conjugates that confers a multidrug resistance phenotype when transfected into drug-sensitive cells. The protein was the first example of a subgroup of the ATP-binding cassette superfamily whose members have three membrane-spanning domains (MSDs) and two nucleotide binding domains. The role(s) of the third MSD of MRP and its related transporters is not known. To begin to address this question, we examined the ability of various MRP fragments, expressed individually and in combination, to transport the MRP substrate, leukotriene C4 (LTC4). We found that elimination of the entire NH2-terminal MSD or just the first putative transmembrane helix, or substitution of the MSD with the comparable region of the functionally and structurally related transporter, the canalicular multispecific organic anion transporter (cMOAT/MRP2), had little effect on protein accumulation in the membrane. However, all three modifications decreased LTC4 transport activity by at least 90%. Transport activity could be reconstituted by co-expression of the NH2-terminal MSD with a fragment corresponding to the remainder of the MRP molecule, but this required both the region encoding the transmembrane helices of the NH2-terminal MSD and the cytoplasmic region linking it to the next MSD. In contrast, a major part of the cytoplasmic region linking the NH2-proximal nucleotide binding domain of the protein to the COOH-proximal MSD was not required for active transport of LTC4.

transporters (1,2). When transfected into drug-sensitive recipient cells, MRP increases resistance to natural product chemotherapeutic agents including epipodophyllotoxins, Vinca alkaloids, and certain anthracyclines (3)(4)(5)(6). The resistance of MRPtransfected cells is associated with an energy-dependent decrease in drug accumulation and an increase in drug efflux (4,6). Thus MRP confers a form of multidrug resistance that shares several characteristics with that caused by overexpression of P-glycoprotein (P-gp).
MRP can also act as a primary active transporter of a wide range of organic, anionic conjugates, some of which are potential physiological substrates (reviewed in Loe et al. (7)). These include a structurally diverse array of glutathione, glucuronide, and sulfate conjugates, with the cysteinyl leukotriene, LTC 4 , being one of the highest affinity MRP substrates characterized to date (8 -11). However, it has not been possible to detect binding or measure primary active transport of unmodified xenobiotics (8,9,12). We have been able to demonstrate MRP-mediated, ATP-dependent transport of vincristine, and more recently, aflatoxin B 1 , but only in the presence of physiological concentrations of GSH (8,13). P-gp (encoded by the MDR1 gene) and MRP share only approximately 18% overall amino acid identity, but despite the lack of primary structure conservation, the hydropathy profiles of the two proteins are quite similar when they are aligned from their COOH termini (14). Like other members of the ABC-superfamily such as the cystic fibrosis transmembrane conductance regulator (CFTR), the nucleotide binding domains (NBDs) of MRP are preceded by a polytopic membrane spanning domain (MSD) that may contain as many as six transmembrane (TM) helices. However, MRP also contains a third, NH 2 -proximal MSD. Topology predictions derived from various protein structure algorithms indicate that this region, MSD1, could span the membrane four to six times (15). Mutagenesis of potential N-glycosylation sites has recently shown that the NH 2 terminus of MRP is extracellular (15). Thus it appears most likely that MSD1 spans the membrane five times. A number of ABC transporters relatively closely related to MRP have now been described that also contain a third NH 2 -proximal MSD (16 -20). Analyses of protein primary structure and gene organization suggest that members of the MRP-related branch of the superfamily may have evolved from a common ancestor that they shared with CFTR (21,22). Thus the additional NH 2 -terminal MSDs of the MRP-related proteins may have been acquired by fusion of a common ancestral gene with one or more genes encoding other membrane proteins. Whether the NH 2 -terminal MSDs of these proteins are now essential for function or play some type of regulatory role is not known.
We demonstrated previously that a MRP fragment containing the two NH 2 -proximal MSDs (MSD1 and MSD2), their associated NBD (NBD1), and most of the region linking this segment of the protein to the COOH-proximal MSD (MSD3) had no demonstrable LTC 4 transport activity. However, a functional transporter could be reconstituted by co-expressing this fragment with another containing MSD3 and NBD2 (23). Here, we have used a similar approach to determine whether molecules lacking all or part of MSD1 and the predicted cytoplasmic loop linking it to the rest of the protein can function as transporters, and whether this domain can reassociate with a COOH-proximal MRP fragment when the two are co-expressed. We have also investigated the possibility of exchanging MSD1 of MRP for that of the related canalicular multispecific organic anion transporter, cMOAT/MRP2 (16). In our initial study, the co-expressed fragments corresponded to those that would be generated by cleaving the molecule at Ala 932 , in the cytoplasmic region linking NBD1 to MSD3. Thus although not physically contiguous, the entire linker region between the two halves of the molecule was present in the reconstituted protein (23). We have now determined whether most of this region, which in CFTR and certain other ABC transporters has been reported to have a regulatory function, is required for the LTC 4 transport activity of MRP.

MATERIALS AND METHODS
Generation of Constructs-The expression cassette pBSMRP-fc-ATG and recombinant donor plasmid pFB-MRP, containing the MRP coding sequence, have been described (23). To generate constructs expressing either MSD1 or the remainder of the protein, pBSMRP-fc-ATG was linearized with BamHI, made blunt-ended using a Klenow fragment, and then digested with SacI and KpnI. This digestion generates one fragment (SacI-BamHI*) containing the sequence encoding amino acids 1-281 of MRP and another (BamHI*-KpnI) encoding the remainder of the protein. The SacI-BamHI* fragment was ligated to the donor plasmid pFASTBAC1 (Life Technologies Inc.) which had been linearized with HindIII, made blunt-ended using Klenow fragment, and then digested with SacI to produce the MRP 1-281 vector. Translation of the inserted fragment terminated at a stop codon in the vector resulting in the addition of six amino acids, QLVEKY.
To generate a construct lacking the NH 2 -proximal 280 amino acids, pFB-MRP was linearized with NcoI, made blunt-ended using Klenow fragment, and then digested with KpnI. The vector backbone that contains the Kozak sequence and initiator methionine originally introduced into pFB-MRP was ligated to the BamHI*-KpnI fragment. The only amino acid introduced during construction of the MRP 281-1531 vector was the initiator methionine.
To create a similar construct encoding MRP 229 -1531 , a fragment containing nucleotides 685-1120 was amplified by PCR. The forward primer used introduced a SacI site preceding an in-frame start codon. The PCR fragment was digested with SacI and BamHI, and the SacI-BamHI fragment was isolated. pBSMRP-fc-ATG was also digested with SacI and BamHI, leaving the 3Ј-proximal sequence beginning at nucleotide 840 attached to the vector backbone. The purified digested vector was ligated to the SacI-BamHI fragment of the PCR product, the fidelity of which was verified by dideoxy sequencing of the modified construct.
To create a construct eliminating amino acids 228 -280 of the protein (MRP ⌬(228 -280) ), a fragment containing nucleotides 1-681 along with the 5Ј UTR of pBSMRP-fc-ATG was amplified by PCR. M 13 -20 primer was used in the forward direction and the reverse primer used was 5Ј-CAGGGATCCAACCCTGTGATCCACCAGA-3Ј. The PCR product was digested with SacI and BamHI and the recovered SacI-BamHI fragment was ligated to a MRP vector as described for MRP 229 -1531 .
To eliminate the first TM helix of the protein, the region between nucleotides 199 and 888 of the MRP coding sequence was amplified by PCR from a recombinant plasmid containing a full-length MRP cDNA. The primers used were 5Ј-CACGAGCTCGCCGCCATGACACCTCT-CAACAAAACC-3Ј (forward primer) which includes a SacI site (underlined) and a consensus Kozak sequence (bold), and 5Ј-CTCCTCATTCG-CATCCACCTTGG-3Ј (reverse primer). The 705-base pair PCR product was digested at the SacI site in the primer and at the BamHI site in the MRP coding sequence to yield a 652-base pair fragment. This fragment was ligated into a MRP vector from which the SacI/BamHI fragment had been removed. Fidelity of the product was confirmed by dideoxy sequencing. Digestion of the modified vector with SacI and KpnI yielded a fragment of 4.3 kilobase pairs encoding amino acids 67-1531, which was cloned into pFASTBAC1.
A similar strategy was used to create a hybrid protein containing the NH 2 -terminal 280 amino acids of cMOAT/MRP2 fused to amino acids 281-1531 of MRP. The cMOAT/MRP2 fragment was created by reverse transcriptase PCR of poly(A) ϩ RNA isolated from the human hepatoma cell line HepG2 using an upstream primer that included a perfect Kozak sequence preceded by a HindIII site and a downstream primer that included a BamHI site. The sequences of the primers were: (forward) 5Ј-CACAAGCTTGCCGCCATGCTGGAGAAGTTCTGCAAC-3Ј, (reverse) 5Ј-CACGGATCCTGATTCTTGTTCAAGCCAGG-3Ј. The fidelity of the PCR product was verified by dideoxy sequencing. It was then co-ligated with the BamHI/KpnI fragment of MRP cDNA into pBluescript II KSϩ that had been digested with KpnI and HindIII. The fusion fragment was then transferred to pFASTBAC1 as an XbaI/KpnI fragment.
Construct MRP 1-858 was created by subcloning from a derivative of the previously described pCEBV7-MRP1 vector (6) which was linearized with XhoI and religated to generate pCEBV7-MRP 1-858 . The SacI-BamHI fragment of pBSMRP-fc-ATG, containing a Kozak sequence and an initiator methionine codon, was isolated and cloned into the vector pGEM-3Zf(ϩ) (Promega). The resulting vector was linearized with BamHI and ligated to the BamHI fragment isolated from pCEBV7-MRP  . Orientation of the BamHI fragment was determined by restriction enzyme digestion. This cassette produces a polypeptide containing amino acids 1-858 of MRP with seven additional amino acids, GRQGRIL.
Viral Infection and Membrane Vesicle Preparation-Recombinant bacmids and baculoviruses were generated as described elsewhere (23). Conditions used for viral infection were similar to those described previously, except that the multiplicity of infection used for each coexpressed construct was adjusted to achieve comparable levels of the two MRP fragments (23). Cells were disrupted by nitrogen cavitation to generate membrane vesicles that were purified by sucrose gradient centrifugation, as described previously (8).
Vesicle Transport of LTC 4 -Uptake of [ 3 H]LTC 4 (50 nM, 132.0 Ci mmol Ϫ1 , NEN Life Science Products) into membrane vesicles was measured at 23°C in the presence of ATP or AMP (4 mM) using a rapid filtration technique, as described elsewhere, with modifications (8,27).
To determine kinetic parameters such as K m and V max , initial rates of LTC 4 uptake were measured for 60 s at various substrate concentrations. Data from initial rate measurements were fitted to the Michaelis-Menten type equation by an iterative nonlinear least-squares method using a MULTI program with a weight of 1/y 2 (28).

Generation of Recombinant
Baculoviruses-A possible topology of MRP derived from computer-assisted hydropathy analyses (7,14,15) and locations of known N-glycosylation sites (15) is shown in Fig. 1. Sf21 cells were transfected with bacmids encoding the regions of MRP indicated (Fig. 1). Whole cell lysates of the transfected cells were then analyzed by SDS-PAGE and immunoblotting using various MRP-specific antibodies (data not shown). Viral stocks were then prepared and their titers determined, as described previously (23).
MSD1 Is Required for Efficient LTC 4 Transport-The cytoplasmic loop connecting MSD1 to MSD2 is predicted to extend from approximately Phe 191 to Thr 320 (Fig. 1). To assess whether MSD1 is required for LTC 4 transport, MRP fragments extending from either Val 229 or Asp 281 to the COOH terminus of the protein were expressed in Sf21 insect cells. SDS-PAGE and immunoblotting of membrane vesicle proteins revealed that MRP 229 -1531 and MRP 281-1531 were expressed efficiently ( Fig. 2A), at levels slightly higher than full-length MRP in comparable vesicle preparations. However, ATP-dependent LTC 4 uptake was approximately 4.0 pmol min Ϫ1 mg Ϫ1 vesicle protein (at 23°C and 50 nM LTC 4 ) (Fig. 2B). This is only 2-3-fold higher than observed with vesicles from cells infected with a control virus encoding ␤-glucuronidase and less than 10% of the rates of uptake typically obtained with vesicles containing comparable levels of full length MRP (data not shown).
Elimination of the First Putative TM Helix of MSD1 Has Little Effect on Accumulation of MRP in the Membrane but Markedly Decreases LTC 4 Transport-To determine whether elimination of just the first putative TM helix of MSD1 influenced accumulation of the protein in the membrane and/or transport activity, we constructed a vector encoding a polypeptide lacking amino acids 1-66 of MRP (MRP 67-1531 ). Immunoblotting of membrane vesicle proteins indicated that MRP 67-1531 accumulated to levels approaching those of full length MRP ( Fig. 2A). However, initial rates of ATP-dependent LTC 4 uptake were very similar to those observed following deletion of the entire MSD1 (Fig. 2B).
Co-expression of MRP 1-281 with either MRP 229 -1531 or MRP 281-1531 Restores ATP-dependent Transport of LTC 4 -To investigate the possibility that co-expression of MSD1 with the remainder of the protein could reconstitute a functional transporter, Sf21 cells were co-infected with constructs encoding MRP 1-281 and MRP 229 -1531 or MRP 281-1531 . The levels of MRP 1-281 were determined by immunoblotting using the MRPspecific mAb, MRPr1 (26). The epitope for this antibody has recently been located to amino acids 238 -247 and thus is known to be present in both MRP 1-281 and MRP 229 -1531 (Fig. 3, left). 2 The levels of MRP 229 -1531 and MRP 281-1531 were estimated with mAb QCRL-1, the epitope for which includes amino acids 918 -924 (29) (Fig. 3, right). The levels of each fragment were normalized by comparison with vesicle proteins from Sf21 cells expressing full-length MRP that were included on immunoblots probed with either mAb MRPr1 or mAb QCRL-1. The multiplicity of infection used for each vector was varied to ensure that similar levels of each of the encoded MRP fragments were obtained. This was achieved by using a 2-fold ATP-dependent LTC 4 uptake was readily detectable with membrane vesicles prepared from cells co-infected with MRP 1-281 and MRP 281-1531 (Fig. 4A). The initial rates of uptake (approximately 25 pmol min Ϫ1 mg Ϫ1 ) were 20 -30% of those obtained with membrane vesicles from cells producing similar levels of intact MRP that were prepared contemporaneously (data not shown). No change in the level of LTC 4 uptake relative to vesicles from control cells was detectable when AMP was used in place of ATP. We also examined membrane vesicles from cells co-infected with MRP 1-281 and MRP 229 -1531 . Despite the fact that these fragments overlap by 52 amino acids in the region predicted to connect MSD1 to MSD2, vesicles containing them displayed initial rates of ATP-dependent uptake of LTC 4 similar to the rate obtained with vesicles containing MRP  and MRP 281-1531 (25-30 pmol mg Ϫ1 min Ϫ1 ).
To further characterize the LTC 4 uptake by vesicles from cells co-expressing MRP 1-281 and MRP 281-1531 , rates of uptake were measured at several LTC 4 concentrations (10 -1000 nM). The apparent K m and V max values for LTC 4 were 136 Ϯ 42 nM and 102 Ϯ 20 pmol min Ϫ1 mg Ϫ1 , respectively (Fig. 4B). This K m value is similar to those reported previously for vesicles from MRP-transfected HeLa T14 cells (105 nM) (8) and HeLa T5 cells (97 nM) (10), but slightly higher than those obtained with vesicles from Sf21 cells expressing either intact MRP (67 nM) or co-expressing both half-molecules (56 nM) (23). Comparable analyses were also performed to determine an apparent K m for ATP of 136 Ϯ 24 M (Fig. 4C), a value similar to that obtained previously with intact protein (100 M) (23).
MSD1 from cMOAT/MRP2 Is Not Exchangeable with MSD1 of MRP-MRP and cMOAT/MRP2 share 49% overall amino acid identity and transport a very similar range of substrates. Comparison of hydropathy profiles suggests that the topology of MSD1 in each protein may be similar and like MRP, cMOAT/ MRP2 has a potential N-glycosylation sequon very near the NH 2 terminus. To determine whether the NH 2 -proximal MSDs of the two proteins might be exchangeable, a vector was constructed that encoded a fusion protein comprised of amino acids 1-280 from cMOAT/MRP2 and amino acids 281-1531 of MRP. Immunoblotting of membrane vesicles from infected cells indicated that the cMOAT 1-280 -MRP 281-1531 fusion accumulated to levels comparable with wild type MRP (Fig. 5A). However, ATP-dependent LTC 4 transport did not differ significantly from cells infected with a control vector (Fig. 5B). To determine if the presence of MSD1 from cMOAT/MRP2 would interfere with the ability of MSD1 of MRP to restore ATP-dependent LTC 4 transport, we co-expressed MRP 1-281 with the hybrid protein. Vesicles from co-infected cells displayed initial rates of ATP-dependent LTC 4 uptake that were very similar to those obtained with co-expressed MRP 1-281 and MRP 281-1531 (Fig.  5B).
The Cytoplasmic Region Linking MSD1 to MSD2 Is Essential for LTC 4 Transport-Results obtained with vesicles containing either MRP 1-281 and MRP 229 -1531 or MRP 1-281 and MRP 281-1531 indicated that duplication of amino acids 229 -281 in the reconstituted transporter had no major effect on its ability to transport LTC 4 . To determine if this was because the region of overlap between MRP 1-281 and MRP 229 -1531 was not critical for function, we constructed a vector encoding a protein with an internal deletion that eliminated amino acids 228 -280 (MRP ⌬ (228 -280) ). The mutant protein was expressed efficiently in Sf21 cells and accumulated to levels which are approximately one-half of those obtained with full-length MRP (Fig.  6A). However, ATP dependent LTC 4 transport in membrane vesicles containing the mutant protein did not differ significantly from the basal activity detected in membranes from control cells (data not shown).
Much of the Region Linking NBD1 to MSD3 Is Not Essential for LTC 4 Transport-Comparison of the primary structures of the human and murine orthologs suggests that the relatively poorly conserved linker region of MRP connecting NBD1 to MSD3, extends from approximately amino acid 820 to 960 (14). This region is thought to be phosphorylated primarily on serine residues, but the protein kinases involved have not been identified (24,30). To investigate the functional requirement for the linker, we co-expressed two MRP fragments, comprising amino acids 1-858 (MRP 1-858 ) and 932-1531 (MRP 932-1531 ). This eliminated the most serine-rich segment of the linker between amino acids 859 and 931. Immunoblotting of membrane proteins from cells co-infected with MRP 1-858 and MRP 932-1531 , confirmed that both fragments were efficiently expressed and accumulated to comparable levels (Fig. 6B). The initial rates of LTC 4 accumulation by vesicles from the co-infected cells were approximately 35% of those obtained with vesicles from Sf21 insect cells producing full-length MRP (Fig. 7A). This initial rate is somewhat higher than we have achieved previously by co-infection with both half-molecules that contained the entire linker region. K m and V max values for LTC 4 transport by MRP 1-858 /MRP 932-1531 vesicles were determined to be 134 Ϯ 18 nM and 218 Ϯ 18 pmol min Ϫ1 mg Ϫ , respectively (Fig. 7B). The K m for ATP was estimated to be 27 Ϯ 3 M (Fig. 7C). This value was lower than obtained previously with vesicles from cells producing the full-length protein (100 M) or co-infected with both half-molecules (44 M) (23).

DISCUSSION
Most ABC transporters are comprised of two polytopic MSDs and two NBDs. In proteins such as the P-gps, CFTR, and STE6, all four domains are contained within a single polypeptide, but this is frequently not the case (2). For example, bacterial transporters such as the maltose and histidine permeases consist of four polypeptides each encoding a single domain (31,32), while mammalian TAP1 and TAP2, which are involved in antigen presentation, each contain one MSD and one NBD, and heterodimerize to form a functional complex (33,34). ABC transporters, such as STE6, CFTR, and P-gp, are thought to have evolved by duplication or fusion of genes encoding half-transporters with domain organizations similar to those of TAP1 and TAP2. This suggestion is supported by studies in which coexpression of half molecules of transporters such as STE6, CFTR, and P-gp has been shown to reconstitute some level of function (35)(36)(37).
MRP was the first example of an ABC transporter with a predicted topology inconsistent with the typical four-domain structure. More recently identified ABC transporters with a third NH 2 -terminal MSD include cMOAT/MRP2, the yeast cadmium resistance factor, and the sulfonylurea receptors, as well as several less well characterized MRP-related proteins (16,18,20). Comparisons of protein primary structure and gene organization suggest that the MRP-related proteins shared a common, four-domain ancestor with CFTR (21). It is also apparent from such comparisons that the NH 2 -terminal MSDs of the MRP-related proteins are poorly conserved relative to the other four domains in these proteins (15,22). These observations raise the question of whether or not the additional MSD is essential for transport activity, or for some other function of the protein. Our results demonstrate that two truncated molecules (MRP 229 -1531 and MRP 281-1531 ) lacking MSD1 can be expressed in Sf21 cells as efficiently as the full-length protein.
However, ATP-dependent transport of LTC 4 by the mutant was less than 10% of that obtained with vesicles containing comparable amounts of intact protein.
If transporters such as MRP evolved by fusion of a common ancestral gene with one or more genes encoding other integral membrane proteins, the additional MSDs might be expected to be capable of folding independently and of being appropriately inserted in cell membranes. In support of this suggestion, we found that a polypeptide corresponding to amino acids 1-281 is efficiently produced and incorporated into membranes. In addition, when this fragment was co-expressed with four-domain fragments lacking MSD1, it increased the ability to transport LTC 4 5-6-fold. The kinetic parameters of the reconstituted transporter were also similar to those of the intact protein.
These observations confirm the importance of MSD1 for transport activity. They also strongly suggest that MSD1 is not simply tethered to the rest of the molecule by the polypeptide backbone, but must be capable of forming stable interactions with other protein domains.
The ability to reconstitute transport activity by co-expressing the 1-281 fragment with a partially overlapping fragment extending from amino acid 229 to the COOH terminus, raised the possibility that the region of overlap in the predicted cytoplasmic linker between MSD1 and MSD2 may not be critical for function. However, elimination of the overlap by creating an internal deletion in MRP (MRP ⌬(228 -280) ) also eliminated LTC 4 transport activity. This suggests either that the region between amino acid 228 and 280 may be involved in LTC 4 binding and/or transport, or that shortening of the loop by 53 amino acids may have altered the topological relationships between MSD1 and other, essential regions of the protein. To try to distinguish between these two possibilities, we co-expressed fragments extending from amino acid 1-227 and 281-1531. This combination of fragments failed to reconstitute LTC 4 transport (data not shown). However, suitable antibodies for determining the level of expression of the 1-227 fragment are currently not available and we cannot exclude the possibility that this fragment is not expressed efficiently.
In addition to their very similar substrate specificities, MRP and cMOAT/MRP2 also share considerable amino acid identity (16,38). The corresponding NBDs of MRP and cMOAT/MRP2 are approximately 70% identical while both MSD2 and MSD3 share approximately 50% identity between the two proteins. In contrast, although the predicted topologies of MSD1 of MRP and cMOAT/MRP2 are similar, their amino acid identity is only approximately 25%. This could reflect independent acquisition of the additional MSD of each protein or a relative lack of functional constraints on the primary structure of the domain. Consequently, we determined if it was possible to exchange MSD1 of MRP for the comparable region of cMOAT/MRP2 without loss of function. Although the hybrid protein could be expressed at high levels in Sf21 cells, it had no detectable LTC 4 transport activity. Interestingly, the presence of MSD1 from cMOAT/MRP2 in the fusion protein, did not interfere with the restoration of transport activity when MSD1 from MRP was co-expressed with the cMOAT 1-280 -MRP 281-1531 hybrid. Given the very similar substrate specificities of the two proteins, these observations suggest that there may be a requirement for specific interactions between MSD1 and other domains of MRP, that cannot be fulfilled by MSD1 of cMOAT/MRP2. MRP has the characteristics of a type IIIb polytopic protein (reviewed in Singer (39)). These proteins are typified by a short NH 2 -terminal extracellular sequence without a cleavable signal peptide and an NH 2 -terminal TM segment that is often followed by a region of relatively positive charge. In MRP, both Asn 19 and Asn 23 are glycosylated and amino acids 1-37 have a net negative charge (15). Thus they are predicted to be on the outside of the membrane while all five charged residues in the first predicted cytoplasmic loop of 21 amino acids are positively charged. In some type IIIb proteins, the NH 2 -terminal TM segment has been shown to be important for efficient insertion into the membrane and for determining the orientation of subsequent TM segments (39). In others, the first TM segment appears to play a passive role (39,40). Our results suggest that this is the case with MRP. We found that removal of the first TM segment and approximately half of the first predicted cytoplasmic loop did not prevent the protein from accumulating in the membrane to levels similar to those of full length MRP. However, elimination of this single TM decreased LTC 4 transport activity to the same extent as removal of all five predicted TMs in MSD1.
The cytoplasmic region connecting NBD1 to the COOH-prox-imal MSD of ABC-proteins such as P-gp and CFTR is highly variable among superfamily members and is relatively poorly conserved among subfamily isoforms (41)(42)(43). It is also frequently post-translationally modified by phosphorylation of serine residues (24, 30, 44 -46). In CFTR, this region forms the "R" domain which mediates cAMP dependent regulation of the chloride channel (45,46). In contrast, phosphorylation status has no apparent effect on the drug transport activity of P-gp (47), although it may influence other functions of the protein (48). Before examining the possible functional implications of phosphorylation of the comparable region of MRP, we investigated whether the region is actually required for ATP-dependent transport. Because creation of a large internal deletion in the linker region could potentially alter interactions between the two NBDs and possibly other domains of the protein, we examined the possibility of reconstituting LTC 4 transport by co-expressing fragments, 1-858 and 932-1531. This eliminated 74 amino acids that included the most serine rich and divergent portion of the linker. The co-expressed fragments actively transported LTC 4 with a K m that was not significantly different from that obtained previously for co-expressed fragments 1-932 and 932-1531 and a V max that was approximately two fold greater. Thus this part of the linker is clearly not essential for transport and presumably is not necessary for interaction between the two halves of the protein. The only difference observed between the transport activity reconstituted from fragments 1-858 and 932-1531 and that of intact MRP was an approximate four fold decrease in the K m for ATP. Whether this indicates that the linker region and/or modifications to it might influence the affinity or catalytic activity of the NBDs remains to be determined.
The results of the present study demonstrate the importance of MSD1 and the first TM segment of MRP for active transport of LTC 4 . The co-expression experiments also suggest that this domain can form stable, functional interactions with other regions of the protein. Studies are in progress to localize the sites of interaction and to determine how this additional domain is integrated into the evolutionarily conserved four domain structure typical of the majority of ABC transporters.