Characterization of the Transport Properties of Cloned Rat Multidrug Resistance-associated Protein 3 (MRP3)*

We have previously cloned rat MRP3 as an inducible transporter in the liver (Hirohashi, T., Suzuki, H., Ito, K., Ogawa, K., Kume, K., Shimizu, T., and Sugiyama, Y. (1998) Mol. Pharmacol. 53, 1068–1075). In the present study, the function of rat MRP3 was investigated using membrane vesicles isolated from LLC-PK1 and HeLa cell population transfected with corresponding cDNA. The ATP-dependent uptake of both 17β estradiol 17-β-d-glucuronide ([3H]E217βG) and glucuronide of [14C] 6-hydroxy-5,7-dimethyl-2-methylamino-4-(3-pyridylmethyl) benzothiazole (E3040), but not that of [3H]leukotriene C4 and [3H]2,4-dinitrophenyl-S-glutathione, was markedly stimulated by MRP3 transfection in both cell lines. TheK m and V max values for the uptake of [3H]E217βG were 67 ± 14 μm and 415 ± 73 pmol/min/mg of protein, respectively, for MRP3-expressing membrane vesicles and 3.0 ± 0.7 μm and 3.4 ± 0.4 pmol/min/mg of protein, respectively, for the endogenous transporter expressed on HeLa cells. [3H]E217βG had also a similarK m value for MRP3 when LLC-PK1 cells were used as the host. All glucuronide conjugates examined (E3040 glucuronide, 4-methylumbelliferone glucuronide, and naphthyl glucuronide) and methotrexate inhibited MRP3-mediated [3H]E217βG transport in LLC-PK1 cells. Moreover, [3H]methotrexate was transported via MRP3. The inhibitory effect of estrone sulfate, [3H]2,4-dinitrophenyl-S-glutathione, and [3H]leukotriene C4 was moderate or minimal, whereas N-acetyl-2,4-dinitrophenylcysteine had no effect on the uptake of [3H]E217βG. The uptake of [3H]E217βG was enhanced by E3040 sulfate and 4-methylumbelliferone sulfate. Thus we were able to demonstrate that several kinds of organic anions are transported via MRP3, although the substrate specificity of MRP3 differs from that of MRP1 and cMOAT/MRP2 in that glutathione conjugates are poor substrates for MRP3.

Although MRP1 is widely expressed in many somatic cells, its hepatic expression is not marked. In the liver, canalicular multispecific organic anion transporter (cMOAT/MRP2), another member of the GS-X pump family, is expressed on the bile canalicular membrane, mediating the efficient biliary excretion of many organic anions (3)(4)(5). The similar substrate specificity of cMOAT/MRP2 and MRP1 has been established by comparing the transport properties across the bile canalicular membrane between normal rats and mutant rats whose cMOAT/MRP2 activity is hereditarily defective (e.g. transport-deficient (TR Ϫ ) rats and Eisai hyperbilirubinemic rats (EHBR)) (3)(4)(5). These mutant rats have been used as an animal model for Dubin-Johnson syndrome found in humans (6 -14). cDNA cloning and functional analysis of its product, along with a mutation analysis (15,16), have been performed in this and other laboratories.
It is possible that transporters other than cMOAT/MRP2 may be also involved in the hepatic transport of organic anions. Indeed, we were able to amplify two kinds of novel transporters, which were initially referred to as MRP-like protein (MLP) 1 and 2 from EHBR liver using RT-PCR with the degenerated primers designed for the highly conserved carboxyl-terminal ABC region of human MRP1 (17). The sequence alignment of the full-length of cDNA indicated that MLP-1 and 2 correspond to MRP6 and 3, respectively (17,18). Northern blot analysis showed that the hepatic expression of MRP3 was significantly enhanced in EHBR compared with Sprague-Dawley (SD) rats, although the extent of MRP6 expression is comparable in the two rat strains (17). It is possible that MRP3 may compensate for the defective expression of cMOAT/MRP2 (17,19). In addition, because Northern blot analysis indicated that there was extensive expression of MRP3 in intestinal tissues of rats and human, it is possible that MRP3 is also responsible for the intestinal excretion of organic anions (17,20).

Materials-[
Preparation of Transfected Cells--Rat MRP3 cDNA, excised with EcoRI from pBluescript II SK(Ϫ) vector (17), was then inserted into the EcoRI site in a mammalian expression vector (pCXN2; supplied by Dr. J. Miyazaki, Osaka University, School of Medicine) (24). After transfection with LipofectAMINE (Life Technologies, Inc.), the LLC-PK1 and HeLa cells were maintained in the presence of 800 g/ml G418 (Geneticin, Life Technologies) to obtain colonies. Three weeks later, the expression level of MRP3 was determined in several independently transfected cell populations using Northern blot analysis. Membrane vesicles were prepared from the cell population exhibiting the highest MRP3 expression.
The cDNA fragment containing the carboxyl-terminal ABC region of rat MRP3 was prepared to undergo Northern hybridization according to the method described previously (9,17). Filters were exposed to Fuji imaging plates (Fuji Photo Film Co., Ltd., Kanagawa, Japan) for 3 h at room temperature and analyzed with an imaging analyzer (BAS 2000, Fuji Photo Film Co., Ltd.).
Transport Studies with HeLa Membrane Vesicles-Membrane vesicles were prepared from 1 ϫ 10 8 of the MRP3-and vector-transfected HeLa and LLC-PK1 cell populations as described previously (25). The membrane vesicles were frozen in liquid nitrogen and then transferred to a freezer (Ϫ100°C) until required.
The transport studies were performed using the rapid filtration technique (26). Briefly, 16 l of transport medium (10 mM Tris-HCl, 250 mM sucrose, 10 mM MgCl 2 , pH 7.4) containing radiolabeled compounds with or without unlabeled substrate was preincubated at 37°C for 3 min and then rapidly mixed with 4 l of membrane vesicle suspension (8ϳ10 g of protein). The reaction mixture contained 5 mM ATP or 5 mM AMP and ATP-regenerating system (10 mM creatine phosphate, 100 g/ml creatine phosphokinase). In some instances, the membrane vesicles were treated with acivicin (final concentration of 6 mM) at 25°C for 15 min before initiation of the transport studies to avoid possible degradation of glutathione-S-conjugates by ␥-glutamyltranspeptidase. Acivicin did not affect MRP3-mediated transport of E 2 17␤G. The transport reaction was stopped by the addition of 1 ml of ice-cold buffer containing 250 mM sucrose, 0.1 M NaCl, 10 mM Tris-HCl (pH 7.4). The stopped reaction mixture was filtered through a 0.45-m membrane filter (GVWP; Millipore Corp., Bedford, MA) and then washed twice with 5 ml of stop solution. Radioactivity retained on the filter was determined using a liquid scintillation counter (LSC-3500, Aloka Co., Tokyo, Japan). The ATP-dependent uptake of ligands was calculated by subtracting the ligand uptake in the absence of ATP from that in the presence of ATP.

Uptake of Glucuronides and Glutathione Conjugates into
Membrane Vesicles-The expression of rat MRP3 in the transfected cells was confirmed by Northern blot analysis (Fig. 1). The length of the transcript in MRP3-transfected LLC-PK1 and HeLa cells was comparable with the band observed in Sprague-Dawley rat colon (17) (Fig. 1) Fig. 2. The ATP-dependent uptake of [ 3 H]E 2 17␤G and [ 14 C]E3040 glucuronide at 10 min was 4.1-fold and 6.9-fold higher in MRP3-transfected LLC-PK1, respectively, compared with the control vector-transfected LLC-PK1 (Fig. 2, a and b). In contrast, the ATPdependent uptake of [ 3 H]DNP-SG and [ 3 H]LTC 4 was not stimulated by MRP3 transfection (Fig. 2, c and d). The same results were obtained in membrane vesicles from MRP3-transfected HeLa cells (Fig. 3). These results indicate that MRP3 preferentially accepts these two glucuronides as substrates, whereas these two glutathione conjugates are poor substrates of MRP3.

Transport Kinetics of [ 3 H]E 2 17␤G--
The ATP-dependent uptake of [ 3 H]E 2 17␤G into membrane vesicles was saturable (Fig.  4). Nonlinear regression analysis of the uptake by MRP3-and control vector-transfected LLC-PK1 cells revealed that the saturable uptake can be described by a K m of 110 Ϯ 20 M and V max of 1570 Ϯ 300 pmol/min/mg of protein and a K m of 13 Ϯ 6 M and V max of 35 Ϯ 11 pmol/min/mg of protein, respectively (Fig. 4a). In MRP3-and control vector-transfected HeLa cells, kinetic parameters for the saturable uptake of [ 3 H]E 2 17␤G were determined to be K m ϭ 67 Ϯ 14 M and V max ϭ 415 Ϯ 73 pmol/min/mg of protein and K m ϭ 3.0 Ϯ 0.7 M and V max ϭ 3.4 Ϯ 0.4 pmol/min/mg of protein, respectively (Fig. 4b).
Characterization of MRP3-mediated Transport of [ 3 H] E 2 17␤G- Table I (Table I). In contrast, E3040 sulfate and 4-methylumbelliferone sulfate enhanced the uptake of    (Table I).
Because MTX inhibited MRP3-mediated transport of [ 3 H]E 2 17␤G, the transport of [ 3 H]MTX was examined in membrane vesicles isolated from LLC-PK1 cells. The uptake of MTX was significantly enhanced by transfection of MRP3 (Fig. 5), indicating that MTX is a substrate for MRP3. DISCUSSION In the present study, we examined the function of the recently cloned rat MRP3 (17) by using membrane vesicles prepared from LLC-PK1 and HeLa cell populations transfected with the corresponding cDNA. The ATP-dependent uptake of glucuronide conjugates (E 2 17␤G and E3040 glucuronide), but not that of glutathione conjugates (DNP-SG and LTC 4 ), was markedly stimulated by MRP3 transfection (Figs. 2 and 3), indicating that MRP3 accepts the glucuronide conjugates much more readily as substrates than the glutathione conjugates. In addition, the transport characteristics of MRP3 are the same if this transporter is expressed on HeLa or LLC-PK1 cells (Figs. 2 and 3, Table I).
The transport characteristics of MRP3 differ from those of MRP1 and cMOAT/MRP2 in that LTC 4 is a much better substrate than E 2 17␤G for the latter two transporters. In MRP1transfected HeLa cells, the V max /K m for the ATP-dependent uptake was 1030, 110, and 42 l/min/mg of protein for LTC 4 , DNP-SG, and E 2 17␤G, respectively (27,28). This result is consistent with the finding that the V max /K m value for the ATP-dependent uptake of LTC 4 was 30 times higher than that of E 2 17␤G in MRP1-transfected HEK cells (29). For murine mrp1, this value for LTC 4 was 200 times higher than that of E 2 17␤G in the transfected HEK cells (29). In rat CMVs, the V max /K m for the ATP-dependent uptake of LTC 4 , DNP-SG, and E 2 17␤G was 268, 58, and 34 ml/min/mg of protein, respectively, indicating that the clearance for uptake mediated by cMOAT/ MRP2 is 8 times higher for LTC 4 than E 2 17␤G. 2 Kinetic analysis revealed that the K m value for MRP3 was 110 M and 67 M in MRP3-transfected LLC-PK1 and HeLa cells, respectively (Fig. 4), which is much higher than that reported for MRP1transfected HeLa and HEK cells (1.5-4.8 M) (28 -30) and in CMVs from SD rats (6.3 M) 3 .
We also examined the inhibitory effect of several anionic compounds on the MRP3-mediated transport of E 2 17␤G (Table   I). LTC 4 inhibited the transport of E 2 17␤G only minimally even at 1 M, which is much higher than its K m value for MRP1 (0.1 M) (27,29,31) and cMOAT/MRP2 (0.25 M) (26). Taking this together with the results of the transport studies shown in Figs. 2 and 3, LTC 4 has, therefore, been shown to be a poor substrate for MRP3. ␣-Naphthyl ␤-D-glucuronide reduced the MRP3-mediated transport of E 2 17␤G with an IC 50 of 20-50 M (Table I), consistent with the hypothesis that this compound is also recognized by MRP3. The inhibitory effect of E3040 glucuronide (IC 50 Ͻ 5 M) and 4-methylumbelliferone glucuronide (IC 50 ϳ 50 M) was in marked contrast to the stimulatory effect of the corresponding sulfates (Table I). Although the mechanism for stimulation still remains unclear, such a stimulatory effect by E3040 sulfate and 4MU sulfate has been demonstrated in uptake of DNP-SG into CMVs from SD rats (32). Moreover, the low affinity of MRP3 toward NAc-DNP-Cys suggests that MRP3 gene may not encode MOAT4, whose transport properties had previously been characterized in mouse L1210 cells (33). It has been shown that MOAT4 mediates the low affinity transport of DNP-SG (K m ϭ 450 M) and exhibits high sensitivity toward NAc-DNP-Cys (K i ϭ 5.0 M) and ␣-naphthyl ␤-D-glucuronide (K i ϭ 8.5 M) (33). Because NAc-DNP-Cys did not affect MRP3-mediated transport, even at a concentration of 500 M, irrespective of the fact that DNP-SG can act as an inhibitor with an IC 50 of 50ϳ100 M (Table I), suggests that MOAT4 differs from MRP3.
Previously, we found that the expression of MRP3 is induced in EHBR liver (17). It is also induced in SD rat liver by phenobarbital treatment and by treatment which increases plasma bilirubin and/or its glucuronide (e.g. the cholestasis induced by common bile duct ligation and by ␣-naphthylisothiocyanate treatment) (17,19). It is plausible that, in EHBR, MRP3 is induced to compensate for the physiological function of cMOAT/ MRP2 to excrete bilirubin glucuronides from hepatocytes (17,19). This hypothesis has been proposed from the previous finding that, in mdr 1a knock-out mice, the hepatic function of mdr 1a is compensated for by the increased expression of mdr 1b, whose substrate specificity resembles that of mdr 1a (34). Although we reported that E3040 glucuronide is taken up by CMVs from EHBR in an ATP-dependent manner (32), the comparison of the transport properties between CMVs and MRP3-expressing membrane vesicles suggested that the uptake into EHBR CMVs may not be mediated by MRP3. This suggestion was proposed based on the finding that neither the uptake of E 2 17␤G 3 nor MTX (35) was stimulated by the addition of ATP if CMVs were isolated from EHBR. Together with the finding that all of the E 2 17␤G, E3040 glucuronide, and MTX are transported via MRP3 (Figs. 2, 3, and 5) suggests that the previously described ATP-dependent transport of E3040 glucuronide in CMVs from EHBR should be attributed to another transporter.
The extensive expression of MRP3 in rat and human intestinal tissues may be related to the intestinal excretion of glucuronides. Indeed, intestinal excretion of ethinylestradiol glucuronide (36) and 1-naphthol glucuronide (37) has been demonstrated in rats in in situ experiments. In addition, the extent of excretion of 1-naphthol glucuronide was comparable in normal and cMOAT/MRP2-deficient rats, suggesting the presence of another transporter responsible for the excretion of this glucuronide (38). It is possible that MRP3 is responsible for the intestinal excretion of this conjugated metabolite.
In conclusion, although MRP3 mediates the transport of several kinds of organic anions, the substrate specificity of MRP3 is in marked contrast to that of MRP1 and cMOAT/ MRP2 in that glutathione conjugates are poor substrates for MRP3. It is possible that MRP3 acts as an inducible transporter compensating for the cMOAT/MRP2 function. In addition, MRP3 may be responsible for the cellular extrusion of glucuronide conjugates in the small intestine.