Introduction

Top Summary Introduction Procedures Results Discussion References

Elimination of endogenous waste products and xenobiotics from the body is mediated by renal and hepatic transport pathways. Excretion of anionic conjugates across liver canalicular (apical) membranes into bile is mediated by the multidrug resistance-associated protein MRP2 (Müller and Jansen, 1997). Initially, this transporter was named cMOAT and characterized by using natural mutant strains of Wistar (TR) and Sprague-Dawley (EHBR) rats (Müller and Jansen, 1997). Recently, cloning of rat Mrp2 revealed that the impaired conjugate transport in canalicular membranes of these rats is caused by a premature termination of the mrp2 gene product (Paulusma et al., 1996; Ito et al., 1997). Similarly, a mutation leading to a truncated MRP2 was identified in a patient with Dubin-Johnson syndrome, a disease that resembles the TR phenotype (Paulusma et al., 1997).

Database analysis revealed that rat Mrp2 is strongly related to the human multidrug resistance-associated protein MRP1, a member of the superfamily of ABC proteins (Büchler et al., 1996; Paulusma et al., 1996; Ito et al., 1997). Originally, MRP1 was identified due to its overexpression in a multidrug-resistant cell line and its ability to confer resistance to chemotherapeutic drugs (Loe et al., 1996). Using isolated membrane vesicles from MRP1-transfected cells, it has been shown that MRP1 is also capable of transporting anionic conjugates in an ATP-dependent manner. Although MRP1 and MRP2 share substrate specificity, these transporters show differences in their tissue distribution. MRP1 is expressed, predominantly intracellularly, in numerous tissues such as lung, heart, and kidney (Flens et al., 1996). In contrast, MRP2 was detected in small intestine and apical (canalicular) membranes of hepatocytes and cells of renal proximal tubules (Büchler et al., 1996; Paulusma et al., 1996; Schaub et al., 1997).

We cloned a rabbit cDNA encoding an ABC-transporter that on basis of similar tissue distribution and highest amino acid homology was considered to be the rabbit homolog of rat Mrp2 (van Kuijck et al., 1996, 1997). On injection of its cRNAs in Xenopus laevis oocytes, we observed in a few cases a cAMP-dependent chloride conductance (van Kuijck et al., 1996). To investigate whether rabbit Mrp2 functions as an ATP-dependent organic anion transporter similar to rat Mrp2, we expressed rabbit Mrp2 in Sf9 cells using recombinant baculovirus and studied uptake of the anionic conjugates E₂17G and LTC₄ into isolated membrane vesicles. In addition, the effect of various inhibitors on Mrp2-mediated [³H]E₂17G transport was investigated.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. [14,15,19,20-³H]LTC₄ (165 Ci/mmol) and [6,7-³H]E₂17G (55 Ci/mmol) were purchased from NEN Life Science Products (Hoofddorp, The Netherlands). ATP, 5'-AMP, LTC₄, E₂17G, MTX, CsA, -naphthyl--D-glucuronide, phenolphthalein glucuronide, and p-nitrophenyl--D-glucuronide were purchased from Sigma (Zwijndrecht, The Netherlands). FL-MTX and FL were purchased from Molecular Probes (Leiden, The Netherlands). Creatine phosphate and creatine kinase were purchased from Boehringer-Mannheim (Almere, The Netherlands). CELLFECTIN and competent DH10BAC Escherichia coli cells were purchased from Life Technologies (Breda, The Netherlands). PNGase F was purchased from New England Biolabs (Westburg, Leiden, The Netherlands). MK571 was a generous gift of Dr. A. W. Ford-Hutchinson (Merck Frosst, Center for Therapeutic Research, Quebec, Canada).

Preparation of antibodies. Rabbit polyclonal antibodies were directed against two different epitopes of rabbit Mrp2 (van Kuijck et al., 1996). Antiserum k78mrp2 was obtained by immunizing rabbits with a glutathione-S-transferase fusion protein containing the 159 carboxyl-terminal amino acids (1405-1564) of Mrp2. Antiserum k51mrp2 was obtained by immunizing rabbits with a synthetic peptide (FQKRQQKKSQKNSRLQGL) corresponding to amino acids 257-274 of Mrp2 coupled to keyhole limpet hemocyanin. Rabbits were immunized with 400 µg of either the fusion protein or the synthetic peptide mixed with Freund's complete adjuvants. At 3-week intervals after priming, rabbits were boosted with 200 µg of proteins supplemented with incomplete adjuvants. Test bleedings were checked for the presence of Mrp2-specific antibodies using enzyme-linked immunosorbent assay.

Expression construct. The vector pFASTBAC1 (Life Technologies) contains an expression cassette that consists of a polyhedrin promoter, a multiple cloning site, and an SV40 poly(A)⁺ signal inserted between the left and right arms of the bacterial transposon Tn7. Cloning of a rabbit mrp2 cDNA into pFASTBAC1 was accomplished in two steps: (1) from the pBluescript KS⁺ construct pBSmrp2, which contains the entire rabbit mrp2 coding sequence (nucleotides 347-5038) (van Kuijck et al., 1996), a 2.7-kb XbaI/PstI fragment (nucleotides 2690-5407) was cloned into the XbaI and PstI sites of the multiple cloning site of pFASTBAC1 to create pFASTBAC-m1; and (2) to minimize the 5'-untranslated region, the 5' coding sequence of rabbit Mrp2 was amplified by polymerase chain reaction using the forward primer Mrp2-F1 (5'-ATGCTGGATAAGTTCTGCAAC-3'; nucleotides 347-368), which contains the ATG start codon (underlined), and the reverse primer Mrp2-R1 (5'-GCAGGAGTAGGCCAGATTAG-3'; nucleotides 844-824). The resulting polymerase chain reaction product of 498 bp was cloned into the SmaI site of pBluescript KS⁺, and its sequence was verified by dideoxy sequence analysis (Sanger et al., 1977). From this construct, a StyI/HincII fragment was removed and replaced by a StyI/EcoRV fragment (nucleotides 480-3007) from pBSmrp2. Next, a 2.5-kb BamHI fragment of this construct, containing the 5'-region of mrp2 (nucleotides 347-2868), was cloned into the BamHI site of pFASTBAC1-m1, and its orientation was determined. The selected construct, designated pFBmrp2, contains a full-length rabbit mrp2 cDNA with the ATG start codon immediately downstream of the polyhedrin promoter.

Production of recombinant baculovirus and viral infection. Baculovirus encoding rabbit Mrp2 was generated using the Bac-to-Bac baculovirus-expressing system (Life Technologies). Competent DH10BAC E. coli cells harboring a baculovirus shuttle vector (bacmid) with a Tn7 attachment site were transformed with the pFBmrp2 construct. On transposition between the Tn7 sites, recombinant bacmids were selected and isolated according to the manufacturer. Subsequently, insect Sf9 cells were transfected with recombinant bacmids using CELLFECTIN reagent. After 3 days, culture medium was collected and used to infect fresh Sf9 cells. Four days after infection, stocks of amplified virus were made. Sf9 cells (10⁶/ml) were grown as 100-ml suspension cultures and infected at a multiplicity of infection of 1-5 with recombinant baculovirus encoding Mrp2. For control experiments, Sf9 cells were infected with recombinant baculovirus encoding -glucuronidase (Life Technologies) or the -subunit of H⁺/K⁺-ATPase (Klaassen et al., 1993). Three days after infection, membrane fractions were isolated (see below).

Isolation of membrane fractions. Crude membrane fractions and membrane vesicles from infected Sf9 cells were isolated as described by Leier et al. (1994) with modifications. Briefly, cells were collected and resuspended in hypotonic buffer (0.5 mM sodium phosphate, 0.1 mM EDTA, pH 7.0) supplemented with protease inhibitors (2 mM phenylmethylsulfonyl fluoride, 5 µg/ml aprotinin, 5 µg/ml leupeptin, 1 mM pepstatin). Cells were stirred gently on ice for 90 min, and the resulting lysate was centrifuged at 100,000 × g for 40 min at 4°. The pellet of crude membranes was resuspended in TS-buffer (10 mM Tris-HEPES, 250 mM sucrose, pH 7.4) using a Potter homogenizer, and the homogenate was centrifuged at 12,000 × g for 10 min at 4°. The postnuclear supernatant was centrifuged at 100,000 × g for 40 min at 4°, and the pellet obtained was resuspended in TS-buffer with a tight-fitting Dounce (type B) homogenizer. The suspension was layered over 38% sucrose in 5 mM HEPES/KOH, pH 7.4, and centrifuged at 100,000 × g for 2 hr at 4°. The interphase was collected and homogenized on ice with a tight-fitting Dounce (type B) homogenizer, and the suspension was centrifuged at 100,000 × g for 40 min at 4°. The resulting pellet was resuspended in TS-buffer and passed through a 27-gauge needle 30 times. Membrane vesicles were frozen and stored at 80° until use. Sidedness of membrane vesicles was assessed by measuring 5'-nucleotidase activity (Doige and Sharom, 1991), and it was determined that ~65% of the vesicles were orientated inside-out.

Deglycosylation studies and immunoblot analysis. Crude membrane fractions from Sf9 cells infected with Mrp2-encoding baculovirus and from rabbit kidney were treated with PNGase F according to the manufacturer. Protein-equivalents (see figure legends) were solubilized in Laemmli's sample buffer supplemented with 100 mM dithiothreitol, heated for 10 min at 65°, subjected to SDS-polyacrylamide gel electrophoresis, and transferred to Hybond-C pure nitrocellulose membrane (Amersham, Buckinghamshire, UK) as described previously (Deen et al., 1996). Transfer of proteins was confirmed by the reversible staining of the membrane with Ponceau Red. Subsequently, the blot was blocked for 60 min with 5% nonfat dry milk powder in Tris-buffered saline supplemented with 0.3% Tween-20 (TBS-T) and washed twice with TBS-T. To detect rabbit Mrp2 proteins, the membrane was incubated overnight at 4° with antiserum k78mrp2 or k51mrp2 diluted 1:5000 in TBS-T. After two times washing for 5 min with TBS-T, the blot was blocked for 30 min as described above. The blot was then washed twice with TBS-T and incubated at room temperature for 60 min with affinity-purified horseradish peroxidase-conjugated goat anti-rabbit IgG (Sigma Immunochemicals, St. Louis, MO) diluted 1:5000 in TBS-T. Finally, the blot was washed twice for 5 min with TBS-T and TBS, respectively. Proteins were visualized using enhanced chemiluminescence (Pierce, Rockford, IL).

Transport studies in membrane vesicles. Uptake of [³H]LTC₄ into membrane vesicles was measured by using a rapid filtration technique (Leier et al., 1994). Briefly, membrane vesicles (20 µg protein-equivalent) were rapidly thawed and incubated at 37° in the presence of 4 mM MgATP, 10 mM MgCl₂, 10 mM creatine phosphate, 100 µg/ml creatine kinase, and 3 nM [³H]LTC₄ in a final volume of 120 µl of TS-buffer (10 mM Tris-HEPES, 250 mM sucrose, pH 7.4). At indicated times, 20-µl samples were taken from the reaction mixture, diluted in ice-cold TS-buffer and filtered through nitrocellulose filters (0.45-µm pore size, Schleicher & Schuell, Dassel, Germany) using a filtration device (Millipore, Bedford, MA). Filters were washed once with 5 ml of TS-buffer and dissolved in liquid scintillation fluid to determine the bound radioactivity. In control experiments, 4 mM MgATP was replaced by 4 mM 5'-AMP. Net ATP-dependent transport was calculated by subtracting values in the presence of 5'-AMP from those in the presence of ATP. Uptake of [³H]E₂17G at a final concentration of 50 nM was done similarly as described for [³H]LTC₄, except that a 50 µg protein-equivalent of membrane vesicles was used.

	Results

Top Summary Introduction Procedures Results Discussion References

Sf9 insect cells were infected with recombinant baculovirus encoding rabbit Mrp2 or control baculovirus. Crude membranes were prepared and subjected to immunoblot analysis using antiserum k78mrp2 and k51mrp2. Both antisera detected a protein of ~180 kDa in membranes from cells infected with baculovirus encoding Mrp2 (Fig. 1A, Sf9-Mrp2) but not in membranes from cells infected with control baculovirus (Fig. 1A, Sf9-c). This size is smaller than cMoat/Mrp2 detected in liver and kidney, which has been reported to have a molecular weight of ~190 kDa (Paulusma et al., 1996; Büchler et al., 1996; Schaub et al., 1997). To investigate whether this difference in molecular weight can be attributed to differences in post-translational modifications, crude membrane fractions from rabbit kidney and Sf9-Mrp2 cells were treated with or without PNGase F and analyzed by immunoblotting using antiserum k78mrp2 (Fig. 1B). Deglycosylation reduced the molecular weight of rabbit kidney Mrp2 from ~190 to 175 kDa, which is the size that can be deduced from the rabbit mrp2 cDNA sequence (van Kuijck et al., 1996, 1997). Treatment with PNGase F reduces the molecular mass of Mrp2 in Sf9 cells only slightly to 175 kDa. This indicates that in Sf9 cells, Mrp2 is less glycosylated than in rabbit kidney.

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Immunoblot analysis of Sf9 cells infected with recombinant baculovirus encoding Mrp2. A, Crude membrane fractions were isolated from Sf9 cells infected with a control baculovirus (Sf9-c) or recombinant baculovirus encoding Mrp2 (Sf9-Mrp2). Proteins (5 µg) were separated on a 6% SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and incubated with Mrp2 polyclonal antiserum k78mrp2 or k51mrp2. Proteins were visualized using enhanced chemiluminescence. Sizes of protein standards are indicated in kilodaltons. B, Protein-equivalents of crude membrane fractions from rabbit kidney (20 µg) and from Sf9 cells infected with recombinant baculovirus encoding Mrp2 (5 µg) were treated without () or with (+) PNGase F and subjected to immunoblot analysis using antiserum k78mrp2 as described in A.

To determine the expression level of rabbit Mrp2 in Sf9 cells, we subjected crude membrane fractions from Sf9-Mrp2 cells and rabbit liver and rabbit renal proximal tubular cells to immunoblot analysis using antiserum k78mrp2 (Fig. 2). A 1 µg protein-equivalent of crude membranes from Sf9-Mrp2 cells was sufficient to detect Mrp2. Approximately 20 µg protein-equivalent of crude membranes from liver and renal proximal tubules was needed to detect a similar amount of Mrp2 protein as present in 4 µg protein-equivalent of Sf9-Mrp2 crude membranes.

View larger version (46K): [in this window] [in a new window]   Fig. 2.   Comparison of Mrp2 protein levels in Sf9 cells and liver and renal proximal tubules. Protein-equivalents of crude membrane fractions isolated from Sf9-Mrp2 cells (1, 4 µg), rabbit liver (5, 20 µg), and rabbit renal proximal tubules (5, 20 µg) were separated on a 7.5% SDS-polyacrylamide gel and subjected to immunoblot analysis using antiserum k78mrp2. Proteins were visualized using enhanced chemiluminescence.

To investigate whether recombinant rabbit Mrp2 is functional, we investigated uptake of [³H]LTC₄ and [³H]E₂17G into Sf9-Mrp2 and Sf9-c membrane vesicles. Sf9-Mrp2 membrane vesicles exhibit net ATP-dependent uptake of both [³H]LTC₄ (Fig. 3, left) and [³H]E₂17G (Fig. 3, right), which was at the 2-min time point ~11-fold higher than in Sf9-c membrane vesicles. In the presence of 5'-AMP, transport of either substrate was hardly detectable in Sf9-Mrp2 membrane vesicles and was similar to that in Sf9-c membrane vesicles in the presence of 5'-AMP or ATP (not shown). Initial rates of uptake for 3 nM [³H]LTC₄ and 50 nM [³H]E₂17G were 75 and 450 fmol/mg/min, respectively.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Time course of net ATP-dependent uptake of [3H]LTC4 and [3H]E217G in Sf9-Mrp2 and Sf9-c membrane vesicles. Membrane vesicles from Sf9-c (, ) and Sf9-Mrp2 (, ) were incubated in TS-buffer (10 mM Tris·HCl, 250 mM sucrose, pH 7.4) at 37° for the times indicated. Uptake was determined for [3H]LTC4 (left) and [3H]E217G (right) at final concentrations of 3 and 50 nM, respectively. Net ATP-dependent transport was calculated by subtracting values in the presence of 4 mM 5'-AMP from those in the presence of 4 mM ATP. Data points represent the mean ± standard error of three or four determinations in a typical experiment.

To confirm that vesicle-associated increase of ligand reflects transport into a vesicular space rather than aspecific binding to the membrane, the medium osmolarity dependence of [³H]E₂17G uptake was investigated. By increasing the extravesicular sucrose concentration from 250 mM (isotonic condition) to 1000 mM, membrane vesicle space will shrink resulting in decreased uptake. As shown in Fig. 4A, initial rates of [³H]E₂17G uptake in Sf9-Mrp2 membrane vesicles decreased linearly with increasing concentrations of sucrose. Transport in Sf9-Mrp2 membrane vesicles should also be dependent on the extravesicular concentration of ATP. Fig. 4B shows that initial rates of [³H]E₂17G uptake increased with ATP concentrations according to Michaelis-Menten kinetics, yielding an apparent K_m value of 623 ± 131 µM and V_max value of 563 ± 32 fmol/mg/min.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Osmolarity dependence and effect of ATP concentration on [3H]E217G uptake in Sf9-Mrp2 membrane vesicles. Membrane vesicles from Sf9-Mrp2 cells were incubated with 50 nM [3H]E217G at 37° for 1 min. A, ATP-dependent transport was determined in the presence of sucrose concentrations ranging from 250 mM (isotonic condition) to 1000 mM. Initial rates of ATP-dependent [3H]E217G uptake were plotted against the inverse sucrose concentration in the reaction mixture. B, ATP-dependent transport was determined at various concentrations of ATP (60-8000 µM). The graph was plotted by fitting the obtained data according to the Michaelis-Menten equation using Prism (GraphPAD, San Diego, CA). Data points in all cases represent the mean ± standard error of three determinations in a typical experiment.

To characterize the inhibitor specificity of rabbit Mrp2, we studied the effect of various compounds on [³H]E₂17G uptake by Sf9-Mrp2 membrane vesicles (Table 1). Phenolphthalein glucuronide exerted a profound inhibition, whereas the other two glucuronides (-naphthyl--D-glucuronide, p-nitrophenyl--D-glucuronide) and FL, a substrate of the classic organic anion transport system (Sullivan et al., 1990), did not inhibit transport up to 1 mM. Uptake was also susceptible to inhibition by LTC₄, MTX, and FL-MTX. Furthermore, we tested the LTD₄-receptor antagonist MK571 (Jones et al., 1989) and the immunosuppressive agent CsA, both of which are inhibitors of human MRP1 and rat Mrp2 (Leier et al., 1994; Büchler et al., 1996) and proved to be inhibitors of rabbit Mrp2.

View this table: [in this window] [in a new window]   TABLE 1 Effect of various compounds on Mrp2-mediated [3H]E217G transport Sf9-Mrp2 membrane vesicles were incubated with 50 nM [3H]E217G at 37° for 1 min with or without (control) the compounds as indicated. Net ATP-dependent transport was calculated by subtracting values in the presence of 4 mM 5'-AMP from those in the presence of 4 mM ATP. Transport was expressed as percent of the control uptake. Data represent mean values of three to six determinations (± standard error).

	Discussion

Top Summary Introduction Procedures Results Discussion References

Mrp2 mediates ATP-dependent elimination of conjugated organic anions from liver and has recently been cloned from rat (Büchler et al., 1996; Paulusma et al., 1996; Ito et al., 1997; Madon et al., 1997) and human (Taniguchi et al., 1996; Paulusma et al., 1997). We cloned a rabbit cDNA encoding an ABC-transporter that on basis of similar tissue distribution and highest amino acid homology was considered as the rabbit homolog of rat Mrp2 (van Kuijck et al., 1996, 1997). On injection of its cRNAs in Xenopus laevis oocytes, we observed in a few cases a cAMP-dependent chloride conductance (van Kuijck et al., 1996). The recent finding that substrates of Mrp2 activate a chloride conductance in hepatocytes of normal rats but not in TR hepatocytes (Weinman and Carruth, 1997) indicates that this phenomenon warrants further investigation.

To investigate Mrp2-mediated transport, we expressed rabbit Mrp2 in Sf9 cells using recombinant baculovirus. In these cells, Mrp2 is highly expressed, although less glycosylated compared with kidney Mrp2. This is in line with results from other studies in which CFTR and MRP1 were expressed in insect cells and detected as an underglycosylated product (Kartner et al., 1991; Gao et al., 1996). Results of functional studies on MRP1- and CFTR-expressing insect cells are comparable to those obtained from transfected eukaryotic cells, indicating that underglycosylation has no significant effect on its function (Kartner et al., 1991; Gao et al., 1996). This was further corroborated by inhibition of glycosylation with tunicamycin in drug-resistant MRP1-expressing human cells (Bakos et al., 1996) and, on basis of our studies, can also be concluded for Mrp2.

Based on studies with intact rats and liver canalicular membranes, the conjugates E₂17G and LTC₄ are considered to be substrates for rat Mrp2 (Büchler et al., 1996; Takikawa et al., 1996). In addition, ATP-dependent transport of LTC₄ has been demonstrated in membrane vesicles isolated from NIH/3T3 cells transfected with a rat mrp2 cDNA, and LTC₄ efflux was found in Mrp2-expressing Xenopus oocytes and COS-7 cells (Madon et al., 1997; Ito et al., 1998). In the current study, we unambiguously demonstrated that rabbit Mrp2 mediates ATP-dependent uptake of both [³H]E₂17G and [³H]LTC₄. The initial uptake rates for [³H]LTC₄ and [³H]E₂17G, as well as the V_max value for ATP using [³H]E₂17G as a cosubstrate, are lower than the values described for rat canalicular membrane vesicles (Büchler et al., 1996; Vore et al., 1996) and membrane vesicles from mrp2-transfected NIH/3T3 cells (Ito et al., 1998). This difference, however, may be explained by the substantially lower substrate concentrations that we used. Uptake of [³H]E₂17G in Sf9-Mrp2 membrane vesicles was inhibited by LTC₄ and phenolphthalein glucuronide. -Naphthyl--D-glucuronide and p-nitrophenyl--D-glucuronide had no significant effect on uptake, although both compounds are thought to be Mrp2 substrates. ATP-dependent uptake of p-nitrophenyl--D-glucuronide into rat canalicular membrane vesicles has been described, whereas in TR rat livers, -naphthyl--D-glucuronide excretion was impaired (de Vries et al., 1989; Kobayashi et al., 1991). These findings suggest that -naphthyl--D-glucuronide and p-nitrophenyl--D-glucuronide are transported with low affinity by Mrp2 and consequently are poor competitive inhibitors themselves. MK571 and CsA are inhibitors of human MRP1 and rat Mrp2 (Leier et al., 1994; Büchler et al., 1996) and, as shown in this study, also inhibit rabbit Mrp2-mediated [³H]E₂17G transport. However, it remains to be elucidated whether these compounds are Mrp2 substrates.

Mrp2 is expressed not only in liver canalicular membranes but also in small intestine and brush-border membranes of renal proximal tubular cells (Büchler et al., 1996; Paulusma et al., 1996; Schaub et al., 1997). However, the functional identification of an ATP-dependent organic anion transporter in membrane vesicles from renal proximal tubular cells, such as in liver canalicular membranes, has never been documented (Pritchard and Miller, 1993). This is mainly due to technical limitations because membrane vesicles of renal proximal tubular cells are exclusively orientated right-side out (Haase et al., 1978). Recently, Masereeuw et al. (1996) identified an energy-dependent transport mechanism for organic anions in isolated renal proximal tubules from killifish using FL-MTX as a substrate. The excretory pathway of FL-MTX was characteristic for its sensitivity to LTC₄, MTX, CsA, and probenecid. In addition, the energy-dependency of this pathway was confirmed by treating cells with KCN, which did not influence FL-MTX uptake but completely abolished luminal excretion. This suggests that FL-MTX may be an Mrp2-substrate for which we provide evidence in this study because FL-MTX strongly inhibits [³H]E₂17G uptake in Sf9-Mrp2 membrane vesicles. In contrast, FL did not inhibit [³H]E₂17G uptake, whereas MTX was only partially inhibitory. It remains to be established whether Mrp2 directly mediates ATP-dependent uptake of FL-MTX.

Besides Mrp2, additional organic anion transporters might be present in brush-border membranes of renal proximal tubular cells. For example, excretion of FL-MTX was shown to be only partially inhibited by probenecid, whereas the probenecid-insensitive mechanism was inhibited completely by verapamil (Masereeuw et al., 1996). In addition, it has been shown that TR rats have impaired hepatic excretion of the conjugates -naphthyl--D-glucuronide and LTC₄, whereas urinary excretion is hardly affected (Huber et al., 1987; de Vries et al., 1989), suggesting that the deficiency of Mrp2 in the kidney can be compensated for by other organic anion transporters. Possible candidates might be the organic anion transporters Oatp1 and Oat-k1, which are both localized in brush-border membranes of renal proximal tubular cells (Bergwerk et al., 1996; Masuda et al., 1997). Although Oatp1 and Oat-k1 are structurally not related to Mrp2, these proteins mediate transport of Mrp2-substrates, such as E₂17G, LTC₄, and MTX (Kanai et al., 1996; Saito et al., 1996; Li et al., 1997). Furthermore, the recently identified family members of human MRP1 (i.e., MRP3, MRP4, and MRP5) are all expressed to some extent in the kidney and might also be involved in renal organic anion transport (Kool et al., 1997).

In conclusion, we demonstrated ATP-dependent transport by recombinant rabbit Mrp2 and provided new data on inhibitor specificity. In future studies, this expression system will be used for identification and characterization of Mrp2-substrates, with emphasis on compounds that are excreted by the kidney.