17α-Ethinyl estradiol (EE), a synthetic estrogen, is an essential constituent of oral contraceptives, which have been widely prescribed since the 1970s. Over 60 million women currently take oral contraceptives, and their safety profile is well established. Numerous examples are known (Stockley, 1999) where coadministration of EE with a range of other drugs can result in decreased plasma levels of EE with the commensurate failure of contraception and breakthrough bleedings. Alterations in EE metabolism and disposition are proposed to occur via induction of hepatic or gut enzymes involved in the metabolism and/or transport of EE and its metabolites.

The pharmacokinetics and metabolism of EE in humans occurs both in gut and liver, respectively, where the mean bioavailability is reported to be 45% (Back et al., 1982; Rogers et al., 1987). EE mainly undergoes sulfation and glucuronidation, resulting in the formation of EE-3-O-sulfate (EE-S) and EE-3-O-glucuronide (EE-G). EE-G and EE-S have been detected in bile, intestinal mucosa, and urine (Maggs et al., 1983; Pacifici and Back, 1988; Orme et al., 1989). However, the mechanisms involved in the transport and elimination of these metabolites are unknown.

Multidrug resistance proteins (MRPs), transporters belonging to the ATP-binding cassette (ABC) superfamily, have been suggested to play an important role in the transport and detoxification of a wide range of endogenous compounds and xenobiotics (Borst et al., 2000; Borst and Elferink, 2002). The MRP family consists of nine members, which are referred to as MRP1–9 (Borst and Elferink, 2002). MRP1 (ABCC1) is localized in the basolateral membranes of polarized cells (Evers et al., 1996; Cole and Deeley, 1998) and is expressed in all tissues except the liver (Zaman et al., 1994). In contrast, MRP2 (ABCC2) is expressed in the canalicular membrane of hepatocytes, the apical membrane of the small intestine, the apical membrane of the proximal tubules of the kidney, and placental trophoblasts (Paulusma and Oude Elferink, 1997; Schaub et al., 1997, 1999; St-Pierre et al., 2000). MRP2 therefore may be involved in hepatobiliary-, renal-, and intestinal excretion of compounds, and protection of the fetus. MRP3 (ABCC3) is localized in the basolateral membranes of polarized cells (Kõnig et al., 1999). It is expressed in the gut, pancreas, liver cholangiocytes, adrenals, and kidney, and it is highly induced in hepatocytes under cholestatic condition (Donner and Keppler, 2001; Scheffer et al., 2002). MRP3 may play a role in enterocytes by transporting compounds from the intestine into the bloodstream, and under cholestatic conditions by removing toxic organic anions from hepatocytes (Keppler and Kõnig, 2000).

MRP1–3 are able to transport conjugated and nonconjugated organic anions (Borst et al., 2000; Borst and Elferink, 2002). The substrate specificity of MRP1 and MRP2 is overlapping and includes glutathione (GSH), glucuronide, and sulfate conjugates, some nonconjugated organic anions, and various neutral or positively charged drugs (Cole et al., 1994; Cui et al., 1999). MRP3 also transports organic anions but prefers glucuronide and sulfate conjugates over glutathione conjugates (Hirohashi et al., 1999, 2000; Zelcer et al., 2003b).

MRP1–3 have been shown to interact with a wide range of substrates, including several physiologically or pharmacologically active organic anions. For instance, GSH plays an important role in MRP1- and MRP2-mediated transport of a number of compounds. In some cases, compounds are cotransported with GSH but in others not (Loe et al., 1998; Renes et al., 1999; Evers et al., 2000; Leslie et al., 2001; Qian et al., 2001). Bodo et al. (2003) showed that some organic anions, including indomethacin, furosemide, probenecid, and some bile acids, significantly stimulated MRP2-mediated 17β-estradiol-17β-d-glucuronide (E₂17βG) uptake, but inhibited its uptake by MRP3. In MDCKII-MRP2 cells, probenecid enhanced vectorial transport of the human immunodeficiency virus protease inhibitors saquinavir, ritonavir, and indinavir (Huisman et al., 2002). So far, the mechanism explaining these complex stimulation and inhibition effects has not been elucidated, but they suggest that MRP1–3 have multiple drug binding sites (Loe et al., 1998; Daoud et al., 2000; Borst and Elferink, 2002; Deeley and Cole, 2003).

Here, we investigate the ability of MRP1–3 to transport EE-G and EE-S using membrane vesicles isolated from baculovirus infected Spodoptera frugiperda (Sf9) insect cells expressing MRP1, MRP2, or MRP3. Although we determined that EE-G was a good substrate for MRP2 and MRP3, neither MRP1, MRP2, nor MRP3 was capable of transporting EE-S. Interestingly, EE-S stimulated strongly MRP2- and MRP3-mediated transport of EE-G and had variable effects on the transport of other substrates. The interpretation of our findings in light of models explaining MRP-mediated transport is discussed. In addition, our data suggest that MRP2 and MRP3 are the relevant transporters in vivo for the export of EE-G from intestine and liver.

Materials and Methods

Materials. [³H]EE-S (10.09 μCi/nmol) and [³H]EE-G (10.09 μCi/nmol) were synthesized by the Department of Drug Metabolism (Merck Research Laboratories, Rahway, NJ) (detailed methods available upon request). The purity of [³H]EE-S and [³H]EE-G was verified by high-performance liquid chromatography (>99.9%). [¹⁴C]Ethacrynic acid glutathione conjugate (EA-SG) (65 mCi/mmol) was synthesized by the Merck Labeled Compound Synthesis Group (Rahway, NJ) by reacting [¹⁴C]ethacrynic acid with glutathione as described previously (Ploemen et al., 1990). [¹⁴C]Ethacrynic acid was obtained by substituting [¹⁴C]paraformaldehyde into the method described in U.S. Patent 3,255,241 for preparation of unlabeled material. The purity of [¹⁴C]EA-SG verified by high-performance liquid chromatography was 98%. [³H]E₂17βG (40.5 Ci/mmol) was obtained from PerkinElmer Life Sciences (Boston, MA). EE-G and EE-S was obtained from Steraloids (Newport, RI). E₂17βG, creatine phosphate, and creatine phosphokinase were purchased from Sigma-Aldrich (St. Louis, MO). All chemicals were of the highest analytical purity grade.

Preparation of Membrane Vesicles. Recombinant baculovirus containing MRP1, MRP2 cDNAs, and β-galactosidase gene (control) were obtained from Solvo Biotechnology (Szeged, Hungary). Baculovirus containing MRP3 cDNA was kindly provided by Dr. Piet Borst (The Netherlands Cancer Institute, Amsterdam, The Netherlands). Sf9 cells in suspension were grown in Sf-900 II SFM medium in the absence of serum (Invitrogen, Carlsbad, CA). About 4 × 10⁷ Sf9 cells were seeded in 175-cm² tissue culture flasks. After the cells became attached, the medium was removed, and 3 ml of medium and 3 ml of virus stock containing MRP1, MRP2, or MRP3 (about 5–8 × 10⁷ virus/ml) was added to infect the cells. One hour after addition of the virus to the cells, cell culture medium was added up to a final volume of 30 ml. After incubation for 72 h at 26°C, the cells were harvested and washed twice in ice-cold washing buffer (50 mM Tris/HCl, 300 mM mannitol, 0.5 mM phenylmethylsulfonyl fluoride, pH 7.0) and centrifuged at 800g for 5 min at 4°C. The cell pellet was resuspended in ice-cold TMEP buffer (50 mM Tris, 50 mM mannitol, 2 mM EGTA-Tris, 2 mM dithiothreitol, aprotinin (8 μg/ml), leupeptin (10 μg/ml), phenylmethylsulfonyl fluoride (50 μg/ml), pH 7.0) and homogenized for 10 min on ice using a tight-fitting Dounce homogenizer. After centrifugation at 800g for 10 min at 4°C, the supernatant was collected and centrifuged at 100,000g for 1 h at 4°C. The pellet was resuspended in TMEP buffer and passed 20 times through a 27-gauge needle. The vesicles were dispensed in aliquots, frozen in liquid nitrogen, and stored at –80°C until use.

Vesicular Uptake Studies. Vesicular uptake studies were performed using the rapid filtration technique as reported previously (Chu et al., 1997). The transport medium contained the radiolabeled ligand, 250 mM sucrose, 10 mM Tris/HCl (pH 7.4), 10 mM MgCl₂, 5 mM ATP, or 5 mM AMP, and an ATP-regenerating system (10 mM creatine phosphate and 100 μg/ml creatine phosphokinase). The uptake study was performed at 37°C. After preincubation for 3 min at 37°C, the uptake study was started by the addition of vesicle suspension (10 μg of protein). The final incubation volume was 20 μl. In the inhibition study, the inhibitors were dissolved in transport buffer and preincubated with ligands for 3 min. At designated time points, transport was terminated by adding 1 ml of ice-cold stop solution containing 10 mM Tris/HCl (pH 7.4), 250 mM sucrose, and 100 mM NaCl. The stopped reaction mixture was filtered through 0.45-μm HA Millipore filters (Millipore Corporation, Bedford, MA) and subsequently washed twice with 5 ml of ice-cold stop solution. For the uptake study with EE-S, glass fiber (type A/E) filters (Gelman Sciences, Dorval, QC, Canada) were used to minimize nonspecific binding to membrane filters. The radioactivity retained on the filter and in the reaction mixture was measured in a liquid scintillation counter (LS 6000; Beckman Coulter Inc., Fullerton, CA). ATP-dependent uptake was determined as the difference in uptake in the presence and absence of ATP.

Western Blot Analysis. Membrane vesicles isolated from baculovirus infected Sf9 cells were solubilized in Laemmli sample buffer (Bio-Rad, Hercules, CA) containing 2.5% of β-mercaptoethanol and separated in a 7.5% denaturing polyacrylamide gel. Gels were immunoblotted onto nitrocellulose membranes. MRP1, 2, and 3 were detected with the specific monoclonal antibodies MRP-R1 (1:500), M₂I4 (1:500), and MRP-III₂₁ (1:40), respectively, as described previously (Bakos et al., 2000). The blots were then developed with the enhanced chemiluminescence kit (Amersham Biosciences Inc., Piscataway, NJ) and exposed to enhanced chemiluminescence hyperfilm (Amersham Biosciences Inc.).

Data Analysis. Kinetic parameters for the ATP-dependent uptake were obtained by fitting the data to the following equation: where V_o is the initial uptake rate of substrate (picomoles per minute per milligram of protein), S is the substrate concentration in the medium (micromolar), K_m is the Michaelis constant (micromolar), and V_max is the maximum uptake rate (picomoles per minute per milligram of protein). To obtain estimates of kinetic parameters, the uptake data were fitted to eq. 1 by a nonlinear least-squares method using KaleidaGraph (Synergy Software, Reading PA).

The inhibition constant (K_i) values for evaluating the inhibitory effect of EE-G on the uptake of [¹⁴C]EA-SG by MRP2 and [³H]E₂17βG by MRP3 were obtained by fitting the following equation to the data as described previously (Chu et al., 1997): V_(+I) and V_(-I) represent the transport velocity in the presence and absence of inhibitor, respectively, and I is the inhibitor concentration. This equation was derived based on the assumptions that first, inhibition was competitive or noncompetitive, and second that the [¹⁴C]EA-SG concentration (2 μM) used was much lower than the K_m value (15 μM for MRP2; data not shown), and that the [³H]E₂17βG concentration (0.4 μM) used was much lower than the K_m value (27 μM for MRP3; data not shown).

Results

Detection of MRP1, MRP2, and MRP3 in Membrane Vesicles. By Western blotting, MRP1, MRP2, and MRP3 were detected with specific antibodies in membrane vesicles isolated from baculovirus-infected Sf9 cells expressing cDNAs encoding MRP1, MRP2, or MRP3, respectively. As shown in Fig. 1, A to C, MRP1 (lanes 3 and 4), MRP2 (lanes 5 and 6), and MRP3 (lanes 7 and 8) were detectable in these vesicles. Levels were similar to those in vesicles used by Bodo et al. (2003) (data not shown). No signal was detected in vesicles isolated from control β-galactosidase-expressing cells (lanes 1 and 2). Because MRP1, MRP2, and MRP3 were detected using different monoclonal antibodies, it was not possible to compare the exact expression levels of the proteins in these membrane vesicles. Bodo et al. (2003), however, detected comparable amounts of MRP1–3 in Sf9 membrane vesicles after staining gels with Coomassie Brilliant Blue. Because we do not know what percentage of MRP was transport competent, the absolute values of the transport measured by different transporters should be compared with care. As a positive control for the quality of the vesicles, functional activity of MRP1, MRP2, and MRP3 was confirmed by ATP-dependent uptake of EA-SG and E₂17βG (data not shown).

Transport of EE-G by MRP1, MRP2, and MRP3. To evaluate whether EE-G was a substrate for MRP1, MRP2, or MRP3, uptake of [³H]EE-G (0.2 μM) into membrane vesicles containing MRP1, MRP2, and MRP3 was investigated (Fig. 2). ATP-dependent uptake of [³H]EE-G (0.2 μM) in MRP1-containing vesicles was low and comparable with that observed in control Sf9 membrane vesicles (Fig. 2A). In contrast, ATP-dependent uptake of [³H]EE-G (0.2 μM) was much higher in MRP2-containing vesicles and showed linearity within 5 min (Fig. 2B). As shown in Fig. 2C, significant time- and ATP-dependent uptake of EE-G was also observed in MRP3-containing vesicles. Several studies have shown that MRP1-mediated transport of some compounds, including 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol glucuronide and estrone-3-sulfate, is GSH-dependent (Loe et al., 1998; Sakamoto et al., 1999; Leslie et al., 2001; Qian et al., 2001). We therefore evaluated whether GSH stimulates MRP1-mediated EE-G uptake. In this experiment, 10 mM dithiothreitol was added to the reaction mixture as a reducing agent (Leslie et al., 2001). Our results, however, indicated that 3 mM GSH did not have an effect on MRP1-mediated EE-G uptake (data not shown).

Concentration dependence of ATP-dependent uptake of EE-G was studied in MRP2 and MRP3 containing membrane vesicles. Because the uptake of [³H]EE-G (0.2 μM) in the presence of 5 mM ATP was linear for up to 5 min (Fig. 2, B and C), the initial rate for the uptake of EE-G at various substrate concentrations was determined after a 5-min incubation. Kinetic analysis revealed that the ATP-dependent uptake of EE-G by MRP2- and MRP3-containing vesicles was saturable with a K_m of 35 ± 3.5 μM and a V_max of 310 ± 10 pmol/min/mg protein for MRP2-mediated uptake (Fig. 3A), and a K_m of 9.2 ± 2.3 μM and a V_max of 58 ± 4.2 pmol/min/mg protein for MRP3-mediated uptake (Fig. 3B).

Transport of EE-S by MRP1, MRP2, and MRP3. To evaluate whether EE-S was a substrate for MRP1, MRP2, or MRP3, uptake of [³H]EE-S (0.2 μM) was investigated in MRP1-, MRP2-, and MRP3-containing vesicles, and control Sf9 membrane vesicles containing β-galactosidase. No ATP-dependent uptake of EE-S higher than in the control vesicles was observed by any of these transporters (Fig. 4, A–C). Our studies also showed that GSH up to 3 mM did not stimulate MRP1-, MRP2-, and MRP3-mediated EE-S uptake (data not shown). Studies using membrane vesicles isolated from MDCKII-MRP2 cells also indicated that EE-S was not a substrate of MRP2 (our unpublished observation).

Effect of EE-G and EE-S on MRP2- and MRP3-Mediated Transport by EA-SG and E₂17βG. Recently, it has been shown that compounds can have both an inhibitory and/or stimulatory effect on MRP-mediated transport both in intact cells and in membrane vesicles (Bakos et al., 2000; Evers et al., 2000; Qian et al., 2001; Huisman et al., 2002). We therefore investigated the effect of EE-G and EE-S on MRP2- and MRP3-mediated uptake by EA-SG and E₂17βG, known substrates for these transporters (Evers et al., 1998; Hirohashi et al., 1999). Uptake of EA-SG and E₂17βG by MRP2 and MRP3, respectively, was ATP-dependent and saturable. The K_m value for EA-SG uptake by MRP2 was 15 μM, and the K_m value for E₂17βG uptake by MRP3 was 27 μM (data not shown). Recent studies showed that the rate of E₂17βG transport by MRP2 increased sigmoidally with half-maximal transport of 120 μM (Bodo et al., 2003; Zelcer et al., 2003a). EE-G significantly inhibited ATP-dependent uptake of EA-SG in MRP2 containing vesicles with a K_i value of 20 ± 1.1 μM (Fig. 5A). EE-G also significantly inhibited MRP3-mediated uptake of E₂17βG in MRP3-expressing vesicles with a K_i value of 5.0 ± 0.6 μM (Fig. 5B). In contrast, EE-S at relatively low concentrations (2–20 μM) stimulated MRP2-mediated uptake of EA-SG up to 1.5-fold (Fig. 6A). EE-S at high concentration (100 μM), however, inhibited MRP2-mediated uptake of EA-SG by approximately 50% (Fig. 6A). Unlike its effect on EA-SG, EE-S only showed a strong stimulatory effect on MRP2-mediated E₂17βG uptake within the concentrations tested (2–100 μM) (Fig. 6B). EE-S at a concentration of 100 μM enhanced E₂17βG uptake up to 17-fold (Fig. 6B). EE-S at low concentration (2–20 μM) also showed a stimulatory effect (3.8-fold) on MRP3-mediated uptake of E₂17βG but slightly inhibited its uptake at a very high concentration (100 μM; Fig. 6C).

Effect of EE-S on MRP1–3-Mediated Uptake of EE-G. A possible stimulatory effect of EE-S on MRP1–3-mediated uptake of EE-G was also evaluated. EE-S at 2 to 100 μM had no stimulatory effect on ATP-dependent uptake of EE-G in MRP1-containing vesicles (Fig. 7A), but rather inhibited its low-level uptake. Because EE-G uptake by MRP1 was not higher than in control vesicles (Fig. 2A), this effect was possibly due to inhibition of a low endogenous MRP-like activity present in Sf9 cells. In contrast, EE-S (2–100 μM) stimulated MRP2-mediated ATP-dependent uptake of EE-G strongly (Fig. 7B). EE-S at a concentration of 20 μM increased the ATP-dependent initial uptake rate of EE-G 15-fold. Stimulation was further enhanced up to 22-fold in the presence of 100 μM EE-S, the highest concentration tested. EE-S (2–100 μM) also showed a significant stimulatory effect on MRP3-mediated uptake of EE-G (Fig. 7C). Strongest stimulation was observed at EE-S concentration of 20 μM (about 5-fold), whereas the stimulatory effect slowly decreased at higher concentrations (Fig. 7C). The effects of EE-S on MRP2- and MRP3-mediated uptake of EE-G were further characterized by determining the initial uptake rates of EE-G at various substrate concentrations in the presence of EE-S (20 μM; Fig. 8, A and B, respectively). In the presence of EE-S (20 μM), the K_m for uptake of EE-G by MRP2 was calculated to be 4.1 ± 0.7 μM, which was about 9-fold lower than the K_m determined in the absence of EE-S (Fig. 3A; K_m = 35 ± 3.5 μM). In the presence of EE-S, the V_max was 435 ± 16 pmol/min/mg protein, which was slightly higher than that in the absence of EE-S (Fig. 3A; V_max = 310 ± 10 pmol/min/mg protein). For MRP3-mediated uptake of EE-G, in the presence of 20 μM EE-S, the K_m was estimated to be 0.5 ± 0.2 μM, which was about 18-fold lower than the K_m obtained in the absence of EE-S (Fig. 3B; K_m = 9.2 ± 2.3 μM). The V_max value was 88 ± 5.6 pmol/min/mg protein, which was again slightly higher than that in the absence of EE-S (Fig. 3B; V_max = 58 ± 4.2 pmol/min/mg protein).

Effect of EE-G on MRP2- and MRP3-Mediated [³H]EE-S Uptake. The stimulatory effect of EE-S on transport of EE-G could be explained by a mechanism in which both compounds are cotransported or by an allosteric effect of EE-S on MRP2 or MRP3 resulting in an increased transport of EE-G. To investigate whether transport of EE-S could be reciprocally stimulated by EE-G, the effect of EE-G on uptake of [³H]EE-S (20 μM) by MRP2 and MRP3 was evaluated. As shown in Fig. 9, EE-G (0.5–100 μM) did not cause a significant stimulation of ATP-dependent uptake of EE-S by MRP2- (Fig. 9A) or MRP3 (Fig. 9B)-containing vesicles. The slight ATP-dependent uptake of EE-S observed in MRP2-containing vesicles at EE-G concentrations of 5, 20, and 100 μM was not significant, as this stimulatory effect could not be confirmed in a subsequent time-course experiment. No significant ATP-dependent uptake of EE-S in the presence of various concentrations of EE-G was detected in these experiments (data not shown).

Discussion

In this study, we show that EE-G is a substrate of MRP2 and MRP3, but not of MRP1 (Figs. 2 and 3). EE-S was not transported by MRP1–3 (Fig. 4) nor in the presence of GSH (data not shown). Because EE undergoes extensive gut (E_g = 44%) and liver (E_h = 25%) first-pass metabolism in humans (Back et al., 1982), MRP2 and MRP3 may be the physiologically relevant transporters for the export of EE-G from these organs. These transporters do not transport EE-S, however. Furthermore, we find a strong allosteric stimulation between two pharmacologically relevant conjugated metabolites of EE, EE-G, and EE-S. The stimulatory effect of EE-S on MRP2- and MRP3-mediated transport of EE-G (Fig. 7) might be relevant in vivo, and therefore could potentially have an impact on the in vivo disposition of EE and its interaction with other drugs.

Interestingly, EE-S showed a pronounced effect on MRP2- and MRP3-mediated uptake of EE-G, but the extent of this interaction varied among these transporters. EE-S stimulated MRP2-mediated uptake of EE-G strongly (Fig. 7B). Uptake was enhanced up to 22-fold at an EE-S concentration of 100 μM. EE-S also significantly stimulated MRP3-mediated uptake of EE-G (5-fold) at an EE-S concentration of 20 μM, but this stimulatory effect was decreased at higher concentration (Fig. 7C). In contrast, EE-S had no stimulatory effect on MRP1-mediated EE-G uptake (Fig. 7A). Such a differential modulatory effect of EE-S on uptake of EE-G by MRP1–3 suggests that the modulator binding sites vary greatly between MRP1–3 in terms of their affinity to various substrates.

To further investigate the mechanism explaining the stimulatory effect of EE-S on MRP2- and MRP3-mediated uptake of EE-G, we compared the kinetic parameters of EE-G uptake in the presence and absence of EE-S. In the presence of EE-S (20 μM), the K_m values for EE-G uptake by MRP2 and MRP3 were considerably decreased compared with those in the absence of EE-S, whereas the V_max values were only slightly higher than those in the absence of EE-S (Figs. 3 and 8). The V_max/K_m, which represents the intrinsic clearance of EE-G transport, was calculated to be 110 in the presence of EE-S (20 μM) and 8.8 in the absence of EE-S for MRP2-mediated uptake, whereas the V_max/K_m ratio was 180 in the presence of EE-S (20 μM) compared with 6.3 in the absence of EE-S for MRP3-mediated uptake. Together, these data demonstrate that EE-S stimulates MRP2- and MRP3-mediated uptake of EE-G by increasing its apparent affinity. Therefore, EE-S could potentially enhance the excretion of EE-G in vivo from the organs where these transporters are expressed. Although it is important to note that EE is administered to humans at a low dose (35 μg/day), local, intracellular concentrations of these metabolites might well be in the micromolar range, as studied here. Therefore, the effects of EE-S on MRP2- and MRP3-mediated transport of EE-G might also be relevant in vivo. Because EE-G and EE-S are highly charged, they do not readily diffuse over the plasma membrane. We therefore could not investigate the effect of EE-S on MRP2 and MRP3-mediated EE-G transport in intact cells by (vectorial) transport studies.

Earlier studies have shown that some modulators of MRP1 and MRP2 were cotransported with substrates (Loe et al., 1998; Evers et al., 2000). The stimulation of MRP2 and MRP3-mediated uptake by EE-S raises the possibility that EE-S might be cotransported with the MRP2 and MRP3 substrates EE-G, E₂17βG, and EA-SG. However, no ATP-dependent uptake of EE-S was observed in the presence of EE-G (Fig. 9), suggesting that EE-S was not cotransported with EE-G. We also determined that EE-S was not cotransported with E₂17βG in MRP3-containing vesicles (data not shown). We can not rule out that some ATP-dependent cotransport of EE-S with EE-G occurs, which we might have missed in our studies due to the relatively high nonspecific binding of EE-S to membrane filters. However, even if this would be the case, the transport of EE-S in the presence of EE-G was very weak compared with its strong stimulatory effect on EE-G transport. Together, the above-mentioned studies suggest that EE-S has an allosteric effect on MRP2 and MRP3, resulting in an increased transport rate of EE-G, and that EE-S is not transported itself at an appreciable rate.

EE-S showed strong stimulatory effects on both MRP2- and MRP3-mediated EE-G and E₂17βG uptake. Hirohashi et al. (1999) and Akita et al. (2001) found that the sulfate conjugates of E3040 and 4-methyl umbelliferone also stimulated both rat and human MRP3-mediated E₂17βG uptake. Our studies demonstrated that the stimulatory effect of EE-S on the uptake of EE-G and E₂17βG varied between MRP2 and MRP3. EE-S showed a stimulatory effect on MRP2-mediated uptake of EE-G and E₂17βG at all concentrations tested (Figs. 6B and 7B), whereas EE-S only showed a strong stimulatory effects on MRP3-mediated EE-G and E₂17βG uptake at lower concentration (<20 μM). This stimulation by EE-S decreased at higher concentrations (Figs. 6C and 7C). This suggests that MRP2 and MRP3 might have a different binding affinity for EE-S. The finding that EE-S showed a much stronger stimulatory effect on MRP2-mediated EE-G and E₂17βG uptake than on EA-SG uptake suggests that sulfate conjugates may stimulate the transport of glucuronides stronger than of glutathione conjugates. Further experiments are needed, however, to verify this possibility.

Recently, various cotransport and allosteric activation models have been proposed to explain the complex interactions between MRP1–3 and substrates. Evers et al. (2000) proposed a working model to explain the cotransport of MRP2 substrates, including sulfinpyrazone or vinblastine, with GSH. However, this model does not explain the lack of cotransport of EE-S with EE-G and other MRP2 and MRP3 substrates, including E₂17βG. Similarly, several recent studies also indicated that some MRP1–3 modulators are not detectably cotransported with MRP substrates, whereas they had a strong stimulatory effect (Leslie et al., 2001, 2003; Qian et al., 2001; Zelcer et al., 2003a).

While this work was in progress, Zelcer et al. (2003a) proposed a transport model for MRP2, in which MRP2 has two similar but nonidentical ligand binding sites: one site from which substrate is transported (S site) and another site able to modulate the transport (M site). Binding of a modulator to the M site would induce a structural change that results in a better fit of the substrate at the S site. According to this model, EE-S might bind only to the M site of MRP2, leading to a conformational change in the S site and there-with causing a strong stimulation of MRP2-mediated transport of EE-G. However, this model may not explain all our experiments. For instance, EE-S stimulated MRP2-mediated uptake of EA-SG at relatively low EE-S concentrations (2–20 μM), but uptake was inhibited at higher concentrations of EE-S (Fig. 6). Based on the “bell-shaped” stimulation curve, we speculate that EE-S could bind to both the M and S site. Therefore, at lower concentrations, it predominantly binds to the M site and stimulates EA-SG uptake; at higher concentrations, it competes with EA-SG for the S site, resulting in the inhibition of EA-SG transport. At present, we have no evidence indicating that the competition between EE-S and EA-SG for the S site could result in transport of EE-S. On the other hand, EE-G might have a much higher binding affinity for the S site than EA-SG. Therefore, also at high concentrations, EE-S can not compete with EE-G for binding to the S site. Like for MRP2, our results also show that EE-S allosterically stimulates MRP3-mediated transport of EE-G and E₂17βG, but was not cotransported with these compounds. We infer that MRP3 might contain similar multiple ligand-binding sites as postulated for MRP2. Together, these observations suggest that binding of some compounds to MRP2 and MRP3 has an allosteric effect resulting in a conformational change in the protein and increasing its affinity for substrates.

In our studies, uptake of EE-G by MRP2 and MRP3 followed simple Michaelis-Menten kinetics. Also upon close inspection of the data from multiple experiments, we were not able to observe a sigmoidal uptake rate versus concentration curve as found by Zelcer et al. (2003a) and Bodo et al. (2003) for MRP2-mediated uptake of E₂17βG. To explain this apparent discrepancy, we hypothesize that EE-G only binds to the S site and has no detectable affinity for the M site.

Our data suggest that the transporters responsible in vivo for the export of EE-G from enterocytes and hepatocytes could be MRP2 and MRP3. Because several organic anions, which show stimulation of MRP-mediated uptake, are physiologically or pharmacologically active, it has been speculated that these might cause clinically relevant drug-drug interactions (Bakos et al., 2000; Huisman et al., 2002). It is important to note, however, that all these studies were performed in vitro with either inside-out membrane vesicles or polarized cell lines. In future studies, it therefore will be important to address whether these effects can be mimicked in animals or perfused organ systems.

The complex interaction between MRP transporters and organic anions suggests that caution needs to be taken in using inhibition studies as a screening tool to predict whether compounds could cause inhibition of MRP-mediated transport under physiological conditions. Probably, such experiments should be performed with physiologically relevant substrates, such as bilirubin glucuronide for MRP2, for example.