Analysis of Methotrexate and Folate Transport by Multidrug Resistance Protein 4 (ABCC4): MRP4 Is a Component of the Methotrexate Efflux System¹

The MRP⁴ family of ATP-binding cassette transporters consists of at least nine members (1, 2, 3), five of which have been determined to be lipophilic anion transporters that are able to confer resistance to a variety of anticancer agents. The substrate ranges and drug resistance properties of MRP1, MRP2 (cMOAT) and MRP3 are the best characterized. MRP1 and MRP2 exhibit the highest potency and broadest resistance profiles with regard to natural product agents (4, 5, 6, 7, 8, 9, 10, 11, 12, 13). By contrast, MRP3 has been reported to confer only low levels of resistance to etoposide (12, 13). With regard to other classes of anticancer agents, all three pumps are capable of conferring resistance to MTX (12, 13, 14), but only MRP2 is able to confer resistance to cisplatin (9, 10, 11). MRPs 1–3 share a high degree of structural resemblance both in terms of protein topology and amino acid alignment (15), and this similarity is evident in their substrate selectivities in that they all have the facility for the MgATP-energized transport of glutathione and glucuronate conjugates (9, 10, 16, 17, 18, 19, 20, 21, 22, 23). However, there are differences in the substrate ranges of these pumps, and these differences in combination with differences in tissue-specific expression patterns and subcellular polarity indicate that they subserve distinct physiological roles. The wide distribution of MRP1 in tissues has led to the notion that it contributes to the phase III detoxification of xenobiotics in many cell types (24, 25), and investigations using MRP1-deficient mice indicate that MRP1, which has a higher affinity for leukotriene C4 than MRP2 and MRP3, also plays a role in immune responses involving efflux of this cysteinyl leukotriene (26, 27). MRP2, is expressed primarily at the canaliculus of hepatocytes, where it functions in the hepatobiliary excretion of conjugates, such as bilirubin glucuronide, and in provision of biliary glutathione (28). The facility of MRP3 for transporting monoanionic bile salts (22, 23), in combination with its induction at basolateral surfaces of hepatocytes in cholestatic conditions (29, 30), suggests that it is a back-up system for protecting hepatocytes from compounds that are ordinarily eliminated by extrusion into the bile.

MRP4 and MRP5 differ in their structures from MRPs 1–3 in that unlike the latter proteins, they do not possess a third (NH₂-terminal) hydrophobic domain (15, 31), and this difference is reflected in their distinctive drug resistance profiles, substrate selectivities, and potential physiological functions. These two pumps do not confer resistance to natural product anticancer agents, but instead are capable of conferring resistance to nucleotide analogues such as 6-MP and 9-(2-phosphonylmethoxyethyl)adenine (32, 33, 34, 35, 36). In addition, MRP4 and MRP5 have the facility for the MgATP-energized transport of cAMP and cGMP, a feature that suggests their involvement in the regulation of intracellular cyclic nucleotide levels (35, 37).

Although MRP4 and MRP5 confer resistance to nucleotide analogues and transport cyclic nucleotides, the substrate ranges and resistance profiles of these two pumps appear to be distinct in several regards. By contrast with MRP5, MRP4 has the facility for mediating the transport of glucuronides, such as estradiol E₂17βG, and it is also a higher affinity transporter of cAMP than is MRP5 (35). Another potential difference concerns the antimetabolite MTX. MRP4, similar to MRPs 1–3, and in contrast to MRP5, has been shown to confer resistance to this widely used antimetabolite (34). Although the potency of MRP4 as assessed in transfected NIH3T3 cells is lower than that reported for MRPs 1–3 expressed in transduced 2008 cells (13, 14), this feature of MRP4 is of interest in that it raises the possibility that it is a component of the previously described energy-dependent efflux system for MTX (38), a system for which MRP3 and MRP1 are now established components (39). However, the notion that MRP4 is able to transport MTX has not been firmly established in that it is currently an inference based entirely upon a previous study in which we found a relatively low level of resistance to this agent and an associated cellular accumulation deficit in a single clone of transfected NIH3T3 cells (34).

Here we analyze the potential involvement of MRP4 in the cellular pharmacology of MTX and related processes, by examining its ability to mediate the transport of this agent and folates in membrane vesicles prepared from insect cells infected with MRP4 baculovirus. In so doing, it is demonstrated that MRP4 is not only competent in the MgATP-energized transport of MTX and folates, but that its affinities for these substrates compare favorably with those we reported previously for MRP1 and MRP3 (39). It is also shown that MRP4 can accommodate a critical property of previously described MTX efflux systems in that its capacity to transport this agent is abrogated by the addition of a single glutamyl residue, as is also the case for MRP1 and MRP3 (39). In addition, it is shown that glutamylation similarly affects the transport activity of MRP2.

On the basis of these results, two conclusions are drawn: (a) MRP4 is a bona fide MTX resistance factor, and it along with MRP1, MRP2, and MRP3 are components of the energy-dependent efflux system for this agent and physiological folates; and (b) MRP4 is a common efflux pump for MTX and certain nucleotide analogues.

[³H]MTX (21.2 Ci/mmol), [³H]MTX-Glu₂ (15 Ci/mmol), [³H]MTX-Glu₃ (17 Ci/mmol), [³H]folic acid (20.2 Ci/mmol), [³H]N⁵-formyl-THF (also known as leucovorin; 17 Ci/mmol), and unlabeled MTX-Glu₂ were purchased from Moravek Biochemicals (Brea, CA). Creatine phosphokinase, phosphocreatine, verapamil, cyclosporin A, probenecid, sulfinpyrazone, trequinsin, zaprinast, ATP, AMP, E₂17βG, cGMP, cAMP, folic acid, and N⁵-formyl-THF were purchased from Sigma Chemical Co. (St. Louis, MO). MK571 and PSC833 were kindly provided by Dr. A. W. Ford-Huchinson (Merck-Frosst Center for Therapeutic Research, Pointe Claire-Dorval, Quebec, Canada) and Sandoz (Tsukuba, Japan), respectively. Sildenafil (Viagra) was kindly provided by Pfizer (Sandwich, Kent). MRP2-transfected LLC/PK1 cells (LLC/cMOAT-1) and parental plasmid-transfected cells (LLC/CMV) were described previously (10). LLC/PK1 cells were grown in M199 medium supplemented with 10% fetal bovine serum, penicillin/kanamycin, and glutamine. Insect cells (Sf9) were cultured and infected with MRP4 baculovirus as described previously (34).

Membrane vesicles were prepared by the nitrogen cavitation method as described previously (40). Transport experiments were performed using the rapid filtration method essentially as described (17) and carried out in medium containing membrane vesicles (10 μg), 0.25 m sucrose, 10 mm Tris-HCl (pH 7.4), 10 mm MgCl₂, 4 mm ATP, 10 mm phosphocreatine, 100 μg/ml creatine phosphokinase, and radiolabeled substrate ± unlabeled substrate in a total volume of 50 μl. Reactions were carried out at 37°C and stopped by the addition of 3 ml of ice-cold stop solution (0.25 m sucrose, 100 mm NaCl, and 10 mm Tris-HCl, pH 7.4) Samples were passed through 0.22 μm Durapore membrane filters (Millipore, Bedford, MA) under vacuum. The filters were washed three times with 3 ml of ice-cold stop solution and dried at room temperature for 30 min. Radioactivity was measured by the use of a liquid scintillation counter. Rates of net ATP-dependent transport were determined by subtracting the values obtained in the presence of 4 mm MgAMP from those obtained in the presence of 4 mm MgATP. Uptake rates were linear for >5 min, and rates for concentration dependence experiments were measured at 5 min.

In a previous study in which MRP4-transfected NIH3T3 cells were used, we found that the pump confers low levels of resistance to MTX (34). To explore the implications of this finding, the capacity of MRP4 to transport this antimetabolite in vitro was examined. For this purpose, MRP4-dependent transport was assayed on density-fractionated membrane vesicles prepared from insect (Sf9) cells infected with MRP4 baculovirus. The relative contribution of MRP4 to overall uptake was assessed in parallel experiments performed on membrane vesicles purified from uninfected insect cells. In a previous study, we found that this was a robust system for the analysis of MRP4-mediated transport of cyclic nucleotides and E₂17βG (35).

[³H]MTX was indeed subject to MRP4-mediated, MgATP-dependent transport (Fig. 1 A). When measured at initial concentrations of 100 μm and at the 5-min time point of the assay, [³H]MTX was taken up by MRP4-enriched vesicles a rate of 103 pmol/mg/min from medium containing MgATP and a rate of 45 pmol/mg/min from medium containing MgAMP. By contrast, the rates of uptake of membranes prepared from uninfected insect cells were <34 pmol/mg/min, under either energized or nonenergized conditions. MgATP-stimulated uptake by MRP4-enriched membranes was linear over the first 10 min of the assay. Uptake by membranes prepared from insect cells infected with parental baculovirus was indistinguishable from that of uninfected insect cells, indicating that infection by itself did not influence MTX transport (data not shown).

MTX, a monoglutamate, is metabolized in the cell by the enzymatic addition of up to six γ-linked glutamate residues, and a critical feature of previously described energy-dependent MTX efflux systems is the ability to transport the parent drug but not its polyglutamylated metabolites (38). To determine whether MRP4 can accommodate this criterion, its capacity to transport [³H]MTX-Glu₂ was assessed. In sharp contrast to [³H]MTX and in strict agreement with the properties expected for a component of the MTX efflux system, 100 μm [³H]MTX-Glu₂ was not transported to any extent by MRP4-enriched membrane vesicles. As shown in Fig. 1 B, the uptake rate of MRP4-enriched membranes in medium containing MgATP was indistinguishable from either the uptake rate of the same membranes in medium containing MgAMP or the uptake rates of membranes prepared from uninfected insect cells under either energized or nonenergized conditions. [³H]MTX-Glu₃ was similarly not subject to MRP4-mediated transport (data not shown).

MgATP-dependent retention of MTX by MRP4-enriched membrane vesicles was predominately a consequence of transport into the intravesicular space as opposed to nonspecific binding to the filters or membranes. Uptake of 100 μm [³H]MTX by MRP4-enriched membrane vesicles in medium containing MgATP increased as a linear function of the reciprocal of sucrose concentration, indicating that the transported substrate was delivered into an osmotically active compartment (Fig. 2). By contrast, substrate retention measured in medium containing MgAMP was only moderately affected by the sucrose concentration. This suggested that under nonenergized conditions, the apparent uptake largely represented nonspecific binding.

Having established that MRP4 is competent in the MgATP-dependent transport of MTX and knowing from previous studies performed in our laboratory that two other MRPs (MRP1 and MRP3) that have this capability also have the facility for transporting physiological folates (39), we next analyzed the capacity of MRP4-enriched membrane vesicles to catalyze the uptake of folic acid, which MTX closely resembles, and N⁵-formyl-THF (leucovorin), a reduced 1-carbon bearing folate.

From these experiments, it was determined that folates are indeed susceptible to MRP4-mediated transport (Fig. 3). When measured at initial concentrations of 100 μm and over the first 5 min of the assay, [³H]folic acid and [³H]N⁵-formyl-THF were taken up from medium containing MgATP by MRP4-enriched vesicles at rates of 142 and 153 pmol/mg/min, respectively. By contrast, the corresponding rates for the same membranes under nonenergized conditions were 67 and 56 pmol/mg/min, respectively, for folic acid and N⁵-formyl-THF and <44 pmol/mg/min for control membranes under either energized or nonenergized conditions.

The kinetic parameters for MRP4-mediated transport of MTX and folates were measured to gain insights into its potency as a MTX resistance factor as well as its potential for influencing folate homeostasis.

The substrate concentration dependence of MgATP-energized [³H]MTX, [³H]folic acid, and [³H]N⁵-formyl-THF transport by MRP4-enriched membrane vesicles approximated Michaelis-Menten kinetics. The initial rates of MgATP-dependent uptake of all three compounds, enumerated as the difference between uptake in medium containing MgATP and uptake in medium containing MgAMP, and measured over a broad range of substrate concentrations, exhibited saturation kinetics (Fig. 4). Nonlinear least squares fitting of the data to the Michaelis-Menten equation (41) yielded K_m and V_max values of 0.22 ± 0.01 mm and 0.24 ± 0.05 nmol/mg/min, 0.17 ± 0.02 mm and 0.68 ± 0.14 nmol/mg/min, and 0.64 ± 0.23 mm and 1.95 ± 0.18 nmol/mg/min for MTX, folic acid, and N⁵-formyl-THF, respectively (Table 1). The efficiencies of transport (V_max/K_m) fell in the rank order folic acid (4.0) > N⁵-formyl-THF (3.0) > MTX (1.1) (Table 1).

The determination herein that MRP4 is able to transport MTX and folates in combination with our previous study in which cyclic nucleotides and E₂17βG were identified as MRP4 substrates (35) indicates that the substrate range of the pump is quite broad. To gain insight into how MRP4 interacts with its diverse substrates, the inhibitory effects of both substrates and other compounds on MRP4-mediated transport of MTX were examined (Table 2).

In the first instance, the inhibitory activities of five transport substrates were analyzed (Table 1). Of these compounds, E₂17βG exerted the highest degree of inhibition (60.1% at 30 μm). By contrast, the inhibitions exerted by cGMP, cAMP, folic acid, and N⁵-formyl-THF were modest (53.5, 51.5, 55.0, and 28.8%, respectively, at 300 μm concentrations). The higher degree of inhibition exerted by E₂17βG by comparison with MTX, folic acid, and N⁵-formyl-THF is commensurate with their respective K_m values (K_m E₂17βG, 30.3 μm; Ref. 35; Table 1). However, cGMP (K_m, 9.7 μm) and cAMP (K_m, 44.5 μm), whose affinities are either higher than or roughly comparable with that of E₂17βG (35), were by comparison unexpectedly weak inhibitors.

The inhibitory activities of three MRP1 reversing agents were examined to test their effects on MRP4 activity. These agents included MK571, a leukotriene D4 receptor antagonist, and two inhibitors of organic anion transporters, probenecid and sulfinpyrazone (17, 42, 43, 44, 45). MK571 was not only the most potent inhibitor of the three MRP1 reversing agents, but it was also the single most efficacious inhibitor of all of the compounds tested (43.6% inhibition at 1 μm, 73.9% inhibition at 3 μm, and complete inhibition at 10 μm; Table 2, and data not shown in Table 2). By contrast, the inhibitions exerted by 300 μm concentrations of probenecid and sulfinpyrazone (50.6 and 76.7%, respectively) were modest.

The inhibitory activities of three P-glycoprotein modulating agents were next examined. These agents included verapamil, a calcium channel blocker, and two bulky lipophilic compounds, PSC833 and cyclosporin A. Of these agents, PSC833 was the most efficacious inhibitor (43.6% at 10 μm). The inhibitions exerted by cyclosporin A (14.5%) and verapamil (46.2%) at concentrations of 10 and 30 μm, respectively, were significantly lower by comparison.

Knowing that MRP4 is able to transport cyclic nucleotides and that inhibitors of phosphodiesterases have been reported to be potent inhibitors of MRP5 (37), we next examined the inhibitory activity of this class of compounds. Three phosphodiesterase inhibitors were selected for analysis, trequinsin, zaprinast, and sildenafil. All three of these agents exhibited significant inhibitory activity (74.3, 65.9, and 54% inhibition at 10 μm concentrations, respectively). Trequinsin was the most potent inhibitor of this group of agents and the second most potent inhibitor overall.

Having determined here that the properties of MRP4 are consistent with it being a component of the cellular MTX efflux system, and in a previous study that MRP1 and MRP3 are similarly able to transport MTX but not polyglutamylated derivatives (39), we next examined the MTX transport properties of MRP2 (cMOAT), the only remaining MRP family member whose expression has been associated with MTX resistance (14) but whose in vitro transport properties with regard to MTX versus MTX polyglutamates has not been determined. For this purpose, membrane vesicles prepared from MRP2-transfected (LLC/cMOAT-1) and parental vector-transfected (LLC/CMV) LLC/PK1 cells were used (10).

Membranes prepared from LLC/cMOAT-1 cells catalyzed the MgATP-dependent uptake of 100 μm [³H]MTX at an initial rate of 61.6 pmol/mg/min at the 5-min time point of the assay, whereas control LLC/CMV membranes did not mediate appreciable transport under energized conditions (Fig. 5,A). However, in strict correspondence with the results obtained with MRP1, MRP3, and MRP4, glutamylation dramatically reduced the capacity of MRP2 to transport MTX, in that 100 μm [³H]MTX-Glu₂ was not transported to any extent by either LLC/cMOAT-1 or LLC/CMV membrane vesicles (Fig. 5 B). Similarly, [³H]MTX-Glu₃ transport was not mediated by MRP2 (data not shown).

In a previous study, we investigated how the in vitro substrate selectivity of MRP4 relates to its ability to confer resistance to antiviral and anticancer nucleotide analogs such as 9-(2-phosphonylmethoxyethyl)adenine and 6-MP and determined that MRP4 is not only a cyclic nucleotide pump that is deployed by the cell for the purpose of effluxing nucleotide analogues, but that it is also competent in the MgATP-dependent transport of the prototypical glucuronate conjugate E₂17βG (35). However, analysis of the drug resistance capabilities of MRP4 also indicates that in addition to nucleotide analogues, the pump also has the facility for conferring resistance to MTX (34), albeit at resistance levels that are 4 to >14-fold lower than those reported for MRPs 1–3 (13, 14), and on this basis it had been inferred that this antimetabolite is also a transport substrate of MRP4. In the present study, we used membrane vesicles prepared from insect cells infected with MRP4 baculovirus and in vitro transport assays to more precisely define the involvement of MRP4 in the cellular pharmacology of MTX and related processes. The results show that MTX is indeed susceptible to MRP4-mediated transport. Moreover, it is shown that addition of a single glutamate residue to MTX abrogates the transport activity of MRP4. Parallel experiments on MRP2 show that glutamylation similarly affects its transport activity. The demonstration that MRP4 and MRP2 are unable to mediate transport of MTX polyglutamates, in combination with a previous study in which we determined that glutamylation also abrogates in vitro transport mediated by MRP1 and MRP3 (39), indicates that all four of the MRP family members that have been implicated in transport of this agent (12, 13, 14, 34, 46) can only mediate the transport of free drug, as had been inferred previously from measurements of free and polyglutamylated drug in cell lines transfected with MRPs 1–3 (13, 14) and from the unusual time dependence of MRP-conferred resistance (12, 13, 14, 34). The inability of these pumps to transport MTX polyglutamates, which are rapidly synthesized in the cytoplasm by the action of FPGS and represent the predominant intracellular species, is significant in that it not only satisfies a critical property of previously described energy-dependent efflux systems for this agent (38) but also constitutes the mechanism by which polyglutamylation engenders metabolic trapping of MTX and an associated massive enhancement in its cytotoxicity. On the basis of this strict selectivity for free drug, it is concluded that at least four MRPs are components of the energy-dependent efflux system for this antimetabolite (Fig. 6). As expected, based upon the structural resemblance of MTX to physiological folates and from our previous determination that MRP1 and MRP3 are able to transport folates, it is shown that MRP4 also has the facility for the low affinity, high capacity transport of folic acid and N⁵-formyl-THF. Although we did not examine this feature of MRP2, its involvement in folate transport has been inferred previously from transport studies using hepatocyte canalicular membrane vesicles prepared from a rat strain that is genetically deficient in this pump (47). In combination, these observations concerning folates, which are also susceptible to polyglutamylation by FPGS, suggest the possibility that their intracellular levels may be influenced by expression of MRPs 1–4.

The results of the present study, in combination with our previous characterization of the in vitro transport properties of MRP4 (35), indicate that the substrate range of this pump is distinguished from that of MRP5 not only by the capacity of MRP4 to transport E₂17βG but also by its ability to transport MTX and folates. Another distinguishing feature was revealed in the present study by the analysis of the inhibitory effects of various compounds on MRP4-mediated transport. We found that MK571, a potent inhibitor of MRP1 transport activity (17), was the single most potent inhibitor of MTX transport by MRP4. This is in sharp contrast to the case with MRP5, for which this compound was reported to have no affect on cGMP transport at concentrations of up to 50 μm (37). This striking difference in the effects of MK571, which resembles a glutathione conjugate (17, 42), may reflect the capacity of MRP4 to transport conjugates, albeit glucuronate conjugates such as E₂17βG. By comparison with MRP4, MRP5 does not appear to have comparable activity with regard to mediating the transport of conjugates in that neither glucuronate nor glutathione conjugates were determined to be in vitro transport substrates of MRP5 in membrane vesicle transport studies (37), although the possibility that the pump has some capabilities with regard to conjugates is suggested by a report in which enhanced cellular efflux of S-(2,4-dinitrophenyl)glutathione was observed in MRP5-transfected cells (33).

In addition to revealing differences between MRP4 and MRP5, analysis of MRP4 inhibitors also demonstrated an interesting similarity between the two pumps with regard to phosphodiesterase inhibitors. The activity of this class of agents, which resemble cyclic nucleotides, as inhibitors of cyclic nucleotide efflux pumps was initially indicated by their ability to attenuate cellular efflux of cAMP (48). More recently, phosphodiesterase inhibitors have been reported to be extremely potent inhibitors of MRP5-mediated cGMP transport (37). In the present study, it is shown that phosphodiesterase inhibitors are also good inhibitors of MRP4-mediated transport, although the degree of inhibition was not comparable with that described for MRP5. The weaker potency of these agents as inhibitors of MRP4 is commensurate with the lower affinity of the pump for cGMP (K_m, 9.7 μm), by comparison with MRP5 (K_m, 2.1 μm; Ref. 35, 37). However, our measurements of inhibition were made in the context of MTX transport, and it is possible that phosphodiesterase inhibitors may be considerably more potent as inhibitors of MRP4-mediated cGMP transport. The susceptibility of MRP4 to inhibition by phosphodiesterase inhibitors may be of substantial pharmaceutical interest in that it suggests the possibility that these agents, which are currently used for the treatment of erectile dysfunction but which also have important potential applications in cardiovascular diseases, may enhance intracellular levels of cGMP not only by inhibiting phosphodiesterases and MRP5-mediated efflux, as suggested previously (37), but by simultaneously inhibiting both of the currently known cGMP efflux pumps (MRP4 and MRP5). This knowledge may be of value in the development of agents that are specifically designed to be potent inhibitors of specific phosphodiesterases, MRP4 and MRP5.

The capability of MRP4 for conferring resistance to MTX in combination with the results of a previous study in which we found that MRP4-transfected NIH3T3 cells are resistant to 6-MP may have special significance for the treatment of childhood acute lymphoblastic leukemia, for which prolonged antimetabolite therapy with these two agents is an important component of remission maintenance therapy. Although our previous observation concerning MRP4 and 6-MP was based upon the analysis of a single transfected cell line and requires confirmation in other MRP4-transfected cell lines, it suggests that induction of MRP4 expression could simultaneously impair both arms of this treatment. Another potential consideration with regard to these two agents pertains to their synergistic cytotoxicity, which has been attributed in part to enhanced incorporation of thiopurine nucleotides into DNA consequent upon MTX-induced reductions in purine pools. Our results suggest that another potential source of synergistic activity might be the inhibition by MTX of MRP4-mediated thiopurine nucleotide efflux (Fig. 6).

In considering the potential for MRP4 to confer clinical resistance to MTX, as well as its capacity to influence folate homeostasis, certain considerations that we discussed in connection with MRP1 and MRP3 should be kept in mind (39). Although the affinities of MRP4 for MTX and folates (Table 1) are higher than those we reported for the same compounds in the case of MRP3 (K_m, 0.62, 1.96, and 1.74 mm, respectively, for MTX, folic acid, and N⁵-formyl-THF, respectively) and for MTX in the case of MRP1 (K_m, 2.15 mm; Ref. 39), the affinities of all of these pumps are quite high by comparison with serum MTX and folate levels (low micromolar concentrations). In addition, in transfected cell lines MTX resistance conferred by MRPs is time dependent in that high levels of resistance are observed only in growth assays in which MTX exposure is limited to the first 1–4 h of the assay (13, 14, 34), and the extent to which this feature conforms to the clinical pharmacokinetics of this agent in patients is unclear. For these reasons, the in vivo pharmacological significance of these pumps requires further investigation.

We thank Dr. Shin-ichi Akiyama (Institute for Cancer Research, Kagoshima University, Kagoshima, Japan) for kindly providing the transfected LLC/PK1 cells and for reviewing the manuscript.