The Multidrug Resistance Protein 5 Functions as an ATP-dependent Export Pump for Cyclic Nucleotides*

Cellular export of cyclic nucleotides has been observed in various tissues and may represent an elimination pathway for these signaling molecules, in addition to degradation by phosphodiesterases. In the present study we provide evidence that this export is mediated by the multidrug resistance protein isoform MRP5 (gene symbol ABCC5). The transport function of MRP5 was studied in V79 hamster lung fibroblasts transfected with a humanMRP5 cDNA. An MRP5-specific antibody detected an overexpression of the glycoprotein of 185 ± 15 kDa in membranes from MRP5-transfected cells and a low basal expression of hamster Mrp5 in control membranes. ATP-dependent transport of 3′,5′-cyclic GMP at a substrate concentration of 1 μmwas 4-fold higher in membrane vesicles fromMRP5-transfected cells than in control membranes. This transport was saturable with a K m value of 2.1 μm. MRP5-mediated transport was also detected for 3′,5′-cyclic AMP at a lower affinity, with a K m value of 379 μm. A potent inhibition of MRP5-mediated transport was observed by several compounds, known as phosphodiesterase modulators, including trequinsin, with a K i of 240 nm, and sildenafil, with a K i value of 267 nm. Thus, cyclic nucleotides are physiological substrates for MRP5; moreover, MRP5 may represent a novel pharmacological target for the enhancement of tissue levels of cGMP.

Lipid membranes are virtually impermeable to cyclic nucleotides, but the appearance of these intracellular signal molecules in blood and urine has been known for decades (1,2). In many cell types formation of 3Ј,5-cAMP has been shown to be accompanied by its secretion from these cells (reviewed in Refs. 3 and 4). Similarly, export of 3Ј,5-cGMP, upon stimulation of guanylyl cyclases in response to nitric oxide or natriuretic peptides, has been demonstrated in many cells, including endothelial cells, vascular smooth muscle cells, kidney epithelial cells, and lung fibroblasts (5)(6)(7)(8). This secretion has been considered as a means of removing the excess of intracellular cyclic nucleotides. Thus, the cellular elimination pathways of these second messengers comprise metabolic degradation by phosphodiesterases, as well as export across the plasma membrane. In addition, a role of extracellular cyclic nucleotides in cell-cell cross-talks has been suggested (8 -10). For example, extracellular cAMP acts as a chemoattractant and differentiation factor in the slime mold Dictyostelium discoideum by binding to surface receptors for cAMP (11). Both cAMP, as well as cGMP, secretion has been shown to be unidirectional and energy-dependent (4 -8, 10). Furthermore, a primary active transport of cGMP has been demonstrated directly in inside-out membrane vesicles from human erythrocytes (12)(13)(14). The transport of both cyclic nucleotides has been shown to be inhibited by probenecid, suggesting that this export is mediated by a transporter for amphiphilic anions (10,13,15).
Members of the multidrug resistance protein (MRP) 1 family have been recognized as export pumps for amphiphilic anions, particularly for conjugates of lipophilic compounds with glutathione or several other anionic residues. The best characterized members with respect to transport function are MRP1 (16), first identified as conjugate export pump in 1994 (17)(18)(19), and the apically localized MRP2, also termed canalicular multidrug resistance protein, cMRP (20), or canalicular multispecific organic anion transporter, cMOAT (21,22) (for reviews see Refs. 23 and 24). The identification of MRP3, MRP4, and MRP5 was mainly based on expressed sequence tag data base analyses (25) followed by cloning of cDNA fragments (26). MRP5 has been shown to be ubiquitously expressed with high transcript levels in brain, skeletal muscle, lung, and heart and only low levels in liver (26 -28). Meanwhile a full-length MRP5 cDNA has been cloned by several groups (27)(28)(29)(30). However, nothing was known so far about the physiological function of MRP5.
In the present study, we stably expressed an MRP5 cDNA, cloned from human brain, in V79 hamster lung fibroblasts. The substrate specificity of the recombinant protein was assessed using isolated membrane vesicles from these cells and control cells. The results identify cyclic nucleotides as physiological substrates of this MRP family member.

EXPERIMENTAL PROCEDURES
Materials- [2, H]cAMP (0.8 TBq/mmol) and  H]cGMP (0.3 TBq/ mmol) were obtained from Hartmann Analytic, Braunschweig, Germany. Unlabeled cyclic nucleotides, 8-bromo-cGMP, N 2 ,2Ј-O-dibutyryl-cGMP, the phosphodiesterase inhibitors Zaprinast and Trequinsin, and the protein standard mixture (26 -180 kDa) for SDS polyacrylamide gel electrophoresis were from Sigma. Sildenafil was extracted from a commercially available 100-mg tablet (Viagra ® , Pfizer; obtained from a local pharmacy) and purified by high performance liquid chromatography using a C 18 Hypersil column as described by Warrington et al. (31). MK * The work on the cloning of MRP5 was supported in part by a research grant from the Deutsche Forschungsgemeinschaft (to G. J.) and a grant from the Wellcome Trust (to B. B.). The stable expression, transport, and inhibitor studies were supported by the Deutsches Krebsforschungszentrum and grants from the Deutsche Forschungsgemeinschaft through Sonderforschungsbereich 601 and 352, Heidelberg, Germany. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (32).
Antibodies-The AMF antibody was raised in rabbits against the 14 carboxyl-terminal amino acids of the deduced MRP5 sequence (AM-FAAAENKVAVKG) coupled to keyhole limpet hemocyanin in a procedure similar to that described previously (20,33).
Cloning of Human MRP5 and Vector Constructions-A full-length MRP5 cDNA was cloned from a human brain 5Ј-STRETCH PLUS cDNA library, prepared from whole cerebral brain of a caucasian male (CLONTECH). Initially, polymerase chain reaction primers were chosen from the partial MRP5 sequence published by Kool et al. (Ref. 26; GenBank TM /EBI accession number U83661) and used for amplification of a 600-bp fragment from the 3Ј-end of the MRP5 sequence. This fragment was used as a probe for screening of this library by plaque hybridization, performed as described (34). The screening yielded a 1.6-kb partial MRP5 clone. Subsequently, the missing 5Ј-half of the full-length sequence was obtained by polymerase chain reaction using gene-specific primers based on the sequence published by Belinsky et al. The complete 4.5-kb cDNA insert was excised from the pGEM ® -T Easy vector and cloned into the NotI site of the mammalian expression vector pcDNA3.1/Hygro (Invitrogen). The correct orientation and integrity of the cDNA in the expression vector was assessed by restriction analysis and sequencing of the cloning site.
Stable Expression in Mammalian Cells-Chinese hamster lung fibroblasts V79 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum and 100 units/ml penicillin/streptomycin. The cells were transfected with the pcDNA3.1/Hygro-MRP5 cDNA construct or the vector only using FuGENE 6 transfection reagent (Roche Molecular Biochemicals). After 48 h, the cells were split, and stable transfectants were selected using medium containing 600 g/ml hygromycin B (Invitrogen). Resistant clones were screened by Northern blot and immunoblot analyses for MRP5 expression. Sodium butyrate (5 mM) was added to the cells 40 h before harvesting to enhance the expression of the recombinant protein (36,37).
Northern Blot Analysis-Total RNA (30 g), isolated from transfected cells using the RNeasy kit (Qiagen), was fractionated on a 1.2% formaldehyde/agarose gel and transferred to Duralon UV membranes (Stratagene). A 32 P-labeled 970-bp fragment of the MRP5 cDNA or a ␤-actin control probe were used for detection. Membranes were hybridized and washed with high stringency as described (34).
Immunoblot Analysis-Membrane fractions were diluted with sample buffer und incubated at 37°C for 20 min prior to separation on a 7.5% SDS polyacrylamide gel. Immunoblotting was performed using a tank blotting system (Bio-Rad) and an enhanced chemiluminescence horseradish peroxidase detection system (NEN Life Science Products).
Vesicle Transport Studies-Plasma membrane vesicles from transfected V79 cells were prepared from hypotonically lysed cells as described (38). ATP-dependent transport of 3 H-labeled substrates into inside-out membrane vesicles was measured by rapid filtration through nitrocellulose filters essentially as described (38). Membrane vesicles (50 g of protein) were incubated in the presence of 4 mM ATP, 10 mM MgCl 2 , 10 mM creatine phosphate, 100 g/ml creatine kinase, and labeled substrate, in an incubation buffer containing 250 mM sucrose, and 10 mM Tris/HCl, pH 7.4. The final incubation volume was 75 l. The substrate and inhibitor concentrations are given in the respective figure legends. For inhibition studies compounds were added from a stock solution in an appropriate solvent (dimethyl sulfoxide or ethanol, final concentration Ͻ 0.5% v/v), and identical concentrations of the vehicle were used in control samples. Aliquots (20 l) of the incubations were taken at the times indicated, diluted in 1 ml of ice-cold incubation buffer, and filtered immediately through nitrocellulose filters (0.2 m pore size, pre-soaked in incubation buffer). Filters were rinsed with 5 ml of incubation buffer, dissolved in liquid scintillation fluid, and counted for radioactivity. In control experiments, ATP was replaced by an equal concentration of 5Ј-AMP. Rates of net ATP-dependent transport were calculated by subtracting values obtained in the presence of 5Ј-AMP as a blank from those in the presence of ATP. For determination of kinetic constants, transport rates were measured at five different substrate concentrations (0.5-10 M for cGMP and 20 -500 M for cAMP). K m values were determined as substrate concentration at half-maximal velocity of transport under these conditions. Similar results were obtained by use of double-reciprocal plots and direct curve fitting to the Michaelis-Menten equation.

RESULTS
Heterologous Expression of MRP5 in V79 Cells-Human MRP5 cDNA was cloned from a human brain cDNA library and characterized as described under "Experimental Procedures." A stable clonal cell line (V79-MRP5) was established after transfection of Chinese hamster V79 fibroblasts with the MRP5-vector construct and selection in hygromycin B. A control hygromycin-resistant clone was obtained from transfection with the parental vector (V79-Co). Expression of MRP5 was analyzed first on the mRNA level by Northern blotting (Fig. 1A) performed on total RNA isolated from these cell lines. A transcript of approximately 4.5 kb was detected only in the MRP5transfected cells under high stringency conditions using a 970-bp fragment of the human MRP5 cDNA as a probe. Further, immunoblot analysis was performed on crude membranes (100,000 ϫ g Pellets) and purified plasma membrane vesicles (Membranes) using the polyclonal antibody AMF directed against the carboxyl terminus of MRP5 (Fig. 1, B and C). This antibody specifically detected a diffuse band, characteristic for glycosylated proteins, at 170 -200 kDa. This is consistent with the molecular mass of about 220 kDa reported for an enhanced green fluorescent protein-MRP5 fusion protein (28). A protein with the same apparent molecular mass was detected also in the control cells and was assumed to represent the endogenous hamster Mrp5. The detection signal was, in all V79-MRP5 membrane preparations, about 5-to 8-fold stronger than in the control cells.
In addition, a significant signal was obtained in membranes of human erythrocytes, indicating the expression of MRP5 in red blood cells (Fig. 1C). The antibody showed no cross-reactivity with human MRP1, MRP2, MRP3, or MRP6 (not shown).
ATP-dependent Transport of Cyclic Nucleotides into Membrane Vesicles from Transfected V79 Cells-ATP-dependent transport of [ 3 H]cGMP and [ 3 H]cAMP, which proceeded into the fraction of inside-out-oriented vesicles, was studied during a 10-min period (Fig. 2). ATP-dependent transport (Fig. 2, left panels) was calculated by subtracting the vesicle-associated radioactivity in the presence of 5Ј-AMP from the values obtained in the presence of ATP. ATP-dependent [ 3 H]cGMP accumulation at the standard concentration of 1 M was 2.3 Ϯ 0.6 pmol ϫ mg protein Ϫ1 at 10 min in vesicles from vector-transfected control cells (V79-Co), which exhibited expression of endogenous Mrp5 (Fig. 1, B and C). The membranes from MRP5-transfected cells (V79-MRP5) showed a 4-fold higher ATP-dependent transport with 9.3 Ϯ 0.9 pmol ϫ mg protein Ϫ1 at 10 min (Fig. 2, upper panels). The amount of [ 3 H]cGMP taken up by the vesicles was markedly decreased by increasing the osmolarity of the extravesicular medium, indicating transport into the intravesicular space. At a concentration of 1 M sucrose (outside the vesicles) the transport rate was 22.2 Ϯ 3.3% (mean Ϯ S.D., n ϭ 3) of the value obtained under standard conditions with 250 mM sucrose. The rate of ATP-dependent transport of [ 3 H]cAMP at the same concentration (1 M) was 2.2 Ϯ 0.4 pmol ϫ mg protein Ϫ1 at 10 min (Fig. 2, lower panels) with V79-MRP5 vesicles. This transport was below the detection limit with membranes from control cells under these conditions and substrate concentration (1 M). At a concentration of 1 M, ATP-dependent transport of [ 3 H]cGMP in V79-MRP5 was about 4-fold higher than [ 3 H]cAMP transport (Fig. 2, lower  right panel). The higher affinity of MRP5 to cGMP is reflected in an apparent K m of 2.1 Ϯ 0.2 M for cGMP compared with a K m value of 379 Ϯ 24 M for cAMP (Fig. 3). The transport efficiency (V max /K m ) was 2100 l ϫ mg protein Ϫ1 ϫ min Ϫ1 for cGMP and only 90 l ϫ mg protein Ϫ1 ϫ min Ϫ1 for cAMP.
Inhibition of MRP5-mediated cGMP Transport-Inhibition studies with several amphiphilic anions are summarized in Table I MRP5 construct (28). Probenecid, a commonly used inhibitor of organic anion transporters, inhibited cGMP transport at the same concentration (50 M) by about 68%. Under the conditions used, inhibition by cAMP was detected only at relatively high concentrations (above 100 M). This is in line with the relatively high K m value for this compound (Fig. 3). MK571, a leukotriene receptor antagonist, which inhibits MRP1-mediated transport in sub-micromolar concentrations (17), had no inhibitory effect on MRP5-mediated transport in concentrations up to 50 M. Several compounds structurally related to cGMP (Fig. 4) and currently used as phosphodiesterase inhibitors were identified as potent inhibitors of cGMP transport. Trequinsin inhibited MRP5-mediated cGMP transport competitively with a K i value of 240 nM (Fig. 3). A similarly potent inhibition was observed with sildenafil, with a K i value of 267 nM (figure not shown).
Transport of Glutathione and Glucuronate Conjugates-Because glutathione and glucuronate conjugates are high affinity substrates for MRP1, MRP2, and MRP3 (24), transport of these compounds was measured in V79-MRP5 and control membranes. No significant MRP5-mediated transport of leukotriene C 4 could be detected in incubations with

DISCUSSION
Cyclic GMP has emerged as a major focus in signal transduction research. Mediating most of the effects of nitric oxide, it plays an important role in smooth muscle relaxation, neutrophil degranulation, inhibition of platelet aggregation, neural communication in the brain, and several other physiological processes (for review see Refs. 39 and 40). On one hand, cellular cGMP levels are determined by the rate of synthesis by guanylyl cyclases and, on the other hand, by the elimination rate. Elimination pathways comprise degradation by phosphodiesterases, as well as active export into the extracellular space (Fig. 5). This is in line with the observation that in cerebral cells and platelets, after stimulation with nitric oxide, the cGMP accumulation is decreased faster than it could be explained solely by the phosphodiesterase activity (41). The knowledge about the structure and function of the protein families involved in cGMP synthesis and degradation has grown vastly the last years, whereas little was known so far about the molecular identity of the proteins that mediate the cellular export. Studies in membrane vesicles from human erythrocytes suggested that cGMP is transported by an organic anion transport ATPase (12)(13)(14). The reported characteristics of this transport system in erythrocytes, including the K m value of 2.4 M for cGMP (13) and the lack of affinity for the MRP1 substrate leukotriene C 4 (14), are similar to our findings on MRP5-mediated cGMP transport (see Fig. 3, Table I, and "Inhibition of MRP5-mediated cGMP Transport"). There is only some discrepancy with respect to cAMP transport. Based on the observation that 100 M of cAMP caused only a poor inhibition of cGMP transport, it was concluded that the export system for cGMP was not shared with cAMP. We could observe MRP5mediated transport of [ 3 H]cAMP, but with a more than 20-fold lower transport efficiency (V max /K m ) than that for cGMP (see Fig. 3 and the second paragraph under "Results"). Therefore,  4. Inhibitors of MRP5-mediated transport and comparison of their structure with that of cAMP and cGMP.
FIG. 5. Proposed scheme of the regulation of intracellular cGMP levels. The cellular accumulation of cGMP in response to nitric oxide (NO) is determined by a balance between the activities of soluble guanylyl cyclases (sGC), which catalyze the formation of cGMP from GTP and the elimination of cGMP by metabolic degradation by 3Ј,5Јcyclic nucleotide phosphodiesterases (PDE), as well as by ATP-dependent export by MRP5. Intracellular cGMP receptors of the downstream transduction pathways include cGMP-dependent protein kinases (PKG) and cGMP-gated ion channels. Compounds like sildenafil, trequinsin, and zaprinast can enhance intracellular cGMP concentrations by a dual action on PDEs and ATP-dependent export. The cGMP pumped into the extracellular space may function, in addition, in cell-cell cross-talks.
we obtained also a poor inhibition of cGMP transport by cAMP at concentrations below 100 M ( Table I). Because of the high K m value for cAMP (379 M, Fig. 3), MRP5-mediated cAMP transport may only be significant under conditions with high intracellular cAMP levels. Because of the reported transport activity, we analyzed MRP5 expression in erythrocyte membranes and obtained a significant signal with the MRP5-specific AMF antiserum (Fig. 1C). This indicates that MRP5 represents at least a major constituent of the cGMP export in human erythrocytes. Secretion of cGMP in response to nitric oxide has been also observed in rat lung fibroblasts (8). This is in line with our observation of a basal cGMP transport in our parental or vector-transfected control cells and the detection of a glycoprotein recognized by the MRP5-specific antibody in the control cells (Fig. 1, B and C).
Little was known so far about the function of MRP5. Studies in human embryonic kidney cells transfected with a green fluorescent protein-MRP5 construct indicated that it mediates export of the anionic dye fluorescein diacetate ATP-dependently, but glutathione-independently (28). This is in agreement with the inhibition of cGMP transport by this compound (Table  I). Unlike MRP1 and MRP2, MRP5 seems not to confer resistance against anthracyclines, Vinca alkaloids, and epipodophyllotoxins (28,29). However, MRP5-mediated low level resistance against thiopurines has been reported (29). Comparison of hydropathy profiles of MRP5 with other members of the MRP family indicated that its structure is most similar to that of MRP4. Both lack the hydrophobic extension of about 200 amino acids (first five transmembrane domains) present in MRP1, MRP2, and MRP3 (27,42). MRP4 has been found to be overexpressed in T-lymphoid CEM-r1 cells resistant to nucleosidebased antiviral drugs, and an energy-dependent efflux of the nucleoside phosphonate 9-(2-phosphonyl-methoxyethyl)adenine (PMEA) and of azidothymidine monophosphate from these cells has been demonstrated (43). Very recently, an MRP5-mediated efflux of PMEA and 6-thioinosine from transfected human embryonic kidney cells and Madin-Darby canine kidney cells has been reported (44). In these cells an enhanced efflux of S-(2, 4dinitrophenyl)glutathione and of glutathione across the basolateral membrane was observed. In isolated membrane vesicles, we could not detect a significant MRP5mediated transport of the MRP1 and MRP2 substrates leukotriene C 4 , 17␤-glucuronosyl estradiol, and glutathione disulfide. In addition, cGMP transport was not inhibited by the cysteinyl leukotriene analog MK571, a potent inhibitor of MRP1-and MRP2-mediated transport. This, however, does not exclude MRP5-mediated transport of other glutathione and glucuronate conjugates or complexes or transport of these compounds at high substrate concentrations.
Our present study identifies for the first time natural cyclic nucleotides as substrates of an ATP-binding cassette transporter of the MRP family. cGMP and cAMP are also the first phosphate substrates for which ATP-dependent transport by an MRP family member was demonstrated directly in isolated membrane vesicles. Other MRP family members, especially MRP4, may function as transporters for cyclic nucleotides, as well. Based on RNase protection assays, however, expression of MRP4 is low in most tissues under normal conditions, whereas MRP5 is abundant in almost every organ, with low expression only in liver (26). Because cGMP is the high affinity substrate for MRP5, this transporter may be a novel pharmacological target to interfere especially with the cGMP elimination from cells and enhance the intracellular cGMP concentration under various pathophysiological conditions, especially to relax vascular smooth muscles in treatment of angina pectoris, arterial hypertension, or erectile dysfunction. We could already demon-strate a relatively potent inhibition of this transport by several compounds known as phosphodiesterase inhibitors, including sildenafil as the most prominent one. Thus, these compounds can enhance intracellular cGMP levels through a dual action on cGMP degradation and export (Fig. 5). Compounds, inhibiting more specifically the transport system, may be identified in future studies and could be useful to determine the role of the ATP-dependent export in signal termination.