pABC11 (Also Known as MOAT-C and MRP5), a Member of the ABC Family of Proteins, Has Anion Transporter Activity but Does Not Confer Multidrug Resistance When Overexpressed in Human Embryonic Kidney 293 Cells*

Several members of the ABC family of proteins have been implicated in multidrug resistance associated with cancer therapies. A novel member of this gene family, designated pABC11, has been identified using degenerate polymerase chain reaction. The full-length cDNA spans 5881 base pairs and encodes an open reading frame of 1437 amino acids predicted to contain two sets of transmembrane domains and two nucleotide binding domains characteristic of ABC proteins. The nucleotide sequence described herein extends that of three recently reported sequences, MRP5 (Kool, M., de Haas, M., Scheffer, G., Scheper, R., van Eijk, M., Juijn, J., Baas, F., and Borst, P. (1997) Cancer Res. 57, 3537–3547), SMRP (Suzuki, T., Nishio, K., Sasaki, H., Kurokawa, H., Saito-Ohara, F., Ikeuchi, T., Tanabe, S., Terada, M., and Saijo, N. (1997) Biochem. Biophys. Res. Commun. 238, 790–794), and MOAT-C (Belinsky, M., Bain, L., Balsara, B., Testa, J., and Kruh, G. (1998) J. Natl. Cancer Inst. 90, 1735–1741), in the 5′ direction. Northern blot analysis detected five transcripts that were differentially expressed in several tissue types, and the gene encoding pABC11 was mapped to chromosome 3. Confocal imaging of HEK293 cells expressing a green fluorescent protein-pABC11 construct confirmed plasma membrane localization of the fusion protein. Overexpression of pABC11 resulted in reduced labeling with the fluorochromes 5-chloromethylfluorescein diacetate, fluorescein diacetate, and 2′,7′-bis-(2-carboxyethyl)-5 (and-6)-carboxyfluorescein acetoxymethyl ester but not with calcein or rhodamine derivatives, consistent with pABC11 being an anion transporter. Fluorochrome export was ATP-dependent but glutathione-independent. We also show that this export pump does not confer resistance to various classes of cytotoxic drugs but does provide small but significant resistance to CdCl2 and potassium antimonyl tartrate.

The ABC gene family encodes a group of structurally related proteins typically composed of one or two transmembrane domains (containing several membrane spanning regions) and one or two nucleotide binding domains characterized by Walker motifs (A and B) and an ATP-binding cassette signature (1,2). Although structurally homologous, diverse biological functions have been ascribed to different members of this gene family. Some ABC proteins are involved in ion channel formation and/or regulation, such as the sulfonylurea receptors (SUR1, SUR2A, and SUR2B), which form K ATP channels (3), and cystic fibrosis transmembrane conductance regulator, which functions as a chloride channel (4). Other members of this family are known to confer resistance to toxic substances. In Saccharomyces cerevisiae yeast cadmium resistance factor (YCF1) contributes to CdCl 2 , antimony and arsenic resistance and has been shown to transport the glutathione-arsenic complex (5,6). Studies in Caenorhabditis elegans have shown that targeted inactivation of the homologue of human multidrug resistanceassociated protein (MRP), 1 mrp-1, results in increased sensitivity to both CdCl 2 and NaAsO 2 (7). P-glycoprotein (Pgp) and MRP are known to be involved in the resistance of some cancerous cell lines to certain cytotoxic drugs (8 -11). Although these two proteins share only approximately 18% amino acid identity, they are both able to confer resistance to a broad spectrum of cytotoxic agents. There is some overlap in the substrate specificities of Pgp and MRP, although the latter has a preference for more anionic substrates, particularly glutathione conjugates (11). However, with observations that drug resistance can often occur in cells not expressing Pgp or MRP, it has become increasingly apparent that these two drug pumps alone cannot account for all of the drug resistance observed. Recently, other ABC proteins have been implicated, including BCRP in anthracycline resistance and MXR1 and MXR2 in mitoxantrone resistance (12)(13)(14). Additional members of the MRP family have also been described (15,16), and investigations have shown that MRP2 (canalicular multispecific organic anion transporter (cMOAT)), when overexpressed in Madin-Darby canine kidney cells, has drug export activity (17). Therefore, it seems that the resistance profile of a given cell line may well involve the activity of several different efflux pumps, and an understanding of the substrate specificity of candidate drug resistance genes would help to unravel the complexities of multidrug resistance.
In this study, we isolated a novel member of the ABC gene family, designated pABC11, using a degenerate polymerase chain reaction strategy. During the course of our work, an almost identical sequence, MOAT-C, has been published (18); therefore, differences between the two sequences will be high-* 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 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF146074.
‡ To whom correspondence should be addressed. Tel.: 44-1865-747100; Fax: 44-1865-748974; E-mail: nmatthews@yam-res.co.uk. lighted in this paper. Our sequence also extends that of an expressed sequence tag (277145) previously identified as an ABC protein (19), which Kool et al. (15) termed MRP5. Examination of a panel of drug-resistant cell lines revealed that MRP5 was up-regulated in three cisplatin-resistant lines but unchanged in others, leading to the conclusion that the role of MRP5 in this kind of drug resistance is uncertain (15). In this study, we have developed a stable cell line that overexpresses pABC11 (MOAT-C/MRP5), examined the subcellular localization of the heterologous protein, and investigated whether it is able to confer resistance to cytotoxic agents. Our results demonstrate that expression of pABC11 results in increases in fluorochrome transport and resistance to certain heavy metal compounds.

EXPERIMENTAL PROCEDURES
Cloning of pABC11-1 g of human pancreas total RNA (CLON-TECH) was reverse transcribed using Superscript II reverse transcriptase (Life Technologies, Inc.) and an oligo(dT) 16 primer (Perkin-Elmer). This first strand cDNA was used as a template for polymerase chain reaction using degenerate primers NDG1 (5Ј-1) and NDG4 (3Ј-1) (20). AmpliTaq polymerase (Perkin-Elmer) was used with the following cycling parameters: 2 cycles of 94°C for 30 s, 45°C for 30 s, and 72°C for 30 s followed by 33 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 30 s and 1 cycle of 72°C for 5 min. A diagnostic restriction digest (HindIII and BamHI) was used to eliminate cystic fibrosis transmembrane conductance regulator products, and the remaining fragments were cloned into pCR2.1 (Invitrogen) and sequenced using a dRhodamine terminator cycle sequencing kit and a 310 Genetic Analyser (PE Applied BioSystems). A brain cDNA library using first strand cDNA synthesized from 1 g of poly(A) ϩ mRNA (CLONTECH) driven by oligo(dT) with avian myeloblastosis virus reverse transcriptase XL (NBL Gene Sciences) (1 h at 42°C) was constructed and screened with novel fragments. Three sequential rapid amplification of cDNA ends reactions (Brain RACE-ready cDNA, CLONTECH) were required to obtain a full-length cDNA using the following gene-specific primers: 5Ј-GTGCTGCCTACATTCAGCATG-3Ј (nt 2814 -2794); 5Ј-CAGGTG-GATGTGCTTGCCTTCTTC-3Ј (nt 1903-1879) and 5Ј-AAAGCCCAG-CATTGTCCACT-3Ј (nt 515-496). The full-length cDNA was inserted into the EcoRI site of the expression vector pEGFP-C1 (CLONTECH).
Cell Culture and Stable Cell Line Generation-All reagents were from Life Technologies, Inc., unless otherwise stated. HEK293 cells (a gift from Professor M Ashford, University of Aberdeen) were cultured in Dulbecco's modified Eagle's medium with Glutamax-1, sodium pyruvate, and 1 mg/ml glucose supplemented with 50 g/ml streptomycin, 50 units/ml penicillin, 2 mM glutamine, and 10% (v/v) fetal calf serum. Monolayers were dispersed with 0.25% trypsin, 0.02% EDTA in PBS (Imperial). Transfections were performed using LipofectAMINE reagent according to the manufacturer's instructions. Stable cell lines were selected in the above medium supplemented with 2 mg/ml G418 sulfate added 24 h after transfection. Single clones were obtained by limiting dilution and clonal populations viewed for EGFP fluorescence with an Axiovert 135 inverted microscope (Zeiss).
Confocal Imaging-Cells were seeded at approximately 1 ϫ 10 5 cells/well of a six-well tissue culture plate (Nunc) containing 25-mmdiameter round coverslips and incubated overnight in Dulbecco's modified Eagle's medium without phenol red, supplemented as above with the addition of 20 mM HEPES. Cells were transferred to a prewarmed heating stage maintained at 37°C and viewed with an Axiovert 135 inverted microscope (Zeiss) equipped with either a ϫ 40 or a ϫ 63 oil immersion lens (Neofluar). Fluorescent images were captured with a digital camera (Hammamatsu C4792 10-bit gray scale CCD camera). Deconvolved images were obtained using an OpenLab 2.0.3 image analysis package (Improvision) by taking 1-m optical sections through the z axis, and background fluorescence was removed using digital deconvolution.
Chromosomal Localization-A monochromosomal somatic cell hybrid DNA panel (human on a mouse/hamster background) was obtained from Human Genome Mapping Project Resource Center (Hinxton Hall, Cambridge, UK). A primer pair combination that yielded species-specific products was used to amplify the panel: 5Ј-AGGAGCATC-CCAAGGGAAAG-3Ј (nt 417-436) and 5Ј-GAAAGCCACGAAAAAGT-CATACAG-3Ј (nt 543-520). 200 ng of each DNA was used per 25-l polymerase chain reaction with cycling parameters of 95°C for 15 min, followed by 33 cycles of 95°C for 1 min, 55°C for 2 min, and 72°C for 3 min, and a final cycle of 72°C for 10 min.
Cytotoxicity Assay-This was as described previously, with slight modifications (22). Cells were plated in 100 l of medium (2.5 ϫ 10 4 /ml). After overnight culture, cytotoxic compounds were added in 100 l of medium and incubated for a further 48 h. Remaining adherent cells were fixed with formaldehyde (3.5% (v/v) in PBS) and stained with crystal violet (0.1% (w/v) in PBS), and the amount of staining was quantitated at 550 nm absorption in a Molecular Devices plate reader. Each agent was tested at several concentrations with three replicates per concentration, and the percentage of cytotoxicity was calculated from the formula 100(a Ϫ b)/a, where a and b are the mean absorbances of wells without and with the test agent, respectively.
Fluorochrome Uptake as Measured by Flow Cytometry-5-Chloromethylfluorescein diacetate (CMFDA), fluorescein diacetate (FDA), 2Ј,7Јbis-(2-carboxyethyl)-5(and-6)-carboxyfluorescein acetoxymethyl ester (BCECF.AM), and tetramethylrosamine chloride (TMR) were obtained from Molecular Probes; other reagents were obtained from Sigma. Cells were detached by minimal treatment with 0.25% trypsin, 0.02% EDTA and suspended at 10 6 /ml in medium. Unless specified otherwise, 500 l of cell suspension was incubated with 500 l of fluorochrome at 5 M in medium for 20 -30 min at 37°C, washed twice with PBS, and resuspended in 600 l of 1% formaldehyde in PBS. Samples were analyzed by flow cytometry using a FACScan with the Fl1 setting for fluorescein derivatives and Fl2 for TMR and daunomycin. In some experiments, a modified protocol was used to measure efflux in which the cells were labeled at 4°C, washed in cold PBS, and resuspended in medium. Replicate cultures were then analyzed after incubation at 37°C for various times thereafter.
Fluorochrome Uptake as Measured by 96-Well Plate Fluorometer-Cells were plated in 100 l of medium (3 ϫ 10 5 /ml) and cultured overnight, and 100 l of fluorochrome (5 M in medium) was added. After incubation for 30 min at 37°C, wells were washed twice with Dulbecco's modified Eagle's medium without phenol red, fixed in 100 l of 1% formaldehyde in PBS, and analyzed with a Fluostar plate fluorometer (excitation at 485 nm and emission at either 520 nm for fluorescein derivatives or 640 nm for TMR). Assays were set up with four or five replicates, and the data are expressed as the mean Ϯ S.D. after subtraction of background, which was fluorescence exhibited by cells in the absence of fluorochrome.

RESULTS
Cloning of a Novel ABC Protein-A novel ABC-related sequence, pABC11.1 (377 bp in length) containing a putative ABC signature (FSVGERQLLCIARAL), together with a Walker B motif (ILILD), was amplified from pancreatic cDNA. A brain-derived expressed sequence tag clone (H17207) was found to be 97% identical to pABC11.1; therefore, we screened a brain cDNA library in order to obtain a full-length cDNA. The resulting clone together with a series of overlapping rapid amplification of cDNA ends products yielded a cDNA that was  Fig. 1). The largest of these transcripts, species A (Ͼ10 kb), hybridized to both 5Ј and 3Ј probes. Species B (approximately 6.0 kb) was considered to correspond to our cloned cDNA (5.8 kb), as it hybridized to both 5Ј and 3Ј probes. Two additional species, C (5.5 kb) and D (2.4 kb), hybridized to the 5Ј probe but not the 3Ј probe.
The relative proportions of the various transcripts differed between the tissue types tested. For instance, skeletal muscle expresses almost equal quantities of A and B, whereas B is the predominant species in brain. Similarly high levels of A, B, C, and D are seen in fetal liver, whereas only very low amounts of A and B are detected in adult liver. Species E was detected in heart, liver, skeletal muscle, and kidney after hybridization to the 3Ј probe but not the 5Ј probe, and its estimated size is 1.2 kb.
To test whether pABC11 homologues exist in other species, we probed a rat MTN with a fragment corresponding to the least conserved region of this cDNA (probe MID). Even after washing at high stringency, an intense signal was seen in thymus and brain (Fig. 1D), with at least two transcripts crosshybridizing and most similar in size to the human A and B forms.
Using gene-specific primers, we mapped pABC11 to chromosome 3. This confirmed the chromosomal location reported by previous authors (15,18,23).
Establishment of a Stable Cell Line Expressing an EGFP-pABC11 Fusion Protein and Subcellular Distribution of EGFP-pABC11-A stable, clonal cell line HEKc10 was established after transfection of wild type HEK293 cells with an EGFP-tagged pABC11 construct and selection in G418. A control G418-resistant clone that did not overexpress EGFP-pABC11, HEKc5, was also obtained from the same transfection experiment. Cells were viewed by fluorescent microscopy, and digital images were captured. As expected, there was no difference in endogenous fluorescence of HEKc5 cells when compared with wild type HEK293 cells. The majority of fluorescence in HEKc10 cells could be visualized at the plasma membrane (Fig. 2). Some punctate fluorescence was also observed; it most probably corresponds to protein trafficking through the Golgi. The distribution of fusion protein in HEKc10 cells was clearly different from that of wild type EGFP, which could be seen in abundance throughout the cell, including within the nucleus (results not shown). Confirmation that HEKc10 cells were expressing the fusion protein was obtained by fluorescence-activated cell sorter (see Fig. 4) and Western blot analysis (Fig. 2, inset, lane 2). The band visualized on our Western blot was diffuse and was estimated FIG. 1. A, MTN blots probed with the 3Ј probe described under "Experimental Procedures." Three species of transcripts were clearly visible: A, Ͼ10 kb; B, 6.0 kb; and E, 1.2 kb. B, the three MTN blots shown in A were stripped according to the manufacturer's instructions, and a control film was exposed to ensure that there was no residual signal left. Blots were then probed with the 5Ј probe. Two additional transcripts, C and D, were observed. C, immune Northern blot probed with the 5Ј probe. D, rat MTN hybridized to the least conserved region of pABC11 (probe MID). Arrows mark two cross-hybridizing transcripts most similar in size to the human A and B forms. Levels of ␤-actin mRNA are also shown.
to be 220 kDa, which is larger than the predicted molecular mass of 166 kDa. This increase in size and diffuse appearance is characteristic of glycosylated proteins (there are eight potential glycosylation sites in pABC11).
Resistance to Cytotoxic Agents by EGFP-pABC11-overexpressing HEKc10 Cells-Because of the sequence homology of pABC11 toYCF1, which confers resistance to Cd ϩ2 in yeast, we initially assessed the cytotoxic effects of CdCl 2 on HEKc10 cells compared with our control cells. As shown in the single experiment in Fig. 3A, HEKc10 cells demonstrated a slight increase in resistance to CdCl 2 toxicity at 1 M compared with three control clones. Because all three control clones behaved similarly, only HEKc5 was used subsequently. In an additional seven experiments, HEKc10 demonstrated increased resistance to 1 M CdCl 2 relative to HEKc5 (Fig. 3B). HEKc10 cells did not show enhanced resistance to the cytotoxic agents daunomycin, vincristine, mitoxantrone, etoposide, cisplatin, colchicine, chloroquine, CDNB, calcein, sodium arsenite, and sodium arsenate (Table I). However, some resistance to potassium antimonyl tartrate was seen (Table I and Fig. 3C). The data in Fig. 3C are from a representative experiment but were confirmed in two additional experiments. This suggests that EGFP-pABC11 may have transporter activity, but it is clearly distinguishable in specificity from Pgp, MRP, and the less well characterized mitoxantrone and cisplatin transporters (13, 24 -28).
Fluorochrome Accumulation/Efflux by HEKc10 Cells-Labeling with fluorochromes enables a more direct measurement of drug efflux. Initially HEKc5 and HEKc10 cells were incubated with fluorochromes at 37°C and washed, and uptake was measured by fluorometry using either a flow cytometer or 96well plate reader. In this variant of the assay (uptake assay), it is assumed that fluorochrome uptake is passive and that the amount of labeling reflects active transport out of the cell.
As analyzed by flow cytometry, HEKc5 and HEKc10 cells labeled comparably with calcein⅐AM, TMR, or daunomycin (Fig. 4), all of which can be exported by MRP-or Pgp-type transporters (28 -30). Because there were no differences between the clones, EGFP-pABC11 does not seem to transport these agents. However, on labeling with CMFDA, a known substrate for MRP (31), a clear difference emerged, with HEKc10 cells having much reduced labeling (Fig. 4).
These results were confirmed and extended using a 96-well plate assay (Fig. 5). Again, there were no differences between the clones in terms of calcein⅐AM, TMR, or rhodamine 123 labeling, but labeling of HEKc10 cells with CMFDA and FDA was greatly reduced, whereas labeling with BCECF⅐AM was reduced by about 50% (Fig. 5). Several conclusions can be drawn from this. First, EGFP-pABC11 is an organic anion transporter. Second, increasing the negative charge (or size) of the substrate as in BCECF and calcein reduces transport efficiency, as does introduction of a positive charge (TMR). Finally, the observation that HEKc10 cells are labeled equally poorly by FDA and CMFDA shows that the chloromethyl group on CM-FDA is not a structural requirement. This is significant in that it is this group that is required for conjugation to reduced glutathione (GSH) for transport as a GSH conjugate by MRP. Therefore GSH conjugation may not be necessary for transport by EGFP-pABC11, and this is discussed further under "Discussion." Modulation of EGFP-pABC11-mediated Fluorochrome Efflux-Fluorochrome efflux by Pgp can be blocked by verapamil, and MRP-mediated efflux can be blocked by probenecid and sulfinpyrazone (28,32,33). With FDA labeling of HEKc10 cells, verapamil and probenecid had minimal effects, whereas sulfinpyrazone had a small effect (Fig. 6A). In contrast, in HEKc5 cells, which do not overexpress EGFP-pABC11, the MRP blockers probenecid and sulfinpyrazone markedly enhanced FDA labeling, implying that HEK293 cells constitutively express a fluorochrome exporter (perhaps of the MRP type).
The requirement for GSH conjugation in drug transport can be tested by depleting cells of GSH by treatment with the GSH synthesis inhibitor DL-buthionine-(S,R)-sulfoximine (BSO) (28,34). BSO treatment enhanced labeling of HEKc5 cells with CMFDA but not HEKc10 cells (Fig. 6B). For HEKc5 cells, this can be interpreted as follows: GSH depletion inhibits conjugation of the fluorochrome with GSH and its subsequent export by the constitutive GSH conjugate transporter. In the EGFP-pABC11-overexpressing HEKc10 cells, export of CMFDA appears to be much less GSH-dependent, in keeping with the observations above. Similar results were obtained on labeling with FDA rather than CMFDA (data not shown).
Depletion of intracellular ATP resulted in enhanced labeling of HEKc10 cells with FDA (Fig. 6C), consistent with EGFP-pABC11 being an ATP-dependent transporter.
Decreased Fluorochrome Labeling of HEKc10 Cells Is Due to Increased Efflux Rather Than Reduced Uptake-In the above labeling experiments, the assumption has been made that the differences observed in fluorochrome labeling between HEKc5 and HEKc10 clones are due to enhanced efflux by HEKc10 cells. To test this directly, cells were labeled with FDA at reduced temperature (to prevent efflux), washed, and then incubated for various periods at 37°C to measure efflux. When labeling was performed at 10°C, HEKc10 cells showed reduced initial labeling relative to HEKc5 cells. This could be due either to differences in dye uptake or to residual efflux activity in HEKc10 at this temperature. The latter seems more likely because reactions carried out at 4°C resulted in comparable labeling. When efflux was subsequently measured in the presence of probenecid (to block the constitutively expressed transporter), HEKc10 cells lost their label much faster than HEKc5 (Fig. 7), confirming that EGFP-pABC11 is involved in drug efflux. In this experiment, the increase in fluorescence with HEKc5 cells at 7.5-15 min is due to conversion of unhydrolyzed (nonfluorescent) FDA to fluorescent fluorescin. In HEKc10 cells, this increase was not seen, presumably due to the efflux rate of fluorescin being higher than the rate of hydrolysis of FDA to fluorescin. Decreased FDA Labeling of HEKc10 Cells Is Due to Preferential Export of the Hydrolyzed Product Rather Than FDA Itself-Fluorochrome labeling was done in the standard 96-well assay with two modifications. First, phenol red-free medium was used so that changes in fluorescence of the whole well contents (cells ϩ supernatant) could be monitored, and second, probenecid was included in the medium to block the constitutive transporter. If EGFP-pABC11 exports FDA (nonfluorescent) in preference to its hydrolysis product fluorescin (fluorescent), then the whole well contents for HEKc10 should exhibit much lower fluorescence than for HEKc5. If anything, HEKc10 cultures had increased fluorescence (data not shown), indicating that EGFP-pABC11 preferentially exports fluorescin relative to FDA.  A novel member of the ABC family of proteins has been cloned. We predict that the initiating codon begins at nt 198 resulting in a protein of 1437 amino acids. Although the sequence surrounding this putative initiating Met, 5Ј-AA-GAUGA-3Ј, is not identical to a consensus Kozak sequence, 5Ј-ACCAUGG-3Ј, the dominant purine at position Ϫ3 is present, rendering the nucleotides at positions Ϫ2, Ϫ1, and ϩ4 less influential in determining the translational start site. Pairwise alignments indicate that this protein shares similar degrees of homology with several ABC proteins of diverse function, including SUR1, YCF1, and MRP, and that the majority of conserved residues lie within the nucleotide binding domains and the second transmembrane domain, with the NH 2 -terminal transmembrane domain being the most divergent portion of the protein.
Although pABC11 may be closest in amino acid composition to a subgroup of ABC proteins, its mRNA expression profile is very different. For example, cMOAT expression is restricted mainly to the liver (17), whereas multiple transcripts of pABC11 are observed in several tissue types. Although it has not been conclusively proven, we suspect that these multiple species arise through alternative splicing of a single gene, a phenomenon already reported for other ABC proteins, including SUR2 (35)(36)(37), the major histocompatibility complex-encoded peptide transporter Tap2 (38), MRP3 (39), and MRP (40). Valuable but limited information can be gained from the protein sequence, mRNA tissue distribution, and homology searching with respect to the function of this novel ABC protein.
Therefore, to get further insight into its function, the effect of overexpression of the protein in HEK293 cells was investigated.
Overexpression of EGFP-pABC11 fusion protein did not increase resistance to a range of anticancer drugs or arsenic. However, small but statistically significant increases in resistance to CdCl 2 (2.4ϫ) and potassium antimonyl tartrate (2.9ϫ) were seen, although it is debatable whether these increases are biologically relevant. This pattern of resistance differs from that found for overexpression of other ABC proteins. For example, with MRP (24,41), similar levels of resistance to potassium antimonyl tartrate were found, but high levels of resistance to anticancer drugs (in the range 5-25ϫ) were also found. This suggests a difference in specificity between pABC11 and other transporters, but other possibilities must be considered. For example, does EGFP tagging affect function? We cannot answer this question for pABC11 because attempts to express untagged or His-tagged protein were unsuccessful. However, for other ABC proteins, tagging was consistent with function (42)(43)(44). Cell background is another possible variable, but the studies on MRP cited above were also done in HEK293 cells (24). These reservations on the function of overexpressed, tagged protein also apply to its subcellular localization, which may be altered relative to the native protein, as has been seen for MRP (discussed by Tommasini et al. (45)).
Our studies on fluorochrome transport also suggest that pABC11 differs in specificity from MRP and Pgp. Like MRP, EGFP-pABC11 overexpression results in reduced labeling with BCECF⅐AM and CMFDA (31,46): BCECF is also a substrate for Pgp (47). A striking observation in our study is the failure of EGFP-pABC11 to reduce labeling by calcein⅐AM and rhodamine derivatives, both classes of compounds being substrates for MRP and Pgp, albeit with differing efficiencies (29). For EGFP-pABC11, there was a clear structure activity relationship for fluorochrome export (Fig. 5) with an increase in size and/or positive charge resulting in decreased efflux. Like MRP (28), EGFP-pABC11 appears to have a preference for the anionic hydrolysis products of fluorochromes rather than the uncharged ester, which is preferred by Pgp (46). In this context, it is significant that antimonyl tartrate is also an organic anion.
Glutathione is necessary for transport of many substrates by MRP and its homologues (reviewed by Ishizawa et al. (48)) but not for calcein or BCECF (28,46). In our studies, BSO treatment to deplete glutathione did not affect fluorochrome transport by EGFP-pABC11 (but did inhibit activity of the constitutively expressed transporter). This shows that glutathione is not necessary for EGFP-pABC11 efflux of the fluorochromes FDA and CMFDA, but we cannot exclude the possibility that it may be required for other substrates.
Although EGFP-pABC11 is functionally different from MRP, it has more in common with MRP than Pgp, as expected from the amino acid homologies. However, one final piece of evidence highlights the distinction between MRP and EGFP-pABC11. Probenecid and sulfinpyrazone are broadly reactive anion transport inhibitors that effectively block MRP when used in the mM range, but at this concentration, these compounds had minimal effect on EGFP-pABC11.
In conclusion, EGFP-pABC11 is an organic anion transporter that appears to be functionally different from previously described ABC transporters. However, it must be stressed that the studies described here were performed using an overexpression system, and the next step is to see whether they can be confirmed with the native protein in a more physiological setting.