The Amino Terminus of the Human Multidrug Resistance Transporter ABCC1 Has a U-shaped Folding with a Gating Function*

Multidrug resistance is a serious problem in successful cancer chemotherapy. Studies using model cell lines have demonstrated that overexpression of some members of the ATP-binding cassette (ABC) transporter superfamily, such as ABCC1, causes enhanced efflux and, thus, decreased accumulation of multiple anticancer drugs, which leads to increased cell survival. Unlike most other ABC transporters, ABCC1 has an additional membrane-spanning domain (MSD0) with a putative extracellular amino terminus of 32 amino acids. However, the function of MSD0 and the role of the extracellular amino terminus are largely unknown. In this study, we examined the structural folding and the function of the amino terminus. We found that it has a U-shaped folding with the bottom of the U-structure facing cytoplasm and both ends in extracellular space. We also found that this U-shaped amino terminus probably functions as a gate to regulate the drug transport activity of human ABCC1.

The ATP-binding cassette (ABC) 3 membrane transporter superfamily comprises over 1000 members from bacteria to humans, with a wide variety of substrate specificity and function (1,2). The activity of ABC transporters appear to be dependent on ATP binding and hydrolysis. In humans, 49 ABC transporters are known (for a complete list, see, on the World Wide Web, nutrigene.4t.com/humanabc.htm), and some of them (e.g. ABCB1 or Pgp, ABCC1 or MRP1, ABCG2 or BCRP) have been associated with multidrug resistance in cancer chemotherapy (3)(4)(5)(6)(7). Overexpression of ABCB1, ABCC1, or ABCG2 alone in cancer cells causes increased drug efflux, which results in decreased cellular accumulation of anticancer drugs, and thus the cells can survive drug treatment.
Unlike most of the known ABC transporters, ABCC1 has an additional membrane-spanning domain (MSD) at its amino terminus with a domain arrangement of MSD0-MSD1-NBD1-MSD2-NBD2 (8,9). The additional MSD0 consists of five putative transmembrane segments with a predicted extracellular amino terminus (Fig. 1). The functional relevance of the MSD0 to the transporter remains elusive. Although it has been shown that deletion of MSD0 (as little as 66 amino acids from the amino terminus, including the first transmembrane segment) reduced 90% of the LTC 4 transport activity of human ABCC1 (10) and mutation of Cys 7 reduced LTC 4 transport and drug resistance (11), another study showed that the carboxyl core domain (MSD1-NBD1-MSD2-NBD2) of human ABCC1 lacking the entire MSD0 is still functional in transporting substrates, such as LTC 4 (12). It was thought that the loop (L0) linking MSD0 and MSD1 is important for ABCC1 function (10,12,13), and it may be involved in drug binding (14). It has also been reported that the MSD0 may play roles in processing and trafficking of human ABCC1 (13,15).
The predicted extracellular location of the amino terminus of ABCC1 has been confirmed by determining glycosylation status (16,17) by epitope insertion (18,19) and by cell-free expression in microsomal membranes (20). However, this prevailing topological orientation of the amino terminus has recently been challenged by a study using a monoclonal antibody (IU2H10) against the amino terminus of human ABCC1, which suggested that part of the amino terminus was exposed intracellularly (21).
To further investigate the exact topological orientation and function of the amino terminus of human ABCC1, we created another monoclonal antibody, IU5C1, and several constructs with HA and FLAG epitope tags at the amino terminus. We found surprisingly that the amino terminus of human ABCC1 has a U-shaped folding with the bottom of the U structure exposed to intracellular space and that this U-shaped structure plays a gating role for the drug transport function of human ABCC1 by plugging into a putative channel.

EXPERIMENTAL PROCEDURES
Materials-Adriamycin, vinblastine, colchicine, VP-16, anti-FLAG monoclonal antibody, and peroxidase-and fluorescein isothiocyanate-conjugated goat anti-mouse IgG were purchased from Sigma. Lipofectamine and G418 were purchased from Invitrogen. ECL reagent and polyvinylidene difluoride membranes were purchased from Amersham Biosciences and Bio-Rad, respectively. Monoclonal antibodies QCRL-1 and MRPr1 were obtained from Kamiya, whereas the monoclonal anti-HA antibody was from Covance. Synthetic peptides were commercially synthesized by Genemed Synthesis, Inc. All other chemicals were purchased from either Sigma or Fisher.
Engineering of Tagged ABCC1 Constructs-To engineer ABCC1 1FLAG2 , amplification of the template construct Yep-FLAG-MRP1-His (22) was performed with primers 5Ј-GGC-TAGCCCACCGGCATGGACTACAAGGACG-3Ј (forward) and 5Ј-GGAACTCTCTTTCGGCTG-3Ј (reverse). The product was cloned into pCR-Blunt (Invitrogen), and the fragment encoding the amino terminus was released by digestion with NheI and BamHI and used to replace the wild type sequence in the ABCC1 WT construct (21).
To mutate the two N-linked glycosylation sites (Asn 19 3 Gln and Asn 23 3 Gln) at the amino terminus, the QuikChange XL site-directed mutagenesis kit (Stratagene) was used with primers 5Ј-CCGCTCTGGGACTGGCAAGT-CACGTGGCAAACCAGCAACCCCGAC-3Ј (forward) and 5Ј-GTCGGGGTTGCTGGTTTGCCACGTGATTGCCAG-TCCCAGAGCGG-3Ј (reverse). The final mutant ABCC1 construct was named ABCC1 QQ . ABCC1 ⌬N32 (with deletion of the first 32 amino acids) was engineered by amplifying a 1209-bp fragment of ABCC1 using a forward primer carrying a NheI site and a Kozak translation initiation sequence 5Ј-CTAGCTAGCGCCGCCATGTTTCA-GAACACGGTC-3Ј and a reverse primer 5Ј-CACAGACATG-AGGTTGAC-3Ј. The PCR products were then digested with NheI and BamHI to generate a 762-bp fragment, which was subsequently subcloned into ABCC1 WT digested with NheI and BamHI, resulting in ABCC1 ⌬N32 . All of the above mutant constructs were confirmed by double-stranded DNA sequencing.
Generation of Monoclonal Antibody IU5C1-Generation of hybridoma-producing anti-ABCC1 antibodies was performed as described previously (21). The hybridoma-producing monoclonal antibody IU5C1 was collected and further characterized in the same way as described previously for IU2H10 (21).
Transfection and Selection of HEK293 Stable Clones-HEK293 cells were plated and grown to subconfluence. 5 g of each construct encoding the wild type and mutant human ABCC1 was transfected into HEK293 cells using Lipofectamine Plus as previously described (21). 2 days after transfection, cells were split, and 1% of the cells were used for selection with G418 (800 g/ml) for 2 weeks. The G418-resistant clones were picked and then expanded for further analysis of ABCC1 expression using Western blot and FACS. These clones were maintained in medium containing G418 at a concentration of 200 g/ml for further studies.
Enzyme-linked Immunosorbent Assay, Western Blot, Confocal Imaging, and FACS Analyses-Enzyme-linked immunosorbent assay, Western blot, confocal imaging, and FACS analyses were performed exactly as previously described (21). For peptide inhibition, different concentrations of various peptides were mixed with the primary antibody and incubated for 1 h at room temperature prior to probing the blot or saponin-permeabilized cells. For FACS analysis in the presence of vanadate, cells were aliquoted into a microtiter plate at a density of 0.5 ϫ 10 6 cells/well. The cells were then incubated with Adriamycin in the presence of sodium vanadate at 37°C for 1 h. The incubation medium was then removed by centrifugation, and the cells were stained with the standard protocol as described above.
Statistical Analyses-All protein activity data (log-transformed EC 50 ) were fitted into a linear mixed model, in which the correlation among paired data were modeled. All p values were calculated based on this mixed model, and p values less than 5% were considered significant. The advantage of this approach is to allow different experiments to share the same variance information. Model diagnosis was also performed, and it appeared that equal variance and normal assumptions worked well in our data set after the EC 50 values were logtransformed. The analysis was conducted in SAS, PROC MIXED (available on the World Wide Web at www.sas.com/).

Characterization and Epitope Mapping of the Monoclonal
Antibodies IU5C1 and IU2H10-The monoclonal antibody IU2H10 has been characterized, and its epitope has been mapped to 10 amino acids ( 8 SADGSDPLWD 17 ) in the amino terminus of human ABCC1 (21). During our screening of monoclonal antibodies raised against the amino terminus of human ABCC1, we found another clone, IU5C1 of the IgG1 subtype. As shown by both Western blot analysis of membranes isolated from human ABCC1-transfected HEK293 or drug-se-lected MCF7 cells and confocal immunofluorescence analysis of ABCC1-transfected HEK293 cells (Fig. 1B), IU5C1 reacted specifically with human ABCC1 similarly as IU2H10 with the exception that IU5C1 had a lower titer than IU2H10.
To determine whether IU5C1 has a different epitope from IU2H10 in the amino terminus of human ABCC1, we first employed the enzyme-linked immunosorbent assay to determine if the epitope of IU5C1 is located in the amino-terminal 19 amino acids. As shown in Fig. 1C, IU5C1 reacted with the synthetic peptides of 19 amino acids as well as with the recombinant peptide immunogen of 33 amino acids representing the amino terminus of human ABCC1. It also reacted with the synthetic peptides of 10 amino acids ( 8 SADGSDPLWD 17 ) (data not shown). Thus, the epitope of IU5C1 is probably located within these 10 amino acids of human ABCC1, the same as IU2H10 (21). To further map the IU5C1 and IU2H10 epitope, we performed a peptide walking experiment using a series of synthetic peptides of six amino acids each (Fig. 1E). These peptides were then tested for their ability to block the reaction of IU5C1 and IU2H10 to human ABCC1 on Western blot. As shown in Fig. 1D, IU2H10 activity was blocked by peptide 6 ( 12 SDPLWD 17 ), whereas IU5C1 activity was blocked by peptide 8 ( 14 PLWDWN 19 ) in a concentration-dependent manner. Thus, probably the epitopes for IU2H10 and IU5C1 are 12 SDPLWD 17 (peptide 6) and 14 PLWDWN 19 (peptide 8), respectively, and their epitopes overlap with a shift of 2 amino acids.
Membrane Orientation of the IU5C1 Epitope-Previously, we have shown that the monoclonal antibody IU2H10 can stain ABCC1 only in permeabilized cells, suggesting that its epitope in the amino terminus is located intracellularly (21), creating a controversial issue regarding the membrane orientation of the amino terminus of human ABCC1. To determine if the results were not due to the use of the specific antibody IU2H10, we performed a similar study using IU5C1. As shown in Fig. 2A, IU5C1 stained ABCC1-expressing cells only when the cells were permeabilized by saponin or by Triton X-100, similar to that stained by IU2H10 shown previously (21). Live cells and cells transfected with vector alone were not stained by IU5C1. In addition, the IU5C1 staining of ABCC1-expressing cells under permeabilized conditions was completely blocked by the peptide 8 but not by peptide 7 (Fig. 2B), suggesting that the IU5C1 staining was specific to human ABCC1. Thus, the IU5C1 epitope in the amino terminus of human ABCC1 is probably also located intracellularly.
The U-shaped Folding of the Amino Terminus of Human ABCC1-Previously, it has been found that an HA tag engineered following the Arg 4 residue (18) and a FLAG tag engineered following Asn 23 with mutations of glycosylation sites (19) at the amino terminus were both found outside of cells. These observations, together with our findings on the relative membrane orientation of IU2H10 and IU5C1 epitopes, raised a possibility that the amino terminus of human ABCC1 may have a U-shaped folding with the bottom of the U (IU2H10 and IU5C1 epitopes) located intracellularly and the two ends (tags) located extracellularly.
To test this possibility, we first engineered a series of constructs with HA (YPYDVPDYAS) or FLAG tags (DYKDDDDK) inserted at different positions in the amino terminus of human ABCC1 (ABCC1 1FLAG2 , ABCC1 15HA16 , and ABCC1 30HA31 ) (Fig. 3A) in addition to the HA-and FLAG-tagged constructs (ABCC1 4HA5 and ABCC1 FLAG-Q ) published elsewhere (18,19). These constructs were transfected into HEK293 cells, and stable clones with similar expression levels of ABCC1 were selected for further characterization. All newly tagged ABCC1 FIGURE 1. Characterization of monoclonal antibody IU5C1 and IU2H10. A, schematic folding of human ABCC1 with sequence of the putative extracellular amino terminus. The glycosylation sites are indicated by branched symbols and asterisks. B, Western blot and immunofluorescence staining. Isolated membranes were used for Western blot analysis using IU2H10 and IU5C1. ␤-Actin was used as a loading control. R, resistant; S, sensitive. Human ABCC1 WTtransfected HEK293 cells were stained by IU2H10 and IU5C1 for immunofluorescence analysis. The nuclei were counterstained with propidium iodide. C, enzyme-linked immunosorbent assay. The recombinant peptide of 33 amino acids and the synthetic peptide of 19 amino acids were used to test IU5C1. D, epitope mapping. IU2H10 and IU5C1 were first incubated with synthetic epitope peptides at 25 g/ml (top panel) or 5 and 10 g/ml (bottom panel) prior to probing the blots. E, sequence of recombinant and synthetic peptides used for epitope mapping.
constructs were well expressed, as shown by Western blot analyses of isolated membranes (Fig. 3, B and C), and all ABCC1 appeared to traffic normally onto plasma membranes as determined by immunofluorescence analysis of permeabilized cells (Fig. 3D). It is noteworthy that both ABCC1 FLAG-Q and ABCC1 QQ have a smaller size than others on the Western blot ( Fig. 3B) due to a lack of glycosylations in the amino terminus.
We next studied the staining pattern of the tagged ABCC1 in live and permeabilized cells using FACS analyses of different antibodies (Fig. 4). The wild type untagged ABCC1 (ABCC1 WT ) was stained by IU2H10 and IU5C1 only under permeabilized conditions (Fig. 4, compare A with B). Neither HA nor FLAG antibody stained the cells expressing ABCC1 WT (Fig. 4C). ABCC1 with the FLAG tag inserted between Met 1 and Ala 2 (ABCC1 1FLAG2 ) was stained by FLAG antibody under both live (Fig. 4D) and permeabilized conditions (Fig. 4F), suggesting that the amino-terminal end of the amino terminus of human ABCC1 is exposed to the extracellular space. ABCC1 with the HA tag inserted between Arg 4 and Gly 5 (ABCC1 4HA5 ) was also stained by HA antibody in live cells (Fig. 4G), suggesting that the HA tag between Arg 4 and Gly 5 is exposed extracellularly. Similar to that in ABCC1 WT , the IU2H10 and IU5C1 epitopes in both ABCC1 1FLAG2 and ABCC1 4HA5 were detected only following cellular permeabilization (Fig. 4, E and H), suggesting that these original epitopes in both ABCC1 1FLAG2 and ABCC1 4HA5 are located intracellularly as ABCC1 WT .
It is interesting to note, however, that the HA staining of ABCC1 4HA5 in the live cells was less than that in the saponinpermeabilized cells (Fig. 4, compare G with I ). The reduced live staining of HA tag in ABCC1 4HA5 occurs probably because the HA tag is too close to the lipid bilayer and cannot be stained effectively by the HA antibody. Saponin is a plant glycoside consisting of a steroid or a triterpene attached to a carbohydrate chain. It may change the membrane lipid fluidity in addition to its ability to permeabilize cells. Treatment with saponin may have helped expose the epitope by changing the membrane fluidity because of its closeness to the lipid bilayer. If this is the case, staining of ABCC1 4HA5 at 37°C, which also increases membrane lipid fluidity, may be increased compared with staining at 4°C. Indeed, the HA staining of ABCC1 4HA5 at 37°C was increased to a level similar to that in the saponin-treated cells (data not shown). Thus, the membrane fluidity affects the staining of HA tags in ABCC1 4HA5 . On the other hand, the IU2H10 epitope cannot be stained at 37°C (data not shown), ruling out the possibility that the inaccessibility of the IU2H10 epitope was due to the lipid fluidity.
Insertion of the HA tag between Leu 15 and Trp 16 (ABCC1 15HA16 ) disrupts the IU2H10 and IU5C1 epitopes (Fig.  3A). As shown in Fig. 4K, no IU2H10 or IU5C1 staining of the ABCC1 15HA16 -expressing cells under either live or permeabilized conditions were observed, consistent with the disruption of their epitopes by the tag insertion. Interestingly, the staining of ABCC1 15HA16 by HA antibody was observed only under permeabilized conditions (Fig. 4, compare J and L). This observation confirms that the central portion of the amino terminus of human ABCC1, where the IU2H10 and IU5C1 epitopes are situated, is located intracellularly.
Previously, it has been observed that the FLAG epitope inserted between Asn 23 and Thr 24 with mutations of Asn 19 and Asn 23 in the amino terminus to Gln residues (ABCC1 FLAG-Q ) was detected extracellularly by the FLAG antibody (19), suggesting that the amino terminus of human ABCC1 is exposed to the extracellular space. It is of interest to determine the relative membrane location of the IU2H10 and IU5C1 epitopes in ABCC1 FLAG-Q . As shown in Fig. 4M, both IU2H10 and IU5C1 can stain live cells expressing the ABCC1 FLAG-Q mutant, suggesting that the IU2H10 and IU5C1 epitopes in the mutant ABCC1 FLAG-Q are located extracellularly. This finding is intriguing, because the central portion containing the IU2H10 and IU5C1 epitopes in the amino terminus of human ABCC1 FLAG-Q was apparently relocated to the extracellular space by the insertion of FLAG tag at the carboxyl side of the IU2H10 and IU5C1 epitopes. To determine whether this relocation was due to the elimination of the two glycosylation sites at the amino terminus of ABCC1 FLAG-Q , we engineered another construct (ABCC1 QQ ) with mutations of the two glycosylation sites (Asn 19 3 Gln and Asn 23 3 Gln) at the amino terminus without insertion of any tags. Staining of the cells expressing ABCC1 QQ by IU2H10 and IU5C1 occurred only when the cells were permeabilized (Fig. 4, compare S and T). Thus, mutation of the glycosylation sites in the amino terminus did not cause the relocalization of the IU2H10 and IU5C1 epitopes in human ABCC1.
It appears that the tag insertion at the carboxyl-terminal side, but not the amino-terminal side, of the IU2H10 and IU5C1 epitopes causes the relocalization of these epitopes. This difference may be due to the fact that the carboxyl-terminal end is inflexible by attaching to the first TM segment anchored in the membranes. To further test this possibility, we engineered another construct by inserting an HA tag between Thr 30 and Lys 31 (ABCC1 30HA31 ), and a stable cell line expressing this construct was tested. As shown in Fig. 4P, the IU2H10 and IU5C1 epitopes could be detected in live cells expressing ABCC1 30HA31 , suggesting that the IU2H10 and IU5C1 epitopes of ABCC1 30HA31 are located extracellularly and confirming that the tag insertion at the carboxyl side of IU2H10 and IU5C1 epitopes relocates these epitopes. Interestingly, the HA tag in the ABCC1 30HA31 could not be detected by HA antibodies in live cells (Fig. 4P). This observation is probably due to the fact that the HA tag inserted following Thr 30 is too close to membranes, which creates steric hindrance. However, a weak staining was observed when the cells were treated with saponin (Fig. 4R), suggesting that the staining of the tag may be enhanced by increasing the membrane lipid fluidity as for the HA tag in ABCC1 4HA5 . To support this argument, we performed a staining at 37°C as described above for ABCC1 4HA5 and also found a weak staining of live cells (data not shown). Thus, the HA tag inserted between Thr 30 and Lys 31 is close to Membranes isolated from stable cell clones expressing wild type (untagged) and tagged or mutant ABCC1 were used for Western blot analysis using antibody MRPr1. Actin was used as a loading control. C, ABCC1 expression level. The relative intensity of ABCC1 in Western blot was determined using Scion imaging software and normalized using actin as a control. The data were from four experiments. D, immunofluorescence staining. Stable cell clones expressing wild type (untagged) and tagged or mutant ABCC1 were fixed and permeabilized before staining with MRPr1 (for ABCC1 15HA16 ) and IU5C1 (for all others). Vector-transfected cells were used as negative controls.  OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41 the extracellular surface of the membrane lipid bilayer, consistent with the predicted topological structure.
anticancer drugs. As shown in Fig. 5, A-D, all of these clones are resistant to vinblastine, Adriamycin, colchicine, and VP-16 compared with cells transfected with vector alone. Although the resistance factors to all drugs are similar between ABCC1 WT , ABCC1 1FLAG2 , and ABCC1 15HA16 , they all appear to be lower than that of ABCC1 30HA31 (Fig. 5E). Further analysis showed that the resistance factors of ABCC1 30HA31 to all drugs are significantly higher than that of ABCC1 WT after normalizing to their expression levels (Fig. 5F), suggesting that ABCC1 30HA31 is more active in effluxing anticancer drug substrates than ABCC1 WT . Because ABCC1 30HA31 has an extracellularly located amino terminus compared with ABCC1 WT , ABCC1 1FLAG2 , and ABCC1 15HA16 , we hypothesized that the amino terminus of human ABCC1 may function as a gate with a U-shaped folding by plugging into a putative channel in ABCC1 WT . The amino-terminal gate may experience transition between two different conformations, open and closed, as shown in Fig. 7C. Insertion of HA epitope at the carboxyl end of the amino terminus (ABCC1 30HA31 ) may have forced the gate to open, which allows the protein to transport anticancer drugs more efficiently.
To test the above hypothesis, we generated another construct (ABCC1 ⌬N32 ) with deletion of the entire putative aminoterminal gate (32 amino acids). We proposed that deletion of the amino terminus would remove the putative gate, which should have an effect on ABCC1 activity similar to or better than that of ABCC1 30HA31 to allow a more efficient transport of substrates. A HEK293 clone with stable expression of ABCC1 ⌬N32 with correct membrane localization (Fig. 6E) was generated and used to determine drug resistance. As shown in Fig. 6, A-D and F, ABCC1 ⌬N32 appeared to be much more active than ABCC1 WT , with significantly higher resistance factors.
To further test the above hypothesis, we analyzed whether the central portion of the amino terminus relocates to extracellular space (conformational change) to open the channel during the transport process and thus expose the amino terminus extracellularly. For this purpose, we used vanadate as an inhibitor to trap the intermediate stage of a putative conformational change to detect the HA epitope in ABCC1 15HA16 , which may be transiently exposed extracellularly. As shown in Fig. 7A, the staining of HA epitope was observed in live cells in the presence of vanadate, suggesting that HA epitope in the central portion of the amino terminus is relocated from the inside to the outside of cells during the transport process and caught outside of cells by using vanadate. This change was not observed in the absence of vandate or in vector-transfected cells. Similar observations have also been made with the ABCC1 WT using IU2H10 as a probe (data not shown). Thus, it is likely that the central portion of the amino terminus of human ABCC1 experiences a conformational change (gate opening) by relocating from extracellularly inaccessible to accessible during the transport process.
If the amino-terminal sequence functions as a gate and plugs into a putative channel, it is possible that the addition of a synthetic peptide representing the amino-terminal gate may block the function of ABCC1 by irreversibly plugging into the channel. We tested this possibility by preincubating isolated membrane vesicles for 30 min with or without a synthetic 19-amino acid peptide (for sequence, see Syn.Pept (19) in Fig. 1E), followed by analysis of ATP-dependent LTC 4 uptake as previously described (11). As shown in Fig. 7B, the ATP-dependent LTC 4 uptake was completely inhibited by the peptide, suggesting that the peptide may directly interact with and inhibit the function of ABCC1.

DISCUSSION
In this study, we further investigated the membrane orientation of the amino terminus of human ABCC1 and its potential function as a gate. Two monoclonal antibodies, IU2H10 and IU5C1, directed against the central portion of the amino terminus of human ABCC1 were generated and used for determining the membrane orientation of their epitopes. We also engineered HA or FLAG tags into different positions at the amino terminus for determination of the folding of the amino terminus. We found that the amino terminus of human ABCC1 probably had a U-shaped folding with the bottom of the U in cytoplasm and the two ends in extracellular space. This U-shaped folding of the amino terminus may serve as a gate for the function of human ABCC1 by plugging into a putative channel formed by the membrane-spanning domains (Fig. 7C).
The epitopes for both IU2H10 and IU5C1 are located in the central portion of the amino-terminal 32 amino acids. Using peptide walking, we mapped their epitopes within six amino acids each. The epitope for IU2H10 has a sequence of 12 SDPLWD 17 , and the epitope for IU5C1 has a sequence of 14 PLWDWN 19 . The two epitopes overlap with a two-amino acid shift. Generation of two different monoclonal antibodies to this region suggests that this portion of the amino terminus is highly immunogenic and further supports the conclusion that the epitopes are located intracellularly at resting state of the protein. It is also noteworthy that the mutants ABCC1 FLAG-Q and ABCC1 QQ , with the mutation of Asn 19 to Gln, which is the last residue in the IU5C1 epitope, are still reactive to IU5C1 (Fig. 4, M and T). This observation suggests that the last residue Asn 19 in the IU5C1 epitope can be changed to a homologous residue Gln. On the other hand, deleting the residue Asn 19 would abolish its reactivity to IU5C1 as demonstrated by peptide walking (Fig. 1).
Within the ABC superfamily of transporters, only a few members (e.g. ABCC1, ABCC2, ABCC3, ABCC6, ABCC8, ABCC9, and ABCC10) have the MSD0 with a predicted extracellular amino terminus. Among these members, the length of the predicted extracellular amino terminus varies, and the sequence is not well conserved (11). Currently, the functional role of the amino terminus and the MSD0 is not well known. For ABCC1, the core domain lacking the entire MSD0 appears to function normally (12), suggesting that MSD0 is dispensable. On the other hand, removal of the first 66 amino acids (10) or mutation of a single Cys (Cys 7 ) residue (11) essentially knocked out ABCC1 function. Despite these findings, the existence of such a domain in several ABC transporters of different functions suggests that it is important for either structure, function, or both, which we currently do not understand. Indeed, it has been found that MSD0 may contribute to the binding of the substrate LTC 4 and glutathione (14), cellular processing and trafficking of ABCC1 (15), and dimerization of ABCC1. 4 In a previous study, we unexpectedly found that the IU2H10 epitope in the central portion of the amino terminus of human ABCC1 can be stained only in permeabilized cells (21), suggesting that it is located intracellularly. This finding contradicts the prevailing model in which the amino terminus is thought to be extracellular inferred from several previous studies (16 -20).
The results from the current study suggest that the putative Vector-transfected cells were used as a negative control. E, Western blot and immunofluorescence analyses of ABCC1 ⌬N32 expression. The relative intensity of ABCC1 in Western blot was determined using a Scion imaging software and normalized using actin as a control. The data were from four experiments. F, relative resistance factor (RRF) between ABCC1 ⌬N32 and ABCC1 WT for different anticancer drugs after normalization to ABCC1 expression level. VLB, vinblastine; Adr, Adriamycin; Col, colchicine. extracellular amino terminus of human ABCC1 probably has a U-shaped structure. Although the bottom of the structure (central portion containing IU2H10 and IU5C1 epitopes) is exposed to intracellular space, both ends of the U-structure are located extracellularly. The intracellular opening of the putative channel may be big enough for antibodies to access their epitopes in the bottom of the U-structure. Alternatively, the amino terminus of human ABCC1 may sit on the putative channel extracellularly rather than traverse it in such a way that the IU2H10 and IU5C1 epitopes are inaccessible on the outside of cells. However, we think that this possibility is less likely. We clearly showed that the access of IU2H10 and IU5C1 epitopes was membrane permeabilization-dependent using several permeabilization agents (Fig. 2) and that temperature (data not shown) and urea (21) treatment, which both could disrupt the extracellular folding of the amino terminus and membrane fluidity but not membrane permeability, did not increase the staining of these epitopes. Furthermore, similar results were observed using an isolated Fab fragment of IU2H10, 5 suggesting that the size of the antibody was not a cause of its inability to stain its epitopes in live cells. FIGURE 7. Conformational changes of the amino terminus and its gating role. A, membrane reorientation. HEK293 cells harboring vector control or ABCC1 15HA16 were first treated with or without vanadate before they were stained by HA antibody. B, effect of amino-terminal peptide on ATP-dependent [ 3 H]LTC 4 uptake in isolated membrane vesicles. Membrane vesicles from HEK293 cells expressing human ABCC1 WT were pretreated with or without 40 mM synthetic peptide Syn.Pep (19) (see Fig. 1). The vesicles were then pelleted and washed before being used for [ 3 H]LTC 4 uptake analysis in the presence of ATP or AMP. Results shown are means Ϯ S.D. of triplicate determinations from a representative experiment. C, schematic model of the gating role of the amino terminus of human ABCC1. Binding of substrate at the cytoplasmic side may trigger the gate to open. ABCC1 with its amino terminus truncated (ABCC1 ⌬N32 ) or relocated (ABCC1 30HA31 ) is constitutively active.
Thus, the gating mechanism can be another level of regulation. It has been shown previously that the amino-terminal domain of ABCC1 contains a substrate-binding site (14). The binding of substrates to this site may trigger the gate to open. Currently, we are testing this hypothesis.
In summary, we have demonstrated in this study that the amino terminus of human ABCC1 has a U-shaped folding with the central bottom portion facing the cytoplasm and the two ends in extracellular space. The U-structure probably consists of an antiparallel ␤-sheet. Four amino acid residues located in the overlapping IU2H10 and IU5C1 epitopes (six amino acid residues each) in the bottom of the U-structure probably form a ␤ turn. This U-shaped structure may function as a gate by plugging into a putative channel formed by the membrane-spanning domain in the carboxyl core of human ABCC1.