Localization of a substrate binding domain of the human reduced folate carrier to transmembrane domain 11 by radioaffinity labeling and cysteine-substituted accessibility methods.

The human reduced folate carrier (hRFC) mediates the membrane transport of reduced folates and classical anti-folates into mammalian cells. RFC is characterized by 12 transmembrane domains (TMDs), internally oriented N and C termini, and a large central linker connecting TMDs 1-6 and 7-12. By co-expression and N-hydroxysuccinimide methotrexate (Mtx) radioaffinity labeling of hRFC TMD 1-6 and TMD 7-12 half-molecules, combined with endoproteinase GluC digestion, a substrate binding domain was previously localized to within TMDs 8-12 (Witt, T. L., Stapels, S. E., and Matherly, L. H. (2004) J. Biol. Chem. 279, 46755-46763). In this report, this region was further refined to TMDs 11-12 by digestion with 2-nitro-5-thiocyanatobenzoic acid. A transportcompetent cysteine-less hRFC was used as a template to prepare single cysteine-replacement mutant constructs in which each residue from Glu-394 to Asp-420 of TMD 11 and Tyr-435 to His-457 of TMD 12 was replaced individually by a cysteine. The mutant constructs were transfected into hRFC-null HeLa cells. Most of the 50 single cysteine-substituted constructs were expressed at high levels on Western blots. With the exception of G401C hRFC, all mutants were active for Mtx transport. Treatment with sodium (2-sulfonatoethyl) methanethiosulfonate (MTSES) had no effect on hRFC activity for all of the cysteine mutants within TMD 12 and for the majority of the cysteine mutants within TMD 11. However, MTSES inhibited Mtx uptake by the T404C, A407C, T408C, T412C, F416C, I417C, V418C, and S419C mutants by 25-65%. Losses of activity by MTSES treatment for T404C, A407C, T412C, and I417C hRFCs were appreciably reversed in the presence of excess leucovorin, a hRFC substrate. Our results strongly suggest that residues within TMD 11 are likely critical structural and/or functional components of the putative hRFC transmembrane channel for anionic folate and anti-folate substrates.

The natural folates are water-soluble members of the B-class of vitamins. They act as one-carbon carriers in a series of metabolic reactions leading to the biosynthesis of purines, thymidylate, methionine, histidine, and serine (1). Thus, folates are essential for cell proliferation and tissue regeneration (1). Because folates cannot be synthesized de novo in mammalian cells, external dietary sources are required. Several patho-physiological states are associated with folate deficiency, ranging from cardiovascular disease to neural tube defects and cancer (2, 3).
As hydrophilic anionic molecules, natural folates show only minimal capacities to cross biological membranes by diffusion alone. Accordingly, sophisticated membrane transport systems have evolved to facilitate membrane translocation of these essential cofactors (4). The reduced folate carrier (RFC) 2 is expressed ubiquitously in tissues and tumors (5) and has long been recognized as the major membrane transporter for uptake of reduced folates in mammalian cells and tissues (4). RFC levels are also critical determinants of the anti-tumor activities of anti-folate drugs such as methotrexate (Mtx) and Pemetrexed (Alimta), and impaired transport is a frequent mechanism of anti-folate resistance (6,7).
RFC cDNAs were first cloned in 1994 and 1995 (8 -13). Hydropathy analyses of the amino acid sequences for both the rodent and human RFCs predict an integral membrane protein with up to 12 transmembrane domains (TMDs), internally oriented N and C termini, and a large central linker connecting TMDs 1-6 and 7-12 (see Fig. 1). Much of this topology has now been experimentally confirmed (14 -16).
"Scanning cysteine accessibility methods" or SCAM is now the method of choice for characterizing membrane topologies and substrate binding sites in polytopic membrane proteins (29 -33). Typically cysteines are inserted into a "cysteine-less" template and functional cysteine mutants are expressed in a suitable cell model. Substrate binding domains within the aqueous accessible membrane "channel" can identified from the loss of activity upon treatment with membrane imper-meable hydrophilic alkylthiosulfonates (32), and surface accessibilities can be established by reactivities with impermeable maleimides (34). Recent hRFC studies with SCAM have corroborated findings of mutant studies that implicated amino acids flanking TMD 1 (e.g. Gly-44, Ile-48) as functionally important and established the membrane topology for TMDs 9 -12 (16,35,36).
In this report, we expand upon our recent studies of hRFC structure and function (16,28,35) and use radioaffinity labeling of hRFC halfmolecules with 2-nitro-5-thiocyanatobenzoic acid (NTCB) digestions, and SCAM with sodium (2-sulfonatoethyl) methanethiosulfonate (MTSES) treatments, to localize a critical substrate binding domain to within TMD 11. This is the first report to definitively localize a substrate-binding region within a specific TMD segment of the hRFC molecule.
Mutagenesis-Single cysteine-substituted hRFC mutants were generated by site-directed mutagenesis using the QuikChange TM kit (Stratagene). Primers for generating Cys substitutions were designed on the Stratagene web site. Sequences for the mutation primers are available upon request. hRFC myc-his6 Cys-less in pcDNA3.1 (35) was used as template to generate the single cysteine-replacement mutant constructs in which each residue from Glu-394 to Asp-420 of TMD 11 and Tyr-435 to His-457 of TMD 12 was replaced individually by a cysteine residue. All mutations were confirmed by DNA sequencing.
hRFC myc-his6 Cys-less and the single cysteine-substituted hRFC myc-his6 constructs (see below) were transfected into transport-defective R5 cells with Lipofectamine Plus reagent (Invitrogen). Typically, R5 cells (2.2 ϫ 10 6 ) were plated in 100-mm dishes, in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum 24 h prior to transfection. Cells were transfected with 15.6 g of plasmid DNA, per the manufacturer's instruction. Cells were harvested after 48 h for the preparation of plasma membranes and Western blotting. For other experiments (MTSES treatments, transport assays), cultures were split 24 h after transfection and assayed after an additional 24 h.
Membrane Transport Assays, MTSES Treatment, and Leucovorin Protection-The uptake of [ 3 H]Mtx (0.5 M) was measured over 2 min at 37°C in 60-mm dishes in an anion-free buffer (20 mM Hepes, 235 mM sucrose, pH 7.3).The uptake of [ 3 H]Mtx was quenched with ice-cold Dulbecco's phosphate buffered saline (PBS). Cells were washed with ice-cold PBS (3ϫ) and solubilized with 0.5 N NaOH. Levels of intracellular radioactivity were expressed as pmol/mg of protein, calculated from direct measurements of radioactivity and protein contents of cell homogenates. Protein assays were based on the method of Lowry et al. (42).
MTSES treatments were performed as described previously (35). Briefly, transfected R5 cells in 60-mm dishes were washed (3ϫ) with PBS and then treated with 10 mM MTSES at 37°C for 15 min. Reactions were quenched by a quick wash with 2-mercaptoethanol (14 mM in PBS), followed by two additional PBS washes and a single wash with anion-free buffer (see above). [ 3 H]Mtx uptake was assayed as described above. To assess the protective effects of a hRFC substrate from MTSES inhibition, leucovorin (300 M final concentration) was added 5 min before adding the MTSES reagent.
Detection of the TMD 11-Loop Boundary with BM/SM-Thiol-reactive reagents were used for mapping the TMD 11-loop boundary region facing the extracellular side. R5 cells, grown in 60-mm dishes expressing hRFC myc-his6 Cys-less and the F416C, I417C, V418C, S419C, D420C, V421C, R422C, and G423C single cysteine-substituted hRFC myc-his6 mutants were treated with 200 M SM (in PBS) at room temperature (25°C) for 30 min followed by treatment with 200 M BM (in PBS), or BM alone at room temperature for 30 min. The cells were briefly treated with 14 mM 2-mercaptoethanol to remove excess reagent and then washed three times with PBS. Cells were harvested, and the plasma membranes were prepared, as described above. For immunoprecipitations, membrane proteins were solubilized in 1 ml of cell lysis buffer A (50 mM Tris (pH 7.5), 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate). The insoluble material was pelleted (12,000 rpm, 10 min) and the supernatant was precleared for 4 h with protein G-plus-agarose beads. The beads were pelleted (12,000 rpm, 30 s) and the supernatant was incubated with anti-myc antibody and fresh protein G-plus-agarose beads overnight at 4°C. The beads were washed twice with lysis buffer A, twice with lysis buffer B (50 mM Tris (pH 7.5), 500 mM NaCl, 0.1% Nonidet P40, 0.05% sodium deoxycholate), and once with lysis buffer C (50 mM Tris (pH 7.5), 0.1% Nonidet P40, 0.05% sodium deoxycholate). The immunoprecipitated proteins were eluted with 70 l of 3ϫ Laemmli sample buffer (containing 2% (w/v) SDS and 2 M 2-mercaptoethanol). Immunodetection of biotinylated hRFC was with peroxidase-linked streptavidin and Lumi-Light PLUS substrate (Roche Diagnostics), whereas total immunoprecipitated hRFC was detected using hRFC-specific antibody (21) and standard Lumi-Light substrate. Images were recorded on x-ray film with multiple exposures.
Undigested and digested NTCB-treated proteins were fractionated for 28 h on 18% polyacrylamide gels with Tris-Tricine buffer (44). The gels were sliced into 2-mm segments, and the pieces were suspended into 1 ml of Soluene-350 (PerkinElmer Life Science) overnight at room temperature followed by 5 ml of Ready-value scintillation mixture (Beckman-Coulter). Radioactivity was detected on a Model 6500 Beckman liquid scintillation counter.
Confocal Microscopy-Confocal microscopy was performed as described previously (28,35). Briefly, cells were fixed with 3.3% paraformaldehyde/PBS, permeabilized with 0.1% Triton X-100, and  stained with anti-myc antibody, followed by goat anti-mouse IgG conjugated with Alexa Fluor 488 (Molecular Probes). Fluorescence staining was observed with a Zeiss laser-scanning microscope 310 using a 63ϫ water immersion lens.
To further refine the region covalently modified with NHS-[ 3 H]Mtx, plasma membranes from hRFC-TMD1-6 HA /TMD7-12 Myc-His10transfected cells were digested with NTCB, followed by electrophoresis on Tris-Tricine gels. NTCB is a cysteine cutter, which selectively cleaves peptide bonds N-terminal to cysteine residues. For the TMD7-12 halfmolecule, up to five theoretical cuts are predicted, generating a range of fragments from 32 to 117 amino acids and molecular masses of 3,177-12,428 Daltons (TABLE ONE). As shown in Fig. 2, NTCB digestions resulted in a reproducible shift of the 40 kDa hRFC-TMD7-12 Myc-His10 half-molecule to ϳ7 kDa. This species could only have arisen from the 396 -457 peptide, which includes TMDs 11 and 12 of hRFC (Fig. 1).
hRFC myc-his6 wt and hRFC myc-his6 Cys-less Exhibit Normal Membrane Targeting and Transport Function in hRFC-null R5 HeLa Cells-A functional hRFC myc-his6 Cys-less construct was previously prepared and found to restore transport to RFC-null MTXRIIOua R 2-4 Chinese hamster ovary (CHO) cells (35). However, the CHO model was restrictive in that it was necessary to isolate stable transfectants over several weeks and, even then, net hRFC expression levels and transport activities were modest (35). For the present study, wild type hRFC myc-his6 wt and hRF-C myc-his6 Cys-less constructs were transiently expressed in hRFC-null R5 HeLa cells (38), whereupon they were detected at high levels and were efficiently targeted to the cell surface to restore transport activity (Fig.  3). On Western blots, the expressed proteins migrated at ϳ58-80 kDa, similar to results with ectopically expressed hRFC myc-his6 Cys-less and hRF-C myc-his6 wt proteins in CHO cells (35). Expression of the hRFC myc-his6 Cysless protein was ϳ60% of that for the wild type protein; transport was 51% of the wild type. Thus, R5 HeLa cells represent a new high throughput cell model that affords us a means to efficiently expand the scope of our SCAM experiments to probe hRFC ligand binding domains.
Expression and Function of Single Cysteine-substituted hRFC Mutants of TMDs 11 and 12-Based on the finding that NTCB cleavage of the radioaffinity-labeled hRFC-TMD7-12 Myc-His10 half-molecule localized substrate binding to within TMDs 11 and 12 (Fig. 2), we used the hRFC myc-his6 Cys-less construct as a template to prepare consecutive single Cys-substituted mutants from Glu-394 to Asp-420 in TMD 11 and from Tyr-435 to His-457 in TMD 12. The single Cys-substituted mutant hRFC myc-his6 constructs were transiently transfected into transport-impaired R5 cells. A representative Western blot of plasma membrane proteins from the 50 mutant transfectants, along with proteins from hRFC myc-his6 Cys-less transfected and untransfected R5 cells, is shown in Fig. 4, A and B. With only a few exceptions (e.g. F416C), all of the single Cys-substituted constructs of hRFC TMDs 11 and12 were expressed in R5 cells within a 2-3-fold range. Although the migrations of 49 of the hRFC mutants were nearly identical to that for hRFC myc-his6 Cys-less, for D420C, high molecular mass bands, presumably corresponding to aggregated hRFC, were reproducibly seen.
To assess the effects of the cysteine substitutions in TMDs 11 and 12 on hRFC function, [ 3 H]Mtx uptake was measured for 2 min for each hRFC myc-his6 Cys mutant. Results are shown in Fig. 4, C and D and are expressed as a percentage of the uptake for hRFC myc-his6 Cys-less, for comparison with R5 cells. Most of the cysteine substitutions were remarkably well tolerated, as reflected by the significantly increased Mtx uptake over the low residual level for the R5 subline. Indeed, of the 27 . Transport results are expressed as the averages Ϯ range for 2 duplicate experiments. C: R5 cells and R5 transfectants expressing hRFC myc-his6 wt and hRFC myc-his6 Cys-less were fixed with 3.3% paraformaldehyde, permeabilized with 0.1% Triton X-100, incubated with mouse anti-myc primary antibody followed by anti-mouse IgG-Alexa Fluor 488-conjugated secondary antibody, and spun onto microscope slides. Slides were visualized with a Zeiss laser-scanning microscope 310 using a 63ϫ water immersion lens. mutant constructs for TMD 11 and the 23 mutants for TMD 12, only G401C-hRFC myc-his6 was completely inactive despite a high level of expression on Western blots. This implies that Gly-401 may be struc-turally or functionally important to hRFC transport. Interestingly, our data also suggest that Cys substitutions may be somewhat better tolerated in TMD 12 than in TMD 11, as reflected in the relative transport activities of the Cys mutants compared with that of the Cys-less hRFC for each of these regions (i.e. nearly half of the TMD 11 mutants were no more than 50% as active as hRFC myc-his6 Cys-less).
Although MTSES inhibition establishes aqueous accessibilities of reactive thiols, evidence for participation of this region in substrate binding involves protection from MTSES inhibition by excess transport substrate. Accordingly, for the T404C, A407C, T408C, T412C, F416C, I417C, V418C, S419C-hRFC myc-his6 transfectants, the MTSES treatments were repeated in the presence of 300 M leucovorin, an established transport substrate for the carrier. As illustrated in Fig. 6, marked (at least 1.6-fold) protection from MTSES was measured for the T404C, A407C, T412C, and I417C hRFC myc-his6 mutants. A lower level of protection (ϳ15%) by leucovorin from MTSES inhibition was also seen for the F416C hRFC myc-his6 mutant. These results strongly suggest that amino acids localized to TMD 11 contribute to the substrate-binding pocket in hRFC.
Mapping the TMD 11 Extracellular Boundary with BM and SM-R5 cells expressing hRFC myc-his6 Cys-less and cysteine-substituted hRFC myc-his6 proteins spanning the predicted TMD 11 extracellular boundary (i.e. F416C, I417C, V418C, S419C, D420C, V421C, R422C, and G423C; Fig. 1) were treated with the surface-labeling reagent BM, with and without the membrane-impermeable blocking agent SM (45), to map the extracellular boundary for TMD 11. S301C, in which the cysteine substitution is located in the connecting loop between TMDs 7 and 8 (16), was included as a positive control. After the treatments, hRFC myc-his6 proteins were immunoprecipitated with anti-myc antibody and resolved by SDS gels, with detection of surface-biotinylated thiols by peroxidase-linked streptavidin. Although all hRFC myc-his6 proteins, including both the cysteine-less and cysteine insertion mutants, were effectively immunoprecipitated (as reflected in the patterns of hRFC immunoreactivity in Fig. 7B), only the S419C, D420C, V421C, R422C, G423C, and S301C mutants were appreciably labeled with BM, and biotinylation was significantly blocked by SM (Fig. 7A). This establishes the extracellular localizations for positions 419 -423. Conversely, no labeling was detected with BM for hRFC myc-his6 Cys-less and the F416C, I417C, and V418C mutants (Fig. 7A), showing these three residues to be buried in the plasma membrane. Based on the patterns of BM reactivity and protection by SM, the TMD 11 extracellular boundary lies between positions 418 and 419.  . Effects of MTSES treatment on the Mtx uptake by single cysteine-substituted hRF-C myc-his6 mutants of TMDs 11 and 12. R5 cells expressing hRFC myc-his6 Cys-less and single cysteine-substituted hRFC myc-his6 mutants were preincubated with and without 10 mM MTSES for 15 min at 37°C. Cells were washed, and 0.5 M [ 3 H]Mtx uptake was assayed at 37°C for 2 min. For each mutant, uptake is presented as a percentage of the level measured in the absence of MTSES. All transport results are expressed as the average values Ϯ range for two separate experiments. Inhibitions of at least 25% are noted with an asterisk. ND, not detected. A, results are shown for single cysteine-substituted hRFC myc-his6 mutants spanning TMD 11. B, results are shown for single cysteinesubstituted hRFC myc-his6 mutants spanning TMD 12.

DISCUSSION
The present report continues our systematic characterization of the structure and function of hRFC, a physiologically important transporter and a member of the Major Facilitator Superfamily of transporters (46). Major Facilitator Superfamily proteins are represented in animals, plants, fungi, lower eukaryotes, bacteria, and eukaryotic organelles and transport a diverse range of substrates in a uniport, symport, or antiport fashion, including amino acids, neurotransmitters, sugars, vitamins, nucleosides, and organic phosphates (46). They typically contain 400 -600 amino acids and a structural motif composed of two halves, each including of six transmembrane ␣ helices connected by a large hydrophilic loop, with cytosolic N and C termini.
X-ray crystallographic structures of the Major Facilitator Superfamily proteins, lactose/proton symporter (LacY) (47) and the inorganic phosphate/glycerol 3-phosphate antiporter (GlpT) (48), were recently reported at resolutions of 3.5 and 3.3 Å, respectively. Comparison of LacY and GlpT identified highly similar structures, although the sequence identity between GlpT and LacY is only 21% (49). By comparative modeling, this structure is highly conserved among other Major Facilitator Superfamily family members (49). In both the GlpT and LacY structures, hydrophilic cavities accommodate the substrate binding sites formed by helices I, IV, and V of the N-terminal domain, and helices VII and XI of the C-terminal domain (47,48).
In this report, we further refined this labeled region to between amino acids 394 and 457, corresponding to TMDs 11 and 12, by cleaving the radioaffinity labeled TMD7-12 polypeptide adjacent to cysteines by treatment with NTCB. To confirm the roles of TMDs 11 and 12 in substrate binding and/or translocation, we expressed single cysteine hRFC mutants, from Glu-394 to Asp-420 in TMD 11 and from Tyr-435 to His-457 in TMD 12 in a cysteine-less background, in hRFC-null R5 HeLa cells. Except for G401C, all mutants restored Mtx uptake, well in excess of the low residual level in R5 cells. The 49 functional cysteine mutants were further characterized by SCAM analysis with MTSES, to establish aqueous accessibilities and location of the Cys-substituted residues within the putative TMD-spanning channel for translocating folate substrates. Whereas MTSES had no effect on the level of Mtx uptake by the 23 Cys mutants in TMD 12, for TMD 11, Mtx uptake by 8 Cys mutants (T404C, A407C, T408C, T412C, F416C, I417C, V418C, and S419C) was markedly inhibited by MTSES. With the T404C, A407C, T412C, and I417C mutants, excess leucovorin afforded appreciable protection from the inhibitory effects of MTSES.
For the LacY and GlpT models, the eight TMDs that form the hydrophilic cavity contain a disproportionate number of glycine and proline residues, well recognized "helix breakers" that result in "irregular" or bent ␣ helices. These structures are believed to provide the necessary flexibilities to assume the different conformations required for substrate translocation (50). A similar consideration applies to TMD 11 of hRFC. Thus, the peptide from Val-402 to Thr-415 is predicted to assume a classical ␣ helical secondary structure, with 3.6 residues/turn (Fig. 8). From a helical wheel model for these 14 residues, the MTSES reactive positions 404, 407, 408, and 412 are all predicted to lie on one (aqueousaccessible) face of a putative helix (Fig. 8). However, position 411, also predicted to lie on the same side of the ␣ helix and implicated as functionally important (27), was apparently unreactive with MTSES. From the patterns of BM reactivity and SM protection, the TMD 11 extracellular boundary lies between Val-418 and Ser-419. Although positions 416 -420 are predicted to exist as a random coil within the plasma membrane, the modifications of F416C, I417C, and V418C by MTSES nonetheless establish their aqueous accessibilities and contributions as structural components of the predicted hRFC transmembrane channel. Because Ser-419 is also accessible to BM and SM, the modification of S419C by MTSES must reflect its localization to the TMD 11 exofacial loop.
Based on their protection by leucovorin, positions 404, 407, 412, and 417 should participate or be spatially juxtaposed to residues involved in FIGURE 8. Predicted secondary structure and helical wheel analysis of TMD 11 of hRFC. Upper panel, the secondary structure of the TMD 11 peptide of hRFC from Glu-394 to Asp-420 was predicted with the PHD (Profile network prediction HeiDelberg) program (52,53). Abbreviations are: H, helix; E, random coil. Lower panel, the helical wheel for amino acids Val-402 to Thr-415 was drawn with a web-based program. FIGURE 7. Mapping the TMD 11 junction with biotin maleimide and stibenedisulfonate maleimide. R5 cells expressing single cysteine-substituted hRFC myc-his6 and hRFC myc-his6 Cys-less proteins were treated with 200 M BM with or without pretreatment with 200 M SM. Membrane proteins were immunoprecipitated by anti-myc antibody and Protein G-plus-agarose beads, followed by Western blotting. Detection of immunoprecipitated proteins was done with streptavidin peroxidase conjugate (A) and hRFC-specific antibody (B), after stripping the polyvinylidene difluoride membrane with 0.2 N NaOH. substrate binding. Although this could also reflect indirect effects of substrate-induced conformational changes at distal binding sites, from our radioaffinity labeling results that localized binding to TMDs 11 or 12, a more likely possibility is that these four residues in TMD 11 directly or indirectly participate in substrate binding, as noted above. Of course, from the lack of leucovorin protection, it would seem that positions 408, 416, 418, and 419 do not directly participate in substrate binding and that the loss of transport activity by MTSES treatment is due to indirect (conformational) effects. However, this result could also partly reflect a loss of reduced folate binding to these mutants.
Of particular interest is Gly-401 which flanks the 402-415 helical stretch of residues and seems to function as a helix breaker (Fig. 8). By analogy with the LacY and GlpT proteins, we propose a critical transport role for Gly-401 through its capacity to facilitate dynamic conformation changes within TMD 11, a notion entirely consistent with our finding that cysteine substitution at position 401 completely abolishes hRFC transport.
In conclusion, our affinity labeling and SCAM results establish an important functional role for TMD 11 of hRFC. As noted above, Lys-411 has been implicated as functionally or structurally important in hRFC transport (27). By random chemical mutagenesis of the murine RFC, mutations of Gly-394 and Ala-400, corresponding to Gly-401 and Ala-407 in hRFC, were associated with anti-folate resistance (51). Our future studies will continue to focus on identification of critical determinants of substrate recognition and translocation for hRFC, an absolute prerequisite to understanding the molecular mechanism of folate and antifolate membrane transport by this physiologically important transport system.