Restoration of Transport Activity by Co-expression of Human Reduced Folate Carrier Half-molecules in Transport-impaired K562 Cells LOCALIZATION OF A SUBSTRATE BINDING DOMAIN TO TRANSMEMBRANE DOMAINS 7–12*

Reduced folates such as 5-methyl tetrahydrofolate and classical antifolates such as methotrexate are actively transported into mammalian cells by the reduced folate carrier (RFC). RFC is characterized by 12 stretches of mostly hydrophobic, (cid:1) -helix-promoting amino acids, internally oriented N and C termini, and a large central linker connecting transmembrane domains (TMDs) 1–6 and 7–12. Previous studies showed that deletion of the majority of the central loop domain between TMDs 6 and 7 abolished transport, but this segment could be replaced with mostly non-homologous sequence from the SLC19A2 thiamine transporter to restore transport function. In this report, we expressed RFC from separate TMD1–6 and TMD7–12 RFC half-molecule constructs, each with a unique epitope tag, in RFC-null K562 cells to restore transport activity. Restored transport exhibited characteristic transport ki-netics for methotrexate, a capacity for trans-stimulation by pretreatment with leucovorin, and inhibition by N hydroxysuccinimide methotrexate, a documented affinity inhibitor of RFC. The TMD1–6 half-molecule migrated on SDS gels as a 38–58 kDa glycosylated species and was converted to 27 kDa by N -glycosidase from separate TMD1–6 and TMD7–12 half-molecule constructs, each with a unique epitope tag, in our well established hRFC-null K562 (K500E) model (32). Our results establish an absolute requirement for both TMD1–6 and TMD7–12 half-molecules for high-level surface expression and restoration of transport activity. Using transfected cells expressing both TMDs 1–6 and 7–12 as separate polypeptides, we covalently labeled the TMD7–12 region with N -hydroxysuccinimide (NHS) [ 3 H]Mtx, a documented radioaffinity ligand for RFC to directly demonstrate an important role for this region in transport substrate binding.

Reduced folates such as 5-methyl tetrahydrofolate and classical antifolates such as methotrexate (Mtx) 1 are actively transported into mammalian cells by the same facilitated transport process termed the reduced folate carrier or RFC (1)(2)(3). For reduced folates, uptake is essential for supplying sufficient cofactors for DNA synthesis and cell proliferation. For Mtx and related antifolates (e.g. Pemetrexed), high levels of transport are necessary to generate sufficient intracellular drug to maximally inhibit dihydrofolate reductase or other enzyme targets and to support synthesis of polyglutamate conjugates (4,5).
The murine and hamster RFCs were cloned in 1994 (6,7) and in 1995 human RFC (hRFC) was cloned and characterized (9 -12). For both the rodent and human carriers, hydropathy analysis of amino acid character predicts 12 stretches of mostly hydrophobic, ␣-helix-promoting amino acids, internally oriented N and C termini, and a large central linker connecting transmembrane domains (TMDs) 1-6 and 7-12 (1, 2, 7) (Fig.  1A). For hRFC, the single N-glycosylation consensus site at asparagine 58 was predicted to be extracellular. This 12 TMD topology model has been experimentally confirmed by a combination of approaches including scanning hemagglutinin (HA) epitope insertion and glycosylation insertion mutagenesis (13,14) and, most recently, by scanning cysteine insertional mutagenesis and accessibility methods (15,16).
In striking contrast to the TMD regions, the 61 amino acids comprising the central connecting loop between TMDs 6 and 7 of hRFC are less conserved (44.3% homology between hRFC and rodent RFCs). Indeed, large deletions in this region of the murine (31 of 66 amino acids) and hamster (45 of 67 amino acids) RFCs preserved membrane targeting and transport activity, as long as a highly conserved stretch of 11 amino acids (corresponding to positions 204 -214 in hRFC) was present (29,30). In hRFC, deletions of 49 or 60 amino acids from the TMD6/7 linker (amino acids 215-263 and 204 -263, respectively) completely abolished transport activity for both Mtx and 5-formyl tetrahydrofolate (5-CHO-H 4 PteGlu) (31), suggesting that some minimal length for the linker is necessary. Further, replacement of the deleted segments with non-homologous 73 or 84 amino acid segments of the structurally analogous thia-mine transporter SLC19A2 (18% homologous to hRFC for the TMD6/7 linker) completely restored transport (31). Thus, with the exception of the conserved 204 -214 amino acid segment, it appears that the primary role of the connecting loop between TMDs 6 and 7 is to ensure the proper spacing between the two halves of hRFC protein for optimal function, and that this is virtually independent of amino acid sequence.
In this report, we further explore the role of the central connecting loop and the relationship between the N-and Cterminal membrane-spanning domains by expressing hRFC FIG. 1. Schematics of hRFC topology and hRFC-TMD1-6 HA and hRFC-TMD7-12 Myc-His10 half-molecules. A shows a computergenerated topology model (8), based on the predicted hRFC amino acid sequence including 12 TMDs, internally oriented N-and C-terminal domains, and a cytosolic loop connecting TMDs 6 and 7. Amino acids are designated by the single letter abbreviations. Conserved amino acids are designated by dark circles. B shows the structure of the hRFC-TMD1-6 HA protein in which the 11 amino acid (YPYDVPDYAVN) HA epitope is inserted at position 226. C shows the structure of the hRFC-TMD7-12 Myc-His10 protein in which a Myc-His10 epitope is inserted at position 537. This construct was also designed to include 18 amino acids (MVPSSPAVEDKQVPVEPG) from the N terminus of hRFC (designated N18).
from separate TMD1-6 and TMD7-12 half-molecule constructs, each with a unique epitope tag, in our well established hRFC-null K562 (K500E) model (32). Our results establish an absolute requirement for both TMD1-6 and TMD7-12 halfmolecules for high-level surface expression and restoration of transport activity. Using transfected cells expressing both TMDs 1-6 and 7-12 as separate polypeptides, we covalently labeled the TMD7-12 region with N-hydroxysuccinimide (NHS) [ 3 H]Mtx, a documented radioaffinity ligand for RFC (33)(34)(35), to directly demonstrate an important role for this region in transport substrate binding.  (36). Restriction and modifying enzymes were obtained from Promega (Madison, WI) or New England Biolabs (Beverly, MA). Synthetic oligonucleotides were obtained from Invitrogen (Carlsbad, CA). Tissue culture reagents and supplies were purchased from assorted vendors with the exception of iron-supplemented calf serum, which was from Hyclone Laboratories, Inc. (Logan, UT).

Reagents
Cell Culture-The Mtx transport-deficient K562 subline, designated K500E, was selected from wild-type K562 cells (American Type Culture Collection) and maintained in complete RPMI 1640 medium containing 10% iron-supplemented calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin, and 0.5 M of Mtx (32). K500E cells were transfected with the wild-type hRFC in pCDNA3 (designated pC43) to generate the K43-6 subline (32), or with a hRFC construct including a C-terminal HA insertion in pCDNA3 (designated hRFC HA12 ) to generate the K43 HA12 subline (14,37). These cells and all other hRFC transfectants (see below) were cultured in complete RPMI 1640 medium with 10% serum and antibiotics, plus 1 mg/ml G418 in a humidified atmosphere at 37°C in the presence of 5% CO 2 .
For experiments in which hRFC-transfected cells were treated with tunicamycin to inhibit N-glycosylation, cells were continuously maintained in 3 g/ml tunicamycin for a minimum of 2 weeks prior to experiment.
For N-glycosidase F (New England Biolabs) digestions of plasma membrane proteins, samples were incubated at 37 o overnight following denaturation for 10 min with 0.5% SDS and 1% 2-mercaptoethanol, followed by addition of 50 mM sodium phosphate (pH 7.5), 1% Nonidet P-40, and N-glycosidase F (1 unit). Samples were diluted with 3ϫ Laemmli sample buffer, fractionated on 10% polyacrylamide gels, and analyzed by Western blotting.
Membrane Transport Assays-Initial rates of [ 3 H]Mtx or (6S)-5-CHO-[ 3 H]H 4 PteGlu uptake were measured over 180 s, as previously described (9,32). Levels of intracellular radioactivity were expressed as pmol/mg protein, calculated from direct measurements of radioactivity and protein contents of the cell homogenates. Protein assays were performed by the method of Lowry et al. (41). Kinetic constants (K t , V max ) were calculated from Lineweaver-Burk plots. To assess the capacity of leucovorin to trans-stimulate Mtx influx via hRFC (32,42), cells were loaded with 50 M leucovorin for 20 min at 37 o , then washed with ice-cold DPBS (3ϫ), and the cell pellets were stored at 0 o . For transport assays, the cell pellets were warmed briefly to 37°and uptake initiated by rapid resuspension into Hank's balanced salts solution Affinity Labeling of hRFC with NHS-Mtx Ester-The preparation of unlabeled and radiolabeled NHS-Mtx was performed as described previously (33,35). The radiospecific activity of the NHS-[ 3 H]Mtx was 28 Ci/mmol. For NHS-Mtx treatments, cells (ϳ5 ϫ 10 7 /ml) were suspended in 20 mM HEPES, 225 mM sucrose, pH 6.8 (with MgO) and were incubated with affinity reagent at 23°(while shaking) for 5 min. Cells were washed with DPBS and assayed for transport (see above) or, for NHS-[ 3 H]Mtx-labeled cells, plasma membranes were prepared by differential centrifugation (35). Membrane pellets were solubilized in 1% SDS and fractionated on 1.5-mm 10% polyacrylamide gels with SDS, the gels sliced into 2-mm segments, and the pieces were suspended into 1 ml of Soluene-350 (PerkinElmer Life Sciences) 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.
In some experiments, the radioaffinity-labeled plasma membrane proteins were digested with endoproteinase GluC (New England Biolabs) using methods described by the manufacturer. The digests were fractionated for 20 h on 14.5% polyacrylamide gels with Tris-Tricine buffer (43). The gels were sliced and processed for radioactivity measurements, as described above.

Restoration of Transport Function by Co-expression of hRFC TMD1-6 and TMD1-7
Half-molecules-For hRFC, major portions of the TMD6/7 linker (amino acids 215-263 and 204 -263) (Fig. 1) could be replaced with non-homologous segments (73 or 84 amino acids, respectively) of the structurally analogous SLC19A2 protein to preserve transport function (31). Because this suggested that the primary role of the linker region was to ensure optimal spacing between the TMD1-6 and TMD7-12 regions and was largely independent of amino acid sequence, we hypothesized that TMDs 1-6 and TMDs 7-12 could be expressed as separate polypeptides to restore transport function.
at Glu 226 (Fig. 1B), and (ii) hRFC-TMD7-12 Myc-His10 , composed of 307 amino acids from Asn 231 to Leu 537 of hRFC and including an identical 5Ј-untranslated region and 18 N-terminal amino acids to the full-length hRFC (KS43) and hRFC-TMD1-6 HA , and with a Myc-His10 epitope at the C terminus (Fig. 1C). Additional control constructs were designed to encompass TMDs 1-12 in their entirety along with specific epitope tags and included (iii) hRFC HA12 , in which a HA epitope was inserted into the C terminus at Gln 587 (14,37), and (iv) hRFC Myc-His10 , in which a Myc-His10 epitope was inserted after Leu 537 .
Collectively, these results demonstrate that the functional properties of hRFC can be largely restored by co-expression of the hRFC-TMD1-6 HA and hRFC-TMD7-12 Myc-His10 half-molecules.
Labeling  A and B) or following sustained growth in the presence and absence of 3 g/ml tunicamycin (TN) (panels C and D). In panel E are shown results for hRFC-TMD1-6 HA /TMD6 -12 Myc-His10 -transfected cells for which plasma membranes were prepared, solubilized, and treated with endoproteinase GluC. Undigested (ϪGluC) and digested (ϩGluC) proteins were fractionated for 20 h on a 14.5% polyacrylamide gel with Tris-Tricine buffers (43). Analysis of bound radioactivity was as described above. For panels A-E, the positions of molecular mass standards (in kDa) are indicated.
To further verify the identities of the affinity-labeled band in hRFC-TMD1-6 HA /TMD7-12 Myc-His10 -transfected cells, cells were treated with tunicamycin (3 g/ml). Predictably, tunicamycin treatment shifted the broadly migrating (ϳ85 kDa) NHS- To further localize the region covalently modified with NHS-[ 3 H]Mtx within TMDs 7-12, plasma membranes from hRFC-TMD1-6 HA /TMD7-12 Myc-His10 -transfected cells were digested with endoproteinase GluC, prior to electrophoresis on Tris-Tricine gels. GluC is a serine proteinase, which selectively cleaves peptide bonds C-terminal to glutamic acid residues (45). Not all glutamic acid residues are equally sensitive to GlyC digestion. For the TMD7-12 Myc-His10 half-molecule, up to 14 theoretical cuts are predicted, generating a range of fragments from 1 to 100 amino acids and molecular masses of 147-10,750 daltons (Table II). As shown in Fig. 9E, GluC digestions resulted in a reproducible shift of the 40 kDa fragment to ϳ20 kDa. Although no ϳ20 kDa peptide was predicted (Table II), this species could only have arisen from the 295-394 and 395-477 peptides. DISCUSSION By topology mapping and deletional mutagenesis studies (13,14,16,29,32), a picture of RFC has emerged of a symmetrical topologic structure in which 2 groups of 6 TMDs are connected by a large cytosolic loop. Neither of the cytosolicfacing N or C termini is directly involved in substrate binding or translocation, and they appear to only slightly influence membrane targeting (30,46). While deletions of the loop domain between TMDs 6 and 7 are deleterious, a large portion can be replaced by non-homologous sequence to restore transport function (31). The present results establish that both the TMD1-6 and TMD7-12 segments are required for proper folding and membrane insertion. Moreover, they suggest that the role of the TMD6-TMD7 connecting loop in hRFC primarily involves providing proper spacing between TMDs1-6 and TMDs7-12 rather than by directly participating in substrate binding and membrane translocation. Thus, co-transfection of both the TMD1-6 and TMD7-12 hRFC half-molecules into transport-impaired K562 cells was essential to the expression of functionally competent carrier at the plasma membrane surface. Restored transport exhibited characteristics typical of native hRFC including Mtx transport kinetics, inhibition by NHS-Mtx, and capacity for trans-stimulation by preloading with high concentrations of leucovorin.
By analogy with other transporters (47,48), it can be assumed that the 12 individual TMDs in hRFC associate with each other to form an entity resembling an aqueous "channel" for transmembrane passage of anionic folate and antifolate substrates, and that these TMD associations may be enhanced by the formation of critical salt bridges between charged residues such as those between Asp 88 in TMD2 and Arg 133 in TMD4 (17). From mutagenesis studies, a number of other amino acids localized to TMDs 1, 3, 4, 8, 10, and 11 have been implicated as important to carrier activity (18 -27). However, with the exception of residues flanking TMD1 (15,28), the functional importance of these amino acids has not been independently verified. Our hRFC-TMD1-6 HA /TMD7-12 Myc-His10 co-transfection model afforded a unique opportunity to shed further light on regions in the hRFC molecule that are important to transport substrate binding. By radioaffinity labeling of the hRFC-TMD1-6 HA /TMD7-12 Myc-His10 -transfected cells with NHS-[ 3 H]Mtx, a non-glycosylated 40 kDa protein was exclusively and specifically labeled, strongly suggesting covalent incorporation of [ 3 H]Mtx into the hRFC-TMD7-12 Myc-His10 half-molecule and implying a direct role for amino acids localized to TMDs 7-12 in substrate binding. By GluC digestions, it was possible to further refine this region to ϳ20 kDa, likely corresponding to amino acids 295-477, including TMDs 8 -12.
Of course, the region labeled by the NHS-[ 3 H]Mtx ester likely represents only amino acids involved in binding the terminal glutamates of Mtx and related substrates since NHS modification involves activation of the Mtx carboxyl group(s) for targeting electrophilic residues (e.g. N, O, S). Other residues/domains undoubtedly also contribute to binding of the pteridine and p-amino acid portions of folate substrates. Indeed, the finding that mutations of amino acids (e.g. Glu 45 , Ile 48 ) localized within TMDs1-6 confer distinct transport properties for (anti)folate substrates with different pteridine modifications (e.g. folic acid versus Mtx or 5,10-dideazatetrahydrofolate) (22,24) strongly implies a role these for distal residues in substrate binding, presumably by associating with the pteridine ring.
Clearly, an important future goal will be further identification of individual domains and amino acids that directly participate in folate and antifolate binding and translocation by RFC. This will be facilitated by our ability to express hRFC as separate half-molecules, as described herein.