A Mutation Inactivating the Mitochondrial Inner Membrane Folate Transporter Creates a Glycine Requirement for Survival of Chinese Hamster Cells*

,

The compartmentalization of folate cofactor interconversion and folate-dependent biosynthetic reactions between cytosol and mitochondria is essential for the proper function of mammalian folate metabolism (1,2). In mitochondria, reduced folates are required for the initiation of mitochondrial protein synthesis, the synthesis of glycine from serine, and the glycine cleavage system, which allows the generation of 10-formyltetrahydrofolate (3,4).
The transport of folates into the cytosol has been studied extensively (5,6). Both the folate receptor (7) and the reduced folate carrier (8) have been defined as assisting in the passage of folates across the plasma membrane. In contrast, very little is known about the mechanism by which folates reach the mitochondrial matrix from the cytosol. Puck and co-workers (1,9) isolated a CHO 1 mutant cell line that was auxotrophic for glycine and that did not accumulate folate cofactors in their mitochondria (15). A recent study in our laboratory (10) led to the isolation of the human gene encoding a protein that complemented the glyB defect. We tentatively named this protein the human mitochondrial folate transporter (MFT) in response to the fact that the structure of this protein had the characteristics of members of the inner mitochondrial membrane carrier family. The human cDNA isolated in that experiment reinstated folate uptake when transfected into glyB cells (10), strong evidence that the MFT was the endogenous protein responsible for the transport of folates into mammalian mitochondria. However, it remained possible that the complementation observed could have been through a compensatory or nonspecific mechanism, and direct evidence that the CHO MFT was defective in glyB cells was needed to conclude that this protein is the endogenous mitochondrial transporter that facilitates the entry of folates into the mitochondrial matrix.
We now report the sequence of the hamster mft cDNA in wild-type CHO and mutant glyB cell lines and functional studies on these cDNAs. We conclude that we have identified the cause of the glyB auxotrophy, that the MFT is the mitochondrial folate transporter, and that we have found a region of this protein vital to its transport function.

EXPERIMENTAL PROCEDURES
Materials-CHO-K1 cells were obtained from ATCC. glyB cells, a CHO-K1 derivative, were generously provided by Prof. Lawrence Chasin from Cornell University. V79 cells were obtained from Prof. Chia C. Chang from Michigan State University. All of the cell lines were negative for mycoplasma as detected by hybridization to a specific ribosomal RNA probe and were routinely grown in ␣-MEM supplemented with 10% fetal calf serum (FCS).
Northern Blot-RNA from CHO and glyB cells was isolated using the TRIzol reagent (Invitrogen). Poly(A) ϩ RNA was prepared using an Oligotex mRNA kit (Qiagen). Poly(A) ϩ mRNA (2 g) was run in each lane of an agarose gel and blotted onto a nylon membrane (Biotrans; ICN). A 726-nt PCR amplified region of CHO mft cDNA spanning a portion of the open reading frame was 32 P-labeled by random priming and added to membranes in a Gilbert's solution-based hybridization mixture at 1 ϫ 10 6 cpm/ml. An overnight hybridization at 65°C was followed by two washes in 2ϫ SSC, 0.1% SDS at room temperature and at 50°C, respectively.
Sequence Analysis of Hamster mft-A cDNA corresponding to a fragment of the hamster mft was amplified from both CHO and glyB mRNA using PCR primers predicted from human mft sequence. The forward primer was 5Ј-GCTGGACCTGGTCAAGATCC-3Ј, and the reverse primer was 5Ј-CTAGCTTACTTTCTTTTCTCGAAGG-3Ј. The PCR conditions were 94°C for 30 s, 58°C for 1.5 min, and 72°C for 1 min for a total of 35 cycles. The optimum PCR product was obtained when using the following conditions: 1.5 mM MgSO 4 , 200 nM 5Ј and 3Ј primers, 0.2 M dNTPs, 1 g of cDNA, 1ϫ PCR enhancer (Invitrogen), 10ϫ PCR amplification buffer, and Taq polymerase in a final volume of 25 l. The 726-nt amplified fragment was subcloned into the pCRII-Topo vector (Invitrogen), and the insert from six colonies from two PCR reactions for each cell line was sequenced in both directions. The sequences were compared using the Gene Jockey program. The alignment of human, CHO, and glyB mft sequences were carried out in the MegAlign program. The genomic organization of human and mouse mfts was deduced by searching the mouse and human genome databases with the cDNA for human mft using the BLAST program.
CHO cDNA Library Screen-A CHO -Zap CMV-XR library (Stratagene) was plated at 50,000 plaque-forming units/plate on 20 150-mm NZY agar plates after incubation with XL1-Blue MRFЈ cells. A 313-nt PCR-amplified fragment from the 5Ј end of CHO mft cDNA was 32 Plabeled by random priming and added to the membranes of duplicate nitrocellulose lifts in a hybridization mixture at 1 ϫ 10 6 cpm/ml. This PCR product was cloned and sequenced prior to use, and its identity as an upstream fragment of the hamster mft was deduced from homology with the human mft. After overnight incubation at 65°C, membranes were washed sequentially to a final stringency of 0.5ϫ SSC, 0.1% SDS at 50°C. Agar plugs containing positive plaques were excised, and secondary and tertiary screening was performed to isolate a single plaque.
Transfection Studies-glyB cells actively growing in ␣-MEM (Invitrogen) supplemented with 10% FCS were plated at 2 ϫ 10 5 cells on each of the 24 100-mm dishes. After 24 h, 5 g of plasmid DNA combined with 2 M CaCl 2 and 2ϫ HEBS (HEPES, NaCl, KCl, Na 2 PO 4 , and glucose) was overlaid on the cells in a total volume of 1 ml (11). After 30 min at room temperature, 9 ml of complete medium was added to each plate and the plates were incubated at 37°C overnight. The following day, a brief Me 2 SO shock was performed and the medium was changed. Cells were placed under single (G418 (1 mg/ml), basal MEM with glycine and containing dialyzed serum) or double selection (G418 (1 mg/ml), glycine-free MEM containing dialyzed serum) 48 h after transfection. Plates were fed daily, and colonies emerging 10 days after transfection were fixed and stained.
Subcellular Distribution of Cellular Folates-Cell cultures (6 ϫ 10 6 / 175 cm 2 ) were grown in ␣-MEM, which contains glycine, supplemented with dialyzed FCS and 0.3 Ci/ml [ 3 H]folic acid (Moravek) for 48 h. Cells were detached with trypsin and pelleted by centrifugation. The pellets were resuspended in 10 ml of PBS (ϩ5% FCS), counted, and re-pelleted. The pellets were resuspended in 2 ml of homogenization solution (0.25 M sucrose and 1 mM EDTA), and the various cellular fractions were isolated using a Dounce homogenization procedure described previously (10). Aliquots of fractions were prepared for scintillation counting under uniform ionic conditions, and radioactivity was converted to folate content by counting an aliquot of the culture medium under identical conditions. Fluorescence in Situ Hybridization (FISH) Analysis-For FISH, metaphase spreads from V79, CHO, and glyB cells were prepared as described previously (12). Two fragments of CHO genomic DNA were PCR-amplified and directly labeled with the Spectrum Orange fluorophore using a nick translation kit (Vysis). These fragments corresponded to the regions of the hamster mft gene spanning predicted exons 2-4 including the two intervening introns (5.7 kb) and the region spanning predicted exons 6 and 7 including the intervening intron (1.3 kb). These two PCR products were cloned and partially sequenced to confirm the identity with the mft gene. Repetitive (Cot-1) DNA was prepared from CHO cells as described previously (13). A mixture consisting of 500 ng of each labeled probe plus 10 g of Cot-1 DNA was precipitated, resuspended in hybridization buffer (Vysis), denatured at 75°C for 5 min, suppression-hybridized for 1 h at 37°C, and then hybridized for 72 h to the metaphase chromosomes. The excess unbound probe was removed by sequential washes, first in 0.4ϫ SSC, 0.3% Nonidet P-40 at 70°C for 2 min followed by 2ϫ SSC, 0.1% Nonidet P-40 at room temperature for 2 min. After air-drying, the chromatin was counterstained with 4Ј,6-diamidino-2-phenylindole/Antifade. For each cell line, at least 15 metaphase spreads were analyzed for the number and chromosomal location of fluorescent signals using a Zeiss Axioskop microscope. Representative images were documented using a Cytovision image analysis system (Applied Imaging).

The Mitochondrial Folate Transporter Is Expressed in glyB
Cells-Earlier studies by Kao et al. (1) and Chasin et al. (14) establish four complementation groups among CHO mutants that were auxotrophic for glycine. A cell line defining one of these complementation groups, designated glyB, was found to have a defect in its ability to accumulate folates in the mitochondria (10, 15), a process required for the activity of mammalian mitochondrial serine hydroxymethyltransferase (16). To understand what was causing this mutant phenotype, we examined the expression of the hamster homolog of a human gene (mft) previously identified (10) to stimulate folate influx into glyB mitochondria. Initially, the human mft cDNA sequence was used to design primers for PCR amplification of the corresponding hamster cDNA. The primers were designed in regions that retained the highest homology among other inner mitochondrial membrane proteins such as the energy sequence motifs and the putative membrane-spanning domains (17)(18)(19)(20). With one set of primers, a single 726-nt fragment was amplified using cDNA from both CHO and glyB cells. The identity of this cDNA fragment as the hamster mft was clear from the homology with the human gene (see below). Thus, it was used subsequently to probe poly(A) ϩ RNA from CHO and glyB cells on a Northern blot (Fig. 1). A diffusely migrating mRNA band was seen at 2.3 kb in both cell lines. With longer exposures, a faint band at 1.6 kb could be seen in both cell lines. This same pattern was observed previously in Northern blots from human cells using the human mft as a probe (10). Hence, it appeared that the glyB phenotype was not caused by a lack of expression of the hamster MFT protein.
A Point Mutation in the mft Gene in glyB Cells-To obtain the full hamster mft sequence, a CHO cDNA library was screened with a 313-nt PCR product amplified from the 5Ј region of the hamster cDNA. Three of the longest clones were sequenced. An open reading frame of 954 nt was identified in these cDNAs that encoded a protein of 317 residues with a calculated mass of 35,125 daltons. The predicted hamster protein was 89% identical to the reported human MFT (Fig. 2). The longest 5Ј-untranslated region found was 203 nt, and the 3Јuntranslated region appeared to be 1189 nt. The sequence can be retrieved from GenBank TM under accession number AY611603. As with the human MFT, hydropathy consider- Poly(A) ϩ mRNA (2 g/lane) from the two cell lines were run on a 1% gel, blotted onto a nylon membrane, and probed with a 726-nt portion of the CHO mft cDNA. A major band at 2.3 kb was detected in both cell lines, and a faint band at 1.6 kb also could be detected with longer exposures. The blot was stripped and re-probed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA for normalization.
ations identified six putative transmembrane domains in the hamster MFT and three internal mitochondrial targeting sequence characteristic of mitochondrial inner membrane proteins were identified, which were placed precisely after the 1st, 3rd, and 5th predicted transmembrane domains (Fig. 2). As with the human gene, the hamster mft belonged to the small subgroup of mitochondrial inner membrane transport proteins that contained the P(I/L)W tripeptide in the second energy transfer motif (see "Discussion").
Using the full-length CHO cDNA sequence as a template, primers placed in the 5Ј-and the 3Ј-untranslated region were designed for PCR amplification of the entire coding region from hamster cells. The PCR product from CHO and glyB cells was gel-purified, cloned, and sequenced. Only a single nucleotide difference (G to A at nt 575) was found between the CHO and glyB cDNAs (Fig 3A). This single mutation was observed in each of six clones isolated in two independent PCR experiments starting with glyB cell cDNA but was not found in six equivalent cloned PCR products from CHO cDNA. This mutation resulted in the substitution of glutamic acid for a glycine at codon 192 in the peptide predicted to form transmembrane domain 4 of the protein.
To determine whether this mutation was occurring in one or both alleles of the glyB mft gene, the region of genomic DNA containing codon 192 was amplified from both CHO and glyB cells. With the completion of the human genome sequence, the exonic structure of the human mft gene could be deduced and the distribution into exons and introns of a mouse dbEST entry (GenBank TM accession number BI251765), which clearly represented the mouse mft homolog, could also be deduced from the recently completed mouse genome sequence (Fig. 3B). Assuming a similar exonic distribution of expressed sequence in the hamster as seen in Fig. 3B for mouse and human genes, it was predicted that codon 192 would be in exon 5 of the hamster genomic locus. Hence, the primers from predicted exons 5 and 6 were used to PCR-amplify the region of the glyB mutation on a larger genomic DNA marked by the presence of intron 5. Sequence obtained from this 626-nt genomic fragment revealed that only an A residue was found in glyB genomic DNA at the position corresponding to the glyB mft mutation and that only a G residue was found in CHO genomic DNA (Fig. 3C). This observation clearly ruled out the presence of one mutated allele and a second wild-type but epigenetically silenced allele in glyB cells. However, by such sequencing alone, we could not differentiate the presence of only one copy of the mft in the CHO genome or a mutation in one allele followed by a gene conversion event, which fixed the same mutation in each of two copies of the gene, or the loss of an allele during the selection of the glyB cells.
CHO Cells Have Two Alleles of the mft Gene, but glyB Cells Have Only One-A FISH analysis was used to determine the number of signals present for the mft gene in the metaphase spreads of CHO, V79, and glyB cell lines. The modal numbers of 21, 22, and 19 chromosomes, respectively, were noted in the metaphase spreads from these cell lines. In the CHO metaphase spreads, two small chromosomes showed bright hybridization signals with the probes for the hamster mft gene (Fig.  4A). Based on the information gained from the FISH and additional GTG-banding studies (data not shown), these signals appeared to localize toward the ends of a Z-12 chromosome and a chromosome 11 (21). The same number of FISH signals was seen on metaphase chromosomes from the hamster V79 lung fibroblast cell line (data not shown), leading to the conclusion that CHO and V79 cells lines had two copies of the mft gene. In contrast, the glyB metaphase spreads had only one chromosome that showed a bright hybridization signal after hybridization with the mft probes (Fig. 4B). The labeled medium-sized chromosome had an atypical morphology and banding pattern, suggesting that these cells had acquired additional chromosomal rearrangement(s) during selection from a CHO background. This conclusion also is supported by the observation of a reduction in chromosome number in the glyB cell line (n ϭ 19) compared with the CHO line (n ϭ 21). We concluded that there are two identical copies of the mft gene in CHO cells but only one mutated copy in glyB cells.
The G192E Mutation Results in Glycine Auxotrophy-To establish whether the mutation identified in the glyB mft gene was causative of the mutant phenotype of this cell line, we stably transfected the mft cDNA from CHO and glyB cells back into glyB recipient cells and determined whether these cDNAs complemented the auxotrophy of these cells. The region of cDNA encoding the open reading frame from both CHO and glyB were re-cloned into the mammalian expression vector pcDNA3.1(Ϫ) and transfected into glyB cells using a calcium phosphate-mediated procedure. After 3 days, replicate plates were placed under single selection with G418 as a transfection frequency control or under double selection using both G418containing and glycine-deficient media. After 10 days of selection, the plates were stained for visualization of colonies. Transfection of the wild-type CHO cDNA into glyB cells resulted in a similar rate of colony formation under single or double selection (Table I), indicating that the CHO mft cDNA could complement the auxotrophy of the glyB cells efficiently. The glyB cells transfected with the mutant construct and placed under double selection were unable to form colonies, but these transfections survived under single selection (Fig. 5A). In every experiment, a few small colonies (Ͻ20 cells/colony) were found on plates transfected with glyB cDNA, but these adherent cell clusters were incapable of continuous growth when transferred in selective medium in 24-well dishes. As seen in previous experiments (10), the human MFT was able to complement the glycine auxotrophy but to a somewhat lesser extent than we observed with the CHO construct. glyB cells transfected with the vector alone had formed colonies only on single selection dishes. Hence, these transfection experiments indicated that the single base pair alteration located at residue 192 in the glyB mft gene is capable of producing the mutant phenotype of glyB cells.
Wild-type MFT Reinstates Folate Levels in Mitochondria of glyB Cells-Colonies of glyB cells transfected with either CHO or glyB mft were selected for clonal expansion. glyB cells expressing the wild-type CHO construct could be maintained in the absence of glycine, but glyB cells transfected with glyB mft cDNA could only survive in medium supplemented with exogenous glycine. To establish a cell line expressing the glyB mft under the control of the cytomegalovirus promoter, glyB cells were transfected with the glyB mft cDNA and exposed to double selection for 10 days. The few weakly adherent small cell FIG. 4. Fluorescent in situ hybridization of a mft probe to metaphase chromosomes from hamster cell lines. A 7.0-kb region from the CHO mft genomic locus was labeled fluorescently and hybridized to metaphase spreads from CHO cell lines. A, in CHO cells, hybridization was detected near the telomeres of two small chromosomes (yellow arrows). B, in glyB cells, one signal could be detected (yellow arrow), indicating that the mft gene is at a haploid locus in this mutant cell line. A faint signal was observed on one copy of chromosome 2 in CHO, which was also seen in V79 and glyB cells, and was thought to represent a region having some underlying sequence similarity to mft.

TABLE I Stable transfection of mft homologs into glyB cells
The open reading frame of mft homologs were subcloned into pcDNA3.1 and stably transfected into glyB cells. In each experiment, triplicate plates were selected in medium containing 1 mg/ml G418 and three more plates were selected in glycine-deficient medium containing G418. Colonies were fixed, stained, and counted after 10  clusters were isolated individually and replated into MEM with G418 and containing glycine. One of these clusters was rescued by the glycine and was expanded into a cell line. The content of folates in subcellular compartments of glyB cells expressing these cDNAs was determined. glyB cells transfected with the wild-type CHO mft were able to restore folates in mitochondria to levels comparable with that seen in wild-type CHO cells (Table II). Cells transfected with the mutant glyB mft had mitochondrial folates equivalent to those seen in glyB cell controls (Table II). We concluded that the MFT is the folate transporter and that the mutation seen in glyB prevented the accumulation of folates in the mitochondria.
Identification of the Mouse and Zebrafish mfts-Several mft candidate homologs from other species were identified easily from the NCBI data base that had high levels of identity to the hamster MFT amino acid sequence (Fig. 6). The putative mouse MFT protein was 96% identical to the hamster MFT, whereas the zebrafish (GenBank TM accession number BC048057) had a 66% identity to the hamster protein. Clones for the mouse and zebrafish candidate mft sequences were purchased from the I.M.A.G.E. consortium of the American Type Culture Collection, and the open reading frames of both sequences were subcloned into pcDNA3.1 and transfected subsequently into glyB cells. The mouse clone allowed the growth of glyB cells equally well under single selection with G418 or with double selection with both G418 and in medium lacking glycine (Fig. 5B and Table I). The zebrafish clone also was able to complement the glycine auxotrophy but to a somewhat lesser extent (44%) than seen with transfections of any of the mammalian cDNAs studied. We concluded that the mouse and zebrafish cDNAs identified in Fig. 6 encoded the mft homologs from these species.
Testing the glyB mft cDNA for a Dominant Negative Effect-Because previous literature (22) indicates that other inner mitochondrial transporters function as homodimers, we had expected that the co-expression of a functional MFT and the glyB MFT in the same cell line would result in an impaired activity of the functional MFT, i.e. that the glyB mft cDNA would have a dominant negative effect. The transfection of either CHO or mouse mft into a glyB cell background (Table I) did not support this expectation. To more formally test this hypothesis, the glyB mft cDNA under the strong cytomegalovirus promoter was transfected into CHO cells and the rate of colony formation was compared in the presence and absence of glycine. Surprisingly, the data from this experiment (Table III) demonstrated that transfection of the glyB mft cDNA was no different from the transfection of the pcDNA control alone, indicating no detectable dominant negative effect and leading us to the conclusion that the glyB MFT was not forming a dimer with co-existing functional MFT species. DISCUSSION We had isolated a human gene previously that could complement the defect of a cell line rendered auxotrophic for glycine and incapable of accumulating folates into their mitochondria. Herein, we report that the homologous hamster protein is mutated in glyB cells and that the single point mutation distinguishing the hamster MFT proteins in CHO and glyB cells explains the mitochondrial folate transport defect in glyB cells. We conclude that the MFT protein can be identified definitively as the endogenous mitochondrial folate transporter.
The point mutation identified in the glyB mft cDNA resulted in the substitution of a glycine residue for a glutamic acid within the fourth predicted transmembrane domain, an almost entirely hydrophobic region of this protein. The placement of a negatively charged amino acid within a hydrophobic membrane-spanning domain probably either disrupts the insertion of the protein into the hydrophobic environment of the inner mitochondrial membrane or interferes with proper folding and results in the targeting of the mutant MFT for proteolytic degradation. Either of these effects would explain the loss of MFT function in the glyB cells. It is thought that the inner mitochondrial membrane transporters function only after assembly as a 12-transmembrane domain homodimer (22). Hence, we had expected that the inactivated glyB MFT transporter might have a dominant negative effect on the function of transfected mft species. However, we have not detected such an effect (Tables I and III), because the glyB cell acts like a null function cellular background. As such, the glyB promises to be a valuable tool for study of the mechanism of the process of transport of folates into the mitochondrial matrix since this function can be selected for in this mammalian cell line.
The high level of sequence identity between human and hamster proteins (Fig. 2) prompted us to search the EST and completed genome databases for probable mft sequences from other organisms. We compare the sequences of an unidentified cDNA of mouse origin with cDNAs from zebrafish and monkey (GenBank TM accession number AB060253), all of which appear to represent mft genes from these species based on homology (Fig. 6) and function (Fig. 5). Also compared are the yeast gene for a mitochondrial inner membrane transporter for flavin (GenBank TM accession number NP012132) and the human ATP/ADP transporter (GenBank TM accession number Q09073). We compared the sequence similarities of the folate-specific carriers and also compared them with related family members in an attempt to deduce structures needed specifically for folate transport. The structural divergence/similarities in several regions of these proteins are noteworthy and suggestive of the roles of these peptides in MFT function as described below.
Energy Transfer Signature Motif-In the hamster MFT (Fig.  2), as with the other mammalian MFTs (Fig. 6), there are three peptides located immediately after the 1st, 3rd, and 5th transmembrane domains that fit the energy transfer signature (ES) consensus ( (27,28). In yeast, these domains bind to proteins TIM9 and TIM10 in the intramembrane space during the course of translocation of mitochondrial transporters to the site of insertion in the inner mitochondrial membrane (27,29,30). When comparing the sequences of the 1st, 2nd, and 3rd energy signal motifs on any one of the MFT proteins, there are clear and notable differences in sequence among the three ES motifs. However, each of the energy signal motifs, taken one at a time, is nearly identical across the several mammalian MFT homologs currently identified. The variation in sequence among the three ES peptides of a single protein, yet near identity at each of the three ES signatures across proteins, is highly suggestive that each ES peptide has a somewhat different but conserved function. It is also very interesting that the ES sequences of MFT are remarkably divergent from those of other inner membrane proteins (e.g. compare the human MFT and human ATP/ADP transporter (Fig. 6) at any of the three ES repeats) (Fig. 6) or even comparing the mammalian sequences with zebrafish within a single ES motif. This would not be expected if the only function of the ES peptides were to bind to a common set of proteins responsible for the trafficking of all of the mitochondrial transporters to their site of insertion in the inner membrane. This seems to imply either that there are substantial differences in binding partners for trafficking of individual inner membrane transporters or that these peptides have other functions involved in the transport of substrates. Others (26) have noted that there is a subgroup of the 75 mitochondrial transport proteins predicted from the yeast genome sequence distinguishable by the fact that they bear a PIW or PLW sequence in the second ES signature. All of the mammalian sequences we identify as mft homologs have the PIW tripeptide with the exception of the zebrafish, which instead has a PVW at this position, placing the MFT transporters in this small subgroup, which includes the yeast flavin transporter flx1.
Inner Membrane Loop Sequences-By homology with other more thoroughly characterized mitochondrial transporters (26,31,32), it is predicted that the MFT proteins are inserted with the N-and C-terminal peptides facing the intermembrane space (Fig. 7). With this type of distribution in the inner membrane, three loops located directly after transmembrane domains 1, 3, and 5 are predicted to extrude into the matrix side of the membrane. The peptides that comprise these loop regions retain profound homology among the predicted mammalian MFT homologs. This does not hold true when one compares across mitochondrial metabolite carriers for other substrates compared with the MFT homologs (compare the features of the mammalian or zebrafish MFT inner loop sequences with those of the human ATP/ADP transporter or yeast flx1 carriers) (Fig.  6). It is also noted that several charged residues are universally present on all of the mammalian folate carriers and are placed identically in these loops. The presence of multiple charged residues in these predicted solvent-exposed loop peptides is to be expected from previous literature (26,33). However, some of these charged residues seem to be conserved only in the predicted MFT homologs, suggesting a role specific to the transport of folates.
Outer Membrane Loop Sequences-Two loop sequences are predicted in the membrane-inserted MFT that project into the intramembrane space after the 2nd and 4th transmembrane domains in addition to the N-and C-terminal peptides. The 1st FIG. 7. Predicted structure of the hamster mitochondrial folate transporter. The projected orientation of the MFT protein in the inner mitochondrial membrane is shown as predicted by hydropathy plot analysis and homology to other mitochondrial inner membrane proteins. The model predicts that both the N and C termini face the intermembrane space, that there are six membrane-spanning regions, and that there three energy motifs are located immediately after membrane-spanning regions 1, 3, and 5. The G to A transition in glyB cells at residue 192 (yellow) would be predicted to be in the fourth membrane-spanning region.

TABLE III Transfection of glyB mft cDNA into CHO cells
The cDNA for the mutant mft found in glyB cells was cloned into pcDNA3.1 and transfected into CHO cells. After 10 days of selection on medium containing glycine and G418 or glycine-deficient medium containing G418, the colonies found on triplicate plates were fixed and stained. Colony sizes were equivalent in all of the groups. and 2nd intramembrane loops of the identified MFTs are highly homologous across species as is the C-terminal peptide, in distinct contrast to the sequences of the N-terminal peptides that differ significantly among even the mammalian MFT proteins (Fig. 6). When compared with peptides of other metabolite carriers in the inner mitochondrial membrane, there are several positions that seem unique to the folate carriers, presumably reflecting function. Similar to what is observed with the internal surface-facing loops, there are several invariant charged residues spaced at intervals of approximately three residues apart. Currently, the role of these residues in recognition and transport of folates is being studied.