Ligand-directed immunoaffinity purification and properties of the one-carbon, reduced folate transporter. Interspecies immuno-cross-reactivity and expression of the native transporter in murine and human tumor cells and their transport-altered variants.

Almost complete purification (> 95%) of the 46-kDa murine, one-carbon, reduced folate transporter (RFT) at a recovery of 20% was obtained by ligand-directed immunoaffinity fractionation from transporter overproducing L1210/R83 cells. These cells were labeled with the N-hydroxysuccinimide ester of [3H]aminopterin (AMT), the isolated plasma membrane alkaline washed to remove nonintegral membrane proteins, detergent-solubilized, and RFT-separated on an anti-AMT antibody-protein G-Sepharose column followed by preparative SDS-polyacrylamide gel electrophoresis. Anti-RFT antibody, subsequently derived, differentially blotted (L1210/R83 >> L1210/0) a 46-kDa protein during SDS-polyacrylamide gel electrophoresis of plasma membrane from L1210/R83 and L1210 cells and in L1210/R83 cells after trichloroacetic acid precipitation. In contrast to that reported for human tumor cells, glycosidase treatment of RFT revealed no common N- or O-linked core oligosaccharides associated with this protein. The same 46-kDa protein at different relative levels was revealed in a Western blot of plasma membrane from other murine tumors. Blotting of plasma membrane from methotrexate resistant, transport defective L1210 cell variants exhibited wild-type levels of a less electrophoretically mobile RFT or greater levels of the same 46-kDa RFT which could not be affinity labeled with N-hydroxysuccinimide-[3H]AMT. The same antibody differentially blotted a 83-kDa plasma membrane protein from human HL-60 and CCRF-CEM cells with different levels of reduced folate transport and affinity labeling of RFT, verifying the conserved nature of this protein consistent with earlier functional studies.

The preferred route for mediated entry of reduced folates and folate analogues through the plasma membrane of tumor cells (reviewed in Refs. 1 and 2) is the one-carbon, reduced folate transport (RFT) 1 system. This transport system is a major determinant of cytotoxicity (2,3) and of therapeutic responsiveness of tumors (3), and genetic alteration of its biochemical properties has been shown (4) to be a common form of acquired resistance to classical folate analogues in many animal and human tumors. More recent studies utilizing affinity-or photoaffinity labeling have characterized the transporter (RFT) for the one-carbon, reduced folate transport system as an integral membrane protein of molecular mass ϭ 43-46 kDa in murine L1210 cells (5)(6)(7), but exhibiting a much higher mass in human CCRF-CEM, K562, and HL-60 cells (8 -10). The larger mass of the human transporter appears to relate to its content of core oligosaccharides (8,9,11). Further studies on the biochemical properties of this transporter and the regulation of its gene expression have been hampered by the relatively low level of its expression (1) in all of the tumor cells studied and the unavailability of facile methods for its purification. Progress in regard to the former limitation was first documented (12) in our studies describing the isolation by one of two methods of variants of the L1210 cell expressing high levels of the transporter. The isolation of similar variants of human CCRF-CEM and K562 leukemia cells by one of these methods was subsequently reported (8,13) by others. More recently (10), we applied the same methodology to the isolation of a similar variant of HL-60 cells.
In a preliminary report (14), a method for purification of the RFT from L1210 cells was recently described. These workers utilized affinity purification by streptavidin "capture" of the transporter labeled with a biotinylated folate analogue to obtain microgram amounts of the transporter. More recently, a second approach was described (15) using lectin-affinity chromatography which appeared to be successful when applied to a highly glycosylated reduced folate transporter (8,11) from an "overproducing" variant of K562 cells. These are extremely interesting approaches with apparent potential for widespread use. However, the application of these approaches in the authors' laboratory for the purification of transporter from L1210 cells "overproducing" the transporter was consistently unsuccessful. As an alternative approach, we now report on the purification of this transporter from an "overproducing" L1210 cell variant by ligand-directed immunoaffinity chromatography. This methodological approach is now described in detail along with data obtained with anti-RFT polyclonal antibodies on the native properties of the transporter and its relative expression and properties in parental and transport altered variants of murine and human tumor cells.

EXPERIMENTAL PROCEDURES
Source of Cells-The isolation and characterization of L1210/R83 variant cells that have increased amounts (30 -40-fold) of the RFT has been described previously (6,12). These cells are maintained in RPMI medium with dialyzed fetal calf serum supplemented with 4 ϫ 10 Ϫ9 M * Supported in part by National Cancer Institute Grant CA56517 and Center Core Grant CA08748 and by the Elsa U. Pardee Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  L-L5CHO-folateH 4 and 6 ϫ 10 Ϫ7 M metoprine. Derivation of human HL-60 or CEM-CCRF cells with elevated levels of one-carbon, reduced folate transport have also been described earlier (9,10). Maintenance of these lines was in RPMI medium with an amount of L-L5CHO-folateH 4 as the sole folate source which will allow maximum growth. Large scale accumulation of L1210/R83 cells in BD2F1 mice has also been described (6). Other murine and human tumor cell lines used in these studies were maintained in cell culture (6,12) in RPMI medium. Transport defective methotrexate resistant variants of L1210 and CCRF-CEM cells were isolated as described (4) and maintained (6,9) in the presence of this antifolate in RPMI medium.
Transport Methodology, Affinity Labeling, and Analytical SDS-PAGE-Variant and parental L1210 cells were incubated in transport medium (6) with varying amounts of [ 3 H]MTX in accordance with protocols and processing methodology described previously (6). L1210/ R83 cells in Hepes Mg 2ϩ sucrose buffer (5 ϫ 10 7 cells/ml) were affinity labeled with 1 M of the N-hydroxysuccinimide ester of [ 3 H]AMT for 10 min at room temperature in a manner similar to that described by Henderson and Zevely (16). The cells were washed once in Hepesbuffered saline (20 mM Hepes, 140 mM NaCl, 10 mM KCl, 2 mM MgCl 2 , pH 7.4) and either stored as a pellet at Ϫ70°C or used immediately for plasma membrane isolation (6). Alkaline washed (6) plasma membrane was dissolved in sample buffer and electrophoresed (17) in a 10% polyacrylamide gel.
Preparation of Anti-AMT Monoclonal Antibodies and Western Blotting-CB6F 1 (BalbC ϫ C57/b16) mice were injected ip with 200 -400 g of AMT linked to keyhole limpet hemocyanin in incomplete Freund's adjuvant and boosted by injection 4 and 8 weeks later by the same antigen in incomplete Freund's adjuvant. Fusion of splenocytes, screening of hybridomas, and cloning was as described in Galfre and Milstein (18) and Hartlow and Lane (19). The antibodies derived were classified as IgG 1A by clonotyping (Southern Biotechnology Associates). Purification of this IgG was by affinity chromatography on a protein G-Sepharose column. Detection of [ 3 H]AMT RFT during Western blotting (18,19) following SDS-PAGE (17) with anti-AMT antibodies was by enhanced chemiluminescence (ECL, Amersham Corp.) using goat horseradish peroxidase-conjugated anti-mouse IgG as the second antibody. Immunoprecipitation of antibody [ 3 H]AMT complexes was with protein G-Sepharose (18).
Preparation of Anti-RFT Antibodies in Rabbits-Approximately 20 -30 g of purified RFT in incomplete Freund's adjuvant were administered interdermally at multiple sites in each of four injections 4 weeks apart to rabbits. Antiserum obtained 2 weeks after the fourth injection was absorbed (18) with (50 -100 g/ml) NHS-[ 3 H]AMT affinity-labeled plasma membrane proteins other than RFT (see below) from L1210/R25 cells that were solubilized, absorbed to the anti-AMT monoclonal antibody affinity column, and eluted (see below). Detection of RFT during SDS-PAGE and Western blotting (20) with anti-RFT antibodies was by ECL (Amersham Corp.) using goat anti-rabbit IgG as the second antibody.

Isolation of Plasma Membranes and Purification of [ 3 H]AMT-Labeled Transporter (RFT)-
The procedure for preparation of plasma membrane from tumor cells was described in Yang et al. (6). The yield was 45-50 mg of plasma membrane protein/10 10 cells. Storage of membranes was at Ϫ70°C in 50 mM potassium phosphate with 150 mM sucrose, pH 7.4. Alkaline washed (6) plasma membrane (8 -10 mg/ml) from [ 3 H]AMT-labeled L1210/R83 cells in the presence of proteinase inhibitors (6) was solubilized at room temperature in 50 mM Tris-Hcl plus 1.5% SDS. The suspension was centrifuged at 16,000 ϫ g for 5 min at room temperature to remove insoluble material. An equal volume of 0.3 M NaCl plus 10 mM EDTA (pH 7.55) was added to the solubilized plasma membrane. Triton X-100 was then added to obtain a final concentration of 15% in order to remove SDS from the protein. The protein solution was vortexed and diluted 4-fold in 25 mM Tris-HCl, 0.15 M NaCl, 5 mM EDTA, and 0.5% Triton X-100 (pH 7.5). After preclearing (18) by successively passing the solution through a column of protein G-Sepharose, the solubilized protein solution was used to resuspend anti-AMT monoclonal antibody linked to protein G-Sepharose, and the suspension was shaken end-to-end at 4°C for 16 -18 h. A column was prepared from the suspension of antigen-antibody complex linked to protein G-Sepharose which was then extensively washed in successive steps with 25 mM Tris-HCl, 5 mM EDTA and 0.5% Triton X-100 (pH 7.5) containing one of the following, 0.15 M NaCl, 0.5 M NaCl, 1.0 M NaCl, or 1 M LiCl; with distilled water and with 10 mM sodium acetate (pH 4.2). Labeled protein was eluted from the antibody-protein G column with 0.1 M glycine HCl with 0.25% CHAPS (pH 2.2) and collected in 2-ml fractions in tubes containing 1.5 M Tris (pH 11.5). The final concentration of Tris-HCl was 0.15 M (pH 8.2). After the addition of SDS to a final concentration of 2%, the pooled material was concentrated at room temperature using a Stircell (Amicon, Inc.). The concentrated material was loaded onto a Bio-Rad 491 preparative cell (10.5% gel) for electrophoresis using the conditions specified by Laemmli (17). After 11 h of electrophoresis at a constant power of 11 watts, the labeled protein was eluted from the gel during further electrophoresis at a flow rate of 0.8 ml/min and 4 ml/fraction. The pooled fractions of peak radioactivity were concentrated with the Stircell and reserved for protein determination (21), analysis, and antigen preparation (see above).

RESULTS
Preliminary Studies with Anti-AMT Antibodies-A number of monoclonal antibodies against AMT were derived in mice immunized with AMT-keyhole limpet hemocyanin that were used for the splenocyte/myeloma cell fusions. Much more immunoprecipitation with these antibodies occurred with [ 3 H]AMT (5 pmol/1 g IgG) then with [ 3 H]AMT-labeled plasma membrane protein (1.2 pmol/1 g IgG). However, immunoprecipitation of the latter was increased (2 pmol/1 g IgG) by heating of the membrane at 50°C for 30 min. In order to maximize the interaction between antibody and [ 3 H]AMT linked to plasma membrane protein, we employed a mixture of three monoclonal antibodies (AMT-1, -2, and -3) for the preparation of antibody-protein G-Sepharose.

Purification of [ 3 H]AMT-RFT from L1210/R83
Cells-There are two essential conditions for successful application of liganddirected immunoprecipitation to AMT-labeled transporter for the one-carbon, reduced folate system. In the first, one must effectively label a large percent of the total transporter present in the plasma membrane. To achieve this, with the cell densities that were required in the labeling buffer and the relatively high surface membrane content of RFT, we found it necessary to expose L1210/R83 cells to relatively high concentrations of NHS-[ 3 H]AMT. This was consistently obtained with one exposure to 1 M of NHS-[ 3 H]AMT for 10 min at room temperature. However, labeling of cells at this concentration resulted in a significant amount of nonspecific labeling of other membrane proteins. To remove most of this nonspecific labeling, which occurred on nonintegral membrane proteins, we carried out two successive alkaline washes of labeled plasma membrane in 0.1 M NaCO 3 and then in 0.1 M EDTA. The radioactive profile obtained following SDS-PAGE of this material is shown in Fig.  1. The radioactive profile of similarly derived and alkaline washed plasma membrane protein from a markedly transport defective, MTX-resistant variant (L1210/R25) is also shown in the figure. From these results it can be estimated that nonspecific labeling represents approximately 10 -15% of total radioactivity. More importantly, we also estimated that nonspecific labeling in the area of the 46-kDa peak (L1210/R25 plasma membrane) represents no more than 3-5% of the total.
Data pertaining to the purification of [ 3 H]AMT-RFT from alkaline washed L1210/R83 plasma membrane are given in Figs. 2 and 3 and Table I. Binding of solubilized NHS-[ 3 H]AMT-labeled proteins to antibody-protein G-Sepharose after preclearing occurred to the extent of 70 -80% and approximately 85% of the bound material could be readily removed from the antibody by elution with glycine HCl buffer. Approximately 75% of the eluted material was recovered in the peak fractions for a total recovery of tritium labeled protein in the range of 50% (Table I). Analytical SDS-PAGE analysis of this material is shown in Fig. 2. This migration profile predicted that the majority of the eluted material would be recovered in the major peak fractions at 46 kDa following preparative electrophoresis. However, the actual recovery of [ 3 H]AMT-RFT in the peak fractions from the Bio-Rad 491 cell usually selected (see Table I and Fig. 3) was considerably less (40%) due to aggregation and proteolysis.
In Table I, we compare the total protein recovered at each purification stage. We also compare the disintegratons/min of radioactivity per mg of protein at each stage. From these values, we estimate that approximately 500-fold purification was required to obtain highly purified RFT from the alkaline washed plasma membrane of these cells and that final recovery was in the range of 20%. Although the data suggest that complete purity was achieved, in actual practice, the purity is probably only in the range of 96 -98%. Following SDS-PAGE (slab gel) of this purified material, duplicate transblots were stained with Coomassie Blue or immunoblotted with anti-AMT monoclonal antibody using ECL detection. In each case a single protein band (approximately 46 kDa) was delineated (Fig. 4) on the blot.
Immunoblotting and Size of Native RFT in Parental and Transport-elevated L1210/R83 Cells-Purified RFT from L1210/R83 cells was utilized to prepare anti-RFT polyclonal antibody in rabbits. The serum after absorption with nonspecifically labeled proteins (see above) was used to immunoblot plasma membrane from parental and L1210/R83 cells following SDS-PAGE. The ECL results in Fig. 5A revealed an immunoreactive band at 46 kDa in both cases with an extremely large differential in intensity favoring L1210/R83 plasma membrane. A similar blot with preimmune serum from the same rabbit did not blot (data not shown) a 46-kDa protein. Minor protein bands in a higher molecular mass range on the gels observed during blotting of material derived from L1210/R83 cells but not L1210 cells with either anti-AMT or anti-RFT immune serum are also seen with purified RFT which was size selected during purification. This reflects aggregation of RFT seen also after affinity labeling (6) of RFT overproducing L1210/R83 cells. These results verify that the absorbed antiserum is highly specific for RFT but does not entirely rule out the possibility that minor amounts of antibodies derived against other membrane proteins that were size selected along with RFT were also present in this serum. Size determination of RFT in L1210/R83 derived plasma membrane, whether by affinity labeling or immunoblotting, was always carried out with purified plasma membrane. Despite the fact that protease inhibitors (6) were routinely used during preparation of these membranes, the possibility exists that the RFT detected was a proteolytic fragment. Therefore, to more rigorously verify that the native size of RFT was, in fact, in the vicinity of 46 kDa, we immunoblotted this protein with anti-RFT antibody following SDS-PAGE of crude membrane preparations obtained by rapid sonication and centrifugation at 0°C or as a trichloroacetic acid precipitate of total cellular protein. Data in Fig. 5, A-C show that the size of L1210/R83-derived RFT blotted is the same (46 kDa) when obtained with plasma membrane (Fig. 5A), trichloroacetic acid precipitate (Fig. 5B), and crude membrane (Fig.  5C), or when purified (Fig. 5C). Blotting with antibody of trichloroacetic acid-precipitated L1210/0 cells did not detect RFT (Fig. 5B) ostensibly because of its low content in the precipitate of whole cell material.
Analysis for the Extent of Glycosylation of Purified RFT-As in prior (5,23,24) studies with human tumor RFT, the availability of anti-RFT antibodies greatly facilitates analytical studies of this protein that requires SDS-PAGE. Accordingly, we (23) analyzed the murine RFT for both N-and O-linked glycosides. In the former case, aliquots of purified transporter from L1210/R83 cells were treated with N-glycanase (24) in the presence of proteinase inhibitors (6). In the latter case, aliquots of the same protein were treated with O-glycanase following treatment with neuraminidase (11,24) in the presence of the same proteinase inhibitors. In both instances, no change in the relative rate of migration of RFT as shown by immunoblotting and ECL detection was discerned (Fig. 6) following SDS-PAGE. As a control for these analyses, we treated under the same conditions (30, 31) three glycosylated proteins, ferritin, ␣-acid p-glycoprotein, and transferrin with N-linked glycanase and determined their migration after staining with Coomassie Blue. In every case, a difference in the rate of migration of these proteins after treatment with glycanase was seen (Fig. 4) during SDS-PAGE.
Immunoblotting of RFT in Parental and Transport-defective Murine Tumor Cells-Using the absorbed anti-RFT serum, immunoblotting of alkaline washed plasma membrane from a series of parental and methotrexate-resistant, transport-defective, murine tumors was carried out. The ECL data in Fig. 7A delineated a single 46-kDa protein band following SDS-PAGE of plasma membrane from L1210, P388, Ehrlich, M5076, and S180 cells. Moreover, the different intensity of the blots indicated that there was some variability in levels of immunoreactive RFT found in each case. All these cells display one-carbon, reduced folate transport as determined in a functional assay (25,26).
The ECL data shown in Fig. 7, B and C, also delineate a single protein band in the vicinity of 46 kDa following SDS-PAGE of plasma membrane from several independently isolated transport defective L1210 cell variants. All of these variants have markedly lower levels (Table II) of one-carbon, reduced folate transport in comparison to parental L1210 cells and in contrast to the latter, trans-stimulation (2) by internalized L-L5CHO-folateH 4 of [ 3 H]MTX influx could not be demonstrated. Also, variants L1210/R26A-D have reduced influx V max , while L1210/R24 and R25 have reduced V max in addition to increased K m for MTX influx. On the basis of affinity labeling and Western blotting, these variants fall into two distinct a Calculated from the specific activity of [ 3 H]AMT and total specific dpm in the 45-47-kDa fraction of protein attained during SDS-PAGE. b Recovery reflects the loss of dpm occurring during each purification step and the fraction size selected for the subsequent step.

FIG. 4. Detection of purified RFT following SDS-PAGE by direct staining and Western blotting with anti-AMT antibody.
Aliquots of two different preparations of purified RFT (RFT#1 and RFT#2) were subjected to SDS-PAGE (10% gel), transblotted to nylon (polyvinylidene difluoride), and either stained with Coomassie Blue (A) or immunoblotted (B). Additional experimental details are given in the legend to Fig. 1 and in the text. classes as shown in Fig. 7, B and C, and Table II. The variants L1210/R26A-D exhibit levels of RFT somewhat higher than parental L1210 cells. However, RFT in all of these variants migrates during SDS-PAGE at a rate slightly reduced relative to parental cell RFT, so that an aberrant molecular size was obtained for each which was approximately 10% higher than that determined for parental cell RFT (46 kDa). Variants L1210/R24 and 25 were even more unusual. RFT in these variants was not affinity labeled by NHS-[ 3 H]AMT (Fig. 2, Table II) but was found in the plasma membrane, based upon the ECL data, at levels 4-(L1210/R24) to 8-fold (L1210/R25) higher than in parental L1210 cells.
Relative Content of RFT in Transport-altered Human Tumor Cells-Plasma membrane from parental HL-60 cells and parental CCRF-CEM cells and variants (27,28) with increased (CEM/7A) or decreased (CEM/MTX-T) one-carbon, reduced folate transport were subjected to SDS-PAGE and immunoblotted with anti-murine RFT antibody. From the ECL data given in Fig. 8, it is seen that an 83-kDa protein was blotted in each case. Moreover, the intensity of the blot was highest in the case of the CCRF-CEM variant with elevated transport and lowest in the case of the variant with decreased transport. In addition to this 83-kDa protein band a more rapidly migrating protein band was also discernable but primarily in the plasma membrane from transport elevated CCRF-CEM cells. As this second band by comparison, decreased disproportionally in the plasma membrane from transport-elevated, parental, and transport-reduced cells and increased in the absence (data not shown) of proteinase inhibitors, we assume that this band is a product of proteolysis. Even though proteinase inhibitors were utilized during the preparation of these plasma membranes, we have usually found (10) that human RFT in comparison to murine RFT was more susceptible to hydrolytic cleavage by proteases particularly when overproduced in transport elevated variants (10). DISCUSSION The results demonstrate that ligand-directed immunoaffinity fractionation with preparative SDS-PAGE is a reasonable approach to the purification of RFT from L1210 cells. Also, the availability of an L1210 cell variant overproducing RFT to a very substantial extent (6) contributed greatly to our realizing adequate amounts of purified RFT for further studies. Although this method of purification has inherent limitations with regard to the maximum degree of purification attainable, it seems likely that the RFT consistently recovered will continue to be greater than 95% in purity. No indication of more substantial amounts of impurity, i.e. manifested as multiple protein bands in the size selected region, was obtained either by direct staining of protein or immunoblotting following SDS-PAGE. Also, the results in Fig. 7, A and B, showing differences in the level of RFT among various tumors and a small but perceptible shift in electrophoretic migration of RFT from four transport-defective variants, could not be obtained if the antiserum to purified RFT contained significant amounts of antibody to contaminating proteins similar in size to RFT. Conversely, since most integral plasma membrane proteins are glycosylated, significant amounts of such proteins as contaminants would have been visible after migration away from glycanase-treated RFT during SDS-PAGE.
The amount of RFT generated by the above methodology was very adequate for eliciting polyclonal antibody formation in rabbits. Western blotting with this immunoabsorbed antibody preparation of plasma membrane from L1210 and L1210/R83 cells detected the 46-kDa RFT in relative amounts commensurate with higher one-carbon, reduced folate transport and NHS-AMT affinity labeling of RFT in L1210/R83 cells (6). Of interest as well were results showing that these antibodies also blotted RFT in plasma membrane of other tumors of different histologic origin and an 83-kDa RFT in plasma membrane from HL-60 and CCRF-CEM cells. Again, the relative extent of blotting was similar to the relative level of one-carbon, reduced folate transport and specific NHS-AMT labeling found (9,27,28) in parental and variant CCRF-CEM cells. These results on relative tissue content and interspecies cross-immunoreactivity of RFT lend considerable support to the notion we have expressed earlier (1), that one-carbon, reduced folate transport is a conserved property among tumor cells of different species.
These and earlier studies from the authors laboratory (6) and others (5,7,29) utilizing alternative methods of size estimation suggested that L1210 cell RFT exists in the plasma membrane as a 43-46-kDa protein in a form with little or no core oligosaccharides attached (14). In the latter case, this was concluded from an experiment examining (14) the effect of N-glycanase on the migration of stained purified RFT during SDS-PAGE. We have extended these studies in greater detail to show that immunologically delineated RFT has no detectable N-or Olinked oligosaccharides attached. These are in sharp contrast to results of very extensive prior studies (11,15,23) by others revealing N-and O-linked oligosaccharides on RFT in plasma membrane of human leukemia cells (K562 and CCRF-CEM). Most importantly, these studies showed that N-glycanase treatment of plasma membrane from cells grown in tunicamycin markedly reduced the molecular size of RFT to approximately 50 kDa, a size similar to that delineated for murine tumor RFT.
The availability of antibody to RFT has made it possible (15) (this study) to examine in more detail the properties of RFT in tumor cell variants with defective one-carbon, reduced folate transport that are resistant to classical folate analogues. The variants examined here revealed interesting differences in the amount and physical properties of RFT in addition to the functional modifications documented in Table II. One group of independently isolated L1210 cell variants (L1210/R26A-D) exhibited an isoform of RFT with a modest, but consistent, decrease in electrophoretic mobility compared to wild-type RFT. Two other variants (L1210/R24 and R25) exhibited markedly increased amounts of RFT despite the fact that these 46-kDa proteins could not be affinity labeled with NHS-AMT. Prior studies (7,29) of similar transport defective variants of the L1210 cell detected no differences in the amount or physical properties of RFT, while a transport-defective variant of CCRF-CEM cells exhibited slightly elevated amounts of a more rapidly migrating isoform of RFT during SDS-PAGE. In contrast, the transport-defective, variant of CCRF-CEM cells examined here and the variants of K562 cells examined by immunoblotting by others (15) showed only lower levels of RFT in the plasma membrane. Thus, it would appear that impaired function of one-carbon, reduced folate transport in tumor cell variants with acquired resistance to folate analogues can result from a variety of alterations affecting both the structure and amount of RFT.
The recent isolation of cDNAs from L1210 (30) and hamster ovary (31) cell recombinant cDNA libraries, which restore MTX transport in transport-defective cells, has potentially important implications for one-carbon, reduced folate transport in tumor cells. It is of interest, therefore, that our results further verify that the molecular size of native RFT expressed in L1210 cells is in the neighborhood of 46 kDa, while pRFC-1 derived from the L1210 cell cDNA library and a related hamster ovary cell cDNA appear to code for a protein of 58 kDa. An NHS-[ 3 H]AMT affinity-labeled protein of 46 kDa was previously delineated (6) among the plasma membrane proteins from L1210/R83 cells during molecular sieve chromatography, most likely ruling out that the 46-kDa RFT reflects an aberrancy in electrophoretic mobility. This discrepancy with regard to the molecular size of RFT remains unexplained at this time, but it clearly has implications for the identity of the protein coding for pRFC-1. Therefore, further work will be required before the exact relationship between pRFC-1 and its hamster and human homologue and RFT is understood, and we present these comments and our findings in this regard with no bias intended.
Acknowledgment-The samples of plasma membrane from the various sublines of CCRF-CEM cells were provided by Doctor G. Jansen, Department of Oncology, Free University of Amsterdam, The Netherlands.  8. Relative amount and properties of RFT in the plasma membrane of human leukemia cells and their variants as detected by Western blotting. The indicated amounts of purified plasma membrane from CCRF-CEM cells and its transport elevated (CEM/7A) or transport reduced (CEM/T) variants and HL-60 cells were run on a 7.5% polyacrylamide gel prior to transblotting. Additional details pertaining to this typical blot are (one of two) provided in the text.