Cellular Folates Prevent Polyglutamation of 5,10-Dideazatetrahydrofolate

Mouse L1210 cell variants were selected for resistance to 5,10-dideazatetrahydrofolate, a potent inhibitor of the first folate-dependent enzyme in de novo purine synthesis, glycinamide ribonucleotide formyltransferase. The drug-resistant phenotype selected was conditional to the folate compound used to support growth: grown on folic acid cells were 400-fold resistant, whereas they were 2.5-fold more sensitive to 5,10-dideazatetrahydrofolate than wild-type L1210 cells when grown on folinic acid. In folic acid-containing media, polyglutamation of 5,10-dideazatetrahydrofolate was markedly reduced, yet folylpolyglutamate synthetase activity was not different from that in parental L1210 cells. Resistance was due to two changes in membrane transport: a minor increase in the K m for 5,10-dideazatetrahydrofolate influx, and a major increase in folic acid transport. Enhanced folic acid transport resulted in an expanded cellular content of folates which blocked polyglutamation of 5,10-dideazatetrahydrofolate. We propose that polyglutamation of 5,10-dideazatetrahydrofolate is limited by feedback inhibition by cellular folates on folylpolyglutamate synthetase, an effect which reflects a mechanism in place to control the level of cellular folates. Although the primary alteration causative of resistance is different from those reported previously, all 5,10-dideazatetrahydrofolate resistance phenotypes result in decreased drug polyglutamation, reflecting the centrality of this reaction to the action of 5,10-dideazatetrahydrofolate.

Biochemical and molecular analysis of the mechanisms of acquired resistance of tumor cells to a drug has historically been a powerful source of information on which steps in the action of a drug are necessary for cytotoxicity. Several tumor cell lines have been selected for resistance to (6R)-DDATHF (18 -20). To date, tumor cell resistance to (6R)-DDATHF has been attributed to decreased drug transport (18), decreased FPGS activity (20), and increased ␥-glutamylcarboxypeptidase (hydrolase) activity (19), all of which result in the reduction in the steady state level of cellular (6R)-DDATHF polyglutamates. In this report, we describe a unique mechanism by which murine leukemic L1210 cells develop resistance to (6R)-DDATHF (and cross-resistance to ZD-1694): an expansion of the intracellular folate pool with consequent blockade of the synthesis of (6R)-DDATHF polyglutamates. This unexpected mutant phenotype appears due to a suspected but heretofore unproven feedback control mechanism on the synthesis of cel-lular (anti)folate polyglutamates, presumably by direct effects of cellular folates on FPGS. H]5-formyltetrahydrofolate were purified by reverse phase HPLC eluted by a linear gradient of methanol and 0.03 M sodium acetate (4 -12% methanol over 12 min for folic acid; 3-15% methanol over 15 min for 5-formyltetrahydrofolate) and were stored at Ϫ20°C for no more than 2 weeks prior to use. [ 3 H]FdUMP was purified by DEAE column chromatography (21). (6S)-H 4 PteGlu and 10-CHO-H 4 PteGlu were prepared as described previously (22).  (23) in a total volume of 30 l of 0.2 M Tris, pH 9.0, containing 0.2 mg/ml bovine serum albumin. The reaction was stopped after 2 h by heating at 100°C for 3 min. The major product was the tetraglutamate derivative, which was identified by cochromatography with authentic DDATHF polyglutamate standards on a reverse phase, paired-ion HPLC column. Product was purified, first on a Sep-Pak C 18 column (Waters Associates, Milford, MA), then on a 10 ϫ 0.46-cm Luna 3 m C-18 (Phenomenex, Torrance, CA) eluted with a multiphase gradient of methanol in acqueous tetrabutylammonium hydrogen sulfate (Pic A reagent, Waters Associates) run at 0.6 ml/min. To generate this gradient, mobile phase methanol concentration was initially 27%, then was increased to 35% in a linear gradient over 10 min; subsequently, a less steep linear gradient was initiated which reached a methanol concentration of 42% after an additional 15 min; the methanol concentration was then held at 42% for the next 10 min. Pooled HPLC fractions containing (6R)-[ 3 H]DDATHF tetraglutamate were dried under vacuum, dissolved in 100 l of water, and passed through a 1-ml column of AG 50W-X8 (Bio-Rad) to remove the ion-pairing reagent; the product was concentrated and stored at Ϫ20°C in 20% ethanol. The amount of tetraglutamate product in an HPLC run was quantitated against the area of a known amount of monoglutamate standard.
Synthesis of 3  H]diethyl ester of (6R)-DDATHF with 1 N sodium hydroxide for 6 h at room temperature. The product was purified by chromatography on a column of DEAE-cellulose and further purified on a paired ion-reverse phase HPLC system and the tetrabutylammonium ion present in the mobile phase was removed as described above. The product was stored at Ϫ20°C in 33% ethanol. The purity and specific activity of the final product was determined by paired ion HPLC and scintillation counting; coinjection of (6R)-[ 3 H]D-DATHF with standard (6R)-DDATHF allowed purity to be estimated at Ͼ98%. (6R)-[ 3 H]DDATHF was stable (Ͻ2% impurities detected by HPLC) for at least 2 weeks when stored at Ϫ20°C. The stability of the radiolabel was studied by HPLC after incubation at 37°C in transport buffer (see below); purity was 95% after 24 h, but dropped to about 70% after 72 h.
Selection of DDATHF-resistant L1210 Cells-Mycoplasma-free mouse L1210 cells were passaged in RPMI 1640 medium supplemented with 10% dialyzed FCS in the presence of increasing concentrations of (6R,6S)-DDATHF until the culture resumed the same growth rate as a parallel culture maintained without drug. Initial drug concentration was 0.05 M; thereafter, drug concentration was increased 2-6-fold every 3-4 weeks. Continuous passage of cells resistant to 3 M (6R,6S)-DDATHF for an additional 5 months in 10 M drug did not allow the emergence of a phenotype which could grow rapidly at that concentration of selective agent. Sublines of L1210 cells which grew in 0.5, 3.0, and 10 M (6R,6S)-DDATHF were initiated from clones grown in soft agarose (24), and used for all comparisons against similarly cloned wild-type L1210 cells.
Growth Inhibition Studies-Exponentially growing cells were transferred to drug containing medium in 24-well plates at an initial density of 2 ϫ 10 4 /ml in a total volume of 1.5 ml. Culture density was determined after 72 h at 37°C and compared with the density of wells determined at time 0. For short term exposure to MTX, cells were incubated with drug for 6 h, washed with prewarmed PBS, and resuspended in drug-free medium; cell density was determined after a total of 72 h of growth. IC 50 values were determined by interpolation (24).
Purification of GARFT for Kinetic Studies-GARFT was purified to electrophoretic homogeneity as described previously (6). This procedure is based on affinity chromatography on a 3-ml column of Sepharose 4B to which 10-formyl-5,8-dideazafolate was attached via an ethylene linker. GARFT was then eluted with a 3 mM solution of 10-formyl-5,8dideazafolic acid and was passed through a 10-ml column of Sephadex immediately prior to kinetic experiments.
Enzyme Assays-GARFT activity was assayed using a spectrophotometric assay which followed the rate of conversion of 10-CHO-5,8dideazafolic acid to 5,8-dideazafolate (6,25). Assays contained 11 M 10-formyl-5,8-dideazafolic acid and 10 M ␣,␤-glycinamide in a 1-cm cuvette. Cellular FPGS activity was determined using (6S)-tetrahydrofolate as a substrate (26). For some experiments, FPGS was partially purified as described previously and kinetic experiments were performed using a charcoal-adsorption based assay with either aminopterin or DDATHF as substrates (27). For determination of FPGS and GARFT activities, enzyme assays were performed on a high speed supernatant fraction produced by centrifugation for 30 min at full speed in a Beckman Airfuge (about 200,000 ϫ g) or for 1 h in a Beckman Ti50 rotor at 160,000 ϫ g. ␥-Glutamylcarboxypeptidase (conjugase) activity was measured by the release of glutamic acid from (6R)-DDATHF tetraglutamate, which was tritium-labeled in the second through fourth side chain glutamate moieties. For conjugase assays, 8 ϫ 10 7 cells were suspended in 1 ml of 50 mM Tris acetate buffer, pH 6.0, containing 50 mM 2-ME, and the cells were broken by 3 ϫ 20 strokes of a hand-held Dounce homogenizer. The lysate was centrifuged for 20 min at 14,000 rpm at 4°C in a Microfuge, and the supernatant was used for assays. Protein (0 -30 g) was incubated with 100 M (6R)-DDATHF- [3, H]Glu 3 (0.08 Ci/assay) in lysate buffer for up to 30 min and the reactions were stopped by the addition of 500 l of a suspension of activated charcoal in 10 mM glutamate, 10 mM 2-ME, and 150 mM KH 2 PO 4 , pH 5.0. The charcoal had been pretreated with Dextran T-70 (27). The reaction mixtures were centrifuged in a Microfuge for 5 min and the supernatant added to scintillation fluid for determination of radioactivity.
Analysis of Intracellular Metabolism of (6R)-DDATHF to Polyglutamates-Log-phase cells, grown in RPMI 1640 medium at a final concentration of 2.0 M folic acid, were treated with (6R)-[ 3 H]DDATHF for 16 h in the presence of 32 M hypoxanthine and 5.6 M thymidine. In some experiments, cells were depleted of intracellular folates by 6 days of exponential growth in RPMI 1640 medium formulated without folic acid and supplemented with 10% dialyzed FCS, 32 M hypoxanthine, and 5.6 M thymidine. Cellular folates were reduced to less than 0.02% of control by this procedure (22). After incubation with (6R)-[ 3 H]D-DATHF, cells were harvested by centrifigation and washed twice with 10 ml of ice-cold PBS containing 5% FCS. A known number of cells were resuspended in 100 l of 5 mM tetrabutylammonium hydrogen sulfate (Pic A reagent, Waters Instruments, Milford, MA), sonicated for 8 -10 1-min pulses in a Heat Systems Ultrasonics sonic disruptor using a cup horn attachment, and the broken cell suspensions were subsequently boiled for 3 min. The samples were filtered through a Microcon 10 filter (Amicon Corp.) and an aliquot was analyzed by HPLC. Recoveries from filtration were noted and used to adjust subsequent calculations. (6R)-DDATHF polyglutamates were separated on a 10 ϫ 0.32-cm column of 3-m pelicular C18 reverse phase column (Applied Biosystems) as described above. Fractions of 200 l were collected, radioactivity was determined by liquid scintillation counting, and the identity of labeled peaks was determined from the retention times of chemically synthesized polyglutamate derivatives of (6R)-DDATHF.
Folate Pool Measurements-For measurement of the total pool of folate derivatives, cells were cultured in RPMI 1640 medium formulated without folic acid to which was added 10% dialyzed FCS and either 2.0 M [ 3 H]folic acid or 60 nM folinic acid spiked with high specific activity (6S)-[ 3 H]5-formyltetrahydrofolate. After 1 week of exponential growth in labeled folate, the specific activity of the intracellular folates becomes equivalent to that in the medium, and pool size can be estimated by the level of intracellular radioactivity (22). Cells were harvested by centrifigation, and the washed cell pellet was dissolved in 0.5 ml of 1 N NaOH. The cell density of an aliquot of the third wash was determined electronically. The lysate was neutralized with HCl and radioactivity was determined by scintillation counting. Standard labeled compound was counted under the same conditions to allow counts/ min to be directly converted to pmol of intracellular folates.
Cellular levels of 10-CHO-H 4 PteGlu n , and the sum of the H 4 PteGlu n and 5,10-CH 2 -H 4 PteGlu n pools were estimated using a modification of procedures (28,29) based on the entrapment of 5,10-CH 2 -H 4 PteGlu n by excess thymidylate synthase and [ 3 H]FdUMP as a macromolecular ternary complex. The content of H 4 PteGlu n and 5,10-CH 2 -H 4 PteGlu n was determined by this method, then the cellular levels of 10-CHO-H 4 PteGlu n were determined indirectly by quantitating the additional H 4 PteGlu n formed in the presence of excess GARFT and glycinamide ribonucleotide. Briefly, cell pellets containing 4 ϫ 10 6 cells were resuspended in 200 l of boiling 10 mM sodium phosphate buffer, pH 7.5, containing 0.1% Triton X-100, 1% 2-ME, and 1% freshly prepared sodium ascorbate. After 3 min, the suspensions were brought to 0°C, and centrifuged at 14,000 rpm for 2 min in a refrigerated Microfuge. The pellets were extracted again and the extracts were combined. Aliquots of extract were added to 125 l of a solution containing 10 mIU of pure L. casei thymidylate synthase, 0.16 M [ 3 H]FdUMP, and 20 M formaldehyde in 10 mM sodium phosphate buffer, pH 7.4, containing 1% 2-ME and 0.2 mg/ml bovine serum albumin, or 200 l of this same solution containing, in addition, 1 mIU of recombinant mouse GARFT (4) and 10 M ␣,␤-glycinamide ribonucleotide. After incubation at 30°C for 2 h, the tubes were boiled for 10 min to denature the ternary complex formed, 200 l of 1% bovine serum albumin was added as carrier, and protein was precipitated by the addition of 4 ml of 8% ice-cold trichloroacetic acid. The final pellets were dissolved in 100 l of 1 N NaOH, neutralized with 750 l of 0.1 N HCl, 0.2 M KCl, and radioactivty was determined by liquid scintillation counting. Standard curves were simultaneously run using both (6S)-H 4 PteGlu and 10-CHO-H 4 PteGlu as controls.
Northern Analysis-RNA from cell lines was extracted using the Triazol reagent (Life Technologies, Inc., Gaithersburg, MD), denatured with glyoxal, and separated by size on a 1.2% denaturing agarose gel in 10 mM sodium phosphate buffer, pH 7.0. The RNA was transferred onto nylon membranes (Biotrans, ICN, Irvine, CA) and hybridized with either a 1.7-kilobase probe representing the downstream sequences of the mouse L1210 cell FPGS cDNA (30) or an 800-base pair probe corresponding to the glycinamide ribonucleotide synthetase domain of the polyfunctional GARFT mRNA (probe pQ0.8 (31)). Probes were random labeled to a specific activity of about 2 ϫ 10 9 cpm/g and were used at a concentration of 1 Ci/ml hybridization solution (0.5 M sodium phosphate, pH 7.0, 7% SDS, 1% bovine serum albumin, and 1 mM EDTA). All blots were hybridized at 65°C and filters were washed to a stringency of 0.2 ϫ SSC and 0.1% SDS at 65°C; a human glyceraldehyde-3-phosphate dehydrogenase probe was used to normalize for RNA loading.
Transport Studies-Exponentially growing cells were harvested by centrifugation, washed once with prewarmed PBS containing 5% FCS, and once with transport buffer (107 mM NaCl, 20 mM Tris-HCl, 26.2 mM NaHCO 3 , 5.3 mM KCl, 1.9 mM CaCl 2 , 1 mM MgCl 2 and 7 mM D-glucose, pH 7.4, at 37°C) (32). For studies on influx rate, cells were resuspended in 0.25 ml of transport buffer at a density of 10 7 cells/ml at 37°C. Transport was initiated by the forceful addition of 0.25 ml of radiolabeled folates in prewarmed transport buffer and quenched at the indicated times by the addition of 10 ml of ice-cold PBS containing 5% FCS. Cells were then washed an additional two times with 10 ml of ice-cold PBS containing 5% serum, and the final pellets were dissolved in 0.5 ml of 1 N NaOH, neutralized with HCl, and radioactivity was determined by liquid scintillation counting. For measurement of folic acid transport, cells were treated with 10 M trimetrexate at 37°C for 10 min prior to harvest, to inhibit dihydrofolate reductase (33). For studies on (6R)-DDATHF efflux, cells were preloaded with radiolabeled drug for 20 min, washed twice with ice-cold PBS containing 5% serum, resuspended in drug-free medium at 37°C, and aliquots of cell suspension were withdrawn with time and processed as above for determination of intracellular label.

Development of L1210 Sublines Resistant to (6R)-DDATHF-
Sublines of murine leukemic L1210 cells were selected by continuous exposure to stepwise increments in concentration of (6R,6S)-DDATHF. Whereas the parental L1210 cells were halfmaximally inhibited by 2.8 ϫ 10 Ϫ8 M (6R,6S)-DDATHF, L1210/ D0.5 and/D3 grew in 0.5 and 3 M selecting agent, respectively, with no detectable change in growth rate (doubling time of 10 -12 h) (Fig. 1). Despite several months of additional selection in 10 M (6R,6S)-DDATHF, more highly resistant cells did not emerge, although a cell line with a slower growth rate, the L1210/D10 cell, was isolated (Fig. 1). The resistance of L1210/ D10 cells to (6R,6S)-DDATHF was found to be stable for at least 9 months during continued growth in the absence of drug; the sensitivity of L1210/D3 cells to (6R)-DDATHF was also found to be stable after growth for 6 months in drug-free media. The highly resistant L1210/D3 cell line showed some interesting characteristics and became the focus of this study. The inhibition of growth of L1210/D3 cells was reversed by either hypoxanthine or aminoimidazole carboxamide (data not shown), indicating that the target of (6R,6S)-DDATHF in these cells was glycinamide ribonucleotide formyltransferase (GARFT), as was the case in parental L1210 cells (34).
The L1210/D3 cell was equivalently resistant to both diastereomers of DDATHF differing in chirality at the C6 position (Table I). This was not surprising; the two compounds have previously been shown to be almost identically inhibitory to the growth of parental L1210 cells, as a result of the remarkably similar biochemical activities of each diastereomer as inhibitors of GARFT and as substrates for FPGS (5). The phenotype displayed by these variant cells was very unusual in that resistance to (6R)-DDATHF was observed only when cells were grown in medium containing folic acid, the medium in which they were derived. When cells were grown in folate-free medium supplemented with 60 nM folinic acid, 3  subline was, paradoxically, more sensitive to drug than the wild-type L1210 (Table I). However, even after a prolonged culture period (Ͼ3 weeks) in folinic acid, resistance to (6R)-DDATHF resumed immediately when cells were transferred back to medium containing folic acid (Table I). From the stability of this phenotype in the absence of selection, we concluded that one or more mutations had been fixed in gene(s) critical to the action of (6R)-DDATHF.
Cross-resistance and Collateral Sensitivity of L1210/D3 Cells to Related Antimetabolites-The sensitivities of L1210 and L1210/D3 cells to other antimetabolites inhibitory to folatedependent enzymes were compared (Table I). With continuous exposure to drug during a 72-h period of growth, L1210/D3 cells showed a 15-fold resistance to the thymidylate synthase inhibitor ZD-1694, but only a 2-fold resistance to the dihydrofolate reductase inhibitor MTX. When drug exposure was limited to the first 6 h of a 72-h growth period, the resistance of L1210/D3 cells to MTX was substantially higher (8-fold). This scheduledependent resistance to MTX was similar to that reported for cells with an impaired capacity for polyglutamation of this drug (36). It was interesting to note that, although deficiency in polyglutamation has been implicated as a mechanism of resistance to fluoropyrimidines (37), L1210/D3 cells were collaterally more sensitive to FUdR compared with parental L1210 cells ( Table I).
Characteristics of GARFT Expressed by DDATHF-resistant Cells-GARFT, the enzyme previously found to be the target of DDATHF in L1210 cells (5,34), was studied as a potential site for the resistance to this drug in L1210/D3 cells. The specific activities of this enzyme were not different in cytosolic preparations from L1210 and L1210/D3 cells (Table II). Northern blots also indicated that the representation of message for the trifunctional GARFT was equally abundant in L1210/D3 and parental L1210 cells ( Fig. 2A) and that the sizes of both the trifunctional message and the monofunctional glycinamide ribonucleotide synthetase message transcribed from the GARFT gene (31,38) were the same in mutant and wild-type L1210 cells. GARFT was purified by affinity chromatography from L1210/D3 cells and the sensitivity of this enzyme to inhibition by (6R)-DDATHF and the pentaglutamate derivative of (6R,6S)-DDATHF was studied. Both were similar to that pre-viously observed for enzyme purified from parental L1210 cells (Table II), and the kinetic characteristics of enzyme purified from L1210/D3 cells were the same as those found for L1210 cells. Because poor inhibition of GARFT did not seem to be responsible for the refractoriness of L1210/D3 cells to drug exposure, it seemed most likely that either membrane transport or cellular metabolism of DDATHF to polyglutamates was impaired in the resistant cells.
FPGS and Conjugase Activities-The characteristics of the late, the equivalent of 60 nM folinic acid (24,35). The proliferation of L1210/D3 cells in medium lacking folic acid but containing 60 nM folinic acid was the same as that of L1210 cells. a Growth inhibition was measured after continuous exposure of cells to the listed drugs for a 72-h growth period, except where noted, and culture densities were determined electronically. Values are represented as mean Ϯ S.D. or 0.5 ϫ range (for n ϭ 2) obtained from n independent experiments.
b These experiments were performed in RPMI 1640 medium formulated without folic acid but containing 6.0 ϫ 10 Ϫ8 M folinic acid. Cells were grown in this medium for 1 week prior to growth inhibition studies.
c After 2 weeks of growth in folinic acid medium, cells were transferred back to standard medium containing 2.0 M folic acid and used for growth inhibition studies in this medium. d In these experiments, cells were exposed to MTX for 6 h, washed, and then resuspended in drug-free medium and culture densities were measured after a total incubation of 72 h.

TABLE II Activity and kinetics of GART and FPGS in L1210
and L1210/D3 cells The folate substrates used for the GART and FPGS activity measurements were 10-formyl-5,8-dideazafolate and (6S)-tetrahydrofolate, respectively (6,26). Kinetic analysis of GART inhibition was performed on GART purified by affinity chromatography from both cell lines. K i values were obtained from Dixon plots using K m values of 75 nM (4). Values are expressed as mean Ϯ S.D. or 0.5 ϫ range (for n ϭ 2) from n experiments. The kinetic parameters for FPGS expressed by L1210 and L1210/D3 cells were determined on enzyme concentrated and purified Ϸ20-fold by (NH 4 ) 2 SO 4 precipitation (27); the V max and first order rate constants (kЈ) for DDATHF isomers are expressed relative to the values for aminopterin measured in each experiment. The limits of detection in these assays. In the same experiments, protein from CEM cells cleaved glutamic acid from DDATHF tetraglutamate at 4.9 Ϯ 0.2 nmol/hr mg protein.
two enzymes involved in maintenance of steady state levels of folylpoly-␥-glutamates, FPGS and ␥-glutamyl carboxypeptidase, were studied in DDATHF-resistant and -sensitive L1210 cells. Using (6S)-tetrahydrofolate as a standard substrate, the activities of FPGS found in high-speed supernatant fractions of L1210 and L1210/D3 cell lines were not significantly different (Table II). Likewise, Northern blots of mRNA from L1210/D3 cells indicated similar abundance and size of FPGS transcripts as found in parental L1210 cells (Fig. 2B). A kinetic analysis performed on FPGS partially purified from both cell lines showed indistinguishable characteristics using aminopterin as a substrate. Likewise, the kinetics of the utilization of either (6R)-DDATHF or (6S)-DDATHF were also identical for FPGS isolated from both lines (Table II). ␥-Glutamyl carboxypeptidase (conjugase) was not detectable in either L1210 or L1210/D3 cells, using experimental conditions under which this enzyme was found at 4.9 Ϯ 0.2 nmol min Ϫ1 (mg protein) Ϫ1 (n ϭ 2) in CCRF-CEM cells, a value similar to that previously reported for this cell line (39).
Membrane Transport of (6R)-DDATHF-In order to determine whether resistance to (6R)-DDATHF in L1210/D3 cells was due to defective membrane transport, high specific activity (6R)-[ 3 H]DDATHF was synthesized, and translocation of drug across the plasma membrane was studied. At 2 M extracellular (6R)-[ 3 H]DDATHF, a concentration which completely suppressed the growth of L1210 but had no effect on that of L1210/D3 cells (see Fig. 1), the uptake of radiolabeled drug was linear in both cell lines over the first 6 min. The initial velocity of transport of (6R)-DDATHF was 0.63 Ϯ 0.10 pmol min Ϫ1 per 10 6 cells for L1210 and 0.38 Ϯ 0.01 pmol min Ϫ1 per 10 6 cells for L1210/D3, respectively (n ϭ 3) (Fig. 3A). Influx of (6R)-[ 3 H]D-DATHF was measured over the initial 4 min after the addition of various concentrations of drug. The transport of (6R)-[ 3 H]D-DATHF in L1210 cells was very similar to that reported by others in CEM human leukemic cells (40). The influx K m for transport of (6R)-DDATHF into L1210/D3 cells was 3-fold higher than that in L1210 cells (5.7 Ϯ 0.2 and 1.7 Ϯ 0.6 M, respectively (n ϭ 3)); however, the influx V max values were not significantly different between the two cell lines (0.93 Ϯ 0.12

FIG. 3. Kinetics of influx and efflux of (6R)-DDATHF from L1210/D3 (filled boxes) and wild-type L1210 cells (open boxes).
Panel A depicts the time course of influx of (6R)-DDATHF at 37°C and an extracellular concentration of 2 M. Panel B shows the concentration dependence of transport into cells measured at 37°C over 4 min. The efflux of label from cells is shown in panel C following a 20-min preexposure of L1210 and L1210/D3 cells to 2 and 6 M (6R)-DDATHF, respectively. The dashed arrow in C indicates the level of the target enzyme for DDATHF, GARFT, in these cells. This value was calculated from the data of Table II, and Ref. 6. For details, see text. and 1.2 Ϯ 0.1 pmol/min per 10 6 cells in L1210 and L1210/D3 cells, respectively (n ϭ 3)) (Fig. 3B). L1210 and L1210/D3 cells were exposed to 2 and 6 M (6R)-DDATHF, respectively, for 20 min to allow accumulation of equivalent intracellular levels of radiolabeled drug, and the rate of efflux was measured in drug-free buffer over a 30-min period. The initial rate of drug efflux was similar in both cell lines (Fig. 3C). It should be noted that these experiments were performed under conditions in which the initial drug concentration in the cells was in excess over the level of the drug target, GARFT, whose cellular content is indicated by the dashed arrow in Fig. 3C. Hence, virtually all of the intracellular drug would be expected to be free drug, not bound to protein, and the initial slope of these curves would indicate efflux rates. However, in wild-type L1210 cells, about one-third of the intracellular drug appeared to be nonexchangeable and this component was absent in the resistant cells (Fig. 3C). This poorly exchangeable fraction appeared to reflect the accumulation of polyglutamate metabolites in L1210 but not in L1210/D3 cells (see below). Hence, there was a minor change in the transport of (6R)-DDATHF in the L1210/D3 cells, namely a 3-fold increase in influx K m , which would contribute to the overall level of resistance in these cells, but could not explain the extent of resistance observed in these cells (Table  I).
Cellular Polyglutamation of (6R)-DDATHF-The metabolism of (6R)-DDATHF to polyglutamates in intact L1210 and L1210/D3 cells was studied by HPLC. When cells were treated with 0.5 M (6R)-[ 3 H]DDATHF for 16 h, wild-type L1210 accumulated 13 pmol of drug per 10 6 cells, 84% of which were in the form of long chain metabolites (Glu 4 -Glu 8 ), with the hexaglutamate metabolite predominating (Fig. 4A). On the other hand, L1210/D3 cells accumulated only 0.82 pmol of drug per 10 6 cells, with long chain polyglutamates accounting for less than 0.34 pmol per 10 6 cells (42% of total intracellular drug). Hence, long chain polyglutamates of DDATHF were found in L1210/D3 cells at less than 3% of the level in parental L1210 cells. Assuming 0.67 l of intracellular water per 10 6 L1210 cells, total long chain polyglutamates of (6R)-DDATHF were present at 0.51 and 10.9 M concentrations in L1210/D3 and L1210 cells, respectively; the former is less than the cellular content of GARFT in L1210 cells (calculated to be 1 M from the results presented in Ref. 6), the latter substantially higher than target enzyme levels. In these experiments, the level of unmetabolized (6R)-DDATHF in L1210/D3 cells was also 3-fold lower than that in L1210 cells, presumably due to the higher K m for drug transport. To allow a legitimate comparison of the rate of polyglutamation between the two cell lines, L1210/D3 cells were treated with higher concentrations of drug (2 and 5 M) in order to compensate for the (K m ) transport defect in these cells and to achieve a steady state level of intracellular (6R)-DDATHF monoglutamate comparable to that in L1210 cells exposed to 0.5 M (6R)-DDATHF. Even at (6R)-DDATHF monoglutamate concentrations in L1210/D3 cells equal to or somewhat greater than those in L1210 cells, accumulation of (6R)-DDATHF polyglutamates was still markedly lower in L1210/D3 cells (Fig. 4B). We concluded that impaired polyglutamation of (6R)-DDATHF in L1210/D3 cells was not secondary to a decrease in drug transport, but rather it reflected a substantial dysfunction in the polyglutamation process per se.
Folate Cofactor Pools in L1210 and L1210/D3 Cells-The low accumulation of (6R)-DDATHF polyglutamates in L1210/D3 cells prompted us to suspect that intracellular endogenous folates would also be low in these cells. When total cellular folates were assessed in cells cultured in [ 3 H]folic acid, surprisingly, L1210/D3 cells showed a 4 -5-fold elevation (Table III). The majority of the cofactors in these cells appeared to be in polyglutamate forms since about 80% of the total radiolabeled material was retained after incubation was continued in medium containing unlabeled folic acid for 12 h (data not shown). The level of the substrate for the GARFT reaction, 10-formyltetrahydrofolate, which, presumably, was present mainly as polyglutamate derivatives, was measured using a modified version of the thymidylate synthase ternary complex formation assay (28,29). The concentration of this cofactor form increased somewhat disproportionally to the increase in the total folate pool, and was 7-fold higher in L1210/D3 cells than in wild-type L1210 (Table III). Conversely, when cells were grown in folate-free medium supplemented with (6S)-[ 3 H]5-formyltetrahydrofolate, the total folate level in the wildtype cells was 4-fold higher than that in the resistant cells (Table III). Because L1210/D3 cells were resistant to (6R)-DDATHF only when they were grown in medium containing folic acid but not in medium formulated with folinic acid, this pattern of intracellular folate levels led us to believe that the resistance phenotype of L1210/D3 cells was caused by the increased cofactor pools in these cells.
Effect of the Expanded Folate Pools on the Accumulation of Intracellular (6R)-DDATHF Polyglutamates in DDATHF-resistant Cells-Drug polyglutamation was studied in parental and drug-resistant cells in which intracellular folate pools were depleted by growth for 10 generations in folate-free medium supplemented with 5.6 M thymidine and 32 M hypoxanthine (22). After incubation with 0.5 M (6R)-[ 3 H]DDATHF for 16 h in folate-free medium, L1210 cells contained 14 pmol of drug per 10 6 cells, 74% of which consisted of long chain polyglutamates; this accumulation and metabolism was similar to that observed in L1210 cells containing normal levels of folate pools (compare Fig. 4, A and C). Following incubation with either 0.5 or 2 M (6R)-[ 3 H]DDATHF, folate-depleted L1210/D3 cells accumulated long chain derivatives of (6R)-DDATHF to an extent that was indistinguishable from that found in L1210 cells (Fig.  4, C and D). Hence, the polyglutamation defect was not manifest in L1210/D3 cells in the absence of cellular folates. We concluded that the low polyglutamation of (6R)-DDATHF in L1210/D3 cells was caused by the elevated folate pools in these cells (Table III).
Polyglutamation of (6R)-DDATHF in Cells Grown in Folinic Acid Medium-The polyglutamation of (6R)-DDATHF was studied in RPMI 1640 medium supplemented with folinic acid rather than folic acid, a condition in which L1210/D3 cells were paradoxically more sensitive to drug-induced growth inhibition than were parental L1210 cells. The results of these studies indicated that there was no impediment to the polyglutamation of DDATHF in L1210/D3 cells (Fig. 5A) when grown in folinic acid and, if anything, polyglutamatate derivatives formed were of longer chain length than in parental L1210 cells (Fig. 5B).
Folic Acid Transport in L1210 and L1210/D3 Cells-In view of the low accumulation of (6R)-DDATHF in L1210/D3 cells grown on folic acid-containing medium, it was not clear why folate pools were expanded in these cells. The membrane transport of [ 3 H]folic acid in both cell lines was examined. In these studies, cells were pretreated with 10 M trimetrexate, a lipophilic inhibitor of dihydrofolate reductase, to prevent the metabolism of [ 3 H]folic acid by dihydrofolate reductase (33). In the presence of 5 M [ 3 H]folic acid, the initial uptake of radiolabeled material at 37°C was barely measurable in the L1210 line, a finding which was consistent with the previous reports that folic acid is a very poor substrate for membrane transport in these cells (K m Ͼ200 M) (32,41). However, under the same conditions, L1210/D3 showed a 50-fold increase in the initial uptake of [ 3 H]folic acid and a 15-fold increase in the steady state level of the radiolabeled compound, measured at 30 min (Fig. 6).

DISCUSSION
In this report, we describe the phenotypic characterization of an L1210 subline selected for resistance to (6R,6S)-DDATHF. We have demonstrated that the underlying primary event in this resistance was an alteration in the substrate specificity of folate transport across the plasma membrane. Resistance was not a direct result of a decreased transport of (6R)-DDATHF; the influx K m of (6R)-DDATHF being increased only 3-fold, compared with the 400-fold resistance to growth inhibition by (6R)-DDATHF. However, transport of folic acid had become remarkably efficient, leading to a substantially expanded cellular folate pool and a surprising subsequent blockade of (6R)-DDATHF polyglutamate synthesis in the resistant cells. In experiments in which the intracellular folates were removed by extensive growth in the absence of extracellular folic acid, polyglutamation of (6R)-DDATHF was normal in the resistant cells. This established that the expanded folate pool was causative of the polyglutamation defect and, hence, the drug resistance of this cell line. A related mechanism of drug resistance was reported recently when it was shown that CHO cells selected for resistance to pyrimethamine lost a folic acid efflux pump which allowed a substantial increase in folic acid accumulation (33).
Previous analysis of variant cell lines has supported the  1 week prior to quantitation of total cellular folates. Cells were then harvested by centrifugation, washed with PBS, solubilized in 1 N NaOH, and radioactivity was determined by scintillation counting. For measurement of cofactor levels, cells were grown in the indicated medium without radiolabeled material, folate cofactors were extracted from cells in boiling 10 mM phosphate buffer, containing 1% ascorbate and 1% 2-ME, and cofactor levels were determined using the ternary complex formation assay (28,29) as described under "Experimental Procedures." Values represents mean Ϯ S.D. or 0.5 ϫ range (for n ϭ 2) from n experiments. concept of a critical role of the metabolism of DDATHF to polyglutamates in the cytotoxicity induced by this drug. A series of cell lines selected for resistance to methotrexate were reported which had decreased levels of FPGS (36); these cells were cross-resistant to (6R)-DDATHF, and the degree of resistance was directly related to the decrease in enzyme levels. Two studies have previously reported the characterization of tumor cells selected directly for resistance to (6R)-DDATHF (19,20). In the one study (19), an increased activity of folyl-␥-glutamyl hydrolase, a carboxypeptidase responsible for the catabolism of folylpolyglutamates, was implicated as a cause of resistance to (6R)-DDATHF in H35 hepatoma cells. In a second report (20), two human leukemic CCRF-CEM cell lines with substantial acquired resistance to (6R)-DDATHF (24-and 217-fold) were shown to have severe impairment of the polyglutamation of the parent drug despite the fact that only a 30 -50% decrease was noted in the activity of FPGS in vitro. The mechanism of decreased polyglutamation in those cells remains unexplained. It is interesting to note that these resistant CCRF-CEM sublines also showed a somewhat elevated endogenous folate pool, suggesting that the phenomenon we now report might have been involved. A third subline from that study expressed very low levels of FPGS activity, despite a steady state content of mRNA for this protein equivalent to that in wild-type cells, suggesting a mutation in the coding region of this enzyme that rendered cells resistant to DDATHF. Hence, it is clear that any of several mechanisms which lead to a decreased cellular content of DDATHF polyglutamates will compromise the cytotoxicity of this agent.
The mechanism responsible for the limited accumulation of (6R)-DDATHF polyglutamates in L1210/D3 cells is fundamentally different from the direct effects on synthesis or degradation of DDATHF polyglutamates reported previously. The defect in polyglutamation in these cells is not a direct consequence of any biochemical alteration in FPGS per se nor an increased availability of folylpoly-␥-glutamate hydrolase activity, but rather appears due to a direct effect of intracellular folates on metabolism of (6R)-DDATHF to polyglutamates. The exact nature of this control mechanism is not yet clear. Previous studies have shown that the K m values for FPGS of most of the monoglutamate forms of the folate cofactors were comparable or even lower than that of (6R)-DDATHF (1, 42) and that K m values for FPGS within a series of homologous compounds generally decrease with increasing polyglutamate chain length (43,44). Hence, a simple mechanism that would explain this effect is that high levels of folate cofactors (or one particular form of cofactor) function as competitive inhibitors for the FPGS reaction blocking the metabolism of DDATHF to forms which are metabolically trapped in the cell. On the other hand, mammalian FPGS has been commonly found to display distinct substrate inhibition at higher concentrations of folate compounds (45), suggesting the possibility of a second site of binding of folates to this enzyme with regulatory significance.
There are aspects of the phenotype displayed by the L1210/D3 cells which are unique but which are explained by the changes in the folate cofactor pools in these cells. Thus, the drug resistance of L1210/D3 cells was displayed only when cells were grown in folic acid-containing medium; when the folate requirement for growth was sufficed by 5-formyltetrahydrofolic acid (folinic acid), L1210/D3 cells were, paradoxically, more sensitive to (6R)-DDATHF than parental L1210 cells. Resistance appears to depend on the folate source because the intracellular pool was only expanded in folic acid-containing medium; in folinic acid medium, cellular folates were diminished. In addition, previous literature has shown that mammalian tumor cells are more sensitive to FUdR in the presence of folinic acid (24,46,47) because of an expansion of the cellular 5,10-methylenetetrahydrofolate pool and a kinetic stabilization of the thymidylate synthase-5,10-methylenetetrahydrofolatefluorodeoxyuridylate ternary complex (48). Hence, it would be predicted that the elevated cellular folate pool in L1210/D3 cells would render them collaterally more sensitive to FUdR; this proved to be the case (Table I).
We initially anticipated that our results were more of theoretical significance than clinical relevance, because the exact mechanism of resistance would only occur when folic acid was a nutritional source of intracellular folate cofactors, and folic acid does not occur naturally. However, the recent preclinical and clinical work on (6R)-DDATHF suggest direct therapeutic relevance of our results (8). 2 Thus, the toxicity of (6R)-DDATHF during early clinical trials had demonstrated that the drug was substantially more toxic to patients than was predictable from preceding preclinical experience in animals (10). When mice were fed a folate-deficient diet for a short period, the toxicity of (6R)-DDATHF increased 1000-fold (8), mirroring the clinical pattern in animals for the first time. The mechanism presumed in the literature for this dramatic change in toxicity is that (6R)-DDATHF accumulates in liver of folate-deficient animals as polyglutamates and these hepatic drug metabolites are then slowly released to the circulation, mimicking a very toxic continuous infusion; presumably this would not occur in animals with higher folate intake. An alternative explanation would be that the polyglutamation and accumulation of DDATHF in the intestinal and bone marrow stem cells is similarly prevented by higher tissue folate stores resultant from higher dietary intake of folic acid. In either case, the toxicity of (6R)-DDATHF to animals or, presumably, man appears to be affected by a mechanism which is formally identical to that uncovered by the L1210/D3 mutation, namely, a blockade of the polyglutamation of (6R)-DDATHF by intracellular folate compounds.
Three distinct routes have been implicated in the membrane transport of folic acid in L1210 cells, including the reduced folate carrier (41), a carrier-mediated system that becomes predominant at low pH (49), and, after selection for the ability to grow at low concentrations of folates, the membrane-bound folate receptor (11,50). In the accompanying article (51), we demonstrate that the L1210/D3 cell has accumulated two point mutations in the reduced folate carrier gene, and that these mutations are causative of the phenotype we now report.