Impact of the reduced folate carrier on the accumulation of active thiamin metabolites in murine leukemia cells.

The thiamin transporter encoded by SLC19A2 and the reduced folate carrier (RFC1) share 40% homology at the protein level, but the thiamin transporter does not mediate transport of folates. By using murine leukemia cell lines that express no, normal, or high levels of RFC1, we demonstrate that RFC1 does not mediate thiamin influx. However, high level RFC1 expression substantially reduced accumulation of the active thiamin coenzyme, thiamin pyrophosphate (TPP). This decreased level of TPP, synthesized intracellularly from imported thiamin, resulted from RFC1-mediated efflux of TPP. This conclusion was supported by the following observations. (i) Efflux of intracellular TPP was increased in cells with high expression of RFC1. (ii) Methotrexate inhibits TPP influx. (iii) TPP competitively inhibits methotrexate influx. (iv) Loading cells, which overexpress RFC1 to high levels of methotrexate to inhibit competitively RFC1-mediated TPP efflux, augment TPP accumulation. (v) There was an inverse correlation between thiamin accumulation and RFC1 activity in cells grown at a physiological concentration of thiamin. The modulation of thiamin accumulation by RFC1 in murine leukemia cells suggests that this carrier may play a role in thiamin homeostasis and could serve as a modifying factor in thiamin nutritional deficiency as well as when the high affinity thiamin transporter is mutated.

The reduced folate carrier (RFC1), 1 first cloned in 1994, mediates transport of reduced folates critical to one carbonrequiring biosynthetic reactions in mammalian cells and is a member of the major facilitator superfamily of transporters (1)(2)(3). RFC1 also delivers MTX and new generation antifolates into a variety of tumors, particularly those of hematopoietic origin (4). RFC1 exchanges folates with a broad spectrum of inorganic and organic anions, and high extracellular concentrations of a variety of organic phosphates competitively inhibit RFC1-mediated folate influx (5)(6)(7). This interaction between RFC1 and organic phosphates results in the uphill transport of folates into cells linked to the organic phosphate gradient across cell membranes (5).
Structurally unrelated to the folates, thiamin plays an essential role in glycolysis and oxidative decarboxylation reactions after conversion to the coenzyme thiamin pyrophosphate by thiamin pyrophosphokinase in cells. Thiamin is also transported across cell membranes by a carrier-mediated process (8). Thiamin deficiency, reflected in a decrease in plasma thiamin concentration and TPP levels in erythrocytes, results in a variety of clinical abnormalities including cardiovascular and neurological disorders (9). Thiamin deficiency due to impaired transport results in the thiamin-responsive megaloblastic anemia syndrome, a disorder also associated with deafness and diabetes mellitus (10,11). Positional cloning with families inheriting this autosomal recessive disease led to the recent identification of the thiamin transporter gene SLC19A2 (12)(13)(14).
The thiamin transporter encoded by SLC19A2 is highly homologous to RFC1, sharing an amino acid identity of 40% and similarity of 55%, and both are predicted to have 12 transmembrane domains. Despite the similarity between these two proteins, the thiamin transporter, when expressed in HeLa cells, was not found to transport folates (15). In the current report, the impact of RFC1 function on thiamin transport and accumulation of its active coenzyme metabolites was studied in murine leukemia cells. Although RFC1 was not found to transport thiamin, it does transport phosphorylated thiamin derivatives, thereby modulating the intracellular accumulation of active thiamin metabolites.  (16), and tritiated thiamin and TPP were used directly after purity was confirmed by HPLC.

Chemicals
Cells Lines and Culture Conditions-G1a, G2, D10, and MTX r A cells were selected from murine leukemia L1210 cells in the presence of MTX, with or without chemical mutagenesis, as previously reported (17)(18)(19)(20)(21). All four cell lines harbor mutations in RFC1 resulting in impaired or absent transport mediated by RFC1. Both R16 and T2 are cell lines with high level expression of RFC1, obtained by transfecting murine RFC1 cDNA into MTX r A or wild-type L1210 cells, respectively (21,22). L7, L15, L44, and L51 are 5,10-dideazatetrahydrofolate-resistant L1210 variants isolated by chemical mutagenesis followed by selection in the presence of this drug. In these cell lines folylpolyglutamate synthetase was mutated, resulting in a marked reduction in activity of the protein, but RFC1 function was not altered (23). All cell lines were grown in RPMI 1640 medium containing 5% bovine calf serum (Hy-Clone), 2 mM glutamine, 20 M 2-mercaptoethanol, penicillin (100 units/ ml), and streptomycin (100 g/ml) at 37°C in a humidified atmosphere of 5% CO 2 . For assay of thiamin accumulation, cells were grown in thiamin-free RPMI medium (custom-made by Life Technologies, Inc.) containing 5% dialyzed bovine calf serum (Life Technologies, Inc.), 2 mM glutamine, 20 M 2-mercaptoethanol, penicillin (100 units/ml), and streptomycin (100 g/ml) supplemented with 30 nM tritiated thiamin. For culture of R16 and T2 cells, G418 at a concentration of 750 g/ml was included in the medium to ensure stable, high level expression of RFC1.
Transport Studies-Cells were harvested, washed twice with HBS (20 mM HEPES, 140 mM NaCl, 5 mM KCl, 2 mM MgCl 2 , 5 mM glucose, pH 7.4), and resuspended in HBS to 1.5 ϫ 10 7 cells/ml. For some experiments, glucose-free HBS was also used. Cell suspensions were incubated at 37°C for 20 min following which uptake was initiated by the addition of [ 3 H]thiamin, [ 3 H]TPP, or [ 3 H]MTX, and samples were taken at the indicated times. Uptake was terminated by injection of 1 ml of the cell suspension into 10 ml of ice-cold HBS. Cells were collected by centrifugation, washed twice with ice-cold HBS, dried, and digested with 1 N NaOH in an 8-ml vial. After fluor was added, radioactivity was assessed in a liquid scintillation spectrometer. For efflux experiments, cells loaded with 0.2 M [ 3 H]thiamin for 1 h, were harvested by centrifugation, washed twice with ice-cold HBS, and resuspended into 9 ml of pre-warmed thiamin-free HBS. Portions were taken, and intracellular tritium was determined as described as above.
HPLC Analysis-Cells (ϳ 8 ϫ 10 6 ) incubated with [ 3 H]thiamin were washed three times with 0°C HBS. One-fourth of the cell pellet was processed for dry weight and total tritium as described in transport studies. The remaining portion was processed according to a slightly modified published protocol (24). Cell pellets were suspended in 250 l of HBS, and 15 l of trichloroacetic acid (100% w/v) was added to precipitate proteins. After centrifugation the supernatant was extracted with 0.5 ml of water-saturated ether for five times and neutralized with 1 N NaOH. After the residual ether was removed in a speedvac, the extract was spiked with unlabeled thiamin, TMP, and TPP and separated on a reversed-phase HPLC column (Waters Spherisorb, 5 m ODS2 4.6 ϫ 250 mm) as described previously (25). Separation of the thiamin, TMP, and TPP was achieved with a linear gradient of from 0 to 60% of acetonitrile in 70 mM phosphate, pH 7.4, over 30 min followed by a 10-min elution with 70 mM phosphate, pH 7.4, at a flow rate of 1 ml/min. Under these conditions, elution times of thiamin, TMP, and TPP were 20, 15.6, and 14 min, respectively. Fractions (0.5 ml) were collected in 8-ml scintillation vials, and radioactivity was assessed as indicated above. The levels of thiamin and its metabolites were normalized to units of nmol/g dry weight of cells.
Accumulation of Thiamin and TPP in Cells under Growth Conditions-Cells (3 ϫ 10 6 ) grown in complete RPMI 1640 were washed twice with thiamin-free RPMI and resuspended into the same medium supplemented with 30 nM [ 3 H]thiamin. After 1 week of exponential growth, cells were harvested, washed twice with ice-cold HBS, and processed for intracellular tritium as described for transport studies.

The Characteristics of Net Thiamin Uptake and Efflux in
Murine Leukemia Cells That Overexpress, or Lack Functional, RFC1-Thiamin uptake was assessed in several murine leukemia cell lines with different levels of RFC1 function. MTX r A is a subline of murine leukemia L1210 cells that lack RFC1 activity due to an alanine to proline substitution at amino acid 130 (21). R16 cells, derived by transfection of RFC1 cDNA into MTX r A cells (21), express about 10 times more RFC1 than wild-type L1210 cells. As indicated in the inset of Fig. 1, MTX uptake in MTX r A cells was negligible over 30 min, whereas MTX uptake in R16 cells was very rapid and reached steady state within 10 min (22). The pattern of uptake of thiamin was reversed in these cell lines. Net uptake in MTX r A cells was roughly three times greater than that of R16 cells by 1 h; in neither cell line was a steady state reached over the interval of observation (upper panel of Fig. 1). HPLC analysis indicated that at 1 h 90 Ϯ 3 and 75 Ϯ 6% (n ϭ 2) of intracellular tritium was the active thiamin metabolite, TPP in R16 and MTX r A cells, respectively; in the latter the remainder of intracellular tritium was thiamin.
The lower panel of Fig. 1 illustrates the decline in cell tritium when MTX r A and R16 cells were loaded with 0.2 M [ 3 H]thiamin for 1 h prior to resuspension into thiamin-free buffer. The decrease of intracellular tritium can be characterized by a single exponential in both cell lines, and the slope of the lines extrapolate through the time 0 points that represent the initial level of intracellular tritium. The rate constant for TPP efflux from R16 cells was 4-fold greater than from MTX r A cells, 0.58 versus 0.14 h Ϫ1 , respectively. However, in both cell lines the rate of loss of TPP was only a small fraction of the rate of thiamin influx (see below). Initial Uptake of Thiamin-Initial thiamin uptake was assessed in L1210 and R16 cells over 20 s at an extracellular concentration of 0.2 M. As indicated in the upper panel of Fig.  2, the initial uptake rate for thiamin in R16 cells was ϳ28% higher than in L1210 cells, much less than the 9-fold difference in RFC1 expression and MTX influx between these cell lines. Moreover, addition of 25 M MTX, which would reduce RFC1mediated influx by at least 70% (based upon the RFC1-mediated MTX influx K t of 7 M (17) and a thiamin influx K i of 2.8 mM, see below), had no effect at all on thiamin influx in either cell line. Hence, RFC1 does not contribute to thiamin influx. The very low affinity of RFC1 for thiamin was confirmed by evaluating the inhibitory effect of this vitamin on MTX influx in L1210 cells. As shown in the lower panel of Fig. 2, there was a gradual increasing inhibition of MTX influx, albeit to a small degree, as the thiamin concentration was increased from 0.1 to 1 mM. At 1 mM thiamin, MTX influx was reduced by only ϳ28%. Based upon the MTX influx K t of 7 M, the K i for thiamin was calculated to be 2.8 mM.
TPP Influx in Murine Leukemia Cells-Since thiamin is rapidly phosphorylated to TPP in these cells, one explanation for low net thiamin uptake in cells with high level RFC1 expression (R16) is that TPP generated in the cell is a substrate for, and is exported by, RFC1. To explore this possibility, the effect of TPP on MTX influx was assessed along with the transport properties of TPP (Fig. 3). TPP inhibited MTX influx with a K i of 32 Ϯ 5 M (n ϭ 3) in L1210 cells (upper panel), a value only ϳ4-fold higher than the MTX influx K t of 7 M, consistent with a previous report (26). Furthermore, TPP influx was directly related to the level of RFC1 activity. Influx in the MTX r A cells was one-fourth that of wild-type L1210 cells, and influx was 7-fold greater in R16 cells than L1210 cells. Furthermore, 25 M MTX markedly decreased TPP influx in R16 and L1210 cells, whereas TPP initial uptake in MTX r A cells was the same in the presence or absence of 25 M MTX (lower panel). Hence, RFC1-mediated influx of TPP is equal to, or larger than, transport mediated by other process(es).
The Effect of Intracellular MTX on Net Thiamin Uptake-If TPP efflux is indeed mediated by RFC1, loading cells to high levels of MTX should competitively inhibit efflux of TPP and augment net TTP cellular accumulation. This was found to be the case. Incubation of R16 and MTX r A cells with 1 mM MTX resulted in intracellular MTX levels of 350 and 220 nmol/g dry wt (100 and 63 M, respectively, based upon a ratio of intracellular water to dry weight of 3.5 l/mg), since MTX enters cells via passive diffusion and possibly routes other than RFC1 at this high concentration. As shown in Fig. 4, intracellular accumulation of thiamin and its metabolites was doubled in R16 cells, to a level comparable to that of MTX r A cells, by pre-and co-incubation with 1 mM MTX. HPLC analysis confirmed that this increase in net uptake was due entirely to an increase in TPP accumulation. On the other hand, there was no effect of 1 mM MTX on thiamin uptake in MTX r A cells, all consistent with the lack of RFC1 function in this cell line.
Energy Dependence of Net Thiamin Uptake-When L1210, MTX r A, and R16 cells are incubated with 10 mM azide in the absence of glucose, intracellular ATP is depleted, an effect at least partially reversed by addition of 5 mM glucose to the transport buffer (5). As indicated in Fig. 5 (upper panel), net thiamin uptake in MTX r A cells was higher than that in L1210 cells without energy depletion. However, thiamin uptake in ATP-depleted cells was decreased to the same level in both L1210 and MTX r A cells. Net thiamin uptake in R16 cells was markedly lower than in L1210 and MTX r A cells regardless of the energy status. This energy dependence of thiamin accumu-

FIG. 2. The effect of MTX on thiamin influx in L1210 and R16 cells (upper panel) and the effect of thiamin on MTX influx in L1210 cells (bottom panel).
Upper panel, L1210 and R16 cells were incubated in HBS at 37°C for 20 min and then exposed to 0.  lation was further explored in L1210 cells by determination of thiamin metabolites by HPLC (lower panel of Fig. 5). Under all conditions, TMP was present at levels that were barely detectable as compared with thiamin and TPP. In energy-depleted cells, the level of thiamin exceeded that for TPP regardless of the time of exposure to tritiated thiamin (10 or 60 min), but some TPP nonetheless was present. There were slightly higher levels of both thiamin and TPP at 1 h of exposure than that at 10 min reflecting continued, albeit slow, thiamin uptake and phosphorylation. In contrast, TPP was the dominant species in energy-replete cells. The increase in TPP accumulation from 10 min to 1 h contributed almost entirely to the increase in net thiamin uptake in L1210 cells. The lower level of thiamin present in energy-replete, versus energy-deplete, cells may be due to the rapid rate of phosphorylation relative to the rate of thiamin entry into cells.

FIG. 3. Dixon plot of TPP inhibition of MTX influx (upper panel) and MTX inhibition of TPP influx in MTX r A, L1210, and R16 cells (bottom panel). Upper panel, after L1210 cells were incubated in HBS at
The Impact of RFC1 Function on Thiamin Accumulation at a Physiological Thiamin Concentration-In most media, including RPMI 1640, in which murine leukemia cells are grown, the thiamin concentration (ϳ3 M) is about 2 orders of magnitude higher than the physiological blood level (10 -30 nM). Hence, in vitro growth conditions are highly nonphysiological with respect to this vitamin. To determine the effect of RFC1 function on thiamin accumulation under more physiological conditions, cells were grown in 30 nM [ 3 H]thiamin for 1 week. In addition to wild-type L1210, MTX r A, and R16 cells, another eight L1210 variants were studied in these experiments. These included the following: (i) L1210-T2 cells obtained by transfection of RFC1 cDNA into wild-type L1210 cells to achieve RFC1 expression at a level about 7-fold higher than in L1210 cells (22); (ii) L7, L15, L44, and L51 cell lines with functional RFC1 but with markedly reduced folylpolyglutamate synthetase activity due to mutations in this enzyme (23); (iii) the D10 line with no expression of RFC1 due to a Gly to Ala substitution in the initiation codon (19); and (iv) G1 and G2 cells with point mutations resulted in S46N and V104M substitutions, respectively, that markedly impair RFC1 activity (17,18). As indicated in Table I, thiamin accumulation generally correlated with levels of RFC1 activity. Thiamin accumulation in R16 and T2 cells was 64 and 42%, respectively, that of L1210 cells, whereas thiamin accumulation in L7, L15, L44, and L51 cells, in which RFC1 function is near normal, was comparable to that of L1210 cells. However, thiamin accumulation in G1a, G2, MTX r A, and D10 cells with markedly impaired RFC1 function was increased by factors of 1.8, 2.4, 2.8, and 2.5, respectively. On average, the level of thiamin and its metabolites in cells that overexpress RFC1 (R16 and T2) was 4 -7-fold greater than in cells in which RFC1 function is absent (MTX r A and D10).

DISCUSSION
Thiamin transport in murine leukemia cells is likely mediated by a facilitative carrier. Accumulation of its active coenzyme form is attributed to rapid phosphorylation of thiamin to TPP which is, to a large extent, retained within cells. This is a common phenomenon for many different substrates of the major facilitator superfamily, as occurs with phosphorylation of nucleosides (27,28) and polyglutamation of folates (29). Consistent with this was the observation that in energy-depleted cells, phosphorylation was impaired and net thiamin uptake was markedly decreased (Fig. 5). Interestingly, the steadystate thiamin level in energy-depleted L1210 cells was ϳ0.3 M (1 nmol/g dry weight), comparable to the extracellular level (0.2 M), consistent with an equilibrating process.
Folate and thiamin, both B family vitamins, differ not only in chemical structure but also in charge. At physiological pH, folate is a bivalent anion, whereas thiamin bears one positive charge. The similarity between RFC1 and the thiamin transporter, especially in the predicted transmembrane domains, raised the possibility that these carriers may share common substrates. This was not the case. Instead, we have demonstrated that RFC1 transports the major thiamin metabolite, TPP, with an influx K i only ϳ4 times greater than the MTX  influx K t . This has important consequences with respect to the level of TTP accumulation. After thiamin enters cells it is phosphorylated to TPP and its molecular charge changes from positive to negative. This, in turn, is associated with an increased affinity for RFC1, an anion exchanger, that mediates TPP efflux, leading to a decrease in net intracellular TPP accumulation. Accordingly, TPP accumulation in these cells is enhanced by the rate of entry of thiamin mediated by the thiamin transporter and the rate of phosphorylation catalyzed by thiamin pyrophosphokinase. TPP accumulation is countered by the efflux of TPP mediated by RFC1. Hence, the net level is determined by balance of these processes. Since efflux studies show that the major portion of intracellular TPP is retained within the cells, there must be a large component that is bound to TPP-dependent apoenzymes in cytosol and mitochondria (30). In addition, TPP may be hydrolyzed, in part, to TMP which is, in turn, exported. TMP also appears to be a good substrate for RFC1 since TMP inhibits RFC1-mediated MTX influx with a K i of 15 M. 2 Increased thiamin accumulation associated with decreased RFC1 activity observed when cells were grown at a physiological thiamin concentration (ϳ30 nM) suggests that the ability of RFC1 to export TPP may have important biological consequences. The requirement for thiamin is ubiquitous, but thiamin-responsive megaloblastic anemia syndrome and dietary thiamin deficiency cause tissue-specific and nonoverlapping defects (9 -11). Metabolically active tissues are vulnerable to thiamin deficiency due to heavy usage of TPP-dependent enzymes. The observation that RFC1 mediates efflux of TPP may provide another important dimension to the understanding of thiamin metabolism. Tissues with high expression of RFC1 may export more TPP, making them more susceptible to metabolic derangement when thiamin is scarce. Hence, RFC1 expression may modulate the tolerance to thiamin deficiency associated with either mutations in SLC19A2, dietary deficiency, or malabsorption. It remains to be established if RFC1 is expressed in thiamin-sensitive cell types and whether metabolic defects associated with thiamin deficiency might be modified by the level of RFC1 expression.
Although TPP is not present in plasma, but accumulates in erythrocytes, TMP is present in plasma at concentrations only slightly lower than that of thiamin (31). As indicated above, TMP is likely a good substrate for RFC1, and transport by this route could be of importance under conditions in which the thiamin transporter is defective. Hence, at high blood levels of thiamin and TMP, substantial delivery of TMP via RFC1 might obviate the consequences of the loss of the thiamin transporter, another potential role for RFC1 as a modifying element in this clinical situation. RFC1-mediated TPP transport could also play a role in thiamin absorption in intestine. Dietary vitamin B 1 exists predominantly as thiamin pyrophosphate that is hydrolyzed to thiamin by a phosphatase in the intestinal lumen before absorption. RFC1 is expressed in intestine and is proposed to mediate folate absorption (32,33). It is possible that some TPP may be directly delivered into mucosal cells by RFC1 before hydrolysis to thiamin is achieved. Furthermore, TMP is derived from dephosphorylation of TPP and/or transphosphorylation of thiamin by a membrane-associated alkaline phosphatase in intestinal mucosa (34,35). TMP has been shown to cross everted rat jejunal sac wall unchanged and enter the serosal fluid (36). It is possible that this process is RFC1-dependent.