Whereas the major folate transporter, the reduced folate carrier (RFC), has a physiological pH optimum, transport activities for folates and antifolates have been detected with low pH optima. Because the interstitial pH in solid tumors is generally acidic, the mechanisms by which antifolates are transported at low pH could be an important determinant of drug activity under these conditions. The current study quantitated the low pH methotrexate (MTX) transport activity in human solid tumor cell lines from the National Cancer Institute tumor panel and other sources. MTX influx at pH 5.5 was equal to, or greater than, influx at pH 7.4 in 29 of 32 cell lines. To assess the role of RFC in transport at low pH in one of these cell lines, a HeLa clonal line (R5) was selected for MTX resistance due to a genomic deletion of the carrier gene. MTX influx was depressed by 70% in R5 versus wild-type HeLa cells at pH 7.4. At pH 6.5, influx in these two lines was similar; as the pH was decreased to 5.5 influx increased in both cell lines. Similarly, whereas net MTX uptake over 1 h was markedly decreased in R5 cells at pH 7.4, net uptake in HeLa and R5 cells was comparable at pH 6.5. Also, as compared with MCF7 breast cancer cells, MTX uptake was markedly decreased at pH 7.4, but only minimally at pH 6.5, in the MDA-MB-231 human breast cancer cell line that lacks RFC expression. When grown with folic acid (2 μm) at pH 7.4, the IC50 for MTX was 14-fold higher in R5 as compared with wild-type HeLa cells; the difference was only 4-fold when cells when grown at pH 6.9; the IC50s were identical at this pH when the medium folate was 25 nm 5-formyltetrahydrofolate. These data demonstrate that transport activity at low pH is prevalent in human solid tumors, is RFC-independent in R5 cells and MDA-MB-231 breast cancer cells, and can preserve MTX activity in the absence of RFC at an acidic pH relevant to solid tumors in vivo.
Transport of folates and antifolates in mammalian cells is mediated by a variety of processes (1) . The most important route of transport is the reduced folate carrier (RFC) characterized in detail in murine and human leukemia cell lines. In these cell systems, RFC operates optimally at physiological pH (2) . However, there are folate transport activities that function optimally at low pH in mammalian cells in which RFC activity is either present or impaired. For instance, in intestinal epithelial cells in vitro, which express high levels of RFC, and no folate receptor, the pH optimum for transport of folates is acidic with only low levels of activity present at physiological pH (3 , 4) . Transport of folates in intestinal loops and everted sacs likewise has a low pH optimum (5) . There have been other reports of folate transport at acidic pH in normal and malignant cell lines, exceeding the rate of transport at physiological pH (6, 7, 8, 9, 10, 11, 12, 13) . In addition to a low pH optimum, there are other properties of transport at low pH that are different from what is observed for RFC-mediated transport at physiological pH. The low pH transport route has an affinity for folic acid that is comparable with that of methotrexate (MTX), unlike RFC that at physiological pH has an affinity for folic acid 2 orders of magnitude lower than that of MTX, or folate receptors that have an affinity for folic acid orders of magnitude higher than for MTX (1) . Furthermore, RFC-mediated influx is energy-independent, whereas influx at low pH is energy-requiring (2) . These observations suggested that the low-pH transport route is RFC-independent. On the other hand, there is evidence to suggest that transport at low pH is related to RFC in intestinal cells. Hence, transfection of RFC into rat intestinal epithelial cells results in augmented transport activity at low (14) as well as physiological pH with their characteristic properties (4) . Furthermore, influx of 5-methyltetrahydrofolate in mouse intestinal epithelia cells is inhibited by antibodies directed against the NH2 and COOH termini of murine RFC (3) , although these domains are predicted to be localized in the cytosol (15, 16, 17) .
The nature of antifolate transport activity at acidic pH has important implications for the application of these drugs in the treatment of solid tumors, because the interstitial or extracellular pH in solid tumors is generally acidic. Measurements by electrode, 31P magnetic resonance spectroscopy, or fluorescence ratio image microscopy indicate pH levels much lower than that of blood and 0.2–0.6 units lower than the intracellular pH often reaching a level of ∼6.5, depending on tumor type and size, and distance from capillary vessels (18, 19, 20) . Although low-pH folate transport activity has been recognized for many years, along with the low pH in the interstitium of solid tumors, there is little information regarding the prevalence, magnitude, and pharmacological significance of this transport activity in human solid tumor cells.
The current study was undertaken to determine the extent to which low pH transport activity is present in a broad spectrum of human solid tumor cell lines from the National Cancer Institute tumor panel and other sources and to compare this with the level of transport at physiological pH. A second objective of the study was to assess the consequence of the loss of RFC expression on MTX transport activities to assess the extent to which the low pH transport route is linked to the presence of RFC. Finally, studies were designed to assess the impact of the low pH transport activity on tumor cell sensitivity to MTX under conditions in which RFC activity is completely lost.
MATERIALS AND METHODS
[3′, 5′, 7-3H]-(6S)-5-formyltetrahydrofolate (5-CHO-THF) was obtained from Moravek Biochemicals (Brea, CA); [3′, 5′, 7-3H] MTX and [3′, 5′, 7, 9-3H] folic acid were purchased from Amersham Corp. (Arlington Heights, IL). Unlabeled 5-CHO-THF, MTX, and folic acid were obtained from Sigma. All of the radiochemicals and their nonlabeled counterparts were purified by high-performance liquid chromatography before use. The stability of radiochemicals once purified was checked by high-performance liquid chromatography on a regular basis and materials repurified as necessary.
Cell Lines and Culture Conditions.
All of the cell lines used in these studies were obtained from the National Cancer Institute Developmental Therapeutics Program except for the HeLa and hepatoma lines. HeLa cells were purchased from American Type Culture Collection. H35 and H35R hepatoma cells (21) were kindly provided by Dr. John Galivan (New York State Department of Health, Albany, NY), and the remaining hepatoma cell lines were obtained from the Liver Center of Albert Einstein College of Medicine. Unless specified, all of the cell lines along with the R7 transfectant (see below) were grown in RPMI 1640 (pH 7.3) supplemented with 10% fetal bovine serum (Gemini Bio-Products), 2 mm glutamine, 20 μm 2-mercaptoethanol, penicillin (100 units/ml), and streptomycin (100 μg/ml) at 37°C in a humidified atmosphere of 5% CO2. In initial studies, HeLa and R5 cells (RFC-null, see below) were grown in DMEM with the same supplementation. For some studies, cells were grown in custom-made folate- and NaHCO3-free RPMI 1640 (Special Media, Phillipsburg, NJ) supplemented with either 25 nm 5-CHO-THF or 2 μm folic acid to which NaCl and NaHCO3 were added to achieve a specified pH and osmolarity. RPMI 1640 was set to pH 6.9 by adjusting the NaHCO3 content to 7.2 mm and increasing the NaCl concentration by 17 mm. Cell cultures were monitored regularly with a mycoplasma detection kit (ATCC) and were shown to be free of this microorganism.
Cell Growth Inhibition Studies.
For growth inhibition studies conducted in pH 7.3 medium, cells were transferred to 96-well plates (500 cells/well) and exposed continuously to a spectrum of antifolate concentrations for 6 days. For studies in pH 6.9 medium, 1000 cells/well were used. Cell growth rates were quantified by sulforhodamine B staining (22) .
Selection of R5 Cells.
R5 cells were generated by a single step exposure of HeLa cells to 200 or 600 nm MTX in DMEM after treatment with the chemical mutagen ethylmethanesulfonate, a methodology used previously in this laboratory for the isolation of antifolate-resistant L1210 leukemia cells (23) . Eighteen MTX-resistant clonal HeLa cell lines were initially selected. Eight clones were screened for expression of RFC mRNA, and one of these lines, R5 (selected with 600 nm MTX), that completely lacked RFC message was used for additional studies. After it was recognized that the IC50 for antifolates was much greater in DMEM than RPMI 1640 likely due to the difference in the folic acid concentrations (12 and 2.0 μm, respectively), all of the cells were grown in RPMI 1640. The growth rate of HeLa cells in these media was similar.
Transfections of RFC cDNA.
R5 cells were transfected with human RFC cDNA cloned into pcDNA 3.1 by Lipofactamine Plus reagent (Invitrogen). Two days later cells were replated and exposed to G418 at a concentration of 600 μg/ml. Surviving clones were expanded and MTX influx determined. One clone, RFC-7, with markedly increased MTX influx was chosen for further study. RFC-7 cells were maintained in RPMI 1640 supplemented with 600 μg/ml of G418.
Total RNA was isolated from HeLa, R5, and RFC-7 cells using the TRIzol reagent (Invitrogen). Total RNA (30 μg) was resolved by electrophoresis on 1% agarose gels containing formaldehyde. RNA was transferred to Nytran N-membranes (Schleicher & Schuell, Keene, NH) and fixed with a Stratalinker UV-cross linker (Stratagene, La Jolla, CA). The blot was first probed with human RFC cDNA then stripped and reprobed with β-actin cDNA.
Southern Blot Analysis.
High molecular weight genomic DNA was extracted from HeLa and R5 cells with the DNeasy Tissue kit (Qiagen, Valencia, CA). DNA (10 μg) was digested with HindIII and PstI (24) and fractionated on 0.6% agarose gels, transferred to Hybond-N membranes (Amersham, Piscataway, NJ), and fixed with a Stratalinker UV-cross-linker (Stratagene, La Jolla, CA). The blot was hybridized with the whole open reading frame of human RFC cDNA.
The following buffers were used for transport studies, HEPES-buffered saline (20 mm HEPES, 140 mm NaCl, 5 mm KCl, 2 mm MgCl2, and 5 mm glucose) adjusted to pH 8.0, 7.4, or 7.0, and 4-morpholinepropanesulfonic acid-buffered saline; and MBS (20 mm 4-morpholinepropanesulfonic acid, 140 mm NaCl, 5 mm KCl, 2 mm MgCl2, and 5 mm glucose) adjusted to pH 6.5, 6.0, 5.5, or 5.0. Transport studies with adherent cells followed a protocol designed for rapid uptake determinations on cells growing in monolayer cultures (25) . Briefly, cells (3–4 × 105) were seeded in 20-ml Low Background glass vials (Research Products International Corp., Prospect, IL) and grown for 3 days to reach early confluence. Cells were washed twice with transport buffer and incubated in this buffer at 37°C for 20 min. The buffer was then removed and uptake initiated by the addition of 0.5 ml of the transport buffer containing [3H]MTX. Uptake was terminated by injection of 5 ml of ice-cold HEPES buffered saline (HBS) after which the vials were washed three times with ice-cold HBS. The adherent cells were dissolved in 0.2 m NaOH (0.5 ml) by incubation at 65°C for 30 min. Lysate (0.4 ml) was transferred to scintillation vials, fluor added, and radioactivity determined; 10 μl was processed for protein determination (BCA, Pierce, Rockford, IL). Cellular uptake is expressed in units of pmol/mg protein.
For determination of tightly bound cellular levels of MTX, HeLa and R5 cells were loaded with 0.5 μm [3H]MTX at pH 6.5 for 10 min then washed three times with HBS-pH 7.4 buffer and incubated in 5 ml of drug-free HBS-pH 7.4 for 60 min. The cells were washed twice with ice-cold HBS-pH 7.4, and tightly bound MTX was determined as described above.
Intracellular Folate Pool Measurement.
Cells grown in the regular RPMI 1640 at pH 7.3 were trypsinized and plated in folate-free RPMI 1640 at pH 7.3 or 6.9 containing either 2 μm [3H]folic acid or 25 nm [3H]5-CHO-THF. The cultures were maintained in these media for 7–9 days and intracellular tritium determined as described for the transport studies.
HeLa and R5 cells (5 × 105 cells/well) were grown in six-well plates in four different RPMI 1640 media: pH 7.3 with 2 μm folic acid; pH 7.3 with 25 nm 5-CHO-THF; pH 6.9 with 2 μm folic acid; and pH 6.9 with 25 nm 5-CHO-THF. [3H]MTX at a concentration of 50 nm, along with 200 μm glycine, 100 μm adenosine, and 10 μm thymidine, were added to the growth media and incubation continued for 48 h. The cells were then washed and intracellular tritium determined as described for transport studies.
MTX Transport Activity in Solid Tumor Cell Lines at Acidic pH.
To assess the prevalence of folate/antifolate transport activity at acidic pH, 32 human solid tumor cell lines were chosen for study, representing a variety of tumor types, most of which were obtained from the National Cancer Institute in vitro drug-screen panel. Human CCRF-CEM leukemia cells were used as a reference point. MTX (0.5 μm) was the reference antifolate because of its high affinity for RFC and low affinity for folate receptors. Folic acid (20 nm) was always present in the assay buffer to suppress binding to, and transport mediated by, folate receptors. A pH of 5.5 was chosen to assess low-pH activity to additionally minimize an RFC-mediated transport component (2) .
Relative MTX influx at pH 5.5 and 7.4 varied among the different tumor cell lines. High activity at pH 5.5 (≥1 pmol/mg protein/2 min) was found in HCT-H15, HeLa, Hep3B, HUH7, and NCI-H460 cell lines (Table 1)⇓ . MTX influx in CCRF-CEM cells at pH 7.4 was three times greater than at pH 5.5 consistent with the low level of transport activity in this and murine leukemia cell lines at acidic pH (11 , 26) . In 9 cell lines (28%), the activity at acidic pH was more than twice that at pH 7.4. In all of the tumor cell lines except 3 (SF-295, HepG2, and PC-3), MTX influx at pH 5.5 was close to, or higher than, MTX influx at pH 7.4. There was no pattern of activities that favored a specific tumor tissue. Two cell lines lacked RFC activity; RFC is not expressed in MDA-MB-231 breast cancer cells due to methylation of the RFC gene promoter (27) , and RFC function is low or absent in H35R cells that were selected for MTX resistance (21) . Hence, in the overwhelming majority of solid tumor cell lines MTX transport activity measured at pH 5.5 was not only present but was comparable with, or greater than, the transport activity at neutral pH.
Selection of an RFC-Null Human Solid Tumor Cell Line.
Folate transport activity at acidic pH has been attributed to RFC-independent and dependent pathways. For example, the low pH activity identified in L1210 leukemia cell lines was thought to be RFC-independent because carrier function was impaired (2 , 28) , whereas a RFC-dependent low-pH activity was expressed in rat intestinal IEC-6 cells transfected with carrier (4 , 14) . Because of the prevalence of a low pH activity in the solid tumor cell lines, studies were undertaken in one of these cell lines in an attempt to determine the extent to which this activity is linked to RFC.
HeLa cells were chosen for study because they exhibited high folate transport activity at low pH and are readily transfected. The approach was to identify a cell line, under MTX selective pressure augmented by chemical mutagenesis, in which RFC expression was lost. Eight MTX-resistant clones were obtained and screened, and four displayed decreased RFC mRNA expression; the R5 clone was chosen for further study because message was completely lost (Fig. 1A)⇓ due to the deletion of the RFC gene as indicated by Southern analysis (Fig. 1B)⇓ .
Influx of 0.5 μm [3H]MTX in the R5 cell line was decreased by ∼70% as compared with wild-type HeLa cells (Fig. 2A)⇓ . This degree of residual transport activity was higher than observed in murine leukemia cells that have lost RFC function (23) , and its basis is unclear. Residual activity was not due to a folate receptor-mediated influx component, because the presence of 50 nm folic acid in the transport buffer would abolish such a component. Accompanying the loss of transport activity, MTX growth inhibition was markedly decreased in R5 cells with a 14-fold increase in the IC50 in RPMI 1640 containing 2 μm folic acid as compared with wild-type cells (Fig. 2B)⇓ .
To exclude the possibility that there were other important changes beyond the loss of MTX transport that contributed to MTX-resistance, transport activity was restored by transfection of RFC cDNA into R5 cells. As indicated in Fig. 3A⇓ , RFC mRNA was highly overexpressed in the R5 RFC transfectant, designated as RFC-7, as compared with HeLa and R5 cells. Consistent with the increase in RFC expression, MTX influx was 4.5-fold greater in the transfectant than in HeLa cells and >12 times greater than in R5 cells (Fig. 3B)⇓ . The MTX IC50 in the RFC-7 transfectant was decreased by a factor of ∼50 to a level one-third that of HeLa cells (Fig. 3C)⇓ . Hence, the loss of RFC expression and the associated transport defect can account entirely for MTX resistance in R5 cells.
A Comparison of the pH Profile for MTX Influx in RFC-Deficient R5- and Wild-Type HeLa Cells.
The pH dependence of RFC-independent MTX influx in R5 cells was assessed and compared with that of RFC-competent wild-type cells. As indicated in Fig. 4A⇓ , the ratio of MTX influx in wild-type HeLa versus R5 cells was substantial at high pH but was reduced to one as the pH was decreased to 6.5–6.0. When the pH was decreased below 6.5 influx increased in both HeLa and R5 cells, although the increase was greater in R5 cells. At pH 5.0, MTX influx in R5 cells exceeded that in HeLa cells by 28% (P < 0.05). The interrupted line indicates the RFC-mediated component in wild-type cells when the residual RFC-independent component measured in R5 cells was subtracted from the total flux. This analysis indicates that RFC-mediated influx is essentially absent below a pH of 6.5.
Folate receptor-mediated transport over this pH spectrum was negligible because, as indicated in Fig. 4B⇓ , 20 nm folic acid had no effect on influx of 0.5 μm MTX at pH 5.5. In contrast, 20 μm folic acid essentially eliminated MTX influx, consistent with a saturable process with a folic acid affinity comparable with that of MTX, as observed for low pH route(s) in other cell lines (2 , 4 , 11 , 26) .
The pH-Dependence of Net MTX Transport in R5 Versus Wild-Type HeLa Cells.
As observed for the influx process (Fig. 4A)⇓ , net MTX uptake in R5 cells was far below that in HeLa cells at pH 7.4 but this difference decreased as the pH was reduced so that at pH 6.5 net cellular levels were the same over an interval of 1 h (Fig. 5)⇓ . At the latter pH, net uptake in HeLa cells was ∼15% higher than at pH 7.4
The free intracellular MTX level was quantitated in HeLa and R5 cells. Cells were loaded with [3H]MTX at pH 6.5 after which the cells were washed, a large volume of MTX-free buffer added, and nonexchangeable cellular MTX level was measured. On the basis of three separate measurements the nonexchangeable MTX level was 1.36 ± 0.04 and 1.13 ± 0.03 (mean ± SE) pmol/mg protein in HeLa and R5 cells, respectively. The free MTX level, which is the difference between net uptake and the nonexchangeable level, was negligible in R5 cells after 1 h of incubation with MTX at pH 7.4. At pH 6.5, the free MTX level in R5 cells at 1 h was more than three times greater than that at pH 7.0. Hence, free intracellular MTX in R5 cells was markedly increased as extracellular pH was decreased from pH 7.4 to 6.5.
Net MTX Uptake in the RFC-Deficient MDA-MD-231 Breast Cancer Cell Line.
The data in HeLa cell lines indicate that loss of RFC function diminished MTX influx and net transport at pH 7.4 but not at pH 6.5. To assess whether this is the case for another human solid tumor cell line, MTX transport was measured in the RFC-null, MTX-resistant MDA-MD-231 breast cancer cell line along with MTX-sensitive MCF-7 cells, which expresses high levels of RFC. Carrier expression is absent in MDA-MB-231 cells due to methylation of the RFC gene promoter (27) . As indicated in Fig. 6⇓ , MTX uptake in MDA-MB-231 cells was much lower than in MCF-7 cells at pH 7.4, consistent with the lack of RFC expression. At pH 6.5, however, MTX uptake in MDA-MB-231 cells was increased by a factor of four and achieved a level 85% that of MCF-7 cells. Hence, as observed in HeLa cells, MTX uptake at low pH in MDA-MB-231 breast cancer cells appears to be mediated, at least in part, by an RFC-independent pathway.
MTX Growth Inhibition in R5 and HeLa Cells at Low pH.
Studies indicated that the emergence of an RFC-independent transport activity at low pH could compensate for the loss of RFC function observed at physiological pH. To determine the impact of the low pH folate transport activity on MTX growth inhibition, an RPMI 1640 medium was used in which the concentration of sodium bicarbonate was decreased from 24 to 7.2 mm. As indicated in Fig. 7A⇓ , the pH of this medium was 6.9 under an atmosphere of 5% carbon dioxide; the pH of the regular RPMI 1640 was 7.3 under the same conditions. The pH in both media was stable for 7 days at 37°C. When cells were present the pH gradually decreased as the cell density increased but the decrease was more precipitous and a greater magnitude in the regular RPMI 1640. By day 5 cells reached full confluence in the regular medium and the pH had stabilized at ∼6.7 but the pH in the low-pH RPMI 1640 continued to drop reaching a value of 6.5 at day 7 at a time when the cells were ∼95% confluent.
MTX growth inhibition was determined in HeLa and R5 cells in the low pH RPMI 1640. The doubling time of both cell lines was comparable in this medium albeit at a slightly lower growth rate than in the regular RPMI 1640. MDA-MB-231 cells did not grow in the low pH medium. Under these conditions the difference in MTX IC50 between R5 and wild-type cells had decreased to a factor of only 4 (Fig. 7B)⇓ , in contrast to the factor of 14 noted in regular RPMI 1640 (Fig. 2B)⇓ .
To evaluate the impact of the folate growth source on MTX growth inhibition at low pH, cells were grown in RPMI pH 6.9 medium containing 25 nm 5-CHO-THF instead of 2.0 μm folic acid. As indicated in Fig. 7C⇓ , the MTX IC50 in R5 and HeLa cells was identical under these conditions. Hence, the complete loss of RFC expression had no impact on the growth inhibitory effect of MTX at low pH when the growth source was 5-CHO-THF.
These observations were then correlated with the levels of intracellular folates under these growth conditions. Two approaches were used to assess folate pools, one direct, and the other indirect. Total folate pools were measured after cells were grown for >1 week in folate-free RPMI 1640 supplemented with either 2 μm [3H]folic acid or 25 nm [3H]5-CHO-THF. As indicated in Table 2⇓ , the total folate pool was increased in R5 cells by 19 and 28% in the pH 7.3 and pH 6.9 media, respectively, as compared with HeLa cells, when cells were grown with 2 μm [3H] folic acid. When cells were grown with 25 nm [3H]5-CHO-THF there was a contraction of folate cofactor pools by a factor of ∼5 in wild-type HeLa cells and ∼10 in R5 cells. However, whereas the total folate cofactor pool in R5 cells was 0.64 that of wild-type HeLa cells at pH 7.3, the levels were comparable in medium at pH 6.9. The second approach was based on the availability of folates for interconversion to dihydrofolate as reflected by the extent of growth inhibition by trimetrexate. As indicated in Fig. 8A⇓ , R5 cells grown in folic acid at pH 7.3 were modestly (2-fold) collaterally sensitive to TMQ. When R5 cells were grown in 5-CHO-THF at the same pH, there was a marked difference (25-fold) in IC50 between R5 and wild-type HeLa cells (Fig. 8B)⇓ due to both an increase in sensitivity of the former (from 18 to 4 nm) and a decrease in sensitivity for the latter (from 35 to 100 nm).
Total MTX accumulation was also assessed after R5 and HeLa cells were grown for 2 days under these conditions. As indicated in Table 3⇓ , whereas MTX accumulation in R5 cells was about half that in wild-type HeLa cells in pH 7.3 RPMI 1640 with folic acid as the folate source, it increased to ∼75% the level in wild-type HeLa cells in pH 6.9 RPMI 1640. Likewise, when cells were grown in 5-CHO-THF, MTX accumulation in R5 cells was 60% that of wild-type HeLa cells, but at low pH the level in R5 cells was increased to ∼82% that of HeLa cells. This pH change had little effect (10–15%) on MTX accumulation in wild-type HeLa cells. Hence, the increase in MTX accumulation in the low pH medium reflected, at least in part, the largely preserved MTX influx and net transport under these conditions (Figs. 4⇓ and 5)⇓ .
These data demonstrate the prevalence and magnitude of MTX transport at low pH in a broad spectrum of human solid tumors, thus confirming and expanding the scope of normal and malignant cells in which this activity has been observed. Because the interstitial pH in solid tumors is generally acidic, this low pH transport activity could play an important role in the internalization of antifolates in the clinical setting. This applies not only to MTX, which is used in the treatment of some solid tumors but primarily in leukemia, but also to new generation antifolates, in particular pemetrexed that is active in the treatment of mesothelioma, non-small cell lung cancer, and other solid tumors (29 , 30) . This transport activity also has very clear physiological relevance in terms of the absorption of physiological folates because there is an acid pH in the microclimate of the small intestinal villi (31) . What is unclear is the mechanistic basis for this low pH activity, the extent to which it is related to RFC, and the number of distinct processes that it might represent.
In addition to small intestine, low-pH folate/antifolate transport has been reported in rat hepatocytes (6 , 7) , rat kidney brush border membranes (8) , rat astrocytes and cerebellar granule cells (9 , 10) , Fisher rat 3T3 fibroblasts (11) , retinal pigment epithelia cells ARPE-19 (12) , and Chinese hamster ovary cells (13) . There is evidence to suggest that the low pH transport activity in intestine is related, at least in part, to RFC. The protein is expressed at the apical surface of intestinal cells (32) , influx in intestinal cells is inhibited by an antibody to RFC (3) , and when mouse RFC was transfected into rat intestinal epithelial cells transport activities at both high and low pH were augmented (4 , 14) . These data suggested that the low-pH transport activity might be related to tissue-specific post-translational modifications of RFC. It is also possible that RFC might interact with another protein that alters RFC transport properties or that this interaction might result in the activation of another distinct protein that is a folate transporter. On the other hand there is evidence that low pH transport activity can also be RFC independent. For instance, a low level of this activity was identified in L1210 murine leukemia cells in which RFC function was impaired due to an alanine to proline substitution at amino acid 130 in the carrier protein (2 , 28) . However, it is also possible that this RFC mutation might alter function at high pH without resulting in a loss of activity at low pH. A low pH activity has also been demonstrated in other cell lines in which RFC function was impaired, although the molecular basis for this defect was not defined (26) .
The results in the present article demonstrate unequivocally that there is low pH transport activity in the R5/HeLa cell line, and likely in the MDA-MB-231 breast cancer cell line, which is RFC-independent, because RFC is not expressed in these two cell lines; and in the case of R5 cells the gene is deleted. However, the data do not rule out the possibility that a component of the low-pH transport in wild-type HeLa cells is contributed by RFC, because it is possible that up-regulation of the RFC-independent low-pH transport activity (Fig. 4A)⇓ may have occurred to compensate for the loss of RFC in R5 cells. The observation that the ratios of the low:high pH transport activities varied among the 32 cell lines evaluated suggests that the expression of the underlying process(es) varies among the different solid tumors.
These studies also demonstrate the pharmacological importance of the low pH transport activity in moderating or eliminating the decline of MTX activity when RFC is lost. The data indicate that the low pH transport activity can fully compensate for the loss of MTX growth inhibition that occurs when RFC is deleted when the folate growth source is 5-CHO-THF. The latter closely corresponds to 5-methyltetrahydrofolate, the physiological folate in the blood of humans and rodents, in terms of their corresponding reduced pteridine ring, metabolic potential, and affinities for RFC. Hence, the data suggest that resistance phenomena for antifolates as they relate to solid tumors may be different from what is observed under cell culture conditions at neutral pH. Of course, because solid tumors vary in size and proximity to capillary blood vessels, not all will have a low interstitial pH, and RFC would remain the major antifolate transporter under physiological conditions. Also relevant to this consideration is the fact that folate receptor-mediated transport has a low pH optimum (33) , and folate receptors could play an important role in antifolate delivery and activity in treatment of solid tumors (34 , 35) . However, in the current study a folate receptor-mediated transport component was minimized by the presence of folic acid.
The low pH folate transport activity is relevant not only as a mechanism of antifolate uptake but as an uptake route for natural folates as well, when cells are at low pH. Hence, the pH in regular RPMI 1640 declined to 6.6 as cells reached confluence, similar to what has been observed for Chinese hamster ovary or L1210 leukemia cell cultures (13 , 26) . Under these conditions, the low pH transport activity would be expected to contribute to folate uptake. This may account for the ability of RFC-null cells, such as the MTXrA line, in which there is negligible residual transport, to grow when 5-CHO-THF is present at very low concentrations (36) . However, the capacity to grow at low pH may be cell-specific, because whereas HeLa cells grow well with a starting pH of 6.9, MDA-MB-231 cells do not grow under these conditions.
It is well established that as the level of folate cofactor pools in cells is decreased, the sensitivity to MTX, and to a varying extent many other antifolates, increases (37, 38, 39) . This is due to a number of factors: (a) enhanced polyglutamation and drug accumulation because there are lower levels of folates available to compete with drug for folypolyglutamate synthetase; and (b) decreased folate substrate to compete at the target enzyme. However, alterations in the total cellular folate pool in HeLa and R5 cells did not always correlate with MTX activity under the different growth conditions in contrast to what has been observed in murine leukemia cells (36 , 39) . The decrease in MTX IC50 that occurred at low pH in the presence of folic acid was associated with an increase in the total folate pool, although the additional decrease in IC50 in the presence of 5-CHO-THF at low pH was associated with a marked drop in the total folate pool. In medium containing 5-CHO-THF the folate pool in R5 cells was half that of HeLa cells comparable with what has been observed previously (36 , 39) and expected because RFC is the major route of transport of this folate into cells. However, in medium containing folic acid, the total folate level in R5 cells was greater than in HeLa cells; the reason for this difference is unclear. At low pH total cell folate was greater in R5 than wild-type HeLa cells despite the increase in sensitivity to MTX. TMQ is a very sensitive indicator of cell folates due to the interconversion of THF cofactors to dihydrofolate, with competition between dihydrofolate and TMQ at the level of dihydrofolate reductase, as the enzyme is inhibited by this agent (37 , 40) . The observed 2-fold collateral sensitivity to TMQ in R5 versus wild-type HeLa cells grown in folic acid and 25-fold collateral sensitivity to this agent in 5-CHO-THF is consistent with increasing contraction of folate cofactors under these conditions. These observations suggest that the folate pools available to support folate-dependent reactions, as measured by equilibrating these cells with tritiated folate substrates, may not be accurately reflected in the total folate cofactor level in these cells. In HeLa cells there may be differences in the distribution of specific folate cofactors, binding or compartmentation of folates, or differences in polyglutamyl chain lengths that may have influenced these results (41 , 42) .
Grant support: Grant CA-82621 from the National Cancer Institute.
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 U.S.C. Section 1734 solely to indicate this fact.
Note: Dr. Gao is currently at the Department of Immunology, Memorial Sloan-Kettering Cancer Center, New York, NY 10021.
Requests for reprints: I. David Goldman, Albert Einstein College of Medicine Cancer Center, Chanin 2, 1300 Morris Park Avenue, Bronx, New York 10461. Phone: (718) 430-2302; Fax: (718) 430-8550; E-mail:
- Received July 21, 2003.
- Revision received September 25, 2003.
- Accepted October 5, 2003.