The antifolate LY309887 is a specific glycinamide ribonucleotide formyltransferase inhibitor that blocks de novo purine synthesis and produces a depletion of purine nucleotides. The activity of LY309887 in six human tumor cell lines has been examined by growth inhibition and clonogenic assay after continuous exposure for three cell doubling times and by ATP depletion at 24 h. Three cell lines (CCRF-CEM, MCF7, and GC3) were sensitive to LY309887-induced growth inhibition (IC50: 5.6–8.1 nm), whereas the other cell lines (COR-L23, T-47D, and A549) were comparatively resistant (IC50: 36–55 nm). Sensitivity to LY309887 cytotoxicity was consistent with sensitivity to growth inhibition in four of five cell lines tested (MCF7/GC3: 0.01% survival and COR-L23/T-47D: 1–5% survival at 100 nm LY309887). LY309887-induced ATP depletion was measured by luciferase-based ATP assay and confirmed by high performance liquid chromatography measurements. There was a linear relationship between ATP depletion and growth inhibition when data were analyzed for all six cell lines (r2 = 0.93; P < 0.0001). Depletion of 24-h cellular ATP concentrations to <1 mm was associated with both cell growth inhibition and cytotoxicity in all cell lines studied. In conclusion, cellular ATP depletion induced by LY309887 can be used to predict growth inhibition and cytotoxicity in human tumor cells.
The folate-requiring enzymes that catalyze the de novo synthesis of nucleotides for DNA replication have been studied as targets for cancer therapy for many years. GARFT3 (EC 22.214.171.124) is the first folate-dependent enzyme in the purine nucleotide synthetic pathway and uses N10-formyl-tetrahydrofolic acid as a cosubstrate. Inhibition of GARFT results in depletion of cellular purine nucleotide pools, which are necessary for cellular energy requiring processes and for the synthesis of DNA and RNA. Lometrexol (6R-5,10-dideaza-5,6,7,8-tetrahydrofolic acid, 6R-DDATHF) is a specific GARFT inhibitor, which has potent antitumor activity against a number of murine tumors and human tumor xenografts (1) . Consistent with the proposed mechanism of action of lometrexol, the ATP and GTP content of L1210 cells was decreased by at least 70% following treatment with lometrexol at concentrations causing growth inhibition (10–100 nm; Ref. 2 ).
LY309887 (6R-2′,5′-thienyl-5,10-dideazatetrahydrofolic acid) is a thiophene analogue of lometrexol, and it is a representative second generation GARFT inhibitor, which is a more potent cytotoxic agent in vitro with greater antitumor activity than lometrexol in vivo (1 , 3) . Measurement of nucleotide pools following LY309887 treatment demonstrated a marked decrease in (d)ATP and (d)GTP levels within 6 h in CCRF-CEM cells (4) . Furthermore, the growth inhibitory activity of both lometrexol and LY309887 was reversed by HPX, but not by thymidine, reflecting the selective inhibition of de novo purine synthesis by both drugs (1 , 2) . During the clinical evaluation of lometrexol and LY309887, antitumor activity was reported in patients with malignant histiocytoma, breast, non-small cell lung, ovarian, and head and neck cancer (5, 6, 7, 8, 9) . In initial clinical trials, lometrexol caused cumulative antiproliferative toxicities in the form of myelosuppression and gastrointestinal damage. Subsequently, it was shown that coadministration of folic acid ameliorates the cumulative antiproliferative toxicities of lometrexol, thereby allowing substantial dose escalation (6) . All clinical studies with LY309887, therefore, have included folic acid supplementation to reduce the cumulative aspects of the toxicity profile.
Using structure-based drug design based on the X-ray structure of GARFT, the antipurine antifolate AG2034 (4-[2-(2-amino-4-oxo-4,6,7,8,-tetrahydro-[3H]pyrimidino[5,4–6][1,4]-thiazin-6-yl)-(S)-ethyl]-2,5-thienoyl-l-glutamic acid) has recently been developed. AG2034 is also a thiophene analogue of lometrexol and has potent in vivo antitumor activity against a variety of murine and human tumor xenografts (10) . As classical antifolates, all of the GARFT inhibitors developed to date are subject to the same biochemical determinants of activity as natural folates, notably membrane transport and intracellular polyglutamation.
The aim of this study was to identify a potential predictor of LY309887 activity in human tumor cell lines with the ultimate objective of using such a predictor in clinical studies. Broadly, there are two groups of the potential cellular determinants of antifolate activity: upstream factors that influence the degree and duration of nucleotide depletion following exposure to a given extracellular drug concentration; and downstream factors that influence the response of the tumor cell to a given level of nucleotide depletion. In the case of LY309887, upstream factors include drug transport and polyglutamation, intracellular folate pools, the degree and duration of GARFT inhibition, and the relative contributions of the de novo and salvage pathways to purine biosynthesis in the target cell. Downstream factors that may influence the response of a cell to a given level of ATP depletion include p53 genotype and functional p53 status or apoptotic propensity, as reflected by, for example, bcl2/bax expression. The relative impact of “upstream” and “downstream” factors on sensitivity to LY309887 can therefore be distinguished by evaluating the relationship between drug exposure and ATP depletion and growth inhibition/cytotoxicity.
In the present study, two classes of tumor cells were identified with regard to sensitivity to LY309887-induced growth inhibition and cytotoxicity. However, the degree of ATP depletion induced by LY309887 predicted growth inhibition, suggesting that upstream factors are primarily responsible for differences between cell lines in their sensitivity to LY309887. Hence, ATP depletion following LY309887 treatment could be used to predict sensitivity to the drug both in vitro and in vivo.
MATERIALS AND METHODS
Reagents and Chemicals.
All routine chemicals and the bioluminescent somatic cell assay kit were obtained from Sigma (Poole, United Kingdom). The BCA protein assay kit was purchased from Pierce (Rockford, USA). LY309887 was a gift from Lilly Research Laboratories, Indianapolis, IN. Raltitrexed was purchased from Zeneca Ltd, Cheshire, United Kingdom, and both LY309887 and raltitrexed were dissolved in water before use.[ 3H]2O was obtained from Amersham (Little Chalfont, United Kingdom).
A549 (human lung carcinoma) cells were obtained from the National Cancer Institute (NIH, Bethesda, MD). MCF7 and T-47D (human breast carcinoma) and CCRF-CEM (human lymphoblastic leukemia) cells were purchased from the American Type Culture Collection (Rockville, MD). COR-L23 (human lung carcinoma) cells were a gift from Dr. Peter Twentyman (Medical Research Council Clinical Oncology and Radiotherapeutics Unit, Cambridge, United Kingdom). GC3 (human colon carcinoma) cells were kindly provided by Lilly Research Laboratories, Indianapolis, IN. All cell lines were adapted for growth in RPMI 1640 medium (Life Technologies, Paisley, United Kingdom) supplemented with 10% (v/v) dialyzed fetal bovine serum at 37°C in 5% CO2 and tested monthly to exclude Mycoplasma infection (11) . The cell population doubling times for the cell lines (determined by daily cell number estimation) were 20 h (GC3), 24 h (A549), 30 h (COR-L23), 38 h (CCRF-CEM), 48 h (MCF7), and 60 h (T-47D).
Growth Inhibition Assay.
Growth inhibition assays were carried out as described previously (12) . Briefly, adherent (all lines except CCRF-CEM) exponentially growing cells were seeded into a 96-well plate at 2–5 × 103 cells/100 μl/well. After 20–24 h at 37°C, the medium was replaced with fresh medium containing either LY309887 or raltitrexed at the appropriate drug concentration. The drug incubation period was varied depending on the growth rate of cell lines to ensure that control cells had undergone three-cell population doublings. After drug treatment, the cells were fixed with Carnoy’s fixative (methanol:acetic acid 3:1, v:v), washed, air dried, and stained with sulforhodamine B as described previously (13) . The absorbance per well was measured at 570 nm on a Dynatech MR 7000 Plate Reader (Billingshurst, United Kingdom).
CCRF-CEM suspension cell cultures were seeded in 24-well plates at 1 × 105 cells/500 μl/well. After 20–24 h of incubation, an equal volume of fresh medium containing LY309887 at two times the final concentration was added. The incubation was continued for a period equivalent to three-cell doubling times (114 h), and the total cell number in each well was counted after fixation with Carnoy’s fixative on a model Z1 Coulter Counter (Coulter Electronics, Luton, United Kingdom).
The cytotoxicity of LY309887 was determined in the five adherent cell lines. Exponential growing cells were seeded into 100-mm Petri dishes at densities ranging from 150 to 4 × 104 cells/dish, and the cell seeding density was adjusted to give an estimated 20–300 colonies/dish following drug exposure. The cells were left to attach for 24 h, and LY309887 was added at the appropriate concentrations, as indicated in the “Results” section, to the dishes. Three dishes were used for each drug concentration, and at least two experiments were carried out under each set of conditions. The cells were exposed to LY309887 for three-cell doubling times, after which the medium was aspirated, the dishes were washed once with warm PBS, and fresh drug-free medium containing 10% (v/v) dialyzed fetal bovine serum was added. In some experiments, 30 μm HPX was added during the drug exposure and/or in the drug-free posttreatment cloning medium. The cells were incubated for an additional 10–16 days until visible colonies appeared, which were fixed with Carnoy’s fixative and visualized by staining cells in 0.4% (w/v) crystal violet. Colonies with >30 cells were counted. Cloning efficiencies were: A549 75%, GC3 37%, T-47D 30%, MCF7 25%, and COR-L23 17%. Cell survival following drug exposure was expressed as the percentage of control cloning efficiency or survival.
Cells were seeded into 24-well plates at 2–5 × 104 cells/ml/well for adherent cells and 1 × 105 cells/0.5 ml/well for CCRF-CEM. After 20–24 h incubation, the medium was replaced with fresh medium containing LY309887 or raltitrexed at the appropriate concentration, as indicated in the “Results” section (adherent cells), or an equal volume of fresh medium added with two times the final concentration of drug (CCRF-CEM cells). After 24 h of drug exposure, the drug-containing medium was removed, and the cells were washed once with cold PBS. Intracellular ATP levels were determined (14) by a bioluminescent somatic cell assay kit according to the manufacturer’s instruction. Briefly, cells in 0.1 ml of PBS/well were mixed with 0.1 ml of Somatic Cell ATP Releasing Reagent on ice for 20 s. Half of the cell extract (0.1 ml) was transferred to a vial containing 0.1 ml of ATP assay mix and mixed for 10 s. The amount of light emitted from each reaction was measured immediately by a luminometer (TD-20e, Turner-Designs, Sunnyvale, CA). The amount of ATP per well was calculated from ATP standard curves, prepared under the same conditions, and analyzed as part of each experiment. The cellular protein levels were determined using the remaining cell extract with the Pierce BCA protein assay. Data were then expressed as moles of ATP per milligram of protein or as the percent control. Ribonucleotide triphosphate pools were measured by HPLC as described previously (4) . To calculate absolute intracellular ATP concentrations in the six human cell lines, cell volumes were estimated by determination of intracellular water following equilibration with[ 3H]2O in buffer as described previously (12) .
Analyses of the relationships between the growth inhibition at three-cell doubling times and ATP depletion at 24 h in six human cell lines and between luciferase-based ATP assay and HPLC measurements were performed using linear regression analysis (GraphPad PRISM, Intuitive Software for Science, San Diego, CA).
Growth Inhibition Studies.
Previous experiments to determine the effects of LY309887 on human tumor cells have concentrated on the human lymphoblastic leukemia cell line CCRF-CEM (4 , 15) . In the experiments described here, five adherent solid tumor cell lines (A549, GC3, MCF7, T-47D, and COR-L23) and one suspension cell line CCRF-CEM were treated with LY309887 for three-cell doubling times. Growth inhibition curves for A549 and MCF7 are shown in Fig. 1⇓ , and a summary of the IC50 values, defined as the concentration of LY309887 required to inhibit growth by 50%, for all six cell lines is given in Table 1⇓ . Two classes of cell lines with differential sensitivity to LY309887 were found. Three of the cell lines (CCRF-CEM, MCF7, and GC3) were sensitive to LY309887-induced growth inhibition (IC50: 5.6–8.1 nm), whereas the other cell lines (COR-L23, T-47D, and A549) were relatively resistant (IC50: 36–55 nm). Furthermore, there was a difference in the behavior of these two classes of cell lines following treatment with LY309887. The sensitive adherent cell lines, i.e., MCF7 and GC3 cells, became detached from the plastic dishes when incubated with high concentrations of LY309887 (100 and 1000 nm), whereas the less sensitive cell lines (COR-L23, T-47D, and A549) remained attached to the plastic at the end of the drug exposure period.
To examine whether the differences in sensitivity observed in the growth inhibition assay reflected differences in sensitivity to LY309887-induced cytotoxicity, clonogenic assays were undertaken. Clonogenic assays, which examine the ability of a single cell to form a colony after drug treatment, have been used previously to measure antifolate-induced cell death (16) . The cytotoxicity induced by LY309887 was examined in the five adherent cell lines. As shown in Fig. 2⇓ , a >99.99% cell kill was observed in two of the sensitive cells (MCF7 and GC3) at 100 nm LY309887, indicating that <0.01% cells were viable and could form colonies after continuous drug exposure for three-cell doubling times. In contrast, only a 1–5% cell kill was found in the less sensitive cells, T-47D and COR-L23, following exposure to 100 nm LY309887. Survival of A549 cells, which was the most resistant cell line in growth inhibition assays, was only 0.02% after exposure to 100 nm LY309887, indicating a discrepancy in the growth inhibition and cytotoxicity data for this cell line. The rate of appearance, size, and morphology of colonies varied depending on the cell line. For example, colonies of A549 cells were very firm round and easy to visualize by crystal violet staining after incubation in drug-free medium for 10 days, whereas COR-L23 colonies had a loose irregular shape, and the cells often had swollen cytoplasmic volumes.
The cytotoxic effect of LY309887 was also studied in the presence of the rescue agent HPX (Fig. 3)⇓ in A549 cells. The absence of added HPX exposure to LY309887 resulted in up to 99.99% cell kill. LY309887-induced cytotoxicity in A549 was completely abolished if 30 μm HPX was present throughout the drug treatment and subsequent incubation periods. If the addition of HPX was delayed until after drug treatment, colony formation was reduced by ∼70%, indicating that even after drug treatment for three-cell doubling times, some 30% of the cells were still viable and able to form colonies when the rescue agent was supplied.
ATP Depletion Studies.
LY309887 is a specific GARFT inhibitor that blocks de novo purine synthesis. To estimate the extent of GARFT inhibition by LY309887, intracellular ATP levels were measured by a luciferase-based assay after cells were incubated with LY309887 at the indicated concentrations for 24 h. The depletion of ATP levels following 24 h-drug treatment showed a very similar concentration dependency to LY309887-induced growth inhibition after three-cell doubling times in all six human tumor cell lines (Fig. 4)⇓ . There was a strong linear relationship between LY309887-induced ATP depletion and growth inhibition in these cell lines (r2 = 0.93; P < 0.0001). To exclude the possibility that LY309887-induced ATP depletion was a secondary consequence of growth inhibition, ATP levels and growth inhibition were studied in two cell lines following treatment with the specific TS inhibitor raltitrexed using the same experimental protocol as used with LY309887. Raltitrexed induced a concentration-dependent growth inhibition in A549 and GC3 cells after continuous exposure for three-cell doubling times. However, no raltitrexed-induced ATP depletion was observed at 24 h in either of the two cell lines (Fig. 5)⇓ . Absolute intracellular ATP concentrations (mm) in the human tumor cell lines were determined using the data from the luciferase-based ATP assay and measurements of cell volumes, performed as described previously (12) . This analysis demonstrated that pretreatment ATP levels were in the range of 2–5 mm (Table 1)⇓ and that the depletion of 24-h cellular ATP concentrations to <1 mm was associated with both cell growth inhibition (>50%) and cytotoxicity (<10% survival) after LY309887 treatment in all of the cell lines examined.
HPLC measurements were used to confirm the data derived from the luciferase-based assay in three of the cell lines, and there was a strong correlation between the data obtained from the two assays in the cells examined (A549, GC3, and COR-L23; r2 = 0.85; P < 0.0001). In addition to ATP depletion after exposure to 100 nm LY309887, GTP levels were also reduced in A549 (Fig. 6)⇓ , GC3, and COR-L23 cell lines (data not shown). However, levels of pyrimidine triphosphates (CTP and UTP) were maintained at pretreatment levels.
The aim of this study was to identify potential determinants and predictors of LY309887 activity in human tumor cell lines. By evaluating the relationship between drug exposure, ATP depletion, and growth inhibition/cytotoxicity, the relative impact of pretarget“ upstream” or posttarget “downstream” factors was investigated. The present studies demonstrated that the GARFT inhibitor LY309887 caused intracellular ATP depletion within 24 h in all of the six cell lines examined. Moreover, the extent of ATP depletion was related to the degree of growth inhibition at three-cell doubling times (Fig. 4)⇓ . In other words, to achieve the same degree of ATP depletion and growth inhibition, higher concentrations of LY309887 were required in resistant cell lines (A549, T-47D, and COR-L23) than in the relatively sensitive cell lines (GC3, CCRF-CEM, and MCF7). Therefore, on the basis of these data, pretarget upstream factors appear to be the primary reason for differences between cell lines in their sensitive to LY309887-induced growth inhibition, and for a given degree of ATP depletion, there is consistent growth inhibition. Such upstream factors could include cell membrane folate transport proteins, folylpolyglutamate synthetase activity, intracellular folate stores, and GARFT levels. Tse and Moran (17) studied the mechanism of lometrexol resistance in an L1210 cell line and showed that an increase in folic acid transport in the resistant cell line resulted in an expanded cellular content of folates that blocked lometrexol polyglutamation. Furthermore, it is of interest to note that GC3 cells were 10-fold more sensitive to raltitrexed than A549 cells (Fig. 5)⇓ , which is consistent with the 7-fold difference in sensitivity to LY309887 (Table 1)⇓ . Considering the very different enzyme targets of these two agents, common pretarget upstream factors, such as those indicated above, are likely to be primary determinants of the activity of these two agents in A549 and GC3 cell lines.
After exposure of COR-L23 and T-47D cells to 100 nm LY309887 for three-cell doubling times, 1–5% of the total cell population had the ability to form colonies (95–99% cell kill). The maximum cytotoxicity level of LY309887 in these two cell lines is therefore similar to that of lometrexol, a first generation GARFT inhibitor, in human WiDr colonic carcinoma cells (16) . These latter authors found that 40 μm lometrexol was required to produce 99% cell kill following a 72-h exposure and that the cytotoxicity of lometrexol was substantially lower than that of the TS inhibitor raltitrexed (99.9% cell kill after a 24-h exposure to 0.04 μm raltitrexed). As described here, substantially greater sensitivity to LY309887 cytotoxicity was found in three other cell lines (A549, MCF7, and GC3; 99.9–99.99% cell kill at 0.04–0.1μ m). Therefore, in four of five cell lines, sensitivity to LY309887 cytotoxicity was consistent with sensitivity in growth inhibition assays (MCF7/GC3-sensitive in both assays and COR-L23/T-47D-relative insensitive in both assays). A549 cells were more sensitive in the cytotoxicity assay than predicted by the growth inhibition data (IC50 = 55 nm, yet only 0.02% survival after exposure to 100 nm LY309887). This discrepancy between the growth inhibition and cytotoxicity data in A549 cells is presently unknown.
The nucleotide pool changes and their effects in cells treated with GARFT inhibitors are different from those produced by TS inhibitors. In TS inhibitor-treated cells, depletion of dTTP following inhibition of TS results in cytotoxicity rather than cytostasis, whereas in GARFT inhibitor-treated cells, depletion of ATP/GTP pools results in cytostasis (15 , 16) . However, the studies reported here suggest that LY309887 exposure can result in cytotoxicity and that there is a spectrum of sensitivity to LY309887-induced cell killing, which in most cell lines (in 4/5 adherent lines), reflected the degree of ATP depletion.
A number of studies have suggested that intracellular ATP levels determine whether a cell dies by apoptosis or necrosis (18, 19, 20) because apoptosis is a highly regulated and energy-requiring process. Using human cell lines, Smets et al. (21) showed that an intracellular ADP/ATP ratio of 0.2 was the critical threshold (ratios of >0.2 resulting in glucocorticoid-induced cell death by apoptosis). Furthermore, Lieberthal et al. (22) reported recently that in cultured mouse proximal tubular cells, the ATP concentration was a determinant of the mode of cell death following treatment with either antimycin or 2-deoxyglucose, with varying concentrations of dextrose. These authors found that, when ATP concentration was below∼ 15% of control, cell death was uniformly by necrosis. In contrast, when the ATP concentration was between ∼25% and 70% of control, the cell death mechanism was apoptosis. The results presented here show that LY309887 caused ATP depletion to decrease below 20% of the control in five of six cell lines following a 24-h exposure to LY309887 (Fig. 4)⇓ . If difficulty was encountered in achieving cytotoxic levels of ATP depletion with LY309887 in clinical trials, the drug could be administrated in combination with a second agent that depletes ATP, such as 2-deoxyglucose, which inhibits the glycolytic generation of ATP. Alternatively, conventional anticancer agents could be used in combination with LY309887 to induce apoptosis, which had been shown to lead to ATP depletion in its own right (23) . Thus, three alternative means of inducing ATP depletion to below the point of cell viability could be combined with the aim of maximizing tumor cell killing.
The response of tumor cell lines to cytotoxic drug treatment can depend on p53 expression, the presence of a functional wild-type p53 gene being associated with increased drug sensitivity in comparison to tumor cell lines harboring mutations in the p53 gene (24) . For four of the cell lines used in the present study, sensitivity to LY309887 did not appear to be related to the published p53 status (24) . Thus, two of the cell lines that were relatively resistant in the growth inhibition assay, A549 and T-47D, had wild-type and mutant p53 status, respectively, whereas two of the sensitive cell lines, MCF7 and CCRF-CEM, also had wild-type and mutant p53 status, respectively. Investigations of the p53 status of all of the cell lines used in this study are presently underway.4
In conclusion, the results of this study suggest that ATP depletion can predict LY309887-induced growth inhibition in human tumor cell lines. Following confirmation of these results in preclinical in vivo studies, tumor ATP levels could be measured by noninvasive 31P nuclear magnetic resonance spectroscopy after treatment with GARFT inhibitors in clinical trials. ATP depletion could be used as an indicator of the extent of GARFT inhibition in individual patients (with the potential for dose modification) and possibly also as a predictor of the likelihood of tumor response.
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.
↵1 Supported by the Eli Lilly Co. and the Cancer Research Campaign (United Kingdom).
↵2 To whom requests for reprints should be addressed, at the Cancer Research Unit, Medical School, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, United Kingdom. Phone: 44-191-222-5712; Fax: 44-191-222-7556; E-mail:
↵3 The abbreviations used are: GARFT, glycinamide ribonucleotide formyltransferase; HPX, hypoxanthine; HPLC, high performance liquid chromatography; TS, thymidylate synthase.
↵4 Xiaochong Lu, Julie Errington, John Lunec, Nicola J. Curtin, Alan V. Boddy, and David R. Newell. The impact of p53 status on cellular sensitivity to antifolate drugs, manuscript in preparation.
- Received August 2, 1999.
- Revision received October 12, 1999.
- Accepted October 13, 1999.