Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

During the past several decades a variety of folic acid analogues have been developed that inhibit the activity of folate requiring enzymes (Jackson, 1984; Schultz, 1995; Hanauske, 1996). Many of these compounds are cytotoxic or cytostatic and are capable of causing tumor regression. Traditionally, antifolates inhibit enzymes that are directly involved in the biosynthesis of purine and pyrimidine nucleotides. For example, methotrexate, an inhibitor of DHFR, limits the production of purine nucleotides as well as thymidylate by preventing the regeneration of tetrahydrofolate (Jackson, 1984). This in turn decreases the supplies of 10-formyl tetrahydrofolate and 5,10-methylenetetrahydrofolate, metabolites that are needed for purine and thymidine biosynthesis, respectively. Other antifolates are inhibitors of enzymes such as TS or GARFT, thus directly affecting pools of nucleotides available for DNA synthesis (Beardsley et al., 1989; Habeck et al., 1994; Jackman and Calvert, 1995; Takemura and Jackman, 1997).

Cytosolic C1-THF synthase is a trifunctional enzyme responsible for maintaining equilibrium among THF, 10-formylTHF, 5,10-methenylTHF and 5,10-methyleneTHF (Tan et al., 1977; Villar et al., 1985; MacKenzie et al., 1988). The activities associated with this enzyme include an NADP-dependent dehydrogenase, a cyclohydrolase and a synthetase. The dehydrogenase activity catalyzes the interconversion of 5,10-methyleneTHF and 5,10-methenylTHF. 5,10-methyleneTHF is a substrate for TS, providing the methyl group for the conversion of dUMP to dTMP. In addition, 5,10-methenylTHF can be converted to 10-formylTHF by the cyclohydrolase activity of C1-synthase. 10-formylTHF is a substrate for GARFT in the purine de novo biosynthetic pathway and can be regenerated by the synthetase activity of C1-synthase. Because 5,10-methyleneTHF and 5,10-methenylTHF are both substrates for the dehydrogenase, the dehydrogenase activity is central to the maintenance of folate cofactors for thymidine and purine synthesis.

Despite the importance of C1-THF synthase, and in particular the dehydrogenase activity, there is very little published data that describes the biochemical and cellular effects of inhibiting this enzyme (Temple et al., 1982; Green et al., 1988). Furthermore, the central role of this enzyme in the maintenance of folate-pool equilibria, makes it difficult to speculate on which nucleoside biosynthetic pathways would be affected, and in what manner, by an inhibitor. In one study, a series of 5- and 10-substituted, 5,10-disubstituted and 5,10-bridged substituted derivatives of THF were examined for inhibition of several folate requiring enzymes (Temple et al., 1982). One analogue, LY354899, was a potent inhibitor of the dehydrogenase activity isolated from porcine liver (IC₅₀ = 100 nM), but was less active against the synthetase and cyclohydrolase activities. However, no significant cytotoxicity was observed against L1210 murine leukemia cells (Temple et al., 1982).

In this study, we confirm that LY354899 is a potent inhibitor of C1-synthase dehydrogenase, but is only weakly antiproliferative toward CCRF-CEM lymphocytic leukemia cells. Despite its poor antiproliferative activity, we observed that LY354899 induced very interesting cell cycle alterations. Although most antifolates can alter the cell cycle, the effects of LY354899 were unique relative to those of GARFT, TS or DHFR inhibitors. The cell cycle profiles indicated that inhibition of C1-synthase dehydrogenase resulted in two temporally distinct events. The first resembled an antipurine effect and was similar to cell cycle effects of GARFT inhibition. The second was similar to a thymineless state induced by TS and DHFR inhibition and ended in apoptosis. Analysis of nucleotide pools confirmed that inhibition of C1-synthase dehydrogenase resulted in a unique biochemical fingerprint. We tested the hypothesis that coadministration of the purine salvage metabolite hypoxanthine, would alleviate the initial purineless state and would potentiate the thymineless apoptotic response. This treatment resulted in increased antiproliferation and in an earlier and more dramatic apoptotic response than was observed with LY354899 alone. The results of this study provide evidence that enzymes involved in the maintenance of folate cofactors represent potentially viable targets for new antifolate chemotherapies.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. The enzymes used in this study were obtained from the following sources: rhTS from Dr. D. V. Santi (University of California at San Francisco, San Francisco, CA); rhDHFR from Anatrace Inc. (Maumee, OH); trifunctional murine GARFT from Dr. R. G. Moran (Medical College of Virginia, Richmond, VA). A truncated version of C1-synthase (Mr 35,000) which contained only the dehydrogenase and cyclohydrolase activities was obtained from Dr. R. E. Mackenzie (McGill University, Montreal, Quebec, Canada).

Synthesis of LY354899. This compound (fig. 1) was synthesized by a modification of the previously published procedure (Temple et al., 1982). A suspension of (6-R,S)-5,6,7,8-tetrahydrofolate (1.50 g, 3.37 mmol), synthesized by the method of Martinelli and Chaykovski (1980), in 60 ml of deionized water under nitrogen was treated with a toluene solution of phosgene [5.0 ml of a 1.93 M solution (Fluka), 9.65 mmol, 2.9 eq]. After stirring vigorously for 5 hr, the aqueous phase was separated, and its pH adjusted to 2.8 by the careful addition of aqueous sodium hydroxide solution. The resulting suspension was stirred at ice bath temperatures for thirty minutes and collected on a filter. Drying under vacuum yielded 1.45 g (90%) of the crude product as a tan solid. This solid was analyzed by HPLC on a Beckman Ultrasphere IP, 80A Pore, 4.6 mm × 25 cm, 1.5 ml/min flow rate, UV detection at 278 nm; elution with solvent A for 5 min followed by a linear gradient to 9% solvent B over 20 min, a 5 min hold at 9% B followed by a linear gradient to 100% B over 5 min; solvent A was 0.1 M ammonium acetate plus 1% acetonitrile adjusted to pH = 5.5 by the addition of glacial acetic acid and solvent B was acetonitrile; the major product eluted at 9.85 min with 67% peak area; at least three impurities were noted (12.43 min, 10 area%; 15.57 min, 6 area%; 18.85 min, 6.75 area%). Portions of the crude product were purified on a Vydac Protein and Peptide C18 column (5 cm × 20 cm) using a modified analytical solvent system; the pooled product fractions were evaporated, dissolved in distilled water and lyophilized; the resulting solid was brought into solution by adjusting the pH of an aqueous suspension to 7.0 with aqueous sodium hydroxide solution and the product reprecipitated by adjusting the pH to 2.8 with 1.0 N aqueous hydrochloric acid solution. Separation of the product diastereomers was not observed during analytical or preparative HPLC. The product was collected on a filter and air dried. Typically 500 mg of crude product yielded 120 mg of (6-R,S)-5,6,7,8-tetrahydro-N⁵,N¹⁰-carbonylfolic acid that was >96 area% pure by analytical HPLC analysis; 1H-NMR (300 MHz, D₂O + DCl) d 7.57 (d, 2H, 8.8 Hz, Ar---H), 7.40 (d, 2H, 8.8 Hz, Ar---H), 4.38-4.33 (dd, 1H, Glu C---H), 4.01 (dd, 1H, J1==J2==7.8 Hz, C---H), 3.57 (d, 1H, 10.1 Hz, C---H), 3.51-3.47 (m, 2H, C---H), 2.99 (dd, 1H, J1==J2==11.2 Hz), 2.32 (t, 2H, 7.0 Hz, Glu CH₂), 2.11-2.04 (m, 1H, Glu CH), 1.94-1.89 (m, 1H, Glu CH); FAB(+)MS = 472 (M + 1); Anal. (C₂₀H₂₁N₇O₇-1.5 H₂O-1.0 HCl) theory: C, 44.91; H, 4.71; N, 18.33; found: C, 45.26; H, 4.50; N, 17.80.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   The structure of LY354899.

Enzymatic assays. TS activity was assayed spectrophotometrically as previously described (Shih et al., 1997). Briefly, deoxyuridylate monophosphate, 6[R]-5,10-methylene-5,6,7,8-tetrahydrofolate and rhTS were incubated together and the formation of 7,8-DHF was monitored at 340 nm. The K_i for LY354899 was determined by measuring the activity of rhTS in the presence of several concentrations of inhibitor using the equation K_iapp = K_i (1 + [S]/K_m) where [S] is the concentration of 6R-MTHF and K_m = 3 µM. Competitive inhibition was assumed. The data were fit to the Morrison equation by nonlinear regression analysis (Morrison, 1969).

Cell culture. CCRF-CEM cells (obtained as a gift from St. Jude's Childrens Research Hospital, Memphis, TN) were grown in RPMI-1640 media (2 µM folic acid), (high folate media) (Whittaker Bioproducts, Walkersville, MD) supplemented with 10% dialyzed fetal bovine serum (Sigma Chemical Co., St. Louis, MO). Cultures were maintained in log phase at 37°C in a humid atmosphere containing 5% CO₂. For all experiments, cells were seeded in fresh media at a density of 1 × 10⁵/ml and allowed to grow for 16 hr before treatment. Nuclei were counted on a Coulter ZM Particle Counter (Coulter Electronics, Hialeah, FL) by treating cells with the stromal lysing agent, Zap-O-Globin (Coulter Inc., Hialeah, FL) before analysis. Cell growth data were fit to a first order exponential equation using DeltaGraph Professional (DeltaPoint, Inc., Monterey, CA).

Analysis of DNA content. DNA content was measured as described previously (Tonkinson et al., 1997a). Briefly, after the indicated treatment, cells were harvested by centrifugation and fixed with 70% methanol in PBS (20°C, 24-72 hr). Fixed cells were pelleted at 2000 × g for 10 min and resuspended in 500 µl of Vindelov's PI stain (Robinson, 1993). PI fluorescence was measured on a Coulter Electronics XL analytical cytometer (Hialeah, FL). Data was analyzed with WinCycle software (Phoenix Flow Systems, San Diego, CA). The histograms in figures 3 and 5 are from the same time course, and the data were collected at the same time. Vertical lines were drawn through channel 360 of each histogram. This represents the approximate division between G₁ (2n) DNA content and S (>2n) DNA content. The histograms were drawn using the full-scale option so that the different y axis scales indicate differences in the height of the tallest peaks.

Microscopic images. Micronucleation was visualized by PI fluorescence using a Nikon Diaphot inverted microscope. CEM cells were fixed in methanol as described above, and then stained with 0.1× Vindelov's PI stain. Images were captured with 20× magnification using Kodak Vericolor III film, ASA = 160.

Nucleotide analysis. After the indicated treatment, CEM cells were harvested, counted and washed twice with PBS. Cells were extracted three times with 60% ethanol in water (Chong and Tattersall, 1995). The extracts were combined, lyophilized and redissolved in 20 mM phosphate, pH 7.0. Ribonucleoside triphosphate pools were measured by HPLC using a Partisil 10 SAX column (4.6 × 250 mm). Nucleotides were eluted isocratically with 1 part acetonitrile and 10 parts ammonium phosphate solution, 0.4 M, pH 4.45. Deoxynucleoside triphosphates were measured after destruction of the corresponding nucleoside triphosphates. This was accomplished by 20 mM NaIO4 and 0.2 M CH₃NH₂ (final) (Garrett and Santi, 1979). The treated extracts were separated by HPLC as described above, except that the pH of ammonium phosphate solution was 3.25.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Enzyme inhibition in vitro and polyglutamylation. The ability of LY354899 to inhibit several folate requiring enzymes as well as enzymes involved in folate metabolism was examined (table 1). LY354899 was a potent inhibitor of the dehydrogenase activity associated with C1-tetrahydrofolate synthase (K_i = 19 nM), but was significantly less potent against the synthetase domain on the same enzyme. Furthermore, the compound was weakly inhibitory against several folate requiring enzymes involved in nucleotide biosynthesis, including human DHFR and TS, and murine GARFT. Although LY354899 possesses a terminal glutamate moiety, it was not a substrate for FPGS (data not shown). This result indicated that LY354899 was probably not polyglutamated inside cells.

Antiproliferative activity and cell cycle effects. The antiproliferative activity of LY354899 was examined against CEM cells grown in high folate media (2 µM folic acid) or low folate media (2 nM folinic acid). The results are shown in figure 2. Cells were counted over a 48-hr time course during continuous incubation with LY354899 (fig. 2a). To determine the dose response, the exponential growth rate constant of each curve was normalized to the growth rate constant of the untreated population and then plotted as a function of concentration (fig. 2c). The antiproliferative activity of LY354899 exhibited a modest dose response with an estimated IC₅₀ = 128 µM. The potency was increased slightly under low folate conditions, with an estimated IC₅₀ = 72 µM (fig. 2 b and c).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   The growth of CEM cells is inhibited by LY354899. CEM cells in high folate media (A) or low folate media (B). Cells were dosed with 0 (), 0.1 µM (+), 1 µM (), 10 µM (), or 100 µM () LY354899 approximately 16 hr after seeding. Cell counts were made at the indicated times. The data shown are from a representative experiment. The data for each treatment were fit to a first order exponential equation, and the rate constants were normalized to the untreated population and plotted as a function of concentration (C); () cells in high folate media, (+) cells in low folate media; each data point is an average of three independent experiments, error bars represent the standard error of the mean.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   DNA content histograms of CEM cells incubated with LY354899 (100 µM). CEM cells were fixed with 70% methanol and stained with Vindelov's PI stain as described in "Materials and Methods." The vertical lines were drawn to approximate the division between G1 and S phase DNA content.

Coadministration of hypoxanthine potentiated the apoptotic activity of LY354899. The cell cycle profiles indicated that two temporally distinct events occurred in response to treatment with LY354899. The first resulted in a cell cycle profile similar to that of CEM cells treated with GARFT inhibitors, and the second resulted in a profile similar to that of cells treated with TS inhibitors and ended with apoptosis. Previously we reported that in CEM cells, a purineless state induced by GARFT inhibition resulted in cytostasis, whereas a thymineless state, induced by TS inhibition, resulted in apoptosis (Tonkinson et al., 1997a). We hypothesized that if LY354899 induced an antipurine effect followed by a thymineless effect, then, bypassing the antipurine effect would result in a more rapid and perhaps potentiated apoptotic response.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   The antiproliferative effect of LY354899 is potentiated by hypoxanthine. CEM cells grown in high folate media (A) or low folate media (B) were incubated with LY354899 (100 µM) () or LY354899 + hypoxanthine (100 µM) (+). Cells were counted, and the exponential growth constants were normalized to control and plotted as a function of concentration; each data point is the average of three independent experiments, error bars represent the S.E.M.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   DNA content histograms of CEM cells incubated with LY354899 (100 µM) plus hypoxanthine (100 µM). CEM cells were harvested at the indicated times, fixed with methanol and stained with Vindelov's PI as described in "Materials and Methods." The vertical line was drawn to approximate the division between G1 and S.

View larger version (47K): [in this window] [in a new window]   Fig. 6.   Fluorescent microscopic images of CEM cells treated with LY354899 and hypoxanthine; A, untreated cells; B, LY354899 (100 µM, 34 hr); C, LY354899 (100 µM, 34 hr) + hypoxanthine (100 µM, 34 hr).

Measurement of ribo- and deoxyribonucleoside triphosphates. Nucleotide pools were measured from cells treated with LY354899 (100 µM) with and without hypoxanthine (100 µM). Cells were harvested and the nucleotides were separated by HPLC as described in "Materials and Methods." LY354899 alone did not dramatically alter the pool sizes of ribonucleoside triphosphates (data not shown). However, in cells treated with the combination of LY354899 and hypoxanthine, a 2-fold increase in ATP was observed after 30 hr of treatment. For 1 × 10⁶ cells, the level of ATP rose from 3.2 to 6.8 nmol.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Inhibition of nucleotide biosynthesis in cell culture leads to a cytostatic-like effect or to apoptosis depending on which nucleotide pools are affected (Tonkinson et al., 1997a). We previously reported that pure inhibitors of GARFT did not induce apoptosis in CEM cells during a 72-hr incubation (Tonkinson et al., 1997a). Instead, the cells entered a quiescent state, with normal respiration, but no detectable DNA synthesis. Smith et al. (1993) reported that WiDR cells treated with the GARFT inhibitor lometrexol, could be maintained in soft agar for up to 3 mo. The cells did not replicate during this time, but there was no evidence of membrane breakdown. In contrast, apoptosis resulted from inhibition of thymidine production in CEM cells. The TS inhibitor ZD1694, the multi-targeted inhibitor, MTA, and methotrexate have all been reported to decrease dTTP levels in CEM cells and to induce apoptosis after 36 to 48 hr of continuous treatment (Pandero et al., 1995; Tonkinson et al., 1997a). It is interesting to note that methotrexate treatment results in apoptosis even though purine and deoxy-purine levels fall coincidentally with dTTP levels.

In contrast to the results of TS, GARFT and DHFR inhibition, the effect of C1-synthase-dehydrogenase inhibition on nucleotide biosynthesis is not fully understood. Extensive biochemical studies have been reported detailing the mechanism, structure and rate constants for this enzyme and the reactions it catalyzes (Tan et al., 1977; Villar et al., 1985; Mejia et al., 1986; Green et al., 1988; MacKenzie et al., 1988; Pelletier and MacKenzie, 1995; Cheung et al., 1997). However, the effects of intracellular inhibition of any of the activities of C1-synthase have not been reported. Temple et al. (1982) reported the synthesis of compounds that inhibited some of the activities of C1-synthase, including the dehydrogenase, but these compounds were reported to be poor cytotoxic agents.

Our data confirm that one of the compounds synthesized by Temple et al. (1982) 5,6,7,8-tetrahydro-N⁵,N¹⁰-carbonylfolic acid, LY354899, is a potent in vitro inhibitor of the dehydrogenase activity of C1-synthase, but is a relatively poor antiproliferative agent. At 100 µM, less than 50% growth inhibition was observed over a 48-hr time course. However, it is clear from our studies that at long incubation times (>48 hr) this compound induces apoptosis. The lack of polyglutamation probably contributes to the lack of potency. The relevant intracellular concentration may be too low to effectively compete with endogenous folate pools. The slight (2-fold) increase in potency observed in CEM cells in low folate media supports this hypothesis. It is also possible that LY354899 is internalized more efficiently in CEM cells adapted to grow in low folate media. This property would increase the intracellular concentration of drug relative to cells grown in normal media.

The poor antiproliferative activity of LY354899 may also be due to continued synthesis of 10-formylTHF from formate and THF via 10-formylTHF synthetase (MacKenzie et al., 1988). Inhibition of the dehydrogenase activity of C1-synthase would only prevent the formation of 10-formylTHF from serine. The fact that LY354899 was antiproliferative suggests that the activity of 10-formylTHF synthetase in CEM cells must be relatively inefficient.

Despite the poor antiproliferative activity of LY354899, some very interesting cell cycle alterations were observed. The effects of LY354899 over a 72-hr treatment appeared to be a temporal composite of purine deoxyribonucleoside triphosphate depletion and thymidine triphosphate depletion. Initially there appeared to be an increase in the number of cells with S phase DNA content, although a distinct G₁ peak was maintained. This profile was very similar to effects that we and others have reported for cells treated with pure GARFT inhibitors (Beardsley et al., 1989; Tonkinson et al., 1997a). CEM cells treated with GARFT inhibitors rapidly entered a quiescent state, with cells in all phases of the cell cycle (Tonkinson et al., 1997a). Unlike cells treated with GARFT inhibitors, the DNA content of cells treated with LY354899 continued to increase, so that after 36 to 48 hr the entire population had a DNA content consistent with cells in early S phase. After 72 hr, the DNA content of all cells had decreased, without a population doubling. This profile along with morphological changes (data not shown) indicated that cell death had occurred. A very similar profile has been observed for CEM cells treated with pure TS inhibitors such as ZD1694, DHFR inhibitors, such as methotrexate, or a multitargeted inhibitor like MTA (Tonkinson et al., 1997a). These compounds induce apoptosis in CEM cells.

Our data demonstrating a potentiated apoptotic response when hypoxanthine was combined with LY354899, supports the hypothesis that inhibition of the dehydrogenase results in two sequential events. When hypoxanthine was added, providing a purine source, the cell cycle profile was no longer a composite, but rather resembled only that of cells treated with compounds that induce thymineless death. This was followed by apoptosis within 34 hr of compound addition rather than 72 hr for LY354899 alone. In addition, thymidine (5 µM) was able to partially overcome the antiproliferative effects of LY354899. Rescue was observed regardless of whether thymidine was added at the start of the experiment or after 36 hr of treatment when the thymineless effect was observed in the DNA histogram. In addition, thymidine reduced the extent of cytotoxicity induced by the combination of LY354899 and hypoxanthine.

Examination of nucleotide pools provided biochemical confirmation of our cell cycle observations. LY354899 alone caused an initial depletion of dATP and dTTP pools. However, after 30 hr, dATP levels increased, although dTTP pools remained low. In cells treated with the combination of LY354899 and hypoxanthine, dATP levels increased at 8 hr and continued to increase during the next 22 hr, although dTTP levels declined. The net result was an increase in the dATP:dTTP ratio before the onset of apoptosis. This profile is distinct from that of methotrexate, where levels of dATP and dTTP fell rapidly and simultaneously (Taylor et al., 1982; Chen et al., 1998). In contrast, TS inhibition results in an increase in the dATP:dTTP ratio before the onset of apoptosis (Chong and Tattersall, 1995; Chen et al., 1998). This profile is virtually identical to the combination of LY354899 and hypoxanthine.

Our results provide information on the cellular and biochemical effects of inhibiting an activity associated with C1-synthase. These effects are distinct from those that result from inhibition of other folate requiring enzymes. It is important to note that our results were limited to CEM lymphocytic leukemia cells. CEM cells have been used to examine the effects of many antifolate compounds. The responsiveness of CEM cells to disruption of metabolic pathways and their susceptibility to apoptosis makes for an attractive model system. It is possible that different results may be obtained in other cells types, especially those derived from solid tumors.