Identification and Biochemical Properties of 10-Formyldihydrofolate, A Novel Folate Found in Methotrexate-treated Cells*

The folate compound 10-formyldihydrofolate (H,folate) has not been found as a component of intracellular folates in normal tissues but has been identified in the cytosol of methotrexate (MTX)-treated MCF-7 breast cancer cells and normal human myeloid precur-sor cells. Its identity was verified by coelution of this compound with a synthetic marker on high pressure liquid chromatography, its reduction to 10-formyl-tetrahydrofolate (H4folate) in the presence of dihydrofolate reductase, and its enzymatic deformylation to dihydrofolate in the presence of aminoimidazolecarboxamide ribonucleotide (AICAR) transformylase. Chemically synthesized monoglutamated or pentaglutamated 10-formyl-H,folate was examined for its interaction with three folate-dependent enzymes: AICAR transformylase, glucinamide ribotide (GAR) transfor-mylase,andthymidylatesynthase. 10-Formyl-H,folate- Glu, was a competitive inhibitor of thymidylate synthase (Ki = 0.16 ~ L M with 5,10-methylene-H4folate-Glu, as substrate and 1.6 pM with


= 2.0 p~) .
It acted as a substrate for AICAR transformylase ( K , = 5.3 pM), and its efficiency was equal to that of the natural substrate 10-formyl-H4folate-Glus. The inhibition of thymidylate synthase by 10-formyl-Hafolate was highly dependent on the inhibitor's polyglutamation state, the -Glus derivative having a 52-85-fold greater affinity as compared to the affinity of -Glul. Polyglutamation of 10-formyl-Hzfolate did not affect its inhibition of GAR transformylase. While the actual role of 10-formyl-H2folate contributing to the cytotoxicity of MTX has not been determined, this compound has the potential to enhance inhibition of GAR transformylase and thymidylate synthase, and at the same time provides additional substrate for AICAR transformylase. The MTX-induced intracellular accumulation of 10-formyl-Hzfolate and H,folate may play a role in the drug-related cytotoxicity through the contribution of these folates to the inhibition of thymidylate synthase and de novo purine synthesis.
Methotrexate (2,4-diamino,l0-methylpteroyl glutamic acid; * 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. 8 To whom correspondence and reprint requests should be addressed Clinical Pharmacology Branch, Bldg. 10, Rm. 12N226, National Cancer Institute, Bethesda, MD 20892, MTX)' (1) is a tight-binding inhibitor of dihydrofolate reductase (DHFR). By virtue of this action, MTX produces a partial depletion of intracellular reduced folates and at the same time a marked accumulation of dihydrofolate (H2folate). The reduced folate pool in normal and malignant tissues consists of a variety of compounds. Two of these are particularly important to DNA synthesis: 10-formyl-H,folate, a substrate for the GAR transformylase and AICAR transformylase reactions, and thus an essential factor for de novo purine synthesis; and 5,10-methylene-H4folate, the folate cosubstrate for thymidylase synthase.
We have previously examined the intracellular folate pool in MCF-7 breast cancer cells and in normal myeloid progenitors isolated from human marrow from normal volunteers and have found relative preservation of lO-formyl-H,folate in MTX-treated cells despite concurrent suppression of the de novo purine pathway (2,3) and marked cytotoxicity (3). Similarly, thymidylate synthesis was inhibited in MTXtreated MCF-7 cells without a quantitative depletion of 5,lOmethylene-H,folate (4). In the absence of depletion of the reduced folates, as has been shown in these studies, alternative mechanisms of MTX cytotoxicity related to direct inhibition of the folate-dependent enzymes of purine and thymidylate may be important. We have identified two types of compounds generated during incubation of cells with MTX that inhibit both thymidylate synthase and AICAR transformylase. One class is the series of MTX polyglutamates, found in both normal and malignant cells, which in contrast to the parent drug potently inhibit enzymes other than DHFR (5,13). The other potential inhibitors of these enzymes are the dihydrofolate polyglutamates, which accumulate within the cells in response to DHFR inhibition by MTX and which inhibit AICAR transformylase (5) and thymidylate synthase (6,7).
During studies of the folate pools in MTX-treated MCF-7 breast cancer cells, we identified a novel folate peak by HPLC, a compound not detectable in untreated tumor cells or normal bone marrow progenitor cells but one that becomes the predominant intracellular folate in MCF-7 breast cancer cells after 12 h of drug exposure. We report here the chemical properties and biochemical action of this compound, which we believe to be 10-formyldihydrofolate. This compound is shown to inhibit both thymidylate synthase and GAR transformylase and to serve as a substrate for AICAR transformylase.
10-Formylfolate was synthesized by the method of Blakley (18). Folic acid-Glu, was dissolved in 98% formic acid and allowed to stand at room temperature for 56 h to complete formylation, as verified by a single peak on HPLC. 10-Formylfolate-Glue was prepared from folic acid-Glu5 by the same method, and its purity was verified by the presence of a single peak on HPLC after enzymatic hydrolysis of the product with porcine kidney conjugase. To test the stability of the polyglutamated peptide bonds under experimental conditions, the polyglutamate profile of a sample of formylfolate pentaglutamate was compared to the starting substrate. Using an HPLC system designed to separate MTX polyglutamates (19), and that we found could separate folate polyglutamates, no degradation to intermediate polyglutamates was noted during the formylation of folic acid pentaglutamate. The solutions of 10-formylfolate-Glul and -GlUs were lyophilized and the powders stored at -40 "e. NH4HC03 (8,9).
Preparation of 10-Formyl-Hzfolate We synthesized 10-formyl-H2folate by three different methods. Preparation of this compound by air oxidation of 10-formyl-H4folate was reported previously (20)(21)(22). In our experience, when a 1 mM solution of 10-formyl-H4folate at pH 7.0 was left at 23 'C for 2-3 h, this procedure resulted in 90-95% oxidation to 10-formyl-Hzfolate. The 5-10% impurity consisted mainly of 10-formylfolate when the oxidation of 10-formyl-H4folate was complete. As an alternative approach, 1 pmol of 2,6-dichloroindophenol (a mild oxidizing agent) was added to 1 ml of 1 mM 10-formyl-H4folate for 10 min at 4 "c, and the oxidant was extracted three times with 10-ml volumes of ether; the residual ether in the aqueous layer was removed by evaporation under nitrogen. The relative purity of 10-formyl-H2folate produced by this method (compound 1) was 97%, as the final preparation included less than 3% of 10-formylfolate. However, the product rapidly oxidized and 10-formylfolate became 5-10% of the preparation content within 20 min at 4 "C (see "Determination of Folate Purity by HPLC"). The UV spectrum of the freshly prepared compound 1 is shown in Fig. 2. In view of its instability when prepared by oxidation of 10-formyl-H4folate, 10-formyl-H2folate was also prepared by reduction of 10-formylfolate by modification of a previously described method (23). Sodium hydrosulfite (200 mg) was added to 10 ml of 2 mM 10-formylfolate in 0.1 M phosphate buffer, pH 7.5. The completeness of 10-formylfolate reduction was determined by disappearance of this peak on HPLC. Excess sodium hydrosulfite was precipitated by cold ethanol (5 ml) and removed by centrifugation at 20,000 rpm for 15 min. The supernatant was lyophilized and the powder, still containing residual sodium hydrosulfite, was essentially free from 10-formylfolate, as determined by HPLC, for up to 2 months at -40 "C (compound 2). A fresh solution was prepared prior to each experiment, and 10-formyl-H2folate concentration was determined at 272 nm and at pH 7.0 using the molar extinction coefficient of 22,000 M" cm" (Fig. 2).
When compound 2 and the freshly prepared compound 1 were compared with respect to activity with each of the enzymes of the study, identical reaction velocities were observed, thus excluding potential interference of sodium hydrosulfite with the catalytic reaction. These two methods for synthesizing 10-formyl-H2folate yielded products that coeluted on HPLC, as did the air oxidation product of lO-formyl-H,folate.
HPLC Analysis of Intracellular Folates Intracellular folates were quantitated according to the method reported by us previously (2,3). Briefly, MCF-7 cells were incubated with [3H15-formyl-H4folate at 0.1 p~ concentration for 4 days, and the labeled intracellular folates were extracted by 1-min boiling in a solution containing 2% 2-mercaptoethanol and 2% ascorbate. After enzymatic hydrolysis of the polyglutamated residues, the various folate pools were separated by HPLC, using a Waters HPLC system as described previously (2,3). The mobile phase consisted of 76% Pic Reagent A, pH 5.2, and 24% methanol, and the compounds were eluted using isocratic conditions and a flow of 2 ml/min. The eluate was collected at 1-min fractions and quantitated by liquid scintillation counting. The identity of radiolabeled folate peaks was verified by coelution with UV peaks of standard folate markers. We used the same HPLC system at pH 7.4 to analyze the changes in composition of radiolabeled folate extracts following their incubation with DHFR and AICAR transformylase (2).

Determination of Folate Purity by HPLC
The purity of the synthetic monoglutamated folates in this study (except 5,10-methylene-H4folate) was determined by HPLC analysis, and the percent purity was expressed as the ratio between the area under the corresponding peak and the total area under all folate peaks in a chromatogram, multiplied by 100. HPLC conditions were similar to those described for separation of intracellular folates, and the pH of Pic Reagent A was 7.4. Since pentaglutamated folates are not eluted in this HPLC system (3) and cannot be quantitated directly, the pentaglutamated folates were first incubated with hog kidney conjugase at 37 "C for 30 min (2, 3) and then analyzed by HPLC.
Elution as a single peak after this procedure was considered as evidence for the purity of the original compound. This method was used to assess purity of 10-formylfolate-Glu6 and 10-formyl-Hzfolate-Glus, while the purity of H2folate-Glu6, H4folate-Glu6, lo-formyl-H4folate-Glu5, and 5,10-methylene-H4folate-Glu~ was spectrally assessed as described previously.

Enzyme Source and Purification
A human breast cancer cell line, MCF-7, was used as the source of AICAR transformylase, GAR transformylase, and thymidylate synthase. The characteristics of this cell line have been previously described (24). Cells were grown in a monolayer culture in rolling flasks and stored at -40 'C. AICAR transformylase was purified 400fold, largely through use of the Affi-Gel Blue affinity chromatography, as described previously (5). The specific activity of the purified enzyme was 216 nmol of H4folate formed per min/mg protein at 37 "C. GAR transformylase was partially purified (24-fold) by a 1% streptomycin sulfate precipitation and ammonium sulfate fractionation. The 3545% ammonium sulfate precipitate was dissolved in 10 mM phosphate buffer, pH 7.2, and dialyzed against the same buffer. GAR transformylase specific activity was 48.2 nmol of H4folate formed per min/mg protein at 37 "C. Thymidylate synthase was purified 150-fold to a specific activity of 0.37 nmol TMP/min/mg protein according to the method described by Dolnick and Cheng (25). Lactobacillus casei DHFR was purchased from the New England Enzyme Center (Boston, MA) (specific activity 0.64 pmol of tetrahydrofolate formed/min/mg at 37 T ) .
Enzyme Assays AICAR Transformylase Assay-Activity was measured using the spectrophotometric assay developed by Black et al. (26). The reaction cuvettes contained 50 mM Tris-HC1, pH 7.4, 50 mM KCI, 50 mM 2mercaptoethanol, 50 p~ AICAR, 4.4-7.4 units of enzyme (1 unit of enzyme defined as amount of enzyme required to form 1 nmol of H4folate), and various amounts of 10-formyl-H4folate-Glul or -Glu6. Reaction rates were monitored with a Gilford 2400-2 recording spectrophotometer equipped with a constant-temperature, water-jacketed sample compartment. Reaction velocities were linear for 10 min with respect to time. The absorbance changes were measured at 298 nm and converted to nanomoles of H4folate formed per min/mg protein, using the extinction coefficient for the reaction of 19,700 M" cm" at 298 nm (26). In assays with 10-formyl-Hpfolate as coenzyme, the absorbance changes were measured at 300 nm and converted to nanomoles of Hpfolate formed per min/mg protein by using an extinction coefficient of 9,500 "' cm". The absorbance at 300 nm of Hzfolate is 17,200 M" cm" and that of 10-formyl-Hzfolate is 7,700 M" cm" (Fig. 1). Thus the change in extinction coefficient for the reaction represented the difference in extinction coefficient of these two compounds at 300 nm. At pH 7.0, the peak molar extinction coefficient for lO-formyl-H,folate (22,000 M" cm") occurs at 272 nm (Fig. 21, while the peak molar extinction coefficient for Hzfolate (22,400 M" cm") occurs at 282 nm (16).
GAR Transformylase Assay-Activity was measured by modification of the spectrophotometric assay of Smith et al. (27). One-ml Curve 1 (solid line) represents the spectrum of 3.2 X M 10-formyl-Hzfolate (product 1) and an equimolar Hnfolate concentration at pH 7. The maximum absorbancy shift during transformylation by AICAR transformylase, with conversion of 10-formyl-H,folate to Hzfolate, was measured at 300 nm (dotted line).
I reactions were carried out in 50 mM Tris, pH 6.8, 25 mM KC1, 0.1 mM EDTA, 300 pM GAR, 1.5-2.4 units of enzyme, and various amounts of 10-formyl-H4folate-Glu, or -GI&. The reaction velocities were calculated in the same manner as for the AICAR transformylase reaction and were linear with respect to time for 10 min.
Thymidylate Synthase Assay-Thymidylate synthase was assayed by modification of the tritium-release assay of Roberts (28) as described in detail elsewhere (12). The reaction velocities were found to be linear for 10 min with 5,10-methylene-H4folate-Glu~, 0.45-4.0 pM, or for 30 min with 5,lO-methylene-H4folate-Glul, 15-90 pM, as substrates, with 0.018-0.03 units of thymidylate synthase (1 unit of TMP/min at 37 "C). enzyme defined as the amount of enzyme required to form 1 nmol of

Data Analysis
All kinetic data were first analyzed by Lineweaver-Burk doublereciprocal plots, and the graphic estimates of parameters derived from these plots were used as initial estimates for computerized curvefitting using a weighted nonlinear least squares method. Curve-fitting was performed by a program designated "enzyme" that was developed by Drs. D. Rodbard and R. Lutz (National Institute of Child Health and Human Development, Bethesda, MD).

Identification of 10-Formyl-Hdolute in Cellular Extracts
The major intracellular folates were identified in extracts of MCF-7 cells after incubation with [3H]5-formyl-H,folate. In the extracts of untreated MCF-7 cells, we identified 10formyl-H,folate, H4folate, 5-formyl-H4folate, and 5-methyl-H,folate (Fig. 2, panel A ) . Analysis of intracellular folates from MCF-7 cells, pretreated with 1 FM MTX for 4 h before the extraction, revealed characteristic changes, i e . a marked decrease in the 5-methyltetrahydrofolate pool (peak 5), 20-40% decrease in 10-formyl-H4folate (peak I ) , and accumulation of H,folate (peak 4 ) , as reported previously (Fig. 2,  with formaldehyde. Peak 0, p-aminobenzoylglutamic acid; peak I , 10formyl-H,folate; peak 2 H4folate or 10-formyl-H2folate; peak 3, 5formyl-Hdfolate; peak 4, H2folate; peak 5, 5-methyl-H4folate; peak 6, 5,lO-methylene-Hlfolate. which corresponded to the H,folate peak in untreated cells (peak 2). To determine the identity of this peak at 9 min in treated cells, extracts from MTX-treated and normal cells were subiected to incubation with 0.1 mM formaldehvde for 1 h at room temperature, a procedure that converts H,folate to 5,lO-methylene-H4folate (Fig. 2, panels C and D). While peak 2 in the MTX-untreated cells underwent quantitative conversion to 5,lO-methylene-H4folate (peak 6) during incubation with formaldehyde (panel C), little effect was seen on the 9min peak from MTX-treated cells (panel D). The migration of this radioactive peak on HPLC was compared to that of authentic unlabeled 10-formyl-H,folate, which by UV analysis eluted 1.5 min before the labeled peak at 9 min (Fig. 3, peak  A). However, when authentic 10-formyl-Hzfolate was treated using an extraction procedure identical to the intracellular folates, its relative elution time was prolonged and corresponded to peak 2 (9.6 min) (Fig. 3, peak B). Thus, the retention time of the material in peak 2 and its failure to react with formaldehyde were consistent with its identification as 10-formyl-Hzfolate. Additional studies showed that ascorbate was responsible for the alteration in HPLC retention time of 10-formyl-H,folate. The deletion of ascorbate from the extraction procedure resulted in a retention time (7.5 min) identical to that of standard 10-formyl-Hzfolate, while the deletion of either 2-mercaptoethanol or the conjugase enzyme preparation had no effect on the prolongation of retention time associated with the extraction procedure. While ascorbate increased the retention time of 10-formyl-H,folate, [14C]ascorbate did not co-migrate with the 10-formyl-Hzfolate peak.
Peak 2 HPLC eluate from MTX-treated MCF-7 cell extracts was pooled, re-analyzed by HPLC under similar conditions, and found to consist of a predominant peak coeluting with authentic 10-formyl-H,folate marker (Fig. 4, panel A), with less than 10% of material eluting in a small second peak of 15 min, corresponding to its oxidative product, 10-formylfolate.
To determine whether the new putative 10-formyl-Hzfolate peak could undergo reduction, material eluting as peak 2 was incubated for 4 h at 37 "C with dihydrofolate reductase (L. A mixture of folate markers, 2-10 pmol each, was separated by HPLC. The elution time of each compound correlated with the elution of the corresponding tritiated intracellular folate (Fig. 2). The elution time of authentic 10-formyl-Hafolate (broken line, peak A) preceded the labeled peak 2 by 1.5 min, but co-eluted with this peak after the authentic compound underwent extraction with 2% mercaptoethanol and 2% ascorbate (broken line, peak B ) . The numbering system used to identify each folate peak is identical to that used in Fig. 2. The early peaks are 2-mercaptoethanol and ascorbic acid. Each experiment also contained -10 nmol of unlabeled 10-formyl-H2folate, and its UV conversion followed closely the changes in elution of the tritiated compound. Peak 1, IO-formyl-H4folate; peak 2, 10-formyl-H2folate; peak 3, = 10-formylfolate; peak 4, Hzfolate; peak 5, folic acid. For clarity, the unrelated UV peaks (mercaptoethanol, ascorbate, AICAR) as well as UV and radiolabeled peaks of p-aminobenzoylglutamic acid were omitted.
formyl-H4folate (Fig. 4, panel B). Authentic 10-formyl-Hzfolate underwent the same conversion to 10-formyl-H,folate in the presence of DHFR. We next examined the ability of peak 2 material to serve as a substrate for 400-fold purified AICAR transformylase. Eluate from peak 2, to which was added authentic unlabeled 10-formyl-Hzfolate, was incubated with AICAR transformylase. Incubation resulted in a conversion of 80% of peak 2 radioactivity to material that eluted with Hzfolate or its further oxidation product folic acid (Fig. 4, panel C, peak 5 ) . An identical rate/extent of conversion was observed simultaneously for the unlabeled 10-formyl-H,folate, as monitored by the elution of UV-absorbing material by HPLC (Fig. 4C).
These studies indicated that the peak 2 material, found only in MTX-treated cells, migrated on HPLC in a pattern consistent with 10-formyl-H,folate and underwent the same transformations with DHFR and AICAR transformylase as observed with authentic 10-formyl-H,folate.

Biochemical Characteristics of 1 O-Formyl-H&ute
To understand the biochemical reactivity of 10-formyl-Hzfolate, we studied the effects of this folate on three folatedependent enzymes: thymidylate synthase, GAR transformylase, and AICAR transformylase.

Inhibition of thymidylate synthase by IO-formyl-Hzfolate-Glu, and Hzfolate-Glu,
The inhibition constants (KJ for each of the inhibitors are tabulated with respect to mono-and pentaglutamated substrate 5,10-methvlene-H,folate. The K , for each is also reported.  5 ) .
Effect of IO-Formyl-Hdolute-Glu, and Hdolate-Glu, on GAR Transformyhe-We studied the effects of mono-and pentaglutamated 10-formyldihydrofolate and dihydrofolate as direct inhibitors of MCF-7 GAR transformylase. Inhibition constants for each compound were determined at a constant GAR concentration of 300 p~ and at variable concentrations of the mono-or pentaglutamated 10-formyl-H,folate (Table  11). The inhibition pattern was competitive with the folate small decrease in K , (6.5 p~ with monoglutamated versus 4.9 p~ with pentaglutamated 10-formyl-H,folate). The effect of polyglutamation state of the inhibitor on GAR transformylase catalysis was further investigated with MTX-Glul to Glu5 (Table 111). A stepwise increase in inhibition potency occurred with the addition of three or more glutamyl groups, with a 32-fold greater inhibition with MTX-Glu5 than MTX-Glul. However, the K, of MTX-Glu5 (2.5 PM) was approximately the same as that of 10-formyl-H,folate-Glu5 when these inhibitors were used in conjunction with the monoglutamated substrate, 10-formyl-H,folate.

DISCUSSION
10-Formyl-Hzfolate has been previously described as a product of 10-formyl-H4folate oxidation, but it was not considered to have physiologic significance (22, 29) and was not detected in prior studies of normal cell extracts. The recognition of this compound in cellular extracts was complicated by the proximity of its peak on HPLC to other folates, and in particular its elution with H4folate under the conditions used in this study. Using the HPLC methodology described in this study, it is not possible to detect 10-formyl-H2folate in measurable amounts in normal untreated cells. It is detectable only after treatment with MTX, as is H,folate. Two questions were addressed in the first part of this study: Is the new compound indeed 10-formyl-H,folate? Secondly, if the compound is 10-formyl-H,folate, is it an experimental artifact of 10-formyl-H4folate oxidation or is it actually generated as a result of MTX effects on folate metabolism? The identity TABLE I1

Inhibition of GAR transformylase by IO-formyl-HJolate-Glu, and Hzfolate-Glu,
The inhibition constants (K;) for each of the inhibitors are tabulated with respect to mono-and pentaglutamated substrate, 10-formyl-H4folate. The corresponding K, is also reported.

Inhibition of GAR transformylase by MTX-Glu,.,
Polyglutamate derivatives of MTX (Glul to Glu,) were tested as inhibitors of the human GAR transformylase with respect to the monoglutamated 10-formyl-H,folate as substrate.  The Michaelis-Menten constants (K,,,) and Vmax with AICAR transformylase and VJK,,, ratios of 10-formyl-H2folate-Glu, and Glu6 were measured and compared to values for 10-formyl-H,folate-Glu,. of lO-formyl-H,folate in the cellular extracts was verified by co-elution of the radiolabeled peak 2 with the unlabeled 10formyl-H,folate marker when both the cellular folate and the standard underwent the same extraction procedure before HPLC separation. Furthermore, treatment of the putative intracellular lO-f?rmyl-H,folate peak with dihydrofolate reductase and with AICAR transformylase revealed characteristic shifts in elution pattern that support the identity of this compound as being 10-formyl-H2folate.
Untreated MCF-7 cells did not contain 10-formyl-H,folate; the folate peak eluting at 9 min included only H,folate. In an extract from untreated cells, conversion of 10-formyl-H, to lO-forrnyl-H,folate during the processing of cell extracts would be possible under conditions allowing oxidation, but this process is almost completely prevented by inclusion of high concentrations of 2-mercaptoethanol and ascorbate; furthermore, the absence of 10-formyl-H2folate in untreated cells indicates that the material found in treated cells was not an experimental artifact. While both MTX-treated and untreated cells contained 10-formyl-H4folate, 10-formyl-H,folate was found only in MTX-treated cells, an observation that further supports its existence in the cells.
We examined the effect of 10-formyl-H,folate on thymidylate synthase, GAR transformylase, and AICAR transformylase, all folate-dependent enzymes that synthesize the precursors of DNA and/or RNA. When the inhibition of thymidylate synthase by 10-formyl-H,folate was compared to inhibition by H,folate, the product of the thymidylate synthase reaction, three main differences emerged 1) a competitive pattern of inhibition with 10-formyl-H,folate versus noncompetitive with H,folate, 2) lower K, values in experiments with 10-formyl-H,folate in comparison to identical experiments with H,folate, and 3) polyglutamation of lO-formyl-H,folate induced relatively less enhancement of inhibition than the polyglutamation of H,folate. These findings may be explained by the existence of separate binding sites present on each of two monomeric units of thymidylate synthase, with H,folate and 10-formyl-H,folate binding to different sites. This model was suggested previously in studies of human thymidylate synthase derived from leukemic blast cells (6). H,folate was noncompetitive, while 10-formyl-H,folate, which is closely related to 10-formyl-H,folate, was a competitive inhibitor with a relatively lower Ki value. The authors suggested that a combination of these two compounds may bind to different sites on the enzyme and inhibit the catalysis without interference of either inhibitor with the binding of the second inhibitor (mutually nonexclusive pattern). A similar relationship probably exists between H,folate and 10-formyl-H2folate, thus producing a stronger net effect than might result from either compound alone. Previous work (2, 3) has shown that the concentration of formyl-H,folate in MTX-exposed cells approaches the inhibition constants for thymidylate synthase and GAR transformylase (0.5-1.0 p~) , suggesting that this compound may contribute to the inhibition of these enzymes during MTX treatment.
10-Formyl-H,folate was also a more potent inhibitor than H,folate when both were tested against GAR transformylase, the first folate-dependent enzyme in the de novo purine synthetic pathway. In contrast to thymidylate synthase, GAR transformylase catalysis and inhibition were minimally affected by the polyglutamation state of the folate compounds. This observation was further tested with MTX-GIU,-~ as inhibitors of this enzyme; surprisingly, MTX-Glu, was 32fold more potent than MTX (Table 111). A similar enhancement of inhibition was found previously in studies of chicken liver GAR transformylase, as inhibited by folic acid and its polyglutamates (30). Although the inhibition of GAR transformylase is enhanced by polyglutamation of MTX or folic acid, this enhancement is modest compared to the 150-2500fold enhancement of inhibition of AICAR transformylase (5) or thymidylate synthase (12) by polyglutamation of MTX and H,folate.
In contrast to the inhibitory effect on the two previously mentioned enzymes, 10-formyl-H,folate was a substrate for human AICAR transformylase. The relative specificity constant (VJK,,,) of the pentaglutamated 10-formyl-H,folate was identical to 10-formyl-H,folate-Glu,, the natural substrate of AICAR transformylase. This is consistent with a recently reported finding using AICAR transformylase isolated from chicken liver (22).
We have previously observed that important intracellular reduced folates, 10-formyl-H,folate and 5,lO-methylene-H,folate, are not significantly depleted with up to 24 h of incubation with cytotoxic concentrations of MTX, at a time when de novo purine and thymidylate synthesis were strongly inhibited (2)(3)(4). We suggest that these pathways may be inhibited directly by polyglutamated metabolites of H,folate and lO-formyl-H,folate. These compounds are formed in the presence of MTX and both induce inhibition of thymidylate synthase and GAR transformylase, the latter being the first of the two folate-related enzymes of the de novo purine pathway. The effects of these two oxidized compounds differed, however, with respect to AICAR transformylase. While Hzfolate inhibits AICAR transformylase, 10-formyl-H2folate acts as a substrate. This substrate activity of 10-formyl-Hzfolate is inconsistent with the assumption that inhibition of purine synthesis results from depletion of folate substrate for AICAR transformylase. We previously demonstrated that in MCF-7 cells treated with MTX the 10-formyl-H4folate pool undergoes little change. The present work demonstrates the formation of an additional substrate, 10-formyl-H,folate, for this enzyme, a compound that does not exist in the untreated cells. This finding adds further evidence for the importance of direct inhibition of de novo purine and thymidylate pathways by H,folate and MTX polyglutamates rather than an indirect inhibition through depletion of the required folate cofactors.
The origin of 10-formyl-Hzfolate remains an enigma. Its appearance only in the extracts from MTX-treated cells is not consistent with a simple oxidation of lO-formyl-H,folate during the extraction procedure. This, however, does not exclude the occurrence of spontaneous oxidation of 10-formyl-Hrfolate within the intact cells prior to the folate extraction. While the cells that were not treated by MTX retain the capacity to reduce 10-formyl-H,folate by DHFR, and thus maintain the 10-formylfolates in the fully reduced tetrahydro form, MTX-treated cells may accumulate lO-formyl-H,folate due to the DHFR block by MTX. This explanation, which assumes spontaneous intracellular oxidation, seems unlikely because of the stabilizing antioxidant intracellular milieu. An alternative explanation appears more likely. Preliminary evidence suggests that 10-formyldihydrofolate may be produced by an enzymatic formylation of dihydrofolate. This hypothesis stems from the observation that this reaction can be measured in MCF-7 cytosol and requires the presence of ATP and sodium formate. No formylation can be measured when MCF-7 cytosol is heat-denatured (31). The dihydrofolate-formylating enzyme appears to represent a new enzyme other than that capable of formylating tetrahydrofolate (32) since its dihydrofolate-formylating activity is unaltered by the presence of excess tetrahydrofolate. The abundant Hzfolate within MTX-treated cells provides a high concentration of substrate, which does not exist in the untreated cells. We have previously shown both in MCF-7 cells (2) and in bone marrow progenitors (3) that after 12 h or longer exposure to MTX, the H2folate level declines while 10-formyl-H2folate continuously increases such that the sum of these two folates remains constant. These findings are consistent with an active enzymatic mechanism for conversion of H,folate to 10-formyl-H,folate. The identification and characterization of this H2folate formylase activity is the subject of continuing investigations.