Antifolates Induce Inhibition of Amido Phosphoribosyltransferase in Leukemia Cells

The pathway for de nouo biosynthesis of purine nucleotides contains two one-carbon transfer reactions catalyzed by glycinamide ribotide (GAR) and 5-aminoimidazole-4-carboxamide ribotide (AICAR) transformylases in which N"-formyltetrahydrofolate is the one-carbon donor. We have found that the antifolates methotrexate (MTX) and piritrexim (PTX) completely block the de nouo purine pathway in mouse L1210 leukemia cells growing in culture but with only minor accumulations of GAR and AICAR to less than 5% of the polyphosphate derivatives of N-formylglycinamide ribotide (FGAR) which accumulate when the pathway is blocked completely by azaserine. This azaserineinduced accumulation of FGAR polyphosphates is completely abolished by MTX, indicating that inhibition of the pathway is at or before GAR transformylase (reaction 3; Lyons, s. D., and Christopherson, R. I. (1991) Biochem. Int. 24,187-197). Three h after the addition of MTX (0.1 PM), cellular 5-phosphoribosyl-l-pyrophosphate has accumulated 3.4-fold while 6-methylmercaptopurine riboside (25 PM) induces a 6.3-fold accumulation. These data suggest that amido phosphoribosyltransferase  catalyzing  reaction 1 of the pathway is the primary site of inhibition. In support of this conclusion, we have found that dihydrofolate-Glu5, which accumulates in MTX-treated cells, is a noncompetitive inhibitor of amido phosphoribosyltransferase with a dissociation constant of 3.41 f 0.08 WM for interaction with the enzyme-glutamine complex in vitro. Folate-Glu,, MTX-Glu,, PTX,  dihydrotriazine benzenesulfonyl fluoride, and AICAR also  inhibit amido phosphoribosyltransferase.


Methotrexate (MTX)' was synthesized as a structural an-
Inhibition of dihydrofolate reductase in human MCF-7 breast cancer cells exposed to MTX (1 p~) results in accumulation of polyglutamated dihydrofolate from undetectable levels to 20% of the total folate pool (3.9 pmol/mg protein; Allegra et al., 1986). Levels of N"-formyltetrahydrofolate were maintained at 84% of the control level after 21 h, indicating that MTX does not induce a cellular deficiency of one-carbon derivatives of tetrahydrofolate. Allegra et al. (1987) found that exposure of MCF-7 cells to MTX (10 p~) for 24 h induced a 3-fold accumulation of AICAR, and they concluded that the purine pathway is blocked at AICAR transformylase (reaction 9) by accumulated polyglutamated dihydrofolate. Pentaglutamyl derivatives of dihydrofolate and MTX are potent inhibitors of AICAR transformylase in vitro with Ki values of 2.7 and 5.9 pM, respectively (Allegra et al., 1985). Pentaglutamyl MTX also inhibits GAR transformylase in vitro with a Ki value of 2.5 PM (Chabner et al., 1985), and polyglutamated dihydrofolate may also inhibit this enzyme.

Inhibition of Purine
Biosynthesis by Antifolates 11039 of GAR or AICAR transformylase (reaction 3 or 9) was more severely inhibited by polyglutamated dihydrofolate accumulating in MTX-treated cells. However, we have shown recently that the primary site of inhibition induced by MTX in the pathway is at or before GAR transformylase (reaction 3; Lyons and Christopherson, 1991). Data presented in this paper indicate that MTX and PTX induce a primary blockade at the first reaction of the de nouo purine pathway catalyzed by amido phosphoribosyltransferase.

EXPERIMENTAL PROCEDURES
Purine Intermediates and Folate Derivatives-GMP, IMP, AMP, AICAR, folate, N5-formyltetrahydrofolate, and MTX were obtained from Sigma. Dihydrofolate and tetrahydrofolate from Sigma were dissolved just before use in degassed water that had been bubbled with nitrogen, and their purity was confirmed by anion exchange, gradient HPLC as described below for nucleotides using solvents purged with helium. MTX-Glu5 was supplied by Shircks Laboratories ( Synthesis of Dihydrofolate-Glu5-Folate-Glu5 was reduced to dihydrofolate-Glu5 with sodium dithionite as described by Coward et al. (1974) and purified by HPLC using solvents purged with helium. The reduced product was applied to a Whatman Partisil 10-SAX column (25 X 0.94 cm) and eluted with a linear gradient from 50 to 1,000 mM ammonium formate adjusted to pH 3.0 with formic acid. Fractions containing dihydrofolate-Glu5 were pooled and desalted immediately by passage over a Brownlee RP-18 column (22 X 0.46 cm) eluted with 100 mM formic acid and then methanol. The pure, desalted product eluted immediately with the methanol and was lyophilized and crystallized from a minimal volume of methanol at -20 "C. The product was then washed with a small volume of methanol, and the methanolinsoluble fraction was dissolved in D20. 'H NMR spectra were recorded in DzO (pH 6.2) at 400 MHz on a Bruker AMX 400 wide bore spectrometer or at 600 MHz on a Bruker AMX 600 spectrometer.  2.28, 2.24, 2.18, 2.02, 1.86, 1.84 (m, polyglutamyl protons). Methanol was used as an internal standard at b = 3.26 ppm. The concentrations of dihydrofolate and dihydrofolate-Glu5 used for inhibition experiments were determined by absorbance at 282 nm using an extinction coefficient of 28,000 1 mol" cm" (Bertino et al., 1965).
Growth of Cells with P'CIGlycine and f4ClFormate-Mouse L1210 leukemia cells were grown in RPMI 1640 medium (bicarbonate-free; Flow Laboratories, Sydney, Australia) containing 20 mM K/Hepes, 13% (v/v) fetal calf serum (Commonwealth Serum Laboratories, Melbourne, Australia), 50 pg/ml gentamycin (Flow Laboratories), 0.5 mM NaHC03, and the final pH was 7.2. Experiments were performed with cells in logarithmic phase, and drugs were added at a density of 5 X lo5 cells/ml. Intermediates of the de nouo purine pathway were radiolabeled by growth of cells in the presence of [I4C]glycine (125 pM, 22.4 Ci/mol) or ["Clformate (50 pM, 58.0 Ci/mol) for 2 h, and then a sample (50 ml) was taken, and metabolites were extracted in ice-cold 0.4 M HC10, which was then neutralized (Sant et al., 198913). Samples of the culture were removed at appropriate times relative to the addition of drug(s) (0 h, 2 h, or 8 h) after radiolabeling for the preceding 2 h.
T o determine the time-dependent effects of MTX or P T X upon the flux through the de novo purine pathway, leukemia cells were "pulse labeled" for 2 h with [I4C]glycine before extraction of metabolites at appropriate times relative to drug addition. The time-dependent effects of MTX upon the azaserine-induced accumulation of FGAR polyphosphates were determined with a similar experiment by pulse labeling cells for 2 h with ['4C]formate. The effects of MTX, PTX, and MMPR upon the azaserine-induced accumulation of FGAR HPLC-Acid-soluble metabolites were separated by gradient anion exchange HPLC on a Partisil 10-SAX column (0.42 X 22 cm, Whatman) and quantified using an LKB model 2140 diode array ultraviolet detector (Bromma, Sweden) and an LKB model 1208 radioactivity monitor (Wallac Oy, Turku, Finland) connected in series as described previously (Sant et al., 1989b).
Thin Layer Chromatography-Leukemia cell extracts (7 pl) were subjected to thin layer chromatography on polyethyleneimine-cellulose plates (1.5 X 18 cm, Machery-Nagel, Doren, West Germany) using 0.34 M NaCl as the developing solvent (Rowe et al., 1978). The chromatograms were dried, sprayed with Enhance (Du Pont-New England Nuclear), and metabolites incorporating [I4C]glycine were detected by autoradiography for 3 months. I4C-Labeled metabolites were identified by co-chromatography with appropriate marker compounds (Sant et al., 1989a).
Assay of P-Rib-PP-A modification of the method of Bokkerink et al. (1986) was used. Samples (10 ml) were taken from cell cultures at appropriate times, and the washed cell pellet was stored at -20 "C for subsequent analysis. The pellet was resuspended in 50 mM Tris. HC1,l mM EDTA (150 pl), and P-Rib-PP was extracted by sonication with a Branson Sonifier Cell Disruptor model B15 (Branson Sonic Power Company, Shelton, CT) at 50 watts for 40 s. The lysate was heated for 30 s in a boiling water bath and clarified by centrifugation. P-Rib-PP was measured in duplicate in mixtures (115 p l ) containing cell extract (100 pl), [carbo~yl-'~C]orotate (Amersham Corp.; 34.8 p~, 26.3 Ci/mol), MgCI, (8.7 mM), and orotate phosphoribosyltransferase, OMP decarboxylase (Sigma; 0.4 mg) and incubated at 37 "C for 90 min. After the addition of 4 M HC10, (50 pl), l4COZ was trapped in 100 p1 of 2 M NaOH which was then transferred to glass fiber discs pretreated with barium acetate (100 pl, 10% (w/v)) and quantified by scintillation counting in a mixture consisting of 3.0 g of 2,5-diphenyloxazole/liter of toluene.
Extraction and Assay of Amido Phosphoribosyltransferase-Mouse L1210 leukemia cells (500 ml, 9 X lo5 cells/ml) were harvested, washed twice, and the cell pellet was resuspended in an equal volume of sonication buffer (0.25 M sucrose, 1.0 mM MgCl,, 20 mM K/Hepes, pH 7.0) at 4 "C. The cells were lysed by sonication (30 watts, 50 s), and cellular debris was removed by centrifugation (10,000 X g, 30 min). The cell-free extract was concentrated 10-fold, and the sonication buffer was exchanged using a Diaflo ultrafiltration cell (Amicon Corporation, Danvers, MA). The extract was stored in small aliquots at -20 "C. Assay mixtures for amido phosphoribosyltransferase contained in a total volume of 50 pl: 50 mM K/Hepes (pH 7.2). 1.0 mM dithiothreitol, 1.0 mM MgC12, 500 p M Mg,P-Rib-PP, 1.0 mM ~-['~C]glutamine (10 Ci/mol), cell-free extract (approximately 37 pg of protein), and appropriate concentrations of a potential inhibitor.
Variations and further details of these procedures appear in figure legends. The reaction was initiated with ~-['~C]glutamine, and four samples (7 pl) were transferred to polyethyleneimine-cellulose chromatograms (1.5 X 10 cm) at appropriate times up to 40 min. Chromatograms were developed immediately by ascending chromatography with 0.28% (v/v) formic acid at 4 "C and autoradiographed overnight. Spots of ~-['~C]glutamate formed were excised and quantified by scintillation counting. P-Rib-PP used in these assays was standardized by complete reaction of limiting concentrations with excess L-['4C]glutamine. The purity of the P-Rib-PP used for the inhibition patterns of Fig. 5 was 85.1%. Analysis of Kinetic Data-Data were fitted to the appropriate velocity equation using the program DNRP53 for nonlinear regression analysis (Duggleby, 1984) with all experimental data points given equal weighting. Data obtained for inhibition by dihydrofolate ( I ) of amido phosphoribosyltransferase saturated with L-glutamine ( E ) , with P-Rib-PP (S) as the varied substrate, were fitted to the velocity equation derived from the model of Scheme 1, in which all species of the enzyme-glutamine complex are in rapid equilibrium and catalysis is the rate-limiting step. It is assumed that the active form of the enzyme is a dimer (Holmes, 1980) and that all enzyme-inhibitor complexes are catalytically inactive. The first-order rate constant for formation of PRA by one catalytic site is k, and cy, 6, and y are interaction factors.

Inhibition of Purine Biosynthesis by Antifolates
The velocity equation derived from Scheme 1 is as follows. where u, is the uninhibited reaction velocity when I = 0, and b and e are constants. When the concentration of I is low and it is assumed that the binding of I does not affect the binding of S ( p = l), then Equation 1 may be written in the form where K i is an apparent inhibition constant for the noncompetitive, initial interaction of an inhibitor ( I ) with the enzyme-glutamine complex. Data obtained for inhibition of amido phosphoribosyltransferase giving linear Dixon plots were fitted to Equation 3.

RESULTS
Growth of mouse L1210 leukemia cells in the presence of [14C ]glycine radiolabels GAR and subsequent intermediates of the de novo purine pathway as shown in Scheme 2. The incorporation of [14C]glycine into GAR, AICAR and subsequent purine nucleotides over a 2-h period, at various times relative to addition of an antifolate, gives a measure of the flux through the two transformylase reactions to ATP and GTP. The time-dependent effects of MTX (0.1 PM) upon the purine pathway in growing leukemia cells were determined by analysis of cell extracts by HPLC. Total levels of purine intermediates, derived from de novo and salvage synthesis, were determined by ultraviolet absorbance (Fig. 1, a-c) and intermediates derived from de m v o synthesis over the previous 2 h, by incorporation of [14C]glycine ( Fig. 1, d -f ) . After exposure of cells to MTX for 2 h, the HPLC elution profile at 260 nm shows significant accumulations of total SAICAR and AICAR, whereas ATP and GTP have decreased substantially (Fig. lb). The peak corresponding to IMP appears to have increased, but this is because of an accumulation of N -(p-aminobenzoy1)-L-glutamate, which has the same retention time, whereas IMP decreases (Lyons and Christopherson, 1991). After 8 h, ATP and GTP have decreased further, and SAICAR, AICAR and IMP have virtually disappeared (

Inhibition of Purine Biosynthesis by Antifolates 11041
with significant ultraviolet absorbance have been computed by integration of the peaks of Fig. 1, a-c, and are summarized in Table I. After exposure to MTX for 2 h, IMP synthesized de m u 0 from [14C]glycine has decreased, ATP and GTP have almost disappeared, and SAICAR and AICAR have accumulated (Fig. le), defining a site of inhibition at AICAR transformylase (AICAR + FAICAR, reaction 9, Scheme 2). After 8 h there is no ['4C]glycine incorporated into purine intermediates (Fig. lf), indicating that de m u 0 synthesis (Scheme 2) over the previous 2 h was completely blocked. The complete disappearance of SAICAR and AICAR synthesized de mu0 after an 8-h exposure to MTX (Fig. l f ) confirms that AICAR transformylase (reaction 9) is not the primary site of inhibition induced by MTX in the pathway (Lyons and Christopherson, 1991). GAR, the substrate for GAR transformylase (reaction 3), is not clearly separated from glycine by this HPLC procedure (Sant et al., 1989a) and was quantified from the same cell extract by thin layer chromatography (Table 11). GAR initially accumulates from a level of 750 amol/cell (approximately 560 pM) to 950 amol/ cell (710 pM) after 2 h while levels of the subsequent intermediates FGAR and AIR decrease, defining a second site of inhibition at GAR transformylase (GAR + FGAR, reaction 3, Scheme 2). Eight h after the addition of MTX, GAR has decreased to 400 amol/cell (300 p~) while FGAR and AIR continue to decrease.
A second culture of leukemia cells was exposed to azaserine (25 p~) , pulse-labeled for 2 h with ['4C]glycine, and extracts were prepared in parallel with the MTX-treated culture. Azaserine is a glutamine antagonist that acts as a potent inhibitor of FGAM synthetase (FGAR FGAM, reaction 4) resulting in the disappearance of purine nucleotides and accumulation of FGAR, FGAR-DP, and FGAR-TP to millimo- Metabolites of interest from the ultraviolet absorbance elution profiles of Fig. 1, a-c were quantified by integration of the appropriate peaks. Levels are calculated as amol/cell for 0, 2, and 8 h after the addition of MTX (0.1 p~) .
Standards for each metabolite were chromatographed, and molar amounts from cell extracts were calculated by direct ratio of peak areas with the standards. SAICAR is not commercially available, and cellular levels were calculated by direct ratio of peak areas for "C-labeled SAICAR and AICAR accumulated after 2 h (Fig. le) with the total level of AICAR from Fig. 16, assuming the specific radioactivities of SAICAR and the subsequent intermediate, AICAR, to be the same (Sant et al., 1989a). SAICAR levels for the 0-h (Fig. la) and 8-h (Fig. IC) samples were calculated by ratio with the integrated ultraviolet absorbance of.the 2 h sample (Fig. 16) 1 p~) and analyzed by HPLC (Fig. 1) were also analyzed by thin layer chromatography, and intermediates incorporating [ "C] glycine were located by autoradiography and quantified by scintillation counting. The specific radioactivities of the abundant purine nucleotides also measurable by ultraviolet absorbance (ATP, ADP, GTP, and GDP) were determined from Fig. 1, a and d, and averaged to give a value of 4.29 & 1.21 dpm/ pmol. It was assumed that the specific radioactivities of the less abundant metabolites, listed below and measurable only by radioactivity, were approximately equal to the end products of the pathway (Sant et al., 1989a)  lar cellular concentrations (Lyons et al., 1990). After exposure of cells to azaserine for 2 h, FGAR polyphosphates quantified from the incorporation of [14C]glycine (cf. Fig. le)' had accumulated from a low level of FGAR (210 amol/cell, Table 11) and undetectable levels of FGAR-DP and FGAR-TP to: FGAR, 11,100 amol/cell(8,380 p~) ; FGAR-DP, 810 amol/cell (610 p~) ;

FGAR-TP, 3,370 amol/cell (2530 p~) .
For the MTX-treated culture, the net accumulations of purine precursors after 2 h were: GAR, 200 amol/cell (150 pM, Table  11); SAICAR, 172 amol/cell (130 p~, Table I The net accumulation of purine precursors induced by MTX is only 4.9% of the accumulation of FGAR polyphosphates induced by azaserine. These observations could be explained if accumulated GAR and AICAR were degraded within leukemia cells, but they are stable when incubated with a whole lysate of mouse L1210 leukemia cells in complete RPMI 1640 medium (4.2 mg of protein/ml) or with an equivalent cell-free extract over 2 h at 37 'C.' These data indicate that during the first 2 h of exposure of cells to MTX (0.1 p~) , the flux through the de nouo purine pathway (Scheme 2) is reduced to 4.9% of the initial rate by a major blockade prior to reaction 3. A "trickle" of purine intermediates is converted through to GAR, where MTX has M. E. Sant induced a blockade at GAR transformylase. Purine intermediates, perhaps preexisting distal to GAR in the pathway, are converted through to AICAR where there is also a blockade at AICAR transformylase (Scheme 2). A primary blockade at reaction 1 induced by MTX would result in accumulation of P-Rib-PP; distal sites of inhibition at reactions 3 and 9 would result in small accumulations of GAR and AICAR with depletion of the end products of the pathway as observed ( Fig. 1; Tables I and 11).
Treatment of leukemia cells with MTX (0.1 p~) induces an accumulation of P-Rib-PP from an initial cellular concentration of 83 amol/cell(62 p~) to 240 amol/cell(l80 p~) after 2 h, 280 amol/cell (210 p~) at 3 h, and 55 amol/cell (41 p~) at 8 h (Fig. 2). Such an accumulation of P-Rib-PP has been reported (Bokkerink et al., 1986) and is consistent with potent inhibition of reaction 1 (P-Rib-PP + PRA, Scheme 2). PTX (0.1 p~) also induced the disappearance of purine intermediates synthesized de mu0 and a similar accumulation of P-Rib-PP to that for MTX (Fig. 2). The total accumulation after 2 h of purine precursors (GAR + SAICAR + AICAR) was only 3.3% of FGAR polyphosphates (FGAR + FGAR-DP + FGAR-TP). Allegra et al. (1987) have attributed total inhibition of de nouo purine biosynthesis by MTX to inhibition of AICAR transformylase by accumulated polyglutamated dihydrofolate. However, if the primary site of blockade of the purine pathway were reaction 9, an accumulation of AICAR of 84-fold would be induced by 0.1 p~ MTX after 2 h rather than the %fold observed here (Table I) 57.0 Ci/mol) and exposed to MTX (25 p~) for 8 h also showed the disappearance of the intermediates of Scheme 2 (Lyons and Christopherson, 1991). MMPR, as the 5'-monophosphate derivative, is a potent inhibitor of amido phosphoribosyltransferase which catalyzes reaction 1 of the de nouo purine pathway (Hill and Bennett, 1969;Nelson and Parks, 1972). MMPR (25 p~) induced metabolic effects similar to MTX (0.1 p M ) and PTX (0.1 p M ) with accumulation of P-Rib-PP (Fig. 2) and the disappearance of purine intermediates synthesized de muo? Lyons and Christopherson (1991) showed that MTX (25 p~) abolished the accumulation of FGAR polyphosphates induced by aza-  Fig. 3 shows data from similar experiments in which [14C]formate was added to cultures with azaserine (25 p~) and 10 p~ concentrations of MTX, PTX, or MMPR. [14C]Formate is incorporated into purine nucleotides during 2 h of radiolabeling (Fig. 3a), and azaserine completely blocks this incorporation and induces accumulation of FGAR, FGAR-DP, and FGAR-TP (Fig. 3b). The presence of MTX (10 p~, Fig. 3c), PTX (10 p~, Fig. 3d), or MMPR (10 p~, Fig. 3e) blocks the accumulation of FGAR polyphosphates confining the primary site of inhibition to the first three reactions (P-Rib-PP + PRA + GAR + FGAR). The addition of 0.1 p~ MTX also abolished the azaserineinduced accumulation of FGAR polyphosphates, but the effect was slower than with 10 p~ MTX (Fig. 4). After 2 h, FGAR, FGAR-DP, and FGAR-TP are still apparent (Fig. 4b) but are not synthesized during the 2-h pulse preceding the 8-h sample (Fig. 4c).
MMPR induces metabolic effects similar to MTX and PTX (Figs. 2 and 3), suggesting that reaction 1 (P-Rib-PP + PRA) may be the primary site of inhibition induced by antifolates. The de nouo purine pathway is subject to regulation by feedback inhibition by end products upon reaction 1. Amido phosphoribosyltransferase is synergistically inhibited by a 6hydroxy purine nucleotide (GMP or IMP) and a 6-amino purine nucleotide (AMP), with positive cooperativity with respect to the substrate P-Rib-PP being accentuated by the presence of these purine nucleotides (Holmes, 1980;Hill and Bennett, 1969;Wood and Seegmiller, 1973). The binding of AMP or GMP to murine amido phosphoribosyltransferase converts the active small form of the enzyme (Mr = 127,000) to the inactive large form (Mr = 292,000) (Holmes, 1980). were as for Fig. 1. Peaks A, E, and C are derived from the corresponding phosphorylated forms of FGAR (Lyons et al., 1990). h relative to the addition of MTX and azaserine. kemia cells exposed to MTX or PTX (0.1 p~) by metabolites that accumulate under these conditions. Inhibition of the enzyme in vitro by such metabolites would provide direct support for this proposal. However, mammalian amido phosphoribosyltransferase is an oxygen-sensitive, iron-sulfur protein, and consequent instability has precluded its purification and characterization (Itakura and Holmes, 1979). Inhibition studies were therefore performed in vitro with amido phosphoribosyltransferase extracted from the mouse L1210 leukemia cells used for the metabolic experiments described above. Although data obtained with heterogeneous enzyme preparations may be more difficult to interpret, there was no choice for these experiments.
A variety of purine and folate derivatives and antifolates was tested as inhibitors of amido phosphoribosyltransferase extracted from mouse L1210 leukemia cells using the assay procedure described under "Experimental Procedures." At 500 p M concentrations, inhibitions of greater than 25% were observed for GMP, IMP, and AMP, consistent with earlier reports (Hill and Bennett, 1969;Wood and Seegmiller, 1973); MTX-Glu5, PTX, DTBSF, dihydrofolate, dihydrofolate-Glu5, folate-Glu5, N5-formyltetrahydrofolate, and AICAR were also effective inhibitors of amido phosphoribosyltransferase. By contrast, tetrahydrofolate, folate, MTX, N-( p-aminoben-zoy1)-L-glutamate, and trimetrexate did not significantly inhibit the enzyme in vitro. Amido phosphoribosyltransferase extracted from cells and used directly for assays exhibited apparent slow binding inhibition as described by Morrison (1982), with progress curves for ~-['~C]glutamate formation turning over in the presence of an inhibitor (500 p~) to reach a lower inhibited, steady-state rate between 60 and 120 min. This time-dependent inhibition was reversible. However, enzyme that had been concentrated 10-fold by Diaflo ultrafiltration gave progress curves for ~-['~C]glutamate formation which were linear for approximately 40 min in the absence and presence of an inhibitor. Holmes (1980) proposed that amido phosphoribosyltransferase has two distinct allosteric, inhibitory sites for AMP and GMP or IMP. The spatial relationship of the inhibitory site for dihydrofolate and dihydrofolate-Glu5 to the catalytic site of the enzyme was investigated with full inhibition patterns with P-Rib-PP as the varied substrate (Fig. 5). Lines at higher inhibitor and lower P-Rib-PP concentrations show upward curvature, consistent with positive cooperativity with respect to P-Rib-PP (Hill and Bennett, 1969;Wood and Seegmiller, 1973). The inhibition patterns intersect to the left of the l / u axis, indicating that dihydrofolate and dihydrofolate-Glua bind at a site distinct from the catalytic site.
Dixon plots ( l / u versus the concentration of dihydrofolate or dihydrofolate-GluJ at constant L-glutamine (1 mM) and P-Rib-PP (500 p~) concentrations also showed upward curvature at higher concentrations, consistent with positive cooperativity for the binding of inhibitor. High concentrations of dihydrofolate (1 mM) completely inhibited enzyme activity, indicating that enzyme-dihydrofolate complexes are catalytically inactive. The data of Fig. 5 were fitted to Equation 1 which describes the model of Scheme 1 with the simplifying assumptions described under "Analysis of Kinetic Data" in "Experimental Procedures." The standard errors were large for some of the parameter values obtained, and the theoretical lines of Fig. 5 should be considered as simulations rather than fits by nonlinear regression. However, the experimental data obtained are consistent with the model of Scheme 1. For dihydrofolate (Fig. 5a), values for the interaction factors a and y are less than 1.0, indicating that the substrate P-Rib-PP and inhibitor are both bound with positive homotropic cooperativity as shown by upward curvature in Lineweaver-Burk plots (Fig. 5) and Dixon plots, respectively. The value for p of 5.4 indicates negative heterotropic cooperativity for binding of substrate or inhibitor to an enzyme complex with the other ligand (Scheme 1). The K, value for binding the first molecule of P-Rib-PP of 109 PM is decreased to 16.8 p~ (aK,) for the second interaction. The Ki for dihydrofolate decreases from 312 p~ (Ki) to 286 p~ (yKi). For dihydrofolate-Glue, values for the interaction factors a and y are again less than 1.0, whereas / 3 is greater than 1.0. The K, value for P-Rib-PP decreases from 138 to 44.6 p~ (aK,) for the second interaction. The Ki for dihydrofolate-Glu5 decreases from 16.1 With the assumption that amido phosphoribosyltransferase contains an allosteric site for folate analogues as well as for to 9.2 p M (yKi).

Inhibition of Purine Biosynthesis
by Antifolates 6-hydroxy and 6-amino purine nucleotides, Dixon plots were obtained for GMP, folate, folate-Glu5, N5-formyltetrahydrofolate, dihydrofolate, dihydrofolate-Glu5, MTX-Glu5, PTX, DTBSF, and AICAR, and the data were fitted to Equation 3. Most of these Dixon plots showed upward curvature at higher inhibitor concentrations consistent with positive cooperativity for the binding of inhibitor as described by Equation 2. However, data from the initial linear portions of these plots were fitted to Equation 3, which describes simple noncompetitive inhibition, and the values obtained are listed in Table  111. Dihydrofolate-Glu5 and PTX are the most potent (noncompetitive) inhibitors of amido phosphoribosyltransferase with apparent Ki values of 3.4 and 6.0 PM for dissociation from the free enzyme compared with dihydrofolate with a Ki of 310 pM. The Ki value of 3.4 ~L M obtained by fitting data to Equation 3 by nonlinear regression may be considered more reliable than the value used in Fig. 5b to simulate theoretical lines from Equation 1.

DISCUSSION
Data presented in this paper have necessitated a reevaluation of the mechanism by which antifolates block the de nouo biosynthesis of purine nucleotides. Four different types of experiments indicate that antifolates induce a primary blockade of the de rwuo purine pathway at reaction 1 (Scheme 2). 1) MTX and PTX abolish the azaserine-induced accumulation of FGAR polyphosphates in growing leukemia cells (Fig. 3). 2) Purine precursors such as SAICAR and AICAR synthesized de nouo from [14C]glycine disappear after an 8-h exposure to MTX (Fig. I), and the net accumulation after 2 h of such precursors is less than 5% of the FGAR polyphosphates which accumulate with azaserine. 3) Dihydrofolate-Glue, MTX-Glu5, PTX, and DTBSF inhibit amido phosphoribosyltransferase in uitro ( Fig. 5 and Table 111). 4) MMPR also induces the disappearance of purine intermediates synthesized de nouo (cf. Fig. I), a similar accumulation of P-Rib-PP (Fig. 2), and abolishes the azaserine-induced accumulation of FGAR polyphosphates (Fig. 3e). This purine nucleoside is converted to the 5'-monophosphate derivative which is an effective inhibitor of amido phosphoribosyltransferase (Nelson and Parks, 1972).

TABLE 111
Apparent dissociation constants for the initial noncompetitive interaction of inhibitors with amido phosphoribosyltransferase The enzyme was extracted from mouse L1210 leukemia cells, and assays were performed as described under "Experimental Procedures." Reaction rates were determined at 10 different concentrations of each inhibitor, and data were plotted as a Dixon plot (l/u uersus I). Amido phosphoribosyltransferase activity measured with 500 p~ P-Rib-PP was approximately 1.64 pmol/min/pg of protein, whereas in the absence of P-Rib-PP glutaminase activity was approximately 0.0817 pmol/min/pg of protein, which was subtracted from the measured rates to give u,. Data from the linear initial portion of Dixon plots (at least six inhibitor concentrations) were fitted to Equation 3 which describes noncompetitive inhibition for single catalytic and allosteric sites.

MTX-Glub
MTX does induce a &fold accumulation of AICAR (Table  I) as reported by Allegra et al. (1987), but this AICAR is probably derived only from the pools of intermediates between the two transformylase reactions (FGAR + FGAM -AIR -CAIR + SAICAR + AICAR) which existed before the MTX addition. MTX and PTX induce small accumulations of GAR, SAICAR, and AICAR derived from [14C]glycine which are inconsistent with complete blockade of the purine pathway. These experiments with growing cells indicate that antifolates induce primary inhibition of the pathway at reactions 1 or 2 (P-Rib-PP + PRA + GAR). MMPR, like MTX and PTX, abolishes the azaserine-induced accumulation of FGAR polyphosphates (Fig. 3) and induces a similar accumulation of P-Rib-PP (Fig. 2), suggesting that reaction 1 is the primary site of inhibition. PRA is very unstable (t, = 38 s at 37 "C and pH 7.5; Schendel et al., 1988), and amido phosphoribosyltransferase is therefore more easily assayed than GAR synthetase. Amido phosphoribosyltransferase was known to be the regulatory enzyme of the pathway, and we have now demonstrated potent inhibition of this enzyme by dihydrofolate-Glu5 in uitro. Dihydrofolate polyglutamates accumulate to 20% of the total folate pool in MTX-treated cells (Allegra et al., 1986). The total cellular concentration of reduced folates in mouse L1210 leukemia cells is about 10 p~ (Seither et al., 1989), suggesting that dihydrofolate polyglutamates would accumulate to approximately 2 p~ in MTXtreated cells. The fractional inhibition of amido phosphoribosyltransferase can be calculated by substitution of appropriate values into Equation 1, the velocity equation which describes the dependence of amido phosphoribosyltransferase activity upon the concentration of P-Rib-PP and dihydrofolate-Glu5, both of which are bound with positive cooperativity. At an initial cellular P-Rib-PP concentration of 56 p~ (average value from Fig. 2), an increase in cellular dihydrofolate-GlU5 from undetectable levels to 2 p~ after the MTX addition would decrease enzymic activity to 51% (Table IV). However, after 2 h, P-Rib-PP has increased to 180 p~, and Equation 1 then predicts 117% of the initial activity (Table IV). If the concentration of dihydrofolate polyglutamates after 2 h were approximately equal to the total cellular pool of reduced folates of 10 PM (Seither et al., 1989), then enzymic activity would have decreased to 16%. After 8 h, P-Rib-PP has decreased to 41 PM, and Equation l predicts 36% of the initial  activity with 2 p~ dihydrofolate-Glu5 and 4.4% with 10 pM dihydrofolate-Glu6.
Thus, inhibition of amido phosphoribosyltransferase predicted from enzyme kinetic studies in vitro does not quantitatively account for the total blockade at reaction 1 observed in growing cells. Inhibition by MTX-Glu5 (Ki = 550 p~) and AICAR (Ki = 1910 p~; Table 111) would not make significant contributions to the blockade, given their likely cellular concentrations (Table I). The more potent inhibition of amido phosphoribosyltransferase by PTX (Ki = 6.0 p~; Table 111) suggests that accumulated PTX may significantly inhibit reaction 1 at later times, but again the inhibition would not be complete as observed in culture. A similar discrepancy would exist for the established feedback inhibition of amido phosphoribosyltransferase by purine nucleoside monophosphates such as GMP (Ki = 1580 p~; Table 111). Amido phosphoribosyltransferase may be part of a multienzyme complex in intact cells (Rowe et al., 1978) which may change the kinetic properties of the enzyme, and the local concentrations of these regulatory metabolites may be higher because of their compartmentation in vivo. The data of Table I11 and Fig. 5 establish that there are strong and specific interactions of dihydrofolate-Glu5, PTX and some other folate analogues with amido phosphoribosyltransferase. Inhibition experiments with amido phosphoribosyltransferase in vitro provide evidence for an inhibitory, allosteric site that binds dihydrofolate-Glu5 and certain folate analogues. This interaction must have a physiological role in the regulation of de m v o purine nucleotide biosynthesis in normal cells. Inhibition of amido phosphoribosyltransferase by dihydrofolate polyglutamates could reduce the flux through the pathway at reaction 1 when there is insufficient N"-formyltetrahydrofolate to convert GAR + FGAR and AICAR + FAICAR. The total concentration of the cellular pool of reduced folates is about 10 p~ (Seither et al., 1989), and a decrease in tetrahydrofolate derivatives would result in an equivalent increase in dihydrofolate derivatives. The rate of reduction of dihydrofolate to tetrahydrofolate may be low because of low dihydrofolate reductase activity during particular phases of the cell cycle or levels of NADPH required for the reduction may be low in cells starved for a carbon source or oxygen. Inhibition of reaction 1 under these conditions would prevent accumulation of GAR and unnecessary consumption of P-Rib-PP, L-glutamine, glycine, and ATP.