Mechanism for the coupling of ATP hydrolysis to the conversion of 5-formyltetrahydrofolate to 5,10-methenyltetrahydrofolate.

5,10-Methenyltetrahydrofolate synthetase catalyzes the irreversible conversion of 5-formyl-tetrahydropteroylpolyglutamates (5-CHO-H4PteGlun) to 5,10-methenyltetrahydropteroylpolyglutamates (5, 10-CH+-H4PteGlun). The equilibrium of the nonenzymatic reaction, which equilibrates slowly in the absence of enzyme, greatly favors 5-CHO-H4PteGlun. The enzyme couples the reaction to the hydrolysis of ATP shifting the equilibrium to favor 5,10-CH+-H4PteGlun. Substrate-dependent non-equilibrium isotope exchange of ADP into ATP was observed, suggesting the formation of a phosphorylated intermediate of 5-CHO-H4PteGlunduring the enzyme-catalyzed reaction. The competitive inhibitor 5-formyltetrahydrohomofolate also supported the ADP to ATP exchange, suggesting that this molecule could also form a phosphorylated intermediate. The initial rates of the ADP-ATP exchange with saturating ADP were about 70 s− for both compounds, while the kcat values for product formation were 5 s− for 5-CHO-H4PteGlun and 0.005 s− for 5-formyltetrahydrohomofolate. Starting with 5-[18O]CHO-H4PteGlun, it was shown by 31P NMR that the formyl oxygen of the substrate was transferred to the product phosphate during the reaction. This further supports the existence of a phosphorylated intermediate. The formyl group of 5-CHO-H4PteGlun is known to be an equilibrium mixture of two rotamers. Stopped-flow analysis of the enzymatic reaction showed that only one of the rotamers serves as a substrate for the enzyme.

5,10-Methenyltetrahydrofolate synthetase (MS) 1 (EC 6.3.3.2), also referred to as 5-formyltetrahydrofolate cyclodehydrase, catalyzes the ATP-dependent conversion of 5-formyltetrahydrofolate (5-CHO-H 4 PteGlu n ) to 5,10-methenyltetrahydrofolate (5,10-CH ϩ -H 4 PteGlu n ) (Reaction I). The enzyme activity was first found in sheep liver acetone powder extracts by Peters and Greenberg (1) and later studied by Kay et al. and Greenberg et al. (2,3). This enzyme has been purified to homogeneity from both prokaryotic and eukaryotic sources (4 -6). The primary structure of the enzyme from rabbit liver has been determined (7). 5-CHO-H 4 PteGlu n ϩ ATP 3 5,10-CH ϩ -H 4 PteGlu n ϩ ADP ϩ P i REACTION I The origin and function of 5-CHO-H 4 PteGlu n has been a source of confusion and debate (8). The acid-catalyzed interconversion of 5-CHO-H 4 PteGlu n and 5,10-CH ϩ -H 4 PteGlu n occurs nonenzymatically with the equilibrium greatly favoring 5-CHO-H 4 PteGlu n at pH 7 (2,9). The rate of the nonenzymatic reaction occurs on the time scale of hours under physiological conditions and is complicated by 5,10-CH ϩ -H 4 PteGlu n also being hydrolyzed to 10-CHO-H 4 PteGlu n . By coupling Reaction I to ATP hydrolysis, the equilibrium is shifted to favor 5,10-CH ϩ -H 4 PteGlu n at pH 7. For many years, neither an enzymatic source nor function of 5-CHO-H 4 PteGlu n was known. This suggested that the function of MS was to serve as a salvage pathway for the reincorporation of this nonenzymatically produced folate derivative into the one-carbon pool. 5-CHO-H 4 PteGlu n has been found to occur in many cells at low concentrations (8). However, its presence in cells is suspect since the harsh procedures used to extract folates are known to convert 10-CHO-H 4 PteGlu n to 5-CHO-H 4 PteGlu n . Recently, it was shown that in Neurospora crassa conidiospores at least 85% of the folate pool is 5-CHO-H 4 PteGlu n (10). The mild extraction procedure was shown not to account for the presence of the 5-CHO-H 4 PteGlu n . Because 5-CHO-H 4 PteGlu n is the only tetrahydrofolate derivative that is stable to oxidative degradation, these studies suggested that a physiological role is to serve as a storage form of this coenzyme in dormant stages of cellular life cycles. It has also been noted that 5-CHO-H 4 PteGlu n is an inhibitor of many enzymes in one-carbon metabolism, suggesting it may also serve a regulation role (8). Making Reaction I irreversible would be beneficial to a cell whether 5-CHO-H 4 PteGlu n is formed nonenzymatically, is used as a storage form for folates, or functions in regulation by inhibiting enzymes in one-carbon metabolism.
5-CHOH 4 PteGlu, clinically known as leucovorin, has been administered to cancer patients to rescue them from toxicity during high dose methotrexate chemotherapy, or more recently to enhance the cytotoxic activity of 5-fluorouracil (11,12). MS is the only known enzyme that utilizes 5-CHO-H 4 PteGlu n as a substrate. For 5-CHO-H 4 PteGlu to be effective as a rescue agent and an enhancer of 5-fluorouracil toxicity, it must first be converted to 5,10-CH ϩ -H 4 PteGlu through MS activity. The effectiveness of leucovorin in chemotherapy is at least partially dependent on the irreversibility of Reaction I. It has been shown that doubling intracellular concentrations of 5-CHO-H 4 PteGlu n , by inhibiting MS activity, suppressed growth of MCF-7 human breast cells by 80% (13). This suggests that MS could be a potentially important enzyme as a target in chemotherapy.
In this paper, we report the mechanism of the coupling of ATP hydrolysis to the conversion of 5-CHO-H 4 PteGlu n to 5,10-CH ϩ -H 4 PteGlu n by MS. By using NMR spectroscopy and isotope exchange techniques, a 5-CHO-H 4 (14). Rabbit livers were from Pel-Freeze Biologicals (Rogers, AK). Livers were quick-frozen and shipped within 24 h on dry ice after removal from the rabbits and stored at Ϫ70°C. MS was purified from the rabbit livers as previously described (15), and the concentration was determined by using the A 278 extinction coefficient of 15 mM Ϫ1 cm Ϫ1 . 5,10-CH ϩ -H 4 PteGlu n Preparation-Either 15 mg of (6R,6S)-5-CHO-H 4 PteGlu or 5 mg of (6S)-5-CHO-H 4 PteGlu 4 were dissolved in 1 ml of H 2 O, and the pH was adjusted to 2 by addition of 6 N HCl. The conversion to 5,10-CH ϩ -H 4 PteGlu n was allowed to proceed for 15 h at 4°C. The molar absorption coefficients (⑀ 360 ϭ 25, 100 M Ϫ1 cm Ϫ1 and ⑀ 287 ϭ 31,500 M Ϫ1 cm Ϫ1 ) were used to determine the concentration of 5,10-CH ϩ -H 4 PteGlu n and 5-CHO-H 4 PteGlu n , respectively (14). The yellow crystals of 5,10-CH ϩ -H 4 PteGlu were collected by centrifugation and washed twice with 10 mM HCl, and the crystals were lyophilized and stored at Ϫ20°C. 5,10-CH ϩ -H 4 PteGlu 4 did not crystallize at pH 2 and was stored at Ϫ20°C without further purification in the 10 mM HCl solution.
Kinetic Analysis-Activity assays during MS purification were performed at 30°C in 1 ml of 50 mM KMES, pH 6.0, containing 1 mM MgATP and 300 M (6R,6S)-5-CHO-H 4 PteGlu. The reaction was initiated by the addition of 0.5 g of pure MS or appropriate aliquots of impure enzyme. The initial rate was determined by measuring the linear increase in absorbance at 360 nm in the first minute of the reaction. Studies of competitive inhibition by 5-formyltetrahydrohomofolate were conducted in the presence of 1 mM MgATP. The value of K i for 5-formyltetrahydrohomofolate was determined by varying the concentration of 5-CHO-H 4 PteGlu from 0.5 to 4 times its K m value (8 M) in the presence of 0.14 -0.56 M 5-CHO-tetrahydrohomofolate. K i values were calculated from double reciprocal plots of initial velocity versus substrate concentration (16).
Non-equilibrium Isotope Exchange-Reaction solutions (0.5 ml) containing 1 mM MgATP, 0.5 mM [ 3 H]MgADP (5 Ci), and either 2 mM (6R,6S) 5-CHO-H 4 PteGlu or 28 M 5-CHO-tetrahydrohomofolate were incubated at 30°C with 2 g of MS in 50 mM KMES, pH 6.0. 100-l aliquots were removed at 30, 60, 90, and 120 min. The protein was removed by centrifugation after heating at 100°C for 1 min. A 40-l aliquot of the supernatant was injected into a Hewlett Packard 1090 HPLC equipped with a Vydac nucleotide separation column. ADP and ATP were separated to base line resolution by isocratic elution with 0.15 M NaH 2 PO 4 , adjusted to pH 3.0 by glacial acetic acid. The fractions of ATP and ADP were collected and counted on a Packard Tri-Carb 1500. In each set of experiments, the control was always performed without the presence of the enzyme. In the study of the effect of ADP concentration on the rate of ADP conversion to ATP, 0.5 ml of the KMES buffer at pH 6.0 containing 1 mM MgATP, 2 mM (6R,6S)-5-CHO-H 4 PteGlu, and 0.25, 0.5, 1.0, or 2.0 mM [ 3 H]MgADP (0.25-2 Ci) were incubated with 2 g of MS. The reaction was allowed to proceed for 30 min at 30°C, and 100-l aliquots were removed and treated as described above.
To determine reversibility of Reaction I, a 1-ml reaction solution containing 0.5 mM [ 3 H]MgADP (2 Ci) and either 200 M 5,10-CH ϩ -H 4 PteGlu or 100 M 5,10-CH ϩ -H 4 PteGlu 4 was incubated with 7.2 g of MS in 50 mM potassium phosphate, pH 7.2, and 100-l aliquots were removed at 30, 60, and 90 min, which were treated the same as described above except that 20 l of a 2.5 mM MgATP solution was added to the extract as an ATP carrier before injection into the HPLC. O] in a 1.5-ml Eppendorf tube, and the pH was adjusted to 1 by adding 10 l of 6 N HCl. The solution was kept at 4°C, and the conversion to 5,10-CH ϩ -H 4 PteGlu 5 was monitored at 360 nm. After 48 h, 64% of the (6S)-5-CHO-H 4 PteGlu 5 had been converted to 5,10-CH ϩ -H 4 PteGlu 5 . The solution was then made 150 mM in potassium phosphate, pH 6.6, with the addition of solid K 2 HPO 4 . The solution was also made 100 mM in 2-mercaptoethanol. The sealed Eppendorf tube was then heated at 90°C for 5 h to convert the 5,10-CH ϩ -H 4 PteGlu 5 to (6S)-5-CHO-H 4 PteGlu 5 (17). The yield of (6S)-5-CHO-H 4 PteGlu 5 , calculated from the absorbance at 288 nm, was greater than 95%. This concentration was confirmed by the ability of 5-CHO-H 4 PteGlu 5 to form a quinonoid complex absorbing at 502 nm with serine hydroxymethyltransferase and glycine (17). The large amount of phosphate in the product solution was then removed by chromatography on a 0.5 ϫ 20-cm BioGel P2 column equilibrated with 5 mM NH 4 HCO 3 . The fractions with maximum absorbance at 288 nm were pooled and lyophilized to dryness. Based on the amount of 5,10-CH ϩ -H 4 PteGlu 5 formed in acid from the initial (6S)-5-CHO-H 4 PteGlu 5 , the product contained 36% 16

RESULTS
Substrate Activity of 5-CHO-Tetrahydrohomofolate-5-CHOtetrahydrohomofolate differs from 5-CHO-H 4 PteGlu by one additional methylene group between the pteridine and aminobenzoyl rings. To form the product 5,11-CH ϩ -tetrahydrohomofolate would involve the formation of a 6-membered ring rather than the 5-membered ring of 5,10-CH ϩ -H 4 PteGlu. However, 5,11-CH ϩ -tetrahydrohomofolate has been previously synthesized and shown to be an inhibitor of the enzyme 5,10-methenyltetrahydrofolate cyclohydrolase, which converts 5,10-CH ϩ -H 4 PteGlu to 10-CHO-H 4 PteGlu (18). We determined that 5,11-CH ϩ -tetrahydrohomofolate could be formed by MS catalysis from 5-CHOtetrahydrohomofolate by observing spectral changes at 316 nm, which characterizes the presence of 5,11-CH ϩ -tetrahydrohomofolate (14). Less than 10% of 5 M 5-CHO-tetrahydrohomofolate was converted to 5,11-CH ϩ -tetrahydrohomofolate after incubation with 0.6 g of MS in 20 mM KMES buffer at 30°C for 1 h. Because of the low turnover rate, it was not possible to determine an accurate K m value for this compound. However, the rate did not change with increasing substrate concentration, suggesting that the K m value was much less than 5 M. Assuming the enzyme was saturated with 5-CHO-tetrahydrohomofolate in these assays, an apparent k cat value of 0.005 s Ϫ1 was determined from the initial velocity, which is 0.1% of the k cat value of 5-CHO-H 4 PteGlu.
The affinity of 5-CHO-tetrahydrohomofolate for MS was determined by observing it as a competitive inhibitor of 5-CHO-H4PteGlu. Using only the monoglutamate form of the substrate and 5-CHO-tetrahydrohomofolate, a competitive inhibition pat-tern was observed in a double reciprocal plot of ⌬A 360 versus 5-CHO-H 4 PteGlu concentration. A K i of 0.1 M was determined. This shows that 5-CHO-tetrahydrohomofolate has a high affinity for the active site of MS but exhibits a low k cat value.
Evidence for Irreversibility of Reaction I-Reversibility of the MS-catalyzed reaction had previously been examined by observing the formation of 5-CHO-H 4 PteGlu from 5,10-CH ϩ -H 4 PteGlu. No 5-CHO-H 4 PteGlu could be detected, suggesting that the reaction was not reversible (6). However, 5,10-CH ϩ -H 4 PteGlu hydrolyzes to 10-CHO-H 4 PteGlu at the neutral pH conditions used in this experiment (19). This decreases the sensitivity of the assay procedure and does not totally rule out that the reaction is reversible. To circumvent this problem, we examined the reversibility of Reaction I by using the more sensitive incorporation of [ 3 H]ADP into ATP. The reaction solution contained 2 Ci of 0.5 mM [ 3 H]ADP, 7.2 g/ml MS, and either 200 M 5,10-CH ϩ -H 4 PteGlu or 100 M 5,10-CH ϩ -H 4 PteGlu 4 . The reaction was allowed to proceed at 30°C, and 100-l aliquots were removed after 30, 60, 90, and 120 min. The reaction was stopped by denaturing the enzyme at 100°C for 1 min, and the protein was removed by centrifugation. MgATP (50 nmol) was added as a carrier, and 40-l aliquots were analyzed by HPLC. The results showed no detectable amounts of [ 3 H]ATP.
Evidence for a Phosphorylated Intermediate-Existence of a phosphorylated intermediate was suggested using nonequilibrium isotope exchange. A K m of 8 M and a k cat of 5 s Ϫ1 for (6S)-5-CHO-H 4 PteGlu were previously reported (7). When 2 mM (6R,6S)-5-CHO-H 4 PteGlu, 1 mM MgATP, and 2 g of enzyme were used, the forward reaction proceeded at a rate which consumed about 80% of the substrate in a 2-h period. To this reaction mixture, 0.5 mM [ 3 H]MgADP (5 Ci) was included, and 100-l aliquots were removed at 30, 60, 90, and 120 min. After removing the protein by heating and centrifugation, 40 l of the supernatant from each time point were injected into a Vydac nucleotide column. This column not only separated ATP from ADP but also separated 5-CHO-H 4 PteGlu from 5,10-CH ϩ -H 4 PteGlu. This permitted the extent of the reaction to also be monitored. Fig. 1 shows the amount of [ 3 H]ATP formed in the presence of (6R,6S)-5-CHO-H 4 PteGlu. The incorporation of 3 H from ADP into ATP shows a decreasing rate with time of incubation when 5-CHO-H 4 PteGlu was used as the substrate.
The exchange experiments were also performed with increasing concentrations of MgADP at saturating levels of 5-CHO-H 4 PteGlu. The first 30 min of radiolabel incorporation into ATP were measured. The results show that the conversion of ADP to ATP depends on the initial concentration of ADP (Fig. 2). A double reciprocal plot of initial rate of [ 3 H]ATP formation versus the concentration of ADP is linear, giving an apparent K m of 1.4 mM for ADP and a k ex for the formation of ATP of 70 s Ϫ1 . A product inhibition constant of 0.3 mM for ADP has been previously determined (7).
[ 31 P]Phosphate NMR Studies-The enzymatic mechanism of coupling ATP to the conversion of substrates to products can often be determined by following the path of oxygen atoms present in the substrate (20,21,27 Rotamer Specificity-In solution, 5-CHO-H 4 PteGlu exists as two slowly interconverting rotamers arising from partial hindered rotation of the formyl-N 5 formamide bond. These two rotamers are present at a ratio of 2.35:1 at 25°C (22,23). The structures of both rotamers have previously been determined by 13 C and 1 H NMR spectroscopy. The more abundant rotamer has the formyl carbonyl oriented toward H 6 and the formyl proton oriented in the same plane as the keto group of C 4 of the pteridine ring.
If only one rotamer of 5-CHO-H 4 PteGlu serves as a substrate for MS, the presence of excess enzyme should rapidly convert this rotamer to 5,10-CH ϩ -H 4 PteGlu with conversion of the second rotamer being controlled by the much slower nonenzymatic interconversion of the two rotamers. Using stopped-flow spectrophotometry, the rate of product formation can be fol- lowed with the use of excess enzyme. Fig. 4 shows the rate of 5,10-CH ϩ -H 4 PteGlu 5 formation at 360 nm when 11.6 M MS was flowed against 11.7 M 5-CHO-H 4 PteGlu 5 and 1.9 mM MgATP at 21°C. The data were fitted to a double exponential equation as represented by the solid line. The slower rate was 0.03 s Ϫ1 and was the same when either 5-CHO-H 4 PteGlu or 5-CHO-H 4 PteGlu 5 was used as substrate. This rate is, within experimental error, the same rate of rotamer interconversion previously determined on our instrument (25). The ratio of the amplitudes for the rapid and slow phases was 2.5. This result is also very close to the equilibrium ratio (2.35:1 at 25°C) determined by NMR spectroscopy (23). These results suggest that MS uses as the substrate only the more abundant rotamer of both 5-CHO-H 4 PteGlu and 5-CHO-H 4 PteGlu 5 . This is the same rotamer bound by dihydrofolate reductase (24) and serine hydroxymethyltransferase (25). DISCUSSION The metabolite 5,10-CH ϩ -H 4 PteGlu n can be hydrolyzed to both 10-CHO-H 4 PteGlu n and 5-CHO-H 4 PteGlu n at neutral pH. Its K eq for formation of 10-CHO-H 4 PteGlu n can be readily determined experimentally, but the K eq for formation of 5-CHO-H 4 PteGlu n can only be determined under acidic conditions and is less accurate. Using the K eq experimentally determined by Kay et al. (2) for Reaction II, we calculate that the equilibrium ratio of product to reactant at pH 7.0 is 1:6.5 ϫ 10 Ϫ5 . Coupling ATP hydrolysis to Reaction II results in Reaction I 2 with a predicted K eq Ј of 66 M. A similar calculation, based on the tetrahydroquinoxaline analog of 5-CHO-H 4 PteGlu n , indicates that the K eq Ј for Reaction I would approach 500 M (2,9). Our failure to find any incorporation of labeled ADP into ATP puts an upper limit on the rate of the reverse of Reaction I of 0.08 min Ϫ1 . This is nearly 4000-fold slower than the enzyme-catalyzed rate of the forward reaction. This suggests that at pH 7.0 the equilibrium of Reaction II does not favor 5-CHO-H 4 PteGlu n to the extent predicted by the studies of Kay et al., since the predicted K eq Ј of 66 M for Reaction I would show some isotope exchange of ADP into ATP.
The studies of non-equilibrium isotope exchange of ADP and ATP with both 5-CHO-H 4 PteGlu and 5-CHO-tetrahydrohomofolate suggest that although the overall reaction is irreversible, there is a reversible step involved on the pathway to the formation of 5,10-CH ϩ -H 4 PteGlu n . Scheme I proposes a mechanism to explain the observed results. The major rotamer of the enolate form of 5-CHO-H 4 PteGlu (structure I) makes a nucleophilic attack on the ␥-phosphoryl group of ATP to form a phosphorylated 5-CHO-H 4 PteGlu (structure II, Scheme I). At this intermediate, enzyme-bound ADP can freely equilibrate with solvent ADP. Reversal of this reaction (II to I) would account for the incorporation of [ 3 H]ADP into ATP (Fig. 1). Intermediate II would be formed with both 5-CHO-H 4 PteGlu and 5-CHO-tetrahydrohomofolate. The studies with varied amounts of ADP show saturation kinetics (Fig. 2). The linear double reciprocal plot suggests a Michaelis-Menten mechanism in which the K m for ADP is 1.4 mM and the rate from ADP to ATP is about 70 s Ϫ1 . This rate is essentially the same rate determined from the study of ADP and ATP isotope exchange in the presence of 5-CHO-tetrahydrohomofolate (66 s Ϫ1 ). The rate of conversion of intermediate II to I is 10 times the k cat of 5 s Ϫ1 determined previously for Reaction I (6).
Intermediate II can proceed further in the reaction by attack of N 10 of 5-CHO-H 4 PteGlu to form a putative tetrahedral intermediate (structure III, Scheme I). This would collapse to eliminate phosphate and form the product 5,10-CH ϩ -H 4 PteGlu (structure IV). It would be this last step that may be essentially irreversible since phosphate would be a poor nucleophile in the back reaction. The value of k cat probably is determined by either the conversion of intermediate II to III or the elimination of phosphate to form product. The observation that k cat for 5-CHO-tetrahydrohomofolate is 1000-fold less than the k cat for 5-CHO-H 4 PteGlu suggests that it is the conversion of intermediate II to intermediate III. The N 11 of the homofolate analog would probably be out of optimum position to make the nucleophilic attack on the phosphorylated intermediate II.
The NMR study provides additional evidence for the formation of the phosphorylated intermediate II. The   the oxygen is transferred to phosphate. An alternative mechanism, which would be consistent with the NMR data, is that ATP forms a phosphorylated enzyme intermediate and that 5-CHO-H 4 PteGlu is phosphorylated by the phosphoenzyme. However, this mechanism suggests that the enzyme would catalyze the equilibration of [ 3 H]ADP and ATP in the absence of the 5-CHO-H 4 PteGlu. We found no evidence for this substrate-independent ADP-ATP exchange, which argues against a phosphorylated enzyme being formed.
Another folate-dependent enzyme that utilizes ATP is N 10formyltetrahydrofolate synthetase, which catalyzes the synthesis of 10-formyltetrahydrofolate from formate and tetrahydrofolate. ATP hydrolysis to ADP and phosphate is coupled to this reaction to shift the equilibrium to favor 10-CHO-H 4 PteGlu. In eukaryotic cells, this activity is part of a trifunctional enzyme, while in bacteria it is monofunctional (26). Mejillano et al. (26) demonstrated that both the prokaryotic and eukaryotic enzymes form a formyl phosphate as an intermediate. However, ADP remains bound to the enzyme so the formate-dependent ADP-ATP exchange is very slow. Schrimsher et al. (27) have shown that aminoimidazole ribonucleotide synthetase catalyzes the ATP-dependent formation of the imidazole ring in purine biosynthesis. Starting with [ 18 O]formylglycinamidine ribonucleotide, these authors showed the existence of a phosphorylated intermediate of the substrate. However, ADP-ATP exchange was only 1/20th the rate of the reaction, again probably the result of a slow release of ADP from the enzyme.
Addendum-During the review of this manuscript, a communication by Kounga et al. (28) was published, which presented similar 31 P NMR data as shown in Fig. 3.