A Methanocaldococcus jannaschii Archaeal Signature Gene Encodes for a 5-Formaminoimidazole-4-carboxamide-1- (cid:1) - D -ribofuranosyl 5 (cid:1) Monophosphate Synthetase A NEW ENZYME IN PURINE BIOSYNTHESIS*

We have identified and characterized a new member of the ATP-grasp enzyme family that catalyzes the ATP- and formate-dependent formylation of 5-aminoimidazole-4-car-boxamide-1- (cid:1) - D -ribofuranosyl 5 (cid:1) -monophosphate (AICAR) to 5-formaminoimidazole-4-carboxamide-1- (cid:1) - D -ribofurano-syl 5 (cid:1) -monophosphate (FAICAR) in the absence of folates. The enzyme, which we designate as PurP, is the product of the Methanocaldococcus jannaschii purP gene (MJ0136), which is a signature gene for Archaea. As is characteristic of reactions catalyzed by this family of enzymes, the other products of the reaction, ADP and P i , were produced stoi- chiometrically with the amount of ATP, formate, and AICAR used. Formyl phosphate was found to substitute for ATP and formate in the reaction, yet the methylene analog, phosphonoacetaldehyde, was not an inhibitor or substrate for the reaction. The enzyme, along with PurO, which catalyzes the cyclization of FAICAR to inosine 5 (cid:1) -monophos-phate, catalyzes the same overall transformation in purine biosynthesis as is accomplished by

The conversion of 5-aminoimidazole-4-carboxamide-1-␤-D-ribofuranosyl 5Ј-monophosphate (AICAR) 1 to inosine 5Ј-monophosphate (IMP) represents the last steps in the de novo bio-synthesis of purines (Fig. 1). In bacteria and eukaryotes, this reaction occurs in two separate steps catalyzed by a single bifunctional enzyme, aminoimidazole carboxamide ribonucleotide transformylase/inosine monophosphate cyclohydrolase (ATIC or PurH), encoded by the purH gene. AICAR transformylase activity resides in the carboxyl terminus of the enzyme, which transfers the formyl group from 10-formyltetrahydrofolate to AICAR, resulting in the formation of 5-formaminoimidazole-4-carboxamide-1-␤-D-ribofuranosyl 5Ј-monophosphate (FAICAR). The product of this first reaction is then cyclized by the IMP cyclohydrolase activity that resides in the amino terminus of the enzyme (1). The two-domain structure of PurH has recently been confirmed by the x-ray crystal structure of the avian enzyme (2).
Until recently, PurH was assumed to function in Archaea, just as in bacteria and eukaryotes. However, new data have led some investigators to question this assumption in some species of Archaea. These data include the following: the observation that N 10 -formyl-tetrahydrofolate is not required for this reaction in Archaea (3), genomic DNA sequence comparisons that indicate the absence of purH genes in Archaea (4 -6), and the identification of purO (MJ0626) that encodes a new IMP cyclohydrolase that cyclizes FAICAR to IMP in Methanocaldococcus jannaschii (7). Our attempts to identify the AICAR transformylase gene by searching for genes close to purO in other genomes have not proved fruitful, due in part to the low abundance of purO in the presently sequenced genomes. Attempts at isolation and purification have proved futile. We thus tested recombinant M. jannaschii proteins predicted to catalyze an analogous reaction in order to find the enzyme catalyzing this reaction. This analogous reaction is one catalyzed by ATP-grasp enzymes, several of which are present in the M. jannaschii genome.
The 5-phosphoribosylglycinamide (GAR) to 5-phosphoribosyl-N-formylglycinamide reaction in purine biosynthesis is conducted by two different routes, each with its own unique enzyme. In one route, catalyzed by GAR transformylase N, N 10 -formyltetrahydrofolate is the substrate and formyl donor; in the other route, catalyzed by GAR transformylase T, ATP and formate serve as substrates (8). This latter route, as well as those catalyzed by phosphoribosylaminoglycine ligase (PurD) and phosphoribosylaminoimidazole carboxylase (PurK), employs members of the ATP-grasp superfamily of enzymes for catalysis (9). M. jannaschii contains 12 genes annotated as ATP-grasp enzymes (10), and based on the comparison with GAR transformylase N and GAR transformylase T, we considered one of them could be the enzyme. We cloned and overexpressed several of these genes and individually tested each for its ability to catalyze the transformylase reaction using ATP and formate as substrates. Only the protein derived from the MJ0136 gene, which we designate as purP, catalyzed the desired reaction. Here we present the data on the characterization of this enzyme and its reaction.

EXPERIMENTAL PROCEDURES
Chemicals-Sodium hydrogencarbonate (99% 13 C) and sodium salt of 3-(trimethylsilyl)propionic acid-2,2,3,3-2 H 4 were purchased from Aldrich Chemical Co. Phosphonoacetaldehyde was a gift from Debra Dunaway-Mariano (University of New Mexico). All other reagents and synthetic precursors were purchased from Sigma unless otherwise specified. PurO was prepared by recombinant methods as described previously (7).
Synthesis of Formyl Phosphate-Formyl phosphate was prepared by the formylation of K 2 HPO 4 in water with acetic-formic anhydride, as described recently by Gill et al. (11) and based on a earlier method by Stadtman (12). The method is simpler than that using formyl fluoride (13). Acetic-formic anhydride was formed by heating a mixture of 0.2 ml of 100% formate and 0.4 ml of acetic anhydride at 50°C for 15 min. After cooling to 0°C, the reaction product was added into 3.0 ml of 2.3 M K 2 HPO 4 with stirring. After 30 min of stirring at 0°C, colorless crystals formed. Acetone (3 ml) was added, and the crystals were separated by filtration and washed three times with 3-ml portions of acetone, dried under vacuum, and stored in the freezer at Ϫ20°C in sealed tubes. The 1 H NMR spectrum of the isolated product in D 2 O, 1 h after solution in D 2 O at room temperature, showed a doublet (␦ 8.4, J H3 P ϭ 3.4 Hz) from the formyl phosphate and a singlet (␦ 8.14) from the formate. The area of formyl phosphate signal was 5.8% of the area of formate resonance signal. These resonances are the same as those reported earlier for formyl phosphate synthesized using formyl fluoride (13). Formyl phosphate in the isolated product was assayed by its conversion into formyl hydroxamic acid, quantitated by formation of its ferric complex (12). This analysis showed, based on dry weight, that only 4 -5% of the isolated product was the potassium salt of formyl phosphate. The formyl phosphate concentrations used in the incubations were calculated based on this latter measurement.
Cloning and Expression of M. jannaschii MJ0136 -The M. jannaschii gene at loci MJ0136 (Swiss-Prot accession number Q57600) was amplified by PCR from genomic DNA using oligonucleotide primers (Invitrogen). The primers used were MJ0136-Fwd (5Ј-GGTGGTCATAT- GATTTCAAAAGATGAG-3Ј) and MJ0136-Rev (5Ј-GATCGGATCCT-TATGAAATAATCTTATC-3Ј). PCR amplification was performed as described previously (14) using a 50°C annealing temperature. The PCR product was purified by using a QIAquick spin column (Qiagen, Valencia, CA), digested with NdeI and BamHI restriction enzymes, and then ligated into plasmid pET17b to make the recombinant plasmid pMJ0136. DNA sequence was verified by dye terminator sequencing at Davis Sequencing, Inc. (Davis, CA). The resulting plasmid was transformed into Escherichia coli strain BL21-Codon Plus (DE3)-PIL (Stratagene), and heterologous expression of the MJ0136 protein was performed by standard methods that included induction with lactose (14). The cells were harvested by centrifugation (4000 ϫ g, 5 min) and frozen at Ϫ20°C.
Purification of the Recombinant MJ0136 Protein from E. coli-Induction of the desired protein was confirmed by SDS-PAGE (12% T, 4% C acrylamide, using a Tris/glycine buffer system) analysis of total cellular proteins. The desired protein was recovered from transformed E. coli cells by sonication in the assay buffer plus dithiothreitol (0.2 M TES/K ϩ , 10 mM Mg 2 , and 20 mM dithiothreitol) and purified by heat treatment as described previously (15). The optimum temperature/heating time used in the purification was established to be 70°C for 10 min. The enzyme was purified further by chromatography on MonoQ as described previously (15). It eluted at a NaCl concentration of ϳ0.5 M. SDS-PAGE with Coomassie Blue staining showed the enzyme was essentially pure. Protein concentrations were determined using the Bio-Rad Protein Assay with bovine serum albumin as the standard.
Measurement of the Native Molecular Weight of PurP-Size exclusion chromatography was performed at room temperature on a Superose 12HR column (1 ϫ 30 cm; Amersham Biosciences) equilibrated with a buffer of 50 mM HEPES/NaOH and 0.15 M NaCl at pH 7.2. PurP in a volume of 160 l was applied to the column and eluted at a flow rate of 0.4 ml/min. Protein standards (horse spleen apoferritin, potato ␤-amylase, yeast alcohol dehydrogenase, hen egg white, conalbumin, bovine erythrocyte carbonic anhydrase, and horse heart cytochrome c) were used to calibrate the column as described previously (16).
Standard Assay for the Measurement of PurP Enzymatic Activity-The Bratton-Marshall assay was used to measure the loss of AICAR as it was converted to FAICAR in a fixed timed assay (17). The standard incubation mixture consisted of 50 l of assay buffer (0.2 M TES/K ϩ and 10 mM Mg 2ϩ buffer at pH 7.3) and 5 l of each of the following solutions: 0.01 M AICAR, 0.1 M ATP, 0.1 M sodium formate, and PurP. Typically 5 l of a diluted sample of a 2.9 mg/ml stock solution of protein containing 0.58 g of protein was sufficient to convert ϳ50% of the substrate to product in 15 min. The solution was incubated in 0.5-ml disposable centrifuge tubes in a water bath at 70°C for 15 min and centrifuged (14,000 ϫ g) for 10 s to reconsolidate the small amount of water evaporated from the sample on the sides of the tube during heating. The remaining AICAR was assayed at room temperature by adding 20 l of the incubated mixture to 400 l of 0.25 M HCl, followed by the addition of 100 l of a 0.22 M aqueous solution of sodium nitrite. The solution was thoroughly mixed. After 1 min, 100 l of 0.66 M aqueous solution of ammonium sulfamate was added, and the solution was mixed. After 2 min, 300 l of a 0.04 M solution of N-(1-naphthyl)ethylenediamine dihydrochloride was added to the solution. Color developed completely in less than 1 min and did not change significantly over time. The absorbance of the assayed solution was measured at 540 nm using a Shimadzu UV-1601 spectrophotometer. The operational extinction coefficient for AICAR was 2.05 ϫ 10 4 cm Ϫ1 M Ϫ1 . Acetate and propionate were tested as alternate substrates for the reaction by their substitution for formate at the same concentration as formate.
Confirmation of Reaction Products Generated by PurP Incubated with AICAR, Formate, and ATP-UV spectroscopic, high pressure liquid chromatography, and NMR methods were used to establish the amount and nature of the products generated after incubation of PurP with AICAR, formate, and ATP. In the first method, FAICAR was converted into IMP by PurO, and IMP was confirmed by UV spectroscopy. Two samples were prepared in a 350-l standard incubation assay buffer and contained 0.7 mol of ATP, 0.35 mol of formate, and 0.21 mol of AICAR; one sample also contained PurP (14 g). Both samples were incubated for 15 min at 70°C. After this initial incubation, PurO (ϳ25 g) was added to both samples, and the incubation was continued for an additional 15 min at 70°C. UV difference spectra were used to obtain the spectrum of the product. Chromatography of the sample on MonoQ (19) before and after PurO incubation showed FAICAR before incubation and IMP after incubation.
The second method recorded the 1 H NMR and 31 P NMR of the reactants and products before and after the incubation period. For both experiments, 350 l of standard incubation assay buffer was mixed with 35 l of 1 M sodium formate, 17.5 l of 1 M ATP, 105 l of 1 M AICAR, and 0.1 mg of MonoQ-purified PurP and then incubated at 70°C in a water bath for 15 min. Under these conditions, ϳ80% of the AICAR was converted into product, based on the observed changes in the concentrations of the reactants and products. Increasing the incubation time to 40 min led to no increase in the amount of product generated. 31 P NMR analysis of the resulting sample was then used to determine the nature of the phosphorylated products. For the 1 H NMR spectra, the water in the sample was removed with a stream of nitrogen gas, and the sample was then dissolved in deuterated water containing the sodium salt of 3-(trimethylsilyl)propionic acid-2,2,3,3-2 H 4 . 1 H NMR spectra were recorded on a Unity-400 spectrometer, and 31 P NMR spectra were recorded on an Inova-400 spectrometer. 1 H NMR of AICAR showed identifying resonances for the H-2 of the imidazole (singlet ␦ 7.52) and a doublet for the anomeric hydrogen of the ribose (␦ 5.66, J H-1Ј3 H-2Ј ϭ 7.2 Hz). Additional resonances (␦ 4 -4.5) for the other ribose hydrogens were also observed. After incubation of AICAR with PurP, formate, and ATP, the intensity of these identifying resonances for AICAR and formate decreased, and they were replaced by resonances for the formamide resonance of FAICAR (␦ 8.41), H-2 of the imidazole (␦ 8.21), and a doublet for the anomeric hydrogen of the ribose (␦ 5.64, J H-1Ј3 H-2Ј ϭ 7.2 Hz). These chemical shifts are the same as Mueller and Benkovic report (18), after correction for their use of dioxane as the internal reference. Mueller and Benkovic (18) report two resonances for the formamide hydrogen (␦ 8.18 and ␦ 7.98). They conclude this splitting arises from two different conformations of the formamide with respect to the rest of the ribonucleotide and results from restricted rotation of the formamide. To confirm the peak assignment for the formamide resonance, the incubation was repeated, substituting [ 13 C]formate. As expected, the singlet for the formamide (␦ 8.41) split into a doublet (J ϭ 207 Hz). We have not established a reasonable explanation for the differences. FAICAR was also prepared as Flaks et al. describe (17) and was purified by MonoQ chromatography. The resulting product gave the same 1 H NMR spectrum as the enzymatically produced compound.
Testing Formyl Phosphate, Phosphonoacetaldehyde, and Acetyl Phosphate-Samples of formyl phosphate were dissolved in the standard incubation assay buffer and added to the standard assay mixture, without the addition of ATP and formate, at a calculated initial concentration of 1.5 mM. The assay was then performed as described earlier.
Phosphonoacetaldehyde, a methylene analog of formyl phosphate, was incubated under the standard assay conditions at concentrations of 0.2, 2, and 20 mM to test whether it inhibited the enzymatic reaction. Acetyl phosphate was tested in the standard assay at a concentration of 14 mM.
Temperature Stability of the Enzyme-Samples of PurP (0.58 g) in 55 l of assay buffer were incubated at 60°C, 70°C, 80°C, 90°C, 100°C, 110°C, and 120°C for 15 min and then cooled to room temperature. Each sample was then assayed for remaining enzymatic activity using the standard assay. The amount of protein selected for this experiment corresponded to the amount of protein producing a linear assay from the 70°C heated sample.
Nucleotide and Metal Ion Specificity of the Enzyme-The nucleotide specificity of PurP was determined by conducting the standard assay and substituting GTP, ITP, UTP, CTP ADP, pyrophosphate (PP i ), and tripolyphosphate (PPP i ) for ATP. The effect of different metal ions on the conversion of AICAR to FAICAR was determined by preparing the assay buffer with 10 mM concentrations of the following metal salts: MgCl 2 , MnCl 2 , BaCl 2 , CuCl 2 , NiCl 2 , ZnCl 2 , and ZnSO 4 . The standard assay was then run with each of the buffers. The amount of PurP was adjusted so that only 50% of the sample was converted into product with MgCl 2 .
pH Optimum-The effect of pH on PurP activity was studied using two different buffer systems, each containing 10 mM MgCl 2 . In the pH range of 3.5-5.5, a 0.1 M sodium acetate buffer was used; in the pH range of 5.5-9.5, a three-component buffer system containing 75 mM bis-Tris, 38 mM HEPPS, and 38 mM CHES (adjusted to pH 5.5-9.5 using either NaOH or HCl) was used (20).
Product Inhibition-To test whether the reaction products inhibited the reaction, the standard assay was run with ATP, formate, and AICAR for 15 min, and then additional AICAR was added to bring the concentration back to the initial assay concentration, followed by continued incubation for an additional 15 min. The amount of AICAR lost was established after each incubation. An additional experiment containing the reaction product ADP at the same concentration of ATP and in addition to ATP was also conducted.
Measurement of Formate Production from CO 2 and H 2 -To 100 l of M. jannaschii cell extract (21) was added 10 l of 10 mM sodium 2-bromoethanesulfonate in anaerobic water and 20 l of 1 M NaH 13 CO 3 , pH 7.0, in anaerobic TES buffer (50 mM TES/K ϩ and 10 mM MgCl 2 at pH 7.5), and the mixture was incubated for 30 min at 70°C under hydrogen (28 p.s.i.). The 2-bromoethanesulfonate was included in the incubation mixture to inhibit the conversion of formate to methane (22)(23)(24). At the end of the incubation, the mixture was cooled to room temperature, and 10 l of 8 M perchloric acid was added with mixing. The resulting sample was centrifuged (14,000 ϫ g, 10 min), and the yellow extract was removed and brought to pH 6.8 by the addition of 6 M NaOH. To this solution was added 100 l of 0.5 M Na 2 HPO 4 at pH 6.8 and 0.5 ml of 100 mM 2,3,4,5,6-pentafluorobenzyl bromide in acetone. The yellow sample was both shaken and heated at 60°C for 14 h in a water bath and, after cooling, was extracted with n-hexane (1 ml). The hexane layer was separated and evaporated to 50 l with a stream of dry nitrogen gas prior to GC-MS analysis. Portions of this procedure were adopted from the previously published method for measuring formate in blood (25). GC-MS analysis of the samples used a VG-70 -70EHF gas chromatography-mass spectrometer operating at 70 eV and equipped with a Varian VF-5MS column (0.32 mm ϫ 30 m) programmed from 60°C to 260°C at 4°C/min. Under these GC-MS conditions, the 2,3,4,5,6-pentafluorobenzyl derivatives of the following compounds had the following retention times (min:s; shown in parentheses) and mass spectral data Quantitation of the formate was based on the ratios of the areas of the separated pentafluorobenzyl formate to pentafluorobenzyl bromide signals obtained by the GC-MS analysis of known samples of formate in the cell extract buffer.
Measurement of Formate in the Cells-To establish the amount of natural formate in the M. jannaschii cells, 290 mg (wet weight) of cells (78.3 mg, dry weight) was mixed with 0.6 ml of 1 M HCl, and the sample was heated for 1 min at 100°C and then centrifuged (14,000 ϫ g, 30 min), and the yellow extract was separated. The resulting pellet was re-extracted two times with water (0.4 ml) following the same procedure. The combined extracts were evaporated to dryness with a stream of nitrogen gas and placed in 0.5 ml of 2 H 2 O, which was then adjusted to pH 7.0 with NaOD. The sodium salt of 3-(trimethylsilyl)propionic acid-2,2,3,3-2 H (0.43 mol) was used as both the chemical shift reference and the internal standard to quantitate the amount of the formate present in the extract. 1 H NMR spectrum was obtained on a Unity-400 spectrometer. The measured formate concentration in the 1 H NMR sample was 0.34 mM formate in the water.

RESULTS
Testing Different ATP-grasp Enzymes-The predicted ATPgrasp enzymes derived from the M. jannaschii genes MJ0136, MJ0776, MJ0815, and MJ0106 were recombinantly expressed in E. coli and tested in the standard assay for their ability to catalyze the formylation of AICAR. Only the MJ0136 protein (PurP) was effective. This enzyme catalyzed the linear loss of AICAR when incubated with ATP and formate (Fig. 2). Under the standard assay conditions, the MonoQ-purified PurP had a specific activity of 1.9 mol/min/mg.
Products Generated after Incubation of AICAR, ATP, and Formate with PurP-The incubation of a mixture of AICAR, ATP, and formate with PurP resulted in the loss of the arylamine of AICAR, confirmed by its inability to undergo the diazotization reaction (17). Because the formamide product would not undergo such a reaction, this strongly indicated that it was the amino group of the arylamine that was formylated during the reaction. Based on its sequence, the MJ0136-derived enzyme, PurP, is considered to be a member of the ATP-grasp family of enzymes (9,26); the other products of the reaction would be expected to be ADP and P i . This reaction course was confirmed by the examination of the reaction products by NMR. 1 H NMR confirmed the production of FAICAR with a spectrum identical to that produced by a sample of the synthetically produced FAICAR. 31 P NMR spectroscopy of an incubated mixture of equal parts AICAR, ATP, and formate with PurP showed the conversion of 80% of the substrates to products. This was confirmed by the identification of 20% of the original ATP by its phosphate resonances consisting of two doublets (␦ Ϫ5.58 and J P␥3 P␤ ϭ 15.9 Hz, ␦ Ϫ10.72 and J P␣3 P␤ ϭ 16.5 Hz) for the ␣ and ␥ phosphates, respectively; a triplet (␦ Ϫ19.16) for the ␤ phosphate of the triphosphate; and a singlet (␦ 3.18) representing 20% of the original AICAR. The products ADP, FAICAR, and P i , corresponding to 80% of the starting material, were identified by two doublets (␦ Ϫ10.26 and J P␣3 P␤ ϭ 18.0 Hz, ␦ Ϫ6.21 and J P␤3 P␣ ϭ 18.3 Hz) for the ADP ␣ and ␤ phosphates, respectively; a singlet each for the orthophosphate (␦ 1.75); and the FAICAR (␦ 3.35). Longer incubation times failed to increase the extent of the conversion of substrates to products. The production of FAICAR was also confirmed by its cyclization to IMP, catalyzed by PurO. IMP was confirmed by its absorbance spectra ( max ϭ 260 nm) and by chromatography on MonoQ, where it eluted at the same retention volume as a known sample of IMP (19).
Alternate Substrates-PurP was found to specifically require formate as its acyl substrate. No activity was observed when acetate or propionate was substituted for formate in the standard assay. Formyl phosphate, but not acetyl phosphate, could substitute for formate and ATP as an alternate substrate for the enzyme. Incubation with formyl phosphate (100 M), replacing ATP and formate, produced one-third the amount of FAICAR as seen in the standard assay. Based on 31 P NMR, 1 H NMR, and direct analysis of the reaction mixtures, ATP and formate converted ϳ80% of the AICAR to FAICAR. No other nucleotide, when substituted for ATP, could function as a substrate. ADP actually inhibited the FIG. 2. Linear loss of AICAR upon incubation with PurP (0.58 g), ATP, and formate. The standard assay was incubated for the indicated times. Based on these data, the enzyme had a specific activity of 1.9 mol/min/mg protein.
reaction by 25% when added to the standard assay at the same concentration as ATP. Phosphonoacetaldehyde, at concentrations as high as 20 mM, showed no stimulation or inhibition of AICAR loss in the standard reaction.
pH Optimum, Metal Ion Dependence on Activity, and Temperature Stability of the Enzyme-The enzyme showed maximum activity at pH 6.25 and still maintained 20% of its maximum activity at pH 4.0 and pH 8.5 (Fig. 3). In the metal ion dependence study, Mg 2ϩ was the best divalent metal in supporting the reaction, with Mn 2ϩ being 87% as active as Mg 2ϩ (Table I). Other metal ions tested supported 20 -40% of the activity of Mg 2ϩ . The enzyme maintained 100% of activity when heated for 15 min at temperatures up to 100°C, but it lost all activity when heated above 100°C for 15 min. 3 -1 H NMR analysis of formate in cell extracts of M. jannaschii showed that the cells contained ϳ0.85 mM formate, assuming that 70% of the dry weight of the cell pellet was intracellular water. Formate recovered from an incubation of cell extract with NaH 13 CO 3 under a hydrogen atmosphere and, in the presence of bromoethanesulfonate, was found to contain 88% 13 C. This observed labeling not only confirmed that the bicarbonate could serve as a carbon source for the formate but also confirmed that the cell extract itself contained formate because 12% of the formate was not labeled with 13 C. In the incubation mixture described, 0.46 mM formate was generated in the 30-min incubation. This corresponded to a production of 3.7 nmol formate/min/mg protein in the cell extract.

Identification and Quantitation of Formate in Cell Extracts and Determination of the Precursor of Formate in Cell Extracts of M. jannaschii Incubated with H 2 and HCO
Product Inhibition-As the reaction progressed beyond 50% completion, its rate began to slow down (Fig. 2). From the NMR experiments, where the incubation was continued for 40 min, the maximum extent of the reaction was never over ϳ80%. Because the reaction is expected to go to completion in the presence of excess formate and ATP, this result indicated that the products of the reaction inhibited the reaction. To confirm this observation, the standard assay was run with ATP, and AICAR for 15 min, and then an additional amount of AICAR was added to bring the concentration of AICAR back to the original concentration in the sample, followed by continued incubation for an additional 15 min. In the first incubation, 80% of the AICAR was consumed and converted into FAICAR, whereas after the second addition, only 10% of the AICAR was consumed. Control experiments showed that the observed loss of activity was not due to heating PurP for such an extended period of time. This result confirmed that FAICAR and/or ADP inhibited the reaction. No inhibition was observed with IMP. DISCUSSION The above results demonstrated that the MJ0136 gene product, PurP, catalyzed the formylation of AICAR to FAICAR in the presence of ATP and formate. Thus, PurP and PurO (7) catalyze the last two steps in purine biosynthesis of M. jannaschii and explain the absence of a gene for purH in this organism (27), as well as in many Archaea. Although PurP is one of the archaeal signature genes (28), the gene for PurO is not; thus, there must be another unknown gene(s) functioning in this capacity in other Archaea.
The data presented here indicate that the mechanism of this enzyme is analogous to that of other ATP-grasp enzymes and proceeds through a bound acyl-phosphate intermediate formed at the active site of the enzyme. The involvement of an acylphosphate intermediate was confirmed by demonstrating that formyl phosphate could serve as an alternate substrate for the PurP reaction. Formyl phosphate has been implicated as an enzyme-bound intermediate in several ATP-grasp enzymes that use formate. These include formyltetrahydrofolate synthe-  3. pH activity profile of PurP. In the pH range of 3.5-5.5, a 0.1 M sodium acetate buffer was used, and in the pH range of 5.5-9.5, a three-component buffer system containing 75 mM bis-Tris, 38 mM HEPPS, and 38 mM CHES adjusted to pH 5.5-9.5 using either NaOH or HCl was used (20). The concentrations of substrates and enzyme were the same as those in the standard assay.
tase (13,29), which uses ATP and formate to generate the N 10 -formyltetrahydrofolate required in the purH step of purine biosynthesis, and GAR transformylase (PurT) (30), which uses ATP and formate to generate 5-phosphoribosyl-N-formylglycinamide in the third step of purine biosynthesis. In each case, the data indicate that bound formyl phosphate is generated from formate and ATP during the first phase of the catalytic cycle. The bound formyl phosphate, without dissociation from the enzyme, then formylates the substrate. Attempts to observe enzymatic activity of PurP with acetyl phosphate, acetate, or propionate and ATP showed no activity, indicating that PurP is very specific for formate.
Considering that the reported half-life of formyl phosphate is 48 min in buffered neutral pH solutions at 20°C, it would have a calculated half-life of only ϳ1.5 min at 70°C at neutral pH. This short half-life is all the more reason for it to function as a bound intermediate. This indicates that the active site of this enzyme must have evolved to protect this unstable carboxylic acid-phosphate anhydride intermediate from hydrolysis, as would also be expected with GAR transformylase (MJ1486) present in M. jannaschii and other thermophiles.
One thing that must be considered in evaluating the function of this enzyme, as well as GAR transformylase, in archaeal purine metabolism is the biochemical origin of the formate. The required formate must be metabolically produced in this atrophic Archaea that is known not to uptake formate (31). The data presented here demonstrate that formate can arise by hydrogen-dependent reduction of bicarbonate and that formate is already present at millimolar concentrations as free formate in the cells. This reaction in cell extracts is consistent with its expected forward direction under the substrate and product concentrations found in methanogens (32). Several different formate dehydrogenases have been described (33)(34)(35)(36), some of which require electron carriers to couple electron flow from hydrogen. The most straightforward enzyme system would be a formate hydrogenase lyase system that could be composed of a hydrogenase-3-type Ni,Fe-hydrogenase (37) coupled directly with a formate dehydrogenase. This enzymatic activity to form formate from CO 2 and H 2 was described some time ago (38,39). Baron and Ferry (40) have shown that the F 420 -hydrogenase and formate dehydrogenase of Methanobacterium formicium can be reconstituted to generate an active system capable of generating formate from CO 2 and H 2 . The enzyme for formate dehydrogenase in M. jannaschii is complex and would appear to be composed of four proteins, whereas in M. formicium, only two subunits are required. Two possible E. coli hydrogenase-3type Ni,Fe-hydrogenase (Ech) are present in M. jannaschii, and either of these could serve as the electron donor to the formate hydrogenase lyase system.
An additional recently recognized route for the generation of formate for the PurP reaction would be by the hydrolysis of formyl-methanofuran. Recently, it was observed that Methylobacterium extorquens AM1 formyl-transferase/hydrolase complex (Fhc) could generate formate by hydrolysis of formylmethanofuran (41). Sequence analysis of the FhcA subunit of the complex showed that the first 60 amino-terminal amino acid residues have sequence identity of about 40% with putative dihydroorotases and hydratoinases (42) deduced from genomic sequences from various organisms and could account for the observed hydrolysis. These sequences are not present in the methanogenic FhcA subunits, and the enzyme from Methanosarcina barkeri was found not to catalyze this reaction (41). It must be concluded that the analogous enzyme from M. jannaschii will also not catalyze the hydrolysis of formylmethanofuran and that such a hydrolysis is not the source of the formate.
Two possible reasons for the involvement of PurP and PurT in archeal purine biosynthesis can be proposed. The first is that the use of formate and ATP for the formation of FAICAR from AICAR represents a more primitive route for purine biosynthesis. By using both the PurP and PurT enzymes for purine biosynthesis, these cells do not have to depend on the presence of a C 1 carrier cofactor and thus have a simpler modular evolution of the purine biosynthetic pathway (43). The other explanation is that when the methanogens Archaea shifted to using methanopterin in place of folate, which cannot function as a N 10 -formyltetrahydromethanopterin derivative (44) required for the PurT and PurH reactions, they had to find another route to IMP. A problem with this idea, however, is that the Halobacteria contain folate as well as PurP. Although PurP is present in all Archaea, PurO, which catalyzes the cyclization of FAICAR, has only a limited distribution in Archaea. It would be interesting to establish the nature of the enzymes that catalyze this reaction in Archaea without PurO.