Mechanism of 4-(beta-D-ribofuranosyl)aminobenzene 5'-phosphate synthase, a key enzyme in the methanopterin biosynthetic pathway.

The first committed step in methanopterin biosynthesis is catalyzed by 4-(beta-D-ribofuranosyl)aminobenzene 5'-phosphate (RFA-P) synthase. Unlike all known phosphoribosyltransferases, beta-RFA-P synthase catalyzes the unique formation of a C-riboside instead of an N-riboside in the condensation of p-aminobenzoic acid (pABA) and 5-phospho-alpha-D-ribosyl-1-pyrophosphate (PRPP) to produce 4-(beta-D-ribofuranosyl)aminobenzene 5'-phosphate (beta-RFA-P), CO(2), and inorganic pyrophosphate (PP(i)). Here we report the successful cloning, active overexpression in Escherichia coli, and purification of this homodimeric enzyme containing two 36.2-kDa subunits from the methanogen Methanococcus jannaschii. Steady-state initial velocity and product inhibition kinetic studies indicate an ordered Bi-Ter mechanism involving binding of PRPP, then pABA, followed by release of the products CO(2), then beta-RFA-P, and finally PP. The Michaelis parameters are as follows: K(m)pABA, 0.15 mm; K(m)PRPP, 1.50 mm; V(max), 375 nmol/min/mg; k(cat), 0.23 s(-1). CO(2) showed uncompetitive inhibition, K(i) = 0.990 mm, under varied PRPP and saturated pABA, and a mixed type of inhibition, K(1) = 1.40 mm and K = 3.800 mm, under varied pABA and saturated PRPP. RFA-P showed uncompetitive inhibition, K(i) = 0.210 mm, under varied PRPP and saturated pABA, and again uncompetitive, K(i) = 0.300 mm, under saturated PRPP and varied pABA. PP(i) exhibits competitive inhibition, K(i) = 0.320 mm, under varied PRPP and saturated pABA, and a mixed type of inhibition, K(1) = 0.60 mm and K(2) = 1.900 mm, under saturated PRPP and varied pABA. Synthase lacks any chromogenic cofactor, and the presence of pyridoxal phosphate and the mechanistically related pyruvoyl cofactors has been strictly excluded.

The first step in methanopterin biosynthesis is catalyzed by 4-(␤-D-ribofuranosyl)aminobenzene 5Ј-phosphate (␤-RFA-P) 1 synthase. This enzyme catalyzes the condensation between para-aminobenzoic acid (pABA) and 5-phospho-␣-D-ribosyl-1-pyrophosphate (PRPP) with concomitant formation of ␤-RFA-P, CO 2 , and inorganic pyrophosphate (PP i ) (1). This enzyme is a phosphoribosyltransferase and a decarboxylase and forms a C-riboside, which is unique among phosphoribosyltransferases and pABA-dependent enzymes. For example, in an early step in tetrahydrofolate biosynthesis, dihydropteroate synthase catalyzes a condensation between the amino group of pABA and dihydropterin pyrophosphate to generate dihydropteroate, eliminating PP i . Thus, ␤-RFA-P synthase and dihydropteroate synthase both use pABA as a substrate and produce PP i as product; however, the amino group is the nucleophile in dihydropteroate synthase, whereas the aromatic ring carbon 4 (C-4) is the nucleophile in ␤-RFA-P synthase (2,3).
How does RFA-P synthase generate an electrophilic center at C-1 of PRPP? How does this enzyme poise ring carbon-4 of pABA for nucleophilic attack on the C-1 of PRPP and activate this position for decarboxylation? The mechanism shown in Scheme 1 is our working hypothesis. When PRPP binds, C-1 is converted into an electrophilic center. Many PRPP-dependent enzymes utilize an S n 1 mechanism involving elimination of PP i to form an oxocarbenium ion that undergoes nucleophilic attack. Deprotonation of the amine group of pABA would promote delocalization of electron density to form a quininoid species with development of negative charge at C-4. What would stabilize this negative charge on a carbon that already contains a negatively charged carboxyl group? It was suggested that ␤-RFA-P synthase contains PLP, which is known to stabilize similar intermediates during decarboxylation or deamination of amino acids, and the partially purified enzyme appeared to contain a chromophore with properties similar to those of PLP (3). However, we show here that ␤-RFA-P synthase lacks PLP and apparently any other cofactor that would be capable of stabilizing negative charge developing during the reaction. Thus, all the interesting chemistry performed in this reaction appears to be promoted by the enzyme per se.
The study of ␤-RFA-P synthase is significant because it is an early step in the biosynthesis of tetrahydromethanopterin (H 4 MPT), which is a modified folate that is of central importance in growth and energy metabolism of methanogens. H 4 MPT is involved in multiple steps in methane formation as in one carbon reactions involved in amino acid and nucleotide metabolism. Even though H 4 MPT is found in Archaea and one class of Bacterium (e.g. Methylobacterium extorquens), the biosynthetic pathway for these two folates (folate and methanopterin) is different suggesting that they play different functional roles in physiology of the cell (4,5). We are targeting this enzyme to inhibit specifically methanogenesis (6), which has the potential impact of reducing the levels of the greenhouse gas methane, which has doubled over the last 200 years.
In this paper, we report the cloning, overexpression, purification, and characterization of the steady-state mechanism of ␤-RFA-P synthase from the hyperthermophilic methanogenic * This work was supported by National Institutes of Health Grant R41 GM67952-01. 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  archaeon Methanococcus jannaschii and show that it lacks cofactors. We propose a mechanism for this reaction based on product inhibition studies and a rationale for how pABA is activated to perform a nucleophilic attack on PRPP. The reaction is believed to proceed via an oxycarbenium intermediate formed from PRPP that forms an adduct with the C-4 of pABA (Scheme 1).

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA), PfuTurbo® DNA polymerase was from Stratagene (La Jolla, CA), and PRPP, CaPP i , and NaHCO 3 were purchased from Sigma, and [ 14 C]pABA was from ICN Pharmaceuticals Inc. (Costa Mesa, CA). Genomic DNA from M. jannaschii was purchased from ATCC (Manassas, VA). All reagents were purchased at the highest quality available.
Enzymatic Synthesis and Purification of ␤-RFA-P-␤-RFA-P was produced enzymatically by incubating the ␤-RFA-P synthase (1 g of protein) with substrates for 3 h at 70°C and isolated using a Shodex AXpak WA-624 (6 ϫ 150 mm) high pressure liquid chromatography column developed first with a linear gradient from solution A to B for 30 min and then washed for 10 min with solution A at a flow rate of 1 ml/min. Solution A contained 0.1 M ammonium formate, pH 3.0/acetonitrile (90:10), and solution B was composed of 0.5 M ammonium formate, pH 3.0/acetonitrile (80:20). The detector was set at 260 nm, and the chromatography was performed at room temperature (22°C). ␤-RFA-P eluted at a retention time of 12.5 min, while the retention times for pABA and PRPP were 22 and 26 min, respectively. ␤-RFA-P was identified and quantified with the Bratton-Marshall assay (3). In the control reaction lacking enzyme, no peak was observed at 12.5 min. The fractions containing ␤-RFA-P were collected and lyophilized, and the powder was stored at room temperature for further usage. To check the stability and purity of the ␤-RFA-P, the powder was dissolved in solution A and run by high pressure liquid chromatography, and only a single sharp peak was observed at 12.5 min.
Cloning and Expression of the RFA-P Synthase Gene-A BLAST search revealed that the M. jannaschii ␤-RFA-P synthase was likely to be encoded by gene MJ1427 (4). We amplified this gene from M. jannaschii chromosomal DNA by the PCR using Pfu polymerase (Stratagene, La Jolla, CA). The commercially synthesized primers were 5Ј-C-ACCATGATAATTCAAACACCATCAAGGA-3Ј and 5Ј-TCACCAAATTT-TATGCCCCACATTATT-3Ј for protein overexpressed with a His tag at the N terminus. The PCR product was separated by gel electrophoresis on a 1% agarose gel, and the appropriate band (1 kb in length) was extracted using a gel extraction kit (Qiagen, Valencia, CA.). The gene sequences were verified by the Genomics Core Research Facility at The University of Nebraska-Lincoln. For expression, the gene was cloned into pET200 (Invitrogen), a Champion TM pET directional TOPO expression plasmid, and transformed into Escherichia coli TOP10 cells. For overexpression, the plasmid was introduced into E. coli Rosetta (DE3)pLysS cells. The cells were grown in 1 liter of Luria-Bertani broth at 15°C to an absorbance at 600 nm of 0.4 -0.6. Protein expression was induced by adding 1 mM isopropyl-␤-thiogalactopyranoside. After 4 h, the cells were harvested by centrifugation at 10,000 ϫ g for 25 min.
The primers used to amplify MJ1427 without a His tag were 5Ј-CA-CCTGAAGAAGGAGAATACATATGATAATTCAAACACCATC-3Ј and 5Ј-TCACCAAATTTTATGCCCCACATTATT-3Ј. The protocol used for cloning and overexpression was identical to the one employed for the gene with a tag at the N terminus. For the tag-less enzyme, the yield was lower (30%) than that for the His-tagged protein. For overexpression, E. coli Rosetta (DE3)pLysS strain was used because it provided a higher yield because of the presence of rare tRNAs required by MJ1427. The pure enzyme had a specific activity ranging from 200 to 250 nmol/min/mg protein.
Purification of ␤-RFA-P Synthase-␤-RFA-P synthase from M. jannaschii was purified to homogeneity by column chromatography and heat treatment. Cells were suspended in lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10 mM imidazole, pH 8.0) and disrupted by sonication. The cell lysate was centrifuged at 15,000 ϫ g for 2 h. The supernatant was loaded on a Ni-NTA column, which was washed with a buffer solution containing 50 mM NaH 2 PO 4 , 300 mM NaCl, and 20 mM imidazole, pH 8.0, to remove proteins that are nonspecifically attached to the resin. The His-tagged protein was eluted with buffer containing 50 mM NaH 2 PO 4 , 300 mM NaCl, and 150 mM imidazole, pH 8.0. Fractions with active enzyme were further purified by heat treatment at 70°C in a water bath for 15 min. The turbid solution was again centrifuged at 10,000 ϫ g for 30 min, and the supernatant was stored at Ϫ80°C until needed.
Size exclusion chromatography was performed using Sephadex G-50 equilibrated with 50 mM Tris/HCl, pH 7.6, 10 mM MgCl 2 , and the protein was eluted at a flow rate of 0.1 ml/min. UV-visible spectra of purified ␤-RFA-P synthase (5 mg/ml in 1 ml of 50 mM Tris-HCl, pH 7.6, 300 mM NaCl, 2 mM DTT, and 10 mM MgCl 2 ) were recorded using a Beckman DU 7400 spectrophotometer.
Assay for PLP and Pyruvoyl Prosthetic Group-The protocol for removal of PLP from the enzyme involved treating the enzyme with 5 mM hydroxylamine for 24 and 72 h as described previously (7). To assay for pyruvate, HCl was added to adjust the pH of the enzyme solution to ϳ3.0. Then the solution was boiled for 1 h, and the pH was adjusted to pH 8.0, and lactate dehydrogenase and NADH were added to measure the amount of pyruvate formed by using the lactate dehydrogenase assay. The total volume of the assay was 1 ml in 50 mM Tris/HCl, pH 8.0, and the reaction was followed spectrophotometrically at 340 nm for 5 min (8).
Assay for ␤-RFA-P Synthase and Protein Determination-Protein concentrations were determined by the Bradford assay with bovine serum albumin as standard (9). Proteins were separated by SDS-PAGE and stained with Coomassie Blue. ␤-RFA-P synthase activity was measured by following the release of 14 CO 2 from p-aminobenzoic acid with a radioactive label in the carboxyl group essentially as described earlier (6).
Steady-state Kinetics of ␤-RFA-P Synthase-Stock solutions of pABA (2.93, 5.86, 14.6, 29.3, 58.66, or 88 mM), 25 mM PP i , 10 mM NaHCO 3 were used. The CO 2 concentration was estimated by the Henderson-Hasselbach equation taking into account the pH of the reaction mixture and the NaHCO 3 concentration. The assays were performed at 70°C for 3 h and were initiated by adding enzyme. In all the assays, the buffer was 100 mM TES, 25 mM MgCl 2 , pH 4.8. To determine the Michaelis parameters of ␤-RFA-P synthase for pABA, the reaction mixture contained PRPP (1, 2.5, 4, or 10 mM) and varying concentrations of pABA (0, 0.05, 0.1, 0.2, 0.5, and 1-3 mM) in 100 mM TES, 25 mM MgCl 2 , pH 4.8.
Product Inhibition Studies-To determine the type of inhibition and the inhibition constants for CO 2 , PP i , and RFAP, the assays were performed at fixed concentrations for each product and variable concentrations of pABA under saturated and unsaturated conditions with respect to PRPP and varied PRPP concentrations under saturated and unsaturated concentrations for pABA. In one set of studies, RFAP (0.2 and 1 mM) was used under saturating pABA (3 mM) and varied PRPP (0, 0.5, 1, 2, 5, and 8 mM). In another set of studies, RFAP (0.1 and 0.25 mM) was used at subsaturating concentrations of pABA (0.1 mM) and varied PRPP. A third set of studies included saturating PRPP (8 mM) and varied pABA (0, 0.05, 0.1, 0.5, and 1-3 mM) and RFAP (0.36, 0.9, and 1.8 mM). For CO 2 and PP i the same concentrations of pABA and PRPP were used but at different concentrations of PP i (1 and 3 mM under saturated and unsaturated PRPP and saturated and unsaturated pABA, respectively), and for CO 2 the following concentrations were used: 2, 4, and 6 mM for both saturated and unsaturated conditions for both substrates.
Analysis of Kinetic Data-The V max , K m , and K i values were determined by globally fitting the data to the nonlinear steady-state equations for bisubstrate sequential (Equation 1) and ping-pong (Equation SCHEME 1. Possible mechanism of reaction followed by ␤-RFA-P synthase. 2) mechanisms. The data were analyzed using Sigma Plot 7.0 (Jandel Scientific, San Rafael, CA). As described under "Results," the fits to a sequential mechanism uniformly gave much lower standard errors than to a ping-pong mechanism. Product inhibition data were fit to the nonlinear equations for competitive, uncompetitive, and mixed modes of inhibition (10).
Isothermal Titration Calorimetry Experiments-For ITC measurements, ␤-RFA-P synthase was dialyzed against a buffer solution containing 50 mM Tris, pH 7.6, 100 mM NaCl prior to analysis. All ITC measurements were carried out using a Microcal VP-ITC isothermal titration calorimeter (Microcal). A 30 M solution of ␤-RFA-P synthase diluted in dialysis buffer was added to the sample cell (about 1.4 ml), and a 0.5-1 mM solution of PRPP and PP i titrant was loaded into the injection syringe. For each titration experiment, a 60-s delay at the start of the experiment was followed by 35 injections of 7.5 l of the titrant solution, spaced apart by 240 s. The sample cell was stirred at 300 rpm throughout and maintained at a temperature of 35°C. Control titrations were performed by titrating titrant solutions into dialysis buffer. Titration data were analyzed using the Origin 5.0 software supplied by Microcal. Data sets were corrected for base-line heats of dilutions from control runs as appropriate. The corrected data were then fit to a theoretical titration curve describing one binding site per titrant. The area under each peak of the resultant heat profile was integrated and plotted against the molar ratio of ␤-RFA-P synthase protein to titrant. A nonlinear best fit binding isotherm for the data was used to calculate the dissociation constant.
Three-dimensional Modeling-The three-dimensional structural prediction for ␤-RFA-P synthase was generated by submitting the amino acid sequence to the three-dimensional-PSSM (threading) server located at www.sbg.bio.ic.ac.uk/ϳ3dpssm/ (13). The best hit had an E value of 1.66e Ϫ08 , and the threading had been done based on this hit. Most interestingly, ␤-RFA-P synthase shares 21% amino acid sequence identity and folding patterns to homoserine kinase (best hit) rather than known N-riboside phosphoribosyltransferases.

RESULTS
The Bratton-Marshall assay (3) that was previously used to measure ␤-RFA-P synthase activity is quite cumbersome. Therefore, we have developed a rapid assay for the enzyme in which p-aminobenzoic acid with a radioactive label in the carboxyl group is reacted with PRPP. This assay facilitates purification and kinetic studies and could easily be adapted for rapid quench kinetic studies.
Cloning, Overexpression, and Purification of ␤-RFA-P Synthase from M. jannaschii-By using the methods described under "Experimental Procedures," we successfully cloned and overexpressed the ␤-RFA-P synthase from M. jannaschii (MJ1427) gene in E. coli.
␤-RFA-P synthase was purified to homogeneity by using a combination of column chromatography and heat treatment (Table I and Fig. 1). The specific activity for the enzyme in the cell extract of induced cells was 10.0 nmol of ␤-RFA-P produced per min/mg of protein, in agreement with the enzyme from Archaeoglobus fulgidus (4). Overexpression of the protein was probed by using anti-His tag antibodies and antibodies raised against the pure protein, which stained a band around 37 kDa, indicating that this is the protein of interest.
The cell extract was loaded onto an Ni-NTA column, which was washed with 10 column volumes of Wash buffer (see under "Experimental Procedures"). ␤-RFA-P synthase (26 mg with a specific activity of 82.5 nmol min Ϫ1 mg Ϫ1 ) was eluted with a solution (Elution buffer) containing 150 mM imidazole. The active fractions were further purified by heat treatment at 70°C for 15 min, which precipitated most E. coli proteins and yielded a pure protein (10 mg of protein per 1 liter of culture medium) with a specific activity of 248 nmol min Ϫ1 mg Ϫ1 .
The ␤-RFA-P synthase gene also has been cloned with a His tag at the C terminus (using a pQE60 expression vector from Qiagen, Valencia, CA) and with no tag (using the same vector as for the gene cloned with a His tag at N terminus), and the specific activities of the pure proteins were indistinguishable, suggesting that the His tag has no influence in the activity of the enzyme. However, in both cases, the levels of protein were lower than that those observed with the gene cloned into the pET 200 expression vector with a tag at N terminus (data not shown).
The recombinant enzyme from M. jannaschii has an optimum pH around 4.9, which is similar to the pH activity profiles of the enzymes from A. fulgidus and Methanosarcina thermophila (2,3). The molecular mass of the native enzyme was estimated by gel filtration on a Sephadex G-50 molecular exclusion column to be 70 kDa, indicating that the protein is a homodimer containing two 36.2-kDa monomers (data not shown).
␤-RFA-P Synthase Does Not Contain PLP or a Pyruvoyl Moiety as Cofactors to Perform Catalysis-The homogeneously purified and fully active enzyme was used to determine whether or not it contains any cofactors to perform catalysis. As described under the "Discussion," it was proposed that ␤-RFA-P synthase might contain PLP (3), which is reasonable given the need to stabilize the build up of charge on the C-4 of pABA during the reaction. However, the UV-visible spectrum of the pure fully active M. jannaschii enzyme at concentration up to 50 M showed no UV-visible or detectable peaks between 300 and 500 nm (Fig. 2), indicating the absence of PLP. If the enzyme sample shown in Fig. 2 contained PLP, there would be an absorption band at ϳ350 nm with an absorbance of 0.2. The  pure enzyme also lacks the characteristic fluorescence spectrum of PLP. Hydroxylamine is routinely used to liberate bound PLP as an oxime that should be separated from the protein by Centricon filtration. However, hydroxylamine treatment for up to 4 days at 4°C does not result in any loss of activity, and incubation of enzyme with PLP does not affect activity (Table II). Because the data just described clearly indicate that PLP is not a cofactor for RFAP synthase, it is possible that another cofactor that does not exhibit an UV-visible spectrum is present (Fig. 2). Instead of PLP, several amino acid decarboxylases contain pyruvate as a covalently bound, catalytically active prosthetic group. However, unlike other pyruvoyl enzymes, ␤-RFA-P synthase does not require auto-processing; the size of the purified protein is consistent with that determined from the gene sequence. Furthermore, the direct lactate dehydrogenase assay for release of bound pyruvate did not indicate any conversion of NADH to NAD ϩ . In Table III, the potential 30 M covalently bound pyruvate that would have been liberated (because 30 M of protein had been used) would have resulted in a change of absorbance at 340 nM because of oxidation of NADH to NAD of 0.19. However, the lack of any change in absorbance of the solution after adding the lactate dehydrogenase reaction mixture strongly indicates that ␤-RFA-P synthase does not contain a pyruvoyl cofactor. Apparently, this enzyme is free of cofactors and prosthetic groups, and the amino acids themselves provide all the catalytic groups that stabilize the oxycarbenium from PRPP and the development of negative charge at C-4 of pABA.
Steady-state Kinetics for ␤-RFA-P Synthase-When the steady-state kinetic data were globally fit to both ordered sequential and ping-pong mechanisms, the initial velocity kinetic experiments clearly support the sequential ternary complex mechanism and are inconsistent with a ping-pong mechanism. The values for the maximal velocity and Michaelis constants for pABA and PRPP are as follows: K m pABA ϭ 150 Ϯ 10 M, K m PRPP ϭ 1600 Ϯ 300 M, and V max ϭ 180 nmol min Ϫ1 mg Ϫ1 (k cat ϭ 0.23 s Ϫ1 ) ( Fig. 3 and Table IV). Fits to a ping-pong mechanism are clearly inferior, based on the statistics shown in Table IV. Given that there is active discussion about whether some PRPP enzymes utilize ping-pong or ternary complex mechanisms (12), we also used isotope exchange methods to further test the steady-state mechanism. In a ping-pong kinetic model, the first product is released before the second substrate binds, forming a modified enzyme intermediate. Such a mechanism would predict the ability of the enzyme to catalyze two independent exchange reactions at rates equal to or greater than the overall reaction rate, namely a PRPP-independent exchange reaction between [CO 2 ] and the [ 14 C]carboxyl group of pABA and a pABA-independent exchange reaction between [ 32 P]PP i and PRPP. We were unable to observe any CO 2 /pABA exchange (see "Experimental Procedures") when up to 250 g of ␤-RFA-P synthase was incubated with 10 mM [CO 2 ] and 3 mM [ 14 C]pABA for 3 h. These results strongly support a ternary complex mechanism.
Product Inhibition Studies-Initial velocity studies in the absence of products cannot provide information about the order of substrate binding or whether the reaction is random or ordered, thus product inhibition studies of the ␤-RFA-P-catalyzed reaction were conducted (Table IV). The pattern of inhibition could be unambiguously assigned on the basis of the quality of the statistical fit of the data to the appropriate inhibition equations in all cases. Table IV lists the patterns of product inhibition predicted for ping-pong and Ordered Bi Ter mechanisms. Although there is an apparent inconsistency in the pattern of inhibition by PP i at saturated PRPP and varied pABA, the data closely match the expected patterns expected for an Ordered Bi Ter mechanism and are inconsistent with a ping-pong or Random mechanism. The product inhibition data also strongly indicate that PRPP is the first substrate that binds followed by pABA and that CO 2 is released first followed by RFA-P and finally PP i .
The prediction of the Ordered Bi-Ter kinetic model we have proposed is that only the first substrate to bind, PRPP, and the  a Following treatment with hydroxylamine, ␤-RFA-P synthase was dialyzed for 2 days as described under "Experimental Procedures." b The assays were conducted in absence or presence of added PLP, and the specific activities represent the average of two experiments.

TABLE III
Lactate dehydrogenase assay to quantify the amount of pyruvate release The activity of lactate dehydrogenase was 100 units/mg. The total volume of the assay was 1 ml, and the buffer used was 50 mM Tris, pH 8.0. The experiments with enzyme have been repeated three times with protein from three different preparations, and the specific activity of the enzyme before acidic hydrolysis was 250 nmol/min/mg. The assays were run for 5 min. To test the proposed mechanism further, isothermal titration calorimetry experiments were used to measure the dissociation constants for the predicted first substrate (PRPP) and last product released (PP i ) ( Table V). As expected, according to the just-described order of substrate/product binding and release, PRPP (K D ϭ 2000 Ϯ 150 M) and PP i (K D ϭ 200 Ϯ 20 M) bind to the free enzyme, whereas we were unable to find evidence for binding of pABA to the free enzyme at concentrations up to 1 mM. These results are consistent with the prediction of an Ordered mechanism in which PRPP binds first and PP i dissociates last and are inconsistent with a ping-pong mechanism or a Random Sequential mechanism.

DISCUSSION
The ␤-RFA-P synthase (MJ1427) from M. jannaschii has been actively overexpressed at a level of about 10 mg per liter of culture medium, which is sufficient for these and further mechanistic studies. Overexpression of this biosynthetic enzyme is important because it is in very low abundance in methanogenic cells; for example, the M. thermophila enzyme required a herculean 2,800-fold purification (4). We focused on this particular enzyme because M. jannaschii is a hyperthermophile, which often makes the proteins more stable than those of mesophilic or even moderately thermophilic orga-

TABLE IV
Results of product inhibition studies as compared with patterns predicted for kinetic models Based on the data the following assignments have been made: A is PRPP; B is pABA; P is CO 2 ; Q is ␤-RFA-P; R is PP i ; UC is uncompetitive; MT is mixed type; C is competitive inhibition. Rsqr designates R 2 , the coefficient of determination, where 1.0 is a perfect correspondence between the data and the model. nisms. This property can become valuable in structural studies of an enzyme. The only other ␤-RFA-P synthase that has been actively overexpressed so far is from A. fulgidus. Although this organism is an archaeon, it is not a methanogen, and because one of our goals is to develop mechanism-based inhibitors to inhibit methanogenesis, we felt it was important to perform mechanistic studies on the enzyme from an organism that is as closely related as possible to ruminant methanogens. Correspondingly, we showed that one of the best inhibitors of the RFA-P synthases from Methanothermobacter marburgensis and the ruminant mesophilic methanogen Methanobrevibacter smithii (6) is equally effective with the M. jannaschii protein (data not shown). Thus, we expect that this particular enzyme will provide a useful model system for future structure-function studies of this unusual reaction. Perhaps the inability to express the M. jannaschii protein earlier (4) derives from the use of a TTG start codon, which could prevent E. coli from initiating translation, because TTG encodes a tryptophan instead of a methionine residue. This substitution of ATG by TTG is fairly common in Archaea (11). In order to express this gene in E. coli, we mutated this codon to an ATG for recognition by the E. coli translational machinery. Based on SDS-PAGE analysis, the purified protein exhibited a single band at 37 kDa and eluted in the molecular exclusion column at the position expected for a 70-kDa protein, indicating that the enzyme in solution is a dimer of identical subunits. Based on three-dimensional modeling, it seems that the ␤-RFA-P synthase shares folding patterns with the GHMP kinase superfamily, which includes proteins that phosphorylate galactose, homoserine, mevalonate, and phosphomevalonate, rather than known and more functionally related Nriboside phosphoribosyltransferases, suggesting that N-and C-riboside proteases are more unrelated than initially believed. Based on the threading it seems that the monomer is composed of two domains with the active site located between the two domains (Fig. 4). By comparison with the homoserine kinaseactive site (with ADP bound in the active site), we propose that the active site of ␤-RFA-P synthase is located in a cleft between the domains. This would be consistent with a conformational change during catalysis (see below). However, the model has to be validated by crystal structure and mutational analysis which are under way.
␤-RFA-P synthase catalyzes the first committed step in the biosynthesis of the essential methanogenic cofactor, methanopterin. This is an unusual PRPP-and pABA-dependent reaction (Scheme 1). Unlike most pABA-dependent reactions, in which the amino group is a nucleophile, the C-4 group of the benzene ring of pABA is the nucleophile. The RFA-P synthase-catalyzed condensation of pABA and PRPP is unique among known phosphoribosyltransferases in that a decarboxylation of the substrate (pABA) occurs and a C-riboside is formed instead of an N-riboside. Given this unusual chemistry, one might expect the participation of a cofactor(s) to stabilize the build up of negative charge on the C-4 group of pABA and/or perhaps to promote formation of an oxocarbenium on PRPP. Correspondingly, based on inactivation of the partially purified ␤-RFAP synthase from M. thermophila by sodium borohydride, the measurement of a fluorescence spectrum characteristic of reduced PLP, and the labeling of a protein the size of the synthase by radioactively labeled sodium borohydride, it was proposed that ␤-RFA-P synthase may contain PLP (3). Furthermore, ␤-RFAP synthase is inhibited by pyridoxol phosphate, an analog of PLP (3), and nonenzymatic decarboxylation of pABA by PLP has been observed (16). However, ␤-RFAP synthase is not inhibited by cysteine or by several carbonyl reactive reagents known to inactivate PLP-dependent enzymes (3). In addition, the partially purified ␤-RFAP synthase from A. fulgidus showed no UV-visible absorbance at wavelengths characteristic of pyridoxal phosphate and a consensus pyridoxal phosphate-binding motif was not found in the amino acid sequence of ␤-RFAP synthase. Thus, it was concluded that the PLP content remained ambiguous and that "further biochemical analysis will be necessary to determine whether pyridoxal phosphate is involved in the mechanism of ␤-RFAP synthase" (4).
Here we have shown that the homogeneous fully active recombinant enzyme expressed in E. coli is unaffected by hydroxylamine under conditions that completely inactivate PLP-dependent cystathionine ␤-synthase (7) and lacks the characteristic UV-visible and fluorescence spectra of PLP-dependent enzymes. If the enzyme contained PLP, there should have been an absorption band centered at ϳ 350 nm with an absorbance of 0.2 (usually 10% of the intensity of the peak at 280 nm) in Fig. 2. Thus, we are confident that ␤-RFA-P synthase does not contain any cofactor with an absorption spectrum in the region between 250 and 600 nm. One might hypothesize that the enzyme uses another nonchromogenic cofactor with a role similar to PLP, such as a bound pyruvoyl cofactor, to perform catalysis; however, the inability to detect released pyruvate and lack of any evidence for processing of the enzyme strongly suggest that pyruvate is not formed in the active site of the enzyme. This information strongly indicates that this enzyme lacks any cofactor and focuses our attention FIG. 4. Three-dimensional modeling of ␤-RFA-P synthase. The threading program used to generate the modeling is located at www.sbg.bio.ic.ac.uk/ϳ3dpssm/. The structure is generated by comparison with homoserine kinase (Protein Data Bank accession number 1FWK) from M. jannaschii. It is proposed that the enzyme is composed of two domains, and the active site is located in the cleft between the domains. ADP, located in the homoserine kinase structure, is shown in the active site. For molecular visualization the program used was University of California, San Francisco Chimera (www. cgl.ucsf.edu/chimera/). on catalytic mechanisms involving only the polypeptide backbone and its side chains.
Kinetic data summarized in Tables IV and V demonstrate that ␤-RFA-P synthase uses an Ordered Bi-Ter reaction mechanism with PRPP binding first, then pABA, followed by the ordered release of CO 2 , then ␤-RFA-P, and finally PP i . The following experimental observations are consistent with the proposed kinetic model. Given that the enzyme uses a Bi-Ter mechanism, why is PRPP required for decarboxylation of pABA? As in many PRPPdependent enzymes (11), one possibility is that binding of the second substrate (pABA) induces a structural/conformational change that would bring the two substrates into proximity and the appropriate catalytic groups into position to induce decarboxylation and C-C bond formation. One important question is whether or not an oxocarbenium ion is formed, as in other phosphoribosyltransferases (11). If so, because the oxocarbenium would be one of the first intermediates in the reaction, PP i must become trapped in a cavity in the protein after it is cleaved from PRPP, because PP i is the last product released during the mechanism. One of the likely candidates for binding PP i is Mg 2ϩ , which is required for ␤-RFA-P synthase activity (6) as well as that of other PRPP-dependent enzymes, such as the hypoxanthine-guanine phosphoribosyltransferases (17,18). Retaining PP i throughout the mechanism could be mechanistically important, because in other PRPP-dependent enzymes, the positive charge on the ribosyl group, which becomes a ribooxocarbenium ion, is stabilized by interactions with the oxygen atoms of the released PP i (19) Another mechanistic feature of the ␤-RFA-P synthase reaction that may be shared with other PRPP-dependent enzymes is the placement of the attacking nucleophile directly above the carbenium carbon, coordinated with formation of the intermediate. This is important because the oxocarbenium could potentially react with other groups in the active site. Furthermore, the structure of pABA in the transition state must be quite unstable. That the enzyme is inhibited by a number of N-alkyl benzoates, but not aniline or the methyl ester of pABA, indicates that the carboxylic acid moiety of pABA must be ionized for the reaction to occur, creating the apparent requirement for developing a negative charge at a carbon adjacent to a negatively charged carboxylate moiety. The need for coordinated development of the reactive oxocarbenium intermediate and movement of the nucleophilic center on pABA above the C-1 position on the ribosyl group suggests that a conformational change might accompany binding of the substrates. The need for a conformational change is a feature of other phosphoribosyltransferases (12,14) and is supported by our threedimensional model, which places the active site in the cleft between the two lobes (or domains) so that closure of the two lobes could bring pABA and PRPP in close proximity in order to promote catalysis and to avoid unnecessary hydrolysis of PRPP. These movements could also place PP i in a position ideally suited for charge stabilization of the oxocarbenium, but distant (or sterically occluded) enough to prevent nucleophilic attack of PP i on the ribooxocarbenium C-1, which would reform PP i . Overall, the bipolar transition state proposed in Scheme 1, with the pABA ring carbon nucleophile building up negative charge and the reactive positively charged ribo-oxocarbenium ion, would require energetic stabilization via both intra-molecular and enzyme-transition state interactions. How the enzyme accomplishes this in the absence of cofactors is the subject of further studies.
Several other key questions remain to be answered. The role of Mg 2ϩ in catalysis, not addressed here, may be to either assist in PRPP hydrolysis and/or to influence chemical steps. Although the formation of an oxycarbenium is the most attractive working hypothesis, it is important to test for this intermediate by primary and secondary kinetic isotope effects studies.

CONCLUSIONS
The gene encoding ␤-RFA-P synthase from a hyperthermophilic methanogen has been cloned and actively overexpressed, and its protein product has been purified to homogeneity. By catalyzing the first committed step in the biosynthesis of the essential cofactor methanopterin, this enzyme is a target for development of inhibitors of methanogenesis. The enzyme follows a strictly ordered Bi-Ter mechanism, which would support a reaction mechanism (Scheme 1) in which a negatively charged aromatic ring carbon 4 from pABA acts as a nucleophile to attack a positively charged ribo-oxocarbenium ion from PRPP to form a C-riboside. ␤-RFA-P synthase was shown to lack any chromogenic cofactor, and the presence of PLP and the mechanistically related pyruvoyl cofactor was strictly excluded. Product inhibition studies indicate an ordered Bi-Ter mechanism, which ensures that both substrates are present at the active site before either substrate undergoes chemical reaction. The details of how this enzyme accomplishes such a remarkable and unique reaction in the apparent absence of cofactors remain to be elucidated. The work also serves as a mechanistic underpinning for the development of inhibitors that may be used eventually to reduce methanogenesis in various anaerobic environments.