Biosynthesis of Riboflavin in Archaea Studies on the Mechanism of 3,4-Dihydroxy-2-butanone-4-phosphate Synthase of Methanococcus jannaschii *

The hypothetical protein predicted by the open reading frame MJ0055 of Methanococcus jannaschii was expressed in a recombinant Escherichia coli strain under the control of a synthetic gene optimized for translation in an eubacterial host. The recombinant protein catalyzes the formation of the riboflavin precursor 3,4-dihydroxy-2-butanone 4-phosphate from ribulose 5-phosphate at a rate of 174 nmol mg−1min−1 at 37 °C. The homodimeric 51.6-kDa protein requires divalent metal ions, preferentially magnesium, for activity. The reaction involves an intramolecular skeletal rearrangement as shown by 13C NMR spectroscopy using [U-13C5]ribulose 5-phosphate as substrate. A cluster of charged amino acid residues comprising arginine 25, glutamates 26 and 28, and aspartates 21 and 30 is essential for catalytic activity. Histidine 164 and glutamate 185 were also shown to be essential for catalytic activity.

The hypothetical protein predicted by the open reading frame MJ0055 of Methanococcus jannaschii was expressed in a recombinant Escherichia coli strain under the control of a synthetic gene optimized for translation in an eubacterial host. The recombinant protein catalyzes the formation of the riboflavin precursor 3,4-dihydroxy-2-butanone 4-phosphate from ribulose 5-phosphate at a rate of 174 nmol mg ؊1 min ؊1 at 37°C. The homodimeric 51.6-kDa protein requires divalent metal ions, preferentially magnesium, for activity. The reaction involves an intramolecular skeletal rearrangement as shown by 13 C NMR spectroscopy using [U-13 C 5 ]ribulose 5-phosphate as substrate. A cluster of charged amino acid residues comprising arginine 25, glutamates 26 and 28, and aspartates 21 and 30 is essential for catalytic activity. Histidine 164 and glutamate 185 were also shown to be essential for catalytic activity.
The biosynthesis of riboflavin in Archaea has not been studied in detail. An in vivo study with Methanobacterium thermoautotrophicum using [U-13 C 2 ]acetate and [1-13 C 1 ]pyruvate as tracers indicated that the xylene ring of the vitamin is assembled from two 4-carbon units (29). It was also shown that compound 2 serves as an intermediate in the biosynthetic pathways of riboflavin as well as the 5-deazaflavin (compound 7) derivative, coenzyme F 420 (30).
Strains and Plasmids-Bacterial strains and plasmids used in this study are summarized in Table I.
Restriction Enzyme Digestion of DNA-DNA was digested at 37°C with restriction enzymes in reaction buffers specified by the supplier. The treated DNA was analyzed by horizontal electrophoresis in 0.8 -3% agarose gels.
Estimation of Protein Concentration-Protein concentration was estimated by a dye binding assay (36) or photometrically (⑀ 280 nm ϭ 10600 mants were selected on LB solid medium supplemented with ampicillin (150 mg/l). The plasmids were reisolated and analyzed by restriction analysis and by DNA sequencing. In all expression plasmids, the genes coding for ribB and for mutated ribB of M. jannaschii are under the control of a T5 promoter and a lac operator.
Gene Synthesis and Construction of an Expression Plasmid-A synthetic 109-bp oligonucleotide was prepared by PCR with the overlapping oligonucleotides MJ-MUT-1 and MJ-MUT-2. In a sequence of eight PCR amplifications, the oligonucleotides listed in Table II were used pairwise starting with MJ-MUT-3 and MJ-MUT-4 for the elongation of each prior amplificate. The final PCR product (651 bp) was digested with EcoRI and BamHI and was ligated into the plasmid pNCO113 (18,38), which had been treated with the same restriction enzymes. The resulting plasmid designated pNCO-MJ-MUT-wild type was transformed into E. coli XL1-Blue cells.
Site-directed Mutagenesis-Mutations were performed by PCR using synthetic oligonucleotides and pNCO-MJ-MUT-wild type as template. All of the sequences were deposited in the GenBank TM sequence data base (Table III).
DNA Sequencing-Sequencing was performed using the Sanger dideoxy chain termination method (39). Plasmid DNA was isolated from cultures (5 ml) of XL1-Blue strains grown overnight in LB medium containing ampicillin (170 mg/l) using QIAprep spin columns from Qiagen.
SDS-Polyacrylamide Gel Electrophoresis-Sodium dodecyl sulfate polyacrylamide gel electrophoresis was performed by previously published procedures (40).
Protein Sequencing-Sequence determination was performed by the automated Edman method using a 471 A Protein Sequencer (PerkinElmer Life Sciences).
Frozen cell mass (5 g) was thawed in 40 ml of 30 mM potassium phosphate, pH 7.0. The suspension was cooled on ice and exposed to 60 pulses of a Branson-Sonifier B-12A (Branson Sonic Power Company, Danbury, CT) set to level 5. The suspension was centrifuged (Sorvall SS34 rotor, 15,000 rpm, 15 min, 4°C). Ammonium sulfate was added to the supernatant to a final concentration of 2 M. After centrifugation (15,000 rpm, 15 min, 4°C), the protein solution was placed on top of a phenyl-Sepharose SP column (40 ml, Amersham Biosciences), which had been equilibrated with 30 mM potassium phosphate, pH 7.0, containing 1 M ammonium sulfate. The column was developed with 30 mM potassium phosphate, pH 7.0. Fractions were combined, concentrated by ultrafiltration (Ultrafree-15 centrifugal filter device, Millipore, Eschborn, Germany), and placed on top of a hydroxy apatite column (40 ml, Amersham Biosciences), which had been equilibrated with 20 mM potassium phosphate, pH 7.0. The column was developed with a linear gradient of 0 -1 M potassium phosphate. Fractions were combined and concentrated by ultrafiltration.
Analytical Ultracentrifugation-Sedimentation equilibrium experiments were performed with an analytical ultracentrifuge Optima XL-I from Beckman Instruments equipped with absorbance optics. Aluminum double sector cells equipped with quartz windows were used throughout. Protein concentration was monitored photometrically at 280 nm. Protein samples were dialyzed against 50 mM potassium phosphate, pH 7.0. The partial specific volume was estimated from the amino acid composition, affording a value of 0.732 ml g Ϫ1 (42).
NMR Spectroscopy--13 C NMR spectra were recorded at 125.6 MHz using Avance 500 spectrometer from Bruker Instruments (Karlsruhe, Germany).  The recombinant protein obtained by heterologous expression of the synthetic gene was purified by the chromatographic procedures described "Experimental Procedures" and was obtained in apparently pure form as judged by sodium dodecyl sulfate polyacrylamide electrophoresis (Fig. 2).
The N-terminal sequence of the recombinant protein was verified by partial Edman degradation affording the sequence MNNVEKAIEALKKGE. A relative mass of 25,799 Da was observed by electrospray mass spectrometry in good agreement with the calculated mass of 25,796 Da.
Sedimentation equilibrium centrifugation of the recombinant M. jannaschii protein afforded a molecular mass of 54   kDa, which indicates a homodimer structure in parallel with the homodimer structure of the E. coli enzyme that has been established by analytical ultracentrifugation and by x-ray crystallography (43,44). The reaction catalyzed by the recombinant protein was observed by 13 C NMR spectrometry in real time using [U-13 C 5 ]ribulose 5-phosphate as substrate. The singlet at 170.9 ppm reflects the formate obtained by fragmentation of the substrate. The other four signals are complex multiplets because of 13 C 13 C and 13 C 31 P coupling. Specifically, the doublet of doublets at 25.7 ppm reflects the methyl group of the enzyme product 3,4-dihydroxy-2-butanone 4-phosphate, which is coupled to the position 2 carbonyl group with a coupling constant of 40.9 Hz and C-3 with a coupling constant of 13.1 Hz. The carbonyl group of 4 resonates at 211.7 ppm and appears as a pseudotriplet because of coupling to C-1 and C-3. The signal of C-3 is a complex multiplet reflecting 13 C 13 C coupling to C-2 and C-4 with coupling constants of 41 and 40 Hz, to C-1 with a coupling constant of 13.1 Hz, and to phosphorus with a coupling constant of 7.5 Hz. C-4 of the product also appears as a multiplet because of coupling to C-3 and to phosphorus. The 13 C spectrum shown in Fig. 3A unequivocally identifies the enzyme product as 3,4-dihydroxy-2-butanone 4-phosphate. The NMR data are summarized in Table IV. The reaction product obtained from unlabeled ribulose 5-phosphate by the catalytic action of the enzyme is shown in Fig. 3C for comparison. The signals of C-3 and C-4 show 13 C 31 P coupling. The use of the 13 C-labeled substrate results in an ϳ90-fold sensitivity enhancement for the detection of formate, which was crucial for the detection of residual catalytic activity in certain mutant proteins described below.
The spectrum in Fig. 3B was obtained in an enzyme experiment using a mixture of [U-13 C 5 ]ribulose 5-phosphate and unlabeled ribulose 5-phosphate proffered at a ratio of 1:25. The spectrum of the product represents a superposition of the spectra in Fig. 3, A and C. This is specifically demonstrated for C-3 in Fig. 4 by the help of spectral simulation, which accounts rigorously for all observed lines.
The enzyme-catalyzed reaction involves breaking of the bonds connecting C-3 and C-5 to C-4 and the formation of a novel bond between C-3 and C-5 of the substrate. Intermolec-ular recombination of fragments from labeled and unlabeled substrate would be conducive to partially 13 C-labeled products, more specifically, to [1,2,3-13 C 3 ]compound 4 and [4-13 C 1 ]compound 4 as shown in Fig. 4, which shows simulated spectra for C-3 of different isotopomers. The 13 C spectrum of compound 4 obtained from a mixture of [U-13 C 5 ]ribulose 5-phosphate and unlabeled ribulose 5-phosphate is virtually identical with the superposition of spectra for compound 4 obtained from unlabeled ribulose 5-phosphate and [U-13 C 5 ]ribulose 5-phosphate. It follows that the reaction proceeds by a strictly intramolecular mechanism.
Because it is known that both enantiomers of 3,4-dihydroxy-2-butanone 4-phosphate can serve as substrates for 6,7-dimethyl-8-ribityllumazine synthase of Bacillus subtilis (25), we determined the configuration of the product of M. jannaschii 3,4-dihydroxy-2-butanone-4-phosphate synthase. The CD spectrum shown in Fig. 5 is that expected for the L-isomer (23). Hence, the M. jannaschii protein converts ribulose 5-phosphate into a mixture of formate and L-3,4-dihydroxy-2-butanone 4-phosphate. It follows that archaea and eubacteria use the same intermediate as precursor for the xylene ring of riboflavin.
Sequence comparison showed that 17 polar amino acids were absolutely conserved in putative orthologs of the 3,4-dihydroxybutanone-4-phosphate synthase from microorganisms and plants. Most notably, a short stretch of charged amino acids extending from position 21 to 30 showed a high degree of overall conservation (Fig. 6).   The hypothetical reaction mechanism (Fig. 7) suggests a crucial role for acid/base catalysis, and polar ligands are probably involved in the interaction of the protein with the essential divalent metal ion (Mg 2ϩ or Mn 2ϩ ). Preliminary NMR studies on enzyme-substrate complexes had also suggested that the conserved amino acid residues Thr-112, Thr-115, Asp-118, Arg-119, and Thr-122 interact with the substrate (45). The x-ray structure of 3,4-dihydroxy-2-butanone-4-phosphate synthase of E. coli in complex with glycerol, which is believed to bind at the active site as a fortuitous substrate analog, is also in line with the hypothesis that the loop comprising the conserved cluster of acidic residues is directly involved in catalysis (46).
The synthetic gene specifying 3,4-dihydroxy-2-butanone 4-phosphate of M. jannaschii could be mutagenized conveniently using various experimental techniques. A total of 33 mutant genes could be expressed efficiently in E. coli. The recombinant proteins were isolated as described under "Experimental Procedures," and their steady-state kinetic parameters were initially determined by a coupled photometric assay using 6,7-dimethyl-8-ribityllumazine synthase as reporter enzyme (Table III). The replacement of glutamate 26, 28, or 185, aspartate 21 or 30, histidine 164, or arginine 25 afforded proteins with relative activities in most cases of Ͻ0.7%. Mutants obtained by the replacement of arginine 119 or 161, threonine 165, or glutamate 166 retained significant enzyme activity, but their K m values were increased ϳ10-fold.
The sensitivity of the photometric assay monitoring the formation of 6,7-dimethyl-8-ribityllumazine is limited by the sta-  bility of 3,4-dihydroxy-2-butanone 4-phosphate and 5-amino-6ribitylamino-2,4(1H,3H)-pyrimidinedione. The relatively large K m and the relatively low catalytic rate of the auxiliary enzyme, 6,7-dimethyl-8-ribityllumazine synthase, limit the accuracy of the assay still further, particularly in the case of mutants with low residual activity.
Formate, the second product of 3,4-dihydroxy-2-butanone-4phosphate synthase, has the advantage of virtually unlimited stability. For this reason, we measured the enzyme-catalyzed formation of formate by 13 C NMR spectroscopy in real time.
To increase the sensitivity of NMR detection, we used [U-13 C 5 ]ribulose 5-phosphate as substrate. With the wild type enzyme, this assay recorded an activity of 174 nmol mg Ϫ1 min Ϫ1 as compared with 148 nmol mg Ϫ1 min Ϫ1 determined with the photometric assay. Energies of activation were 61 and 55 kJ mol Ϫ1 , respectively, as calculated from the Arrhenius plot (290 -330 K) for the NMR and the photometric assay. The two assays also gave similar values with mutant proteins, which display relatively high activities (Table III).
Mutant proteins that had little or no enzyme activity according to the photometric assay were reanalyzed by the NMR assay. To maximize the sensitivity, the assay mixture contained protein in the millimolar range. Operating thus under single turnover conditions, the detection limit was 40 pmol mg Ϫ1 min Ϫ1 . That value is equivalent to one molecule of product formed per subunit during a period of 16 h. Even at that level of assay sensitivity, the formation of formate could not be detected with following mutants: D21E, D21N, E26D, E26Q, E28D, D30E, H164N, and E185D. DISCUSSION The data reported in this paper show that the formation of the riboflavin precursor, 3,4-dihydroxy-2-butanone 4-phosphate, from ribulose 5-phosphate proceeds similarly in eubacteria, fungi, plants, and archaea. An intramolecular rearrangement precedes the elimination of formate.
The synthetic gene for the expression of the M. jannaschii enzyme enabled the production of the protein in high yield in E. coli. Because the synthetic gene comprises a closely spaced set of unique restriction sites, it is also ideally suited for mutagenesis by the replacement of short cassettes. This enabled the rapid construction of a large set of mutants addressing all conserved polar amino acid residues, which may be essential for catalysis.
The published assay of 3,4-dihydroxy-2-butanone 4-phosphate activity (47) uses an auxiliary enzyme, 6,7-dimethyl-8ribityllumazine synthase, to monitor 3,4-dihydroxybutanone 4-phosphate. This assay is suitable for samples with a relatively high catalytic activity but becomes error-prone in the case of mutant proteins with minimum activity.
A novel assay reported in this paper detects the formate formed by 3,4-dihydroxy-2-butanone 4-phosphate by 13 C NMR spectroscopy. Formate has the advantage of virtually unlimited stability under the experimental conditions. To enhance the sensitivity of the method, we used [U-13 C 5 ]ribulose 5-phosphate as substrate that can be easily obtained from [U-13 C 6 ]glucose (24). Using this assay, relative activities of 0.02% as compared with the wild type could be detected.
3,4-Dihydroxy-2-butanone-4-phosphate synthases have a characteristic motif of polar amino acids extending from residue 21 to 30 in case of the M. jannaschii enzyme (Fig. 6). Replacement of aspartate 21 or 30 or glutamate 26 or 28 reduces the enzyme activity in most cases to levels of Ͻ0.02% as compared with the wild type enzyme. Outside the patch of charged amino acids, the replacement of threonine 112, histidine 164, or glutamate 185 resulted in the reduction of enzyme activity by at least three orders of magnitude.
The hypothetical reaction sequence in Fig. 7 suggests a large number of proton transfer reactions (around a dozen). Therefore, it is not surprising that charged amino acids could play a major role in such a complex reaction trajectory.