|Home | About | Journals | Submit | Contact Us | Français|
Thiamin diphosphate is an essential cofactor in all forms of life and plays a key role in amino acid and carbohydrate metabolism. Its biosynthesis involves separate syntheses of the pyrimidine and thiazole moieties, which are then coupled to form thiamin monophosphate. A final phosphorylation produces the active form of the cofactor. In most bacteria, six gene products are required for biosynthesis of the thiamin thiazole. In yeast and fungi only one gene product, Thi4, is required for thiazole biosynthesis. Methanococcus jannaschii expresses a putative Thi4 ortholog that was previously reported to be a ribulose 1, 5-bisphosphate synthase [Finn, M. W. and Tabita, F. R. (2004) J. Bacteriol. 186, 6360–6366]. Our structural studies show that the Thi4 orthologs from M. jannaschii and Methanococcus igneus are structurally similar to Thi4 from Saccharomyces cerevisiae. In addition, all active site residues are conserved except for a key cysteine residue, which in S. cerevisiae is the source of the thiazole sulfur atom. Our recent biochemical studies showed that the archael Thi4 orthologs use nicotinamide adenine dinucleotide, glycine and free sulfide to form the thiamin thiazole in an iron-dependent reaction [Eser, B., Zhang, X., Chanani, P. K., Ealick, S.E., and Begley, T.P. (2015) submitted]. Here we report X-ray crystal structures of Thi4 from M. jannaschii complexed with ADP-ribulose, the C205S variant of Thi4 from S. cerevisiae with a bound glycine imine intermediate, and Thi4 from M. igneus with bound glycine imine intermediate and iron. These studies reveal the structural basis for the iron-dependent mechanism of sulfur transfer in archael and yeast thiazole synthases.
Thiamin pyrophosphate (ThDP 4) plays a key role in branched amino acid biosynthesis and carbohydrate metabolism, and is an essential cofactor in all living systems. Its biosynthesis requires the separate formation of the pyrimidine and thiazole moieties, which are then coupled to form thiamin monophosphate (ThMP 3) which is then phosphorylated to form the cofactor (Scheme 1). The mechanism of thiazole formation in bacteria has been elucidated and involves a complex oxidative condensation in which five different enzymes and a sulfide carrier protein catalyze reactions involving three substrates (1-deoxy-D-xylulose-5-phosphate, cysteine, and glycine or tyrosine).1–3In yeast, fungi, and plants, only one gene product, Thi4, is required for thiazole biosynthesis, and nicotinamide adenine dinucleotide (NAD) 5, glycine 8 and cysteine are the substrates.4–6Our previous results showed that Saccharomyces cerevisiae Thi4 (ScThi4) is a suicide thiazole synthase that uses a conserved Cys205 residue as the sulfur source, and that the sulfur transfer reaction is Fe(II) dependent.7We have recently shown that the C205A variant of the yeast enzyme can utilize sulfide as the sulfur source.8A mechanistic proposal for the Thi4 catalyzed formation of adenosine diphosphate thiazole (ADT 17) is shown in Figure 1.7
Putative orthologs of Thi4 are also found in archaea. The Thi4 ortholog in Methanococcus jannaschii (MjThi4, also known as Mj0601) was first characterized as an NAD-dependent ribulose 1, 5-bisphosphate (RuBP) synthase, which converts ribose 1,5-bisphosphate to RuBP.9 M. jannaschii is a thermophilic methanogenic archae with 85 °C as the optimum living temperature10and is found in anaerobic environments with high levels of sulfide.11We recently showed that MjThi4, which lacks the equivalent of ScThi4 Cys205, synthesizes the thiamin thiazole moiety (ADT 17) using NAD, glycine, and sulfide in an iron-dependent reaction.8
Here, we report the structure of MjThi4 with bound ADP-ribulose (ADPrl 7), which copurified with the enzyme. We also report the structure of the C205S variant of ScThi4 with a bound glycine imine intermediate 12, and the structure of Methanococcus igneus Thi4 (MiThi4) with 12 and Fe(II). These structures suggest a mechanism for iron-dependent sulfur transfer in the formation of the thiamin thiazole. Finally, we were unable to demonstrate RuBP synthase activity using purified MjThi4, suggesting that in archaea the only function for the Thi4 orthologs is biosynthesis of the thiamin thiazole.
The overexpression plasmid for the M. jannaschii Thi4 gene was constructed at the Cornell Protein Production Facility. Plasmid DNA was purified with a Fermentas GeneJet miniprep kit and DNA fragments were purified from agarose gel with a Zymoclean Gel DNA Recovery kit from Zymo Research. Escherichia coli strain MachI (Invitrogen) was used as a recipient for transformations during plasmid construction and for plasmid propagation and storage. Phusion DNA polymerase (New England Biolabs) was used for PCR following the manufacturer’s instructions. PfuTurbo DNA polymerase (Stratagene) was used for mutagenesis. All restriction enzymes were purchased from New England Biolabs. The plasmid THT, which was purchased from Novagen, is a derivative of pET-28 and incorporates an N-terminal polyhistidine tag cleavable by TEV protease and a short sequence to improve solubility.
The Thi4 gene was amplified using standard PCR conditions with desired primers. The amplified product was purified and digested with NdeI and XhoI, and inserted into similarly digested pTHT vector to yield MjThi4/pTHT and MiThi4/pTHT. Mutagenesis was performed following a standard PCR-mediated site directed protocol using primers with the desired mutation. The wild type ScThi4/p28 and MjThi4/pTHT clones were used as the template to generate the ScThi4C205S/p28 and MjThi4H164C/THT variants, respectively.
The Thi4 plasmids were transformed into E. coli BL21(DE3) (Novagen) cells and grown with shaking overnight at 37 °C in a 20 mL starter culture of Luria-Bertani (LB) media containing 30 µg/mL kanamycin. The overnight starter culture was then transferred into 1.5 L of either LB medium or M9 minimal medium and incubated at 37 °C with shaking until reaching an OD600 of ~0.6. In the case of M9 minimal medium, the medium was supplemented with 0.2% glucose, 1 mM MgSO4 and 0.1 mM CaCl2. The culture was then induced with 0.5 mM isopropyl 1-β-D-galactopyranoside (IPTG) and incubation was further carried out for 18 – 24 h at 15 °C. Cells were harvested by centrifugation at 1000 g for 30 min at 4 °C. The cell pellet was collected and frozen at −20 °C for storage.
The N-terminal hexahistidine-tagged MjThi4 was first purified by Ni-nitrilo acetic acid (Ni-NTA) affinity chromatography. Cell pellets from 2 L of culture were resuspended in a lysis buffer containing 20 mM Tris, 500 mM NaCl, 10 mM imidazole, pH 8.0, and lysed by sonication. The lysate was centrifuged at 4500 g for 30 min at 4 °C and the resulting supernatants were subjected to a column containing 2 mL Ni-NTA resin (Qiagen) preequilibrated with the lysis buffer. The column was washed with 10 column volumes of buffer containing 20 mM Tris, 500 mM NaCl, 30 mM imidazole, pH 8.0, and bound protein was recovered with 20 mM Tris, 500 mM NaCl, 250 mM imidazole, pH 8.0. The eluted samples were further purified using a Superdex G200 column (GE Healthcare) with 10 mM Tris, 50 mM NaCl, pH 8.0, and the peaks corresponding to MjThi4 were pooled, concentrated to 15 mg/mL using an Amicon concentrator (10 kDa MWCO filter, Millipore), flash frozen, and stored at −80 °C. MiThi4 was purified using the same procedure.
N-terminal hexahistidine-tagged ScThi4C205S was expressed in M9 minimal media and purified in a glove box under anaerobic conditions. Cell pellets from 2 L of culture were lysed by sonication on ice in 20 mM Tris, 500 mM NaCl, 10 mM imidazole, pH 8.0. The resulting supernatant was subjected to a Ni-NTA column and the column was washed with 10 column volumes of buffer containing 20 mM Tris, 500 mM NaCl, 50 mM imidazole, pH 8.0. Bound protein was recovered by elution with 20 mM Tris, 500 mM NaCl, 250 mM imidazole, pH 8.0. The eluted samples were buffer exchanged into 10 mM Tris, 50 mM NaCl, pH 8.0 using an Econo-Pac 10DG desalting column (Bio-Rad). The purified protein was concentrated to 10 mg/mL using an Amicon Ultra-0.5 concentrator (10 kDa MWCO filter, Millipore) and incubated at room temperature for 1 h with 2 mM NAD, 2 mM glycine, 0.5 mM ferrous ammonium sulfate (FeAS), and 2 mM dithiothreitol.
Crystals of MjThi4 were grown using the vapor diffusion hanging drop method. Subsequent analysis showed that MjThi4 selectively crystallized as the MjThi4/ADPrl 7 complex. Equal volumes of protein and reservoir solutions were mixed and equilibrated at 22 °C against a total volume of 400 µL well solution. The initial crystallization condition was determined using the commercially available Wizard I sparse matrix screen (Emerald Biosystems). The optimized condition for MjThi4/ADPrl 7 was 1.05 M potassium/sodium tartrate, 0.1 M N-cyclohexyl-2-aminoethanesulfonic acid (CHES)/sodium hydroxide, pH 9.5 and 0.2 M lithium sulfate. Rod shaped crystals grew in about six days to their maximum size of 0.5–0.6 mm × 0.1–0.2 mm. The crystals were cryoprotected for data collection in a solution composed of the mother liquor supplemented with 2.5% (v/v) ethylene glycol.
Crystals of ScThi4C205S were grown under anaerobic conditions using the vapor diffusion hanging drop method. Equal volumes of protein and reservoir solutions were mixed and equilibrated at 22 °C against a total volume of 400 µL well solution. The initial crystallization condition was determined using the commercially available Wizard IV sparse matrix screen (Emerald Biosystems). The optimized conditions were 25% (w/v) PEG1500, 0.0125 M succinic acid, 0.044 M sodium dihydrogen phosphate, and 0.044 M glycine, pH 8.5. ScThi4C205S crystals grew in about 10 days to their maximum size of ~0.1 mm × 0.1 mm × 0.4 mm. Crystals were cryoprotected for data collection in a solution composed of the mother liquor supplemented with 15% (v/v) glycerol.
MiThi4 (300 µL, 500 µM) was incubated anaerobically with 1.5 equivalents of FeAS at 22 °C for 30 min. The mixture was then incubated with 1.5 equivalents of adenosine diphosphate ribose (ADPr) at 80 °C for 30 min, and finally 1.5 equivalents of glycine were added to the reaction system and incubated at 80 °C for a further 30 min to form MiThi4/Fe/glycine imine 12. The protein complex was concentrated to 15 mg/mL and removed from the glove box for further crystallization experiments. The initial crystallization conditions for MiThi4/Fe/glycine imine 12 were determined using the commercially available Wizard II sparse matrix screen (Emerald Biosystems). The optimized condition was 1.0 M sodium citrate, 0.1 M CHES/sodium hydroxide, pH 9.5. Bipyrimidal-shaped crystals grew in about 4 days to a maximum dimension of 0.4–0.5 mm. Crystals were cryoprotected for data collection in a solution composed of the mother liquor supplemented with 2.5% (v/v) ethylene glycol.
X-ray diffraction data for MjThi4/ADPrl 7, ScThi4C205S/glycine imine 12 and MiThi4/Fe/glycine imine 12 were collected at the Advanced Photon Source Northeastern Collaborative Access Team (NE-CAT) beamline 24-ID-C using an ADSC Q315 CCD detector with a crystal to detector distance of 475 mm for MjThi4/ADPrl 7, 260 mm for ScThi4C205S/glycine imine 12, and 320 mm for MiThi4/Fe/glycine imine 12. The wavelength was 0.9791 Å and the data collection temperature was 100 K. Individual frames were collected using 1 s for each 1.0° oscillation over a range of 180° for MjThi4/ADPrl, 120° for ScThi4C205S/glycine imine 12, and 150° for MiThi4/Fe/glycine imine 12. X-ray diffraction data were indexed, integrated, scaled, and merged using the program HKL2000.12Data collection and processing statistics are shown in Table 1.
The structure of MjThi4/ADPrl 7 was determined by molecular replacement using Phaser13as implemented in the PHENIX14 program package. An octamer of Arabidopsis thaliana thiamin thiazole synthase (AtThi1) (PDB entry 1RP0)15, which shares a 32% sequence identity with MjThi4, was used as the template to generate a search model using Chainsaw16. The initial refinement resulted in an R-factor of 33.1% and R-free of 39.7%. Side chains were added and the model was manually adjusted using COOT17and refined using Refmac 5.0.18After several cycles of refinement and model building, ADPrl 7 was built into the active site based on the Fo - Fc electron density, and then water molecules were added in the model. Geometry was verified using COOT17and PROCHECK.19The final refinement statistics are summarized in Table 1.
The structure of ScThi4C205S/glycine imine 12 was determined by molecular replacement using Phaser13as implemented in the PHENIX14program package. A dimer of wild-type ScThi4 (PDB entry 3FPZ)7was used as the search model. The initial refinement resulted in an R-factor of 25.7% and R-free of 21.3%. The model was manually adjusted using COOT17and refined using Refmac 5.0.18After several cycles of refinement and model building, the glycine imine intermediate 12 was built into the active site based on the Fo - Fc and composite omit maps, and then water molecules were added to the model. Geometry was verified using COOT17and PROCHECK.19The final refinement statistics are summarized in Table 1.
The structure of MiThi4/Fe/glycine imine 12 was determined by molecular replacement using Phaser13as implemented in the PHENIX14program package. An octamer of wild-type AtThi1 (PDB entry 1RP0),15which shares 29% sequence identity with MiThi4, was modified using Chainsaw16to generate a search model. The initial molecular replacement solution was refined to an R-factor of 40.8% and R-free of 46.9%. All side chains were added and the model was manually adjusted using COOT.17After several cycles of refinement using COOT17, Fe, the glycine imine intermediate 12, and water molecules were added to the model. The final model was refined to an R-factor of 19.9% and R-free of 21.6%. Refinement statistics are summarized in Table 1.
All assays were performed in a glove box either at 25 °C or 75 °C with components prepared with degassed, deionized water. Sodium sulfide was prepared by passing H2S through an aqueous solution of NaOH. The final concentration of the sulfide solution (5–20 mM Na2S) was determined by both the 5,5´-dithiobis-(2-nitrobenzoic acid) and methylene blue methods.20, 21The sulfide solutions were stored at highly basic solutions (pH>12) to prevent escape of H2S and under anaerobic conditions to prevent oxidation. Stock solutions of ADPr 7, NAD 5, and glycine 8 were prepared at concentrations of 1 mM, 10 mM, and 100 mM, respectively. Buffer A (50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.1, 0.1 M KCl, and 10% glycerol) was added to make the final volume of each mixture 100 µL. Purified MjThi4 (20–100 µM) was first incubated with FeAS (400 µM to 1 mM) for approximately 10 min to assure binding of iron to the enzyme and to prevent precipitation of iron with sulfide. The conversion of NAD 5 to ADT 17 by wild-type MjThi4 was carried out in the presence 100 µM MjThi4, 300 µM NAD 5, 1 mM glycine 8, 400 µM FeAS and 800 µM Na2S at 25 °C, and quenched with 8 M guanidinium chloride. The activity assay for the H164C variant was performed in the presence or absence of 400 µM Na2S with 190 µM variant enzyme, 500 µM ADPr 7, 1 mM glycine 8, 500 µM FeAS, and 1 mM dithionite at 75 °C. After incubation of the assay ingredients for 1 h for the sample with Na2S or for overnight for the sample without Na2S, the samples were quenched with equal volumes of 8 M guanidinium hydrochloride. The ScThi4 variants (C205S, D207A, and H237N) were similarily assayed. All the reaction components were allowed to sit for 2 h in the glovebox for complete deoxygenation. Purified enzyme was first incubated with freshly prepared FeAS (final concentration 400 uM) for 10 min followed by the addition of sulfide (final concentration 600 uM), glycine (final concentration 1 mM) and NAD or ADPr (final concentration 500 µM). Reactions were incubated at room temperature for 6 h followed by quenching with 8 M guanidinium hydrochloride. The reaction mixture was deproteinized by passing through a 10 kD cut off filter and then analyzed by HPLC using a Supelcosil LC-18 reversed-phase column (150 mm × 4.6 mm, 3 µm ID). The following linear gradient, at a flow rate of 1 mL/min, was used. Solvent A is water; solvent B is 100 mM K2HPO4, pH 6.6; and solvent C is methanol. 0 min: 100% B; 3 min: 10% A, 90% B; 10 min: 25% A, 60% B, 15% C; 14 min: 25% A, 60% B, 15%; 19 min: 30% A, 40% B, 30% C; 21 min: 100%B; 30 min 100% B.
The derivatization reaction uses 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) as a coupling reagent and pyridine as a catalyst.22 Formation of RuBP 18 (Figure 2A) was tested by derivatization of 3-phosphoglycerate (3-PGA 19), which would form upon reaction of RuBP with ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO).9The substrate proposed for RuBP 18 formation is ribose 1, 5-bisphosphate,9which is not available commercially. Ribose 1, 5-bisphosphate was generated from 5-phospho-α-D-ribose 1-diphosphate (PRPP) by incubating with Mg2+ at ~80 °C for a few hours at neutral pH.9To test for MjThi4 catalyzed RuBP 18 formation, apo MjThi4 (~500 µM final) was incubated, in a glove box, with ribose 1, 5-bisphosphate (~5 mM final) and NAD 5 (final ~500 µM) in 50 mM HEPES (pH 7.5). After a few hours, RuBisCo (~0.2 µM final) and NaHCO3 (20 mM final) were added and the resulting mixture was anaerobically incubated overnight at room temperature. A control reaction, containing FeAS (final ~ 500 µM) in addition to the above components, was also run. The reaction mixture was then treated with calf intestinal phosphatase (CIP) for 1 h at 37 °C. For the derivatization (Figure 2)22, 200 µL of the dephosphorylated reaction mixture were mixed with 2,4-DNPH (80 µL of 40 mM solution in 1:1 0.2 M HCl/ethanol) and EDC (100 µL of 125 mM solution in 1.5% pyridine, 50% ethanol, 48.5% water) and incubated at 80 °C for 1 h. The samples were then centrifuged to remove the precipitate and analyzed by HPLC using Supelcosil LC-18 reversed-phase column (150 mm × 4.6 mm, 3 µm ID). Two mobile phases were used for HPLC analysis; (A) acidified water with HCl at pH 4.5, (B) acetonitrile. Elution was performed isocratically in 90% (A) and 10% (B) at a flow rate of 1 mL/min. The 2,4-DNP hydrazide 22 of 3-PGA could be detected with a sensitivity of 30 – 50 µM.
MjThi4 (500 µM) was loaded with ~50% iron (to simultaneously assay the apo and the metalloenzyme) and reacted overnight in a glove box with ~5 mM PRPP and ~25 mM MgCl2 at 50 °C in the presence and absence of NAD 5 (0.5 mM final). The samples, in 1 mL of D2O:H2O (1:10), were transferred to 5 mm NMR tubes and 31P NMR spectra were acquired on a Bruker Avance III 400 Mhz spectrometer at 161.98 MHz. A 45° read pulse was used for 100 scans with Waltz-16 proton decoupling. For comparison, the NMR spectrum of a RuBP 18 standard was also taken.
All figures were prepared using PyMOL23or ChemDraw (Cambridge Biosoft), and compiled in Photoshop (Adobe).
MjThi4 and MiThi4 show approximately 30% sequence identity to ScThi47and AtThi115and the four structures are very similar (Figure 3). All Thi4 orthologs are homooctamers. Each protomer consists of 12 β-strands organized as one six-stranded and two three stranded β-sheets flanked by nine α-helices. The Thi4 octamer has 422 symmetry and can be thought of as a tetramer of closely packed dimers. The buried dimer interface has an area of ~1500 Å2. The octamer contains a channel along the fourfold axis with a maximum opening of about 30 Å.
The active site of MjThi4 is located near the central channel and involves two adjacent protomers. Most of the residues come from one protomer, and a loop containing Gly162′-Asp166′ from a fourfold related protomer completes the active site. Analysis of MjThi4 samples expressed in LB media showed high levels of NAD 5, ADT 17, and flavin mononucleotide and smaller amounts of ADPr 6 and ADPrl 7 eluting with the protein (data not shown). Based on the Fo - Fc electron density in the active site, the intermediate ADPrl 7 was included in the model (Figure 4A). The structure of MiThi4 also showed ADPrl 7 unless further treated with glycine and iron. The interactions of ADPrl 7 with the surrounding residues are illustrated in Figure 4B. The N1, N3, N6, and N7 atoms of the adenine base hydrogen bond to the Val138 amide nitrogen atom, the Arg67 amide nitrogen atom, the Val138 carbonyl oxygen atom and the Ser184 side chain hydroxyl group, respectively. The side chain carboxylate group of Glu66 hydrogen bonds to the O2′ and O3′ hydroxyl groups of the ADPrl. The ADPrl a-phosphate hydrogen bonds to the Gly74 amide nitrogen atom and the His164′ side chain from the fourfold related protomer. The β-phosphate hydrogen bonds to the Ser47 side chain and amide group, the Met230 amide group and two water molecules. The ribulose moiety hydrogen bonds to the carbonyl groups of Gly74 and Gly242, and both carboxylate oxygen atoms of Asp166′ from the adjacent protomer.
Electron density for the ScThi4C205S variant treated with NAD and glycine showed the glycine imine intermediate 12 in the active site (Figure 5A and S1). Interactions between the intermediate and the active site residues are shown in Figure 5B. The N1 and N6 atoms of the adenine base hydrogen bond to the Val170 amide and carbonyl groups, analogously to MjThi4/ADPrl 7, and the Ser98 amide donates a hydrogen bond to N3. The side chain carboxylate group of Glu97 hydrogen bonds to the O2′ and O3′ hydroxyl groups of the ADPrl 7 ribose, analogously to MjThi4/ADPrl 7, and also hydrogen bonds to the Ser98 and a water. The ADP α-phosphate group hydrogen bonds to the Gly105 amide nitrogen atom and two water molecules. The ADP β-phosphate hydrogen bonds to the Met291 amide group and two water molecules. The hydroxyl groups of the ribulose fragment only interact with water molecules. The nitrogen atom of the glycine fragment forms two strong hydrogen bonds with Asp207′ of the adjacent protomer.
MiThi4 and MjThi4 share 78% sequence identity and the ADP portion of the binding site in MiThi4/Fe/glycine imine 12 is essentially identical to that of MjThi4/ADPrl 7 (Figures 2, ,6A6A and S2). Likewise, the glycine imine 12 conformation is very similar to that observed for the ScThi4C205S complex; however, a small movement takes place as required to form the octahedral iron binding site. The ribulose and glycine fragments of the glycine imine 12 occupy three positions of the octrahedral Fe(II) binding site through coordinate covalent bonds with a ribulose oxygen atom and the glycyl nitrogen and carboxylate atoms (Figure 6B). The other three sites are occupied by His176, Asp161′ from the adjacent protomer and a water molecule. The Fe(II) to oxygen or nitrogen distances range from 2.0 Å to 2.2 Å. Analogous to ScThi4C205S, the glycyl carboxylate group of the glycine imine 12 forms a strong salt bridge with Arg236 through two strong hydrogen bonds (Figure 6C).
The activity assays of MjThi4 and H164C MjThi4 were performed under anaerobic conditions and HPLC was used to detect the reaction products. Wild-type MjThi4 catalyzes thiazole biosynthesis with the substrates NAD 5, glycine 8 and sodium sulfide in the presence of FeAS (Figure 7A). The enzyme can only use free sulfide as the sulfur source for ADT 17 formation, and the H164C variant still requires free sulfide for the reaction (Figure 7B). The ScThi4C205S variant is catalytic in the presence of exogenous sulfide, the ScThi4D207A variant is inactive, and the ScThi4H237N variant only catalyzes the conversion of ADPr 6 to ADPrl 7.
Analysis of standards showed that concentrations of RuBP 18 as low as 30–50 µM can be detected using the RuBisCO/2, 4-DNPH assay (Figure 2A). Using this assay, we demonstrated that MjThi4 does not catalyze the formation of RuBP 18 (Figure 2C). Neither NAD 5 nor Fe(II) had any effect on RuBP 18 formation (Figure 2C).9
The 31P-NMR spectra for reaction samples with and without NAD 5 did not exhibit any peak that corresponds to the 31P-NMR spectrum of the RuBP 18 standard (Figure 8) indicating no detectable RuBP 18 synthase activity.
A structural similarity search for MjThi4 was performed using the DALI24online server. ScThi4, AtThi1, MjThi4, and MiThi4 are homooctamers, and the protomers are structurally similar (Figure 9). Unsurprisingly, the most similar structures are Thi4 orthologs. The structure of AtThi1 (PDB ID 1RP0), which was used as the search model during molecular replacement, showed the highest similarity with a Z score of 34.6. The structure of ScThi4 (PDB ID 3FPZ) shows a Z score of 32.4. MjThi4 shares a 29% sequence identity with ScThi4 and superimposes to ScThi4 with an rmsd for 158 Cα atoms of 1.0 Å, while MjThi4 and AtThi1 share 32% sequence identity and superimpose with an rmsd for 180 Cα atoms of 0.9 Å. The key difference between MjThi4 and ScThi4 or AtThi1 is the length of the active site loops β10 - α6 and α5 - β9. The β10 - α6 loop and α5 - β9 loop in MjThi4 are 3–5 residues shorter than the corresponding loops in ScThi4 and AtThi4, which makes the active site of MjThi4 more open to the solvent (Figure 9B).
In addition to Thi4 orthologs, similar structures include NAD(FAD)-utilizing dehydrogenases (e.g., PDB ID 2IOZ with an rmsd of 2.6 and a Z score of 20.3). The NAD(FAD) binding domains are structurally similar to the NAD binding domain of Thi425; however, NAD serves as a cofactor for the dehydrogenases and as a substrate for Thi4.
Comparison of the MjThi4 active site with those of ScThi4 or AtThi1 reveals a highly conserved arrangement around the ribulose moiety of ADPrl 7 (Figure 9C). Active site residues Asp166′ and His164′ (MjThi4 numbering) are from loop α5 - β9 of the adjacent monomer, His181 and Glu182 are from loop β10 - α6. Three highly conserved residues, Asp166′, Arg240 and Gly242, interact with the thiazole moiety in the ScThi4 and AtThi1 structures (Figure 9C). Residues His189 and Asp190 in AtThi1 are involved in the coordination of a zinc ion in the active site13. These two residues are conserved in ScThi4, while in MjThi4 Asp190 is replaced by Glu182 (Figure 9C). All of these active site residues near the thiazole moiety in ScThi4 and AtThi1 are highly conserved or partially conserved in MjThi4, except for His164, which corresponds to Cys205 of ScThi4. ScThi4 was identified as a suicide enzyme and Cys205, which is conserved in all eukaryotic Thi4 orthologs, was shown to be the sulfur donor in a suicide reaction for thiazole synthesis5. The multiple sequence alignment result of eukaryotic Thi4 and archaea Thi4 homologs shows that both MjThi4 and MiThi4 lack this cysteine residue (Figure S3), and our studies show that thiazole synthesis in MjThi4 and MiThi4 utilizes sulfide as the sulfur source.
Together with the previously reported structure of ScThi4/ADT (PDB ID 3FPZ) 7and AtThi1/ADT (PDB ID 1RP0)15, complexes of four Thi4 orthologs are now available. Different orthologs were used because the crystallization behavior varied, and not all intermediates could be trapped for a single ortholog. While multiple intermediates were detected for purified MjThi4 and MiThi4, both selectively crystallized with bound ADPrl 7. Previous studies showed the presence of the glycine imine 12, the final intermediate 16, and the product ADT 17, in purified ScThi4; however, ScThi4 selectively crystallized with ADT 17. We were only able to trap the glycine imine 12 intermediate using the ScThi4C205S variant, and only MiThi4 crystallized with bound glycine imine and iron.
Superposition of the active sites of MjThi4/ADPrl, ScThi4C205S/glycine imine, MiThi4/Fe/glycine imine 12 and ScThi4 (PDB ID 3FPZ) provide snapshots along the Thi4 reaction coordinate (Figure 10). The residues involved in binding the ADP moiety are highly conserved and the ADP moieties from the four structures superimpose closely (Figures 3, ,44 and and5).5). Arg301 (ScThi4 numbering) anchors both intermediates and the product through hydrogen bonds with the carboxylate tail. This arginine residue is conserved in all Thi4 orthologs. The glycine imine intermediate 12 undergoes a conformational change upon Fe(II) binding, resulting in the formation of 23. Two residues, Asp161 and His176 (MiThi4 numbering), which chelate Fe(II) are conserved in MjThi4, MiThi4, ScThi4, and other orthologs indicating a similar Fe(II) binding site for ScThi4. The iron is released from the enzyme after the conversion of the glycine imine 12 to a structure that is no longer able to function as a tridentate ligand (Figure 11).
C205S, D207A, and H237N. Cys205 has previously been identified as the sulfide donor in the ScThi4 catalyzed reaction and the C205A variant can catalyze the formation of ADT 17 in the presence of sulfide. In contrast, the “dead enzyme” with dehydroalanine at position 205 is unable to utilize sulfide for thiazole formation. We therefore tested the C205S variant to further explore the constraints on the use of exogenous sulfide and were surprised to find that this variant could also use sulfide (Figures S4 and S5). This suggests that planarization of the alpha carbon at position 205 locks the enzyme into an inactive conformation.
Asp207 in ScThi4 corresponds to the iron ligating Asp161 of MiThi4 and is likely to have multiple functions. The D207A variant is inactive and unable to catalyze the conversion of NAD 5 to ADPr 6 or the conversion of ADPr to ADPrl 7 (Figure S6). This suggests that Asp207 plays a role in the initial N-glycosyl bond cleavage of NAD and in the tautomerization of 6 to 7 (see Figure 5B). H237 in ScThi4 corresponds to the iron ligating His176 of MiThi4 and is also likely to have multiple functions. The H237N variant catalyzes the conversion of ADPr 6 to ADPrl 7 but not the conversion of ADPr 6 to the glycine imine 12, suggesting that in addition to serving as a ligand for the iron, His237 plays an essential role in the conversion of 7 to 12 (Figure S7).
Cysteine, thioacetate, thiosulfate, sulfide, and glutathione were previously tested as possible sulfur sources in the MjThi4 activity assays, but only sulfide supported ADT 17 formation.8Because His164 corresponds structurally to Cys205 in ScThi4, it was mutated to cysteine to determine if a cysteine residue in this position could serve as the sulfur source for thiazole formation. Neither free cysteine nor cysteine at position 164 in the protein functioned as the sulfur source. Methanococci inhabit strictly anaerobic environments with high levels of sulfide9 and require high sulfide concentrations in the medium for cultivation (~4.5 mM).26 In addition, recent studies on iron-sulfur cluster, thiouridine and methionine biosynthesis in Methanococcus maripaludis suggests that the sulfur in these molecules was predominantly derived from exogenous sulfide.27,28 Therefore, we conclude that free sulfide is the likely sulfur donor for the in vivo biosynthesis of thiazole in M. jannaschii.
Our previous biochemical studies showed that the biosynthesis of the thiamin thiazole in yeast is Fe(II)-dependent.7We recently showed that ADT 17 formation catalyzed by MjThi4 is also dependent on metal; however, in addition to Fe(II) and Fe(III), many divalent metals including Mn(II), Co(II), Zn(II), Ni(II), Mg(II), and Ca(II) also support activity of MjThi4 at varying levels.8 We further showed that upon reaction of MjThi4 with Fe(II), ADPr 6 and glycine 8, a chromophore at 380 nm was observed.8 The crystal structure of this chromophoric intermediate showed that the iron is ligated by the carboxylate of the glycine imine intermediate 12, Asp161, His176, and one water molecule (Figure 6). The formation of the 380 nm absorbing species was only observed in the presence of Fe(II), suggesting a possible charge-transfer interaction between Fe(II) and the glycine imine intermediate. After the addition of free sulfide, the ADT-tautomer 16 was observed and the 380 nm peak decayed simultaneously, indicating that the loss of the 380 nm signal is due to the formation of dihydrothiazole 16.
MjThi4 and ScThi4 both use NAD 5 and glycine 8 to form the glycine imine intermediate 12 and the residues involved in glycine imine binding and conversion to the thiazole 17 are highly conserved. We therefore suggest that both enzymes use a similar mechanism for thiazole formation and that the structure of the MiThi4/glycine imine intermediate 12 with bound iron (Figure 6) provides important insight into the previously poorly understood role of iron in the ScThi4 mediated sulfur transfer chemistry. Based on this structure, we propose that sulfide from the reaction medium displaces the ketone oxygen from the iron, thus allowing the carbonyl carbon to reorient to satisfy the stereoelectronic requirements for sulfide addition to the carbonyl carbon (23 to 24, Figure 11) in the active site of MjThi4. While the reorientation requires nearly a 180° flip of the carbonyl group, the conformational change is consistent with the crystal structures of 23 and the ADT product complex. For the ScThi4, sulfide, generated from Cys205 by an elimination reaction, forms 24. Cys205 is in a flexible loop but in the C205S structure, His200, which hydrogen bonded to Glu171, could serve as a base to deprotonate Cys205 Ca. The reaction then proceeds by a common pathway involving addition of the iron-bound sulfide to the carbonyl of 24 followed by loss of water to give 26. No protein side chain is positioned to serve as a base; however, the β-phosphate of the ADP could serve this function. As the thiol of 26 cannot reach the glycine Ca, we propose that Fe(II) dissociation followed by bond rotation gives 27. Thioenolization, followed by addition of the resulting thiol to the imine gives 15. Water elimination and tautomerization completes the thiazole formation.
While our studies elucidate the sulfur transfer chemistry involved in thiamin thiazole biosynthesis, they also show that there is no mechanistic requirement for enzyme autoinactivation in the sulfur transfer. Why then does yeast use an active site cysteine rather than exogenous sulfide for thiazole biosynthesis? We do not yet know the answer to this interesting question. One possibility is that thiazole formation triggers a posttranslational modification and that “inactive” Thi4 has additional functions in yeast.
We thank the staff of NE-CAT at the Advanced Photon Source for assistance with data collection, Dr. Cynthia Kinsland for cloning MjThi4, MiThi4 and the ScThi4C205S variant, Dr. Howard Williams for measuring the 31P NMR spectra and Leslie Kinsland for assistance in preparing the manuscript.
This work was supported by NIH grants DK67081 (to SEE) and DK44083 (to TPB) and by the Robert A. Welch Foundation (A-0034 to TPB). This work is based upon research conducted at the Advanced Photon Source on the Northeastern Collaborative Access Team beamlines, which are supported by award GM103403 from the NIH. Use of the Advanced Photon Source is supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.
‡The coordinates for ScThi4C205S/glycine imine, MjThi4/ADPrl, and MiThi4/Fe/glycine imine have been deposited in the Protein Data Bank under accession codes 4Y4L, 4Y4M, and 4Y4N, respectively.
Supplementary Figure 1 is the ribbon diagram of the overall structure of ScThi4C205S.
Supplementary Figure 2 is the ribbon diagram of the overall structure of MiThi4.
Supplementary Figure 3 is the multiple sequence alignment result of archaeal Thi4 homologs with eukaryotic Thi4 homologs.
Supplementary Figures 4 and 5 show the activity of the ScThi4C205S variant.
Supplementary Figure 6 shows the activity of the ScThi4D207A variant.
Supplementary Figure 7 shows the activity of the ScThi4H237N variant.
This material is available free of charge via the Internet at http://pubs.acs.org.