The present paper describes the biosynthesis of the thiamin thiazole in Bacillus subtilis and Saccharomyces cerevisiae. The two pathways are quite different: in B. subtilis, the thiazole is formed by an oxidative condensation of glycine, deoxy-D-xylulose 5-phosphate and a protein thiocarboxylate, whereas, in S. cerevisiae, the thiazole is assembled from glycine, NAD and Cys205 of the thiazole synthase.
- Bacillus subtilis
- Saccharomyces cerevisiae
- thiamin biosynthesis
- thiamin thiazole
The major thiamin biosynthetic pathway in bacteria is outlined in Figure 1 [1,2]. In this pathway, glycine (3) undergoes an oxidative condensation with DXP (deoxy-D-xylulose 5-phosphate) (5) and ThiS thiocarboxylate (6), to give the thiazole tautomer (7), which then aromatizes to form carboxythiazole (8) [3,4]. The thiamin pyrimidine (15) is formed by a remarkable rearrangement of AIR (5-aminoimidazole ribotide) (14), an intermediate on the purine pathway . Coupling the thiazole and the pyrimidine, with concomitant decarboxylation, yields thiamin phosphate (2) [6,7]. A final phosphorylation gives thiamin pyrophosphate (1), the biochemically active form of the cofactor .
Our understanding of this biosynthetic pathway is now at an advanced stage. All of the biosynthetic genes have been identified and cloned, all of the enzymes have been overexpressed, reconstituted and structurally characterized, and mechanisms for all of the biosynthetic reactions, except for the pyrimidine synthase (ThiC) are reasonably clear [1,5]. The entire biosynthetic pathway has been fully reconstituted using pure enzymes. In the present paper, we describe the biosynthesis of the thiamin thiazole in Bacillus subtilis and compare this pathway with the very different thiazole biosynthesis recently elucidated in Saccharomyces cerevisiae.
Thiamin thiazole biosynthesis in B. subtilis
Each of the steps involved in the assembly of the thiamin thiazole in bacteria are described in the following sections.
The ThiO gene product encodes a flavin-dependent glycine oxidase that catalyses the oxidation of glycine (3) to the glycine imine (4) . In the absence of the other thiazole biosynthetic enzymes, the glycine imine is hydrolysed to glyoxal.
The structure of this enzyme, with N-acetylglycine bound at the active site, has been determined (PDB code 1NG3). This structure and studies with substrate analogues are consistent with a hydride-transfer mechanism for glycine oxidation .
ThiS thiocarboxylate formation
The chemistry involved in the formation of ThiS thiocarboxylate is outlined in Figure 1. Activation of ThiS-COOH (9), at its C-terminus, by adenylation, gives 10, which then acylates the IscS persulfide to give 13. Reduction of 13, by DTT (dithiothreitol) in the reconstitution reaction mixture, gives the ThiS thiocarboxylate (6) [13,14]. The biochemical reduction of 13 is not yet understood. In some bacteria, an additional protein (ThiI) mediates the sulfur transfer to 10 . ThiS thiocarboxylate (6) can also be efficiently synthesized by treating intein-activated ThiS-COOH with ammonium sulfide .
The structures of the ThiF–ThiS complex and the ThiF–ATP complex have been determined [17,18] (PDB codes 1ZUD and 1ZFN). The IscS protein probably does not form a specific complex with ThiS/ThiF because all four IscS paralogues in B. subtilis are competent persulfide donors [14,19].
Protein thiocarboxylates as sulfide carriers in other biosynthetic pathways
Protein thiocarboxylates have now been found to play a role as sulfide carriers in several other biosynthetic pathways, and sequence analysis suggests that this strategy may be quite general (Figure 2).
In molybdopterin biosynthesis, MoaE catalyses the transfer of sulfide from MoaD thiocarboxylate to give 19 [20,21] (Figure 2A). A protein thiocarboxylate-dependent cysteine biosynthetic pathway has been found in Mycobacterium tuberculosis (Figure 2B). In this pathway, CysM thiocarboxylate reacts with phosphoserine (21) in a PLP (pyridoxal 5′-phosphate)-mediated reaction to form a thioester (22). This then undergoes a nitrogen–sulfur acyl shift to give 23, followed by release of cysteine in a hydrolysis reaction catalysed by the Mec protease [22–26]. A closely related pathway for the biosynthesis of homocysteine was discovered in Wolinella succinogenes (Figure 2C). In this pathway, HcyS thiocarboxylate (25) adds to O-acetylhomoserine (26) to give a thioester (27). A nitrogen–sulfur acyl shift to give 28, followed by HcyD-catalysed amide hydrolysis generates homocysteine (30) [27,28]. A fourth example is found in the biosynthesis of the siderophore thioquinolobactin (34) (Figure 2D). In this pathway, QbsE thiocarboxylate forms a mixed thioanhydride (33) with quinolobactin (31). Hydrolysis of 33 generates the siderophore 34 [29,30]. A reagent for the sensitive detection of protein thiocarboxylates in a proteome, which uses a click reaction between the protein thiocarboxylate and a fluorophore-tagged sulfonyl azide, has been described .
Formation of the thiazole tautomer (7)
The bacterial thiazole synthase catalyses the condensation of DXP (5), ThiS-COSH (6) and the glycine imine (4) to form the thiazole tautomer (7)  (Figure 1). A mechanistic proposal for this reaction is outlined in Figure 3. In this mechanism, DXP (5) forms an imine with Lys96 of the thiazole synthase. Tautomerization to 38 followed by thiocarboxylate addition gives 39. An oxygen–sulfur acyl shift followed by loss of water generates thioketone (41). Tautomerization of 41, followed by loss of ThiS-COOH generates 43. Addition of the glycine imine (4) followed by transimination gives the thiazole tautomer (7).
In support of this mechanism, enzyme-catalysed exchange of the DXP carbonyl oxygen has been observed and the DXP/Lys96 imine has been trapped by borohydride reduction and characterized by MS analysis. Intermediate 37 is supported by the observation of enzyme-catalysed exchange of the C3 proton of DXP. The unanticipated oxygen–sulfur acyl shift to give 40 is supported by the observation of oxygen incorporation from DXP and not the buffer into the nascent ThiS-COOH (9). Thioenol (42) has also been trapped and characterized by MS analysis and the final product (7) has been fully characterized by spectroscopic analysis [3,14].
The structure of the ThiG–ThiS complex, with phosphate bound at the active site, has been determined (PDB code 1TYG). In this structure, the phosphate and Lys96 define the DXP-binding site, which suggests that Glu98 and Asp182 are also likely to play a role in the catalysis of thiazole formation .
The thiazole tautomer (7) is surprisingly stable and the aromatization reaction to produce the thiazole (8) requires enzymatic catalysis. In B. subtilis, the TenI protein has recently been identified as the thiazole tautomerase .
The structure of the enzyme product (8) complex has been determined (PDB code 3QH2). A model of the enzyme substrate complex generated from this structure suggests that His122 mediates the deprotonation at C2 and that the substrate phosphate group functions as the proton donor for the exocyclic double bond protonation . TenI shows high sequence similarity to thiamin phosphate synthase, and the two enzymes are frequently incorrectly assigned in genome annotation.
Thiamin thiazole biosynthesis in S. cerevisiae
The thiamin biosynthetic pathway in S. cerevisiae is outlined in Figure 4 . The biosynthesis of the thiazole and the pyrimidine heterocycles (5 and 10) occurs by very different chemistry from that used for the bacterial biosynthesis. Labelling studies have demonstrated that the thiazole is formed from an unidentified C5 carbohydrate, glycine (3) and cysteine (11) [34–36] and that the pyrimidine (10) is formed from histidine (48) and PLP (49) [37–39]. Thiamin biosynthesis in yeast requires fewer enzymes than in the bacterial pathway. The biosynthesis of the thiazole requires only one protein (THI4p) in contrast with the bacterial pathway, which requires six (ThiO, ThiF, ThiS, ThiG, IscS and TenI).
All attempts to reconstitute the THI4p-catalysed reaction, using a variety of C5 carbohydrates, initially failed. However, a breakthrough was achieved by the detection of three metabolites (56, 63 and 64 in Figure 5) released from the protein by heat denaturation [40,41].
The identification of product 64 demonstrated that complete thiazole biosynthesis could be achieved using THI4p expressed in Escherichia coli. In addition, this structure demonstrated that the thiazole was adenylated, suggesting that NAD (45), and not a simple pentose, might be the donor of the C5 carbohydrate. Initial attempts to detect Thi4p-catalysed modification of NAD failed. However, after the structure of THI4p was determined (PDB code 3FPZ) , it was possible to prepare an active-site mutant (C204A) that was free of the tightly bound metabolites 56, 63 and 64 . This form of the enzyme catalysed the conversion of NAD (45) and glycine (3) into 56 via intermediates 51 and 52 and confirmed NAD as the C5 carbohydrate donor .
The discovery that metabolite-free THI4p could also be isolated when the E. coli overexpression strain was grown at low iron concentrations provided a source of native enzyme with an unoccupied active site. Treatment of NAD and glycine with this form of the enzyme generated intermediate 56. Addition of Fe(III) to this reaction mixture resulted in the transfer of sulfide from Cys205 of THI4p to generate 63 and 64. MS analysis of the protein in this reaction mixture confirmed Cys205 as the sulfide donor . These observations led to the mechanistic proposal outlined in Figure 5.
In this proposal, hydrolysis of the N-glycosyl bond of NAD (45) gives 51. Ring opening, tautomerization and imine formation give 53. Tautomerization, loss of water and a second tautomerization generate compound 56, the most labile of the three intermediates released in the heat denaturation experiment. Tautomerization to 57 followed by sulfide transfer from Cys205 of THI4p gives 60. Cyclization and two dehydrations give the thiazole tautomer (63), the second of the heat-released metabolites. A final tautomerization completes the thiazole formation. Our mechanism suggests that THI4p may be a single-turnover enzyme. This was confirmed by demonstrating a 1:1 ratio of THI4p to thiamin produced.
In conclusion, we have explored the mechanistic biochemistry of thiamin thiazole biosynthesis in B. subtilis as a representative prokaryote and in S. cerevisiae as a representative eukaryote. The biosynthetic routes are quite different between the two systems, and the reasons for these differences are not yet known. Mechanistic studies on thiazole biosynthesis in bacteria are at an advanced stage, whereas our understanding of the mechanism of thiazole biosynthesis in yeast is still growing, with many unanswered questions remaining. We have not yet identified most of the residues involved in catalysing the conversion of 45 into 64. We also do not yet understand the role of iron in the sulfur transfer or the physiological role of inactive THI4p.
The thiamin project was supported by the Robert A. Welch Foundation [grant number A-0034] and the National Institutes of Health [grant numbers DK44083 (to T.P.B.), DK67081 (to S.E.E.) and GM16609 (to F.W.M.)].
The research described was a collaborative effort between the Begley, Ealick and McLafferty groups. We thank the capable graduate students and postdoctoral associates who carried out all of the experimental work. Begley Group: Dinuka Abeydeera, Alison Backstrom, Kristin Burns, Abhiskek Chatterjee, Pieter Dorrestein, Amrita Hazra, Amy Godert, Neil Kelleher, Cynthia Kinsland, Kalyan Krishnamoorthy, Rung-Yi Lai, Sean O'Leary, Joo-Heon Park and Sean Taylor. Ealick Group: Jessica Chiu, Ying Han, Chris Jurgenson, Christopher Lehmann, Ethan Settembre, Tim Tran and Yang Zhang. McLafferty Group: Sabine Baumgart, Ying Ge, Mi Jin, Neil Kelleher and Huili Zhai. Their individual accomplishments are listed in the references.
Frontiers in Biological Catalysis: Biochemical Society Annual Symposium No. 79 held at Robinson College, Cambridge, U.K., 10–12 January 2012. Organized and Edited by David Leys (Manchester, U.K.), Andrew Munro (Manchester, U.K.), Emma Raven (Leicester, U.K.) and Martin Warren (University of Kent, U.K.).
Abbreviations: DXP, deoxy-D-xylulose 5-phosphate; PLP, pyridoxal 5′-phosphate
- © The Authors Journal compilation © 2012 Biochemical Society