Chapter 43

Biosynthesis of Thiamin

ROBERT L. WHITE and IAN D. SPENSER

INTRODUCTION

The biosynthesis of thiamin (vitamin B₁) (Fig. 1, structure 7) serves as an object lesson in biochemical diversity. The vitamin is generated in most microorganisms, but prokaryotes and eukaryotes utilize different pathways for its construction. Thiamin is composed of two subunits, the pyrimidine moiety, 4-amino-5-hydroxymethyl-2-methylpyrimidine (Fig. 1, structure 1) (referred to here as the pyrimidine unit), and the thiazole moiety, 5-(2-hydroxyethyl)-4-methylthiazole (Fig. 1, structure 3) (referred to here as the thiazole unit). Not only is there diversity in the routes from primary precursors to each of the two subunits, but there are also differences, in bacteria and in yeasts, even in the final stages of the biosynthetic process, that is, in the union of the two subunits to form thiamin pyrophosphate (cocarboxylase) (Fig. 1, structure 8), the coenzyme of transketolase (EC 2.2.1.1), pyruvate decarboxylase (EC 4.1.1.1), and several other enzymes.

Fig. 1Summary of the final stages of the biosynthesis of thiamin (vitamin B1; structure 7) and its pyrophosphate ester (cocarboxylase; structure 8), including the enzymes and genes (E. coli genes unless otherwise indicated) that are implicated in individual steps.

Source:

The designation of vitamin B₁ as thiamin was recommended in 1965 by the International Union of Biochemistry (see IUPAC-IUB Tentative Rules: Vitamins, Coenzymes, etc., Rule M-5 [replacing V-5][27]): "The cation 7, 3-(4-amino-2-methylpyrimidin-5-yl-methyl)-5-(2-hydroxyethyl)-4-methylthiazolium, also known as vitamin B₁, aneurin(e), or thiamine, should be designated thiamin."

It is firmly established that the first step in the formation of thiamin involves the reaction of the pyrimidine pyrophosphate ester (Fig. 1, structure 4) with the thiazole phosphate ester (Fig. 1, structure 5). What has not yet been established with certainty, however, is whether the biosynthesis of the two subunits leads to the free pyrimidine (Fig. 1, structure 1) and the free thiazole (Fig. 1, structure 3) or whether the corresponding phosphate esters (Fig. 1, structures 2 and 5, respectively) are the primary products while the free subunits (structures 1 and 3) represent salvage intermediates.

Thiamin biosynthesis in prokaryotes is emphasized in this chapter, but references to parallel work with other organisms, particularly eukaryotes, are provided in order to make a balanced perspective on the biochemical diversity in thiamin biosynthesis accessible. One reason for the slow progress in the elucidation of the biosynthesis of thiamin was the influence of the dogma of biochemical unity, which led to conceptual difficulties in accepting the notion that pathways to thiamin might be different in different organisms. Only when biochemical diversity in the biosynthesis of thiamin was accepted as a possibility were many apparently contradictory results, which simply could not be accommodated by a single pathway, reconciled. Another formidable obstacle to progress in the elucidation of thiamin biosynthesis is the exceedingly small amount of thiamin synthesized by microorganisms.

We stress the advances that have been made during the past 10 years, that is, since the publication of the first edition of this monograph. That chapter by G. M. Brown and J. M. Williamson (6) provides a detailed summary of the state of our knowledge of the biosynthetic process in prokaryotes in 1984. A review by Young (99) covered the literature to June 1985 for both prokaryotes and eukaryotes.

UNION OF THE TWO SUBUNITS

Figure 1 shows the steps leading from the two subunits to thiamin (structure 7) and its pyrophosphate ester, cocarboxylase (structure 8). This figure summarizes the current state of knowledge of the enzymes and the genes controlling the final phase of thiamin and thiamin pyrophosphate biosynthesis. This information has been repeatedly reviewed (3, 5, 6, 42, 99). Intermediates have been identified, and most of the enzymes controlling the processes have been partially purified. Hydroxymethylpyrimidine kinase (EC 2.7.1.49) from Escherichia coli (51) was recently purified 3,000-fold. Table 1 summarizes the current state of knowledge and provides adequate documentation.

Table 1Summary of the final steps of the biosynthesis of thiamin pyrophosphate: union of the subunits

Genetic analysis by Nakayama and his collaborators of mutant strains of E. coli, each lacking one of the enzymes of thiamin biosynthesis, led to identification of the gene corresponding to thiamin phosphate pyrophosphorylase (thiB) and the genes (thiDKLMN) corresponding to the five kinases that are implicated in the biosynthesis of thiamin pyrophosphate (Fig. 1, structure 8) (25, 26, 30, 50, 51). These six genes are located at four distinct sites on the E. coli genome: 9 min (thiL), 25 min (thiK), 46 min (thiDMN), and 90 min (thiB). The thiK (thiamin kinase) and thiL (thiamin monophosphate kinase) genes have been cloned. E. coli transformed with the cloned genes contained levels of thiamin phosphate 30 times higher than normal (20). Recent studies indicate that thi-33 (thiB) mutants are thz auxotrophs (82). The structural gene for thiamin phosphate pyrophosphorylase in E. coli has been defined as thiE (1). The gene product of thiE, a gene originally assigned to the thiazole pathway (82), has been overexpressed, purified, and shown to catalyze the coupling of structure 4 with structure 5 (Fig. 1) (1). The K_m values for the two substrates were 1 and 2 μM, respectively. In eukaryotes, a gene involved in subunit phosphorylation or coupling, thi4 (101), and a regulatory gene, thi2 (pho6), controlling the expression of acid phosphatase and other thiamin enzymes (32, 55), have been cloned and characterized. Genes coding for thiamin pyrophosphokinase (thi80) (59a) and for the bifunctional enzyme thiamin phosphate pyrophosphorylase/hydroxyethylthiazole kinase (thi6) (60a) have been isolated from Saccharomyces cerevisiae and characterized.

ORIGIN OF THE SUBUNITS

The search for the primary precursors of the pyrimidine and of the thiazole unit of thiamin started 35 years ago. Current work is still at the first stage of biosynthetic investigations, that is, attempting to identify precursor-product relationships. The reader is referred to the original literature for details of the methods that were used to establish the sites of incorporation of the labeled substrates employed to trace the origin of the pyrimidine and the thiazole unit. Only those contributions that in our view gave conclusive results are referred to here.

Although some progress in identifying the primary building blocks of each of the two units has been made, particularly during the past 15 years, no committed intermediate other than 5-aminoimidazole ribotide (Fig. 1, structure 9) has been recognized, and the enzymology of the biosynthetic steps leading to the two subunits is entirely unknown. The first genetic results in this phase of thiamin biosynthesis appeared only recently (82).

Origin of the Pyrimidine Unit

It is now firmly established that prokaryotes and eukaryotes employ entirely different modes of construction of the pyrimidine unit. Formate is the only precursor that is utilized by both systems: it supplies C-2 in prokaryotes (Fig. 2) (16, 38, 94) and C-4 in eukaryotes (10, 21, 81, 94) (but see reference 21).

Fig. 2Biosynthesis of the pyrimidine unit (structure 1) in facultative anaerobic bacteria: precursor-product relationships.

Source:

Glycine as a Precursor.

In a glycine (Fig. 2, structure 10)-requiring mutant of Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium) (53), radioactivity from [1-¹⁴C]- and [2-¹⁴C]glycine entered C-4 (16) and C-6 (12), respectively, of the pyrimidine unit. Confirmation of these results in E. coli came from a mass spectrometric investigation of the incorporation of [1-¹³C]-, [2-¹³C]-, and [¹⁵N]glycine (87). It was inferred, but has not yet been demonstrated by direct experiment, that the C-C-N chain of glycine is incorporated as a unit, giving rise to C-6, C-4, and N-1, respectively, of the pyrimidine. In eukaryotes, glycine is not utilized in the generation of the pyrimidine unit (74, 89, 95).

5-Aminoimidazole Ribotide as a Precursor.

The recognition by Newell and Tucker (54) that, in S. typhimurium, the pyrimidine moiety of thiamin shares precursors and the initial steps of its biosynthetic pathway with the purines (36) and that 5-aminoimidazole ribotide (Fig. 2, structure 9) is the last common intermediate, was a milestone in the elucidation of thiamin biosynthesis. Radioactivity from a ¹⁴C-labeled sample of the corresponding riboside (prepared biosynthetically from [1-¹⁴C]glycine and therefore presumably [5-¹⁴C]-5-aminoimidazole riboside) was incorporated into the pyrimidine in S. typhimurium without change in specific activity (54). Label from [3-¹⁵N]- and [¹⁵NH₂]-5-aminoimidazole riboside entered N-1 and -NH₂, respectively, of the pyrimidine in the same microorganism (14). In E. coli, one of these two nitrogen atoms, N-1, predictably originates from glycine (87), and the other, -NH_2, originates from the amide nitrogen of glutamine, as does the third nitrogen atom, N-3 (74). There is no doubt that the aminoimidazole nucleus of 5-aminoimidazole riboside or ribotide serves as the major source of the pyrimidine nucleus, accounting for all atoms except C-2', i.e., the CH₃ group, and the C₂ unit C-5,5'.

The nucleic acid purines are derived from 5-aminoimidazole ribotide also in Saccharomyces cerevisiae and presumably in eukaryotes in general (36). However, incorporation of labeled glucose (21, 78), histidine (71, 75), and pyridoxol (72, 76) into the pyrimidine unit of thiamin demonstrates that in S. cerevisiae, this nucleus is not derived from 5-aminoimidazole.

Glucose and Ribose as Precursors.

Label from [6-¹⁴C]glucose entered C-5' of the pyrimidine unit of thiamin in E. coli (92, 93), and it was suggested that the C₂ unit C-5,5' of the pyrimidine was derived from C-5,6 of glucose via C-4,5 of ribose, possibly directly from the ribose unit of 5-aminoimidazole ribotide.

That this is indeed the case was shown by Estramareix and David in a series of experiments with Newell and Tucker’s S. typhimurium strain. When the bacterium was incubated with [U-¹³C]glucose together with unlabeled 5-aminoimidazole riboside, label did not enter the pyrimidine unit, but in the converse experiment, with unlabeled glucose and [U-¹³C]5-aminoimidazole riboside, ¹³C entered the pyrimidine. Furthermore, radioactivity from a sample of 5-aminoimidazole [U-¹⁴C]riboside entered the pyrimidine and was present mainly at C-2' and C-5,5', the three carbon atoms that do not originate from the 5-aminoimidazole ring (19). Subsequent work with ¹⁴C-labeled (13) and ¹³C- and ¹⁵N-labeled (14) samples of 5-aminoimidazole riboside confirmed that these three carbon atoms of the pyrimidine originate intramolecularly from three of the five riboside carbon atoms: as suggested earlier, the C₂ unit, C-5,5', of the pyrimidine is derived from the C₂ unit C-4,5 of the ribose moiety, and surprisingly, the CH₃ group, C-2', of the pyrimidine is derived from C-2 of the ribose (Fig. 2). This result points to a most remarkable intramolecular rearrangement of 5-aminoimidazole riboside (or possibly of 5-aminoimidazole ribotide if the primary biosynthetic product is not 4-amino-5-hydroxymethyl-2-methylpyrimidine but the corresponding 5'-phosphate ester).

Origin of the Thiazole Unit

As in the biosynthesis of the pyrimidine unit, distinct sources of the fragments whose union leads to the formation of the thiazole unit in facultative anaerobic bacteria and eukaryotic microorganisms have been established. Some of the available information suggests that aerobic prokaryotes use the same pathway as eukaryotic microorganisms (73; R. L. White, unpublished results), whereas green plants employ that of facultative anaerobic bacteria. In yeasts, glycine (45, 46, 47, 89) and a pentulose (88, 90, 95, 96) are the precursors, whereas in E. coli and S. typhimurium, tyrosine and a deoxypentulose provide the nitrogen and the carbon atoms of the thiazole moiety (Fig. 3).

Fig. 3Biosynthesis of the thiazole unit (structure 3) in faculative anaerobic prokaryotes and in eukaryotes: precursor-product relationships.

Source:

Tyrosine as a Precursor.

[U-¹⁴C] tyrosine labels C-2 and no other carbon atom of the thiazole (17), and the α carbon atom of tyrosine is incorporated into C-2 of the thiazole unit in E. coli (17) and S. typhimurium (2) and presumably into the same site in higher plants (29) (Fig. 3). The nitrogen atom of the thiazole moiety is derived from [¹⁵N]tyrosine in E. coli (73, 86) and Enterobacter aerogenes (73). The CN fragment, α-C,N of tyrosine, probably enters as an intact unit, and the benzylic fragment of tyrosine is probably split off as p-hydroxybenzyl alcohol (Fig. 3, structure 16) (84). In contrast, label from [α-¹⁴C]glycine, the precursor of the N,C-2 fragment of the thiazole in eukaryotes, is not incorporated into the thiazole unit by S. typhimurium (2), and [¹⁵N]glycine is not incorporated into the thiazole unit by either E. coli or Enterobacter aerogenes (73).

Two compounds that are closely related to the thiazole unit, 5-(1,2-dihydroxyethyl)-4-methylthiazole (80) and 5-(2-hydroxyethyl)-4-methylthiazole-2-carboxylic acid (18), have been isolated from E. coli mutants. Their structural similarity and the fact that label from [carboxyl-¹⁴C]tyrosine is incorporated into the latter molecule suggest that steps in the pathway to the thiazole unit are used in their formation. Evidence to support the view that either compound represents a biosynthetic intermediate rather than a dead end byproduct is not available.

Glucose, Glycerol, and Pyruvate as Precursors.

The carbohydrate-derived five-carbon chain of the thiazole unit arises by the union of a C₃ unit (giving rise to C-5,6,7) with a C₂ unit (giving rise to C-4',4). The thiazole unit of thiamin, isolated from an E. coli culture that had been incubated with [U-¹³C₆]glucose, was enriched by two or three ¹³C atoms, and one ¹³C was lost from the ¹³C₃-enriched thiazole moiety when the terminal carbon (C-7) of the hydroxyethyl side chain was removed by fragmentation (83). That C-5 and the hydroxyethyl substituent C-6,7 of the thiazole moiety were derived from carbons 4, 5, and 6 of glucose was further indicated by incorporation of label from [6,6'-²H₂]glucose and [5,6,6'-²H₃]glucose. That it was derived from a C₃ unit was made likely by the incorporation pattern of label observed when (R)-[(RS)-1-²H]glycerol and (R)-[(RS)-1-²H,1-¹⁸O]glycerol (85) served as substrates.

Support for formation of the five-carbon unit from two smaller fragments is provided by the multiple incorporation of deuterium from [1-²H]glucose and by the intact incorporation of the C²H₃ group of [3-²H₃]pyruvate (83). Furthermore, label from [3-¹⁴C]- and [2-¹⁴C]pyruvate was incorporated into the thiazole unit in E. coli (91). [U-¹⁴C]alanine, a close metabolic relative of pyruvic acid, was incorporated in E. coli (91) but not in S. typhimurium (2).

1-Deoxy-D-Xylulose as a Precursor.

Extensive investigations by Serge David and Bernard Estramareix and their coworkers (8, 9, 79) have led to the important discovery that 1-deoxy-d-xylulose (synonym: 1-deoxy-d-threo-pentulose; Fig. 3, structure 15) serves as a precursor of the carbohydrate-derived C₅ chain of the thiazole unit. Deuterium from a sample of 1-deoxy-d-threo-[1-²H₃,5-²H₁]pentulose was delivered into the thiazole unit with retention of deuterium at C-4' and C-7 (8, 9), indicating intact incorporation of the C₅ substrate. The corresponding d-erythro isomer was not incorporated (79).

Formation of the Schiff base of 1-deoxy-d-xylulose with tyrosine (Fig. 3, structure 14) is probably the first step in formation of the thiazole unit in E. coli.

1-Deoxy-d-xylulose has been isolated from bacterial cell extracts supplemented with pyruvate (Fig. 3, structure 12) and d-glyceraldehyde (Fig. 3, structure 13) (97). Pyruvate dehydrogenase (EC 1.2.4.1) from E. coli catalyzes the condensation of pyruvate with d-glyceraldehyde. This process, which is accompanied by decarboxylation, requires thiamin pyrophosphate as a coenzyme (97). The formation of 1-deoxy-d-xylulose in extracts of Bacillus subtilis correlates with pyruvate dehydrogenase activity (98). No product is formed in the presence of bromopyruvate, an inhibitor of pyruvate dehydrogenase (98). 1-Deoxy-d-xylulose is also synthesized, in smaller amounts, by Streptomyces hygroscopicus (24) and by various fungal extracts (97).

Two remarkable facts stand out. First, it is important to note that 1-deoxy-d-xylulose is implicated in the biosynthesis of pyridoxine (see chapter 45 of this volume). Second, if, as seems more than likely, the process catalyzed by pyruvate dehydrogenase plays a crucial part in elaboration of the thiazole unit, then thiamin is a cofactor in its own biosynthetic pathway. Several other B vitamins (e.g., pyridoxine [see chapter 45] and biotin [64]) are also implicated in their own biosynthetic pathways.

Sulfur-Containing Precursors.

Insight into the origin of the sulfur atom has been sought on the basis of incorporation efficiency, a most unreliable criterion when investigations have to be carried out with intact systems. Since sulfur is introduced as an isolated atom, it will not be possible to determine its source with certainty until the enzymology of this process is clarified. An early investigation with E. coli (37) indicated equal efficiency of incorporations of ³⁵S from [³⁵S]cysteine or [³⁵S]methionine, but more recently, higher incorporations have been reported from cysteine in E. coli (77, 91). Cysteine, but not methionine, dilutes the incorporation of label from [³⁵S]sulfate (15) (but see reference 23) and [³⁴S]thiocysteine (11). An investigation of S. typhimurium, on the other hand, showed a reduced incorporation of cysteine in the presence of methionine (1a). It appears to be generally believed that cysteine (Fig. 3, structure 17) is a more direct precursor than methionine.

GENETICS OF STEPS TO SUBUNITS

A cluster of five genes (thiCEFGH), located at 90 min on the E. coli genetic map, was cloned (82; T. P. Begley, personal communication). These genes are associated with the pathways to the individual subunits, since plasmids containing thiC complement Hmp^– mutants (thi mutants that utilize 4-amino-5-hydroxymethyl-2-methyl pyrimidines), whereas Thz^– mutants (mutants that utilize 4-methyl-5-hydroxyethyl thiazole) are complemented by the thiEFGH genes. The genes were sequenced, and polypeptides of 70, 21, 27, 34, and 43 kDa, expressed from thiC, thiE, thiF, thiG, and thiH, respectively, were detected by using polyacrylamide Laemmli gel electrophoresis and an in vivo T7 expression system. A database search revealed that the predicted polypeptide sequences of thiC, thiE, thiG, and thiH did not match any known sequence. However, the predicted amino acid sequence encoded by thiF was 44% homologous with the product of moeB, a protein thought to be involved in sulfur insertion in the biosynthesis of molybdopterin. 1-Deoxy-d-xylulose did not replace the thiazole requirement in the Thz^– mutants, indicating that the gene products of thiEFGH are involved in steps beyond this intermediate on the pathway to the thiazole unit. Another E. coli gene, nuvC, or a closely related gene located between 42 and 46 min, has also been implicated in the biosynthesis of the thiazole unit (63, 82). Factor C, the protein product of nuvC, catalyzes the transfer of sulfur from cysteine to tRNA uridine (63).

Thus, the thiazole pathway in E. coli contains at least four steps between the deoxypentulose and the thiazole. Although polypeptides for individual genes have been expressed, not enough information is available to characterize the steps on the route from primary precursors to the thiazole unit, or to substantiate the identity of the intermediates that have been proposed on the basis of the results of isotopic tracer experiments (see above).

S. typhimurium mutants that are blocked in purF, the gene associated with the first step of the purine biosynthetic pathway, appear to be able to grow without thiamin. This suggests that AIR (structure 9) may be formed by a route other than that of the purine pathway (11c). This alternative route appears to be expressed under anaerobic culture conditions (11a) and used for thiamin biosynthesis when S. typhimurium is supplied with exogenous purines (11b).

To date, a smaller number of genes for the pyrimidine and the thiazole pathways have been found in the higher plant Pisum sativum (thiC and thiA) (39) and in the eukaryotes Schizosaccharomyces pombe (thi2, thi3, and thi4) (49, 65, 66, 100, 101) and Saccharomyces cerevisiae (thi4 [mol1]) (62).

CONCLUSION

Whereas the late stages of thiamin biosynthesis are well documented, the steps on the routes to the pyrimidine and the thiazole unit remain largely unexplored. The results of isotopic tracer experiments have fully established the primary precursors of each of the two units in facultative anaerobic bacteria. Committed intermediates of the pyrimidine and thiazole pathways have not been identified. Because of the minute quantities of thiamin that are biosynthesized in microorganisms, this information will probably be obtained only when overexpressed gene products whose synthetic capabilities can be examined become available.

The search for the genes that control the pathways to the two subunits is at an early stage, and since each of the pathways is likely to consist of several steps, several genes probably remain to be discovered. In E. coli five genes that are implicated in the biosynthesis of the thiazole unit have been located, but only one gene related to the pyrimidine pathway has been located. Even though some of these genes have been cloned and the gene products have been overexpressed, none of the enzymic activities of these proteins have been identified.

It is our view that further progress in the elucidation of the pathways to the two subunits will be very slow if the groups that are studying the genetics, enzymology, biochemistry, and bioorganic chemistry of the pathways continue to work in isolation. Only close collaboration between the groups that are employing these various lines of attack will yield solutions to what has turned out to be a set of difficult problems. It is our hope that this review will serve as a catalyst for such collaboration.

ACKNOWLEDGMENTS

We acknowledge with thanks research grants from the National Institute of General Medical Sciences, U.S. Public Health Service (GM50778) (to I.D.S.), and from the Natural Sciences and Engineering Research Council of Canada (to R.L.W.).