Identification of the Two Missing Bacterial Genes Involved in Thiamine Salvage: Thiamine Pyrophosphokinase and Thiamine Kinase

Abstract

The genes encoding thiamine kinase in Escherichia coli (ycfN) and thiamine pyrophosphokinase in Bacillus subtilis (yloS) have been identified. This study completes the identification of the thiamine salvage enzymes in bacteria.

Thiamine pyrophosphate biosynthesis occurs by a complex multistep pathway (2). Therefore, bacteria have evolved a set of salvage kinases to utilize dephosphorylated intermediates present in growth medium (Fig. 1). Pyrimidine alcohol and thiazole alcohol salvage requires pyrimidine alcohol kinase (ThiD) and thiazole kinase (ThiM), two previously characterized enzymes (3, 4, 7, 9). Thiamine salvage occurs in two steps, with thiamine kinase (5) catalyzing the formation of thiamine phosphate, and thiamine phosphate kinase (ThiL) (12), a biosynthetic enzyme, catalyzing the conversion of this intermediate to thiamine pyrophosphate. An alternative to this two-step thiamine salvage pathway has been found in Saccharomyces cerevisiae, where thiamine pyrophosphokinase (THI80) catalyzes the direct conversion of thiamine to thiamine pyrophosphate (1, 6). This single-step thiamine salvage pathway was previously thought to occur only in eukaryotes. Here we describe the characterization of the genes encoding thiamine kinase from Escherichia coli (ycfN) and thiamine pyrophosphokinase from Bacillus subtilis (yloS). This study reveals different thiamine salvage strategies in the two bacteria and completes the identification of the thiamine salvage enzymes in bacteria.

FIG. 1.

Thiamine salvage pathway in E. coli and B. subtilis. The thiamine kinase and thiamine pyrophosphokinase described in this paper are shown in bold. Thiamine kinase is found in E. coli but not in B. subtilis, and thiamine pyrophosphokinase is found in B. subtilis but not in E. coli. HMP, pyrimidine alcohol; Thi-PP, thiamine pyrophosphate; OP, phosphate; OPP, pyrophosphate.

The thiamine kinase gene in E. coli was previously mapped between fabD and purB and was subcloned on a 2.6-kb fragment containing ClaI, MluI, SalI, HpaI, and NruI restriction sites (T. Fujio, M. Hayashi, A. Iida, T. Nishi, and T. Hagihara, 1991, Thiamine phosphates and their enzymic manufacture using recombinant Escherichia, European Patent Application EP 417953 A1). Of these, the MluI, SalI, and HpaI sites were contained within the gene. This information was sufficient to identify a region on the E. coli chromosome that contained a ClaI site, two closely spaced MluI sites, a SalI site, two closely spaced HpaI sites, and an NruI site on a 2.9-kb region (bases 1161161 to 1164160). The MluI, SalI, and HpaI sites were all within the ycfN gene, strongly suggesting that ycfN is the thiamine kinase gene. A protein BLAST search of the YcfN sequence was uninformative, indicating only weak sequence similarity to a glycosidase.

The E. coli ycfN gene was readily overexpressed in E. coli as a His₁₀-tagged 34-kDa protein. This protein was present primarily in inclusion bodies, was unstable, and lost all activity during purification (Fig. 2, lanes 1 to 3). Assays to test the functional prediction were therefore carried out using cell extract and measuring thiamine phosphate formation after its conversion to the highly fluorescent thiochrome phosphate. High-performance liquid chromatography (HPLC) analysis of this reaction mixture (8) clearly demonstrated that YcfN catalyzed the phosphorylation of thiamine (37% conversion of a 3.3 μM solution in 1 h), while cell extract from the host strain did not show this activity.

FIG. 2.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of YcfN and YloS overexpression. Lane 1, total protein from the cell extract of the YcfN overexpression strain before induction; lane 2, total protein from the cell extract of the YcfN overexpression strain after induction; lane 3, soluble protein from the cell extract of the YcfN overexpression strain after induction; lane 4, total protein from the cell extract of the YloS overexpression strain before induction; lane 5, total protein from the cell extract of the YloS overexpression strain after induction; lane 6, soluble protein from the cell extract of the YloS overexpression strain after induction; lane 7, YloS after purification on a HisBind column.

B. subtilis can also salvage thiamine but does not contain a YcfN ortholog. The absence of this ortholog suggests that this microorganism may utilize a different strategy for thiamine salvage such as that found, for example, in S. cerevisiae, where thiamine is directly pyrophosphorylated (6). A protein BLAST search, using the S. cerevisiae thiamine pyrophosphokinase amino acid sequence (THI80), revealed that B. subtilis contains a protein (YloS) of previously unidentified function with low similarity to this sequence (17% sequence identity over 262 residues). To test the function of YloS, the gene was overexpressed in E. coli as a His₆-tagged 24.5-kDa soluble protein and readily purified on a HisBind resin (Fig. 2, lanes 4 to 7). YloS was assayed for thiamine pyrophosphokinase activity by treating thiamine and ATP with the enzyme and assaying for thiamine pyrophosphate formation after its conversion to the highly fluorescent thiochrome pyrophosphate. HPLC analysis of the reaction mixture clearly demonstrated the presence of thiamine pyrophosphate (30% conversion of a 300 μM solution in 1 h). A continuous assay (9), based on the coupling of AMP production to NADH consumption, was used to determine the kinetic parameters of the enzyme (k_cat, 0.2 s⁻¹; K_m of thiamine, 20 μM; K_m of ATP, 1 mM).

The functional assignment of yloS and ycfN completes the identification of the thiamine salvage genes in bacteria. These genes show a surprisingly narrow distribution. YloS orthologs are found only in bacilli and lactobacilli, and YcfN orthologs are found only in the γ-proteobacteria. Neither gene clusters with other thiamine biosynthetic genes. Based on these functional assignments, we propose that YcfN be renamed ThiK and that YloS be renamed ThiN.

Standard methods were used for DNA restriction endonuclease digestion, ligation, and transformation (11). PCR-derived DNA was sequenced and shown to contain no errors. The yloS gene was PCR amplified from B. subtilis 168 genomic DNA and overexpressed using pET28a in E. coli DH5α grown at 37°C in Luria-Bertani medium. The overexpressed protein was soluble and was purified on a HisBind column (Novagen). The ycfN gene was PCR amplified from E. coli genomic DNA and overexpressed using pET-16b(+) in E. coli DH5α grown at 37°C in Luria-Bertani medium.

YcfN and YloS were assayed for thiamine kinase and thiamine pyrophosphokinase activity, respectively, with the thiochrome assay (8, 10). For the thiamine pyrophosphokinase assay, YloS (51 μg, pure), thiamine alcohol (300 μM), ATP (300 μM), MgCl₂ (2.2 mM), and KCl (11 mM) were incubated in 100 μl of 100 mM Tris HCl (pH 8) buffer. For the thiamine kinase assay, YcfN cell extract (10 μl of a 4.8-mg/ml solution) was added to a solution of thiamine (3 μM) and ATP (5 mM) in 290 μl of 0.1 M Tris (pH 8)-5 mM MgCl₂. The reaction mixtures were incubated at 37°C, and 50-μl aliquots were removed after 0 and 60 min and quenched with 50 μl of 10% trichloroacetic acid. The denatured protein was removed by centrifugation (16,000 × g, 3 min), and 50 μl of 4 M potassium acetate followed by 50 μl K₃Fe(CN)₆ (30 mg/ml in 7 M NaOH) was added to the supernatant. The resulting solution was mixed by vortexing. After 1 min, 58 μl of 6 M HCl was added to quench the reaction. The resulting mixture was analyzed by HPLC (8).

The kinetic parameters for YloS were determined by using a coupled assay involving NADH consumption (9). A stock solution containing KCl (112 mg), MgCl₂ (61 mg), NADH (4.3 mg), phosphoenolpyruvate (5.7 mg), pyruvate kinase (2.3 mg), lactate dehydrogenase (1.1 mg), and myokinase (1 mg) in 15 ml of 100 mM Tris HCl (pH 8) was prepared. Aliquots (250 μl) of this solution were placed in a quartz cuvette and diluted with 250 μl of 100 mM Tris HCl (pH 8) containing various concentrations of ATP and thiamine alcohol. The reaction was initiated by adding YloS (5 μl of a 10.1-mg/ml solution), and the consumption of NADH was measured by monitoring the decrease in absorbance at 340 nm (ɛ_340,NADH = 4.3 cm⁻¹ mM⁻¹; 2 NADH were consumed for each thiamine pyrophosphate produced).