The structural characterization of proteins expressed from the genome is a major problem in proteomics. The solution to this problem requires the separation of the protein of interest from a complex mixture, the identification of its DNA-predicted sequence, and the characterization of sequencing errors and posttranslational modifications. For this, the "top down" mass spectrometry (MS) approach, extended by the greatly increased protein fragmentation from electron capture dissociation (ECD), has been applied to characterize proteins involved in the biosynthesis of thiamin, Coenzyme A, and the hydroxylation of proline residues in proteins. With Fourier transform (FT) MS, electrospray ionization (ESI) of a complex mixture from an E. coli cell extract gave 102 accurate molecular weight values (2-30 kDa), but none corresponding to the predicted masses of the four desired enzymes for thiamin biosynthesis (GoxB, ThiS, ThiG, and ThiF). MS/MS of one ion species (representing approximately 1% of the mixture) identified it with the DNA-predicted sequence of ThiS, although the predicted and measured molecular weights were different. Further purification yielded a 2-component mixture whose ECD spectrum characterized both proteins simultaneously as ThiS and ThiG, showing an additional N-terminal Met on the 8 kDa ThiS and removal of an N-terminal Met and Ser from the 27 kDa ThiG. For a second system, the molecular weight of the 45 kDa phosphopantothenoylcysteine synthetase/decarboxylase (CoaBC), an enzyme involved in Coenzyme A biosynthesis, was 131 Da lower than that of the DNA prediction; the ECD spectrum showed that this is due to the removal of the N-terminal Met. For a third system, viral prolyl 4-hydroxylase (26 kDa), ECD showed that multiple molecular ions (+98, +178, etc.) are due to phosphate noncovalent adducts, and MS/MS pinpointed the overall mass discrepancy of 135 Da to removal of the initiation Met (131 Da) and to formation of disulfide bonds (2 x 2 Da) at C32-C49 and C143-C147, although 10 S-S positions were possible. In contrast, "bottom up" proteolysis characterization of the CoaBC and the P4H proteins was relatively unsuccessful. The addition of ECD substantially increases the capabilities of top down FTMS for the detailed structural characterization of large proteins.

The thiO gene of Bacillus subtilis encodes an FAD-dependent glycine oxidase. This enzyme is a homotetramer with a monomer molecular mass of 42 kDa. In this paper, we demonstrate that ThiO is required for the biosynthesis of the thiazole moiety of thiamin pyrophosphate and describe the structure of the enzyme with N-acetylglycine bound at the active site. The closest structural relatives of ThiO are sarcosine oxidase and d-amino acid oxidase. The ThiO structure, as well as the observation that N-cyclopropylglycine is a good substrate, supports a hydride transfer mechanism for the enzyme. A mechanistic proposal for the role of ThiO in thiazole biosynthesis is also described.

While most of the proteins required for the biosynthesis of thiamin pyrophosphate have been known for more than a decade, the reconstitution of this biosynthesis in a defined biochemical system has been difficult due to the novelty of the chemistry involved. Here we demonstrate the first successful enzymatic synthesis of the thiazole moiety of thiamin from glycine, cysteine, and deoxy-d-xylulose-5-phosphate using overexpressed Bacillus subtilis ThiF, ThiS, ThiO, ThiG, and a NifS-like protein. This has facilitated the identification of the biochemical function of each of the proteins involved:  ThiF catalyzes the adenylation of ThiS; NifS catalyzes the transfer of sulfur from cysteine to the acyl adenylate of ThiS; ThiO catalyzes the oxidation of glycine to the corresponding imine; and ThiG catalyzes the formation of the thiazole phosphate ring. The complex oxidative cyclization reaction involved in the biosynthesis of the thiamin thiazole has been greatly simplified by replacing ThiF, ThiS, ThiO, and NifS with defined biosynthetic intermediates in a reaction where ThiG is the only required enzyme.

Bacillus subtilis gene products TenA and TenI have been implicated in regulating the production of extracellular proteases, but their role in the regulation process remains unclear. The structural characterization of these proteins was undertaken to help provide insight into their function. We have determined the structure of TenA alone and in complex with 4-amino-2-methyl-5-hydroxymethylpyrimidine, and we demonstrate that TenA is a thiaminase II. The TenA structure suggests that the degradation of thiamin by TenA likely proceeds via the same addition-elimination mechanism described for thiaminase I. Three active-site residues, Asp44, Cys135, and Glu205, are likely involved in substrate binding and catalysis based on the enzyme/product complex structure and the conservation of these residues within TenA sequences. We have also determined the structure of TenI. Although TenI shows significant structural homology to thiamin phosphate synthase, it has no known enzymatic function. The structure suggests that TenI is unable to bind thiamin phosphate, largely resulting from the presence of leucine at position 119, while the corresponding residue in thiamin phosphate synthase is glycine.

Two Bacillus subtilis genes encoding two proteins (currently annotated ThiD and YjbV) were overexpressed and characterized. YjbV has 4-amino-5-hydroxymethyl-2-methylpyrimidine and 4-amino-5-hydroxymethyl-2-methylpyrimidine pyrophosphate kinase activity and should be reannotated ThiD, and B. subtilis ThiD has pyridoxine, pyridoxal, and pyridoxamine kinase activity and should be reannotated PdxK.