Accelerated Publications Mechanistic Studies of a Flavin-Dependent Thymidylate Synthase†

The ThyA gene that encodes for thymidylate synthase (TS) is absent in the genomes of a large number of bacteria, including several human pathogens. Many of these bacteria also lack the genes for dihydrofolate reductase (DHFR) and thymidine kinase and are totally dependent on an alternative enzyme for thymidylate synthesis. Thy1 encodes flavin-dependent TS (FDTS, previously denoted as TSCP) and shares no sequence homology with classical TS genes. Mechanistic studies of a FDTS from Thermotoga maritima (TM0449) are presented here. Several isotopic labeling experiments reveal details of the catalyzed reaction, and a chemical mechanism that is consistent with the experimental data is proposed. The reaction proceeds via a ping-pong mechanism where nicotinamide binding and release precedes the oxidative halfreaction. The enzyme is primarily pro-R specific with regard to the nicotinamide (NADPH), the oxidation of which is the rate-limiting step of the whole catalytic cascade. An enzyme-bound flavin is reduced with an isotope effect of 25 (consistent with H-tunneling) and exchanges protons with the solvent prior to the reduction of an intermediate methylene. A quantitative assay was developed, and the kinetic parameters were measured. A significant NADPH substrate inhibition and large KM rationalized the slow activity reported for this enzyme in the past. These and other findings are compared with classical TS (ThyA) catalysis in terms of kinetic and molecular mechanisms. The differences between the FDTS proposed mechanism and that of the classical TS are striking and invoke the notion that mechanism-based drugs will selectively inhibit FDTS and will not have much effect on human (and other eukaryotes) TS. Since TS activity is essential to DNA replication, the unique mechanism of FDTS makes it an attractive target for antibiotic drug development. Deoxythymidine 5′-monophosphate (dTMP)1 is one of the four building blocks of DNA. Until recently, dTMP in all cellular organisms was thought to be formed de novo by thymidylate synthase (TS, encoded by ThyA). TS catalyzes the reductive methylation of deoxyuridine 5′-monophosphate (dUMP) to form dTMP (1). Classical TSs use R-N5,N10† This work was supported by NIH Grant R01 GM65368-01 and NSF Grant CHE-0133117 to A.K. and the NIH NIGMS Protein Structure Initiative (Grant GM59965-02) to P.K. * Corresponding author. Tel: 319-335-0234. Fax: 319-335-1270. E-mail: [email protected]. § University of Iowa. ‡ Genomics Institute of the Novartis Research Foundation. ⊥ The Scripps Research Institute. 1 Abbreviations: FDTS, flavin-dependent thymidylate synthase (previously denoted as TSCP); KIE, kinetic isotope effect; RP HPLC, reverse phase high-pressure liquid chromatography; LSC, liquid scintillation counter; dUMP, 2′-deoxyuridine-5′-monophosphate; dTMP, 2′deoxythymidine-5′-monophosphate; CH2H4folate, R-N5,N10-methylene5,6,7,8-tetrahydrofolate; H2folate, 7,8-dihydrofolate; H4folate, 5,6,7,8tetrahydrofolate; FAD, flavin adenine dinucleotide; Mdpm, million decompositions per min. © Copyright 2004 by the American Chemical Society Volume 43, Number 32 August 17, 2004 10.1021/bi0490439 CCC: $27.50 © 2004 American Chemical Society Published on Web 07/20/2004 methylene-5,6,7,8-tetrahydrofolate (CH2H4folate) both as a methylene donor and as a reductant (hydride donor) leading to 7,8-dihydrofolate (H2folate) formation (2). Because 5,6,7,8tetrahydrofolate (H4folate) and its derivatives are essential for a variety of biological processes, H2folate formed by TS is rapidly reduced to H4folate by dihydrofolate reductase (DHFR). This coupling of TS and DHFR proteins was thought to be essential for de novo thymidylate synthesis in virtually all dividing cells (1). Recently, several organisms were identified that lack ThyA in their genome (3). Most of these also lack genes for DHFR and thymidine kinase (an enzyme that enables salvage of thymidine derivatives from the growth media), suggesting that an alternative TS activity must be present. A new family of genes was identified that encodes for enzymes that convert dUMP to dTMP via an alternative pathway (4). These genes are called Thy1 (or ThyX) and are widely distributed in bacterial and archaeal organisms including several human pathogenic bacteria (3, 4). This alternative TS is named TS complementing protein (TSCP) (3) or flavin-dependent TS (5) (denoted below as FDTS). Since it has no sequence homology with TS, drugs targeting FDTS may be specific to those pathogens with little, if any, toxic effect in humans. Development of such drugs would greatly benefit from understanding of the FDTS catalytic mechanism, the reactants binding order, and identification of the rate-limiting step, intermediates, and possible transition states (of which analogues may lead to the most promising inhibitors). Classical TSs have been extensively studied in terms of structure, kinetics, and mechanism. They are among the most highly conserved enzymes with approximately 18% of residues strictly conserved (2). Numerous X-ray structures of free TS and bound enzyme-substrate-cofactor analogues have been determined, and several hundred mutants have been cloned and studied (2). The mechanism of classical TS is relatively well understood, and mechanism-based drugs are in extensive clinical use (e.g., 5-fluorouracil). The indepth studies included investigation of environmentally coupled hydrogen tunneling in the hydride transfer step (6) and the functional dynamics of the protein (7). The mechanism of the FDTS, on the other hand, is not yet well understood (3, 4). The crystal structure of FDTS from Thermotoga maritima (TM0449) was solved by Kuhn et al. in 2002 (8). This is a tetramer with four identical subunits with a molecular weight of 26005 Da each (220 amino acid residues), and no structural similarities are apparent between FDTS and classical TS. Recently, Mathews et al. (3) have shown an absolute activity dependency on flavin adenine dinucleotide (FAD), CH2H4folate, and reduced nicotinamide adenine dinucleotide (NAD(P)H). The activity assay used by Mathews et al. was based on a qualitative thin-layer chromatography (TLC) analysis of [2-14C]-dUMP. Yet that assay indicated that NADPH enhances the enzyme activity better than NADH. Accordingly, NADPH was used in the current work. In ref 3, eight crystal structures of TM0449 were solved with several ligands (including FAD, dUMP, and 5-fluoro-dUMP). These structures clearly identified the FAD and dUMP binding sites. Structural features that directly affect the current study include the following: (i) the adenine-pyrophosphate moiety of the FAD is anchored (noncovalently) deep inside the protein, while its reactive isoalloxazine ring is exposed to the protein surface and appears to be fairly flexible; (ii) there is no space for the adenine-pyrophosphate moiety of NADPH in the monophosphate ribose binding pocket of dUMP; (iii) the (p-aminobenzoyl)-glutamate moiety of CH2H4folate has no apparent binding pocket close to the dUMP with or without bound FAD; (iv) the only nucleophile in the active site is Ser88, which is about 4 Å from the flexible pyrimidine ring of dUMP. This serine is at position 84 of FDTS from Helicobacter pylori for which either the mutants at Ser84 are inactive or their activity is partly rescued by Ser85 (for which glycine is present for the TM0449 enzyme), indicating a critical role of this serine in catalysis (5). The analogy that springs to mind is the dichotomy between serine proteases and cysteine proteases: the OH and SH nucleophile contrast is present there too. In this paper, we present an experimental analysis of isotopically labeled reactants and products and a spectroscopic investigation of the TM0449-encoded FDTS. The experimental data and computer modeling shed light on the FDTS-catalyzed reaction and suggest a kinetic and chemical mechanism for the function of the enzyme. The stereospecificity, H-tunneling contribution, and other mechanistic features of the enzyme are also discussed and compared to the classical TS (ThyA). MATERIALS AND METHODS