Thymidylate synthase (TS) shares with ribonucleotide reductase the distinction of being responsible for one of the two chemical differences between DNA and RNA. Just as ribonucleotide reductase carries out the production of deoxyribose at the nucleoside di- or triphosphate level, TS generates the methyl group of thymine, acting at the deoxyribonucleoside 5′-monophosphate level:
![tmp10D-28_thumb[1]](http://what-when-how.com/wp-content/uploads/2011/05/tmp10D28_thumb1.jpg "tmp10D-28_thumb[1]"){kind=small}
This is the only known biological methylation reaction not involving S-adenosylmethionine or the synthesis of methionine itself. Aside from this distinctive role, the enzyme is of interest from several standpoints—(1) mechanistically, (2) as the first enzyme identified as an anticancer drug target, (3) as one of the most active current targets for computer-assisted drug design, (4) as the first prokaryotic protein shown to result from processing of a split gene, (5) as an agent in translational autoregulation, and (6) as a component of multifunctional proteins and multienzyme complexes.
1. A Target for Fluorinated Pyrimidines
In 1957, Friedkin and Kornberg (1) described a tetrahydrofolate-dependent enzyme that converts deoxyuridine monophosphate (dUMP) to thymidine monophosphate (dTMP) (Eq. (1)). The methyl group of dTMP was shown to originate as the methylene carbon of 5,10-methylenetetrahydrofolate (CH2 = THF). This discovery of thymidylate synthase was intertwined with the early work of Heidelberger, who synthesized 5-fluorouracil and 5-fluorodeoxyuridine as potential anticancer drugs (2). Heidelberger had noted that tumor cells metabolize uracil much more rapidly than normal cells do, and he expected that substitution of fluorine for a similarly sized hydrogen atom might inhibit the cellular uptake and usage of uracil selectively in tumor cells. In 1958, the action of both fluorinated pyrimidines was shown by Cohen et al (3) to result from their conversion in vivo to the deoxyribonucleoside monophosphate 5-fluorodeoxyuridylate (FdUMP), which they showed to act as a potent inhibitor of TS. Fluorine has a van der Waals radius close to that of hydrogen, for which it is substituted in the analogues. This substitution creates a kinetically irreversible inhibition, which eventually was shown to result from formation of a covalently bonded enzyme-inhibitor complex. This is a ternary complex, whose formation requires CH2 = THF as well.
These factors suggested that the structure of this complex would yield clues to the mechanism of action of the enzyme.
Another important early development was the demonstration by Friedkin and colleagues that the folate coenzyme serves as the agent, not only for transfer of a single-carbon functional group but also for reduction of that group from the methylene to the methyl level (4). Radioisotope labeling studies showed that both the methylene group and a hydride ion—the hydrogen linked to C6 of CH2 = THF—were transferred to dTMP essentially without dilution. This is the only known reaction
in which tetrahydrofolate serves as a redox cofactor. These findings suggested a bridged intermediate in which the methylene carbon of CH2 = THF is linked transiently to C5 of the pyrimidine ring. Inhibition by FdUMP was postulated to result from formation of a similar complex, with the fluorine atom acting to prevent cleavage of the N5-methylene carbon bond.
But how does FdUMP bind covalently to the enzyme? Degradation of the ternary complex revealed a cysteine sulfur atom linked to C6 of the pyrimidine ring (reviewed in 5). These findings suggested a mechanism for the reaction in which the cysteine thiol group initiates nucleophilic attack on C6, converting C5 into a nucleophile that attacks C11, the methylene carbon of 5,10-methylenetetrahydrofolate. Much evidence now supports the mechanism that is outlined in Figure 1.
Figure 1. A plausible mechanism for the thymidylate synthase reaction, based on radiolabeling studies and the structure of the FdUMP—CH2 = THF—TS ternary complex. In this complex, the C—F bond cannot be broken, which blocks the reaction at the stage of conversion of III to IV.
2. Structure of Thymidylate Synthase
TS from nearly all sources investigated is a homodimeric protein with a protomer size of 30 to 40 kDa. The enzyme is one of the most highly conserved proteins known, with some 18% of its residues invariant among the two dozen known sequences. The first X-ray crystallography structure reported was of the Lactobacillus casei TS (6). The structure shows two substrate-binding clefts, each comprising residues from both polypeptide chains. One such cleft is shown in Figure 2a, which also depicts conserved residues found within the cleft. Cys198 is the initiating nucleophile, and the nearby Arg218 is thought to lower the pKa of Cys198, converting the thiol group to the reactive thiolate ion. Arg179′, from the other polypeptide chain, is thought to interact with the phosphate group on the substrate. Several nearby lysine residues (not shown) are thought to interact with the polyglutamate tail of the folate cofactor, which explains the observation that thymidylate synthase binds folate polyglutamates about 100-fold more tightly than the monoglutamate.
Figure 2. The crystal structure of thymidylate synthase. (a) One subunit of the L. casei TS dimer, showing (with numbers) conserved residues in the active site. Reproduced with permission from L. W. Hardy, J. S. Finer-Moore, W. R. Montfort, M. O. Jones, D. V. Santi, and R. M. Stroud (1987) Science 235, 448-455. (b) One subunit of the ternary complex between E. coli TS, FdUMP, and CH2 = THF. The ligands are bold, with the folate cofactor above and dUMP below.
Several other crystal structures are now known for thymidylate synthase. Figure 2b shows the structure of the FdUMP—CH2 = THF—enzyme ternary complex for one subunit of Escherichia coli thymidylate synthase. Detailed analysis of this structure (7) and others confirmed the interactions postulated from studies of the free enzyme, and it also helped to resolve some stereochemical anomalies of the reaction that had been revealed by radioisotope labeling experiments.
The availability of several TS structures, including that of the human enzyme, plus the attractiveness of TS as a chemotherapeutic target, has made this enzyme one of the most active subjects of computer-assisted drug discovery research; compounds are being sought that bind as tightly as FdUMP but with greater selectivity and/or lower toxicity. In one approach (8), a docking program was used to identify known compounds that bind tightly in the TS active site. Phenolphthalein, a compound with no evident similarity to substrate or cofactor, was found to bind with an affinity in the micromolar range. Another approach (9) used computational analysis of the binding site to predict structures that would bind tightly in the cofactor site. One such compound, shown below, has no evident structural relationship to methylenetetrahydrofolate, but it binds the enzyme with an inhibition constant (K;) of around 30 nM.
Although the thymidylate synthase molecule contains two structurally identical substrate binding sites, considerable evidence indicates that TS displays half-of-the-sites reactivity, ie, that just one of the two sites is catalytically active. Consistent with this, Maley et al (10) found that simply mixing two inactive mutant forms of E. coli TS led to restoration of full wild-type activity, suggesting that mutant subunits had exchanged. Because one of the mutations affected the critical active-site cysteine residue (Cys146 in E. coli), only one functional active site per enzyme molecule was evidently enough to give full activity.
3. Thymidylate Synthase and Prokaryotic Introns
In bacteriophage T4, a virus-specific TS is encoded by gene td. In 1984, biologists were startled to learn that split genes were not confined to the eukaryotic kingdom, when Chu et al (11) showed that the T4 td gene contains a 1-kbp intron. Further analysis showed this and a handful of other prokaryotic introns to be of the Group I, self-splicing type. These and subsequent findings (12) have contributed to a lively debate over whether introns are evolutionarily ancient or recent.
4. Thymidylate Synthase and Translational Autoregulation
In eukaryotic cells, thymidylate synthase gene expression is closely aligned with the cell cycle, a fact that has led to detailed investigations of TS promoter structure and gene regulation. What is most novel about TS gene regulation, however, is that it is one of the first eukaryotic genes shown to be autoregulated at the level of translation (see Translation Repressors). TS protein has been shown in vitro to bind specifically to its own mRNA and to prevent translation of that mRNA (13). The ability of TS protein to inhibit its own synthesis was blocked by TS substrates or by FdUMP. The physiological significance of this autoregulation is not yet known.
5. Multienzyme Complexes and Multifunctional Proteins
Thymidylate synthase is associated structurally with functionally related enzymes. This was first shown in bacteriophage T4, where at least 10 enzymes of DNA precursor biosynthesis interact to form a complex, called dNTP synthetase, which evidently facilitates the flow of DNA precursors into DNA. Within this complex, an especially strong interaction links thymidylate synthase with the phage-coded deoxycytidylate deaminase, the source of most of the dUMP that is used by TS (14). Similar complexes have been described in eukaryotic cells although a role in coupling dNTP synthesis to DNA replication has not yet been established. Surprisingly, of those TS molecules that have been shown to be localized within the nucleus (a prerequisite for the coupling of dNTP synthesis to replication), most are located in the nucleolus (15), a finding that suggests possible additional metabolic roles for the TS protein.
Quite different is the situation in protozoa and some plants, where TS exists as a bifunctional protein also containing dihydrofolate reductase activity. Kinetic analysis of the Leishmania DHFR-TS (16) shows that the dihydrofolate released by the TS reaction is channeled to the DHFR active site— transported from site to site without release from the enzyme surface—even though the active sites are 40 A apart. Elcock et al (16) have ascribed this unusual behavior to the charge distribution on the enzyme surface. Because of its unusual structure and function, the bifunctional DHFR-TS is being actively studied as a target for antiparasitic drugs.
