Nature often adopts several approaches to crack the same problem. The finding that the mechanism of a crucial enzyme in certain disease-causing bacteria differs from that in mammals offers scope for drug discovery.
On page 919 of this issue, Koehn et al.1 propose that, in certain microorganisms, a previously unknown biochemical mechanism underpins the function of an enzyme that is essential to the microorganisms' survival. In mammals, the activity of this enzyme — thymidylate synthase — depends on an 'anchor' in its active site that binds covalently to the enzyme's substrate. But the authors find that, in some microbes (including many that threaten human life), thymidylate synthase is active in the absence of such an anchor. The mechanism that explains this behaviour is a potential target for antibiotic drugs that would be toxic to microbes, but not to humans.
This difference1 between taxonomic groups is a clear example of how some cells evolved to have well-developed enzyme mechanisms that have high energy costs (as in mammalian thymidylate synthase), whereas others make do with less-specialized mechanisms that have lower energy costs (as in the microbial enzyme). Thymidylate synthase produces a deoxythymidine nucleotide (dTMP), which is necessary for DNA synthesis. Classic biochemical2 and proteomic studies3 have clearly shown that there are two kinds of thymidylate synthase, each having distinct evolutionary origins (based on their different mechanisms of action and structures). Those found in humans and other mammals are known as TS enzymes, whereas the other group, found in 30% of microbial genomes4, is known as the FDTS family of enzymes.
The mechanistic differences between the two groups hinge on the cofactors required and on the reactions that occur between the enzymes and their substrate, a deoxyuridine nucleotide (dUMP). In mammalian TS, synthesis of dTMP begins when a cysteine amino-acid residue in TS forms a covalent bond to a specific carbon in dUMP (see Fig. 1 here, and Fig. 1a on page 920). This bond anchors the substrate to the enzyme. In the next step, a carbon atom is transferred from a cofactor to the carbon adjacent to the anchor. This is a high-energy process, which occurs only because the anchor aligns the reacting groups perfectly for reaction. In the final step, the newly attached carbon atom is converted into a methyl group and the anchor breaks, releasing the resulting dTMP product.
But Koehn et al.1 have found that microbial FDTS enzymes do not covalently anchor dUMP, reducing the energy cost of their reactions. To prove this, they performed conceptually simple experiments on the active site of FDTS from the microbe Thermotoga maritima. It had previously been thought that a serine amino-acid residue in the active site acts as an anchor for dUMP, in the same way as a cysteine residue does in TS enzymes. The authors therefore mutated the serine to alanine, whose side chain is incapable of reacting with dUMP to form a covalent anchor. The mutant FDTS retained its activity, thus showing that substrate anchoring is not necessary to drive the enzyme's reaction.
The authors also made a serine-to-cysteine mutant, which was expected to be active by analogy with the TS enzymes. In fact, it was less active than the non-mutated enzyme. A crystal structure1 of the mutant revealed that the cysteine residue does not form a covalent bond to dUMP, yet Koehn et al. obtained other evidence suggesting that the cysteine does form a complex with the substrate. Taken together, these results suggest that the observed cysteine–dUMP complex is a dead end that doesn't form part of the FDTS catalytic cycle. These data serve as a reminder that, although mutagenesis experiments are often very useful, local changes to protein structures can translate into mostly unpredictable long-range effects. Indeed, previous mutagenesis experiments4,5on the FDTS of the bacterium Helicobacter pylori produced misleading evidence about the mechanism of action of the enzymes.
The reactivity of the dUMP molecule (specifically, of its uracil nucleotide base) in the FDTS active site is very different from its reactivity in the TS active site, providing another indication of differences between the two classes of enzymes. Koehn et al.1 propose that a third component of FDTS reactions, the 'FADH2 cofactor', is responsible for this difference. They claim that an unusual reaction occurs in which a hydrogen atom is transferred from FADH2 to a carbon in the uracil base — the same carbon that is anchored covalently in TS enzymes (see Fig. 1c on page 920).
The authors found evidence for this theory by introducing hydrogen isotopes into the various components of the FDTS reaction, and following the movements of the isotopes as the reaction progressed. They thus showed that FADH2 is the source of the hydrogen atom that ends up in the uracil core of dUMP. Further evidence came from the authors' X-ray crystal structures of FDTS in complex with analogues of dUMP and with FAD (the side product formed from FADH2). The distance between the active-site serine and the uracil carbon atom is too great to allow a covalent bond to form between them (as occurs in TS enzymes). Instead, the FAD molecule lies close to the reactive carbon atom, as would be expected if the FADH2 cofactor transfers a hydrogen atom to dUMP's uracil. The complete picture emerging from Koehn and colleagues' multidisciplinary work convincingly shows that the reaction pathway occurring in certain microbial thymidylate synthases is different from that occurring in mammals.
The unravelling of the entire FDTS pathway through to the end products remains incomplete, and other loose ends must still be tied up — for example, the formation of the 'dead-end complex'. As it is unlikely that FDTS is simply a reaction vessel for dUMP and the two cofactors, the enzyme's contribution to the mechanism must also be clearly characterized in the future.
Although Koehn and colleagues work1 raises the possibility of developing antibiotics that selectively block microbial FDTS, but not mammalian TS, this remains to be demonstrated, and obstacles certainly exist: only a few inhibitors6 of FDTS have been discovered since the protein was first reported in 2002. Nevertheless, the prospect of such an antibiotic is exciting, as several pathogenic bacteria, including that responsible for tuberculosis, should be susceptible. More broadly, the authors have discovered a new concept in antibiotic drug discovery — compounds that resemble an enzyme's substrate, but that incorporate significant changes to the core structure (such as that occurring in dUMP when it accepts a hydrogen atom from FADH2), might selectively attack bacteria, and so be less toxic to humans.
Koehn, E. M. et al. Nature 458, 919–923 (2009).
Myllykallio, H. et al. Science 297, 105–107 (2002).
Lesley, S. A. et al. Proc. Natl Acad. Sci. USA 99, 11664–11669 (2002).
Leduc, D. et al. Proc. Natl Acad. Sci. USA 101, 7252–7257 (2004).
Agrawal, N., Lesley, S. A., Kuhn, P. & Kohen, A. Biochemistry 43, 10295–10301 (2004).
Esra Oenen, F. et al. Bioorg. Med. Chem. Lett. 18, 3628–3631 (2008).
About this article
Discovery of an Acyclic Nucleoside Phosphonate that InhibitsMycobacterium tuberculosisThyX Based on the Binding Mode of a 5-Alkynyl Substrate Analogue
Proceedings of the National Academy of Sciences (2012)
Archives of Biochemistry and Biophysics (2010)