Development of fragment-based approaches to build chemical tools for biology

Enzymes catalyse chemical reactions within living cells, very quickly and very precisely. One way to cure (or kill) an organism is to slow down or stop an enzyme by using a small molecule called an enzyme inhibitor. Enzyme inhibitors have to be very selective (so they only affect the target enzyme) and they have to bind very tightly to that enzyme, so that they are not pushed off by the enzyme's normal substrate. By understanding how to inhibit an enzyme we can often learn something about how the enzyme works. The design and synthesis of enzyme inhibitors is also central to the pharmaceutical industry. Many medicines are enzyme inhibitors e.g. penicillin, pills for high blood pressure etc. So it is not surprising that ways to discover new enzyme inhibitors are very important. In the last ten years the main approach industry has used to find new enzyme inhibitors is to test millions of compounds very quickly to see if promising compounds can be found. These are then modified to be more potent and have the properties they need to be a medicine. However there is now growing interest in a completely different approach to discovering enzyme inhibitors. This involves finding a quite small molecule (called a fragment) that binds to the enzyme, and then using this as an anchor point to build up bigger more potent inhibitors. These 'fragment-based' approaches depend on methods of finding fragments. There are several ways to this, and we propose to use a number of these, some of which are well established and others which are more novel. The methods use techniques including NMR spectroscopy, mass spectrometry, and calorimetry. All rely on some way of showing the fragment binds to the enzyme, with the exception of virtual screening, where computational approaches are used to try and identify fragments that look like they should bind to the enzyme. Once we have identified fragments that bind to our target enzymes we will try and grow them into bigger molecules. One general way to do this is to mix a lot of fragments and let them react with each other in every possible combination (called dynamic combinatorial chemistry), and let the enzyme select out the one that binds to it most efficiently. It may be that the enzyme acts as a template to bring together two fragments that bind to it separately. If a lot of enzyme is used, it may lead to more of the key compound that binds to it being formed, allowing this compound to be identified as the major product. A more elegant way to identify the best binding compound, which uses less of the valuable enzyme, is to form the mixture of compounds in the presence of a crystal of the enzyme. If a compound binds to the enzyme, its identity can be deduced by solving the crystal structure of the enzyme with the molecule bound to it. We will devlop these ideas by trying to make inhibitors of two enzymes involved in making vitamin B5 (pantothenate). There is evidence to suggest that inhibitors of these enzymes may be useful against tuberculosis. Another reason to target these enzymes is that they bind molecules called NADPH and ATP, which are also used by other enzymes, so we might learn some useful general information about how to inhibit these other enzymes.

There is a revolution occurring in the way enzyme inhibitors are identified. After a decade of high throughput screening, there is a family of related new approaches that are gaining acceptence. The common theme is that they involve identifying small molecular fragments (MW<250, variously referred to as 'needles', 'shapes', binding elements' 'seed templates') which bind weakly to the enzyme. These are then iteratively elaborated (or grown) to make larger more potent inhibitors. This application describes projects using these approaches to generate chemical tools to be used in biology, specifically inhibitors that bind competitively at the NADPH site of dehydrogenases, or the ATP site of synthetases. These will be of general utility, but in the project are being designed specifically against ketopantoate reductase (KPR) and pantothenate synthetase (PtS) respectively. These two enzymes on the pantothenate pathway are potential antimicrobial targets (e.g. inhibitors may attenuate virulence in Mycobacterium tuberculosis). We have previously solved the crystal structures of both enzymes and have extensive experience in their molecular biology and enzymology. We have done extensive groundwork for this project, including virtual screens, preliminary 1D 1H NMR WaterLOGSY screens, and some proof of principle experiments using non-covalent mass spectrometry, surface plasmon resonance and isothermal titration calorimetry. We plan to use all these techniques and x-ray crystallograpy to identify the molecular fragments. We plan to use three strategies to build up the inhibitors: (i) Growing out of a fragment anchor. There is a growing understanding of why fragments provide a good starting anchor point for inhibitor synthesis. The approach will involve iterative rounds of synthesis and biophysical characterisation of the mode of binding. (ii) Dynamic combinatorial chemistry in solution. This approach now has reasonable precedent, but is still far from straightforward. We plan to use reductive amination and disulfide exchange strategies. To increase the chance of success, and to provide an anchor for one half of the inhibitor our initial screens will include an adenosyl group. (iii) Dynamic combinatorial x-ray crystallography. This is dynamic combinatorial chemistry in the presence of a single enzyme crystal. The identity of the best inhibitor is solved by solving the x-ray structure of the enzyme-ligand complex. Inhibitors will be tested against KPR and PtS, and against functionally related enzymes.