Investigations Into Fragment Ligand Binding Using Quantitative STD and WaterLOGSY NMR Spectroscopy

Ligand-observed NMR spectroscopy is frequently employed in early-stage drug discovery, often as an initial screen to narrow the field of potential drug-like molecules. However, its use is limited to this early stage. More information regarding binding mode can be extracted from these experiments via quantification, and this should help extend the remit of these experiments beyond simple screening functions.

Initially, it was shown that the amount of signal that could be produced from an STD NMR experiment could be dramatically increased by careful consideration of the selective saturation pulse. By systematically shortening the Gaussian pulse and positioning it at specific offset positions, it was shown that these dramatic increases in signal are genuine and need not result in false positives.

Quantitative STD NMR spectroscopy as applied to Hsp90 and a series of small fragment ligands provided evidence to suggest that the precise inter-atomic distances between a protein and ligand within a crystal structure correlate with both initial rates of STD build up, and T1-adjusted STD values. This precise correlation has implications for chemotype clustering and initial binding mode selection, something which should be useful in the absence of a crystal structure.

Taking the same quantitative principles and applying to LOGSY experiments elucidated another, discrete property of protein-ligand binding. Examining the ‘LOGSY difference’ signal for protons of a ligand allows us to see what protons are in close proximity to conserved, bound water at the protein-ligand binding interface. This is fundamentally different to the information gained from STD experiments.

Applying the insights to a protein of a different nature, Ras, it was shown that quantitative STD can be applied to proteins of both different size and structure. Furthermore, more evidence was acquired to suggest that conserved, bound water in the binding site really is responsible for generating LOGSY signal. In the absence of these molecules, as in Ras, proximity of a proton to an exchangeable tends to dominate. In addition we were able to show that these quantitative methods can be used together to help eliminate incorrect computationally generated docking poses.

The work presented in this thesis provides evidence for the advantages of STD and LOGSY NMR spectroscopy in fragment-based drug discovery. The information that can be extracted from relatively simple ligand-observed NMR experiments should be used to provide more evidence at an earlier stage of the drug discovery process, hopefully reducing late-stage attrition and helping us get to the therapeutic drug

molecules we need a little more quickly.