Fragment based ligand discovery: Library design and screening by thermal shift analysis

The central idea in Fragment Based Ligand Discovery (FBLD) is to identify small, low molecular weight compounds (MW < 250) that bind to a particular protein active site. Hits can be used to efficiently design larger compounds with the desired affinity and selectivity.

Three approaches to FBLD are described in this thesis.

The first topic is the development and assessment of different chemoinformatics procedures to select those fragments that maximally represent the chemical features
of a larger compound library. Such a fragment library could be of great value in the so-called “SAR by Catalogue" approach, where the initial stage of fragment growth is by selecting existing compounds that contain sub-structures
of the hit fragments. Five schemes implemented in the
Pipeline Pilot software are described.

The second project was to develop improved approaches to processing Thermal Shift Analysis (TSA) data. The shift in melting temperature can indicate that a ligand binds and thus stabilises a protein. A program, MTSA, has been written which allows more straightforward processing of the experimental data than existing available software. However, detailed analysis of fragment screening
data highlighted difficulties in defining the melting temperature and suggest that TSA is not sufficiently reliable for routine screening use.

Finally, a number of proteins were assessed experimentally for suitability for FBLD: N-myristoyl transferase (NMT), the bacterial homologue of a GlcNAcase enzyme (BtGH84) and the model system hen egg white lysozyme (HEWL). It was not possible to produce suitable NMT material due to the inherent instability of the protein produced in York. The screening results of HEWL with a new Surface Plasmon Resonance (SPR) assay, a cell based activity assay and TSA were inconsistent and difficult to interpret. However, BtGH84 was suitable for screening by both TSA and SPR. The resulting fragment hits are suitable starting points for further evolution.