Integrative approaches in fragment-based drug discovery

Thesis

This thesis combines experimental and computational methods to investigate aspects of fragment identification and elaboration in fragment-based ligand design, a promising approach for identifying small molecule drugs, to target the pharmacologically relevant bromodomain PHIP(2). The research covers various aspects of the process, from initial crystallographic fragment screening to validation of follow-up compounds.

Chapters 1 and 2 provide an overview of relevant perspectives and methodologies in fragment-based drug discovery. Chapter 3 reports a crystallographic fragment screening against PHIP(2), resolving 47 fragments at the acetylated-lysine binding site, and evaluates the abilities of crowdsourced computational methods to replicate fragment binding and crystallographic poses. This chapter highlights the challenges associated with using computational methods for reproducing crystallographic fragment screening results with submissions performing relatively weakly. Chapter 4 demonstrates the advantages of X-ray crystallographic screening of crude reaction mixtures generated robotically, showcasing reduced time, solvent, and hardware requirements. Soaking crude reaction mixtures maintains crystal integrity which led to the identification of 22 binders, 3 with an alternate pose caused by a single methyl addition to the core fragment and 1 hit in assays. It demonstrates how affordable methods can generate large amounts of crystallographic data of fragment elaborations. Chapter 5 develops an algorithmic approach to extract features associated with crystallographic binding, deriving simple binding scores using data from Chapter 4. The method identifies 26 false negatives with binding scores enriching binders over non-binders. Employing these scores prospectively in a virtual screening demonstrated how binding features can be exploited to select further follow-up compounds leading to low micromolar potencies. Chapter 6 attempts to integrate more computationally intensive methods to identify fragment follow-up compounds with increased potency through virtual screening enhanced with free energy calculations. Only two out of six synthesised follow-up compounds showed weak binding in assays, and none were resolved in crystal structures.

This thesis tackles critical challenges in follow-up design, synthesis, and dataset analysis, underlining the limitations of existing methods in advancing fragment-based drug discovery. It emphasises the necessity of integrative approaches for an optimised “design, make, test” cycle in fragment-based drug discovery.

Authors: Grosjean, H

• DOI: 10.5287/ora-mz7ozkzpn
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: Fragment-Based Drug Discovery; Structural Biology; Biophysics; Pleckstrin-homology domain interacting protein; Computational chemistry