Starting with just 500 fragments, crystallography allowed the researchers to identify more than 30 fragments that bound in the ATP-binding site, all of which made at least one hydrogen bond to the so-called “hinge region” of the protein. Three of these fragments were optimized using structure-based design and medicinal chemistry, with the most successful yielding AT7519.
The figure shows the progression of the series, starting from the initial fragment, along with the IC50s and the ligand efficiencies of key milestones. Some notable decisions included replacing the indazole moiety with a pyrazole, which resulted in a 30-fold drop in biochemical potency but did not notably reduce the ligand efficiency, and the replacement of a fluorobenzene moiety with a piperidine, which led to a loss in biochemical potency but an improvement in solubility and cell potency. AT7519 showed 86% tumor growth inhibition in an ovarian mouse xenograft model when dosed IP at 7.5 mg/kg. The full paper is well-written and provides an elegant example of taking a fragment all the way to a clinical compound.