Abstract
Manual synthesis of small molecules can represent a rate-limiting step in medicinal chemistry. This study describes the application of an automated, modular synthesis platform ('Chemputer') to a drug discovery project targeting a model of KRAS-mutant colorectal cancer (K-CRC). A 4-anilinoquinazoline-based compound library was synthesized using automated and digitized protocols for nucleophilic aromatic substitution (SnAr) and Suzuki cross-coupling reactions. Chemical synthesis is guided by phenotypic screening of a transgenic Drosophila line engineered to model the genetic profile of a patient's K-CRC tumour. This integrated system enables iterative synthesis and screening cycles. An initial run identified the hit compound AP2-83, which strongly improves animal survival. Kinase profiling and genetic validation find that AP2-83 activity is mediated in part through inhibition of CLK1 and PI3K. A subsequent optimisation effort, informed by these results, produced AP4-43. AP4-43 demonstrates increased efficacy in the Drosophila model and greater potency than regorafenib in a mammalian CRC organoid growth assay. Functional analysis indicates AP4-43 acts as a multi-kinase inhibitor, with its enhanced activity associated with the inhibition of a network including CLK1 and NEK4. This work demonstrates the utility of a digital synthesis platform for generating and optimising lead compounds in a complex, preclinical drug discovery context.
