Abstract 4191: Combining artificial intelligence and human expertise in pursuit of novel PolQ inhibitors

Abstract DNA polymerase theta (POLθ or POLQ) recently emerged as a new promising drug target for the treatment of homologous recombination (HR) deficient tumours. This unique fusion protein comprises both DNA polymerase (Pol) and DNA helicase domains that function together to facilitate multiple DNA repair mechanisms. Small molecule approaches to target both domains have been described throughout the last decade and have shown potent and selective HR-deficient tumor cell death in vitro and in vivo. Despite the growing number of advanced PolQ polymerase inhibitors reported in the literature, significant ADME liabilities remain associated with the current reported scaffolds. To address this issue, machine-based learning methodologies were envisioned to act as a potential aid for medicinal chemistry efforts and help drive identification of chemical matter that might reside outside of the typical toolbox of small molecule design ideas. Using the Makya generative AI platform, synergistically with traditional knowledge-based approaches, we have identified novel small molecule PolQ polymerase domain inhibitors. One chemotype, with an embedded oxime motif, has demonstrated potent biochemical activity and comparable cell activity to known tool compounds. A follow-up library of functionalised oxime derivatives, with varying compositions of donor and acceptor motifs, was synthesized to probe the chemical space further which demonstrated improved biochemical potency on the original hit matter. Initial ADME profiling suggests that these ligands possess properties that represent a good starting point for a lead optimisation campaign. Herein, we report key findings from the initial phase of development of our small molecule PolQ inhibitors. Citation Format: Allan M. Jordan, Ageo Miccoli, Annabelle Charnock, Alexis Denis, Euan Fordyce, Cameron Franklin, Ennys Gheyouche, Ian Henderson, Christopher Housseman, Maxime Laugeois, Jean-Christophe Meillon, Stuart Onions, Quentin Perron, Clementine Pescheteau, Emilie Pihan, Mark Reeves, Kirsty Rooney, Hannah Rowlands, Kamaldeep Sumal, Franck Le Vaillant, Graeme Walker. Combining artificial intelligence and human expertise in pursuit of novel PolQ inhibitors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4191.