The quest for ligands against kinesin motor protein Eg5 using CADD

The kinesin motor protein Eg5 is a well-validated anticancer target due to its essential role in mitotic spindle formation. However, despite numerous clinical trials, no Eg5 inhibitor has yet achieved regulatory approval, largely due to challenges with modest efficacy, dose-limiting toxicities, and drug resistance. This highlights a critical need for novel chemical scaffolds with superior therapeutic profiles. This dissertation addresses this challenge by employing a comprehensive, dual-track computational strategy to discover next-generation Eg5 inhibitors. The first track utilized a validated, structure-based virtual screening cascade on the ~11 million compound ZINC Anodyne library. The second, more exploratory track deployed a state-of-the-art DeepDocking active learning campaign to navigate the vast, 1-billion-compound ZINC-ML chemical space. The conservative Track 1 successfully identified four novel, active chemotypes, with the lead candidate ZINC243914928 demonstrating an in vitro IC₅₀ of 28.97 µM. This experimental validation of a computationally derived hit provided a crucial proof-of-concept for the entire workflow. More significantly, the AI-driven Track 2 discovered multiple, structurally diverse classes of inhibitors with superior in silico profiles. The top candidates from this track exhibited calculated binding free energies that were profoundly more favorable than both the reference inhibitor and the experimentally validated Track 1 hit. Crucially, the final hit compounds from both tracks exhibited favorable drug-like properties, conforming to Lipinski's rules, and demonstrated significant structural novelty with low Tanimoto similarity to known Eg5 chemotypes, confirming the success of the workflow in achieving genuine scaffold hopping. In conclusion, this thesis not only delivers multiple novel and validated chemical scaffolds for Eg5 inhibition but also demonstrates a powerful, multi-faceted computational strategy for tackling challenging drug targets. The exceptional computational profiles of the Track 2 candidates, supported by the validated success of the overall methodology, strongly suggest their potential for high potency and mark them as highly promising starting points for the development of a clinically successful Eg5-targeted therapeutic.