The structures predicted by AlphaFold can provide unprecedented opportunities for docking screens; however, experimentally validated examples of using the apo-form are limited. This study reports novel chemotype inhibitors targeting the leucine-rich repeat kinase 2 (LRRK2) kinase domain through a docking screen using one of the ensemble structures starting from the template deposited by AlphaFold2. The MODELLER software generated the ensemble. The conformer that showed the best early enrichment of true positives with the mixture of known ligands and their property-matched decoys was selected. The docking screen against approximately 1.3 million small molecules and enzyme-based assays with the LRRK2 kinase domain followed. We selected 17 molecules, excluding those similar to all known kinase inhibitors. Combined with analogs-by-catalog, ten new small molecules with Ki values below 15 μM were discovered, including one sub-μM inhibitor. To test selectivity, enzyme assays with a mutant and six additional kinases, including known off-targets of existing LRRK2 inhibitors, were performed using three inhibitors. The data suggest that the novelty in chemical structure may be insufficient for providing selectivity. Our approach is generally applicable to cases where information on known binders is available but experimental structure is not.
Baral et al. (Thu,) studied this question.