RESUMO
Despite decades of intensive search for compounds that modulate the activity of particular protein targets, a large proportion of the human kinome remains as yet undrugged. Effective approaches are therefore required to map the massive space of unexplored compound-kinase interactions for novel and potent activities. Here, we carry out a crowdsourced benchmarking of predictive algorithms for kinase inhibitor potencies across multiple kinase families tested on unpublished bioactivity data. We find the top-performing predictions are based on various models, including kernel learning, gradient boosting and deep learning, and their ensemble leads to a predictive accuracy exceeding that of single-dose kinase activity assays. We design experiments based on the model predictions and identify unexpected activities even for under-studied kinases, thereby accelerating experimental mapping efforts. The open-source prediction algorithms together with the bioactivities between 95 compounds and 295 kinases provide a resource for benchmarking prediction algorithms and for extending the druggable kinome.
Assuntos
Inibidores de Proteínas Quinases/farmacologia , Proteínas Quinases/metabolismo , Algoritmos , Benchmarking , Crowdsourcing , Bases de Dados de Produtos Farmacêuticos , Aprendizado Profundo , Descoberta de Drogas , Avaliação Pré-Clínica de Medicamentos , Humanos , Cinética , Aprendizado de Máquina , Modelos Biológicos , Modelos Químicos , Inibidores de Proteínas Quinases/química , Inibidores de Proteínas Quinases/farmacocinética , Proteínas Quinases/química , Proteômica , Análise de RegressãoRESUMO
Continuous exposure of a pancreatic cancer cell line MIA PaCa-2 (MiaS) to gemcitabine resulted in the formation of a gemcitabine-resistant subline (MiaR). In an effort to discover kinase inhibitors that inhibited MiaR growth, MiaR cells were exposed to kinase inhibitors (PKIS-1 library) in a 384-well screening format. Three compounds (UNC10112721A, UNC10112652A, and UNC10112793A) were identified that inhibited the growth of MiaR cells by more than 50% (at 50 nM). Two compounds (UNC10112721A and UNC10112652A) were classified as cyclin-dependent kinase (CDK) inhibitors, whereas UNC10112793A was reported to be a PLK inhibitor. Dose-response experiments supported the efficacy of these compounds to inhibit growth and increase apoptosis in 2D cultures of these cells. However, only UNC10112721A significantly inhibited the growth of 3D spheroids composed of MiaR cells and GFP-tagged cancer-associated fibroblasts. Multiplexed inhibitor bead (MIB)-mass spectrometry (MS) kinome competition experiments identified CDK9, CLK1-4, DYRK1A, and CSNK1 as major kinase targets for UNC10112721A in MiaR cells. Another CDK9 inhibitor (CDK-IN-2) replicated the growth inhibitory effects of UNC10112721A, whereas inhibitors against the CLK, DYRK, or CSNK1 kinases had no effect. In summary, these studies describe a coordinated approach to discover novel kinase inhibitors, evaluate their efficacy in 3D models, and define their specificity against the kinome.