Dona Flor and her two husbands: Discovery of novel HDAC6/AKT2 inhibitors for myeloid cancer treatment.

Unlabelled: Hematological cancer treatment with hybrid kinase/HDAC inhibitors is a novel strategy to overcome the challenge of acquired resistance to drugs. We collected IC datasets from the ChEMBL database for 13 cancer cell lines (72 h cytotoxicity, measured by MTT), known inhibitors for 38 kinases, and 10 HDACs isoforms, that we identified by target fishing and literature review. The data was subjected to rigorous biological and chemical curation leaving the final datasets ranging from 76 to 8173 compounds depending on the target. We generated Random Forest classification models, whereby 14 showed greater than 80% predictability after 5-fold external cross-validation. We screened 30 hybrid kinase/HDAC inhibitor analogs through each of these models. Fragment-contribution maps were constructed to aid the understanding of SARs and the optimization of these compounds as selective kinase/HDAC inhibitors for cancer treatment. Among the predicted compounds, 9 representative hybrids were synthesized and subjected to biological evaluation to validate the models. We observed high hit rates after biological testing for the following models: K562 (62.5%), MV4-11 (75.0%), MM1S (100%), NB-4 (62.5%), U937 (75.0), and HDAC6 (86.0%). This aided the identification of and as potent anticancer inhibitors with IC of 0.2-0.8 µM in three cancer cell lines, linked to HDAC6 inhibition below 2 nM, and blockade of AKT2 phosphorylation at 2 μM, validating the ability of our models to predict novel drug candidates.

Highlights: Novel kinase/HDAC inhibitors for cancer treatment were found using machine learning61 QSAR models for hematological cancers and its targets were built and validatedK562, MV4-11, MM1S, NB-4, U937, and HDAC6 models had hit rates above 62.5% in tests and presented potent IC of 0.2-0.8 µM in three cancer cell lines and inhibited HDAC6 below 2 nM, and blockade of AKT2 phosphorylation at 2 μM.