Deep learning model enables the discovery of a novel BET inhibitor YD-851.

Abstract

BET inhibitor is a novel strategy in tumor therapy based on targeting epigenetic mechanism. In recent decades, dozens of clinical trials have been conducted to validate the potential efficacy of the first-generation BET inhibitors in refractory cancer and non-cancerous ailments. However, limited efficacy and significant toxicity were observed in clinical trials for treating solid tumors. Here, we proposed a novel inhibitor strategy as well as an effective and low toxicity agent that can effectively kill tumor cells and exhibited low toxicity. A ring-closure scaffold hopping approach and high-precision deep learning models was leveraged to furnish a series of rationally designed carboline derivatives as desired BET inhibitors. These most potent compounds were synthesized by an efficient and facile multistep route. Subsequent evaluations identified a potent BET inhibitor YD-851 and it can effectively inhibit tumor cell proliferation. In addition, YD-851 causes tumor shrinkage and significantly suppresses tumor growth in multiple xenograft solid tumor models. Moreover the results of toxicity texting and pharmacokinetic properties support further development of YD-851. We obtain an effective and low toxicity preclinical candidate for BET inhibitor to treat solid tumors. And the success of our strategy encourages the implementation of similar methods in the drug discovery of other targets.