Deep learning identification of cardiac transporters as targets for approved drugs.

Abstract

We re-tasked a novel ligand-centered deep learning drug discovery method to identify molecular targets for approved drugs in the heart. This approach uses information about small-molecule effectors to map and probe the pharmacological space of functionally relevant targets in the heart. Here we studied the calcium pump (Sarcoplasmic reticulum Ca2+-ATPase, SERCA) as a proof-of-principle target to test our method. We chose SERCA because it plays a major role in the excitation-contraction-relaxation cycle in normal and pathological muscle, and it represents a major pharmacological target in the heart. We applied this method to demonstrate that SERCA is a pharmacological target for statins, a group of FDA-approved HMGCoA inhibitors used as lipid-lowering medications. We used in situ enzymatic assays and atomistic simulations to demonstrate that that these approved drugs are SERCA inhibitors at micromolar concentrations, inhibiting the pump by binding to two different effector sites. These proof-of-concept studies support the applicability of our approach for off-target identification and drug repurposing, ultimately minimizing the translational gap in drug development targeting the heart.