Identification of new potent A₁ adenosine receptor antagonists using a multistage virtual screening approach

The use of antagonists for each adenosine receptor (AR) subtype as potent clinical candidates is of growing interest due to their involvement in the treatment of various diseases. The recent resolution of several A₁ and A_2A ARs X-ray structures provides opportunities for structure-based drug design. In this study, we describe the discovery of novel A₁AR antagonists by applying a multistage virtual screening approach, which is based on random forest (RF), e-pharmacophore modeling and docking methods. A multistage virtual screening approach was applied to screen the ChemDiv library (1,492,362 compounds). Among the final hits, 22 compounds were selected for further radioligand binding assay analysis against human A₁AR, and 18 compounds (81.82% success) exhibited nanomolar or low micromolar binding potency (K_i). Then, we selected six compounds (pK_i > 6) to further evaluate their antagonist profile in a cAMP functional assay, and we found that they had low micromolar antagonistic activity (pIC₅₀ = 5.51-6.38) for the A₁AR. Particularly, four of six compounds (pK_i > 6) showed very good affinity (pK_i = 6.11-7.13) and selectively (>100-fold) for A₁AR over A_2AAR. Moreover, the novelty analysis suggested that four of six compounds (pK_i > 6) were dissimilar to existing A₁AR antagonists and hence represented novel A₁AR antagonists. Further molecular docking and molecular dynamics (MD) studies showed that the three selective compounds 15, 20 and 22 were stabilized (RMS_lig value ≤ 2 Å) inside the binding pocket of A₁AR with similar orientations to the docking pose in 100-ns MD simulations, whereas they escaped from the binding area of A_2AAR with larger values of RMS_lig (RMS_lig ≥ 2 Å). We hope that these findings provide new insights into the discovery of drugs targeting A₁AR and facilitate research on new drugs and treatments for A₁AR-related human pathologies.