Fig 1.
Structures of ASP5854 (A), 2-amino-4-phenyl-8-(pyrrolidin-1-ylmethyl)-5H-indeno[1,2-d]pyrimidin-5-one (B) and 4-(5-methylfuran-2-yl)-6-[3-(piperidine-1-carbonyl)phenyl]pyrimidin-2-amine (C), which were found to be dual A1/A2A AR antagonists.
Table 1.
Test results of the DNN classification model under different batch sizes.
Table 2.
Test results of the CNN classification model under different batch sizes.
Fig 2.
Pharmacophore features of pharmacophore hypotheses AADR_4 and AAADR_1.
AADR_4 is displayed in green, and AAADR_1 is displayed in red.
Table 3.
Validation of the pharmacophore hypotheses.
Fig 3.
Schematic overview of the discovery of dual A1/A2A AR antagonists by performing multistage virtual screening against the ChemDiv library.
Fig 4.
Chemical structures of compounds C1—C19.
Table 4.
Potencies and binding activities at human A1 and A2A ARs of compounds C1 –C19.
A1/A2AAR Affinity Experiments.
Fig 5.
Chemical structures of DPCPX (A) and ZM241385 (B).
Table 5.
Glide XP scores, RMSDlig and MM-GBSA binding free energies of ligands C8 and C9 against A1AR and A2AAR.
Fig 6.
Superposition of the starting structure (in cyan) and the final snapshot (in orange) in the MD trajectory of Compounds C8 (A) and C9 (B) in the orthosteric binding area of A1AR. Superposition of the starting structure (in green) and the final snapshot (in orange) in the MD trajectory of Compounds C8 (C) and C9 (D) against A2AAR. The protein is shown as a gray cartoon, the ligands are shown as sticks, and the residues in the binding pocket of A1 and A2A AR are represented as lines. The hydrogen bonds and π-π stacking between the ligands and ARs are represented by dashed lines.
Fig 7.
(A) Superposition of the starting structure (in cyan) and the last snapshot (in orange) in the MD trajectory of compound C10 in the orthosteric binding area of A1AR. (B) Superposition of the starting structure (in green) and the final snapshot (in orange) in the MD trajectory of compound C10 against A2AAR. The protein is shown as a gray cartoon, the ligands are shown as stick, and the residues in the binding pocket of A1 and A2A AR are represented as lines. The hydrogen bonds and π-π stacking between compound C10 and ARs are represented by dashed lines.
Fig 8.
Architectures of neural networks.
(A) Architecture of deep neural networks. (B) Architecture of convolutional neural networks.