Discovery of novel dual adenosine A1/A2A receptor antagonists using deep learning, pharmacophore modeling and molecular docking.

Mukuo Wang,Jianping Lin,Dongmei Li,Yu Wei,Shujing Hou,Avner Schlessinger

doi:10.1371/journal.pcbi.1008821

Abstract

Adenosine receptors (ARs) have been demonstrated to be potential therapeutic targets against Parkinson’s disease (PD). In the present study, we describe a multistage virtual screening approach that identifies dual adenosine A1 and A2A receptor antagonists using deep learning, pharmacophore models, and molecular docking methods. Nineteen hits from the ChemDiv library containing 1,178,506 compounds were selected and further tested by in vitro assays (cAMP functional assay and radioligand binding assay); of these hits, two compounds (C8 and C9) with 1,2,4-triazole scaffolds possessing the most potent binding affinity and antagonistic activity for A1/A2A ARs at the nanomolar level (pKi of 7.16–7.49 and pIC50 of 6.31–6.78) were identified. Further molecular dynamics (MD) simulations suggested similarly strong binding interactions of the complexes between the A1/A2A ARs and two compounds (C8 and C9). Notably, the 1,2,4-triazole derivatives (compounds C8 and C9) were identified as the most potent dual A1/A2A AR antagonists in our study and could serve as a basis for further development. The effective multistage screening approach developed in this study can be utilized to identify potent ligands for other drug targets.

Highlights

Parkinson’s disease (PD) is a common and complex neurodegenerative disorder that is characterized by the early prominent death of dopaminergic neurons in the substantia nigra pars compacta and the abnormal aggregation of the α-synuclein protein, called Lewy bodies and Lewy neurites [1]
The results showed that the multistage virtual screening approach can effectively identify potent dual adenosine A1/A2A receptor antagonists
We used the self-developed Python script based on the algorithm from reference 50 to construct the deep neural networks (DNNs) and convolutional neural networks (CNNs) classification models of the dual A1/A2A Adenosine receptors (ARs) antagonists using a library containing 310 bioactive dual A1/A2A antagonists (Ki < 40 nM) and 405 nonbioactive antagonists (Ki > 1000 nM)

Summary

Introduction

Parkinson’s disease (PD) is a common and complex neurodegenerative disorder that is characterized by the early prominent death of dopaminergic neurons in the substantia nigra pars compacta and the abnormal aggregation of the α-synuclein protein, called Lewy bodies and Lewy neurites [1]. L-3,4-Dihydroxyphenylalanine (L-dopa), which is the direct precursor of dopamine, is commonly used as dopamine replacement therapy in the treatment of PD motor symptoms [5]. Dopamine receptor agonists, catechol O-methyltransferase inhibitors, monoamine oxidase B (MAOB) inhibitors, amantadine and anticholinergic drugs are available on the market for the treatment of PD [7,8]. These drugs are mainly used to replace the concentration and/or effect of dopamine in the brain and only solve the motor symptoms but not the nonmotor symptoms [9].

Objectives

Methods

Results

Conclusion