In silico evaluation of 4-quinolone ligands interacting with human G protein-coupled receptor kinase 5 (GRK5) using message passing neural networks (MPNNs), molecular docking, and dynamics simulation

Abstract Background Cardiovascular disease (CVD) remains the leading cause of death globally, with a significant burden in low- and middle-income countries. The increasing prevalence of CVD is attributed to factors such as aging populations, urbanization, and lifestyle changes. Effective interventions, including lifestyle modifications and pharmacological treatments, have been shown to reduce the incidence of coronary heart disease, a major contributor to CVD. This study aimed to identify potential GRK5 inhibitors from the 4-quinolone class of compounds. Materials and methods An initial substructure search for 4-quinolones was carried out using PubChem, which returned over 210,000 compounds. These compounds were filtered using Lipinski’s Rule of Five and medicinal chemistry properties such as QED and SAScore. To further reduce the number of compounds, a machine learning model trained on known GRK5 inhibitors was used, identifying 26,373 potential inhibitors. These predicted inhibitors were then screened against GRK5 using structure-based screening to evaluate their binding affinity. Finally, molecular dynamics simulations of the lead compound and the reference compound were performed at 100 ns to confirm the stability of the protein–ligand complexes. Result In this study, five hit compounds were identified with better binding affinities than the reference ligand KR-39038 (− 8.922 kcal/mol). These include Compound 91,786,186 (− 10.252 kcal/mol), 14,795,803 (− 9.847 kcal/mol), 45,920,092 (− 9.220 kcal/mol), 155,494,220 (− 9.137 kcal/mol), and 152,753,635 (− 9.125 kcal/mol). Free binding energy calculations further supported the strong binding of these hits. ADMET predictions indicated that all five compounds have high gastrointestinal absorption, making them favorable for oral administration and systemic bioavailability. In addition, the compounds displayed overall drug-like properties. Finally, molecular dynamics simulations confirmed the stability of the lead compound within the active site of GRK5, showing improved stability compared to the reference ligand. Conclusion This study highlights the potential of 4-quinolones as a promising scaffold in the search for more potent GRK5 inhibitors. However, since the study is entirely computational, further in vitro and in vivo validation is necessary to confirm these findings.