Identifying effects of adrenaline and dopamine binding on the beta2-adrenergic receptor structure and function using machine learning

Abstract: The beta2-adrenergic receptor is a G-protein coupled receptor, involved in several physiological processes, which enables signaling through the cell membrane. To study the effect of dopamine and adrenaline binding on the receptor structure and function, we used machine learning methods applied to data from molecular dynamics simulations. We found that the three machine learning methods Random Forest, Kullback-Leibler divergence, and Principal Component Analysis generated results that correspond to previous studies. When comparing the active state of the receptor with or without a ligand bound, we found that residues around Ser203 and Asn301 of the orthosteric binding pocket and residues around Ala91 of the TM2 differed. When instead comparing the active state of the receptor with adrenaline or dopamine bound, we found that residues around Thr68 differed. Additionally, we also found that adrenaline and dopamine cause different structural changes in the intracellular parts of TM5 and TM6. These findings indicate ligand-specific effects on the receptor, providing potentially useful information for the understanding of the interaction of adrenaline and dopamine with the beta2-adrenergic receptor.