Identification of ligand-specific G-protein coupled receptor states and prediction of downstream efficacy via data-driven modeling

Abstract

<jats:title>Abstract</jats:title><jats:p>G protein-coupled receptors (GPCRs) shift between inactive non-signalling states and active signalling states, to which intracellular binding partners can bind. Extracellular binding of ligands stabilizes different receptor states and modulates the intracellular response via a complex and not well understood allosteric process. Despite the recent advances in structure determination and spectroscopy techniques, a comprehensive view of the ligand-protein interplay remains a challenge. We derived the activation free energy of the β2 adrenergic receptor (β2AR) bound to ligands with different efficacy profiles using enhanced sampling molecular dynamics (MD) simulations. The resulting free energy landscapes reveal clear shifts towards active-like states at the G protein binding site for receptors bound to partial and full agonists compared to antagonists and inverse agonists. Not only do the ligands control the population of states, they also modulate the receptor’s conformational ensemble by flipping allosteric protein microswitches. We find an excellent correlation between the conformation of the microswitches close to the ligand binding site and in the transmembrane region and experimentally reported cAMP signaling responses, highlighting the predictive power of our apprch. Using dimensionality reduction techniques, we could further assess the similarity between the unique states induced by different ligands. Two hotspots governing agonism on transmembrane helix 5 (TM5) and TM7, including the conserved NPxxY motif, formed the endpoints of an allosteric pathway between the binding sites. Our results demonstrate that computational methods can be used to assess GPCR drug candidates with specific efficacy profiles.</jats:p>

Cooperative protein allosteric transition mediated by a fluctuating transmission network

“…For example, a repacking of a hydrophobic core as well as switching of polar interaction partners was reported for the nitrogen regulatory protein C. 17 Moreover, the above described lever-arm effect is reminiscent of structural changes in G protein-coupled receptors, where small changes in the ligand binding site are structurally amplified to large-scale protein surface changes at the G protein binding site. 39,40 A similar effect was also reported for the microbial rhodopsin bacteriorhodopsin, where the retinal cofactor pushing against Trp182 causes a large-scale outward motion of helix F, 41 and for the connection between ATP hydrolysis and protein conformational changes in heat shock protein 90. 20 The interplay between rigid secondary elements (such as α-helices and β-sheets) and flexible protein sections (such as loops or linkers) to mediate structural rearrangements was also discussed by Nussinov and Thirumalai.…”

Distinct Activation Mechanisms Regulate Subtype Selectivity of Cannabinoid Receptors

“…(2) KullbackLeibler divergence (K-L divergence) of apo simulation with respect to agonist bound active simulation (3) K-L divergence of apo simulations with respect to agonist bound active simulations. 47 Here, features were normalized for better comparison using these metrices. As expected, all the features which show distinction in crystal structures for both proteins have larger values as shown in Table 1 and 2.…”

Multiscale reweighted stochastic embedding (MRSE): Deep learning of collective variables for enhanced sampling

“…21 t-SNE has been used to analyze unbiased MD trajectories. [63][64][65][66] We introduce here a parametric and multiscale variant of a SNE method aimed at learning CVs for atomistic simulations. In particular, we focus on using the method within enhanced sampling simulations, where we need to consider biased simulation data.…”