Computational and experimental studies of novel β-adrenoceptor ligands

With over 30% of available drugs targeting them, G protein-coupled receptors (GPCRs) are of significant pharmaceutical interest. Efforts to understand protein-ligand interactions among this group of proteins have been aided by the increase in available x-ray crystallography structures. β1-Adrenergic receptor (β1-AR) antagonists are used as treatment in patients with cardiovascular and airway conditions. However, current widely used medications are considerably prone to off-target side effects due to the lack of selectivity between the β1- and β2-AR subtypes. Therefore, a deeper understanding into the structural differences in characteristics is necessary to utilise them as a means of increasing ligand selectivity and therefore reducing the prevalence of off-target side effects. Here, two characteristics of the β1-AR are targeted as a means of increasing receptor selectivity. The first being receptor plasticity - recent research has shown that β-ARs contain a fissure between transmembrane helices 4 and 5 (TM4, TM5) (dubbed the ‘keyhole’) that differ slightly between β-AR subtypes that may accommodate for extended moieties or ligand entry and exit via the intramembrane space. The second characteristic being receptor dimerization. Receptor dimerization among GPCRs remains an active area of research, that so far has many pharmacological implications. Targeting receptor homodimerization has been proposed to be a method of improving receptor specificity within GPCRs. Research into β-AR dimers and findings from X-ray crystallography have shown that β1-AR homodimers may indeed align with a TM4/TM5 interface, aligning the ‘keyhole’. By combining and exploring both characteristics, we designed and computationally validated bivalent ligands capable of taking advantage of two unique β1-AR structural features as a means of improving ligand selectivity. Most current attempts at bivalent ligands in GPCRs explore using the extracellular space as a spacing route, leading to longer ligands, undesirably affecting molecular weight, lipophilicity, and viability. However, to validate our ligand design, we computationally demonstrate – by analyzing all-atom molecular dynamics (MD) simulations – that those ligands long enough to extend beyond the receptor via the keyhole can bind canonically and maintain key interactions that have previously been pharmacologically verified, as well as investigate structure activity relationships (SARs) of differing steric and electronic configurations of ligand components exposed to the intramembrane space. Bivalent ligand linkers were designed and computationally investigated within a GPCR dimer system to determine whether flexibility of the linker impacts the pharmacophores’ ability to maintain key canonical interactions. Long timescale coarse grained simulations of a membrane-bound β1-AR dimer showed the dimer interface to be stable, so shorter all-atom simulations could be used with confidence to aid bivalent ligand design. Bivalent ligands of the nature discussed in this work required at least one pharmacophore to enter/exit the receptor orthosteric binding site via the keyhole route. In house enhanced sampling computational methods were developed to study and validate the feasibility of this entry and exit route. Ligand exit pathways were generated by performing self-avoiding walk MD on protein-ligand complexes, then used to define starting and end points for weighted ensemble molecular dynamics (WEMD) to predict kinetic rate constants. These rate constants were then verified against pharmacologically derived β-AR kinetics data to validate the method, model, and ligand entry/exit pathway. The designed ligands would then lead to shorter and less hindering spacing between orthosteric sites