Influence of the Structural Accuracy of Homology Models on Their Applicability to Docking-Based Virtual Screening: The β₂ Adrenergic Receptor as a Case Study

Abstract

How accurate do structures of the β 2 adrenergic receptor (β 2 AR) need to be to effectively serve as platforms for docking-based virtual screening campaigns? To answer this research question, here we targeted through controlled virtual screening experiments 23 homology models of the β 2 AR endowed with different levels of structural accuracy. Subsequently, we studied the correlation between virtual screening performance and structural accuracy of the targeted models. Moreover, we studied the correlation between virtual screening performance and template/target receptor sequence identity. Our study demonstrates that docking-based virtual screening campaigns targeting homology models of the β 2 AR, in the majority of the cases, yielded results that exceeded random expectations in terms of area under the receiver operating characteristic curve (ROC AUC). Moreover, with the most effective scoring method, over one third and one quarter of the models yielded results that exceeded random expectation also in terms of enrichment factors (EF1, EF5, and EF10) and BEDROC (α = 160.9), respectively. Not surprisingly, we found a detectable linear correlation between virtual screening performance and structural accuracy of the ligandbinding cavity. We also found a detectable linear correlation between virtual screening performance and structural accuracy of the second extracellular loop (EL2). Finally, our data indicate that, although there is no detectable linear correlation between virtual screening performance and template/β 2 AR sequence identity, models built on the basis of templates that show high sequence identity with the β 2 AR, especially within the ligand-biding cavity, performed consistently well. Conversely, models with lower sequence identity displayed performance levels

Performance of virtual screening against GPCR homology models: Impact of template selection and treatment of binding site plasticity

“…Large sets of models with different side chain conformations must hence be considered to evaluate the quality of a template. This finding makes it difficult to compare our results to previous studies that were mainly based on a single homology model per template [24,32,34]. In agreement with expectations, the most accurate predictions of the 5-HT 2A R binding site were obtained based on closely related 5-HT 2 subtypes (66-75% TM sequence identity) and these models showed excellent virtual screening performance.…”

“…Reliable techniques for GPCR modeling would make it possible to extend the use of structure-based ligand discovery to many unexplored drug targets. An increasingly employed strategy to evaluate models is to use molecular docking to assess if the binding site can identify known active compounds among decoys [18][19][20][21][22][23][24]. A model that displays high enrichment of known ligands is considered to be a good representative of the receptor structure and suitable for virtual screening.…”

Locating ligand binding sites in G-protein coupled receptors using combined information from docking and sequence conservation

“…ConDockSite also fails with the muscarinic M1 receptor (PDB 5cxv) where the distance between the predicted and actual triotropium binding sites was >10 Å. Overall, our results with ConDockSite are consistent with benchmark modeling results that show that GPCR homology models of modest accuracy from templates with low sequence identity are still sometimes useful for docking and virtual screening (Lim et al, 2018;Costanzi et al, ). In some of the failed cases (histamine H1, muscarinic M1, and serotonin 5HT2B), while the homology models were sometime accurate at the level of backbone atoms in the 7tm region, the loops were modeled poorly and disrupted the modeled ligand binding pocket.…”

Structure-Based Virtual Screening for Ligands of G Protein–Coupled Receptors: What Can Molecular Docking Do for You?

“…Virtual screening performance of homology models can be enhanced when known actives are available. Models generated with different methods and/or templates can then be evaluated by calculating the ligand enrichment, and the structure with the optimal performance can be used in the prospective virtual screen (Lim et al, 2018;Costanzi et al, 2019;Jaiteh et al, 2020). Most of the successful virtual screens based on homology models used a structure of a closely related receptor as template (Carlsson et al, 2011;Kolb et al, 2012;Vass et al, 2014b;Lam et al, 2015;Ranganathan et al, 2015;Sz} oll} osi et al, 2015;Kaczor et al, 2016;Weiss et al, 2018).…”