Decoding ligand recognition at GPR55 using in silico predictions and functional assays
Publication Date
December 10, 2025
Creator
Abstract
GPR55 is a non-canonical cannabinoid receptor that has emerged as a promising therapeutic target for cancer, inflammation, and metabolic disorders. Despite its therapeutic potential, the pharmacology of GPR55 remains poorly defined due to a limited understanding of the structural mechanisms underlying receptor function, inconsistent ligand activity across cellular assays, a lack of potent ligands, and the absence of standardised screening methods. To address these challenges, this study combines computational and experimental approaches to elucidate the molecular basis of GPR55 ligand recognition and develop a sensitive, robust assay for ligand screening.
Computationally, we applied molecular docking, interatomic fingerprinting, and machine learning analysis to a curated dataset of 563 GPR55 ligands across five receptor structures. CatBoost classifiers and SHAP analysis were also used to identify recurring interactions and visualise the data in a way that distinguishes agonists from antagonists. Experimentally, we replicated the established cell-based β-arrestin recruitment and G-GASE biosensor assays in stable HEK293TR cell lines. To overcome the limitations of cell-based assay, we developed a BRET membrane-based mG protein recruitment assay (MeGaBRET) to provide a more direct and controllable readout of GPR55 activation.
The computational analysis of ~5600 ligand binding poses identified the eleven residues W177ECL2, P1554.59x60, M167ECL2, Q2717.36x35, F1023.33x33, V1494.53x53, M2747.39x38, F2466.55x55, M172ECL2, L2707.35x34, and I1564.60x61 as key determinants of ligand pharmacology at GPR55. The CatBoost prediction model showed moderate performance, with positive MCC values between 0.25 and 0.39, demonstrating that our model was able to capture meaningful trends in agonist and antagonist classification. Experimentally, the β- arrestin recruitment assay reproduced literature trends but with ~2-fold lower potency than the reported data. THC showed no β-arrestin recruitment despite literature reports of GPR55 agonism. Unfortunately, multiple attempts to set up the G-CASE biosensor assays failed due to poor signal-to-noise ratios and overlapping concentration-response curves. In contrast, the optimised MeGaBRET assay produced response windows ~3 times larger than β-arrestin assays and successfully revealed the G protein bias of THC and the endogenous ligand LPI. MeGaBRET also detected mGi protein coupling for the potent ligands ML184 and compound 26, with EC50 values of 243 nM and 65 nM, respectively. Further optimisation is still required to measure mG12 and mGq coupling using MeGaBRET.
This study demonstrates that combining computational modelling with advanced functional assays provides a more comprehensive understanding of GPR55 pharmacology. Key receptor residues influencing ligand activity were identified, and the MeGaBRET assay was successfully developed for GPR55 ligand screening. This novel, robust assay expands the toolkit for ligand screening and provides a more controlled and sensitive approach to characterising ligand activity, helping to address the inconsistencies observed in conventional GPR55 assays. Together, these approaches offer a validated framework for rational drug design and orphan GPCR characterisation that can be applied beyond GPR55.
Item Type
ethesis
Thesis Type
MRes
Supervisors
Subjects (LC)
Associated Schools / Departments
School of Life Sciences
eprints ID
82833
UoN Repository URI
Except where otherwise noted, this item's license is described as
File(s)![Thumbnail Image]()
Name
VunganaiCleopatra 20135580 final submission.pdf
Type
Full-text
Description
Examined
Size
13.07 MB
Format
Adobe PDF
Checksum (MD5)
cd23640044c51691567c5a79a1ce8a1d