Orphan receptor ligand discovery by pickpocketing pharmacological neighbors

Abstract

Understanding the pharmacological similarity of G protein-coupled receptors (GPCRs) is paramount for predicting ligand off-target effects, drug repurposing, and ligand discovery for orphan receptors. Phylogenetic relationships do not always correctly capture pharmacological similarity. Previous family-wide attempts to define pharmacological relationships were based on three-dimensional structures and/or known receptor-ligand pairings, both unavailable for orphan GPCRs. Here, we present GPCR-CoINPocket, a novel contact-informed neighboring pocket metric of GPCR binding-site similarity that is informed by patterns of ligand-residue interactions observed in crystallographically characterized GPCRs. GPCR-CoINPocket is applicable to receptors with unknown structure or ligands and accurately captures known pharmacological relationships between GPCRs, even those undetected by phylogeny. When applied to orphan receptor GPR37L1, GPCR-CoINPocket identified its pharmacological neighbors, and transfer of their pharmacology aided in discovery of the first surrogate ligands for this orphan with a 30% success rate. Although primarily designed for GPCRs, the method is easily transferable to other protein families.

Figure 1. Ligand contact map of Class A GPCRs

Ligand contact strength fingerprints of receptor residues located in the transmembrane (TM) and extracellular loop two (ECL2) regions of crystallized Class A GPCRs, as collated in the Pocketome. The area of the circles represents the relative strength of the ligand contact and residue position is numbered according to the GPCRdb numbering scheme. Residue positions overlapping the binding site in Gloriam et al are highlighted in red. Ligand contacts within three residues of the highly conserved ECL2 cysteine have been portrayed (LPA1 receptor ECL2 contacts excluded). Contacts from residue backbone atoms (except for glycine), ECL1 and ECL3 were also excluded. Colors of the circles represent the chemical class of the GPCR’s endogenous ligand: gold, retinal; blue, biogenic amines; red, nucleosides; green, lipids; purple, peptides.

Figure 2. Comparison matrix of Class A GPCR binding site and transmembrane sequence similarities compared to GPCR-CoINPocket

Heat map representation of GPCR sequence similarities based on transmembrane (TM) domains (a), unweighted generic GPCR binding site residues (b), or contact profiling of GPCR binding sites using GPCR-CoINPocket (c). In the heat maps, the GPCRs have been ordered according to TM sequence similarity clustering, with orphan clusters colored in red. The matrix diagonals are colored in gray. SOG, somatostation, opioids, and galanin

Figure 3. Organization of Class A GPCRs based on GPCR-CoINPocket

Dendrogram of Class A GPCRs clustered by contact strength-profiled binding site sequence similarities (GPCR-CoINPocket). All receptors have been named and abbreviated in accordance with the IUPHAR/BPS Concise Guide to PHARMACOLOGY (www.guidetopharmacology.org) except for the adrenoceptors where the suffix AR has been used for figure clarity. GPR37L1 is marked with a red asterisk. Receptor names are colored according to the chemical class of the GPCR’s endogenous ligand: gold, retinal; blue, biogenic amines; red, nucleosides; green, lipids; purple, peptides; orange, melatonin; black, orphans.

Figure 4. The pharmacological neighborhood of orphan receptors, ACKR3 and GPR37L1

GPCR-CoINPocket and transmembrane (TM) similarity of receptors, relative to ACKR3 (a) and GPR37L1 (b) were plotted against each other. The top ranking binding site-related receptors are highlighted in black while in (b), the phylogenetically-related endothelin receptors are highlighted in red. Dotted line represents a linear regression of the data. For ACKR3, GPCR-CoINPocket correctly identified the pharmacological similarity with the distantly related CXCR4 receptor. However, for GPR37L1, we can see that receptors had rather moderate GPCR-CoINPocket scores. (c) The respective binding site residues are shown for GPR37L1 and its related receptors, as defined by contact strength fingerprints. The aggregated contact strength across the Class A GPCR Pocketome entries is represented by blue circles and by the color intensity of the alignment in the respective positions. Receptors have been named according to the IUPHAR/BPS Concise Guide to PHARMACOLOGY.

Figure 5. Screening for GPR37L1 surrogate ligands

HEK293 cells were transiently transfected with pcDNA3 or GPR37L1 and a cAMP response element luciferase reporter (CRE-luciferase) 24 h prior to the addition of ligands. (a) Initial single concentration screen of various bombesin, orexin and neuropeptide S receptor antagonists, with triangles indicating increasing or decreasing selectivity for one OX receptor subtype over the other. n = 3 independent biological replicates. (b) Chemical structures of the five ligands that reduced GPR37L1-mediated CRE-luciferase levels. Concentration-response curves for SHA-68, n = 3, (c), JNJ 10397049, n = 4 (d) and SB 674042, n = 4 (e), following confirmation of GPR37L1 specificity by ligand aggregation counter-screening. The mean and s.e.m is reported for each point.