Virtual Screen for Ligands of Orphan G Protein-Coupled Receptors

Abstract

This paper describes a virtual screening methodology that generates a ranked list of high-binding small molecule ligands for orphan G protein-coupled receptors (oGPCRs), circumventing the requirement for receptor three-dimensional structure determination. Features representing the receptor are based only on physicochemical properties of primary amino acid sequence, and ligand features use the two-dimensional atomic connection topology and atomic properties. An experimental screen comprised nearly 2 million hypothetical oGPCR-ligand complexes, from which it was observed that the top 1.96% predicted affinity scores corresponded to "highly active" ligands against orphan receptors. Results representing predicted high-scoring novel ligands for many oGPCRs are presented here. Validation of the method was carried out in several ways: (1) A random permutation of the structure-activity relationship of the training data was carried out; by comparing test statistic values of the randomized and nonshuffled data, we conclude that the value obtained with nonshuffled data is unlikely to have been encountered by chance. (2) Biological activities linked to the compounds with high cross-target binding affinity were analyzed using computed log-odds from a structure-based program. This information was correlated with literature citations where GPCR-related pathways or processes were linked to the bioactivity in question. (3) Anecdotal, out-of-sample predictions for nicotinic targets and known ligands were performed, with good accuracy in the low-to-high "active" binding range. (4) An out-of-sample consistency check using the commercial antipsychotic drug olanzapine produced "active" to "highly-active" predicted affinities for all oGPCRs in our study, an observation that is consistent with documented findings of cross-target affinity of this compound for many different GPCRs. It is suggested that this virtual screening approach may be used in support of the functional characterization of oGPCRs by identifying potential cognate ligands. Ultimately, this approach may have implications for pharmaceutical therapies to modulate the activity of faulty or disease-related cellular signaling pathways. In addition to application to cell surface receptors, this approach is a generalized strategy for discovery of small molecules that may bind intracellular enzymes and involve protein-protein interactions.

Discovery and Validation of Novel Peptide Agonists for G-protein-coupled Receptors

“…Computational screening can supplement experimental efforts as a preliminary screening tool for efficient identification of natural ligands, especially peptides (10). Bioinformatics can serve as a powerful tool that provides reliable predictive measures to select for the high potential candidates and provide a spotlight pointed at potential new candidates for experimental discovery, thereby enabling higher success rates in identification of novel GPCR ligands (11).…”

Receptor-based QSAR studies of non-peptide human oxytocin receptor antagonists

“…The cardinal preoccupation is that only the crystal structure of bovine rhodopsin (bRho) is available, therefore the homology modelling (at least 20% sequence identity is necessary) is the only way to model the 3D structure of these receptors. After the homology modelling the automated docking and virtual screening [9][10][11][12] is possible.…”

Structure‐Based Approaches to Target Fishing and Ligand Profiling

“…[90][91][92] In a pioneering work, Bock et al [90] used rather standard 2-D topological and atomic descriptors for ligands, physicochemical properties of amino acid sequences for receptors, and concatenate feature vectors for both the receptor and the ligand in a single fingerprint. A support vector machine (SVM) model was trained on 5319 receptor-ligand pairs from the PDSP Ki database [93] to predict the Ki of any ligand to any GPCR and used to propose novel ligands for orphan GPCRs.…”