Machine Learning Assisted Approach for Finding Novel High Activity Agonists of Human Ectopic Olfactory Receptors.

Olfactory receptors (ORs) constitute the largest superfamily of G protein-coupled receptors (GPCRs). ORs are involved in sensing odorants as well as in other ectopic roles in non-nasal tissues. Matching of an enormous number of the olfactory stimulation repertoire to its counterpart OR through machine learning (ML) will enable understanding of olfactory system, receptor characterization, and exploitation of their therapeutic potential. In the current study, we have selected two broadly tuned ectopic human OR proteins, OR1A1 and OR2W1, for expanding their known chemical space by using molecular descriptors. We present a scheme for selecting the optimal features required to train an ML-based model, based on which we selected the random forest (RF) as the best performer. High activity agonist prediction involved screening five databases comprising ~23 M compounds, using the trained RF classifier. To evaluate the effectiveness of the machine learning based virtual screening and check receptor binding site compatibility, we used docking of the top target ligands to carefully develop receptor model structures. Finally, experimental validation of selected compounds with significant docking scores through in vitro assays revealed two high activity novel agonists for OR1A1 and one for OR2W1.

Machine learning decodes chemical features to identify novel agonists of a moth odorant receptor.

Odorant receptors expressed at the peripheral olfactory organs are key proteins for animal volatile sensing. Although they determine the odor space of a given species, their functional characterization is a long process and remains limited. To date, machine learning virtual screening has been used to predict new ligands for such receptors in both mammals and insects, using chemical features of known ligands. In insects, such approach is yet limited to Diptera, whereas insect odorant receptors are known to be highly divergent between orders. Here, we extend this strategy to a Lepidoptera receptor, SlitOR25, involved in the recognition of attractive odorants in the crop pest Spodoptera littoralis larvae. Virtual screening of 3 million molecules predicted 32 purchasable ones whose function has been systematically tested on SlitOR25, revealing 11 novel agonists with a success rate of 28%. Our results show that Support Vector Machine optimizes the discovery of novel agonists and expands the chemical space of a Lepidoptera OR. More, it opens up structure-function relationship analyses through a comparison of the agonist chemical structures. This proof-of-concept in a crop pest could ultimately enable the identification of OR agonists or antagonists, capable of modifying olfactory behaviors in a context of biocontrol.

Computational modeling of the olfactory receptor Olfr73 suggests a molecular basis for low potency of olfactory receptor-activating compounds.

The mammalian olfactory system uses hundreds of specialized G-protein-coupled olfactory receptors (ORs) to discriminate a nearly unlimited number of odorants. Cognate agonists of most ORs have not yet been identified and potential non-olfactory processes mediated by ORs are unknown. Here, we used molecular modeling, fingerprint interaction analysis and molecular dynamics simulations to show that the binding pocket of the prototypical olfactory receptor Olfr73 is smaller, but more flexible, than binding pockets of typical non-olfactory G-protein-coupled receptors. We extended our modeling to virtual screening of a library of 1.6 million compounds against Olfr73. Our screen predicted 25 Olfr73 agonists beyond traditional odorants, of which 17 compounds, some with therapeutic potential, were validated in cell-based assays. Our modeling suggests a molecular basis for reduced interaction contacts between an odorant and its OR and thus the typical low potency of OR-activating compounds. These results provide a proof-of-principle for identifying novel therapeutic OR agonists.

Agonists of G-Protein-Coupled Odorant Receptors Are Predicted from Chemical Features.

Predicting the activity of chemicals for a given odorant receptor is a longstanding challenge. Here the activity of 258 chemicals on the human G-protein-coupled odorant receptor (OR)51E1, also known as prostate-specific G-protein-coupled receptor 2 (PSGR2), was virtually screened by machine learning using 4884 chemical descriptors as input. A systematic control by functional in vitro assays revealed that a support vector machine algorithm accurately predicted the activity of a screened library. It allowed us to identify two novel agonists in vitro for OR51E1. The transferability of the protocol was assessed on OR1A1, OR2W1, and MOR256-3 odorant receptors, and, in each case, novel agonists were identified with a hit rate of 39-50%. We further show how ligands' efficacy is encoded into residues within OR51E1 cavity using a molecular modeling protocol. Our approach allows widening the chemical spaces associated with odorant receptors. This machine-learning protocol based on chemical features thus represents an efficient tool for screening ligands for G-protein-coupled odorant receptors that modulate non-olfactory functions or, upon combinatorial activation, give rise to our sense of smell.

The Broadly Tuned Odorant Receptor OR1A1 is Highly Selective for 3-Methyl-2,4-nonanedione, a Key Food Odorant in Aged Wines, Tea, and Other Foods.

Key food odorants are the most relevant determinants by which we detect, recognize, and hedonically evaluate the aroma of foods and beverages. Odorants are detected by our chemical sense of olfaction, comprising a set of approximately 400 different odorant receptor types. However, the specific receptor activity patterns representing the aroma percepts of foods or beverages, as well as the key food odorant agonist profiles of single-odorant receptors, are largely unknown. We aimed to establish comprehensive key food odorant agonist profiles of 2 unrelated, broadly tuned receptors, OR1A1 and OR2W1, that had been associated thus far with mostly non-key food odorants and shared some of these agonists. By screening both receptors against 190 key food odorants in a cell-based luminescence assay, we identified 14 and 18 new key food odorant agonists for OR1A1 and OR2W1, respectively, with 3-methyl-2,4-nonanedione emerging as the most potent agonist for OR1A1 by 3 orders of magnitude, with a submicromolar half maximal effective concentration. 3-Methyl-2,4-nonanedione has been associated with a prune note in oxidized wine and is an aroma determinant in tea and apricots. Further screening against the entire set of 391 human odorant receptors revealed that 30 or 300 µmol/L 3-methyl-2,4-nonanedione activated only 1 receptor, OR1A1, suggesting a unique role of OR1A1 for the most sensitive detection of this key food odorant in wine, tea, and other food matrices.

Discovery of novel ligands for mouse olfactory receptor MOR42-3 using an in silico screening approach and in vitro validation.

The ligands for many olfactory receptors remain largely unknown despite successful heterologous expression of these receptors. Understanding the molecular receptive range of olfactory receptors and deciphering the olfactory recognition code are hampered by the huge number of odorants and large number of olfactory receptors, as well as the complexity of their combinatorial coding. Here, we present an in silico screening approach to find additional ligands for a mouse olfactory receptor that allows improved definition of its molecular receptive range. A virtual library of 574 odorants was screened against a mouse olfactory receptor MOR42-3. We selected the top 20 candidate ligands using two different scoring functions. These 40 odorant candidate ligands were then tested in vitro using the Xenopus oocyte heterologous expression system and two-electrode voltage clamp electrophysiology. We experimentally confirmed 22 of these ligands. The candidate ligands were screened for both agonist and antagonist activity. In summary, we validated 19 agonists and 3 antagonists. Two of the newly identified antagonists were of low potency. Several previously known ligands (mono- and dicarboxylic acids) are also confirmed in this study. However, some of the newly identified ligands were structurally dissimilar compounds with various functional groups belonging to aldehydes, phenyls, alkenes, esters and ethers. The high positive predictive value of our in silico approach is promising. We believe that this approach can be used for initial deorphanization of olfactory receptors as well as for future comprehensive studies of molecular receptive range of olfactory receptors.