Intense bitterness of molecules: Machine learning for expediting drug discovery.

Drug development is a long, expensive and multistage process geared to achieving safe drugs with high efficacy. A crucial prerequisite for completing the medication regimen for oral drugs, particularly for pediatric and geriatric populations, is achieving taste that does not hinder compliance. Currently, the aversive taste of drugs is tested in late stages of clinical trials. This can result in the need to reformulate, potentially resulting in the use of more animals for additional toxicity trials, increased financial costs and a delay in release to the market. Here we present BitterIntense, a machine learning tool that classifies molecules into "very bitter" or "not very bitter", based on their chemical structure. The model, trained on chemically diverse compounds, has above 80% accuracy on several test sets. Our results suggest that about 25% of drugs are predicted to be very bitter, with even higher prevalence (~40%) in COVID19 drug candidates and in microbial natural products. Only ~10% of toxic molecules are predicted to be intensely bitter, and it is also suggested that intense bitterness does not correlate with hepatotoxicity of drugs. However, very bitter compounds may be more cardiotoxic than not very bitter compounds, possessing significantly lower QPlogHERG values. BitterIntense allows quick and easy prediction of strong bitterness of compounds of interest for food, pharma and biotechnology industries. We estimate that implementation of BitterIntense or similar tools early in drug discovery process may lead to reduction in delays, in animal use and in overall financial burden.

BitterDB: taste ligands and receptors database in 2019.

BitterDB (http://bitterdb.agri.huji.ac.il) was introduced in 2012 as a central resource for information on bitter-tasting molecules and their receptors. The information in BitterDB is frequently used for choosing suitable ligands for experimental studies, for developing bitterness predictors, for analysis of receptors promiscuity and more. Here, we describe a major upgrade of the database, including significant increase in content as well as new features. BitterDB now holds over 1000 bitter molecules, up from the initial 550. When available, quantitative sensory data on bitterness intensity as well as toxicity information were added. For 270 molecules, at least one associated bitter taste receptor (T2R) is reported. The overall number of ligand-T2R associations is now close to 800. BitterDB was extended to several species: in addition to human, it now holds information on mouse, cat and chicken T2Rs, and the compounds that activate them. BitterDB now provides a unique platform for structure-based studies with high-quality homology models, known ligands, and for the human receptors also data from mutagenesis experiments, information on frequently occurring single nucleotide polymorphisms and links to expression levels in different tissues.

Biomimetic Sensors for the Senses: Towards Better Understanding of Taste and Odor Sensation.

Taste and smell are very important chemical senses that provide indispensable information on food quality, potential mates and potential danger. In recent decades, much progress has been achieved regarding the underlying molecular and cellular mechanisms of taste and odor senses. Recently, biosensors have been developed for detecting odorants and tastants as well as for studying ligand-receptor interactions. This review summarizes the currently available biosensing approaches, which can be classified into two main categories: in vitro and in vivo approaches. The former is based on utilizing biological components such as taste and olfactory tissues, cells and receptors, as sensitive elements. The latter is dependent on signals recorded from animals' signaling pathways using implanted microelectrodes into living animals. Advantages and disadvantages of these two approaches, as well as differences in terms of sensing principles and applications are highlighted. The main current challenges, future trends and prospects of research in biomimetic taste and odor sensors are discussed.

The taste of toxicity: A quantitative analysis of bitter and toxic molecules.

The role of bitter taste-one of the few basic taste modalities-is commonly assumed to signal toxicity and alert animals against consuming harmful compounds. However, it is known that some toxic compounds are not bitter and that many bitter compounds have negligible toxicity while having important health benefits. Here we apply a quantitative analysis of the chemical space to shed light on the bitterness-toxicity relationship. Using the BitterDB dataset of bitter molecules, The BitterPredict prediction tool, and datasets of toxic compounds, we quantify the identity and similarity between bitter and toxic compounds. About 60% of the bitter compounds have documented toxicity and only 56% of the toxic compounds are known or predicted to be bitter. The LD50 value distributions suggest that most of the bitter compounds are not very toxic, but there is a somewhat higher chance of toxicity for known bitter compounds compared to known nonbitter ones. Flavonoids and alpha acids are more common in the bitter dataset compared with the toxic dataset. In contrast, alkaloids are more common in the toxic datasets compared to the bitter dataset. Interestingly, no trend linking LD50 values with the number of activated bitter taste receptors (TAS2Rs) subtypes is apparent in the currently available data. This is in accord with the newly discovered expression of TAS2Rs in several extra-oral tissues, in which they might be activated by yet unknown endogenous ligands and play non-gustatory physiological roles. These results suggest that bitter taste is not a very reliable marker for toxicity, and is likely to have other physiological roles. © 2017 IUBMB Life, 69(12):938-946, 2017.

Bitter or not? BitterPredict, a tool for predicting taste from chemical structure.

Bitter taste is an innately aversive taste modality that is considered to protect animals from consuming toxic compounds. Yet, bitterness is not always noxious and some bitter compounds have beneficial effects on health. Hundreds of bitter compounds were reported (and are accessible via the BitterDB http://bitterdb.agri.huji.ac.il/dbbitter.php ), but numerous additional bitter molecules are still unknown. The dramatic chemical diversity of bitterants makes bitterness prediction a difficult task. Here we present a machine learning classifier, BitterPredict, which predicts whether a compound is bitter or not, based on its chemical structure. BitterDB was used as the positive set, and non-bitter molecules were gathered from literature to create the negative set. Adaptive Boosting (AdaBoost), based on decision trees machine-learning algorithm was applied to molecules that were represented using physicochemical and ADME/Tox descriptors. BitterPredict correctly classifies over 80% of the compounds in the hold-out test set, and 70-90% of the compounds in three independent external sets and in sensory test validation, providing a quick and reliable tool for classifying large sets of compounds into bitter and non-bitter groups. BitterPredict suggests that about 40% of random molecules, and a large portion (66%) of clinical and experimental drugs, and of natural products (77%) are bitter.

The bitter pill: clinical drugs that activate the human bitter taste receptor TAS2R14.

Bitter taste receptors (TAS2Rs) mediate aversive response to toxic food, which is often bitter. These G-protein-coupled receptors are also expressed in extraoral tissues, and emerge as novel targets for therapeutic indications such as asthma and infection. Our goal was to identify ligands of the broadly tuned TAS2R14 among clinical drugs. Molecular properties of known human bitter taste receptor TAS2R14 agonists were incorporated into pharmacophore- and shape-based models and used to computationally predict additional ligands. Predictions were tested by calcium imaging of TAS2R14-transfected HEK293 cells. In vitro testing of the virtual screening predictions resulted in 30-80% success rates, and 15 clinical drugs were found to activate the TAS2R14. hERG potassium channel, which is predominantly expressed in the heart, emerged as a common off-target of bitter drugs. Despite immense chemical diversity of known TAS2R14 ligands, novel ligands and previously unknown polypharmacology of drugs were unraveled by in vitro screening of computational predictions. This enables rational repurposing of traditional and standard drugs for bitter taste signaling modulation for therapeutic indications.

BitterDB: a database of bitter compounds.

Basic taste qualities like sour, salty, sweet, bitter and umami serve specific functions in identifying food components found in the diet of humans and animals, and are recognized by proteins in the oral cavity. Recognition of bitter taste and aversion to it are thought to protect the organism against the ingestion of poisonous food compounds, which are often bitter. Interestingly, bitter taste receptors are expressed not only in the mouth but also in extraoral tissues, such as the gastrointestinal tract, indicating that they may play a role in digestive and metabolic processes. BitterDB database, available at http://bitterdb.agri.huji.ac.il/bitterdb/, includes over 550 compounds that were reported to taste bitter to humans. The compounds can be searched by name, chemical structure, similarity to other bitter compounds, association with a particular human bitter taste receptor, and so on. The database also contains information on mutations in bitter taste receptors that were shown to influence receptor activation by bitter compounds. The aim of BitterDB is to facilitate studying the chemical features associated with bitterness. These studies may contribute to predicting bitterness of unknown compounds, predicting ligands for bitter receptors from different species and rational design of bitterness modulators.

Modeling of human prokineticin receptors: interactions with novel small-molecule binders and potential off-target drugs.

BACKGROUND AND MOTIVATION: The Prokineticin receptor (PKR) 1 and 2 subtypes are novel members of family A GPCRs, which exhibit an unusually high degree of sequence similarity. Prokineticins (PKs), their cognate ligands, are small secreted proteins of ∼80 amino acids; however, non-peptidic low-molecular weight antagonists have also been identified. PKs and their receptors play important roles under various physiological conditions such as maintaining circadian rhythm and pain perception, as well as regulating angiogenesis and modulating immunity. Identifying binding sites for known antagonists and for additional potential binders will facilitate studying and regulating these novel receptors. Blocking PKRs may serve as a therapeutic tool for various diseases, including acute pain, inflammation and cancer. METHODS AND RESULTS: Ligand-based pharmacophore models were derived from known antagonists, and virtual screening performed on the DrugBank dataset identified potential human PKR (hPKR) ligands with novel scaffolds. Interestingly, these included several HIV protease inhibitors for which endothelial cell dysfunction is a documented side effect. Our results suggest that the side effects might be due to inhibition of the PKR signaling pathway. Docking of known binders to a 3D homology model of hPKR1 is in agreement with the well-established canonical TM-bundle binding site of family A GPCRs. Furthermore, the docking results highlight residues that may form specific contacts with the ligands. These contacts provide structural explanation for the importance of several chemical features that were obtained from the structure-activity analysis of known binders. With the exception of a single loop residue that might be perused in the future for obtaining subtype-specific regulation, the results suggest an identical TM-bundle binding site for hPKR1 and hPKR2. In addition, analysis of the intracellular regions highlights variable regions that may provide subtype specificity.

Structure-based identification of catalytic residues.

The identification of catalytic residues is an essential step in functional characterization of enzymes. We present a purely structural approach to this problem, which is motivated by the difficulty of evolution-based methods to annotate structural genomics targets that have few or no homologs in the databases. Our approach combines a state-of-the-art support vector machine (SVM) classifier with novel structural features that augment structural clues by spatial averaging and Z scoring. Special attention is paid to the class imbalance problem that stems from the overwhelming number of non-catalytic residues in enzymes compared to catalytic residues. This problem is tackled by: (1) optimizing the classifier to maximize a performance criterion that considers both Type I and Type II errors in the classification of catalytic and non-catalytic residues; (2) under-sampling non-catalytic residues before SVM training; and (3) during SVM training, penalizing errors in learning catalytic residues more than errors in learning non-catalytic residues. Tested on four enzyme datasets, one specifically designed by us to mimic the structural genomics scenario and three previously evaluated datasets, our structure-based classifier is never inferior to similar structure-based classifiers and comparable to classifiers that use both structural and evolutionary features. In addition to the evaluation of the performance of catalytic residue identification, we also present detailed case studies on three proteins. This analysis suggests that many false positive predictions may correspond to binding sites and other functional residues. A web server that implements the method, our own-designed database, and the source code of the programs are publicly available at http://www.cs.bgu.ac.il/∼meshi/functionPrediction.