"Bitter Taste Signals Toxicity": Assessing the Paradigm using Chemoinformatics

Bitter taste is one of the basic taste modalities and is commonly assumed to signal toxicity and alert animals against consuming harmful substances, but no quantitative study has been conducted to validate this assumption until this point. It is known that some toxic compounds are not bitter and many bitter compounds have negligible toxicity while presenting profound health benefits. Furthermore, the receptors for bitter compounds, Tas2Rs - a subfamily of G-protein-coupled receptors (GPCRs) - are expressed in extra-oral tissues. All this suggests additional physiological roles that may not be related to food choices or toxicity. Building upon the recently established database of bitter compounds – the BitterDB – and using several datasets of toxic compounds, we set to quantify the bitterness-toxicity relationship. We found that the overlap between the bitter and toxic compounds is only partial. Additionally, bitter acute oral LD₅₀ values are much higher compared to the toxic datasets. We therefore assess the similarity of the bitter and the toxic datasets and apply the newly developed bitterness-prediction algorithm. While more than half of the bitter compounds have some toxicity, only a third of the toxic compounds are expected to be bitter. The generalistic approach used allowed quantif​ication of the bitter and toxic chemical spaces and clarification of​ their relationship, suggesting that bitter taste ​is not merely a​ gatekeeper for toxicity.