Knowledge-Driven Approaches to Create the MTox700+ Metabolite Panel for Predicting Toxicity

Abstract

Endogenous metabolite levels describe the molecular phenotype that is most downstream from chemical exposure. Consequently, quantitative changes in metabolite levels have the potential to predict mode-of-action and adversity, with regulatory toxicology predicated on the latter. However, toxicity-related metabolic biomarker resources remain highly fragmented and incomplete. While development of the S1500+ gene biomarker panel has accelerated the application of transcriptomics to toxicology, a similar initiative for metabolic biomarkers is lacking. Our aim was to define a publicly-available metabolic biomarker panel, equivalent to S1500+, capable of predicting pathway perturbations and/or adverse outcomes. We conducted a systematic review of multiple toxicological resources, yielding 189 proposed metabolic biomarkers from existing assays (BASF, Bowes-44 and Tox21), 342 biomarkers from databases (Adverse Outcome Pathway Wiki, Comparative Toxicogenomics Database, QIAGEN Ingenuity Pathway Analysis, and Toxin and Toxin-Target Database), and 435 biomarkers from the literature. Evidence mapping across all eight resources generated a panel of 722 metabolic biomarkers for toxicology (MTox700+), of which 462 (64%) are associated with molecular pathways and 575 (80%) with adverse outcomes. Comparing MTox700+ and S1500+ revealed that 418 (58%) metabolic biomarkers associate with pathways shared across both panels, with further metabolites mapping to unique pathways. Metabolite reference standards are commercially available for 646 (90%) of the panel metabolites, and assays exist for 578 (80%) of these biomarkers. This study has generated a publicly-available metabolic biomarker panel for toxicology, which through its future laboratory deployment, is intended to help build foundational knowledge to support the generation of molecular mechanistic data for chemical hazard assessment.

From Pipeline to Plant Protection Products: Using New Approach Methodologies (NAMs) in Agrochemical Safety Assessment

“…A panel of over 700 metabolites has been proposed as biomarkers of toxicity based upon a compendium of publicly available databases. 61 Validating such proposed metabolite biomarkers and developing appropriate rapid, nonanimal assays to query metabolite biomarkers is a significant challenge requiring sufficient positive and negative control data sets and extensive systems biology knowledge.…”

“…There have been numerous applications for metabolomics in drug discovery, precision medicine, agriculture, food and beverage tests, and petrochemical analysis. − From an agrochemical discovery standpoint, metabolomics may be a valuable tool to identify novel biomarkers that are associated with undesirable modes of action (i.e., toxicity) and to monitor the effects of known biomarkers as analytical endpoints for the compounds of interest. A panel of over 700 metabolites has been proposed as biomarkers of toxicity based upon a compendium of publicly available databases . Validating such proposed metabolite biomarkers and developing appropriate rapid, nonanimal assays to query metabolite biomarkers is a significant challenge requiring sufficient positive and negative control data sets and extensive systems biology knowledge.…”

Derivation of metabolic point of departure using high-throughput in vitro metabolomics: investigating the importance of sampling time points on benchmark concentration values in the HepaRG cell line

“…Metabolites present in the HepaRG-specific library were searched against the MTox700+ biomarker list (Sostare et al 2022), which contains toxicologically relevant polar metabolic and lipid biomarkers obtained through interrogation of existing literature, databases, and analytical assays. These biomarkers should in principle be equivalent to panels of genes used in transcriptomics studies such as the S1500+ panel developed by the US National Toxicology Program (Mav et al 2018).…”

“…As anticipated, there were few lipid molecules detected that overlapped with MTox700+ , mostly because this biomarker list contains relatively few lipids. These low values are not surprising considering a) the diversity of lipidome and b) that not every putative annotation of a lipid molecule has a corresponding HMDB ID, which is the unique identifier used when searching the MTox700+ metabolite panel (Hu et al 2019;Sostare et al 2022). Next, the putatively annotated lipids were assigned lipid classes according to LipidSearch, as shown in Fig.…”

“…Summary of polar metabolites and lipids measured in the extracts of the HepaRG cell line using HILIC and RP C 30 UHPLC-MS/MS assays in positive and negative ionisation modes Range refers to the worst-and best-case scenarios of a metabolite being present in the MTox700+ biomarker list(Sostare et al 2022), where a metabolite is excluded or included from the count if it has multiple annotations, respectively…”