Citation:

Collette, Tim, D. Ekman, P. Bradley, AND Q. Teng. Prioritizing Contaminants in Complex Mixtures Using In vitro-based metabolomics and Multivariate Statistics. 2018 SETAC NA Annual Meeting, Sacramento, CA, November 04 - 08, 2018.

Impact/Purpose:

Abstract is for submission to the 39th Annual Society for Environmental Toxicology and Chemistry North American Meeting.

Description:

Chemicals occur as highly complex mixtures in developed-watershed streams, and the makeup of these mixtures may differ dramatically based on land use and a variety of other factors. While it is commonly accepted that some anthropogenic chemicals (and chemical mixtures) are detrimental to ecosystem health, there is rarely compelling evidence to suggest which chemicals, chemical classes, or defined mixtures are the most biologically impactful. To help address some of these needs, the USGS and the EPA partnered to conduct an extensive field study of 38 stream sites across the US during 2012-2014. Water was collected from these sites and subjected to expanded target analysis (916 total analytes). In addition, water samples were split and submitted to a variety of effects-based tools aimed at evaluating the biological impacts of chemicals (and other stressors) in these waters. Here, we present results from an untargeted NMR-based metabolomic assessment of these split water samples, using cultured zebrafish liver cells (ZFL). One goal of this assessment was to use partial least squares (PLS) modeling to determine which of the chemicals and other stressors most strongly co-varied with changes in metabolite profiles, and conversely, which of the stressors showed little to no covariance and thus potentially less biological relevance. Of the 916 total analytes targeted in the contaminant monitoring, 280 displayed significant covariance with impacts on the ZFL metabolome. To provide a coarse ranking of the strength of covariance, the 280 analytes were sorted into quartiles according to the strength of this covariance. Interestingly, those analytes that occupied the top quartile (i.e., most strongly covaried with metabolomics dataset) were not among those detected at the highest concentrations and/or showed the greatest number of detects. Furthermore, most comprised a mix of both legacy contaminants (e.g., dioxins) and contaminants of emerging concern. In addition, a variety of non-chemical stressors (e.g., pH) displayed considerable covariance, highlighting the ability of the untargeted in vitro-based metabolomics approach to capture these types of stresses as well. These and other aspects of this project will be presented to showcase the utility of multivariate statistics and in vitro-based metabolomics for prioritizing contaminants in complex mixtures.

URLs/Downloads:

Record Details: