Development and application of a metabolomic tool to assess exposure of an estuarine amphipod to pollutants in the environment.

Identifying major adverse effects on aquatic organisms in environmental samples is still challenging, and metabolomic approaches have been utilized as non-target screening techniques in the context of ecotoxicology. While existing methods have focused on statistical tests or univariate analysis, there is the need to further explore a multivariate analytical method that captures synergetic effects and associations among metabolites and toxicants. Here we show a new tool for screening sediment toxicity in the environment. First, we constructed predictive models using the metabolomic profiles and the result of exposure tests, to discriminate the toxic effects of target substances. The developed models were then applied to sediment samples collected from an actual urban area that contain chromium, nickel, copper, zinc, cadmium, fluoranthene, nicotine, and osmotic stress, incorporated with exposure tests of the benthic amphipod Grandidierella japonica. As a result, the fitted models showed high predictive power (Q > 0.71) and could detect toxicants from mixed chemical samples across a wide range of concentrations in test datasets. The application of the constructed models to river sediment and road dust samples indicated that almost all target substances were less toxic compared with the effects at LC50 levels. Only zinc showed slight increasing trends among samples, suggesting that the proposed method can be used for prioritization of toxicants. The present work made a direct connection between chemical exposures and metabolomic responses, and draws attention to the need for further studies on interactive mechanisms of metabolites in toxicological assessments.