The potential of a metabolomics method to detect statistically significant perturbations in the metabolome of an organism is enhanced by excellent analytical precision, unequivocal identification, and broad metabolomic coverage. While the former two metrics are usually associated with targeted metabolomics and the latter with non-targeted metabolomics, a systematic comparison of the performance of both approaches has not yet been carried out. The present work reports on the development and performance evaluation of separate targeted and non-targeted metabolomics methods. The targeted approach facilitated determination of 181 metabolites (quantitative analysis of 18 amino acids, 11 biogenic amines, 5 neurotransmitters, 5 nucleobases and semi-quantitative analysis of 50 carnitines, 83 phos-phatidylcholines, and 9 sphingomyelins) using ultra-performance liquid chromatography - tandem mass spectrometry (UPLC-MS/MS) and flow injection-tandem mass spectrometry (Fl-MS/MS). Method accuracy and/or precision were assessed using replicate samples of NIST SRM1950 as well as fish liver and brain tissue from Gilthead Bream (Sparus aurata). The non-target approach involved UPLC-high resolution (Orbitrap) mass spectrometry (UPLC-HRMS). Testing of ionization mode and stationary phase revealed that a combination of positive electrospray ionization and HILIC chromatography produced the largest number of chromatographic features during non-target analysis. Furthermore, an evaluation of 4 different sequence drift correction algorithms, and combinations thereof, revealed that batchCorr produced the best precision in almost every test. However, even following correction of non-target data for signal drift, the precision of targeted data was better, confirming our existing assumptions about the strengths of targeted metabolomics. Finally, the accuracy of the online MS2-library mzCloud was evaluated using reference standards for 38 different metabolites. This is among the few studies that have systematically evaluated the performance of targeted and non-targeted metabolomics and provides new insight into the advantages and disadvantages of each approach.

Toxicometabolomic studies involving zebrafish embryos have become increasingly popular for linking apical endpoints to biochemical perturbations as part of adverse outcome pathway determination. These experiments involve pooling embryos to generate sufficient biomass for metabolomic measurement, which adds both time and cost. To address this limitation, we developed a high-throughput toxicometabolomic assay involving single zebrafish embryos. Incubation, microscopy, embryo extraction, and instrumental metabolomic analysis were all performed in the same 96-well plate, following acquisition of conventional toxicological endpoints. The total time for the assay (including testing of 6 doses/n = 12 embryos per dose plus positive and negative controls, assessing conventional endpoints, instrumental analysis, data processing and multivariate statistics) is <14 days. Metabolomic perturbations at low dose were linked statistically to those observed at high dose and in the presence of an adverse effect, thereby contextualizing omic data amongst apical endpoints. Overall, this assay enables collection of high resolution metabolomic data in a high throughput manner, suitable for mode of action hypothesis generation in the context of pharmaceutical or toxicological screening.

A procedure was developed for rapid screening of xenobiotic biotransformationproducts (bioTPs) in single zebrafish (ZF; Danio rerio) embryos. Exposure was carried out from 0-120 hours post fertilization (hpf) to 6 different concentrations of the model compound propranolol (PPL). Following in-plate extraction and non-target instrumental analysis by high resolution mass spectrometry, suspected bioTPs were identified using custom data filtration scripts and matching to in silico structural predictions. A total of eight PPL bioTPs were identified (two at a level 1 confidence and six at a level 2-3 confidence). These findings supplement previously generated toxicometabolomic models derived from the same dataset, and were obtained without conducting additional exposure experiments. In addition to facilitating assessments of inter-individual variability in bioTP production in ZF embryos, we demonstrate that bioTPs can be elucidated using extremely small quantities of biomass. To the best of our knowledge, this is the first time bioTP elucidation has been carried out in single ZFembryos.

We report here on the high throughput determination of biotransformation products(bioTPs) and toxicometabolomics in single zebrafish (ZF) embryos exposed to carbamazepine(CBZ). Exposures were carried out in 96-well plates with six CBZ concentrations ranging from 0.5 μg/L to 50 mg/L (n=12 embryos/dose). In the 50 mg/L dose group 33% of the embryos developed edema during the exposure (120hpf) while hatching was significantly delayed in three of the lower dose groups (0.46, 3.85 and the 445 μg/L) compared to the control at 48 hpf. Toxicometabolomic analysis together with random forest modelling revealed a total of 80 significantly affected metabolites (22 identified via targeted lipidomics and 58 via non-target analysis). The wide range of doses tested enabled observation of both monotonic and nonmonotonic dose-responses which fit the known mode of action of CBZ. A novel pathway was also proposed based on changes in PE-Cer (d16:2/24:1; CL 2-3) which could be the result of CBZ induced apoptosis via induction of the enzyme SAMD8. In addition, 2 CBZ bioTPs were identified without additional exposure experiments. Overall this work showcases the potential of toxicometabolomics and bioTP determination in single ZF embryos for improved and comprehensive chemical hazard assessment.