Transcriptomic Points of Departure Calculated from Rainbow Trout Gill, Liver, and Gut Cell Lines Exposed to Methylmercury and Fluoxetine

Abstract

Ethical and resource limitation concerns are pushing chemicals management to develop alternatives to animal testing strategies. The objective of our study was to determine whether transcriptomic point of departure (tPOD) values could be derived from studies that followed Organisation for Economic Co‐operation and Development (OECD) Test No. 249 (rainbow trout gill cell line), as well as from studies on trout liver and gut cells. Gill, liver, and gut cell lines were exposed to methylmercury and fluoxetine. Concentrations causing 50% cytotoxicity (LC50) were derived, the whole transcriptome was sequenced, and gene tPOD and pathway benchmark dose (BMD) values were derived from transcriptomic dose–response analysis. Differences in LC50 and transcriptomic responses across the cell lines were noted. For methylmercury, the tPODmode values were 14.5, 20.5, and 17.8 ppb for the gill, liver, and gut cells, respectively. The most sensitive pathway (pathway BMDs in parentheses) was ferroptosis in the gill (3.1 ppb) and liver (3.5 ppb), and glutathione metabolism in the gut (6.6 ppb). For fluoxetine, the tPODmode values were 109.4, 108.4, and 97.4 ppb for the gill, liver, and gut cells, respectively. The most sensitive pathway was neurotrophin signaling in the gill (147 ppb) and dopaminergic signaling in the gut (86.3 ppb). For both chemicals, the gene tPOD and pathway BMD values were lower than cytotoxic concentrations in vitro, and within 10‐fold below the in vivo LC50s. By bringing together transcriptomics and dose–response analysis with an OECD test method in three cell lines, the results help to establish an in vitro method yielding tPOD values that are hypothesized to be protective of in vivo concentrations associated with adverse outcomes, and also give insights into mechanisms of action. Environ Toxicol Chem 2022;41:1982–1992. © 2022 SETAC

Characterizing Variability and Uncertainty Associated with Transcriptomic Dose–Response Modeling

“…Overall, the analyses presented in Figures and suggest that the dose range and spacing have a greater impact than the replicate number on the final tPOD values because there is more variability in tPOD values along rows than along columns. This observation is consistent with our previous findings that low-replicate TDRA experiments give similar results to high-replicate ones . Thus, given the fixed sample size due to financial or logistical constraints, we suggest that a lower replicate number be used ( n = 2 or 3 per group) so that a wider range of doses can be included .…”

“…This observation is consistent with our previous findings that lowreplicate TDRA experiments give similar results to highreplicate ones. 15 Thus, given the fixed sample size due to financial or logistical constraints, we suggest that a lower replicate number be used (n = 2 or 3 per group) so that a wider range of doses can be included. 31 The need for an adequate dose range is underscored by previous recommendations that the lowest dose should be below the most sensitive transcriptomic response for accurate determination of all gene BMD values.…”

“…Our analysis was based on the results from a single study system: Japanese quail embryos exposed to CPF. In relation to other common toxicological models, this study system contains a moderate amount of variability (e.g., embryos contained in eggs have less variability than adult animals, 34 but more than genetically identical cell lines 15 ). In addition, our dataset contains both males and females, randomly allocated across the dose groups.…”

“…As uptake of TDRA by the toxicology community increases, − there is a need for practical guidance concerning experimental design, data analysis, and interpretation of the results. This guidance must be founded upon a solid understanding of uncertainty (lack of knowledge) and variability (unexplained differences between observations in or between populations) of each component of the analysis at the gene, pathway, and transcriptome levels. , Extensive work has gone into establishing best practices for the data analysis component of TDRA, for example, studies that have investigated variation of results in relation to different normalization, filtering, and gene BMD summary methods. ,, In these studies, the actual datasets were fixed while aspects of the data analysis methods varied.…”

Framework for multi-stressor physiological response evaluation in amphibian risk assessment and conservation

“…Changes in gene expression can reveal which genes are upregulated or downregulated, thereby identifying perturbations in specific biochemical pathways regardless of the origin of the stressor. The magnitude of the response could indicate functional points of departure (e.g., Ewald et al, 2022;Mittal et al, 2022). Transcriptomics data, generated from controlled laboratory exposures, provide a comprehensive view of gene expression changes comparable to traditional apical endpoints.…”