You are here:
Identifying Structural Alerts Based on Zebrafish Developmental Morphological Toxicity (TDS)
Heish, J., N. Kleinstreuer, J. Franzosa, A. Sedykh, AND R. Tice. Identifying Structural Alerts Based on Zebrafish Developmental Morphological Toxicity (TDS). Presented at 2nd ToxCast Data Summit, RTP, NC, September 29 - 30, 2014. https://doi.org/10.23645/epacomptox.5197081
Identifies structural characteristic of environmental compounds that elicit developmental toxicity in zebrafish.
Zebrafish constitute a powerful alternative animal model for chemical hazard evaluation. To provide an in vivo complement to high-throughput screening data from the ToxCast program, zebrafish developmental toxicity screens were conducted on the ToxCast Phase I (Padilla et al., 2012) and Phase II chemical libraries (Truong et al., 2014). The screens had several differences in experimental design and data analysis. Padilla et al. assigned Toxicity Scores based on descriptive data (lethality, hatching) and malformation assessments (binary and numeric). The Toxicity Scores were used to plot concentration response curves and estimate half-maximal activity concentrations. Truong et al. evaluated morphological endpoints by assigning binary incidence counts and computing lowest effect levels. This binomial count approach preserves independent morphological information and does not constrain dose-response behavior (e.g., response can be non-monotonic). In contrast, the Toxicity Score approach aggregates morphological data and assumes monotonic dose-response behavior. Utilizing the relative strengths of both approaches, we conducted a study to identify and analyze correlated morphological endpoints based on the Phase II data. We found endpoints that are closely related, for example, the jaw, snout, and pericardial and yolk sac edema, and grouped them into developmental process endpoints. Based on the new endpoints, we generated Severity Scores to plot concentration response curves. Based on the curve assumed to be monotonic, we calculated the point-of-departure (POD), which is the concentration where a response exceeds the baseline noise level. Quantitative structure activity relationship (QSAR) analyses using the CASE Ultra structural alert mining system yielded predictive models for chemicals likely to induce developmental malformations in zebrafish. The externally validated balanced accuracies of the QSAR models ranged from 0.63-0.74. Several structural alerts related to zebrafish developmental toxicity were identified such as the ortho-dimethoxyaryl group (e.g., as in colchicine), thioamide group (e.g., in ziram), and benzothiazole scaffold (e.g., in 2-mercaptobenzothiazole). In summary, we identified 220 compounds in the Phase II library that elicit adverse developmental responses (including general toxicity) under specific process based mechanisms with POD values interpolated from dose-response curves. We further developed QSAR models to evaluate large chemical libraries, such asTox21, and prioritize chemicals for further investigation.