Neurotoxic effects in zebrafish embryos by valproic acid and nine of its analogues: the fish-mouse connection?

Since teratogenicity testing in mammals is a particular challenge from an animal welfare perspective, there is a great need for the development of alternative test systems. In this context, the zebrafish (Danio rerio) embryo has received increasing attention as a non-protected embryonic vertebrate in vivo model. The predictive power of zebrafish embryos for general vertebrate teratogenicity strongly depends on the correlation between fish and mammals with respect to both overall general toxicity and more specific endpoints indicative of certain modes-of-action. The present study was designed to analyze the correlation between (1) effects of valproic acid and nine of its analogues in zebrafish embryos and (2) their known neurodevelopmental effects in mice. To this end, zebrafish embryos exposed for 120 h in an extended version of the acute fish embryo toxicity test (FET; OECD TG 236) were analyzed with respect to an extended list of sublethal endpoints. Particular care was given to endpoints putatively related to neurodevelopmental toxicity, namely jitter/tremor, deformation of sensory organs (eyes) and craniofacial deformation, which might correlate to neural tube defects caused by valproic acid in mammals. A standard evaluation of lethal (LC according to OECD TG 236) and sublethal toxicity (EC) merely indicated that four out of ten compounds tested in zebrafish correlate with positive results in mouse in vivo studies. A detailed assessment of more specific effects, however, namely, jitter/tremor, small eyes and craniofacial deformation, resulted in a correspondence of 75% with in vivo mouse data. A refinement of endpoint analysis from an integration of all observations into one LCx or ECx data (as foreseen by current ecotoxicology-driven OECD guidelines) to a differential evaluation of endpoints specific of selected modes-of-action thus increases significantly the predictive power of the zebrafish embryo model for mammalian teratogenicity. However, for some of the endpoints observed, e.g., scoliosis, lordosis, pectoral fin deformation and lack of movement, further experiments are required for the identification of underlying modes-of-action and an unambiguous interpretation of their predictive power for mammalian toxicity.