Abstract
The zebrafish has been widely used as a predictive model in safety and toxicology. Low cost high-throughput screening can be achieved with this model, and the genome contains orthologues of the majority of human disease genes. However, previous studies indicate that the predictivity of the zebrafish model in toxicology varies between compound and compound class. We examined this issue by screening 24 compounds from two different compound classes, metals and biocides (pesticides/insecticides) for toxicity in the zebrafish model and looked at the effects on hatching, morphology and predictivity for mammalian toxicity. Wild-type zebrafish embryos were exposed to test compounds in 96-well plates for 96 hours starting at 24 hours post fertilization. Hatching was either delayed or accelerated depending on the compound. Three types of alteration in behavioural responses were noted: (i) hypoactivity; (ii) hyperactivity; and (iii) biphasic response (a dose-dependent shift between hypo- and hyperactivity). LC50 of compounds was calculated and compared to published LD50 values in rodents. The zebrafish-rodent values were poorly correlated for both metals and biocides. We conclude that, although the zebrafish is a good model for some aspects of toxicology, its predictivity for mammalian toxicity needs to be determined per compound class.