Drug discovery is a complex, expensive, time-consuming, and challenging process.
Continuous drug discovery pipeline still fails to fully address the diversity of patient’s cancer including bad responders and latent off-target responders. Therefore, predictive in vivo models combining short and cost-efficient studies able to predict efficacy and toxicity are highly sought after. Zebrafish (Danio rerio) is a vertebrate sharing 70% genetic homology with human. Its transparency during external embryonic development makes it a model of choice for high-content in vivo studies in the respect of the 3R regulation. Disease modelling and toxicological tests can be performed through flexible approaches. The zebrafish model has already contributed to several successful phenotype-based drug discovery. Here, we describe a 3-steps pipeline for drug discovery that addresses both efficacy and toxicity in cancer drug discovery context using the zebrafish model. We first developed a method to study in vivo tumor progression and its microenvironment modification. Stable fluorescent glioblastoma and melanoma cells where respectively xenografted in zebrafish embryos. Tumor growth, metastatic process, and cancer cells interaction with vascular and immune system were analyzed in 3D to study neovascularization or macrophages dynamic in real time. This first phenotype-based screening realized in 96 well plate allowed the selection of the most efficient compounds. The oncostatic, antimetastatic or potentiator effects of these compounds were characterized in detail on our zebrafish model. Finally, we assessed toxicology through a multisystemic high-content approach. We performed toxicity studies of selected hits on several organs (Heart, Liver, Kidney & Central Nervous System) known as major attrition factor during drug development. Heart anomalies (cardiac rhythm, arrythmia & cardiac arrest), liver abnormalities (hepatomegaly & steatosis), Kidney failures and cystic fibrosis, as well as the identification of neuroactive compounds which may alter behavior have been analyzed. This pipeline offers a powerful tool to select efficient compounds and de-risk drug discovery process before regulatory preclinical studies. The same approach is also applicable to other therapeutic fields (CNS disorders, genetic diseases, disease models).