Optimizing Quantitative Phenotyping of Embryonic Disease Models

Using Xenopus tropicali as a model for polycystic kidney disease, as well as light sheet microscopy and deep learning, researchers at the University of Zurich established a framework for higher-throughput characterization of embryonic model organisms.

The team used CRISPR/Cas9 to target genes known to play a role in cystic kidney disease. “Our novel frog models develop cysts in the kidneys within only a few days, allowing us to observe these disease processes in real time for the first time,” says Thomas Naert, first author of a paper published in Development today. While most genetic studies are performed on mice, frogs have features that make them well-suited for larger scale studies. “One frog couple can produce hundreds or even thousands of eggs,” says Naert. “

To analyze the data from such a large number of animals, the team employed light-sheet microscopy, which produced a 3D reconstruction of the entire tadpole and all its organs. Much like magnetic resonance imaging, light-sheet techniques make it possible to see through tissues in tadpoles to find disease-affected organs. The collected data was then processed using a deep learning approach to allow rapid, automated assessment of disease.

“While it would normally take my team several days or even weeks to analyze data from hundreds of tadpoles, artificial intelligence can now do this task in a matter of hours,” explains study leader Soeren Lienkamp.

The findings from frog models analyzed in this way provide new insights into the early processes of polycystic kidney disease, as well as show that deep-learning approaches can be harnessed to accelerate and automate accurate quantitative phenotyping of embryonic disease models.