220-LB: Characterizing the Role of 99 Cardiometabolic Candidate Genes Using CRISPR/Cas9, In Vivo Imaging, and Deep Learning in Zebrafish Larvae

Introduction: Genome-wide association studies identified 100s of loci that are associated with the risk of common cardio-metabolic diseases (T2D and CVD). Here we functionally characterise 99 candidate genes for a role in relevant traits using zebrafish larvae.

Methods: Fifty-six T2D and 46 CVD candidate genes (three overlap) were functionally characterised by targeting their zebrafish orthologs using CRISPR/Cas9. On day 9-11 post-fertilisation, we acquired optical sections of the pancreatic islet, liver and vasculature using semi-automated fluorescence microscopy, in 16,141 CRISPR/Cas9-edited zebrafish larvae, followed by image analysis using deep learning.

Results: Perturbing all nine T2D genes with at least moderate prior evidence affects ≥1 T2D trait (i.e., beta cell mass, beta cell insulin expression, liver fat, and/or glucose content), while no consequence is observed for 22 genes without prior evidence of a role in T2D. For 12 genes with at most modest prior evidence, perturbation also affects at least 1 T2D trait. Perturbation of 23 genes prioritised for a role in CVD affects at least one vascular trait, while no consequence is observed for 15 non-prioritised genes. Effects of mutations in mice and/or humans are published for 14 of 23 genes, 78% of which show directionally consistent effects across zebrafish larvae and mammals. Of the 15 genes influencing liver fat upon CRISPR/Cas9 editing, mutations in 12 previously showed an effect on liver fat in mice and/or humans, with 75% congruence in direction of effect across species.

Conclusion: Systematically characterising candidate genes for a role in image-based T2D, liver fat and CVD traits in CRISPR/Cas9-edited zebrafish larvae shows highly congruent results with mice and humans, and prioritises promising genes for further in-depth characterisation.