Interpreting genetic screens using
Thomas M. Norman, Britt Adamson, Marco Jost, Luke Gilbert, Max Horlbeck, Jonathan S. Weissman
Department of Cellular & Molecular Pharmacology, Howard Hughes Medical Institute, University of California, San Francisco, San Francisco, CA 94158, USA; California Institute for Quantitative Biomedical Research, San Francisco, CA 94158, USA
Functional genomics efforts typically face a tradeoff between the number of perturbations examined and the complexity of measured phenotype. To bridge this gap, we developed Perturb-seq, an approach that pairs droplet-based single-cell RNA sequencing with a robust cell barcoding strategy that encodes the identity of a CRISPR-mediated perturbation in an expressed transcript. It is thus now possible to obtain rich phenotypic profiles of hundreds of perturbations in pooled format with single-cell resolution.
The Perturb-seq approach is particularly helpful for interpreting genetic screens, as it is often unclear why particular hits appear. For example, our pilot study used a traditional genome-wide screen to find 66 genetic stressors whose depletion induced the unfolded protein response, including both well-characterized genes and many unknowns. Leveraging the rich phenotypes afforded by Perturb-seq, we were able to determine how each of these hits activated the UPR, through separate, combined, or heterogeneous induction of the three canonical pathways. Additionally, we have developed vectors that allow combinatorial knockdown or activation of multiple genes. These allow the construction of large-scale genetic interaction maps in which all pairwise phenotypes are measured to identify synthetic interactions that control cancer cell growth. Leveraging improvements in the Perturb-seq approach, we can shed light on why these interactions occur. Together these studies highlight how high-content phenotypes have the potential to transform the way genetics is done.