Executive Vice President Genentech, California, United States
Pooled perturbation screens with CRISPR/Cas9 are a powerful approach to functionally dissect molecular circuits in physiological and disease contexts. Traditionally, CRISPR screens measure population guide RNA (gRNA) enrichment, necessitating selection and limiting phenotypes studied. Several years ago, we developed Perturb-Seq to combine pooled perturbation screens with single-cell RNA-seq read-outs: each cell is matched to its perturbation and corresponding transcriptional profile. Such pooled, high-content screens allow identification of gene programs impacted by perturbations, can be performed in heterogeneous biological models, and applied in the context of combinatorial (simultaneous) genetic perturbations, both essential for deciphering disease mechanisms, from rare disease, complex disease, and cancer. Here, I will present three methods we have recently developed to enable such studies, and how we applied them in cancer and complex disease. First, I will describe Perturb-CITE-Seq to yield single cell multiomic (RNA and proteins) readouts, and its extension with marker-based enrichment, Perturb-ME, to increase the scale and efficiency of screens. I will show their application to investigate a program of immune cell evasion in patient melanoma cells co-cultured with matched tumor-infiltrating lymphocytes (TILs) and for a genome-wide screen of regulators of MHC Class I expression in melanoma cells. Second, I will describe a new approach we developed to generate stepwise, genome-edited, human models of melanoma, involving combinatorial perturbation sof up to six pathways, and how we coupled them to both scRNA-seq and histology in vivo to reveal the effects of mutational landscape on both tumor cell intrinsic mechanisms and microenvironment properties. Finally, I will present in vivo Perturb-Seq, for pooled CRISPR screens in animal models or organoids, and its application to study genes important for autism spectrum disorders. I will close with the prospects for computational inference of the effect of combinatorial perturbation, and the experimental and algorithmic approaches that enable it.
