Identification of true dependencies in cancer is pivotal to the elucidation of novel therapeutic strategies to increase prospects for cancer patients. Unfortunately, in vivo identification of genetic dependencies has long relied on expensive and time-consuming breeding of genetically engineered animal models. Recently, in vitro CRISPR/Cas9 screens provided a new method for rapid and genome-wide identification of genetic dependencies. Nevertheless, genetic dependencies would ideally be identified using in vivo cancer models initiated by clinically relevant oncogenic driver or tumor suppressor insults. Here, we report a new methodology called CRISPR/Cas9-mediated Negative Selection Identification of genetic Dependencies (CRISPR-NSID) that allows in vivo elucidation of cancer cell vulnerabilities in genetic cancer models. The methodology hinges on the fact that for a genetic dependency there is an incapability for recovering tumors carrying biallelic frameshift mutations in this gene. We demonstrate how integrating experimentally determined, or in silico predicted, probabilities of frameshift editing for any given gRNA can be employed to ascertain negative selection pressure on inactivation of a genetic dependency during tumorigenesis. As a proof-of-principle, we use CRISPR-NSID to identify ezh2 and creb3l1 as genetic dependencies in desmoid tumors (desmoid-type fibromatosis) occurring in a Xenopus tropicalis cancer model driven by apc mutations. Bridging CRISPR-NSID to a clinically unmet need, we further demonstrate the promise for EZH2 inhibition as a new therapeutic strategy for desmoid tumors. This study establishes a new methodology for rapid identification of genetic dependencies in monoclonal disorders with wide adaptability to other model systems and organisms.

CRISPR-NSID: an in vivo CRISPR/Cas9 negative selection screen reveals EZH2 as a druggable dependency factor in a genetic desmoid tumor model