An Interrogation of ORF Versus CRISPRa Pooled-Screening Technologies Used to Define Cancer Drug-Resistance Landscapes.

Abstract

Resistance to cancer therapies is an ever-present problem, and preemptively understanding the underlying genetic causes will improve patient care, help predict clinical response rates, and elucidate new drug targets. Within the last several years, studying the drug-resistance landscape of a cancer type has been made easier with two gain-of-function pooled genetic-screening systems – open reading frames (ORFs) and CRISPR activation (CRISPRa). The ORF and CRISPRa screening systems produce the same overexpression phenotypes, but with ORF technology the gene of interest is exogenously expressed as a cDNA. Although directed overexpression of single genes provides valuable information, the genes are expressed at non-physiological levels, which can cause artifacts and non-meaningful biological insights. CRISPRa technology has the advantage of activating the endogenous gene transcript, and its splice isoforms, while expanding the capacity to conduct large-scale genetic screens. Despite these differences between the two technologies, both have proven successful in unfolding some of the unknown mechanisms of drug-resistance in cancer. However, recent works have shown that the two gain-of-function mechanisms provide minimally concordant gene hit lists when screened in the same NRAS-mutant melanoma cell line. This study examines why the two technologies yield some discrepancies by generating follow-up, pooled ORF and CRISPRa libraries. These libraries contain the same set of genes hits from the primary screens, and secondary screens are performed in the same cell line. The data show that genes that were identified with only one of the methodologies in the primary screens, are confirmed more strongly with the same methodology in the secondary screens. Both ORF and CRISPRa validate the same percentage of gene hits, but unique sets of ‘ORF-only’ and ‘CRISPRa-only’ genes are defined. Follow-up assays show that certain genes do not emerge in an ORF screen because the ORF itself is lethal or is not expressed post-transduction. Other genes do not score with CRISPRa because the overall significance of the hit is decreased by averaging the magnitudes of all the sgRNAs for the gene. Altogether, the two activation systems produce distinctive results due to ineffective ORF and CRISPRa constructs, library design, and unaccountable biological factors specific to the cancer model of choice.