Expanding cancer therapy options through genome-scale identification of synthetic lethal paralogs

Abstract

Synthetic lethal therapies have the potential to expand cancer treatment options. Synthetic lethality is a form of context-dependent gene essentiality in which dual inactivation of a gene pair leads to cell death but single-gene inactivation does not affect viability. When one synthetic lethal gene is inactivated in cancer, targeting the remaining gene results in tumor cell death while normal cells remain unaffected. Therefore, synthetic lethal therapies may show substantially reduced toxicity compared to chemo- or radiotherapy and oncogene-targeted therapies. The successful use of PARP inhibitors to treat BRCA1 or BRCA2-mutant breast, ovarian, and prostate cancers shows that synthetic lethal therapies are a viable approach. Yet PARP-BRCA remains the only synthetic lethal target with a clinically-approved inhibitor to date. There thus is an urgent need to find more targetable synthetic lethal interactions in cancer. Given the Berger lab’s focus on lung cancer, we developed a dual knockout CRISPR method to systematically identify synthetic lethal human gene pairs in lung cells. Previous studies showed that less than 3% of unrelated human gene pairs are synthetic lethal, while duplicated yeast genes showed a 25% synthetic lethal hit rate. We thus chose to identify new synthetic lethal lung cancer drug targets by developing a pooled dual-targeting CRISPR-Cas9 library called paired guide RNAs for paralog genetic interaction mapping (pgPEN). pgPEN targets over 2,000 human paralogs, or duplicated genes. We applied pgPEN to lung and cervical cancer cell lines and found that 12% (n = 122) of paralog pairs exhibited synthetic lethality in at least one context. These synthetic lethal paralogs represent new potential cancer therapeutic targets. We also developed two computational methods to ensure that pgPEN can be applied by other researchers and that targeting paralogs is a viable synthetic lethal therapy approach. The first tool, paired guide mapper (pgMAP), is a user-friendly software pipeline that reproducibly maps sequencing reads from large, dual-targeting CRISPR sequencing datasets. The second method leverages existing gene expression and drug sensitivity datasets to identify 68 cases where loss of one paralog led to significantly decreased cancer cell viability upon treatment with a drug targeting that gene family. These analysis methods enable other groups to find synthetic lethal paralog pairs in any cancer type and support targeting paralogs as a potentially viable therapeutic approach. This work reveals over 100 novel, potentially targetable synthetic lethal interactions in human cancer cells that can be further tested and translated to the clinic. Additionally, the pgPEN CRISPR library and the pgMAP pipeline will enable other researchers to identify targetable synthetic lethal interactions in other cancer types. Synthetic lethal therapy has the potential to provide a relatively low-toxicity treatment approach that can expand cancer therapy options, help address the challenge of acquired drug resistance, and improve patient outcomes.