Abstract 1821: Genome-wide genetic screens define the drug resistance landscape of BRAF mutant colon cancer

Background: Almost 10% of colon cancers harbor the canonical activating V600 BRAF mutation that in melanoma cancers has been shown to dramatically sensitize cells to targeted therapy and prolong survival. In the colon, such tumors are associated with worse survival. Recent clinical trials of combination therapies targeting the EGFR and MAPK pathways in these tumors have demonstrated impressive response rates. However, the clinical benefit has been limited by the rapid emergence of drug resistance. Identifying and targeting the genes complicit in this process and thus re-sensitize resistant cancer cells may be of clinical benefit. The landscape of drug resistance in cancer patients is composed of (a) loss-of-function (LoF), (b) gain-of-function (GoF) and (c) point mutation-mediated perturbations of genes, with often no single mechanism being predominant. Therefore, capturing the full breadth of resistance genes for any drug would require multiple genetic screens to be executed in parallel. Methods: We performed 3 genome-wide genetic screens in parallel in BRAF mutant colon cancer cells treated with a BRAF/MEK/EGFR inhibitor combination, to capture all of the resistance mechanisms described above. To screen for LoF events we used a genome-wide CRISPR/Cas9 sgRNA library. For GoF resistance events, a lentiviral-based insertional mutagenesis vector was used to randomly integrate the SFFV enhancer/promoter sequence throughout the genome. Finally, to saturate the genome of each cell line with point mutations we used N-Ethyl-N-nitrosourea (ENU) to randomly mutagenize every base in the genome. Each genetic screen was analysed separately to identify the relevant genes and pathways that confer drug resistance before aggregating the outputs of all three screens for a more comprehensive view of the drug resistance landscape. Results: The most recurrent and strongest hit from the CRISPR knockout screen was the inhibitor of Src family kinases, CSK. A Gaussian Kernal Convolution analysis for recurrent integrations detected in resistant clones following the insertional mutagenesis screen identified five genes as likely resistance candidates - BRAF, SOS1, MET, FRS2 and KRAS. Finally, Illumina exome sequencing of resistant clones following ENU mutagenesis revealed statistical enrichment for non-synonymous point mutations in three genes in the MAPK pathway, namely MAP2K1, NRAS and KRAS. Of note, loss of CSK was also shown to confer resistance to targeted therapies in other ca