Predicting resistance mechanisms in pancreatic cancer organoids under KRAS G12D inhibition via pooled CRISPR-Cas9 druggable library screen

Background: Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer mortality, with a median 5-year survival of 13%. Acquired resistance to single agent targeted inhibitors plays a critical role in progression, however there is a lack of predictive tools on resistance that incorporate functional genomics. Here, we evaluate the resistance mechanisms with early treatment of MRTX1133, a small non-covalent inhibitor of KRAS G12D , using a pooled CRISPR-Cas9 lentivirus screen in patient derived cancer organoids (PCO). Methods: PCOs were collected and transduced at a 1:5 cell to lentivirus ratio using a lentiviral-based CRISPR-Cas9 library with 2,292 genes targets from the ‘druggable genome' (Milipore-Sigma). Transduced PCO's were expanded in Cultrex matrix and underwent puromycin selection for 6 days. A baseline group was collected post selection to control for basal expression over the course of media treated control. In parallel, PCOs were treated with MRTX1133 (30nM) or control with collection at 6 days post-treatment. Digital PCR was used to normalize lentiviral transduction efficiency to background. PCR libraries were prepared against targets and DNA sequencing was done to assess resistance mechanisms via the Model-based Analysis of the Genome-wide CRISPR/Cas9 Knockout (MAGeCK) analysis. Results from MAGeCK were also used for HALLMARK gene set analysis to evaluate pathway disruption specific to MRTX1133 treatment. Results: Digital PCR showed optimal lentiviral copy number for baseline (0.863 ± 0.025) post puromycin selection. Pearson correlation plot showed increased variance between control groups when treatment was extended from a 6-day collection (PC1 16%, PC2 15%, PC total 31%) to a 9-day collection (PC1 18%, PC2 18%, PC total 36%). MAGeCK analysis revealed a list of 130 significant gene targets from the druggable screen (p<0.05). Top targets from basal PDAC expression included epithelial-mesenchymal transition pathways with knockout of interleukin-6 (p<0.001) and the E2F transcription family pathway with knockout of CDKN1B (p<0.05). HALLMARK analysis revealed 168 expressed pathways expressed in the control against PCO's treated with MRTX1133 following a 0.05 false discovery rate threshold. These pathways include KRAS signaling (p<0.05), adipogenesis (p<0.05), and E2F targets (p<0.05). Individual gene knockouts observed consistent targets included KCNQT (p<0.05) in KRAS signaling, UQCRC1 (p<0.01) in adipogenesis, and CDKN1B (p< 0.01) in E2F targeting. Conclusions: By scaling a druggable lentiviral based CRISPR-Cas9 screen, we show potential signaling pathways that aid in resistance to tool compound MRTX in PDAC organoids. Thus, addressing the unmet need of predictive resistance modeling in PDAC organoids. Further work includes selective knockout of key targets to confirm synthetic lethality with MRTX1133.