Abstract LB108: Validation of computationally predicted synthetic lethal interactions for KRAS

Abstract Despite dedicated research in precision medicine, identification of therapeutic strategies for oncogenic drivers remains a major challenge as many mutations are difficult to target directly. Synthetic lethality-based target prediction becomes a powerful approach for drug discovery. We previously developed a computational platform, Mining Synthetic Lethals (MiSL), that can output a list of possible synthetic lethal partners of a specific gene when fed with primary human tumor genomic and transcriptomic data. The principal logic of the MiSL algorithm is based on the Boolean implication that the SL partners of a mutation will never be deleted in the presence of the mutation in the primary tumors. High frequency of KRAS mutation in lung and colon cancers makes it an important therapeutic target. While drugs directly targeting KRAS are yet to be fully developed, we applied the principle of synthetic lethality using the MiSL platform to predict potential KRAS specific therapeutic targets. We identified 556 putative SL partners of KRAS mutations combining lung and colon cancers. To efficiently test all predictions, we used customized shRNA and CRISPR lentiviral libraries to validate all the predicted SL interactions for KRAS in colon cancer cell lines HT29 (KRAS WT) and SW620 (KRAS G12V mutant). Eight shRNAs and four sgRNAs were designed for each gene in the shRNA and CRISPR libraries. The shRNAs and sgRNAs expressed at baseline and at the end of each screen were analyzed by barcode sequencing. Synthetic lethal interactions with KRAS were determined by identifying individual sh/sgRNAs whose mean values at the end relative to the start of the screen were significantly reduced in KRAS Mut cell lines compared to WT. We also compared the results of the in vitro to an in vivo screen of the focused libraries in human tumors xenografted into mice using the same cell lines. We compared the top 50 SL candidates from each screen as ranked by MAGeCK (Model-based Genome-wide Analysis of CRISPR/Cas9 KO) and found that only 19 were shared between the in vitro and in vivo screens. Further validation of several top SL candidates coming from both in vitro and in vivo screens was performed by testing individual shRNAs/sgRNAs for growth inhibition in KRAS Mut (SW620) vs. KRAS WT (HT29) cell lines in vitro and in vivo. Individual in vitro CRISPR/Cas9 knock out of two of the candidate SL partners reduced cell viability by 50% and 57% respectively in KRAS mutant cells compared to KRAS WT cells. Furthermore, induced individual KD or KO of four SL partners resulted in inhibition of tumor growth by 84%, 68%, 95% and 60%, respectively. In summary, MiSL has shown to be a powerful platform for discovering novel targets for specified cancer types. This study has identified and validated several genes as promising targets of KRAS dependent colon cancer for further therapeutic investigation. Citation Format: He Gong, Claire Repellin, Puja Patel, Merrill Knapp, Mingze He, Jun Li, Lisa Wu, Xiaohe Liu, Subarna Sinha, Lidia Sambucetti. Validation of computationally predicted synthetic lethal interactions for KRAS [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr LB108.