Abstract
Activating RAS mutations are among the most common pathogenetic events in a broad spectrum of hematologic malignancies and epithelial tumors. However, oncogenic RAS has thus far not proven to be a tractable target for therapeutic intervention. An alternative to direct targeting of known oncogenes is to perform “synthetic lethality” screens to identify genes that are selectively required for cell viability in the context of specific cancer-causing mutations. Using this approach, we have discovered a synthetic lethal interaction between mutant KRAS, the most frequently mutated oncogene in human cancer, and inactivation of the gene encoding the STK33 serine/threonine protein kinase. To identify genes that are essential for cell viability in the context of mutant KRAS, we performed high-throughput loss-of-function RNA interference (RNAi) screens in eight human cancer cell lines (mutant KRAS, n=4; wildtype KRAS, n=4), representing seven different tumor types (acute myeloid leukemia, multiple myeloma, colon cancer, breast cancer, ovarian cancer, prostate cancer, glioblastoma), as well as normal human fibroblasts and mammary epithelial cells. We screened each cell line with a subset of the short hairpin RNA (shRNA) library developed by the RNAi Consortium (http://www.broad.mit.edu/genome_bio/trc/rnai.html) that consists of 5,024 individual shRNA constructs targeting 1,011 human genes, including the majority of known and putative protein kinase and phosphatase genes and a selection of known cancer-related genes. In these cell lines, suppression of STK33 preferentially inhibited the viability and proliferation of cells that were dependent on mutant KRAS. The differential requirement for STK33 based on oncogenic KRAS dependency was confirmed in 16 additional cell lines using in vitro transformation assays and human tumor xenograft models. Biochemical analyses support the hypothesis that STK33 promotes cell growth and survival in a kinase activity-dependent manner by regulating the activity of S6K1 as well as BAD-induced apoptosis selectively in mutant KRAS-dependent cells. Notably, molecular genetic characterization of cancer cell lines and analysis of patient-derived genomic data sets indicate that STK33 is not frequently mutated or overexpressed in human tumors. These observations identify STK33 as a potential target for the treatment of mutant KRAS-driven cancers that may have a broad therapeutic index in normal versus malignant cells, and illustrate the potential of RNAi for discovering critical functional dependencies created by oncogenic mutations that cannot be identified using other genomic technologies.
Statistical methods for analysis of high-throughput RNA interference screens