The mutation status of cancer driver genes may correlate with different degrees of malignancy of cancers. The doubling time and multidrug resistance are 2 phenotypes that reflect the degree of malignancy of cancer cells. Because most of cancer driver genes are hard to target, identification of their synthetic lethal partners might be a viable approach to treatment of the cancers with the relevant mutations.
The genome-wide screening for synthetic lethal partners is costly and labor intensive. Thus, a computational approach facilitating identification of candidate genes for a focus synthetic lethal RNAi screening will accelerate novel anticancer drug discovery.
We used several publicly available cancer cell lines and tumor tissue genomic data in this study.
We compared the doubling time and multidrug resistance between the NCI-60 cell lines with mutations in some cancer driver genes and those without the mutations. We identified some candidate synthetic lethal genes to the cancer driver genes APC, KRAS, BRAF, PIK3CA, and TP53 by comparison of their gene phenotype values in cancer cell lines with the relevant mutations and wild-type background. Further, we experimentally validated some of the synthetic lethal relationships we predicted.
We reported that mutations in some cancer driver genes mutations in some cancer driver genes such as APC, KRAS, or PIK3CA might correlate with cancer proliferation or drug resistance. We identified 40, 21, 5, 43, and 18 potential synthetic lethal genes to APC, KRAS, BRAF, PIK3CA, and TP53, respectively. We found that some of the potential synthetic lethal genes show significantly higher expression in the cancers with mutations of their synthetic lethal partners and the wild-type counterparts. Further, our experiments confirmed several synthetic lethal relationships that are novel findings by our methods.
We experimentally validated a part of the synthetic lethal relationships we predicted. We plan to perform further experiments to validate the other synthetic lethal relationships predicted by this study.
Our computational methods achieve to identify candidate synthetic lethal partners to cancer driver genes for further experimental screening with multiple lines of evidences, and therefore contribute to development of anticancer drugs.