Analysis of Cross-Association between mRNA Expression and RNAi Efficacy for Predictive Target Discovery in Colon Cancers

SIMPLE SUMMARY: This study focused on finding genes for which mRNA expression was able to predict the anticancer efficacy of its RNAi treatment. Predictive target discovery is of critical importance in developing biomarker-based strategies of precision medicine. We demonstrated this carrying out cross-association analysis on collateral mRNA expression and RNAi treatment data of ~12,000 genes on a colon cell line panel. The analysis revealed several genes with significant association between mRNA expression level and the inhibitory efficacy of its RNAi treatment. The experimental validation confirm that this simple approach has general applications for studying gene association between omics data from diverse cancer lineages. ABSTRACT: The availability of large-scale, collateral mRNA expression and RNAi data from diverse cancer cell types provides useful resources for the discovery of anticancer targets for which inhibitory efficacy can be predicted from gene expression. Here, we calculated bidirectional cross-association scores (predictivity and descriptivity) for each of approximately 18,000 genes identified from mRNA and RNAi (i.e., shRNA and sgRNA) data from colon cancer cell lines. The predictivity score measures the difference in RNAi efficacy between cell lines with high vs. low expression of the target gene, while the descriptivity score measures the differential mRNA expression between groups of cell lines exhibiting high vs. low RNAi efficacy. The mRNA expression of 90 and 74 genes showed significant (p < 0.01) cross-association scores with the shRNA and sgRNA data, respectively. The genes were found to be from diverse molecular classes and have different functions. Cross-association scores for the mRNA expression of six genes (CHAF1B, HNF1B, HTATSF1, IRS2, POLR2B and SATB2) with both shRNA and sgRNA efficacy were significant. These genes were interconnected in cancer-related transcriptional networks. Additional experimental validation confirmed that siHNF1B efficacy is correlated with HNF1B mRNA expression levels in diverse colon cancer cell lines. Furthermore, KIF26A and ZIC2 gene expression, with which shRNA efficacy displayed significant scores, were found to correlate with the survival rate from colon cancer patient data. This study demonstrates that bidirectional predictivity and descriptivity calculations between mRNA and RNAi data serve as useful resources for the discovery of predictive anticancer targets.