Construction of ceRNA networks in endocrine therapy resistance model systems

Abstract

BackgroundEndocrine therapy resistance (ETR) in breast cancer is achieved via multiple pathways including a decrease in ER, dysregulation of cell cycle genes, and/or mutations in ER/co-activators/co-repressors. We have reported earlier that high expression of Jumonji domaining containing protein 6 (JMJD6) induced ETR by depleting ER expression. In this study, 3 cellular models representing distinct ETR pathways; Tamoxifen resistant (TAMR), Long-term Estrogen deprived (LTEDI), JMJD6 overexpressing (JOE) cells, and parental MCF7 were subjected to RNA-sequencing, CNC, and ceRNA network analysis. We hypothesised that post-comparison RNA regulations that are common to all cell lines, will reveal actionable markers and targets. These will be shared by all patients with ET-resistant disease, independent of the initiating event. Results170 differentially expressed genes were found, of these, 73 maintained the same directionality in expression (ETR cassette genes). These genes segregated TCGA ER+ tumors into two groups, one intermixing with ER-tumors. Pathway-based curation of ETR genes identified 21 genes (7 up- and 14 down-regulated) that participated in multiple cancer hallmark pathways. Genes upregulated in ETR cells were less expressed in ER+ tumors at diagnosis when compared to normal breast samples but their higher expression indicated adverse survival outcomes. Next, these genes were used for CNC and ceRNA network construction and a triad FLT4:MIR503HG:miR-497/195/424 was discovered. The expression levels of miRNAs were predicted via network analysis and quantitative RT-PCR was used to validate the down regulation of miR-497/195/424 and upregulation of their targets, FLT4 and MIR503HG in ETR cells. ConclusionsWe show that total RNA-seq data can be successfully used to predict actionable miRNAs that achieve drug resistance. Re-expression of ETR genes such as FLT4 in tumor cells, that are less expressed at diagnosis, may be indicative of ETR onset. Finally, ETR may arise due to suppression of miR-424/497/195 leading to higher expression of FLT4 and MIR503HG. We posit that FLT4 may be a suitable target and RT-PCR analysis of this RNA triad could be developed as a detection strategy for ETR in ER+ breast cancer.