Meta-analysis of breast carcinoma transcriptomes demonstrates correlation between differential splicing of RAC GTPases and tumor aggressiveness

Abstract

Cancer, one of the leading causes of death worldwide can be broadly categorized into two types: Sarcoma and Carcinoma. Carcinoma, a cancer originating in epithelial tissues, makes up most of the most common forms of cancer, such as breast, lung, and prostate. While recent advances in surgical and chemotherapeutic techniques have increased the survival rate for localized cancer, the 5-year prognosis for metastatic cases has not improved nearly as much. To this end, research has been done to identify early indicators of metastatic potential, as well as to elucidate the method of metastasis. Carcinomas can only present an invasive phenotype by initiating Epithelial to Mesenchyme Transition (EMT), a process of cell differentiation that occurs naturally during development and wound healing, which downregulates tight cell junctions and upregulates cell motility. Previous work has suggested that there is a correlation between EMT potential in breast cancers and a change in splicing patterns of the gene RAC1, a signaling GTPase with effects on the cytoskeleton. Collecting multiple RNAseq datasets of breast carcinoma cells, this paper examines whether this differential splicing pattern holds true for other members of the RAC GTPase family. Exons of RAC GTPases had their expression level in each dataset calculated in terms of RPKM, the number of RNAseq reads per million total reads per kilobase length of the exon. Exon RPKM in each cell type was then plotted against that cell type’s expression level of canonical EMT markers. These results suggest that RAC splice isoforms with higher affinity for GTP or lower GTPase efficiency, both of which cause the proteins to spend more time in its active conformation, are more common in metastatic breast tumors. This underscores the importance of changes in cytoskeletal regulation in the progression of carcinomas and the potential for events other than coding mutations to have deep effects on tumor behavior.