Abstract
BACKGROUND/OBJECTIVES: Human endogenous retroviruses (ERVs) are genomic sequences integrated into the human genome from ancestral exogenous retroviruses and are epigenetically silenced under normal conditions. Growing evidence has shown that they can be reactivated in human diseases such as cancers and autoimmune diseases. However, their clinical implications in colon cancer are yet to be explored. METHODS: RNA-seq data were downloaded from RNA Atlas and TCGA for cell lines and tissue samples, respectively. After alignment, ERV expression was quantified against comprehensively compiled ERVs (3220). ERV expression profiles were compared between sequencing protocols, cancer and normal cells, and matched tumor and normal tissue pairs. Unsupervised clustering was used to identify ERV-defined tumor subtypes and their associations with clinical and other molecular features. ERV association with disease-specific survival (DSS) was performed using the Cox regression model. RESULTS: PolyA and total RNA protocols were comparable in ERV expression detection. Cancer cells had significantly increased ERV expression and reactivation. Upregulated ERVs were significantly enriched in viral protein interactions with cytokine and cytokine receptors. ERV expression-defined tumor classes were significantly associated with tumor mutation burden and immuno-phenotypes such as antigen processing and presenting machinery and tumor immune infiltration score. Survival analysis identified 152 ERVs to be independently associated with DSS. CONCLUSIONS: ERV abnormal expression is common in colon cancer. The ERV-defined subtypes are associated with tumor immunity, and some ERVs are independently associated with patient outcomes.