Abstract
The class of circular RNA (circRNA) is characterized by head-to-tail bonds between exons formed by backsplicing. Here, we provide a resource of circRNA expression in a comprehensive panel of 60 lung cancer and non-transformed cell lines (FL3C dataset). RNA sequencing after depletion of ribosomal RNA quantified the expression of circRNA and linear RNA. We detected 148,811 circular RNAs quantified by 2.8 million backsplicing reads originating from 12,251 genes. The number of identified circRNAs was markedly higher using rRNA depletion compared to public polyA-enriched RNA-seq datasets. CircRNAs almost never started in the first exon nor ended in the last exon and started more frequently in earlier exons. Most circRNAs showed high cell line specificity and correlated positively with their linear RNA counterpart. Known cancer genes produced more circRNAs than non-cancer genes. Subsets of circRNAs correlated with cell proliferation, histological subtype or genotype. CircTNFRSF21 was translated crossing the backsplice site in two different reading frames. Overexpression of circPVT1, circERBB2, circHIPK3, circCCNB1, circSMAD2, circTNFRSF21 and circKIF5B significantly increased colony formation. In conclusion, our data provide a comprehensive map of circRNA expression in lung cancer cells and global patterns of circRNA production as a useful resource for future research into lung cancer circRNAs.