The landscape of chimeric RNAs in bladder urothelial carcinoma.

BACKGROUND

Gene fusions and products have been identified as oncogenic drivers in many cancers, making them attractive diagnostic markers and therapeutic targets. However, the landscape of fusion transcripts in bladder cancer has not been fully characterized.

METHODS

To identify fusion transcripts with potential therapeutic or diagnostic values, TCGA bladder urothelial carcinoma RNA-sequencing dataset was used. In order to avoid false positives, we applied multiple criteria including filtering out fusions detected in normal samples from GTEx dataset. We validated a subset of candidate fusions with a collection of bladder cancer and adjacent normal samples.

RESULT

We identified 19,547 high confidence fusion genes from 414 bladder cancer samples. After filtering off M/M fusions, fusions in GTEx normal samples, and occurrence frequency <5, we obtained a list of 271 gene fusions, 13 of which were novel and specific to cancer samples. Six of those fusions were validated using cell lines and clinical samples. We discovered that two chimeric RNAs, BCL2L2-PABPN1 and CHFR-GOLGA3, were detected to be expressed significantly higher in bladder cancer samples compared to adjacent normal samples. Impressively, the wild-type of the parental genes were not differentially expressed. Mechanistically, we demonstrated that these two fusions are generated by cis-splicing between adjacent genes. These two fusions were detected mainly in the fraction of cell nucleus, suggesting a potential long noncoding RNA role.

CONCLUSION

Our findings provide a panoramic view of the landscape of chimeric RNAs in bladder cancer. Some frequent chimeric RNAs are generated by intergenic splicing, and represent a new repertoire for potential biomarkers.