Structured elements drive circular RNA translation and expand the human proteome, part4

The human genome encodes tens of thousands circular RNAs (circRNAs) whose levels correlate with many disease states. While studies have focused on the non-coding functions of circRNAs, emerging evidence suggests that a handful of circRNAs encode proteins. Translation canonically starts by recognition of mRNA 5’cap and scanning to the start codon; how circRNA translation initiates remains unclear. Here, we developed a high-throughput screen to systematically identify and quantify RNA sequences that can direct circRNA translation. We identify and validate over 17,000 circRNA internal ribosome entry sites (IRES) and reveal that 18S rRNA complementarity and a structured RNA element on the IRES are important for facilitating circRNA cap-independent translation. With genomic and peptidomic analyses of the IRES, we identified nearly 1,000 putative endogenous protein-coding circRNAs and hundreds of translational units encoded by these circRNAs. We further characterized circFGFR1p, a protein encoded by circFGFR1, functions as a negative regulator of FGFR1 to suppress cell growth under stress conditions. The circRNA proteome may be important links among circRNA, biological control, and disease.