Circular RNAs (circRNAs) are a large class of transcripts in the mammalian genome. Although the translation of circRNAs was reported, additional coding circRNAs and the functions of their translated products remain elusive. Here, we demonstrate that an endogenous circRNA generated from a long noncoding RNA encodes regulatory peptides. Through ribosome nascent-chain complex-bound RNA sequencing (RNC-seq), we discover several peptides potentially encoded by circRNAs. We identify an 87-amino-acid peptide encoded by the circular form of the long intergenic non-protein-coding RNA p53-induced transcript (LINC-PINT) that suppresses glioblastoma cell proliferation in vitro and in vivo. This peptide directly interacts with polymerase associated factor complex (PAF1c) and inhibits the transcriptional elongation of multiple oncogenes. The expression of this peptide and its corresponding circRNA are decreased in glioblastoma compared with the levels in normal tissues. Our results establish the existence of peptides encoded by circRNAs and demonstrate their potential functions in glioblastoma tumorigenesis.

As a well-known phenomenon, total mRNAs poorly correlate to proteins in their abundances as reported. Recent findings calculated with bivariate models suggested even poorer such correlation, whereas focusing on the translating mRNAs (ribosome nascent-chain complex-bound mRNAs, RNC-mRNAs) subset. In this study, we analysed the relative abundances of mRNAs, RNC-mRNAs and proteins on genome-wide scale, comparing human lung cancer A549 and H1299 cells with normal human bronchial epithelial (HBE) cells, respectively. As discovered, a strong correlation between RNC-mRNAs and proteins in their relative abundances could be established through a multivariate linear model by integrating the mRNA length as a key factor. The R² reached 0.94 and 0.97 in A549 versus HBE and H1299 versus HBE comparisons, respectively. This correlation highlighted that the mRNA length significantly contributes to the translational modulation, especially to the translational initiation, favoured by its correlation with the mRNA translation ratio (TR) as observed. We found TR is highly phenotype specific, which was substantiated by both pathway analysis and biased TRs of the splice variants of BDP1 gene, which is a key transcription factor of transfer RNAs. These findings revealed, for the first time, the intrinsic and genome-wide translation modulations at translatomic level in human cells at steady-state, which are tightly correlated to the protein abundance and functionally relevant to cellular phenotypes.

It has been a long debate whether the 98% ‘non-coding’ fraction of human genome can encode functional proteins besides short peptides. With full-length translating mRNA sequencing and ribosome profiling, we found that up to 3330 long non-coding RNAs (lncRNAs) were bound to ribosomes with active translation elongation. With shotgun proteomics, 308 lncRNA-encoded new proteins were detected. A total of 207 unique peptides of these new proteins were verified by multiple reaction monitoring (MRM) and/or parallel reaction monitoring (PRM); and 10 new proteins were verified by immunoblotting. We found that these new proteins deviated from the canonical proteins with various physical and chemical properties, and emerged mostly in primates during evolution. We further deduced the protein functions by the assays of translation efficiency, RNA folding and intracellular localizations. As the new protein UBAP1-AST6 is localized in the nucleoli and is preferentially expressed by lung cancer cell lines, we biologically verified that it has a function associated with cell proliferation. In sum, we experimentally evidenced a hidden human functional proteome encoded by purported lncRNAs, suggesting a resource for annotating new human proteins.