Nature Short Video Download Full Screen ##TOP##

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Some circular RNAs (circRNAs) were found to be translated through IRES-driven mechanism, however the scope and functions of circRNA translation are unclear because endogenous IRESs are rare. To determine the prevalence and mechanism of circRNA translation, we develop a cell-based system to screen random sequences and identify 97 overrepresented hexamers that drive cap-independent circRNA translation. These IRES-like short elements are significantly enriched in endogenous circRNAs and sufficient to drive circRNA translation. We further identify multiple trans-acting factors that bind these IRES-like elements to initiate translation. Using mass-spectrometry data, hundreds of circRNA-coded peptides are identified, most of which have low abundance due to rapid degradation. As judged by mass-spectrometry, 50% of translatable endogenous circRNAs undergo rolling circle translation, several of which are experimentally validated. Consistently, mutations of the IRES-like element in one circRNA reduce its translation. Collectively, our findings suggest a pervasive translation of circRNAs, providing profound implications in translation control.

Nature Short Video Download Full Screen

Download Zip 🔥 https://tinurll.com/2y2M54 🔥

To study the scope of circRNA translation, we developed a cell-based reporter system to screen a random library of short sequences that drive circRNA translation. Through a near-saturated screen and subsequent bioinformatics analyses, we identified 97 IRES-like hexamers that are clustered into 11 groups with AU-rich consensus motifs. The IRES-like elements are significantly enriched in human circRNAs compared to all linear RNAs, suggesting that they are positively selected in circRNAs. Since these IRES-like hexamers account for ~2% of all hexamers (97/4096), any sequences longer than 50-nt should contain such an element by chance, implying that most circRNAs in human cells can potentially be translated through the IRES-like short elements. Consistently, circRNAs containing only the coding sequences can indeed be translated in a rolling circle fashion, presumably from internal IRES-like short elements in the coding region. We further identified hundreds of circRNA-coded proteins with mass spectrometry datasets, and explored the potential roles of the circRNA-coded proteins and the mechanism of their translation. Collectively, our data indicate that short IRES-like elements can drive extensive circRNA translation, which may represent a general function of circRNAs.

Surprisingly, we found that three of the four short poly-N sequences designed as negative controls for IRESs could also promote GFP translation from circRNAs (Supplementary Fig. 1a). This observation indicates that certain short elements other than known IRESs are sufficient to initiate circRNA translation. To systematically identify additional sequences that drive circRNA translation, we adopted an unbiased screen approach originally developed to identify splicing regulatory cis-elements23,24,25. Briefly, a library of random 10-nt sequences was inserted before the start codon of circRNA-coded GFP26, which was transfected into 293T cells to generate circRNAs that can be translated into intact GFP (Methods and Supplementary Data 1). The green cells with active circRNA translation can be recovered with fluorescence-activated cell sorting (FACS), and we used RT-PCR with junction-specific primers to amplify the inserted sequences from circRNAs. The inserted decamers can be subsequently sequenced to identify the IRES-like elements that drive circRNA translation (Fig. 1a).

We further compared the average frequency of the IRES-like hexamers vs. the control set of hexamers in different types of RNAs. In the linear mRNAs (all annotated mRNAs from RefSeq), the average frequencies of the IRES-like hexamers were similar to random hexamers or the hexamers depleted in green cells (Fig. 2a, left panel). Surprisingly, the average frequencies of IRES-like hexamers were significantly higher in all tested circRNA datasets4,7,32,33 compared to the control hexamer sets (Fig. 2a), indicating that endogenous circRNAs are enriched with short IRES-like elements. Since these IRES-like hexamers were independently identified from unbiased screen of random sequences, such enrichment strongly suggests that circRNAs may be positively selected for their ability to be translated.

PABPC1 is an abundant protein that binds poly-A or AU-rich sequences41. We further examined the role of PABPC1 in regulating circRNA using reporters contain a short IRES-like element (poly-A) (Fig. 3e, left panel). We found that over-expression of PABPC1 and its paralogue PABPC4 can indeed promote translation of circRNA-coded GFP, whereas the circRNA translation was slightly affected by another polypyrimidine binding protein PTBP1 that was previously reported to enhance IRES activity. Interestingly, for unknown reason, the co-expression of U2AF2 seemed to inhibit circRNA translation. As a control, over-expression of these proteins showed no effect on the circRNA reporters containing the (G)10 sequence (Fig. 3e, right panel). Furthermore, we used RNA-IP to examine the binding between PABPC1 with circRNAs. As expected, the PABPC1 indeed binds more tightly to the circRNAs containing IRES-like short elements compared to the control sequences (Supplementary Fig. 3c), supporting its role to promote circRNA translation by directly recruited to circRNAs through IRES-like elements. Taking together, our results showed that certain RBPs are capable to recognize these IRES-like elements to promote cap-independent translation of circRNAs, exemplifying a new model for circRNA translation initiation independent of canonical IRESs.

Detecting translation products of endogenous circRNAs is technically difficult because they differ from the canonical products only at the back-splice junctions. Generating new antibodies against the short fragments coded by the back-splice junctions is time consuming and sometimes unreliable, and thus we labeled the candidate circRNA products with an epitope tag. We selected three circRNAs (circPSAP, circPFAS, and circABHD12) with high-quality MS evidence (Fig. 5a and Supplementary Fig. 6a), and constructed back-splicing reporters to ectopically express these circRNAs in two different cell types26. An in-frame V5 epitope tag was inserted into the rcORFs to facilitate the detection of translation products (Fig. 5b). We found multiple protein bands in 293T and SH-SY5Y cells transfected with all circRNAs tested, suggesting that these circRNAs undergo robust rolling circle translation to produce protein concatemers (Fig. 5c and Supplementary Fig. 6b). Interestingly, the cells transfected with circPSAP produced a faint band of protein concatemer and a much stronger band corresponding to the product of a single cycle of circRNA translation, suggesting a low translation processivity for this circRNA.

Since the translation of rcORFs can generate protein concatemers with repeat sequences, they are likely to induce protein mis-folding and aggregation that often leads to rapid protein degradation. To measure the stability of these proteins translated from rcORF, we treated the cells with short exposure of MG132 or chloroquine, and found that the translation products of two circRNAs (circPSAP and circPFAS) were increased by MG132 treatment but not by chloroquine while the control protein GFP was not affected (Fig. 5d). These findings suggested that some rolling circle translation products were rapidly degraded in cells through proteasome pathway, which are consistent with the observation that the circRNA-coded proteins are generally low abundant inside cells (Supplementary Fig. 5). Intriguingly, the translation product of circABHD12 seems to be relatively stable despite containing repetitive sequences.

We next examined whether the rolling circle translation of endogenous circRNAs is indeed driven by the newly identified IRES-like short elements. The circPFAS contains two IRES-like hexamers, and we made individual mutations on both hexamers to examine their effects on circPFAS translation (Fig. 5e, left). The mutation on one IRES-like hexamer (AAGAAG) dramatically reduced the translation of circPFAS, whereas the mutation on the other element (AATTCA) had no effect, suggesting that the AAGAAG hexamer is the major translation initiation element in circPFAS (Fig. 5e and Supplementary Fig. 6c). Interestingly, a trace amount of translation product was observed even when mutating both IRES-like elements, suggesting that other unknown cis-elements may also help to promote circPFAS translation. This observation is not a surprise since we used a strong cutoff (enrichment z score >7) in the identification of the 97 IRES-like hexamers, and thus may miss some elements with weak activity. We further mutated the downstream AUG start codons to confirm that the second AUG was the major translation start site of circPFAS, as its mutation significantly reduced the translation efficiency (Fig. 5e, right). Interestingly, the circRNA translation is reduced but not completely abolished with the AUG mutation, and additional mutations of nearby in-frame CUG site can further reduce translation (Fig. 5e). This result suggests that the cap-independent translation driven by IRES-like hexamers can also use non-canonical start codon (e.g., CUG), which is consistent with previous reports54.

Several recent studies indicated that circRNAs can function as template to direct protein synthesis; however, the nature of circRNA translation is still under debate because other studies failed to detect the significant association of circRNA with polysomes4,55,56. In addition, the mechanism of circRNA translation is not clear. While the cap-independent translation of circRNA generally requires a viral or endogenous IRES, we have demonstrated that a short element containing m6A modification is sufficient to drive circRNA translation13. Here we surprisingly found that the requirement of IRESs can be easily fulfilled, and many short sequences (~2% of all hexamers) are capable of driving cap-independent translation of circRNAs. This finding is also consistent with an earlier report that in vitro synthesized circRNAs with mutated Kozak sequences can still be translated22. As a result, thousands of cytoplasmic circRNAs can be potentially translated, among which hundreds of circRNA-coded peptides were supported by mass spectrometry evidences. We further identified many RNA binding proteins that specifically recognize these IRES-like short elements and function as trans-factor to promote cap-independent translation, providing a new paradigm on the mechanism of circRNA translation. ff782bc1db