Rolling Circle Translation of Circular RNA in Living Human Cells

Abstract

We recently reported that circular RNA is efficiently translated by a rolling circle amplification (RCA) mechanism in a cell-free Escherichia coli translation system. Recent studies have shown that circular RNAs composed of exonic sequences are abundant in human cells. However, whether these circular RNAs can be translated into proteins within cells remains unclear. In this study, we prepared circular RNAs with an infinite open reading frame and tested their translation in eukaryotic systems. Circular RNAs were translated into long proteins in rabbit reticulocyte lysate in the absence of any particular element for internal ribosome entry, a poly-A tail, or a cap structure. The translation systems in eukaryote can accept much simpler RNA as a template for protein synthesis by cyclisation. Here, we demonstrated that the circular RNA is efficiently translated in living human cells to produce abundant protein product by RCA mechanism. These findings suggest that translation of exonic circular RNAs present in human cells is more probable than previously thought.

Figure 1. Rolling circle amplification of DNA (A) or peptide (B) on a small circular template.

Figure 2. Synthesis of circular RNAs.

(A) A scheme for the synthesis of circular RNAs used in this study. Transcribed linear RNAs were annealed to its complementary DNA oligomer and then ligated using T4 RNA ligase 2 to produce the circular RNA. (B,C) Verification of their circularity of the RNAs. The RNAs were incubated with RNase R and the reactions were analysed by denaturing PAGE. The gels were visualised by SYBR Green II staining.

Figure 3. Design of circular RNAs and their translation in rabbit reticulocyte lysate (RRL).

(A) Schematic representation of the circular RNAs used in this study. The grey dashed line represents the point of ligation of the linear RNA to form a circle. (B) Nucleotide sequence of 4× FLAG circular RNA. The first nucleotide, G, is linked with the last nucleotide, U, in this context. Sequences of the other RNAs are shown in Supplementary Table S1. (C) Western blot analysis of the translation reaction in RRL. The three circular RNAs and their linear precursors (1 μg) were incubated at 30 °C overnight in a 25 μL reaction.

Figure 4. Translation of a circular 8× FLAG RNA in living HeLa cells.

HeLa cells were transfected with 8× FLAG linear or circular RNA by lipofection. (A) Western blot analysis of the lysate using an anti-FLAG antibody. β-Actin was detected as a loading control. (B) Microscopic imaging of the translation product in HeLa cells. The subcellular localisation of the FLAG-containing product was analysed by immunofluorescence staining using an anti-FLAG antibody and anti-mouse IgG antibody labelled with Alexa Fluor 488 (FLAG, green). Cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; blue) to visualise nuclei.

Figure 5. Bicistronic reporter assay for IRES activity in the repeating FLAG-coding sequence.

(A) Schematic representation of the bicistronic plasmid constructs. pβGal–CAT contains no insert between the two cistrons, which encoded chloramphenicol acetyltransferase (CAT) and β-galactosidase (β-gal). pβGal–4× FLAG–CAT contained a repeated (four) FLAG sequence and pβGal–IRES–CAT contained an IRES sequence derived from EMCV in the region between the two cistrons. The plasmid, pIRES, was used as a negative control. (B,C) Expression levels of β-gal (B) and CAT (C) in the cell lysate after the transfection of these plasmids into HeLa cells. The amounts of CAT and β-gal were determined by enzyme-linked immunoabsorbent assay. Results obtained from mock-transfected control are also shown. The plotted data are the means ± standard deviation of three independent experiments. (D) Relative IRES activities were calculated from the data shown in (B,C). The ratio of CAT/β-gal expression for pβGal–CAT was set at 1.0.

Figure 6. Synthesis of circular RNAs that encode human growth factors and their translation in rabbit reticulocyte lysate.

(A) Schematic representation of the circular RNAs used in this study. The Kozak consensus sequence, the FLAG coding sequence and growth factor-coding sequences are indicated. The nucleotide sequences of the Kozak sequence and its mutant are shown. (B–D) Western blot analysis of the translation reaction in rabbit reticulocyte lysate. Circular RNAs and their linear precursors (2.4 μg in B, C and 1.2 μg in D) were incubated at 30 °C for 6 h in a 20 μL reaction. Anti-FLAG (B, left panel of C and D) or anti-EGF antibody (right panel of C) were used to detect the product. RNAs tested are indicated in the figure. Letters (L,C) in the figure denote linear or circular RNA, respectively.