Metazoan tRNA introns generate stable circular RNAs in vivo

Abstract

We report the discovery of a class of abundant circular noncoding RNAs that are produced during metazoan tRNA splicing. These transcripts, termed tRNA intronic circular (tric)RNAs, are conserved features of animal transcriptomes. Biogenesis of tricRNAs requires anciently conserved tRNA sequence motifs and processing enzymes, and their expression is regulated in an age-dependent and tissue-specific manner. Furthermore, we exploited this biogenesis pathway to develop an in vivo expression system for generating "designer" circular RNAs in human cells. Reporter constructs expressing RNA aptamers such as Spinach and Broccoli can be used to follow the transcription and subcellular localization of tricRNAs in living cells. Owing to the superior stability of circular vs. linear RNA isoforms, this expression system has a wide range of potential applications, from basic research to pharmaceutical science.

Keywords: circular RNA; expression system; noncoding RNA, ncRNA; tRNA processing.

FIGURE 1.

Detection of endogenous tricRNAs in Drosophila melanogaster. (A) RNA-seq reads spanning the circular junction of the intron of tRNA:Y1:28C-RA (tRNA:Tyr:GUA, CR31905). rRNA-depleted pharate adult RNA-seq data were mapped using Bowtie2 (for end-to-end and partially mapped reads) and a modified version of Vicinal (for circular junction-spanning reads). The structure of the pre-tRNA gene is shown under the tracks, with the thinner line representing the intron and the thicker lines representing the exons. The junction-spanning reads (counts ≥100) are shown beneath the gene structure. (B) Distribution of tRNA intron sizes among select eukaryotic genomes. The box plots show the first, second, and third quartiles; the whiskers represent a 1.5 interquartile range. Box thickness is proportional to the square root of the number of tRNA introns in each genome. Outliers are drawn as dots; the Drosophila CR31905 intron is marked by a blue box. The number of introns for each species is as follows: Saccharomyces cerevisiae 59, D. melanogaster 16, Anopheles gambiae 29, Aedes aegypti 32, Apis mellifera 11, Mus musculus 24, Rattus norvegicus 10, Caenorhabditis elegans 32, Tribolium castaneum 21, Homo sapiens 39. The plot was drawn using the boxplot R package. (C) Conservation of predicted secondary structure for tric31905 among Drosophilids. Sequences of the orthologous CR31905 tRNA intron from various species of Drosophila are shown. Note the compensatory base changes that maintain base-pairing within the predicted structure. (D) RT-PCR detection of tric31905 from total larval RNA. Cartoon shows schematic of an intronic tRNA gene with convergent (Con) and divergent (Div) primer pairs. Four pairs of primers (two Div and two Con) were tested. The ladders of PCR products correspond to amplification of concatemers of the cDNA via rolling circle reverse transcription. Experimental results agree well with the expected lengths (in base pairs) of the PCR products. Lane 1: 82 + 113n; lane 2: 75 + 113n; lane 3: 100 + 113n; lane 4: 78 + 113n; n = 0, 1, 2, … etc. (E) Abnormal migration of the 113-nt circular RNA tric31905. Total larval RNA samples were electrophoresed through 10% or 6% TBE-urea gels and tric31905 was detected by Northern blotting using a radiolabeled circular RNA junction oligomer. (F) Total larval RNA samples were treated with or without RNase R and run on a 10% TBE-urea gel. RNA was imaged using SYBR Gold (left panel), whereas tric31905 (right upper panel), U1 and U4 snRNAs (right lower panel) were detected by Northern blotting.

FIGURE 2.

Expression and processing of tricRNAs. (A) Northern blot of tric31905 during fly development. tRNA:Leu_CAA and tRNA:Tyr_GUA were used as loading controls. (E) embryo, (1–4) 1- to 4-d-old larvae, (5–8) 1- to 4-d-old pupae, (M) male, (F) female, (O) ovary, (S2) S2 cells. (B) Cartoon of metazoan tRNA splicing and processing pathway illustrating the bulge-helix-bulge (BHB) splicing motif, tRNA splicing endonuclease (TSEN) cleavage sites (junctions of black and gray lines), 5′-OH and 2′,3′-cyclic phosphate groups. 5′ leader, 3′ trailer, and CCA nucleotides are not shown. In animal cells, RtcB is required for ligation of tRNA exons, but the intron ligase is unknown. (C) Knockdown of CG9987 reduces expression of tric31905 as well as that of mature tRNA:Tyr_GUA. RT-PCR was performed on total RNA samples from larvae or pupae. Error bars represent standard deviations from three biological replicates. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, Student's t-test. (D) Processing of tric31905 into small RNAs. Structure track: the predicted secondary structure of tric31905 in linear format. sRNAs track: diagram of the RNAs observed in small RNA-seq data. The solid arrows represent the most prominent species, whereas the dashed arrows represent the opposite strands. Genome Browser track: UCSC Genome Browser view of the read pileup (vertical scale 0–15,000). Read starts: numbers of reads starting at each position in the Genome Browser track (vertical scale: 0–15,000). Heatmap: the numbers of reads that start at each position for every small RNA-seq data set. Each row of the heatmap is normalized to a range of 0–1. (E) Cartoon showing locations of small RNAs along the secondary structure of tric31905, revealing a processing pattern similar to that of miRNAs.

FIGURE 3.

In vivo expression of stable, heterologous tricRNAs in human cells. (A) tricRNA vectors used for circular RNA expression, showing the tricRNA scaffold. The gray arrows represent the A and B boxes of the intragenic tRNA promoters. External 5S and U6 promoter constructs were also generated. (B) pTRIC31905 (tRNA:Tyr_GUA) and pTRIC31143 (tRNA:Leu_CAA) native intron vectors were transfected into HeLa cells for 24 h and their expression was measured by qRT-PCR. β-actin mRNA was used as normalization standard. U1 snRNA was used as a negative control (Control). tric31905-c and tric31143-c are convergent primer pairs, whereas tric31905-d and tric31143-d are divergent primer pairs used to measure expression of tric31905 and tric31143, respectively. Standard deviations were calculated from three biological replicates. (C) Secondary structures of RNAs expressed from pTRIC-Y and pTRIC-L scaffolds, highlighting the TSEN cleavage sites (arrows), anticodons (blue) and NotI–SacII restriction sites (gray). (D) Vectors for two circular Spinach2 constructs, pTRIC-Y:Sp2 (blue) and pTRIC-L:Sp2 (red), were transfected into HeLa cells and expression was determined using qRT-PCR. A control pTRIC31905 native intron vector was also used (black). The three experiments were analyzed using convergent (Spinach2-c) and divergent (Spinach2-d) primer pairs. β-actin mRNA was used as normalization standard and U1 snRNA was used as a negative PCR control (Control). Standard deviations were calculated from three biological replicates. (E) In-gel detection of Broccoli-tagged RNAs expressed from three different pTRIC-Y constructs bearing tRNA, U6 sRNA, or 5S rRNA promoters. HEK293T cells were transfected and total RNA was run on a 6% TBE-urea gel. The gel was washed to remove the urea, renatured in buffer, and stained with DFHBI-1T to reveal Broccoli (right panel), then stained with SYBR Gold to reveal total RNA (left panel). (RnR) RNase R treatment prior to gel electrophoresis. In vitro transcribed pre-tRNA-Broccoli (5 ng) was used as a positive control (first lane). In the experimental lanes, the upper bands (bracket) correspond to the precursor tRNAs; the lower band (arrow) corresponds to the mature circular Broccoli tricRNA. (F) Capping of the pre-tRNA improves production of circular RNAs. Two different U6 snRNA promoter constructs were tested, one that includes the first 4 nt of human U6 snRNA (U6) and a construct that includes the first 27 nt (U6*). HEK293T cells were transfected for the indicated times prior to harvesting total RNA. Electrophoresis, gel-detection, and band labels as described in E.

FIGURE 4.

Stability and live-cell detection of fluorescent tricRNAs. (A) HEK293T cells were transfected with pTRIC-Y:Broccoli-U6, treated with actinomycin D (ActD) and harvested after the indicated times. Total RNA was run on a 10% TBE-urea gel, renatured, stained with DFHBI-1T for Broccoli (right panel) and then SYBR Gold for total RNA (left panel). (B) Quantification of linear and circular Broccoli RNA bands shown in F; (a.u.) arbitrary units. (C) Broccoli RNA detection by flow cytometry. HEK293T cells were transfected with either a mock or circular Broccoli-expressing plasmid (pTRIC-Y:Broc-U6). mCherry was also expressed from a separate plasmid as a transfection efficiency control (y-axis). In addition, Broccoli-tricRNA expressing cells were also treated with actinomycin D for 2 h before sorting. All cells were stained with DFHBI-1T and analyzed by flow cytometry in green and red channels. (D) Localization of circular Broccoli in living cells. Light microscopy of the same cells as described in C, stained with DFHBI-1T (Broccoli) and Hoechst. Scale bar, 20 µm.