Discovery of circular transcripts of the human BCL2-like 12 (BCL2L12) apoptosis-related gene, using targeted nanopore sequencing, provides new insights into circular RNA biology

Abstract

Circular RNAs (circRNAs) constitute an RNA type formed by back-splicing. BCL2-like 12 (BCL2L12) is an apoptosis-related gene comprising 7 exons. In this study, we used targeted nanopore sequencing to identify circular BCL2L12 transcripts in human colorectal cancer cells and investigated the effect of circRNA silencing on mRNA expression of the parental gene. In brief, nanopore sequencing following nested PCR amplification of cDNAs of BCL2L12 circRNAs from 7 colorectal cancer cell lines unraveled 46 BCL2L12 circRNAs, most of which described for the first time. Interestingly, 40 novel circRNAs are likely to form via back-splicing between non-canonical back-splice sites residing in highly similar regions of the primary transcripts. All back-splice junctions were validated using next-generation sequencing (NGS) after circRNA enrichment. Surprisingly, 2 novel circRNAs also comprised a poly(A) tract after BCL2L12 exon 7; this poly(A) tract was back-spliced to exon 1, in both cases. Furthermore, the selective silencing of a BCL2L12 circRNA resulted in a subsequent decrease of BCL2L12 mRNA levels in HCT 116 cells, thus providing evidence of parental gene expression regulation by circRNAs. In conclusion, our study led to the discovery of many circular transcripts from a single human gene and provided new insights into circRNA biogenesis and mode of action.

Keywords: Alternative back-splicing; Alternative splicing; CircRNAs; Colorectal cancer; Gene expression regulation; Third-generation (long-read) sequencing.

Fig. 1

Annotated reads representing the back-splice junctions (BSJs), obtained from bioinformatics analysis of the publicly available RNA-seq datasets previously used to deduce full-length circRNA sequences currently deposited in public circRNA repositories. Non-colored text indicates a mismatch between the read and the reference sequence

Fig. 2

Representation of BCL2L12 circRNAs and their nanopore sequencing reads in the 7 colorectal cancer (CRC) cell lines, using the Integrative Genomics Viewer (IGV). (A) The alignment of BCL2L12 circRNAs against chromosome 19 (chr19) of the Homo sapiens genome assembly GRCh38.p14 (hg38) was done using Minimap2. The circRNAs cumulatively (iii) cover a much wider region of the primary transcript of this gene (i), compared to all mRNAs together (ii). The acceptor back-splice site of each circRNA constitutes the starting position of each aligned sequence (iv). Distinct colors of aligned sequences indicate subgroups; cyan: single-exon circRNAs; purple: circRNAs comprising one or two novel cryptic exons; jazzberry jam and red: circRNAs with one or both (respectively) back-spliced exons being already known and of full length. (B) This Sashimi plot demonstrates the plurality of splice junctions formed between usual or cryptic splice sites, including those of 4 novel cryptic exons harbored by BCL2L12 introns. Only the most frequent splice junctions (covered by > 50 reads) are shown. The vertical axis showing the coverage is drawn in log scale. (C) The coverage of the BCL2L12 genomic region by the nanopore sequencing reads of circRNAs differed among the 7 CRC cell lines. Intronic regions of BCL2L12 are much less represented than all known exons of this gene. The numbers in parentheses indicate the frequency range

Fig. 3

Agarose gel electrophoresis of the back-splice junction (BSJ)-specific amplicons, unique for the vast majority of the identified circRNAs. The cDNA pools used as PCR templates were produced from total RNA extracts of CRC cell lines, without (A) or with (B) prior RNase R treatment

Fig. 4

The formation of the back-splice junction (BSJ) between non-canonical back-splice sites in the identified BCL2L12 circRNAs. (A) Two non-canonical back-splice sites residing in highly similar or even identical regions of a primary transcript, either in exons (i) or introns (ii), are joined together to form the BSJ of the circRNA. However, a different biogenesis pathway may lead to the formation of circRNA(s) with poly(A) tracts resulting from the poly(A) tail added during the maturation of the primary transcript (iii). (B) Only one of the two short identical sequences is present in the circRNA sequence and comprises the BSJ, while the other one is spliced out. (C) The BCL2L12 circRNAs with poly(A) tracts are composed exclusively of the main known exons of this gene (all or most of them)

Fig. 5

Sanger sequencing chromatograms of a common region of circ-BCL2L12-9 and circ-BCL2L12-55, including the poly(A) tract and the common back-splice junction (BSJ) of these two circRNAs. The green shade-colored arrows indicate the sense primers, whereas the red shade-colored arrows indicate the antisense primers

Fig. 6

Normalized expression of circ-BCL2L12-92 and BCL2L12 mRNA in HCT 116 cells transfected with a siRNA duplex targeting the back-splice junction (BSJ) of circ-BCL2L12-92, along with the appropriate controls (HCT 116 cells transfected with scrambled siRNA and untransfected cells). The binding specificity of siRNA against circ-BCL2L12-92 BSJ yet not the 3′-untranslated region (3′-UTR) of BCL2L12 mRNA was previously demonstrated in a separate experiment. The expression levels of circ-BCL2L12-92 and BCL2L12 mRNA were determined using the relative quantification (2^−∆∆Ct) method: ciRS-7 served as an endogenous control for circ-BCL2L12-92, whereas GAPDH, ACTB and ACTG1 mRNAs served as references for BCL2L12 mRNA; the cDNA from untransfected HCT 116 cells was used as a calibrator. The graph shows the mean expression of each of circ-BCL2L12-92 and BCL2L12 mRNA along with its standard error; 3 biological replicates were used in each case