Circular RNA (circRNA) is a newly validated class of single-stranded RNA, ubiquitously expressed in mammalian tissues and possessing key functions including acting as microRNA sponges and as transcriptional regulators by binding to RNA-binding proteins. While independent studies confirm the expression of circRNA in various tissue types, genome-wide circRNA expression in the heart has yet to be described in detail.
Researchers at the Genome Institute of Singapore performed deep RNA-sequencing on ribosomal-depleted RNA isolated from 12 human hearts, 25 mouse hearts and across a 28-day differentiation time-course of human embryonic stem cell-derived cardiomyocytes. Using purpose-designed bioinformatics tools, they uncovered a total of 15 318 and 3017 cardiac circRNA within human and mouse, respectively. Their abundance generally correlates with the abundance of their cognate linear RNA, but selected circRNAs exist at disproportionately higher abundance. Top highly expressed circRNA corresponded to key cardiac genes including Titin (TTN), RYR2, and DMD. The most abundant cardiac-expressed circRNA is a cytoplasmic localized single-exon circSLC8A1-1. The longest human transcript TTN alone generates up to 415 different exonic circRNA isoforms, the majority (83%) of which originates from the I-band domain. Finally, the researchers confirmed the expression of selected cardiac circRNA by RT-PCR, Sanger sequencing and single molecule RNA-fluorescence in situ hybridization.
Global landscape of human cardiac circRNA
(A), Schematic showing a sequencing back-spliced junction-read leading to the identity of the single-exon human circRNA circSLC8A1. (B), Chart showing circRNA detected in human hearts. (C), A map of circRNA isoforms expressed from the TTN gene locus. Larger loops represent back-splice junctions spanning across multiple exons. Note small black loops spanning single or fewer exons, especially in the A-band domain. circTTN-366 (blue loop) is by prediction the longest cardiac circRNA that spans 153 exons. (D), Integrated panels to show the relationship between linear protein coding or non-coding gene expression and their corresponding circRNA expression. Heatmap shows the average expression of linear genes ranked by expression abundance (highest FPKM: yellow; lowest FPKM: blue). Scatterplots show the expression abundance of circRNA isoforms corresponding to their protein-coding or non-coding linear genes in the rank position of the heatmap. The size of each point in the scatterplots is adjusted according to the number of circRNA isoforms generated by each linear gene. (E), Scatterplot showing the relationship between the abundance of each back-spliced junction representative of the circRNA (x-axis) to their corresponding linear spliced junction representing the linear transcript (y-axis), in a healthy human heart sample. Some genes express significantly more circRNA than their corresponding linear transcript such as SLC8A1, TTN, and RYR2. Otherwise, most genes such as TPM1 and TNNT2 express more linear transcripts than circRNA. (F), Boxplot showing the relationship between linear gene length and the number of their circRNA isoforms per linear gene. Linear genes were divided into quartiles based on their gene lengths. (G), Barplot showing that introns flanking circRNA are significantly longer than the average intron. P-value <0.01. (H), Barplot showing that introns flanking circRNAs are significantly enriched with Alu repetitive elements. P-value <0.01. P-values were obtained by non-parametric bootstrapping.
These data provide a detailed circRNA expression landscape in hearts. There is a high-abundance of specific cardiac-expressed circRNA. These findings open up a new avenue for future investigation into this emerging class of RNA.