Trypanosomatid mitochondrial RNA editing: dramatically complex transcript repertoires revealed with a dedicated mapping tool

Abstract

RNA editing by targeted insertion and deletion of uridine is crucial to generate translatable mRNAs from the cryptogenes of the mitochondrial genome of kinetoplastids. This type of editing consists of a stepwise cascade of reactions generally proceeding from 3' to 5' on a transcript, resulting in a population of partially edited as well as pre-edited and completely edited molecules for each mitochondrial cryptogene of these protozoans. Often, the number of uridines inserted and deleted exceed the number of nucleotides that are genome-encoded. Thus, analysis of kinetoplastid mitochondrial transcriptomes has proven frustratingly complex. Here we present our analysis of Leptomonas pyrrhocoris mitochondrial cDNA deep sequencing reads using T-Aligner, our new tool which allows comprehensive characterization of RNA editing, not relying on targeted transcript amplification and on prior knowledge of final edited products. T-Aligner implements a pipeline of read mapping, visualization of all editing states and their coverage, and assembly of canonical and alternative translatable mRNAs. We also assess T-Aligner functionality on a more challenging deep sequencing read input from Trypanosoma cruzi. The analysis reveals that transcripts of cryptogenes of both species undergo very complex editing that includes the formation of alternative open reading frames and whole categories of truncated editing products.

Figure 1.

(A) A gene map of the coding region of the Leptomonas pyrrhocoris mitochondrial maxicircle. Edited transcript regions are hatched, and never-edited transcripts are shown in blue. (B) Read mapping performance compared for bowtie2 (blue bars), bowtie2-mod (purple bars) and T-Aligner (green bars) on all genes. Read mapping performance was quantified in reads per one nucleotide of the genomic sequence (100 nt flanks on both sides were also counted). The inset shows genes expressed at a low level.

Figure 2.

Reconstruction of limited edited domains of the L. pyrrhocoris ND7 (A), COIII (B), CYb (C) and MURF2 (D) maxicircle cryptogenes. Only 5′ parts of transcripts are shown; for full reconstructions see Supplementary Figure S2. The number of Ts in the genomic sequence is shown with stacked red bars at top. Each center panel shows editing states at each site found in at least two reads (light-grey dots). Proportion of reads supporting an editing state is color-coded, with the blue dot being the most supported and black, the next most supported. Cloud coverage diagrams of editing states at each edited site are shown in the bottom panel. Each read is plotted as a semi-transparent circle. ORFs visualized as a path through editing states are shown, with the red line representing the canonical ORF in each figure. Minor alternative ORFs are represented in blue and green. Red triangles, start codons. For ND7 (A), non-canonical edited sites within the blue ORF are marked with arrows. Editing sites are numbered in COIII (B).

Figure 3.

Overview of L. pyrrhocoris ND8 transcript editing and reconstruction of the canonical edited product and two abundant alternative products. The number of Ts in the genomic sequence is shown with stacked red bars at top. (A) Visualization of editing states found in at least two reads (light-gray dots). Proportion of reads supporting an editing state is calculated for each site separately and color-coded with the blue dot being the most supported and black, the next most supported. (B) Absolute coverage bar graph with proportions of insertion edited (blue), deletion edited (pink), or never-edited reads (green). The proportion of edited reads at each site is also shown with black vertical lines. Y-axis values for both metrics are provided. (C) ORFs visualized as paths through editing states. Translatable editing states (those included into at least one ORF >60 aa in length) are boxed in green and overlaid with the canonical ORF (red) and alternative ORFs (black and blue). Where editing differs in the black and blue ORF compared with the canonical ORF, it is indicated by dots between panels C and D. Red triangles, start codons; red squares, stop codons. (D) Cloud coverage diagrams of editing states at each edited site. Each read is plotted as a semi-transparent dot. The canonical and alternative ORFs shown in C are also plotted.

Figure 4.

Overview of L. pyrrhocoris RPS12 transcript editing and reconstruction of the canonical edited product and two abundant alternative products. The number of Ts in the genomic sequence is shown with stacked red bars at top. (A) Visualization of editing states found in at least two reads (light-gray dots). Proportion of reads supporting an editing state is calculated for each site separately and color-coded, with the blue dot being the most supported and black, the next most supported. (B) Absolute coverage bar graph with proportions of U-indel insertion edited (blue), deletion edited (pink), or never-edited reads (green). The proportion of edited reads at each site is also shown with black vertical lines. Y-axis values for both metrics are provided. (C) ORFs visualized as paths through editing states. Translatable editing states (those included into at least one ORF >60 aa in length) are boxed in green and overlaid with the canonical ORF (red) and alternative ORFs (black and blue). Where editing differs in the black and blue ORF compared with the canonical ORF, it is indicated by dots between panels C and D. Red triangles, start codons; red squares, stop codons. (D) Cloud coverage diagrams of editing states at each edited site. Each read is plotted as a semi-transparent dot. The canonical and alternative ORFs shown in C are also plotted.

Figure 5.

Overview of T. cruzi RPS12 transcript editing and reconstruction of the canonical edited product, a well-supported 5′ truncated alternative ORF, and a full-length alternative ORF with an N-terminal shifted reading frame. The number of Ts in the genomic sequence is shown with stacked red bars at top. (A) Visualization of editing states found in at least two reads (light-grey dots). Proportion of reads supporting an editing state is calculated for each site separately and color-coded, with the blue dot being the most supported and black, the next most supported. (B) Absolute coverage bar graph with proportions of U-indel insertion edited (blue), deletion edited (pink), or never-edited reads (green). The proportion of edited reads at each site is also shown with black vertical lines. Y-axis values for both metrics are provided. (C) ORFs visualized as paths through editing states. Translatable editing states (those included into at least one ORF >60 aa in length) are boxed in green and overlaid with the canonical ORF (red) and alternative ORFs (black and blue). Where editing differs in the black and blue ORF compared with the canonical ORF, it is indicated by dots between panels C and D. Red triangles, start codons; red squares, stop codons. (D) Cloud coverage diagrams of editing states at each edited site. Each read is plotted as a semi-transparent dot. The canonical and alternative ORFs shown in C are also plotted. Blue horizontal lines in B, C and D indicate a domain of editing that remains unedited in the ‘blue’ truncated alternative product.

Figure 6.

Percentages of L. pyrrhocoris mapped reads with at least 1 to at least 20 alternatively edited sites (i.e. sites with editing states different from those within the canonical open reading frame and different from the genomic sequence). All maxicircle protein-coding genes except for G3 and MURF5 (due to low coverage; Supplementary Table S1) were analyzed. Non-edited transcripts are shown in blue, minimally edited in green, and pan-edited in red. Several transcript groups can be distinguished based on the relative abundance and extent of alternatively edited reads: (i) non-edited and minimally edited transcripts with very short edited domains; (ii) MURF2, and A6 with a longer edited domain; (iii) pan-edited transcripts.