Simultaneous processing and degradation of mitochondrial RNAs revealed by circularized RNA sequencing

Abstract

Mammalian mitochondrial RNAs are unique as they are derived from primary transcripts that encompass almost the entire mitochondrial genome. This necessitates extensive processing to release the individual mRNAs, rRNAs and tRNAs required for gene expression. Recent studies have revealed many of the proteins required for mitochondrial RNA processing, however the rapid turnover of precursor RNAs has made it impossible to analyze their composition and the hierarchy of processing. Here, we find that circularization of RNA prior to deep sequencing enables the discovery and characterization of unprocessed RNAs. Using this approach, we identify the most stable processing intermediates and the presence of intermediate processing products that are partially degraded and polyadenylated. Analysis of libraries constructed using RNA from mice lacking the nuclease subunit of the mitochondrial RNase P reveals the identities of stalled processing intermediates, their order of cleavage, and confirms the importance of RNase P in generating mature mitochondrial RNAs. Using RNA circularization prior to library preparation should provide a generally useful approach to studying RNA processing in many different biological systems.

Figure 1.

Schematic illustrating the preparation and analysis of circularized RNA libraries. Purified RNA was treated with a thermostable single-stranded nucleic acid-specific ligase (CircLigase II) to achieve intramolecular ligation of individual molecules. After ligation any remaining linear RNA was degraded using RNase R and the circular RNAs were reverse transcribed using random primers incorporating an adaptor sequence for subsequent sequencing. A second adaptor sequence was added by extension of the cDNA using a 3΄-end-blocked oligonucleotide as a template. The cDNA was then amplified by limited PCR. Primers were designed to incorporate sequences for subsequent deep sequencing.

Figure 2.

Analysis of coverage from libraries produced by circularized RNA sequencing compared to standard RNA-Seq. The mitochondrial genome is separated into the region containing the rRNAs (A) and mRNAs (B). Coverage is shown in dark grey for heavy strand-encoded RNAs and in light gray for light strand-encoded transcripts. The ratio of coverage from circularized RNA sequencing compared to RNA-Seq is shown as log₂ fold change with regions of increased coverage in red and decreased coverage in blue. The locations of genes are shown in a central schematic, where genes encoding rRNAs are shown in green, mRNAs are shown in light blue and tRNAs in gray. The ‘mt-’ prefix is omitted from each gene name for clarity.

Figure 3.

Analysis of partially processed transcripts in libraries produced by circularized RNA sequencing. Abundance of reads which map between multiple genes are shown across the mitochondrial genome in both wild-type mouse mitochondria and Mrpp3 knock out mitochondria (A). A log₂ fold change (log₂FC) highlights partially processed regions that are enriched in the absence of MRPP3 with regions of increased coverage in red and decreased coverage in blue. Partially processed transcripts containing the 12S rRNA are shown in (B) and the region containing mt-Th, mt-Ts2 and mt-Tl2, is highlighted by dotted lines. The locations of genes are shown in a schematic, where genes encoding rRNAs are shown in green, mRNAs are shown in light blue and tRNAs in gray.

Figure 4.

Analysis of partially processed transcripts containing mt-Atp8/6. (A) Circularized RNA sequencing of wild-type mouse mitochondrial RNA. Reads mapping between mt-Atp8/6 and other mitochondrial RNAs are illustrated by lines between the 5΄ and 3΄ nucleotide positions. The locations of genes are shown in a schematic, where heavy strand genes are shown in the outer ring and light strand genes in the inner ring. Genes encoding rRNAs are shown in green, mRNAs are shown in light blue, tRNAs in gray, the non-coding D-loop in white, and the mRNA of interest, mt-Atp8/6, is shown in orange. (B) Circularized RNA sequencing of mitochondrial RNA from Mrpp3 knock out mice. (C) Circularized RNA sequencing data illustrating the increase in coverage of unprocessed transcripts containing mt-Atp8/6 in the absence of MRPP3. (D) PARE data illustrating the reduction in 5΄ and 3΄ processing of mt-Tk in the absence of MRPP3. (E) Northern blotting of heart RNA from wild-type and Mrpp3 knock out mice using a probe specific for mt-Atp8/6. (F) The 5΄ ends of reads mapping with the genomic region containing the 5΄ end of mt-Co3 are shown from the circularized RNA sequencing and PARE data. The putative 5΄ end position of mt-Co3 is highlighted in red in the data and the mt-Co3 mRNA is highlighted in orange in the schematic. (G) A 5΄-phosphate-dependent exonuclease (terminator exonuclease) was used to digest isolated mouse mitochondrial RNA using either a high specificity reaction buffer (lane indicated by a +) or a high activity reaction buffer (lane indicated by ++). Specific RNAs were subsequently detected by northern blotting.

Figure 5.

Analysis of partially processed transcripts containing mt-Co1. (A) Circularized RNA sequencing of wild-type mouse mitochondrial RNA. Reads mapping between mt-Co1 and other mitochondrial RNAs are illustrated by lines between the 5΄ and 3΄ nucleotide positions. The locations of genes are shown in a schematic, where heavy strand genes are shown in the outer ring and light strand genes in the inner ring. Genes encoding rRNAs are shown in green, mRNAs are shown in light blue, tRNAs in grey, the D-loop in white, and the mRNA of interest, mt-Co1, is shown in orange. (B) Circularized RNA sequencing of mitochondrial RNA from Mrpp3 knock out mice. (C) PARE data illustrating the reduction in 5΄ and 3΄ processing of mt-Td in the absence of MRPP3 and also the position of mt-Co1's 3΄-UTR (12). (D) Northern blotting of heart RNA from wild-type and Mrpp3 knock out mice using a probe specific for mt-Co1.

Figure 6.

Polyadenylation of mRNA decay intermediates. (A) 5΄ rapid amplification of cDNA ends (5΄-RACE) validates the 5΄ ends of mt-Co1. 3΄-RACE identified total (B) and polyadenylated (C) processing intermediates that include mt-Co1. Circularized RNA sequencing captured authentic small RNAs (D). The abundance of small RNAs (sRNAs) of different sizes identified by circularized RNA sequencing were compared to those detected by classical sRNA sequencing. The positions of the 19 and 22 nt sRNAs previously identified in the human mitochondrial transcriptome (27) are indicated by arrows. (E) Locations of sRNA 5΄ and 3΄ ends within mt-Co1 and the fold change between wild-type and Mrpp3 knockout mitochondrial circularized RNA libraries are shown. Small RNAs derived from mt-Co1 were increased in abundance in the Mrpp3 RNA libraries (Student's t-test, P = 0.0034). (F) A proportion of sRNA degradation products derived from mRNAs are polyadenylated. The proportion of polyadenylated sRNAs derived from the following mRNAs were significantly increased upon loss of MRPP3 (Student's t-test, P ≤ 0.05, indicated by an asterisk): mt-Co1, mt-Co2, mt-Atp8/6, mt-Co3, mt-Nd6 and mt-Cytb.

Figure 7.

Analysis of partially processed transcripts containing mt-N4l/4. (A) Circularized RNA sequencing of wild-type mouse mitochondrial RNA. Reads mapping between mt-Nd4l/4 and other mitochondrial RNAs are illustrated by lines between the 5΄ and 3΄ nucleotide positions. The locations of genes are shown in a schematic, where heavy strand genes are shown in the outer ring and light strand genes in the inner ring. Genes encoding rRNAs are shown in green, mRNAs are shown in light blue, tRNAs in grey, the D-loop in white, and the mRNA of interest, mt-Nd4l/4, is shown in orange. (B) Circularized RNA sequencing of mitochondrial RNA from Mrpp3 knock out mice. (C) PARE data illustrating the various positions of mt-Nd4l/4's short 5΄-UTRs. (D) Northern blotting of heart RNA from wild-type and Mrpp3 knock out mice using a probe specific for mt-Nd4l/4.