ANGEL2 phosphatase activity is required for non-canonical mitochondrial RNA processing

Abstract

Canonical RNA processing in mammalian mitochondria is defined by tRNAs acting as recognition sites for nucleases to release flanking transcripts. The relevant factors, their structures, and mechanism are well described, but not all mitochondrial transcripts are punctuated by tRNAs, and their mode of processing has remained unsolved. Using Drosophila and mouse models, we demonstrate that non-canonical processing results in the formation of 3' phosphates, and that phosphatase activity by the carbon catabolite repressor 4 domain-containing family member ANGEL2 is required for their hydrolysis. Furthermore, our data suggest that members of the FAST kinase domain-containing protein family are responsible for these 3' phosphates. Our results therefore propose a mechanism for non-canonical RNA processing in metazoan mitochondria, by identifying the role of ANGEL2.

Fig. 1. DmANGEL and MmANGEL2 are mitochondrial proteins required for normal respiratory chain function.

a Confocal imaging of human skin fibroblasts expressing GFP-tagged DmANGEL, HsANGEL1 or HsANGEL2 constructs. (Scale bar = 20 mm; Zoom = 4.5X). Representative images are shown of three independent experiments with 20 cells analysed per experiment. b Western blot analysis of sub-cellular fractionations from human skin fibroblasts expressing ANGEL1-FLAG or ANGEL2-proteinC fusion proteins expressed in human skin fibroblasts, decorated with antibodies against nuclear (Histone 3), cytosolic (GAPDH) or mitochondrial (HSP60) fractions. Representative experiment of two independent experiments, with two technical replicas. c Diagram depicting the sub-mitochondrial localisations of DmANGEL, ANGEL1 and ANGEL2. Mitochondria are illustrated as outer mitochondrial membrane (OMM), inner mitochondrial membrane (IMM) and matrix. Isolated respiratory chain enzyme activities for NADH:ubiquinone oxidoreductase (CI), NADH:cytochrome c oxidoreductase (CI + III), succinate:ubiquinone oxidoreductase (CII), succinate: cytochrome c oxidoreductase (CII + III), and cytochrome c oxidase (CIV), in isolated mitochondria from (d) 4-days-after-egg-laying (dael) DmANGEL KO (red; DmAngel KO1,2) and control (light grey; w^Dah) larvae, and (e) 16-week-old mouse hearts (blue; Angel2 KO) and controls (light grey; Angel2 WT). (Data in d, e are represented as mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001 with two-tailed Student’s t test; Dm n = 5, Mm wt n = 4, Mm KO n = 3 biologically independent samples). f Proteomic levels of individual subunits of the five OXPHOS complexes, accessory, and OXPHOS assembly factors in 4-dael DmAngel^KO larvae. Data are presented as 25 to 75 percentile box with median, and whiskers represent ±1.5× inter-quartile range (n = 4 biologically independent samples). No statistical test was performed. g Gene set enrichment analysis of proteomic data in 16-week-old mouse hearts, relative to controls. p values are given as false discovery rates after multiple testing adjustment (n = 3 biologically independent samples).

Fig. 2. Deletion of DmAngel or MmANGEL2 affects mitochondrial gene expression.

Relative mitochondrial transcript steady-state levels in 16-week-old mouse hearts (blue; Angel2 KO) and controls (light grey; Angel2 WT), as determined by Northern blot analysis (n = 5 biologically independent samples) (a, b) or qRT-PCR (n = 4 biologically independent samples with 3 technical replicas) (c). 18S rRNA was used as loading control. d Illustration of mouse mtDNA (NC_005089). Black arrows depict transcription initiation sites and tRNAs are indicated in orange. Relative mitochondrial transcript steady-state levels in 4-dael DmANGEL KO (red; DmAngel KO1,2) and control (light grey; w^Dah) larvae, as determined by Northern blot analysis (e, f) or qRT-PCR (g). RP49 was used as loading control (n = 5 biologically independent samples). h Illustration of Dm mtDNA (NC_001709). Black arrows depict transcription initiation sites and tRNAs are indicated in orange. (All data are represented as mean ± SD; *p < 0.05, **p < 0.01, ***p < 0.001 with two-tailed Student’s t test).

Fig. 3. Non-canonical processing of mitochondrial gene junctions results in the formation of 3′ phosphates that are hydrolysed by DmANGEL or MmANGEL2.

a Mitochondrial polyadenylation tail (MPAT) length assay in MmAngel2^KO mouse heart samples, performed with or without CIP or PNK pre-treatment, as indicated. b Mitochondrial polyadenylation tail (MPAT) length assay in DmAngel^KO fly samples, performed with or without CIP or PNK pre-treatment, as indicated. CIP calf intestinal phosphatase, hydrolyses phosphomonoester bonds. PNK T4 polynucleotide kinase, hydrolyses phosphomonoester bonds or 2′,3′ cyclic phosphodiester from RNA ends. Gene junctions and their sequences are indicated. Non-coding sequences are lower case. Sequences not annotated are shown in lower case grey. tRNAs are shown in orange with their single letter code. Poly(A) tails (-A_n) and 3′ phosphates (-℗) are shown. A representative experiment is shown of at least three independent experiments performed with biologically independent samples.

Fig. 4. ANGEL2 and DmANGEL are terminal phosphatases in vivo and in vitro.

MPAT length assay using RtcB ligase in a MmAngel2^KO mouse heart or b DmAngel^KO fly samples. c In vitro phosphatase activity of recombinant Drosophila (DmAngel), mouse (MmANGEL), or human (HsANGEL) ANGEL or ANGEL2 protein, in the presence of RNA oligos (20 nts) containing 3′-, 2′-, or 2′,3′-cPs as indicated. PNK treatment was used as positive control. Representative experiments are shown of at least two independent experiments performed with biologically independent samples.

Fig. 5. Incomplete non-canonical processing due to loss of DmANGEL results in aberrant mitochondrial translation.

a De novo translation in isolated mitochondria from 4-dael control (w^Dah) and DmAngel^KO (KO 1,2) larvae. Mitochondrial proteins are indicated. Porin was used as loading control. A representative experiment is shown with n = 2 biologically independent samples, of two independent experiments performed. b Western blots of ribosome gradient fractions from 4-dael control and DmAngel^KO larvae, decorated with antibodies against the small (MRPS16) and large (MRPL12) mitochondrial ribosome subunits. The small (28S), large (39S), and monosome (55S) fractions are indicated. A representative experiment is shown of four independent experiments performed with biologically independent samples. c, d Abundance of mitoribosome footprints in mitochondrial transcripts from ribosome profiling experiments in control and DmAngel^KO as indicated, normalised to total read count and expressed as proportion of total footprints (n = 2 biologically independent samples).

Fig. 6. Members of the FASTK family are required for non-canonical processing of mitochondrial transcripts.

a Volcano plot of proteomics data from immunoprecipitation (IP) experiments, using 4 dael larvae from DmANGEL-FLAG versus DmANGEL overexpressing larvae (n = 4). Proteins in the category “mitochondria” are highlighted in blue. Limma moderated t-test, p values are given as false discovery rate after adjustment for multiple testing (n = 4 biologically independent samples). b Heatmap of the enrichment of DmANGEL and CG13850 in CG13850-FLAG IP experiments (n = 4 biologically independent samples). c Neighbour-joining protein MUSCLE alignment of 30 FASTK family members (see Supplementary Table 1) from five species (as indicated) shown in an unrouted tree layout. CG13850 and CG31643 are the two Drosophila FASTK family orthologs. d Northern blot analysis of selected mitochondrial transcripts in 4-dael control (daGAL4 and RNAi) larvae and larvae with silenced CG13850 (CG13850 RNAi). RP49 was used as loading control. Representative experiment is shown with n = 2 biologically independent samples, of two independent experiments performed. e BN-PAGE and in-gel activity staining of NADH dehydrogenase (Complex I), cytochrome c oxidase (Complex IV), and ATP synthase (Complex V) with a Coomassie stain as a loading control. Dissociation of the ATPase F₁ subunit is indicated by *. Representative experiment is shown with n = 2 biologically independent samples.

Fig. 7. ANGEL2 phosphatase activity is required for the maturation of mRNAs that undergo non-canonical processing.

Model of the non-canonical processing of mitochondrial RNA. The mitochondrial RNA polymerase, POLRMT, transcribes long, polycistronic transcripts, which are cleaved by RNase P and ELAC2 at the tRNA gene junctions. Non-canonical junctions are recognised by members of the FASTK family (FASTK*), resulting in the formation of 3′ phosphates. Phosphatase activity by ANGEL2 hydrolyses the 3′ ends of mt:Nd5 and mt:Atp8/6, allowing further their polyadenylation by the mitochondrial poly(A) polymerase, MTPAP.