KREH2 helicase represses ND7 mRNA editing in procyclic-stage Trypanosoma brucei by opposite modulation of canonical and 'moonlighting' gRNA utilization creating a proposed mRNA structure

Abstract

Unknown factors regulate mitochondrial U-insertion/deletion (U-indel) RNA editing in procyclic-form (PCF) and bloodstream-form (BSF) T. brucei. This editing, directed by anti-sense gRNAs, creates canonical protein-encoding mRNAs and may developmentally control respiration. Canonical editing by gRNAs that specify protein-encoding mRNA sequences occurs amid massive non-canonical editing of unclear sources and biological significance. We found PCF-specific repression at a major early checkpoint in mRNA ND7, involving helicase KREH2-dependent opposite modulation of canonical and non-canonical 'terminator' gRNA utilization. Terminator-programmed editing derails canonical editing and installs proposed repressive structure in 30% of the ND7 transcriptome. BSF-to-PCF differentiation in vitro recreated this negative control. Remarkably, KREH2-RNAi knockdown relieved repression and increased editing progression by reverting canonical/terminator gRNA utilization. ND7 transcripts lacking early terminator-directed editing in PCF exhibited similar negative editing control along the mRNA sequence, suggesting global modulation of gRNA utilization fidelity. The terminator is a 'moonlighting' gRNA also associated with mRNA COX3 canonical editing, so the gRNA transcriptome seems multifunctional. Thus, KREH2 is the first identified repressor in developmental editing control. This and our prior work support a model whereby KREH2 activates or represses editing in a stage and substrate-specific manner. KREH2's novel dual role tunes mitochondrial gene expression in either direction during development.

Graphical Abstract

Figure 1.

KREH2 localization and association with proteins in REH2C in BSF cells. (A) Immunofluorescent microscopy of BSF T. brucei. Cells were imaged for marker proteins in editing complexes: KREH2 (REH2C) and KREL1 (RECC). White arrows point to DAPI-stained kDNA in each cell. (B) Western blot of mt-extracts (input) and KREH2 immunoprecipitation (KREH2-IP) from a BSF dyskinetoplastic cell line that we made via acriflavine treatment as described in the methods section, which is devoid of kDNA (–kDNA). The parental cell line that was not treated with acriflavine and contains kDNA (+kDNA) was included as a control. These BSF cell lines are conditional nulls of KH2F1 that constitutively express KH2F1-v5 tagged protein, as described in the methods section. KREH2, KH2F1-v5, KH2F2 (all in REH2C) and RESC2 and RESC13 (in RESC) were examined. KREH2 is often fragmented (fragment) in our mitochondrial extract preparations.

Figure 2.

Comparison of ND7 3′ total editing in PCF and BSF cells. ‘Snapshots’ of typical ND7 3′ amplicon RNA-seq data sets from (A) PCF and (B) BSF. Stacked histograms show all possible editing events at each site in replicate mtRNA (Mito) samples from cells containing a KREH2-RNAi construct –Tet (i.e. uninduced), as in all other panels below, which used the same uninduced construct. Color-coded nucleotides are just 3′ to canonical sites for U-insertion (Ins, red), U-deletion (Del, blue), or sites not expected to change in mature mRNA (black) (see Supplementary Table S1; glossary of terms). Bars represent the percentage of canonical insertion (red), deletion (blue), or non-canonical edits (yellow) at canonical sites or edits at sites not expected to change (black). Blocks of canonical editing thought to be directed by known gRNAs are indicated by indigo lines. (C) Site-by-site analysis of total edits across gRNA-1 and gRNA-2 (through site 55) in mtRNA from PCF versus BSF cells. (D) Cumulative total edits in PCF versus BSF mtRNA. (E) Cumulative total edits from mtRNA, RESC6-RNA-Immunoprecipitation (RIP), and RESC1-RIP. The cumulative value at the most 5′ site (site 117) in the amplicon was plotted. Full-amplicon analyses are in Supplementary Figure S2C. As noted above, all cells in this figure are uninduced. Average and error bars representing the standard deviation of biological replicates to determine P-values *** < 0.005, ** < 0.05, * < 0.5 were annotated. See Materials and methods and Supplementary Table S3 for additional details on statistical analysis.

Figure 3.

Analyses of ND7 3′ editing domain total editing upon KREH2-RNAi and KH2F1-RNAi in PCF and KREH2-RNAi in BSF. (A) Site-by-site, and (B) cumulative total edits across gRNA-1 and gRNA-2 (through site 55) in PCF mtRNA ± KREH2-RNAi. RNAi knockdowns decreased the targeted protein by ∼80% (see methods section). Sites 33 and 37–43 of particular interest in ND7 3′ early editing, as described in the text, are highlighted. (C) Cumulative total edits up to the most 5′ site (site 117) in the ND7 3′ domain amplicon in PCF mtRNA ± KREH2-RNAi or KH2F1-RNAi and BSF mtRNA ± KREH2-RNAi. (D) Cumulative total edits as in panel C but in RESC6- and RESC1-RIPs in PCF KREH2-RNAi or KH2F1-RNAi. Full-amplicon analyses are in Supplementary Figure S3. Black bars are +Tet, and gold bars –Tet (labeled ±T). Statistical analyses are as in Figure 2.

Figure 4.

Analyses of ND7 3′ domain editing fidelity (NC/C) upon KREH2-RNAi or KH2F1-RNAi in mtRNA or RESC. (A) Site-by-site and (B) cumulative NC/C ratios across gRNA blocks 1–2 in PCF mtRNA ± KREH2-RNAi. NC/C ratios are the percentage of non-canonical reads divided by the percentage of canonical reads at the same site. Sites 33, 37–43 and 49 (highlighted) exhibited particularly high NC/C ratios in the replicates. (C) Cumulative NC/C ratios at the most 5′ site (site 117) in the ND7 3′ domain amplicon in PCF mtRNA ± KREH2-RNAi or KH2F1-RNAi and BSF mtRNA ± KREH2-RNAi. (D) Cumulative NC/C ratios as in panel C but in RESC6- and RESC1-RIPs in PCF ± KREH2-RNAi or KH2F1-RNAi. (E) Cumulative canonical and (F) cumulative non-canonical edits at the most 5′ site (site 117) in mtRNA and RESC6-RIP in PCF or BSF ± KREH2-RNAi. All RNAi is ±Tet (labeled ±T). Note that the +Tet decreased NC/C, i.e. increased editing fidelity. Full-amplicon analyses are in Supplementary Figure S4 and Supplementary Table ST3. The highest NC/C in the amplicon at site 33 (arrowhead) caused the largest KREH2-promoted pause at site 32 (PPS32) in the sequence examined. Other annotations and statistical analyses are as in Figures 2 and 3.

Figure 5.

An abundant non-canonical 3′ high-frequency element (HFE) sequence in the ND7 3′ editing domain in PCF and BSF, and its modulation by REH2C proteins. (A) Canonically edited ND7 3′ terminus. Color-coded letters are just 3′ of sites requiring insertion (red), deletion (blue), or no changes (black). ORF, 3′ UTR and never-edited regions are indicated. The first editing site (ES1) is at position 17, counting from the 3′ end. Illumina sequenced gRNA isoforms: gRNA-1 B1 and gRNA-2 B2.alt in strain Lister 427 (31) correspond to gRNA-1 gND7(1269–1320) and gRNA-2 gND7(1240–1268) in strain EATRO 164 (22), with identical guiding function at block 1 and block 2, respectively. These guides produced the best match with the ND7 3′ edited pattern in this study and prior Sanger sequencing in our strain (31). Color-coded arrowheads indicate sites with a dominant NC read representing >30% (black) or less in the color-coded scale of all reads in RESC6-RIPs in uninduced (–Tet) PCF cells containing a KREH2-RNAi construct. (B) Canonical and non-canonical guiding potential of the initiator gRNA-1 B1. Alternative uninterrupted alignment of the 3′ terminus in gRNA-1 would involve non-canonical addition of + 2 U at site 33 and 0 U at site 31. (C) Actual percentage of the dominant NC read at each indicated site versus all reads (black) or versus all NC reads (white) in mtRNA and RESC6 or RESC1-RIPs in PCF. (D) Consensus 3′ High-Frequency Element (3′ HFE) made by the dominant NC reads at sites 33–43. The top two 3′ HFE isoforms found in all samples examined show the dominant NC reads in gray. General 3′ HFE long or short forms, where sites 34–36 included any T nucleotide number. Bottom: ∼62-nt extended 3′ element, including the 3′ HFE and 3′ terminal fully edited sequence. (E) Frequency of 3′ HFE (left) or canonically edited block 2 (right) in mtRNA or RESC6-RIPs in PCF ± KREH2-RNAi or KH2F1-RNAi. (F) Frequency of 3′ HFE, and either block 2 read type: canonically edited, pre-edited, and other NC (‘partial’) in PCF and BSF ± KREH2-RNAi cells. ±Tet (also labeled ±T). Statistical analyses are as in Figures 2 and 3.

Figure 6.

Relative changes in ND7 3′ edited blocks 1–4 upon KREH2-RNAi knockdown and in 3′ HFE upon in vitro differentiation. (A) RNA-seq analyses of the total reads (%) of canonically edited ND7 3′ domain blocks 1 through 4 in mtRNA or RESC6-RIPs in PCF ± KREH2-RNAi ± Tet. (B) qPCR analyses of relative levels for hallmark transcripts during in vitro differentiation. BSF strain Lister 427 cells before (day 0; d0) and after (day 2, d2, or day 8, d8) in vitro differentiation to PCF via 2 days of incubation at 37°C in BSF media + pCPTcAMP (cAMP) and 6 more days of incubation in PCF media with citrate/cis-aconitate at 26°C (full details of in vitro differentiation protocol in Materials and Methods). (C, D) Independent in vitro differentiation experiments and qPCR analyses (from two different labs; see Materials and Methods for more details) of 3′ HFE, canonically edited, and pre-edited ND7 normalized to mt-RNA polymerase mRNA (C) or TERT (D) as reference transcripts. Ratios in parent cells, PCF/BSF, or post/prior differentiation (±cAMP + CCA)/BSF are compared for the independent experiments in Lister 427 cells grown separately in two different labs for decades.

Figure 7.

Most frequent ND7 amplicons that contain the non-canonical 3′ HFE sequence. Sequence alignment of the top ten amplicons with the short-form 3′ HFE (sites 37–43) in a representative sample of mtRNA in PCF KREH2-RNAi (A) uninduced -Tet or (B) induced+Tet. The last edit (gray) in each unique sequence is indicated as a percentage in the top 100 amplicons. Sequence 5′ to the last edit is pre-edited or includes a variable length non-canonical editing junction. The count of each amplicon in the top 100 amplicons in that sample examined is shown. The cognate initiator gRNA-1 and gRNA-2 with an anchor (box) and guiding region (dashed line) in blue, a terminator gRNA matching the 3′ HFE (straight line) in red, and the first canonical editing site in ND7 3′ (position 17) are depicted. Color-coded letters are as in prior figures. (C) Junction length (JL) from panels A and B, counted as the number of sites containing any edits upstream of site 43. The site containing the last edit in the junction is indicated. Amplicons carry canonically edited sequence 3′ to the 3′ HFE. Junction length (JL) analyses of other samples were also performed (Supplementary Figures S9–S11).

Figure 8.

Coordinated non-canonical usage of the initiator gRNA-1 and a novel terminator guide may install the extended 3′ element in ND7 3′. (A) The extended 3′ element from sites 17 to 43 in ND7 3′, as in Figure 5D, aligned with gRNA-1 isoforms in strains BSF EATRO 1125, PCF EATRO 164, and PCF Lister 427 (used in this study) with non-canonical +2U at site 33 (boxed). A novel gRNA mO_235(Ia)_gCOX3(206–250) identified in strain EATRO 1125 (transcript #1) or isoforms in other strains (transcripts #2–6) match the 3′ HFE except for a single C:A mismatch (either C in a CC doublet) at site 35 or 36 in mRNA. Guide isoform #6 in strain Lister 427 has no matches but ends prematurely at site 41 in the 3′ HFE. (B) Sequence alignment of gRNA mO_235(Ia)_gCOX3(206–250) with its cognate mRNA COX3. Predicted ΔG values in kcal/mol) of the duplexes are indicated. The original stop codon created by gRNA-1 in ND7 3′, a never-edited sequence, the reverse (R) primer 5′ end (red line), and anchor regions (blue) are annotated. (C) Number of gRNA transcripts in available databases in PCF and BSF. Initiator gRNA-1 isoforms in EATRO 164, gND7 (1269–1320), Lister 427, gND7 B1, and EATRO 1125, mO_095(II)_gND7(1186–1229). gRNAs in EATRO 164 and EATRO 1125 were determined in total mtRNA; gRNAs in Lister 427 were determined in RESC6-IPs and available alignments online (31). Transcripts no. 2 (&), no. 5 (#), and no. 4 (*). gND7(1240–1268) B2.alt ($), and mO_094(II)_gND7(1186–1129) (^).

Figure 9.

Structural models determined by DMS-MaPseq of the (A) pre-edited (PE) and (B) 3′ HFE-containing ND7 3′ terminus. Nucleotides are color-coded by normalized DMS signal. The 3′ HFE is highlighted in a green box; nucleotides used in binding the anchor region in cognate ND7 gRNA-1 and gRNA-2 and non-cognate gCOX3 are highlighted in blue, orange, and royal blue, respectively. Both structures carry a complete anchor binding site for cognate gND7 gRNA-1. Only the 3′ HFE-containing structure carries a complete element (sites 33–43) and complete anchor binding sites for cognate gND7 gRNA-2 and non-cognate gCOX3. R² is Pearson's R² in structures from n = 2 experiments. DMS reactivity is calculated as the ratiometric DMS signal per position normalized to the highest number of reads in the displayed region, which is set to 1.0.

Figure 10.

Model of REH2C-dependent developmental ND7 editing regulation via modulation of gRNA selection and RNA structure. In this model, KREH2 maintains a state of ND7 3′ editing repression in PCF involving modulation of gRNA selection and installation of a proposed RNA structure. PCF-specific repression was defined at two levels: (A) at a major early editing checkpoint where KREH2 exhibits concurrent positive (+) control on usage of a novel ‘terminator’ guide (gCOX3) and negative (–) control on usage of the canonical cognate gRNA-2 (gND7). KREH2 promotes preferential usage of the terminator over gND7. Editing using the terminator installs a non-canonical 3′ HFE in ∼30% of the ND7 transcriptome, which has two main effects: first, it derails all upstream canonical editing, and second, it forms a stable fold, potentially hindering downstream gRNA-2 binding and ‘repair’ via re-editing to replace the 3′ HFE with canonical sequence, and (B) globally, with gRNA fidelity changes by KREH2 promoting ‘repressive’ non-cognate guiding and inhibiting canonical cognate guiding at most ND7 3′ sites examined. Thus, inhibition of ND7 3′ editing maturation in PCF is mainly at the early checkpoint but also occurs globally in transcripts that did not receive the 3′ HFE. KREH2-mediated repression in ND7 3′ was not observed in BSF cells.