Comprehensive high-resolution analysis of the role of an Arabidopsis gene family in RNA editing

Abstract

In flowering plants, mitochondrial and chloroplast mRNAs are edited by C-to-U base modification. In plant organelles, RNA editing appears to be generally a correcting mechanism that restores the proper function of the encoded product. Members of the Arabidopsis RNA editing-Interacting Protein (RIP) family have been recently shown to be essential components of the plant editing machinery. We report the use of a strand- and transcript-specific RNA-seq method (STS-PCRseq) to explore the effect of mutation or silencing of every RIP gene on plant organelle editing. We confirm RIP1 to be a major editing factor that controls the editing extent of 75% of the mitochondrial sites and 20% of the plastid C targets of editing. The quantitative nature of RNA sequencing allows the precise determination of overlapping effects of RIP factors on RNA editing. Over 85% of the sites under the influence of RIP3 and RIP8, two moderately important mitochondrial factors, are also controlled by RIP1. Previously uncharacterized RIP family members were found to have only a slight effect on RNA editing. The preferential location of editing sites controlled by RIP7 on some transcripts suggests an RNA metabolism function for this factor other than editing. In addition to a complete characterization of the RIP factors for their effect on RNA editing, our study highlights the potential of RNA-seq for studying plant organelle editing. Unlike previous attempts to use RNA-seq to analyze RNA editing extent, our methodology focuses on sequencing of organelle cDNAs corresponding to known transcripts. As a result, the depth of coverage of each editing site reaches unprecedented values, assuring a reliable measurement of editing extent and the detection of numerous new sites. This strategy can be applied to the study of RNA editing in any organism.

Figure 1. The rip1 mutant is the only mutant in our study with a severe defective phenotype.

rip1, rip1-wt, rip3-2 and rip3-2-wt were grown under long-day growth room conditions (14 h light/10 h dark) while the other plants were grown in short day conditions (10 k light/14 h dark). rip3-2 mutant plants show a slight delay in development compared to their wild-type siblings. At the time of photography, 2 out of 8 (1/4) rip3-2 seedlings (30 days post-sowing) have flowered while all the wild-type seedlings have flowered (white spots in the middle of each plant are the flowers). rip4-1, rip4-2, and rip5-1 do not show any phenotypic difference with their wild-type (a rip5-1 wild-type was not available for the picture, but its genetic background is Columbia like rip4-2).

Figure 2. Examples of newly identified editing sites having sequences in their putative cis elements similar to known editing sites.

The sequences shown are 20 nt upstream and 5 nt downstream around the target C for editing (underlined and capitalized). Other Cs that are edited are underlined. The upper (lower) sequence belongs to a new (known) editing site. The names of the sites are given on the right of each sequence with the average editing extent found in the wild-type in between parentheses. Identical sequences are highlighted by red squares. Upon visual inspection, gaps were introduced in order to increase the similarity between cis elements.

Figure 3. Validation by PPE of two plastid sites detected by RNA-seq.

(A) Acrylamide gels separate the PPE products obtained from some samples used in this study; −/−, T-DNA mutant; +/+, wild-type. E, edited; P, primer; U, unedited. The name of the site assayed is given above each gel. (B) The quantification of editing extent derived from the measure of the band's intensity is represented by a bar below each lane of the acrylamide gels.

Figure 4. Measurement of RNA editing extent by RNA-seq is accurate and highly reproducible.

(A) The editing extent of 226 organelle editing events, 106 mitochondrial :9 sites and 3 to 19 genotypes assayed per site and 120 plastid: 12 sites and 7 to 11 genotypes assayed per site, was measured by RNA-seq and PPE assay. A high correlation with PPE assay (R² = 0.97), the most precise method to measure editing extent, demonstrates the robustness of RNA-seq to evaluate organelle editing extent. (B) Two libraries with different indexes were prepared from the same sheared cDNA obtained from rip1 mutant. The editing extent measured between the two libraries shows a very high correlation (R²>0.99).

Figure 5. Biological replicates exhibit a high correlation of editing extent.

Each graph represents a pairwise comparison of editing extent that was measured on two libraries obtained from cDNAs of two plants belonging to the same genotype grown in the same conditions and harvested at the same time.

Figure 6. Relative importance of RIPs on mitochondrial editing.

(A) Number of mitochondrial sites under the control of RIPs (ΔEE≥0.1, P<1.6 e-6). RIP8 results were obtained from VIGS (all the mitochondrial sites were counted in this analysis). Numbers of sites for RIP3 refer to rip3-1 (rip3-2). (B) Examples of mitochondrial sites falling into one of the eight categories described in the Venn diagram shown in (A). The background color reflects the range of editing extent from red (low: 0–0.2) to dark green (high: 0.8–1).

Figure 7. Relative importance of RIPs on plastid editing.

(A) Number of plastid sites under the control of RIPs (ΔEE≥0.1, P 2.7<10⁻⁵). RIP2 and RIP9 results were obtained from VIGS. (B) Examples of plastid sites falling into one of the eight categories described in the Venn diagram shown in (A). The background color reflects the range of editing extent from red (low: 0–0.2) to dark green (high: 0.8–1).