Multiple organellar RNA editing factor (MORF) family proteins are required for RNA editing in mitochondria and plastids of plants

Abstract

RNA editing in plastids and mitochondria of flowering plants changes hundreds of selected cytidines to uridines, mostly in coding regions of mRNAs. Specific sequences around the editing sites are presumably recognized by up to 200 pentatricopeptide repeat (PPR) proteins. The here identified family of multiple organellar RNA editing factor (MORF) proteins provides additional components of the RNA editing machinery in both plant organelles. Two MORF proteins are required for editing in plastids; at least two are essential for editing in mitochondria. The loss of a MORF protein abolishes or lowers editing at multiple sites, many of which are addressed individually by PPR proteins. In plastids, both MORF proteins are required for complete editing at almost all sites, suggesting a heterodimeric complex. In yeast two-hybrid and pull-down assays, MORF proteins can connect to form hetero- and homodimers. Furthermore, MORF proteins interact selectively with PPR proteins, establishing a more complex editosome in plant organelles than previously thought.

Fig. 1.

The MORF1 protein is required for RNA editing at multiple sites. (A) Sample sequences of the more than 40 editing sites affected in morf1-1 EMS mutant plants. The first five sites show editing reduced to different degrees. At the last site, editing increases in the mutant in comparison with wild-type plants of A. thaliana ecotype Columbia. (B) Structure of the MORF1 gene and the MORF1 protein. The location of the morf1-1 single-nucleotide alteration changing a proline to a serine codon and the T-DNA insertion site in morf1-2 are indicated. LB denotes the location of the left border of the T-DNA. The darker shading in the MORF protein marks the conserved MORF domain. (C) Stable transformation of morf1-1 mutant plants with the wild-type Col gene under control of a 35S promoter complements the editing defects. (D) The T-DNA insertion line morf1-2 is homozygous lethal; homozygous seed growth is aborted (arrows) in pods on a selfed heterozygous plant. (E) Wild-type Col plants show the full seed set.

Fig. 2.

The MORF family of proteins contains nine genes and a potential pseudogene in A. thaliana. (A) The cladogram of similarities between the MORF proteins shows that the plastid editing factors MORF2 and MORF9 are rather distant from each other and more similar to the mitochondrial proteins MORF3 and MORF1, respectively. Predictions (marked mt or cp) and experimental data obtained by GFP-fusion protein localization (only MORF2) or proteomics MS data (marked with an asterisk) for the respective organellar locations are indicated. The MORF8 protein encoded by At3g15000 has been found in mitochondria in three independent assays. Proteins investigated here for their function are boxed. The conserved ∼100-amino acids domain is shaded; the other sequences show much less conservation (SI Appendix, Fig. S2). The potential pseudogene (At1g53260) contains only the C-terminal part of this conserved region. (B) Exon structures of the MORF3, MORF4, and MORF6 genes are similar to the MORF1 locus and contribute similar fragments but differ in their C-terminal extensions. MORF3 is a mitochondrial editing factor involved at more than 40 sites. Locations of the T-DNA insertions in the mutants morf3-1, morf4-1, and morf6-1 are shown. LB denotes the location of the left border of the T-DNA. (C) Numbers of editing sites affected by T-DNA insertions in the respective MORF genes. In the mutants morf4-1 and morf6-1, only one noncoding site each shows somewhat reduced editing.

Fig. 3.

MORF2 and MORF9 are required for RNA editing in plastid mRNAs. (A) Exons of the MORF2 and MORF9 genes yield similar-sized proteins, although the intron structures vary. Sites of the T-DNA insertions in the homozygous mutants morf2-1 and morf9-1 are shown. (B) Phenotype of the morf2-1 mutant shows a complete lack of chlorophyll biosynthesis in light, and plantlets have to be grown on sugar-containing medium. This mutant is allelic to the dag and dal mutants described in Antirrhinum and Arabidopsis, respectively. (Scale bar, 1 mm.) (C) In the morf9-1 mutant, the cotyledons are fully green but the leaves show a variegated appearance with spots of green on otherwise whitish leaves. (Scale bar, 1 mm.) (D–F) Sample plants of the morf9-1 mutants show the variation of the green islands in intensity and distribution between individuals. These plants are able to grow autotrophically on soil. (Scale bars, 1 cm.) (G) Several of the affected editing sites are shown that document the differing influence of the MORF2 and MORF9 genes. Site ndhD-2 canonically requires both intact MORF proteins. Several sites cannot be edited without intact MORF2 (e.g., site psbZ-50); others require functional MORF9 proteins (e.g., site petL-5). Most of the sites show reduced editing in the absence of either factor, suggesting that the two MORF proteins act in concert at the same sites and that heterodimeric combinations of the two proteins are required for optimal editing.

Fig. 4.

MORF and MEF proteins can physically interact in yeast two-hybrid assays. (A) MORF proteins interact with each other. Reciprocal assays with the MORFs in bait (pGBKT7) or prey (pGADT7) vectors in a yeast two-hybrid analysis reveal that these proteins can interact with themselves in homodimers and with each other in heterodimers. The least specific appears to be the mitochondrial MORF1 protein, which can contact all other MORFs in either direction. Another mitochondrial protein, MORF3, forms strong heterodimers only with the likewise mitochondrial MORF1. The plastid proteins MORF2 and MORF9 interact with each other and with the mitochondrial MORF1, but only weakly with the mitochondrial MORF3. “Empty” is the control for autoactivation. +++ indicates a strong interaction; + represents fewer, slower-growing colonies formed; - indicates no colonies. (B) MORF and MEF proteins interact in yeast two-hybrid assays. Respective MEFs are indicated for each plate, and their protein structures are shown. The MORFs tested for binding are numbered in their respective quadrants. The mitochondrial editing protein MORF1, for example, interacts with the mitochondrial editing factors MEF1 (weakly), MEF9, and MEF21, but not with MEF11. These results show that principally MORF and MEF proteins interact rather unspecifically, as, for example, the binding of the plastid proteins MORF2 and MORF9 with the mitochondrial MEF1, MEF9, and MEF21 PPR proteins shows. However, some combinations are preferred, and others do not occur. No-growth quadrants show that there is no autoactivation. (C) The interactions between MEF and MORF proteins documented in B are interpreted as strong interactions (+++), weak (+), or no (-) binding. The results show that the two DYW-containing proteins, MEF1 and MEF11, interact weakly or not at all with the MORF proteins, whereas the MEF9 and MEF21 proteins, which terminate after the E domain and do not contain a DYW extension, connect more readily and promiscuously with MORF proteins in the yeast two-hybrid assays. The specific target site of MEF21 (cox3-257) also requires MORF1, and MEF21 indeed does interact strongly with the MORF1 protein. pGBKT7 is the bait and pGADT7 is the prey vector.

Fig. 5.

MORF and MEF proteins interact in pull-down assays. (A) In the MORF-MORF pull-down experiment, the GST-His-S-tag-MORF1-His-GFP protein was bound to Ni-NTA agarose beads (Right). A parallel bound GST-His-S-tag-GFP protein served as control (Left). MORF1 and MORF2 proteins tagged N-terminally with a maltose binding protein (MBP) extension and, as a control, MBP only, were added in separate assays, washed, released, spread on an SDS/PAGE gel, and visualized with an MBP antibody system. (Right) The MORF1 protein binds efficiently to the immobilized MORF1 protein. The MORF2 protein binds less effectively: A 20-fold–higher amount of input protein is required to obtain a signal of comparable intensity. (Left) Weak interactions of the MBP-MORF1 and MBP-MORF2 proteins to the GST-His-S-tag-GFP control are revealed; to make this background detectable, an approximately fivefold excess of the GST-His-S-tag-GFP control was loaded, as documented by Coomassie stain (CBB; Bottom). MBP protein alone is not detectably retained by either the GST-His-S-tag-MORF1-His-GFP protein or the GST-His-S-tag-GFP protein. Agarose beads (400 μL) were loaded with 3.5 nmol of GST-His-S-tag-MORF1-His-GFP protein or 35 nmol of GST-His-S-tag-GFP protein. Input protein was 0.5 nmol of MBP-MORF1, 10 nmol of MBP-MORF2, and 10 nmol of MBP. (B) For this MORF-MEF pull-down analysis, the GST-His-S-tag-MEF19-His-GFP protein (Center) or the GST-His-S-tag-MEF21-His-GFP protein (Right) were immobilized on glutathione agarose beads and probed for interaction with the MBP-fused MORF1 and MORF2 proteins. Retained MORF proteins were detected in the gel blot with an MBP antibody system. Comparison with the control glutathione agarose-bound GST-His-S-tag-GFP (Left) shows that the MBP-tagged MORF1 and MORF2 proteins do not bind detectably to the GST-His-S-tag-GFP protein when present in amounts comparable to the MEF19 and MEF21 proteins; the weak signals obtained with excess amounts of the control are shown in A. Both MEF19 and MEF21 are able to bind and retain the mitochondrially located MORF1 but not the plastid-targeted MORF2. This result confirms the interaction pattern seen in the yeast two-hybrid assays (Fig. 4), where MEF21 strongly interacts with MORF1 but only weakly with MORF2. The interactions observed between MORF1 and MEF19, and MORF1 and MEF21, agree with the RNA editing site analysis, with MORF1 and MEF19 and MORF1 and MEF21 targeting the same respective RNA editing sites in mitochondria. The agarose beads (400 μL) were loaded with 3.5 nmol of the GST-His-S-tag-MEF19-His-GFP, the GST-His-S-tag-MEF21-His-GFP, or the GST-His-S-tag-GFP protein. Input protein was 0.5 nmol of MBP-MORF1 or MBP-MORF2 and in the control 1 nmol of MBP. In the Coomassie stain (Bottom), not all partial MEF proteins that contain the N-terminal GST-His-S tag but not the C-terminal His-GFP tag are documented. The weak signal seen of free MBP retained by immobilized MEF21 in the input lane of MBP-MORF1 is either a much shorter bacterial translation product or a result of protein cleavage before or during the protein preparation from the bacteria.