Multiple PPR protein interactions are involved in the RNA editing system in Arabidopsis mitochondria and plastids

Significance

RNA editing is a posttranscriptional regulation process essential for organellar function. The PPR protein family plays a major role in RNA editing, but the precise function of each member and its interactions in editosome complexes remain unclear. This research demonstrates that three PPR proteins (SLO2, NUWA, and DYW2) interact in mitochondria and provides evidence for the following: (i) P-type PPR protein NUWA assists in the interaction between SLO2 and DYW2, and (ii) DYW2 participates in E⁺ editosomes. We hypothesize that this mechanism is common in both mitochondria and chloroplasts, where the E⁺-type PPR proteins, such as SLO2, target the specific RNA sites, with DYW2 suggested to provide the editing catalytic domain and NUWA assisting the interaction between the former ones.

Abstract

Recent identification of several different types of RNA editing factors in plant organelles suggests complex RNA editosomes within which each factor has a different task. However, the precise protein interactions between the different editing factors are still poorly understood. In this paper, we show that the E⁺-type pentatricopeptide repeat (PPR) protein SLO2, which lacks a C-terminal cytidine deaminase-like DYW domain, interacts in vivo with the DYW-type PPR protein DYW2 and the P-type PPR protein NUWA in mitochondria, and that the latter enhances the interaction of the former ones. These results may reflect a protein scaffold or complex stabilization role of NUWA between E⁺-type PPR and DYW2 proteins. Interestingly, DYW2 and NUWA also interact in chloroplasts, and DYW2-GFP overexpressing lines show broad editing defects in both organelles, with predominant specificity for sites edited by E⁺-type PPR proteins. The latter suggests a coordinated regulation of organellar multiple site editing through DYW2, which probably provides the deaminase activity to E⁺ editosomes.

The PPR family is classified into the P and PLS subfamilies, depending on the PPR signature domains. The P subfamily was described to participate in RNA stabilization, cleavage, splicing, and translation (1). Their members can bind specific RNA sequences to protect them from endonucleases or modify their secondary structure to recruit other RNA maturation factors (2). The P-type PPR protein PPME was the first one described with a role in RNA editing via RNA binding (3). The PLS subfamily is further divided into PLS, E, E⁺, and DYW subgroups, based on the C-terminal extensions and plays a major role in C-to-U RNA editing in plant organelles. Among them, E and E⁺-type PPR proteins do not have the DYW domain, which is considered to be a part of the catalytic domain due to the similarity to cytidine deaminase (2), suggesting the requirement of forming protein complexes with a DYW protein for a complete editing event. Indeed, the E-type PPR protein CRR4 forms a complex with DYW1, which consists of only a DYW domain, to edit an ndhD site in chloroplasts (4). However, such transassociated DYW-type PPR protein in an editing complex was reported only at this site in chloroplasts. It is still unclear whether this scenario is universal for numerous RNA editing sites in plant mitochondria. Not only the composition but the stoichiometry of the proteins in the editosome complexes remains unsolved (5). Some PPR proteins can homodimerize to bind RNA, although dimerization can also occur in the absence of the RNA target (6).

Most of the PPR proteins are located in mitochondria (65%) or chloroplasts (17%) (7). Systematic localization experiments and data integration showed that PPR dual targeting occurs more frequently than expected (7, 8). Interestingly, some heat shock proteins (HSPs) have also been associated with editing processes, translocating the editing factors into the required organelle (9).

Previously, we characterized the mitochondrial E⁺-type PPR protein SLO2, which participates in the editing of mttB-144, mttB-145, nad4L-110, nad7-739, mttB-666, nad1-2, and nad1-40 sites, affecting four proteins of complex I in the mitochondrial electron transport chain (10). However, SLO2 lacks a catalytic domain responsible for the editing task. In this report, we aimed to elucidate which proteins participate in the working mechanism of SLO2 by identifying its interacting partners.

Results

Immunoprecipitation with SLO2 Reveals PPR- and HSP-Type Candidate Interactors.

As a first approach to identify interacting partners of the E⁺-type PPR protein SLO2, we performed an immunoprecipitation and mass spectrometry (IP–MS) assay. Total protein extracts from Arabidopsis seedlings expressing a green fluorescent protein (GFP)-tagged SLO2 protein (SLO2-GFP), either under control of the SLO2 or the CaMV 35S promoter, were taken as starting material. Among the top 25 interactors, after selection of significant data (P value <0.05) and sorting based on the ratio of pSLO2::SLO2 to the Col-0 wild-type control (Fig. S1), three types of proteins were chosen: PPR proteins with a DYW domain (DYW2-At2g15690 and MEF57-At5g44230), P-type PPR proteins (NUWA-At3g49240, P486-At3g60980, and P487-At3g60960), and mtHSP60 chaperones (HSP60.2-At2g33210, HSP60.3A-At3g13860, and HSP60.3B-At3g23990), considered to be part of the import apparatus specific to the mitochondrial matrix (11) (Fig. 1). The limited number of PPR motifs in the DYW2 protein (Fig. 1), is likely to be insufficient for specific recognition of target RNAs and is reminiscent of the DYW1 protein in chloroplast (4).

Fig. S1.

Putative SLO2 interacting proteins. (A) Overview of the IP–MS results. The table shows an overview of the first 25 interactors with an indication of the ratio vs. WT and the P value obtained from the three repeats for each genotype used. A P value <0.05 was taken as cutoff value for the level of significance (Perseus statistical analysis). Significant data were ordered based on the pSLO2::SLO2-GFP ratio values and the resulting top 25 candidates are shown. Both IP experiments for pSLO2::SLO2-GFP and p35S::SLO2-GFP are shown with similar results. Red text indicates bait, green text indicates GFP, blue text indicates selected proteins. For row number 1, the actual bait protein is Q9SIT7 (red), but based on the identified peptides, MS analysis cannot distinguish between this protein and Q9FFG8 (blue). (B) Coverage of the detected peptides by IP–MS. The coverage from A is highlighted in gray in the respective protein sequences.

Fig. 1.

Scheme of the primary structure of SLO2 and its interacting partners. The names/ID and locus codes and the type of PPR protein are indicated on the left, and the sizes in base pairs (bp) and amino acids (aa) on the right side of each scheme. The coding/exon fragments are indicated with gray boxes. The different PPR protein motifs (13, 37) are indicated with colors and capital letters (L1, brown; L2, red; S, yellow; P, orange; E, light green; E⁺, dark green; and DYW, blue).

Several SLO2 Interacting Partners Are Uniquely Localized in Mitochondria; Others Are Dual Localized in Mitochondria and Chloroplasts.

As SLO2 is localized in mitochondria (10), its interactors are also expected to show mitochondrial localization. All of the SLO2 interacting partners were predicted to be localized in mitochondria by the SUBAcon program (12), except MEF57, predicted to be in plastids. For DYW2 and NUWA, proteomic data and reporter studies indicated dual targeting to mitochondria and chloroplasts (7, 8). NUWA mitochondrial localization was confirmed in Arabidopsis transgenic seedlings under the native promoter (13). In Fig. S2A, we show that the GFP signal of all GFP-tagged SLO2 candidate interactors consistently colocalized with the mitochondrial marker mitotracker/mt-rb in Nicotiana leaves, corroborating their mitochondrial localization, including for MEF57. In addition, a remarkably strong signal of NUWA-GFP was more frequently detected in chloroplasts (Fig. S2B) than in mitochondria. Likewise, DYW2-GFP showed strong chloroplast signals in several cells, in a dotted pattern (Fig. S2B), supporting dual localization for both proteins. The background signal at high laser gain, including chloroplast autofluorescence, is shown in Fig. S3A. HSP60 proteins, apart from the mitochondrial signal, also exhibited a signal associated with the cytosol (Fig. S2A).

Fig. S2.

Subcellular localization of SLO2 and its interacting partners MEF57, DYW2, P486, P487, NUWA, HSP60.2, HSP60.3A, and HSP60.3B. (A) N. benthamiana leaves infiltrated with SLO2-/MEF57-/DYW2-/P486-/P487-/NUWA-/HSP60.2-/HSP60.3A-/HSP60.3B-GFP constructs and analyzed by confocal microscopy 3 d after infiltration. Green and red fluorescences are indicative of the localization of the GFP protein and the mitotracker/mt-rb mitochondria markers, respectively. Representative individual cells of at least two independent experiments are shown, including their merged and light fields. Single cells were cropped from the original image. (Scale bar, 50 μm.) White arrows point to a mitochondrial signal dot. The contrast/brightness was adjusted for good visualization. (B) N. benthamiana leaves infiltrated with DYW2-/NUWA-GFP constructs and analyzed by confocal microscopy 3 d after infiltration, as indicated in A. Yellow arrows point to a chloroplast signal. A second image with lower laser gain is shown for the DYW2-GFP construct.

Fig. S3.

BiFC experimental setup. (A) Background of the GFP/YFP signal in the confocal microscopy analysis. N. benthamiana leaves infiltrated without any construct (negative control) and analyzed by confocal microscopy 3–6 d after infiltration. Green and red fluorescences are indicative of the localization of the reconstituted whole YFP protein (protein interaction) and the mitotracker/mt-rb mitochondria markers or mRFP-SKL peroxisome marker, respectively. Representative individual cells of at least two independent experiments are shown, including their merged and light fields. Single cells were cropped from the original image. (Scale bar, 50 μm.) The contrast/brightness was adjusted for good visualization. (B) Positive controls for protein interaction under BiFC experimental conditions. N. benthamiana leaves infiltrated with c/nYFP-APEM9 and n/cYFP-PEX6 constructs and analyzed by confocal microscopy 3 d after infiltration, as indicated in A. Purple arrows point to a peroxisomal signal dot. (C and D) Negative controls for protein interaction under BiFC experimental conditions. N. benthamiana leaves infiltrated with SLO2-/MEF57-/DYW2-/P486-/P487-/NUWA-/HSP60.2-/HSP60.3A-/HSP60.3B-nYFP (C) or SLO2-/MEF57-/DYW2-/NUWA-/HSP60.2-/HSP60.3A-/HSP60.3B-cYFP (D) together with the respective c/nYFP-APEM9 constructs and analyzed by confocal microscopy 3–6 d after infiltration, as indicated in A.

SLO2 and Its Binding Partners Interact in Vivo with HSP60 Import Factors in Mitochondria.

The suggested interaction between SLO2 and the mtHSP60 chaperones (Fig. S1), motivated us not only to confirm that these interactions occur in vivo in mitochondria, but also to check a possible interaction between the SLO2 binding PPR proteins and mtHSP60 import factors. All interactions were tested by bimolecular fluorescence complementation (BiFC). Both potential protein partners were tagged with the respective halves of the split-yellow fluorescent protein (YFP) protein and coinfiltrated pairwise in Nicotiana leaves. As summarized in Fig. 2A, the YFP signal was recovered when combining SLO2 or the selected PPR proteins with each of the mtHSP60 proteins, corroborating their interaction. No signal was recovered when combining them with ABERRANT PEROXISOME MORPHOLOGY 9 (APEM9) protein (14) (Fig. 2A and Fig. S3). The recovered YFP signal colocalized with the signal of the mitotracker/mt-rb mitochondrial marker (Fig. S4), proving that the interactions occur in mitochondria. To further substantiate the interaction of SLO2 and its binding PPR partners with the HSP60 chaperones, we coimmunoprecipitated human influenza hemagglutinin (HA)-/GFP-tagged SLO2, DYW2, or NUWA proteins together with GFP-/HA-tagged HSP60.2 and HSP60.3B proteins. Fig. 2B shows that HSP60.2 and HSP60.3B immunoprecipitated together with SLO2, DYW2, or NUWA, whereas APEM9 did not (Fig. S5A). Taken together, these results support the protein interaction of SLO2, DYW2, and NUWA with HSP60.2 and HSP60.3B in mitochondria.

Fig. 2.

Protein interaction of SLO2 and its partners MEF57, DYW2, P486, P487, and NUWA with mtHSP60 import factors. (A) Summary of the protein interactions between SLO2 or its partners and the HSP60 import factors in mitochondria, detected by BiFC assay in Figs. S3 B–D and S4. “Strong” and “weak” indicate the strength of the interactions according to the signal intensity under similar conditions. No, no interaction; n.c., not checked. (B) Total protein extracts (input) and GFP-immunoprecipitated proteins (GFP-IP) from N. benthamiana leaves infiltrated with SLO2-HA or HSP60.2-/HSP60.3B-HA together with HSP60.2-/HSP60.3B-GFP or DYW2-/NUWA-GFP constructs, 3 d after infiltration, analyzed by Western blot with anti-HA and anti-GFP antibodies (HA-Ab and GFP-Ab). N. benthamiana leaves infiltrated with the corresponding HA-tag constructs alone were taken as negative controls of the immunoprecipitation. Representative blots of at least two independent experiments for each combination are shown. Blots were cropped to the bands of interest (full-length blots in Fig. S5B). The contrast/brightness was adjusted for good visualization.

Fig. S4.

Interaction of SLO2 and its PPR partners MEF57, DYW2, P486, P487, and NUWA with mtHSP60 import factors in mitochondria. N. benthamiana leaves infiltrated with SLO2-/MEF57-/DYW2-cYFP or MEF57-/P486-/P487-/NUWA-nYFP together with HSP60.2-/HSP60.3A-/HSP60.3B-nYFP or HSP60.2-/HSP60.3A-/HSP60.3B-cYFP constructs, respectively, and analyzed by confocal microscopy 3 d after infiltration, as indicated in Fig. S3A. White arrows point to a mitochondrial signal dot.

Fig. S5.

CoIP experiment setup. (A) Negative controls for protein interaction under CoIP experimental conditions. Total protein extracts (input) and GFP-immunoprecipitated proteins (GFP-IP) from N. benthamiana leaves infiltrated with SLO2-HA or HA-APEM9 together with GFP-APEM9 or DYW2-/NUWA-GFP constructs, 3–6 d after infiltration, analyzed by Western blot with anti-HA and anti-GFP antibodies (HA-Ab and GFP-Ab). N. benthamiana leaves infiltrated with the corresponding HA-tag constructs alone were taken as negative controls of the immunoprecipitation. Representative blots of at least two independent replicates for each combination are shown. Blots were cropped to the bands of interest (full-length blots in B). The contrast/brightness was adjusted for good visualization. (B) Full-length blots of the CoIP experiments. Both antibodies were checked consecutively on the same membrane (first anti-GFP, second anti-HA) and pictures after each antibody checking are shown. Black arrows point at the bands of interest for each antibody. White squares hide nonrelevant runs/samples for visual and interpretation clarity. The contrast/brightness was adjusted for good visualization. Size markers are as follows: 170, 130, 100, 70 (red), 55, 40, 35, 25, and 15 kDa. The respective molecular weights are as follows: SLO2-HA, 84.03 kDa; SLO2-GFP, 105.04 kDa; DYW2-GFP, 92.56 kDa; MEF57-GFP, 100.62 kDa; NUWA-HA, 77.23 kDa; NUWA-GFP, 98.24 kDa; HSP60.2-HA, 67.92 kDa; HSP60.2-GFP, 88.93 kDa; HSP60.3B-HA, 66.41 kDa; HSP60.3B-GFP, 87.42 kDa; HA-APEM9, 43.41 kDa; and GFP-APEM9, 64.42 kDa.

SLO2 Interacts in Vivo with DYW2 and NUWA Proteins in Mitochondria.

Next we verified by BiFC whether the interaction of SLO2 with the P- and DYW-type PPR proteins detected by immunoprecipitation occurs in mitochondria. As presented in Fig. 3A, the recovery of the YFP signal when combining SLO2 with each PPR protein, confirmed the physical interaction of SLO2 with all of the tested PPR proteins. The colocalization of the recovered YFP signal with the mitotracker/mt-rb mitochondrial marker (Fig. S6A) certified that the interaction takes place in mitochondria. However, the recovery of the YFP signal was observed with a low frequency (less than 10% of the cells) and low intensity, except when combining SLO2 with NUWA (Fig. 3A and Fig. S6A). Because BiFC analyses tend to also detect rather weak and transient interactions, to corroborate the stable interaction between SLO2 and the DYW proteins or the P-type PPR protein NUWA, we coimmunoprecipitated tagged DYW proteins (MEF57-GFP and DYW2-GFP) and P-type PPR protein (NUWA-HA), together with tagged SLO2 (SLO2-HA or SLO2-GFP). As shown in Fig. 3B and Fig. S5A, SLO2 was immunoprecipitated together with all of the candidates tested, except with MEF57 and APEM9 control. These results endorse the specific interaction of SLO2 with all of the candidates tested, except for MEF57.

Fig. 3.

Protein interaction between SLO2 and its PPR interacting partners MEF57, DYW2, P486, P487, and NUWA. (A) Summary of the protein interactions between SLO2, its DYW- and P-type PPR partners in mitochondria, detected by BiFC assay in Figs. S3 B–D and S6, as indicated in Fig. 2A. (B) Total protein extracts (input) and GFP-immunoprecipitated proteins (GFP-IP) from N. benthamiana leaves infiltrated with SLO2-HA together with MEF57-/DYW2-GFP, or NUWA-HA together with SLO2-/DYW2-GFP, or infiltrated with SLO2-HA together with DYW2-GFP or NUWA-His, or the combination of the three constructs. Proteins were extracted 3 d after infiltration and analyzed by Western blot as indicated in Fig. 2B (full-length blots in Fig. S5B). (C) N. benthamiana leaves infiltrated with SLO2-nYFP together with MEF57-/DYW2-cYFP, in the absence or presence of NUWA-HA construct, analyzed by confocal microscopy 3–6 d after infiltration. Green fluorescence is indicative of the localization of the reconstituted whole YFP protein (protein interaction). Representative individual cells of at least three independent experiments are shown. Single cells were cropped from the original image. (Scale bar, 50 μm.) Images for combinations without NUWA-HA (Left), with really low fluorescent signal, were taken at a higher laser gain condition, whereas the images where NUWA-HA was added (Right) were taken under lower gain conditions. For a proper comparison between proteins, images for both DYW2 and MEF57 proteins were taken under the same conditions. Images were extracted from Fig. S6A.

Fig. S6.

Interaction between SLO2 and its PPR interacting partners. (A) Interaction of SLO2 with MEF57, DYW2, P486, P487, and NUWA in mitochondria. N. benthamiana leaves infiltrated with SLO2-nYFP together with MEF57-/DYW2-cYFP in the absence or presence of HSP60.2-/HSP60.3A-/HSP60.3B-/P486-/P487-HA or NUWA-HA constructs, or with SLO2-cYFP together with P486-/P487-/NUWA-nYFP constructs, analyzed by confocal microscopy 3–6 d after infiltration, as indicated in Fig. S3A. White arrows point to a mitochondrial signal dot. Images for combinations without NUWA-HA or NUWA-nYFP, with really low fluorescent signal, were taken at a higher laser gain condition. For a proper comparison between proteins, images for both DYW-type PPR proteins (MEF57 and DYW2) were taken under the same conditions. For the combinations in the presence of HSP60.2-/HSP60.3A-/HSP60.3B-/P486-/P487-HA constructs, the SLO2-nYFP/MEF57-cYFP + HSP60.3A-HA and the SLO2-nYFP/DYW2-cYFP + P486-HA images are shown as representative of all of the listed combinations. (B) Interaction of the SLO2 binding partners MEF57 and DYW2, with P486, P487, and NUWA in mitochondria. N. benthamiana leaves were infiltrated with MEF57-/DYW2-cYFP together with P486-/P487-/NUWA-nYFP constructs, analyzed by confocal microscopy 3 d after infiltration, as indicated in Fig. S3B. White arrows point to a mitochondrial signal dot and yellow arrows to a chloroplast signal. Images for combinations without NUWA-nYFP, with really low fluorescent signal, were taken at a higher laser gain condition. For a proper comparison between proteins, images for both DYW-type PPR proteins (MEF57 and DYW2) were taken under the same conditions.

NUWA Enhances the Interaction Between SLO2 and DYW2 in Mitochondria in Vivo.

Because some HSPs have been described to enhance mRNA or protein stability (15–17), we investigated whether the three SLO2 associated HSP60 proteins, interacting also with both DYW proteins (Fig. 2A), could stabilize the DYW proteins or the SLO2/DYW complexes, directly or through binding their target RNAs. Such stabilization could result in an increased interaction signal between SLO2 and the DYW proteins. To this end, we tested the interaction by BiFC in the presence and absence of the HSP60 partners (HSP60-HA), in parallel experiments, in different halves of the same Nicotiana leaf. The addition of HSP60 chaperones did not result in an increased interaction signal (Fig. S6A).

P-type PPR proteins also have been described to possess RNA or complex stabilization properties (2). Therefore, we checked whether the addition of the P-type PPR proteins detected as SLO2-interacting partners could improve the interaction between SLO2 and the DYW proteins. Interestingly, a clear increase in the SLO2 interaction with DYW2, appreciated by a high frequency (in around 90% of the cells) and intensity of the recovered YFP signal in parallel experiments, was detected in mitochondria in the presence of the P-type PPR protein NUWA (NUWA-HA) (Fig. 3 A and C and Fig. S6A), whereas coexpression of the other two P-type PPR proteins did not enhance the interaction (Fig. S6A). The increase in the interaction of SLO2 with DYW2 in the presence of NUWA (NUWA-His) was corroborated by coimmunoprecipitation (Fig. 3B), as reflected by a more intense SLO2-HA signal in the GFP immunoprecipitate (GFP-IP) compared with the input signal. The addition of NUWA, on the other hand, did not result in an enhanced signal for the SLO2/MEF57 interaction (Fig. 3 A and C and Fig. S6A).

DYW2 and NUWA Proteins Interact in Mitochondria and Chloroplasts.

To elucidate whether the specific enhancement of the SLO2/DYW2 interaction by NUWA was conferred through their direct protein interactions, we compared affinities between each P-type PPR protein and the two DYW-type PPR proteins by BiFC assay. Although all three P-type PPR proteins could interact with both DYW-type PPR proteins in mitochondria, NUWA showed significantly stronger BiFC signal than the other two P-type PPR proteins (Fig. 3A and Fig. S6B). Interestingly, the interaction between NUWA and DYW2 took place mainly in mitochondria but, in some cells, the signal was detected in chloroplasts (Fig. S6B). This result is in agreement with the identification of NUWA and DYW2 as interactors of the chloroplastic E⁺-type PPR CLB19 (18). Furthermore, the interaction between NUWA and DYW2 proteins was corroborated by coimmunoprecipitation (Fig. 3B and Fig. S5A).

DYW2 and NUWA Proteins Form Homomers.

We further investigated the capacity of SLO2 and its interacting PPR partners to homomerize by BiFC assay. Only the PPR proteins DYW2 and NUWA showed the capacity to form homomers (Fig. 4). Moreover, whereas NUWA homodimers were formed in both mitochondria and chloroplasts, DYW2 homomerization was restricted to chloroplasts (Fig. 4).

Fig. 4.

Dimerization of DYW2 and NUWA. N. benthamiana leaves infiltrated with DYW2-/NUWA-nYFP together with DYW2-/NUWA-cYFP constructs and analyzed by confocal microscopy 3 d after infiltration. Green and red fluorescence are indicative of the localization of the reconstituted whole YFP protein (protein interaction) and the mitotracker mitochondrial marker, respectively. Representative individual cells of at least two independent experiments are shown, including their merged and light fields. Single cells were cropped from the original image. (Scale bar, 50 μm.) White arrows point to a mitochondrial signal dot and yellow arrows to a chloroplast signal. The contrast/brightness was adjusted for good visualization.

DYW2 Is Involved in the Editing of the E⁺-Type PPR Edited Sites.

Finally, to clarify the biological relevance of the SLO2/DYW2 interaction, because we could not analyze homozygous dyw2 knockout lines due to their embryo lethality (18), we generated Arabidopsis plants overexpressing the GFP-tagged DYW2 gene (Fig. S7). Two independent lines, containing at least fourfold higher levels of DYW2/DYW2-GFP transcripts than the DYW2 transcripts in wild type (Fig. S7A), were isolated. They showed a slow growth phenotype (Fig. 5 A and B), reminiscent of other PPR loss-of-function mutants. Editing analysis revealed a high number of affected editing sites, mainly down-edited, and some completely (100%) affected (Dataset S1). In particular, the E⁺-type PPR SLO2 edited sites (10), including one identified in this work (Fig. S8), were down-edited (Dataset S1). Interestingly, this down-editing was also the case for most of the sites described to be edited by other E⁺-type PPR proteins, but only for a few ones edited by DYW- or E-type PPRs, both in mitochondria and chloroplasts (Fig. 5 C and D). This result suggests that DYW2-mediated editing events are basically linked to E⁺-type PPR proteins. Interestingly, as shown in Fig. S7B, the NUWA expression levels were higher in these DYW2-GFP overexpressing (DYW2-GFP-OE) lines compared with the wild-type plants, indicative of coordinated regulation. The work of Guillaumot et al. (18) on dyw2 and nuwa loss-of-function mutants (18) also revealed a clear editing defect in most of the E⁺-type PPR edited sites, including the SLO2 sites, corroborating a predominant role of DYW2 and NUWA proteins in E⁺-type PPR-associated editing. Moreover, some sites not described to be regulated by known PPR proteins were also altered in the DYW2-GFP-OE lines (Dataset S1). This result points toward a hub function for DYW2; however, further investigation will be needed to establish a clear connection between DYW2 and non-PPR editing factors. The knockout line of the other DYW-type PPR partner of SLO2, MEF57 (Fig. S9), shows RNA editing defect at the nad9-92 site (Fig. S8), but no alteration at the SLO2 edited sites.

Fig. S7.

DYW2 and NUWA expression levels in DYW2-GFP-OE lines. Total RNA from rosette leaves of WT and two independent DYW2-GFP-OE lines analyzed by RT-qPCR with specific primers for DYW2 (A) and NUWA (B). The relative expression levels with respect to three reference genes (Actin 2, Elongation factor 1 ALPHA, and RGS1-HXK1 INTERACTING PROTEIN 1) are shown. The columns and error bars represent the mean and SD values, respectively, of three biological replicates (represented by dots).

Fig. 5.

Characterization of DYW2-GFP-OE lines. (A and B) Morphology of adult plants. Images of adult plants from the homozygous DYW2-GFP-OE2 line and the WT control, grown in parallel (A), and a detail of the morphology in two independent heterozygous DYW2-GFP-OE lines (B). Representative images of at least two independent experiments are shown. (C and D) RNA editing analysis of DYW2-GFP-OE lines. Editing analysis of rosette leaves from WT and two independent DYW2-GFP-OE lines are shown. Each independent line was analyzed once. The media and SD of the differences in editing percentage in both DYW2-GFP-OE lines with respect to the WT (Δ%) were calculated (Dataset S1). The number of sites modified by a particular type of PPR protein is represented. Both mitochondrial and chloroplastic sites are shown, split into down-edited (C) and nonaffected (D). A value of Δ% = 24% was taken as cutoff. The consistent data in both OE lines are shown. The list of PPR associated with editing sites was formed according to the literature (2, 20, 38) and this work.

Fig. S8.

RNA editing analysis of slo2 and mef57 mutants. (A) Editing analysis of seedlings from WT, slo2-2, slo2-3, complemented slo2-2 and complemented slo2-3 lines, for the nad1-743 site. The sequencing chromatograms are shown. The editing site is marked with a square. The amino acid change is indicated on the left. The WT line was analyzed twice and each independent mutant line once. (B) Editing analysis of WT, mef57, and complemented mef57 lines, for the nad9-92 site, as indicated in A. Seedlings, leaves, and siliques of each line were analyzed once, with similar results.

Fig. S9.

Isolation of a homozygous T-DNA insertion mef57 line and the respective complemented line. (A) Scheme of the primary structure of MEF57 and T-DNA insertion site. The ID code is indicated on the Left and the sizes in base pairs (bp) and amino acids (aa) on the Right. The different PPR protein motifs are indicated with colors and capital letters (L1, brown; L2, red; S, yellow; P, orange; E, light green; E⁺, dark green; DYW, blue). The T-DNA insertion site is indicated with a big white triangle, with the name of the corresponding mutant line inside and the distance to the start codon in base pairs outside. The position and name of the primers used for genotyping and expression analyses are indicated with continuous and dashed arrows, respectively. (B) Genotyping of the mef57 and the complemented mef57 lines. gDNA from leaves analyzed by standard PCR with specific primers, as indicated. The contrast/brightness was adjusted for good visualization. (C) MEF57 expression levels in mef57 and the complemented mef57 lines. Total RNA from siliques of WT, mef57, and complemented mef57 lines were analyzed by RT-qPCR with MEF57 specific primers. The relative expression levels with respect to two reference genes (Elongation factor 1 ALPHA and RGS1-HXK1 INTERACTING PROTEIN 1) are shown. The columns and error bars represent the mean and SD values, respectively, of three biological replicates (represented by dots).

Discussion

PPR Editing Complexes.

To date, the interactions between PPR proteins were restricted to the chloroplastic PPR protein DYW1, which can function in trans with the E-type PPR protein CRR4 (4). In this paper, we present evidence for the interaction between three types of PPR proteins (E⁺, DYW, and P type) in mitochondria (SLO2, DYW2, and NUWA) (Figs. 1 and 3). The high NUWA expression levels in the DYW2-GFP-OE lines (Fig. S7B) reinforces the idea that NUWA is required by DYW2 to form a functional editosome with other editing factors. In addition, the dual localization and interaction of the DYW2 and the NUWA partner proteins (Figs. S2 and S6) suggest a coordinated regulation of organellar editing, supported by the identification of NUWA and DYW2 as interactors of the chloroplastic E⁺-type PPR protein CLB19 (18) and the broad effect of DYW2 in E⁺-type PPR edited sites (Fig. 5) (18). Taken together, we propose that E⁺-type PPR proteins target the specific RNA sites, whereas P-type PPR protein NUWA assists the interaction between the E⁺ partner and DYW2. Based on the current evidence (5), we hypothesize that DYW2 would provide the editing catalytic domain, both in mitochondria and chloroplasts. A model for the proposed PPR editosome mechanism is shown in Fig. 6.

Fig. 6.

Model of the tripartite PPR protein interaction in editosomes in mitochondria and chloroplasts.

Although the interaction between SLO2 and MEF57 detected by IP–MS could be confirmed by BiFC, it was not corroborated by coimmunoprecipitation (CoIP) (Fig. 3). The immunoprecipitation was realized from stable transgenic Arabidopsis plants, whereas the coimmunoprecipitation and BiFC assays were performed on transiently transformed Nicotiana leaves. These differences suggest the possibility that a factor present in Arabidopsis but not in Nicotiana could assist in the SLO2/MEF57 interaction. However, the possibility of absence of MEF57/SLO2 interaction would also be plausible. Indeed, the RNA editing analysis suggests that MEF57 protein is not required for the SLO2 editing ability (Fig. S8).

The observed down-editing effect in the DYW2-GFP-OE lines (Fig. 5 and Dataset S1) could likely be explained by a dominant negative effect of an inappropriate stoichiometry of the complex (19) or through capture of DYW2 interacting factors required by other editing complexes that target the same sites or by competition with other factors for the same editing sites (20–22). Furthermore, it was reported that the addition of a C-terminal tag to DYW proteins could inhibit their editing function (23, 24), while allowing interaction with other editing partners (4, 25–27). This might lead to a nonfunctional editosome, explaining the down-editing effect in these overexpressing lines. The observed DYW2-GFP-OE phenotype (Fig. 5) could be explained by the broad editing defect that would affect the organellar functions, impairing plant growth. This broad defect in the essential RNA editing process, would justify the lethality of the homozygous dyw2 loss-of-function mutants (18).

Dual Localization of PPR Proteins.

The PPR proteins DYW2 and NUWA have dual localization in both mitochondria and chloroplasts (Fig. S2). However, the protein characteristics and mechanisms responsible for this dual localization remain unclear. A significant number of proteins are dual targeted to both organelles via ambiguous targeting signals (28). It has been proposed that cytosolic chaperones may play a role in determining targeting specificity (29). In Arabidopsis, the mitochondrial HSP60 family has three members, considered to be part of the import apparatus specific to the mitochondrial matrix (30). We would expect that mutants of the interacting HSP60 proteins in this work affect RNA editing. However, if the three mtHSP60 proteins have any redundant function, a triple mutant would be needed to observe such editing defects.

PPR mRNA and Protein Complex Stabilization.

The increased interaction of SLO2 and DYW2 when adding NUWA (Fig. 3) and the protein interaction of NUWA with both SLO2 and DYW2 (Fig. 3) suggest a possible scaffold or complex stabilization role of NUWA for the interaction between SLO2 and DYW2 in the editosome. In this context, a non-PPR editing factor (MORF8) was described to assist the interaction between an E⁺-type PPR protein (MEF13) and another non-PPR factor (MORF3) (31). The fact that the P-type PPR protein NUWA assists in the interaction between an E⁺ (SLO2) and a DYW (DYW2) protein, could reflect a more general mode of action of PPR proteins, hence representing the key for the discovery of similar tripartite PPR interactions and shedding light on the complex PPR editosome formation and action mechanisms.

Our results suggest that HSP60 proteins do not participate in the protein complex stabilization (Fig. S6A). They may assist in mitochondrial import of the partner proteins, as previously mentioned, or support RNA binding of the editing complexes. The chaperone HSP60 was suggested to have a highly specific nucleic acid-binding activity, presumably implicated in mtDNA stability (32). In protozoa, it has been demonstrated that a chaperone activity in the editosome increases the flexibility of U residues in the pre-mRNAs to facilitate the binding of gRNAs (33), providing a rational explanation for the U specificity of the editing reaction. P-type PPR subfamily members can also bind to specific RNA sequences (2), protecting the RNA from endonucleases and/or modifying its secondary structure to recruit general factors involved in RNA maturation. In this context, it will be interesting to check RNA binding activity of the HSP60 and P-type PPR proteins detected in this work.

Further investigation will be needed to understand the precise function of SLO2, DYW2, NUWA, and the HSP60 proteins in the SLO2 editosome complex and beyond.

Materials and Methods

IP–MS Assay.

The IP–MS assay was performed as previously described (34), without modifications. Three biological replicates were used for each line.

BiFC.

The predicted subcellular localization was analyzed by SUBAcon and SUBA3 programs (12, 35). For experimental subcellular localization and BiFC, the corresponding binary vectors were introduced into Agrobacterium tumefaciens strain GV3101. Agrobacterium was used to transiently transform young leaves of Nicotiana benthamiana grown on soil under long day conditions. For details, see SI Materials and Methods.

CoIP.

A modified μMACS Epitope Tag Protein Isolation Kit (Miltenyi) protocol (34) was followed. For details, see SI Materials and Methods.

T-DNA Insertion, Complemented, and Overexpressing Lines.

Arabidopsis thaliana T-DNA insertion mutant line N585176 from the SALK collection was used and complemented with the 35S::MEF57-GFP construct. The DYW2-GFP-OE lines were generated transforming Col-0 A. thaliana plants with the 35S::DYW2-GFP construct.

RNA Editing Analysis.

Specific cDNAs were generated as described previously (36). The sequence data for each gene-specific RT-PCR product were obtained commercially and compared for analysis (Macrogen). Ratios between heights of C and T signals were calculated with the DNADynamo software (BlueTractorSoftware). The Arabidopsis slo2-2, slo2-3, complemented slo2-2, and complemented slo2-3 lines were taken from our previous work (10).

A detailed description of the plasmid construction and full methods are described in SI Materials and Methods.

Data Availability.

All data generated or analyzed during this study are included in this article (and Supporting Information).

SI Materials and Methods

Plant Materials and Growth Conditions.

Arabidopsis thaliana ecotype Col-0 and Nicotiana benthamiana were used. Seeds were stratified for 2 d at 4 °C. Growth conditions were a 16-h light/8-h dark photoperiod and white fluorescence light (75 μM m^–2⋅s^–1) at 22 °C. For growing plants in soil, Jiffy-7 peat pellets were used. For growing seedlings on plates, seeds were surface sterilized with 30% bleach and 0.02% Triton X-100 and germinated on 1/2 MS medium supplemented with 0.5 g/L Mes, the pH adjusted to 5.7 with KOH, 10 g/L sucrose, and 8 g/L agar.

Plasmid Construction.

The complete coding sequences of interest were obtained from A. thaliana cDNA by HiFi PCR using attB-flanked primers (Dataset S2), which introduced the adequate sites for the subsequent cloning by the Gateway system (Invitrogen) into the pDONR221 vector via BP reaction and were confirmed by sequencing. The At5g44230 (MEF57) coding sequence has a single nucleotide mutation at position 1089, but it does not change any amino acid in the protein sequence.

The verified fragments were cloned under the control of the constitutive CaMV 35S promoter via LR reaction into pK7WGF2 vector for C-terminal fusion with GFP; pB4GWnY and pB4GWnY vectors [kindly provided by Shoji Mano (39)] for C-terminal fusion with the n- or c-split-YFP; pGWB514 or pGWB508 vector [kindly provided by Tsuyoshi Nakagawa (40)] for C-terminal fusion with 3×HA epitope or His-tag. The SLO2 coding sequence was cloned also under the control of its own promoter as described previously (10). The mitochondria-mCherry marker plasmid [mt-rb (41)] was used as mitochondrial marker. The pB4nYFPPEX6 and pB4cYFPAPEM9 plasmids [kindly provided by Tsuyoshi Nakagawa (40)] were used as BiFC controls.

Subcellular Localization and BiFC.

Two-dimensional cultures of Agrobacterium tumefaciens harboring the desired constructs and the silencing-suppressor p19 plasmid (42) were pelleted, diluted to an OD₆₀₀ of 0.1–0.5 in infiltration buffer [0.5% (wt/vol) d-glucose, 10 mM Mes, 10 mM MgCl2, 0.1 mM acetosyringone] and coinfiltrated in young N. benthamiana leaves. At 3–6 d after infiltration, small pieces were excised and visualized under confocal imaging. The MitotrackerOrange CMTMRos dye (Invitrogen) at 250 nM in 1/2 MS liquid medium was infiltrated 4 h before imaging to stain mitochondria. Plasmids carrying the mitochondria-mCherry marker {mt[29 aa ScCOX4]-mCherry-rb (41)} and the mRFP-SKL marker for mitochondria and peroxisomes, respectively, were coinfiltrated together with the constructs of interest.

Leaves infiltrated without GFP/YFP constructs were used as a negative control to reveal the background signal with a high laser gain (Fig. S3A), visualizing chloroplast autofluorescence as well as cellular structure. Samples with sufficiently strong fluorescent signal were assessed under reduced laser gain conditions, thus avoiding chloroplast autofluorescence or signal saturation. Although under these conditions detection of weaker signals can be lost, such as the mitochondrial signal, it allows for identifying strong signals in the chloroplasts or clear strong mitochondrial signals without saturation.

The interacting APEM9 and PEX6 proteins (14) were used in the same system as a positive control for the interaction as YFP signal recovery (Fig. S3B). The combination between each of the constructs and the protein APEM9 (14) was used as a negative control of the interactions (Fig. S3 C and D).

For weak signals, such as the one from the combination MEF57/HSP60.2 and MEF57/HSP60.3B, the interaction was also tested by tagging the proteins with the other half of the split YFP protein.

The subcellular localization experiments were repeated, with similar results, three independent times (agro-infiltrations), except for HSP60.3A-GFP, which was repeated twice.

The protein interaction experiments by BiFC were repeated, with similar results, three independent times (agro-infiltrations), except for combinations with HSP60.3A-/P486-/P487-nYFP and the combinations NUWA-nYFP/HSP60-cYFP, MEF57-cYFP/NUWA-nYFP, cYFP-APEM9/NUWA-nYFP, nYFP-APEM9/NUWA-cYFP, and DYW2-nYFP/DYW2-cYFP, which were repeated twice.

CoIP and Western Blot.

Plant material expressing the constructs of interest (transiently transformed N. benthamiana) was grinded in extraction buffer [50 mM Tris⋅HCl pH 7.5, 150 mM NaCl, 1% Nonidet P-40 (Igepal CA630), 2× protease inhibitor mix mixture without EDTA (EDTA-free Complete, Roche), 1× phosphatase inhibitor (PhosSTOP, Roche), 50 µM MG132 proteasome inhibitor (C2211, Sigma), 5 mM ATP and 1 mM PMSF]. Nonidet P-40 was diluted to 0.2% by adding extraction buffer without Nonidet P-40 before filtering with 40 µM nylon mesh and centrifuging 1 min at 12,000 rpm (5417R centrifuge, F45-30-11 rotor; Eppendorf, Hauppauge, New York) at 4 °C. A total of 50 μL anti-GFP magnetic Microbeads (Miltenyi Biotec REF 130–091-125) was added and incubated for 2 h under rotation at 4 °C. After equilibrating the μ-column (in the magnetic field) with 200 μL extraction buffer containing 0.1% Nonidet P-40, the cell lysate was applied. The column was washed four times with 200 μL extraction buffer with 0.1% Nonidet P-40 and once with 20 mM Tris⋅HCl, pH 7.5. The coimmunoprecipitated protein was eluted with 50 μL elution buffer preheated at 95 °C [50 mM Tris HCl pH 6.8, 50 mM DTT, 1% SDS, 1 mM EDTA, 0.005% bromophenol blue, 10% glycerol] after incubating the column with 20 μL of preheated elution buffer for 5 min at room temperature. The coimmunoprecipitated protein samples were incubated with elution buffer or sample buffer [250 mM Tris⋅HCl, pH 6.8, 20% glycerol, 4% SDS, 10% β-mercaptoethanol, and 0.025% bromophenol blue] for 5 min at 65 °C. Thirty microliters of the coimmunoprecipitated protein samples was run in a 8% SDS/PAGE gel and blotted on a 0.45-µm nitrocellulose membrane (Amersham Protran) at 4 °C, which was probed with 1:100 anti–HA-HRP (clone 3F10 Roche) or 1:100 anti–GFP-HRP (A10260 Invitrogen) in T-TBS buffer (20 mM Tris⋅HCl, pH 7.5, 0.5 M NaCl, 0.1% Tween-20) and developed by colorimetric assay (0.5 mg/mL DAB in 0.1 M imidazole pH 7 with 0.1 µl/mL 30% H₂O₂ and 54 µl/mL 0.6% CoCl₂). Both antibodies were checked consecutively on the same membrane, with a long washing step (T-TBS buffer for 30 min) in between. The PageRuler prestained protein ladder was used as size marker of 170, 130, 100, 70 (red), 55, 40, 35, 25, and 15 kDa (Thermo Scientific, CN26616). The respective molecular weights were: SLO2-HA, 84.03 kDa; SLO2-GFP, 105.04 kDa; DYW2-GFP, 92.56 kDa; MEF57-GFP, 100.62 kDa; NUWA-HA, 77.23 kDa; NUWA-GFP, 98.24 kDa; HSP60.2-HA, 67.92 kDa; HSP60.2-GFP, 88.93 kDa; HSP60.3B-HA, 66.41 kDa; HSP60.3B-GFP, 87.42 kDa; HA-APEM9, 43.41 kDa; and GFP-APEM9, 64.42 kDa. Full-length blots, including the size marker, are shown in Fig. S5B.

As a negative control of the coimmunoprecipitation (protein interaction), we used a combination of SLO2 (SLO2-HA), DYW2 (DYW2-GFP), or NUWA (NUWA-GFP) with an N-terminus GFP- or HA-tagged APEM9 (GFP-APEM9 or HA-APEM9). As negative controls of the immunoprecipitation (binding to the antibody beads), N. benthamiana leaves infiltrated only with the corresponding HA-tag construct were used.

A positive interaction is shown by a higher signal of the HA constructs in the GFP immunoprecipitate (GFP-IP) with respect to the input signal, when combining the HA and GFP constructs compared with the negative controls for the immunoprecipitation/binding to the antibody beads (HA construct alone). A negative interaction is indicated by a lower or practically equal signal.

The coimmunoprecipitation experiments were repeated, with similar results, three independent times with a different biological replicate (agro-infiltration) each, for the combinations SLO2-HA/MEF57-GFP and SLO2-HA/DYW2-GFP; for the rest, the experiment was repeated two times with a different biological replicate each, except for the combinations SLO2-GFP/NUWA-HA and HA-APEM9/NUWA-HA, which were performed once with two biological replicates.

T-DNA Insertion, Complemented, and Overexpressing Lines.

A. thaliana T-DNA insertion mutant line N585176 from the SALK collection was genotyped from genomic DNA of adult plant leaves by standard PCR with specific primers for the T-DNA and the gene of interest (Dataset S2). The PCR products were visualized on ethidium bromide-stained 2% agarose gel (Fig. S9). The absence of a second T-DNA insertion indicated in the database was confirmed also by standard PCR, after selection for the T-DNA insertion of interest through several generations. The loss of expression of the corresponding gene was checked from RNA by quantitative PCR (qPCR) (Fig. S9).

The homozygous T-DNA insertion mutant line or Col-0 A. thaliana plants were transformed with the 35S::MEF57-GFP and 35S::DYW2-GFP plasmids (in pK7WGF2 backbone), respectively, by the floral-dip method (43). Transformed plants were selected based on Kanamycin resistance. Two independent DYW2-GFP-OE lines were isolated from independent transformation experiments. The presence of the insert was confirmed by standard PCR on genomic DNA with specific primers (Dataset S2) and the PCR products were visualized on ethidium bromide-stained 2% agarose gel (Fig. S9). The expression of the corresponding genes was checked from RNA by qPCR (Figs. S7 and S9).

Quantitative PCR.

Total RNA was extracted from adult plant leaves, seedlings, or siliques, as indicated, using an RNeasy plant mini kit (Qiagen). RNA was quantified by UV spectrophotometry (NanoPhotometer, IMPLEN) and its integrity was visually assessed on ethidium bromide-stained 2% agarose gels. Treatment with DNase I Amplification Grade (DNase I, Amp Grade, Invitrogen) was performed.

RNA was reverse transcribed to cDNA by the RNase H+, iScript cDNA synthesis kit (Bio-Rad). Real-time qPCR was performed in a iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad) with the iQ5 Optical System Software Version 2.0 (Bio-Rad). Specific primers for the genes of interest and three reference genes (Dataset S2) were used. Three biological replicates were taken for each plant line, and three technical repeats were performed for each biological replicate. The data were analyzed with the qbasePLUS2 software (Biogazelle).

Equipment and Settings.

The blot images were taken with a Motorola MotoG3 camera at 2340 × 4160 pixel resolution and 24 bits per pixel. Images were cropped to the bands of interest using PowerPoint. The entire blot images (including the protein size marker) are shown in Fig. S5B. Images of adult plants were taken with the same camera and settings and cropped to the area of interest.

The confocal images were taken from a Nikon ECLIPSE TE2000-S microscope, with fixed spectrum filters, with a Nikon D-ECLIPSE C1 confocal scanhead and the software NIS Element 4.10 from Nikon. The GFP and YFP fluorescences were detected with excitation at 488 nm and emission at 500–530 nm; the red fluorescence from the MitotrackerOrange CMTMRos dye (Invitrogen), the mitochondria-mCherry marker plasmid {mt[29 aa ScCOX4]-mCherry-rb (41)} and the mRFP-SKL marker plasmid (44) were detected with excitation at 552 nm and emission at 593–640 nm. Pictures were taken with a 40× WI objective at 1024 × 1024 pixel resolution (12 bit). The LUT is linear and covers the full range of the data. Single fluorochrome images were merged with the Adobe Photoshop program. Images were cropped to a single representative cell with the PowerPoint program.

Statistical Analysis.

For the IP–MS experiment, a standard two-sided t test with false discovery rate = 0.05 and S0 = 1 was performed with Perseus software (34). Number of biological replicates was n = 3, for each line. No samples were discarded as outliers.