Maize PPR-E proteins mediate RNA C-to-U editing in mitochondria by recruiting the trans deaminase PCW1

Abstract

RNA C-to-U editing in organelles is essential for plant growth and development; however, the underlying mechanism is not fully understood. Here, we report that pentatricopeptide repeat (PPR)-E subclass proteins carry out RNA C-to-U editing by recruiting the trans deaminase PPR motifs, coiled-coil, and DYW domain-containing protein 1 (PCW1) in maize (Zea mays) mitochondria. Loss-of-function of bZIP and coiled-coil domain-containing PPR 1 (bCCP1) or PCW1 arrests seed development in maize. bCCP1 encodes a bZIP and coiled-coil domain-containing PPR protein, and PCW1 encodes an atypical PPR–DYW protein. bCCP1 is required for editing at 66 sites in mitochondria and PCW1 is required for editing at 102 sites, including the 66 sites that require bCCP1. The PCW1-mediated editing sites are exclusively associated with PPR-E proteins. bCCP1 interacts with PCW1 and the PPR-E protein Empty pericarp7 (EMP7). Two multiple organellar RNA editing factor (MORF) proteins, ZmMORF1 and ZmMORF8, interact with PCW1, EMP7, and bCCP1. ZmMORF8 enhanced the EMP7–PCW1 interaction in a yeast three-hybrid assay. C-to-U editing at the ccmF_N-1553 site in maize required EMP7, bCCP1, and PCW1. These results suggest that PPR-E proteins function in RNA editing by recruiting the trans deaminase PCW1 and bCCP1, and MORF1/8 assist this recruitment through protein–protein interactions.

PPR-E proteins function in RNA editing by recruiting trans deaminase PCW1 where bCCP1 and MORF1/8 assist the recruitment through protein–protein interactions in mitochondria.

IN A NUTSHELL.

Background: Mitochondria produce energy for all cell activities and contain genomes harboring about 60 genes. Mitochondrial gene transcripts undergo complex posttranscriptional modifications, including RNA C-to-U editing, intron splicing, maturation of transcript ends, and RNA stabilization. RNA C-to-U editing is widespread in vascular plants and plays an essential role in organellar gene expression and plant growth and development. However, the underlying mechanism of RNA editing is not fully understood. PLS-class pentatricopeptide repeat (PPR) proteins are divided into three subclasses, PPR-E, PPR-E+, and PPR–DYW, and function as critical factors in C-to-U editing of RNAs. The DYW domain in PPR–DYW proteins provides the cytidine deaminase activity for RNA C-to-U editing and PPR-E+ proteins recruit an atypical PPR–DYW protein to function in RNA editing.

Question: The PPR-E proteins lack the E+ and DYW domains but function in RNA editing. How do PPR-E subclass proteins mediate RNA C-to-U editing?

Findings: We identified bZIP and coiled-coil domain-containing PPR 1 (bCCP1) and an atypical PPR–DYW protein, PCW1, which are required for the C-to-U editing at 66 sites and 102 sites in maize (Zea mays) mitochondria, respectively. Surprisingly, the 66 editing sites overlap entirely with the 102 editing sites and are solely associated with known PPR-E class proteins. Furthermore, protein interaction analyses show that bCCP1, PCW1, and the multiple organellar RNA editing factor (MORF) proteins ZmMORF1/8 interact with each other, and the PPR-E protein Empty pericarp7 interacts with bCCP1 and ZmMORF1/8. Thus, we propose a model for PPR-E protein function in RNA editing in which the PPR-E protein recognizes the target site and recruits the trans cytidine deaminase PCW1 with the assistance of bCCP1 and MORFs.

Next steps: The editing at 102 sites requires PCW1, but only 66 of these sites require bCCP1; therefore, one or more bCCP1-like proteins are predicted to function as a bridge between the PPR-E proteins and PCW1. What is the other bCCP1-like protein? Are there other non-PPR editing factors in the PPR-E complexes? What is the role of the non-PPR proteins in the editosome?

Introduction

Mitochondria, which produce energy for all cell activities, contain genomes inherited from their α-proteobacterial ancestor. Plant mitochondrial genomes retain about 60 genes encoding rRNA, tRNA, and proteins in the translation and oxidative phosphorylation machinery (Unseld et al., 1997; Notsu et al., 2002; Clifton et al., 2004). Transcripts of the mitochondrial genes undergo complex posttranscriptional modifications, including RNA C-to-U editing, intron splicing, maturation of transcript ends, and RNA stabilization (Giege and Brennicke, 2001; Hammani and Giege, 2014). RNA C-to-U editing is widespread in vascular plants, with more than 300 editing sites in mitochondria (Giege and Brennicke, 1999; Notsu et al., 2002; Mower and Palmer, 2006; Bentolila et al., 2013; Wang et al., 2019b) and 20–40 editing sites in plastids (Tsudzuki et al., 2001; Tillich et al., 2005). The editing often restores the conserved amino acids (AAs) critical to protein functions (Liu et al., 2013; Wang et al., 2021), facilitates intron splicing (Xu et al., 2020), enhances tRNA precursor processing (Fey et al., 2002), and generates start or stop codons (Kadowaki et al., 1995; Kotera et al., 2005). RNA C-to-U editing thus plays an essential role in organellar gene expression.

RNA C-to-U editing involves many nucleus-encoded factors, many of which are pentatricopeptide repeat (PPR) proteins (Sun et al., 2016; Small et al., 2020). PPR proteins are classified into the P-class, harboring bona fide P-motifs with 35 AAs, and the PLS-class, comprises P-, L- (35–36 AAs), and S-motifs (31–32 AAs) (Small and Peeters, 2000; Lurin et al., 2004). PLS-class PPR proteins are divided into PPR-E, PPR-E+, and PPR–DYW by the presence of the E, E+, and DYW domains, respectively (Lurin et al., 2004). Almost all of the reported PLS-PPRs have a function in RNA C-to-U editing (Barkan and Small, 2014). Tandem PPR motifs recognize the nucleotide sequences upstream of the target Cs through specific interaction with the 6, 1′ AA residues in the adjacent two PPR motifs (Barkan et al., 2012).

The DYW domain harboring conserved cytidine deaminase (CDA)-like zinc-binding signature residues (HxE(x)nCxxC) provides cytidine deaminase activity (Oldenkott et al., 2019; Hayes and Santibanez, 2020; Takenaka et al., 2021). However, the PPR-E and PPR-E+ subclasses lack the DYW domain. Thus, the cytidine deaminase activity is proposed to be provided in trans (Okuda et al., 2006). Potential trans deaminases include PPR–DYW proteins or other cytidine deaminases. The atypical PPR-E protein CHLORORESPIRATORY REDUCTION 4 specifically recognizes and binds to the sequence upstream of the ndhD-1 site and recruits the DYW1 in trans to carry out the editing at this site (Boussardon et al., 2012). In addition, other PPR-E+ proteins are shown to recruit an atypical PPR–DYW protein DYW2 in the editing, where the P-class PPR protein NUWA assists the interaction in Arabidopsis (Arabidopsis thaliana) (Andrés-Colás et al., 2017; Guillaumot et al., 2017), providing a model for PPR-E+ proteins in editing. In contrast, the PPR-E proteins lacking the E+ domain, and the DYW domain have the function in RNA editing (Sun et al., 2015a; Qi et al., 2017; Wang et al., 2019a; Ren et al., 2020). Therefore, how PPR-E subclass proteins mediate the RNA C-to-U editing remains to be resolved.

In addition to PPR proteins, multiple organellar RNA editing factor/RNA-editing factor interacting proteins (MORFs/RIPs) function as key editing factors. MORF/RIP proteins contain nine members (MORF1–9), which are required for the editing at ∼80% of mitochondrial target Cs and almost every editing site at plastid transcripts in Arabidopsis. MORF8 is involved in the editing at over 300 sites in mitochondria and several sites in plastids (Bentolila et al., 2012; Takenaka et al., 2012; Bentolila et al., 2013). MORF1 and MORF3 are required for more than 50 editing events in mitochondria, while MORF2 and MORF9 were shown to act at almost all the editing sites in plastids (Takenaka et al., 2012; Bentolila et al., 2013). Several MORFs were shown to interact with themselves and other MORFs, and selectively with PPR-E, PPR-E+, and PPR–DYW proteins (Bentolila et al., 2012; Takenaka et al., 2012; Zehrmann et al., 2015; Bayer-Csaszar et al., 2017). Co-crystal structural analysis of MORF9 associated with synthetic PLS-PPR and electrophoretic mobility shift assays showed that the association between MORF9 and PLS-PPR increases the affinity of the PPR motifs for the target RNA (Yan et al., 2017; Royan et al., 2021). In addition, a different hypothetical role for MORF proteins has been proposed in which they bring together PPR-E/E+ and PPR–DYW proteins, which provide the specificity for editing sites and cytidine deaminases (Bayer-Csaszar et al., 2017; Small et al., 2020). However, we lack direct evidence to support this hypothesis.

Through molecular characterization of two seed development mutants in maize (Zea mays), we identified a novel protein, bZIP and coiled-coil domain-containing PPR 1 (bCCP1), and an atypical PPR–DYW protein, PPR motif, coiled-coil, and DYW domain-containing protein 1 (PCW1), which are required for the C-to-U editing at 66 sites and 102 sites in mitochondria, respectively. Surprisingly, the 66 editing sites overlap entirely with the 102 editing sites and are solely associated with known PPR-E class proteins. Protein interaction analysis showed that bCCP1, PCW1, ZmMORF1, and ZmMORF8 interact with each other. Using the PPR-E protein Empty pericarp7 (EMP7) as a test for these interactions showed that bCCP1 strongly interacts with EMP7 and PCW1. ZmMORF1 and ZmMORF8 interact with PCW1, EMP7, and bCCP1. A yeast three-hybrid assay showed that ZmMORF8 could enhance the interaction between PCW1 and EMP7. Based on these results, we proposed a model for PPR-E protein editing in which the PPR-E protein recognizes the target site and recruits the trans cytidine deaminase PCW1 with the assistance of bCCP1 and MORFs. This model may represent a major form of editing in maize mitochondria and is reminiscent of the model for PPR-E+ editosomes characterized in Arabidopsis.

Results

Phenotypic and genetic characterization of bccp1-1

We isolated a seed mutant bccp1-1 from the UniformMu maize (Z. mays) population (McCarty et al., 2005). Selfed ears of bccp1-1 heterozygotes segregated severe empty pericarp (emp) kernels in a 3:1 ratio (wild-type:emp = 643:202 = 3:1, χ² = 0.54, P = 0.46; Figure 1A; Supplemental Table S1), indicating a nuclear recessive mutation. The mutant kernels at 14-day after pollination (DAP) were much smaller than the wild-type kernels. The mutant embryo was barely visible, and the endosperm was tiny (Figure 1C). Paraffin sectioning showed that the wild-type kernels developed a complete embryonic structure at 14 DAP (Figure 1, D and F), whereas the bccp1-1 embryo only reached the transition stage, and the endosperm remained tiny (Figure 1, E and G). These results show that embryogenesis is arrested at the transition stage, and the endosperm appears to be arrested at the cellularization stage. As such, bccp1-1 is embryo lethal.

Figure 1.

Embryogenesis and endosperm development is severely arrested in the bccp1-1 mutant. A. A selfed ear at 14 DAP produced by the bccp1-1 heterozygous plant. The emp seeds are marked by red arrows. B and C, The embryo (em) and endosperm (en) of the wild-type (WT) (B) and bccp1-1 mutant (C) at 14 DAP. D–G, Paraffin section of WT (D and F) and bccp1-1 (E and G) kernels at 14 DAP. SC, scutellum; COL, coleoptile; LP, leaf primordia; SAM, shoot apical meristem; RAM, root apical meristem; COR, coleorhiza. Bars = 1 cm (A) and 1 mm (B–G). H, Schematic structure of the bCCP1 gene. The coding region of bCCP1 is shown in black and the position of the Mu insertion in bccp1-1 is marked by triangles. The mutation sites +1,189 (A>C) and +1,192 (G>T) in bccp1-2 are shown. I, PCR amplification of fragments of bCCP1 (1,328 bp) and bCCP1-Mu (3,527 bp) from genomic DNA (gDNA) of the WT and bccp1-1 mutant, respectively. The size of the bands is shown on the right. J, RT–PCR analysis of the transcripts in WT and bccp1-1 kernels at 14 DAP. Primers are shown on the right. The templates were normalized by ZmActin.

The Mu-seq analysis was used to identify the causal mutation of bccp1-1 (Settles et al., 2007; Liu et al., 2016). A Mu insertion 1,508-bp downstream from the translation start codon in AC218148.2_FG008 was identified to be linked with the bccp1-1 mutant in a six mutant and six wild-type family members (Figure 1H). Analysis of an additional 32 F2 individuals showed that only the plants harboring the Mu insertion segregated the emp kernels, indicating a tight linkage between the Mu insertion and the phenotype (Supplemental Figure S1B). Sequencing of the polymerase chain reaction (PCR)-amplified Mu-containing fragment identified a Mu7 element inserted into AC218148.2_FG008 (Figure 1, H and I). The AC218148.2_FG008 transcript was expressed in the developing wild-type kernels but not in the bccp1-1 kernels, suggesting that the insertion interrupted the expression of this gene (Figure 1J). To further test whether bCCP1 is the causal gene, we isolated a bccp1-2 allele from the maize ethyl methanesulfonate-induced mutant database (Lu et al., 2018). bccp1-2 contains a 1189^A→C mutation causing Lys→Gln and an 1192^G→T mutation producing a stop codon, causing a truncation of 269 AAs at the C-terminus of AC218148.2_FG008 (Figures 1, H and 2, A; Supplemental Figure S2, A and B).

Figure 2.

bCCP1 is dual-localized in the mitochondria and nucleus. A, The protein structure of bCCP1. TP, target peptide; bZIP, basic region/nuclear localization signal leucine zipper domain. The position of the Mu7 insertion in the bccp1-1 mutant is marked by triangles. The mutation site in bccp1-2 is shown. B, The pBI221-bCCP1-GFP and pGWB5-bCCP1-GFP vectors were transiently transformed into Arabidopsis protoplasts and tobacco leaf epidermal cells, respectively. The GFP signals are merged with mitochondria stained by MitoTracker Red and nuclei stained by DAPI. Fluorescence signals were detected by laser confocal microscopy ZEISS LSM 880. DIC, differential interference contrast. Bars = 10 μm.

Self-pollinated bccp1-2 heterozygotes segregated a quarter of the emp kernels (Supplemental Figure S2C). All the emp kernels were found to be homozygous for bccp1-2 (Supplemental Figure S2A). The bccp1-2 mutant kernels at 12 DAP developed a barely visible embryo and an endosperm that was much smaller than that of the wild-type (Supplemental Figure S2, D and E). Paraffin sectioning of the developing kernels at 12 DAP showed similar results (Supplemental Figure S2, F–I). Reciprocal crosses between heterozygotes bccp1-1 and bccp1-2 produced the emp kernels at a 25% ratio (Supplemental Figure S3, A–F), suggesting that AC218148.2_FG008 is the causal gene of the emp phenotype.

To further confirm this, we created transgenic maize over-expressing bCCP1 (bCCP1-OE) by placing bCCP1 under the control of the Ubiquitin promoter. Seven independent lines (bCCP1-OE1 to bCCP1-OE7) were generated, and reverse transcription–quantitative PCR (RT–qPCR) analysis showed increases of 5 to 110 times in expression levels compared with levels in the wild-type (Supplemental Figure S4A). bCCP1-OE3 was crossed with bccp1-1 heterozygotes. In the F2 progeny, four viable seedlings homozygous for bccp1-1 were identified, all of which contain the bCCP1-OE transgene (Supplemental Figure S4, B and C). These four plants can complete the life cycle (Supplemental Figure S4C). Thus, the emp phenotype in the bccp1-1 and bccp1-2 mutant is caused by the mutation of AC218148.2_FG008.

bCCP1 encodes a novel protein that is dual-targeted to mitochondria and nuclei

bCCP1 encodes a protein with several unique domains. In contrast to the PPR proteins, bCCP1 contains a basic region/nuclear localization signal leucine zipper motif (bZIP) at the N-terminus, eight PPR motifs in the middle, and a coiled-coil(CC) domain at the C-terminus (Figure 2A); thus, it is named bCCP1. BLAST analysis shows that bCCP1 shares 37.82% sequence similarity with AtNUWA and 18.72% similarity with Arabidopsis GLUTAMINE-RICH PROTEIN23 (AtGRP23). The putative orthologs of AtNUWA and AtGRP23 are GRMZM2G074599 (ZmNUWA) and GRMZM2G164018 (ZmGRP23) in maize, with a sequence similarity of 46.45% and 37.58%, respectively (Supplemental Figure S5). Instead, bCCP1 shares a sequence similarity of 50.32% and 20.32% with ZmNUWA and ZmGRP23, respectively (Supplemental Figure S5), suggesting that bCCP1 is not a recent duplication of either AtNUWA or AtGRP23. Thus, bCCP1 may be a novel protein specific to the maize lineage, hence not present in Arabidopsis.

Phylogenetic analysis shows two major clades, one consisting of bCCP1, ZmNUWA, and AtNUWA, and the other composed of GRP23 homologs, but the bCCP1 homologs are clustered in a subclade (Supplemental Figure S6). Analysis of 94 sequenced species suggests that bCCP1 is probably Poaceae specific (Supplemental Figure S7). To determine whether bCCP1 is divergent from AtNUWA or whether they have an overlapping function, we over-expressed bCCP1 in the Arabidopsis nuwa mutant. The nuwa mutant shows an embryo lethal phenotype that arrests embryogenesis before the 16-cell dermatogen stage (He et al., 2017). Analysis of 48 T₂ transformants indicated that over-expressing bCCP1 could not rescue the nuwa phenotype (Supplemental Figure S8B), supporting the notion that bCCP1 may have a distinct function and is monocot specific. bCCP1 is predicted to have a mitochondrion target peptide according to the TargetP (http://www.cbs.dtu.dk/services/TargetP) and Predotar algorithms (https://urgi.versailles.inra.fr/predotar/ predotar.html). To experimentally localize bCCP1, we fused bCCP1 with GFP in a binary vector pBI221 (pBI221-bCCP1-GFP) and pGWB5 (pGWB5-bCCP1-GFP) for transient expression in Arabidopsis protoplasts and tobacco (Nicotiana tabacum) leaf epidermal cells, respectively. Green fluorescent protein (GFP) signals were detected in small punctate dots merged with the mitochondria stained with MitoTracker Red (Figure 2B). GFP signals also appeared in the nuclei stained with DAPI (Figure 2B), indicating that bCCP1 is dual-targeted to mitochondria and nuclei.

bCCP1 is required for the C-to-U editing at 66 sites in mitochondria

Because bCCP1 is related to AtNUWA, which functions in the C-to-U editing at many sites in Arabidopsis mitochondria and plastids (Andrés-Colás et al., 2017; Guillaumot et al., 2017), we tested whether bCCP1 is involved in RNA editing in maize. The editing profiles of the 35 mitochondrial protein-coding genes were analyzed in the bccp1 mutants and the wild-type siblings using strand- and transcript-specific RNA-seq (STS-PCR-seq) (Bentolila et al., 2013). The results showed that the C-to-U editing at 66 target Cs is severely decreased in the bccp1-1 and bccp1-2 mutants compared with in the wild-type (Figure 3, A and B; Supplemental Data Set 1). These sites are distributed in 23 mitochondrial transcripts (nad1, -2, -4L, -5, -6, -7, -9; atp4, -6; rps1, -2A, -2B, -3, -4, -13; rpl16; cox2, -3; matR; ccmB; ccmF_C; ccmF_N; and cob). The editing efficiency of 30 and 32 sites was reduced by >50% in the bccp1-1 and bccp1-2 mutants, respectively, compared with in the wild-type (Figure 3, A and B). To verify this result, we used RT–PCR to amplify the cDNA and directly sequenced the amplicons in the mutant and the wild-type. The results are consistent with the STS-PCR-seq results (Supplemental Figure S9). The absence of editing at most sites results in AA changes in the encoded proteins (Supplemental Figure S9). In addition, the editing at 23 and 29 sites in mitochondrial transcripts was increased in the bccp1-1 and bccp1-2 mutants, respectively, compared with in the wild-type (Supplemental Figure S10, A and B), which is probably an indirect effect of the mutation. It is possible that the impaired process in bccp1-1 and bccp1-2 mitochondria enhances the expression of certain editing factors, which leads to increased editing at certain sites. Another possibility is that the loss of bCCP1 increases the formation of other editing complexes. It was reported that the MEF8 protein might play an inhibitory role in mitochondrial editing (Diaz et al., 2017). It is also possible that bCCP1 plays a similar role as MEF8 in editing. These results indicate that bCCP1 is required for the C-to-U editing at 66 sites in maize mitochondria.

Figure 3.

Editing at 66 mitochondrial sites is defective in the bccp1-1 and bccp1-2 mutants. The editing efficiency is shown by T/(T+C)% in the WT and in the bccp1-1 (A), and bccp1-2 (B) mutants. The relevant raw data is reported in Supplemental Data Set 1. Each RNA sample was extracted from the embryo, and endosperm was sampled from six independent selfed ears. Two independent RNA samples were used to analyze the editing by STS-PCR-seq. The bars show the mean ± sd (standard deviation; n = 2).

bCCP1 interacts with an atypical PPR–DYW protein PCW1

The loss-of-function of bCCP1 decreasing the editing at 66 sites implies that bCCP1 might play a general role in mitochondrial RNA editing. A cytidine deaminase is required for bCCP1. Thus, we explored whether a trans cytidine deaminase is associated with bCCP1. Atypical PPR–DYW proteins were considered the primary candidates because they are implicated in the C-to-U editing at many sites in mitochondria and chloroplasts (Andrés-Colás et al., 2017; Diaz et al., 2017; Guillaumot et al., 2017). Unlike canonical PPR–DYW proteins, atypical PPR–DYW proteins do not possess the E1 and E2 domains. DYW2 is involved in editing numerous sites in Arabidopsis mitochondria (Andrés-Colás et al., 2017; Guillaumot et al., 2017). Five atypical PPR–DYW proteins have been identified in Arabidopsis, namely DYW2, DYW3, DYW4, MEF8, and MEF8S (Diaz et al., 2017; Guillaumot et al., 2017).

We analyzed these atypical PPR–DYW proteins and found only three members in maize, ZmDYW2A (GRMZM2G017821), ZmDYW2B (GRMZM2G073551), and PCW1 (GRMZM2G158645), and two members in rice (Oryza sativa), OsDYW2 (LOC_Os04g09530) and OsPCW1 (LOC_Os01g53610). Phylogenetic analysis shows three clades, one consisting of ZmDYW2A, ZmDYW2B, OsDYW2, and AtDYW2, one harboring PCW1 and OsPCW1, and the other composed of MEF8, MEF8S, AtDYW3, and AtDYW4 (Supplemental Figure S11A). Alignment analysis shows that PCW1 and OsPCW1 share a 58.09% sequence identity, and PCW1 shows higher sequence similarity with Arabidopsis MEF8 and MEF8S than with AtDYW2 (Supplemental Figure S11A). However, the sequence identities of PCW1 with the two Arabidopsis proteins are merely 28.80% and 29.10%, respectively (Supplemental Figure S11, A and B). ZmDYW2A and ZmDYW2B share a 44.48% and 43.84% sequence identity with AtDYW2, and 46.91% and 46.55% sequence identity with OsDYW2, respectively. ZmDYW2A and ZmDYW2B share an 85.23% sequence identity, suggesting they might be divergent duplicates.

These three genes were cloned and tested for interaction with bCCP1 in a yeast two-hybrid assay (Y2H). As shown in Figure 4A and Supplemental Figure S12, bCCP1 strongly interacted with PCW1 in yeast but did not show interaction with ZmDYW2A or ZmDYW2B. The strong interaction between bCCP1 and PCW1 was also detected in a firefly luciferase (LUC) complementation imaging (LCI) assay (Figure 4B) and a bimolecular fluorescence complementation (BiFC) assay (Figure 4C). As shown in Figure 4B, bCCP1-CLuc and PCW1-NLuc co-expression in the tobacco leaf epidermis led to intense LUC activity. No LUC activity was detected in tobacco leaves co-expressing bCCP1-CLuc/Nluc or PCW1-NLuc/CLuc (Figure 4B). In addition, after co-expressing the fusion proteins bCCP1-cYFP and PCW1-nYFP in Arabidopsis protoplasts, the punctated dots of YFP, which are merged with mitochondria stained by MitoTracker Red, were detected (Figure 4C; Supplemental Figure S13A). When there was co-expression of bCCP1-cYFP/PPR14-nYFP, PPR–SMR1-cYFP/PCW1-nYFP, or PPR14-nYFP/PCW1-cYFP, no YFP signal was observed in the protoplasts (Figure 4C). The PPR14-nYFP/PPR–SMR1-cYFP set was used as the positive control in the BiFC assay (Wang et al., 2020). In addition, we found that PCW1 might form homodimers in the Y2H and BiFC assay (Figure 4, A and C). These results show that bCCP1 may physically interact with the atypical PPR–DYW protein PCW1 but not with ZmDYW2A or ZmDYW2B in maize mitochondria.

Figure 4.

bCCP1 interacts with PCW1 in Y2H, LCI, and BiFC assays. A, Y2H assay tests of the interaction between bCCP1 and PCW1. The colony pictures were taken after 3 days at 30°C in DDO, QDO, and QDO+X-α-gal plates. The BD-NUWA AD-DYW2 set was used as a positive control. B, LCI assay tests of the interaction between bCCP1 and PCW1 in tobacco leaf epidermal cells. C, BiFC assay tests of the interaction between bCCP1 and PCW1 in Arabidopsis protoplasts. The mitochondria were stained by MitoTracker Red. Fluorescence signals were detected by laser confocal microscopy with a ZEISS LSM 880. The PPR14-nYFP/PPR–SMR1-cYFP set was used as a positive control. Bars = 10 μm.

PCW1 is localized in mitochondria

PCW1 harbors six PPR motifs and a C-terminal DYW domain separated by a CC domain (Figure 5A), thus it is named PCW1. PCW1 was predicted to localize in mitochondria by the TargetP and Predotar algorithm. Subcellular localization of PCW1 by PCW1–GFP fusion confirmed the mitochondrial localization (Figure 5B). The GFP signals of PCW1–GFP were observed in punctate dots merged with mitochondria stained by MitoTracker Red (Figure 5B). No GFP signal was detected in other compartments in cells (Figure 5B). Thus, PCW1 is exclusively localized in mitochondria.

Figure 5.

PCW1 is a mitochondrion-targeted atypical PPR–DYW protein. A, Schematics structure of the PCW1 protein. TP, target peptide. B, Subcellular localization of PCW1 in Arabidopsis protoplasts. The mitochondria were stained by MitoTracker Red. Fluorescence signals were detected by laser confocal microscopy with a ZEISS LSM 880. Bars = 10 μm.

Loss-of-function of PCW1 arrests kernel development in maize

To genetically analyze the function of PCW1, we created mutants using clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9)-based genome editing. Two independent mutants, pcw1-1 and pcw1-2, were isolated, which carry a deletion of 775C and two bases of AT from position 161 from the start codon in PCW1, respectively (Figure 6A). The mutations cause a frameshift that presumably knocks out PCW1 expression. The selfed progenies of pcw1-1 and pcw1-2 heterozygotes segregated about 25% empty pericarp (emp) kernels (Figure 6B; Supplemental Figure S14A), and reciprocal crosses between pcw1-1 heterozygotes and pcw1-2 heterozygotes also produced approximately one-fourth emp kernels (Supplemental Figure S14, D and E). Genotyping by sequencing confirmed that the emp kernels are homozygous for either pcw1-1 or pcw1-2 (Figure 6A). The pcw1-1 and pcw1-2 mutant kernels at 12 DAP with an invisible embryo and a tiny drop-shaped endosperm were much smaller than wild-type kernels (Figure 6D; Supplemental Figure S14C), which was confirmed by inspection of sectioned tissue (Figure 6, E–H). Thus, loss-of-function of PCW1 severely arrests embryogenesis and endosperm development.

Figure 6.

Null mutation of PCW1 leads to the arrest of embryogenesis and severely impairs endosperm development. A, The sequence and sequencing peaks of the pcw1-1 and pcw1-2 mutants constructed by CRISPR/Cas9-based genome editing. B, A selfed ear at 12 DAP produced by the pcw1-1 heterozygous plant. The empty pericarp seeds are marked by red arrows. C and D, The embryo (em) and endosperm (en) of the WT (C) and pcw1-1 mutant (D) at 12 DAP. E–G, Paraffin section of WT (E and G) and pcw1-1 (F and H) kernels at 12 DAP. SC, scutellum; COL, coleoptile; LP, leaf primordia; SAM, shoot apical meristem; RAM, root apical meristem; COR, coleorhiza. Bars = 1 cm (B) and 1 mm (C–H).

PCW1 is required for the editing at 102 target Cs in maize mitochondria

Because PCW1 directly interacts with bCCP1, which is required for the editing at numerous sites, the editing profiles at 35 predicted mitochondrial protein-coding transcripts were analyzed using an STS-PCR-seq assay and by directly sequencing the amplicons in the wild-type and the pcw1 mutants. The editing at 102 sites distributed in 31 mitochondrial genes (nad1, -2, -3, -4, -4L, -5, -6, -7, -9; matR; mttB; ccmB, -F_C, -F_N; cob; cox1, -2, -3; atp1, -4, -6, -9; rps1, -2A, -2B, -3, -4, -7, -12, -13; and rpl16) was abolished or significantly decreased in the pcw1 mutants compared with in the wild-type (Figure 7A; Supplemental Figure S15; Supplemental Data Set 1). Interestingly, the 66 defective editing sites in the bccp1 mutant are also defective in the pcw1 mutants (Figure 7B), strongly supporting the finding that bCCP1 and PCW1 interact to facilitate the editing of the 66 target Cs in mitochondria. In addition, loss-of-function of PCW1 leads to increased editing at 47 mitochondrial target Cs (Supplemental Figure S16). These results indicate that PCW1 is a general factor in the C-to-U editing in maize mitochondria.

Figure 7.

Editing at 102 mitochondrial sites is severely defective in the pcw1 mutant. A, Sites at which editing was abolished or decreased in the pcw1-1 and pcw1-2 mutants compared with in the WT. The editing efficiency is shown by T/(T + C)%. The relevant raw data are reported in Supplemental Data Set 1. Each RNA sample was extracted from the embryo, and endosperm was sampled from six independent selfed ears. Two independent RNA samples were used to analyze the editing by STS-PCR-seq. The bars show the mean ± sd (n = 2). B, The number of defective editing sites in the bccp1 and pcw1 mutants. C, The PPR specificity of PCW1 and bCCP1 was estimated by counting the dependent and independent editing sites associated with PPR–DYW, PPR-E, and PPR-E+.

PCW1 and bCCP1 are involved in the editing at sites that are mediated by PPR-E proteins

PPR-E/E+ proteins are thought to bind to the upstream sequences of the target Cs and then recruit other editing factors to carry out the C-to-U editing (Barkan et al., 2012). In maize, the reported PPR-E proteins EMP7, SMK4, DEK10, and DEK55 specifically recognize ccmF_N-1553, cox1-1489, cox2-550, and matR-1877, ccmF_N-287, and rps13-56, and loss-of-function of these proteins abolished the editing of these six sites (Sun et al., 2015a; Qi et al., 2017; Wang et al., 2019a; Ren et al., 2020), among which six and three sites depend on PCW1 and bCCP1, respectively (Figure 7C; Supplemental Data Set 2). The PPR-E+ proteins analyzed are essential to the editing at 16 target Cs in maize (Li et al., 2014; Wang et al., 2017; Yang et al., 2017; Ding et al., 2019; Ren et al., 2019b; Wang et al., 2022). However, the editing at only nad4L-110 out of these 16 sites mediated by PPR-E+ is partially decreased in the pcw1 mutants compared with in the wild-type (Figure 7C; Supplemental Data Set 2). None of these sites are mediated by bCCP1 (Figure 7C; Supplemental Data Set 2). In addition, the reported PPR–DYW proteins specifically target 21 mitochondrial target Cs in maize, among which none of the sites overlap with the sites affected by bCCP1 and PCW1 (Figure 7C; Supplemental Data Set 2). As bCCP1 interacts with PCW1 but not ZmDYW2A or ZmDYW2B, these results suggest that the atypical PPR–DYW protein PCW1 is probably the deaminase recruited by PPR-E, and the bCCP1–PCW1 complex-mediated editing events are associated with PPR-E proteins rather than with PPR-E+ proteins.

EMP7, bCCP1, and PCW1 interact in the editing at the ccmF_N-1553 site

The PPR-E protein EMP7 is required for the C-to-U editing at the ccmF_N-1553 site in maize (Sun et al., 2015a), whose editing also requires both PCW1 and bCCP1 (Figure 8A). Thus, EMP7 offers a model to examine the relationships among these three proteins. EMP7 and bCCP1 did not exhibit an interaction in a Y2H assay (Figure 8B) but showed clear interaction in a BiFC assay in both Arabidopsis protoplasts and tobacco leaf epidermal cells (Figure 8D). YFP signals were detected in punctate dots merged with the mitochondria stained by MitoTracker Red, suggesting the interaction occurs in mitochondria (Figure 8D; Supplemental Figure S13A). No YFP signal was detected in the negative controls (Supplemental Figure S13, B and C). The interaction between bCCP1 and EMP7 may be too weak or too transient to be detected in the yeast assay. It might also be the case that the molecular environment differs in a way that prevents interaction between bCCP1 and EMP7 in the Y2H assay. We also performed an LCI assay to verify this interaction independently. Co-expression of bCCP1-CLuc and EMP7-NLuc in tobacco leaf epidermal cells produced an intense LUC activity (Figure 8C). These results suggest that bCCP1 may interact with EMP7 in vivo. However, no interaction was detected between EMP7 and PCW1 in Y2H, BiFC, or LCI assays (Figure 8, B–D). Because bCCP1 interacts with PCW1 and EMP7, bCCP1 may also facilitate the association between EMP7 and PCW1. Together, these proteins probably form a complex to carry out the C-to-U editing at the ccmF_N-1553 site in vivo.

Figure 8.

EMP7, bCCP1, and PCW1 interact in the editing at the ccmF_N-1553 site. A, Loss-of-function of bCCP1, PCW1, and EMP7 abolished or severely decreased the editing at the ccmF_N-1553 site. The ccmF_N-1553 sites are marked by arrows. B, Y2H assay tests of the interactions among bCCP1, PCW1, and EMP7. The colony pictures were taken after 3 days at 30°C in DDO, QDO, and QDO+X-α-gal plates. C, LCI assay tests of the interactions among bCCP1, PCW1, and EMP7 in tobacco leaf epidermal cells. D, BiFC assay tests of the interaction among bCCP1, PCW1, and EMP7 in Arabidopsis protoplasts and tobacco leaf epidermal cells. The mitochondria were stained by MitoTracker Red. Fluorescence signals were detected by laser confocal microscopy with a ZEISS LSM 880. Bars = 10 μm.

bCCP1, PCW1, and EMP7 interact with ZmMORF1 and ZmMORF8

Besides the PPR proteins, MORFs/RIPs are involved in RNA editing at many target Cs in mitochondria and/or plastids in Arabidopsis (Bentolila et al., 2012; Takenaka et al., 2012; Bentolila et al., 2013). Several MORFs were shown to form hetero- or homo-dimer and selectively interact with PLS-PPR proteins (Bentolila et al., 2012; Takenaka et al., 2012; Zehrmann et al., 2015; Bayer-Csaszar et al., 2017). A proposed role for MORFs is that they assist in the interaction between PPR-E/E+ and PPR–DYW (Bayer-Csaszar et al., 2017; Small et al., 2020). To test this hypothesis, we first examined whether PCW1 and EMP7 interact with ZmMORFs. The Y2H assay showed that both ZmMORF1 and ZmMORF8 interacted with EMP7 and PCW1 (Figure 9A). These interactions were displayed in BiFC and LCI assays. Co-expression of PCW1-NLuc/ZmMORF1-CLuc, PCW1-NLuc/ZmMORF8-CLuc, EMP7-NLuc/ZmMORF1-CLuc, or EMP7-NLuc/ZmMORF8-CLuc in tobacco leaf epidermal cells produced a strong LUC activity (Figure 9C). In addition, when the combinations of EMP7-cYFP/ZmMORF1-nYFP, EMP7-cYFP/ZmMORF8-nYFP, or PCW1-cYFP/ZmMORF8-nYFP were co-expressed in Arabidopsis protoplasts, punctated dots of YFP were detected that were merged with mitochondria (Figure 9B; Supplemental Figure S13A). However, no signal was detected when PCW1-cYFP and ZmMORF1-nYFP were co-expressed in Arabidopsis protoplasts (Figure 9B). No YFP signal was detected in the negative controls (Supplemental Figure S13C). These results revealed that both EMP7 and PCW1 potentially interact with ZmMORF1 and ZmMORF8, implying that ZmMORF1 and/or ZmMORF8 may play a role in facilitating the interaction between EMP7 and PCW1. The Y3H assay was employed to test this hypothesis. Co-expression of activation domain (AD)-PCW1 and binding domain (BD)-EMP7 with ZmMORF8 resulted in the growth of yeast in –Leu–Trp–His–Met dropout, –Leu–Trp–His–Ade dropout (QDO), and QDO + α-gal plates (Figure 9D). In contrast, no colony was produced in yeast expressing AD-PCW1 and BD-EMP7 in these plates. These results support the hypothesis that ZmMORF8 may play a role in enhancing the interaction between the PPR-E protein EMP7 and the atypical PPR–DYW protein PCW1. Because ZmMORF1 displayed strong auto-activation when fused to the BD domain, whether ZmMORF1 plays a similar role cannot be tested in the Y3H assay. In addition, PCW1 showed an interaction with ZmMORF5 and ZmMORF6 in the Y2H assay (Supplemental Figure S17).

Figure 9.

bCCP1, PCW1, and EMP7 interact with ZmMORF1 and ZmMORF8. A, Y2H assay tests of the interactions among bCCP1, PCW1, EMP7, ZmMORF1, and ZmMORF8. The colony pictures were taken after 3 days at 30°C in DDO, QDO, and QDO+X-α-gal plates. BD-ZmMORF1 displays auto-activation. B, BiFC assay tests of the interactions among bCCP1, PCW1, EMP7, ZmMORF1, and ZmMORF8 in Arabidopsis protoplasts. The mitochondria were stained by MitoTracker Red. Fluorescence signals were detected by laser confocal microscopy with a ZEISS LSM 880. Bars = 10 μm. C, LCI assay tests of the interactions among bCCP1, PCW1, EMP7, ZmMORF1, and ZmMORF8 in tobacco leaf epidermal cells. D, Y3H assay tests of whether ZmMORF8 enhances the interaction between PCW1 and EMP7. The colony pictures were taken after 3 days at 30°C in DDO plates and after 9 days at 30°C in –Leu–Trp–His–Met, QDO, and QDO-X-α-gal plates.

Although no interaction was detected between bCCP1 and ZmMORFs in the Y2H assay (Figure 9A), bCCP1 strongly interacted with ZmMORF1/8 in the BiFC and LCI assay (Figure 9, B and C). When co-expressing the fusion proteins bCCP1-cYFP/ZmMORF1-nYFP or bCCP1-cYFP/ZmMORF8-nYFP in Arabidopsis protoplasts, the YFP signals were observed in mitochondria (Figure 9B). Furthermore, co-expression of bCCP1-CLuc/ZmMORF1-NLuc and bCCP1-CLuc/ZmMORF8-NLuc in tobacco leaf epidermal cells produced a strong LUC activity (Figure 9C). Thus, bCCP1 may interact with ZmMORF1 and ZmMORF8. In addition, ZmMORF1 was found to interact with ZmMORF8 (Figure 9, B and C). Because ZmMORF1 and ZmMORF8 may directly interact with EMP7, PCW1, and bCCP1, whereas bCCP1 strongly interacts with EMP7 and PCW1, we proposed a model that bCCP1 and ZmMORF8 and/or ZmMORF1 probably facilitate the interaction between EMP7 and PCW1 to carry out RNA editing in mitochondria.

The C-terminal region containing the CC domain of bCCP1 is crucial to its function

The bccp1-2 allele is probably a null mutation as the G→T mutation causes a stop codon interrupting the coding of bCCP1 in the PPR motif region (Supplemental Figure S2, A and B). However, the bccp1-1 allele may offer insight into the domain function of bCCP1. The Mu7 element inserted in the bccp1-1 allele creates a stop codon right after the PPR motifs, potentially resulting in bCCP1 with a deletion of the C-terminal region that includes the CC domain (Supplemental Figure S18, A and B). RT–PCR analysis using the primer pair bCCP1-F7/TIR8 indicated that bCCP1 transcribed into the Mu7 element in the bccp1-1 mutant at a comparable level to that in the wild-type (Supplemental Figure S18C). The transcript includes the created stop codon (Supplemental Figure S18A). However, we were unable to verify if the truncated bCCP1 is produced in bccp1-1 because of the lack of a bCCP1 antibody. Given the loss of function in the bccp1-1 mutant, the result suggests that the C-terminal region, including the CC domain, is vital to bCCP1 function.

We then investigated whether the deletion of the C-terminal region of bCCP1 disrupts the interaction between bCCP1 and PCW1, ZmMORF1/8, and EMP7. First, the interaction between the bCCP1^ΔC and PCW1 was tested in a Y2H assay. The results showed that the deletion of the C-terminal in bCCP1 does not disrupt the bCCP1–PCW1 interaction in the Y2H assay (Supplemental Figure S19), which was proved in an LCI assay. When PCW1-NLuc and bCCP1^ΔC-CLuc were co-expressed in tobacco leaf epidermal cells, strong LUC activity was always produced (Supplemental Figure S20). Therefore, the interaction between bCCP1 and PCW1 could not rely on the C-terminal in bCCP1. To determine the interaction domain of bCCP1 mediating the bCCP1-PCW1 interaction, bCCP1^ΔbZIP, bCCP1^ΔC, and bCCP1_bZIP were constructed into pGADT7 and pGBKT7. A Y2H assay showed that deletion of the bZIP domain in bCCP1 completely abolished the interaction between bCCP1 and PCW1, while only the bZIP domain in bCCP1 interacts with PCW1. These results suggest that the bCCP1–PCW1 interaction may depend on the bZIP domain in bCCP1 (Supplemental Figure S19). It has been found that bCCP1 can interact with ZmMORF1/8 and EMP7 in LCI and BiFC (Figures 8, D, 9, B and C). We tested whether the deletion of the C-terminal in bCCP1 disrupts the interaction between bCCP1 and ZmMORF1/8 or EMP7 in an LCI assay. The results indicated that co-expression of ZmMORF1-NLuc/bCCP1^ΔC-Cluc or ZmMORF8-NLuc/bCCP1^ΔC-Cluc always produces a strong LUC activity in tobacco leaf epidermal cells (Supplemental Figure S20). Thus, the C-terminal containing the CC domain in bCCP1 may not be involved in the interaction between bCCP1 and ZmMORF1/8. However, when we co-expressed EMP7-NLuc/bCCP1^ΔC-Cluc in tobacco leaf epidermal cells, no LUC activity was detected (Supplemental Figure S20). This result suggests that the C-terminal in bCCP1 may be important for the interaction between bCCP1 and EMP7.

The CC motif is often involved in protein–protein interactions (Hao et al., 2013). We then investigated whether the bCCP1–PCW1, ZmMORF1–PCW1, and ZmMORF8–PCW1 interactions depend on the CC domain of PCW1 in yeast. The results indicated that the CC domain in PCW1 does not interact with bCCP1, ZmMORF1, or ZmMORF8, and the interactions of PCW1^ΔCC-DYW with bCCP1 and PCW1^ΔCC-DYW with ZmMORF8 are not weaker than the interactions of PCW1^ΔDYW with bCCP1 or PCW1^ΔDYW and ZmMORF8 in yeast (Supplemental Figure S21). Thus, the CC domain of PCW1 may not participate in the bCCP1–PCW1, ZmMORF1–PCW1, and ZmMORF8–PCW1 interactions. Because BD-PCW1^ΔDYW, BD-PCW1^ΔCC-DYW, and BD-ZmMORF1 display auto-activation (Figure 9A; Supplemental Figure S21), the interactions of PCW1^ΔDYW with ZmMORF1 and PCW1^ΔCC-DYW with ZmMORF1 cannot be tested.

Discussion

bCCP1 and PCW1 are general editing factors essential to embryogenesis and endosperm development in maize

Through genetic and molecular characterization of two severely defective seed development mutants in maize, we cloned and uncovered the functions of two genes, bCCP1 and PCW1. Both proteins contain unique structural characteristics strikingly different from related proteins. Unlike canonical P-type PPR proteins, bCCP1 contains the special bZIP and CC domains in addition to the PPR motifs (Figure 2A). PCW1 also harbors a CC domain sandwiched between the PPR motifs and DYW domain (Figure 5A). These features are not found in the large family of PPR proteins. In addition, bCCP1 orthologs cannot be found in Arabidopsis or other sequenced eudicots, only in Poaceae (Supplemental Figures S6 and S7), suggesting that it might be Poaceae specific (Supplemental Figure S7). PCW1 is functionally related to AtDYW2, AtDYW3, MEF8, and MEF8S but not the ortholog (Supplemental Figure S11, A and B). This phenomenon may reflect the fast evolution of PPR proteins after the rapid amplification of PPR proteins in land plants (O'Toole et al., 2008; Gutmann et al., 2020). Functional analysis identified that bCCP1 and PCW1 are required for the C-to-U editing at 66 and 102 sites in maize mitochondria, respectively. The absence of either protein abolishes the editing at many sites while substantially reducing the editing at other sites (Figures 3, A, B, and 7, A; Supplemental Figures S9 and S15). Most of the RNA editing factors identified so far function in the editing of one or a few sites in either chloroplasts or mitochondria, except DYW2 (Andrés-Colás et al., 2017; Guillaumot et al., 2017), MEF8 (Diaz et al., 2017), NUWA (Guillaumot et al., 2017), MORFs (Bentolila et al., 2012; Takenaka et al., 2012; Bentolila et al., 2013), ORRMs (Sun et al., 2013; Shi et al., 2015, 2016), and OZ1 (Sun et al., 2015b). These mentioned proteins are all general C-to-U editing factors. Thus, we conclude that bCCP1 and PCW1 are general C-to-U editing factors that have a general role in the editing in maize mitochondria.

RNA C-to-U editing is essential to the expression of organelle genes. RNA editing often restores the nucleotide to code for conserved AAs (Liu et al., 2013; Wang et al., 2021), to maintain structural integrity for intron splicing (Xu et al., 2020), to enhance tRNA precursor processing (Fey et al., 2002), or to generate start or stop codons (Kadowaki et al., 1995; Kotera et al., 2005). The absence of editing at one (such as nad7-836 in smk1 [Li et al., 2014] and ccmF_N-1553 in emp7 [Sun et al., 2015a]) or a few sites (such as atp6-635 and cox2-449 in emp18 [Li et al., 2019], rps4-335 and ccmB-43 in emp9 [Yang et al., 2017], atp4-59, nad7-383, and ccmF_N-302 in dek36 [Wang et al., 2017]) in mitochondrial transcripts leads to dysfunctional mitochondria and thereby severely inhibits maize seed development. The defective editing at 66 sites and 102 sites in bccp1 and pcw1 mutants severely affects the mitochondrial gene expression (Figures 3, A, B, and 7, A). Among these sites, the editing at 26 sites within 14 transcripts in the bccp1 mutant (Figure 3, A and B; Supplemental Figure S9) and 84 sites within 25 transcripts in the pcw1 mutant (Figure 7A; Supplemental Figure S15) is abolished or nearly abolished. And the absence of editing at most of the sites results in AA changes in the encoded proteins (Supplemental Figures S9 and S15). These proteins are essential to the function of mitochondria, particularly the OXPHOS machinery for cellular energy production. Defects in expression of the OXPHOS components lead to arrested embryogenesis and endosperm development in maize, rice, and Arabidopsis (de Longevialle et al., 2007; Francs-Small et al., 2012; Li et al., 2014; Sun et al., 2018; Xiao et al., 2018; Ren et al., 2019a; Xiu et al., 2020). Thus, the editing defects in the bccp1 and pcw1 mutants provide a plausible explanation for the severely arrested embryogenesis and endosperm development phenotype (Figures 1, A–G and 6, B–H; Supplemental Figures S2, C–I and S14, A–E). The block of the cytochrome pathway of the respiratory chain often induced the expression of ZmAOX genes. Indeed, we did find the enhancement of ZmAOX expression (Supplemental Figure S22, A–C).

PPR-E proteins recruit PCW1 as the trans deaminase in RNA C-to-U editing

C-to-U editing is chemically a deamination reaction of specific cytidines in RNA (Blanc et al., 1995; Yu and Schuster, 1995). This event requires a mechanism of target specification and an enzyme with cytidine deaminase activity. The tandem PPR motifs in PPR proteins serve the function of target site specification by direct binding to the upstream sequence of the target site via the 6, 1′ AA residues in the two adjacent PPR motifs (Barkan et al., 2012; Yin et al., 2013). The DYW domains in PPR–DYW proteins containing the CDAs-like zinc-binding signature residues (HxE(x)nCxxC) provide the deaminase activity (Oldenkott et al., 2019; Hayes and Santibanez, 2020; Takenaka et al., 2021). However, unlike the PPR–DYW proteins, the PPR-E and PPR-E+ class proteins lack the DYW domain. PPR-E and PPR-E+ proteins are more prevalent than PPR–DYW proteins in higher plants. For example, 47 PPR-E, 60 PPR-E+, and 87 PPR–DYW proteins are found in Arabidopsis, and 76 PPR-E, 49 PPR-E+, and 82 PPR–DYW proteins are present in maize (Lurin et al., 2004; Wei and Han, 2016). PPR-E/E+ proteins are thought to recruit a trans cytidine deaminase to carry out the C-to-U editing (Boussardon et al., 2012; Andrés-Colás et al., 2017; Diaz et al., 2017; Guillaumot et al., 2017). It has been proposed that PPR-E+ proteins recruit the atypical PPR–DYW protein DYW2 in Arabidopsis (Andrés-Colás et al., 2017; Guillaumot et al., 2017).

In this study, we have provided multiple lines of evidence indicating that the CC domain-containing PCW1 is the trans deaminase recruited by the PPR-E subclass proteins. PCW1 is a general editing factor required for the editing at 102 sites in maize mitochondria (Figure 7, A and B), suggesting that PCW1 plays a general role in C-to-U editing. These 102 sites were found to be associated almost exclusively with the PPR-E subclass proteins but not with PPR–DYW proteins (Figure 7C). Only one site, nad4L-110, recognized by a PPR-E+ protein was partially decreased in the pcw1 mutants. The association of PCW1 is different from DYW2, an atypical DYW protein required for editing over 300 sites and associated with the PPR-E+ proteins in Arabidopsis mitochondria (Guillaumot et al., 2017). It appears that PPR-E subclass proteins recruit PCW1, and PPR-E+ subclass proteins recruit DYW2 as the trans deaminase in the RNA C-to-U editing in mitochondria. To further examine the specific correlations between PCW1 and E-subclass PPR proteins, we analyzed the PLS–PPR proteins that function in mitochondrial editing in rice and Arabidopsis. In rice, six PPR–DYW, two PPR-E, and one PPR-E+ protein have been characterized for their functions in editing thus far (Supplemental Data Set 2). However, only one site (nad5-1580) overlaps with the PCW1-affected sites, and the responsible editing factor is MPR25, a PPR-E type protein (Toda et al., 2012). We also examined Arabidopsis, which might not be suitable for testing the correlation because bCCP1 and PCW1 are monocot-specific (Supplemental Figures S7 and S11A). However, we found that three PPR-E protein-targeted sites and three PPR-E+ protein-targeted sites overlap with the sites edited by PCW1, but not PPR–DYW-targeted sites (Supplemental Data Set 2). Interestingly, one of the PPR-E+ proteins in Arabidopsis appears to convert to PPR-E in maize and rice based on homology, which is required for the editing at nad5-1580, that is, MPR25 in rice (Toda et al., 2012). Intriguingly, all the PPR-E protein-affected sites in Arabidopsis showed an 80% decrease or more in the pcw1 mutant, in contrast to the PPR-E+ affected sites with a ∼20% decrease in the pcw1 mutant (Supplemental Data Set 1). We speculate that this might be a clue for the evolutionary diversification in editing machinery, that is, a subset of PPR-E proteins adopting the PPR-E/(bCCP1/MORF1/8)/PCW1 editosome in monocots, no longer dependent on DYW2 or MEF8 homologs.

The specific recruitment of PCW1 by PPR-E proteins and DYW2 by PPR-E+ proteins may have a structural basis. The crystal structure of the DYW domain was recently resolved (Takenaka et al., 2021). The DYW domain can undergo a transition from an inactive ground state to an active state. The transition is regulated by a conserved gating domain within the deaminase fold in the DYW domain. The gating domain, which is composed of an amphipathic α-helix and two anti-parallel β-sheets, is shared by the DYW domains of almost all PPR–DYW proteins of all plant clades. It is essential for cytidine deaminase activity (Takenaka et al., 2021). Protein structure modeling suggests that AtDYW2 and ZmDYW2A/2B show a less conserved N-terminus of the gating domain (Supplemental Figure S23A) (Takenaka et al., 2021). However, we analyzed 5,107 PPR-E+ proteins from various species and found that the E+ domain harbors the strictly conserved N-terminus of the gating domain (Supplemental Figure S23B). Thus, it is plausible that the missing N-terminus of the gating domain in DYW2 could be provided in trans by the E+ domain in the PPR-E+ proteins. The interacting PPR-E+/DYW2 reconstitutes a functional cytidine deaminase while the PPR-E+ recognizes the target C.

In contrast, the PPR-E proteins do not have the E+ and DYW domain, thus lacking this highly conserved N-terminus of the gating domain. However, PCW1 contains this highly conserved N-terminus of the gating domain (Supplemental Figure S23A). The PPR-E/PCW1 combination offers the target site recognition and cytidine deaminase activity, reconstituting an editing complex. Based on this hypothesis, PPR-E proteins cannot recruit DYW2 to constitute a functional cytidine deaminase in RNA C-to-U editing because of the lack of an N-terminus in the gating domain.

Furthermore, we tested this hypothesis by using the PPR-E protein EMP7, which, together with bCCP1 and PCW1, is required for the C-to-U editing of the ccmF_N-1553 (Sun et al., 2015a) (Figure 8A). The results show that bCCP1 strongly interacts with EMP7 and PCW1 (Figures 4, A–C and 8, C and D), suggesting that bCCP1 serves a bridge function to link EMP7 and PCW1. It is possible that EMP7 may interact with PCW1 weakly in vivo, and bCCP1 enhances this interaction. These results support a model of how PPR-E proteins function in RNA C-to-U editing. The PPR-E protein recognizes the upstream sequences of the target site and recruits the trans cytidine deaminase PCW1 with the assistance of bCCP1; then, PCW1 catalyzes the deamination of the target site. Given that the editing at 102 sites requires PCW1, whereas only 66 of these sites require bCCP1, one or more bCCP1-like proteins are predicted to function as a bridge between the PPR-E proteins and PCW1. In addition, about 14 sites in the pcw1 mutants showed only a partial reduction of RNA editing, suggesting that the editing at a specific site may involve more than one editosome. This phenomenon has been observed in quite some cases where the loss of one editing factor only partially affects the editing level, for example, in morf mutants (Bentolila et al., 2013). We speculate that functional redundancy is widely involved in editing.

Recruitment of PCW1 by PPR-E proteins is facilitated by bCCP1 and MORFs

The mechanism underlying PPR-E proteins recruitment of PCW1 remains a question. Using EMP7 as a PPR-E example, we did not find an interaction between EMP7 and PCW1 (Figure 8, B–D). It is possible that the interaction is too weak or transient to be detected. However, strong interactions were found between bCCP1 and PCW1 and between bCCP1 and EMP7 (Figures 4, A–C, 8, C and D), suggesting that EMP7 recruits PCW1 via bCCP1, that is, bCCP1 serves as a bridge between PCW1 and EMP7. This model is reminiscent of the PPR-E+ recruitment of DYW2, where NUWA helps to enhance the interaction in Arabidopsis (Andrés-Colás et al., 2017; Guillaumot et al., 2017). Although bCCP1 appears to be grass-specific (Supplemental Figure S7), this recruitment mechanism via a bridge protein seems to have converged in eudicots and monocots.

This study provided evidence that ZmMORF1 and ZmMORF8 can also facilitate the interaction between the PPR-E protein EMP7 and PCW1. MORF/RIP proteins have been identified as editing factors required for the editing of most targeted Cs in mitochondria and plastids (Bentolila et al., 2012; Takenaka et al., 2012; Bentolila et al., 2013). MORFs can form homo- and heterodimers and interact with PPR-E, PPR-E+, and PPR–DYW proteins selectively (Bentolila et al., 2012; Takenaka et al., 2012; Zehrmann et al., 2015; Bayer-Csaszar et al., 2017). MORF1 pulled down the P-class protein NUWA and the atypical PPR–DYW protein DYW2 in Arabidopsis. MEF8 and MEF8S interact with all the mitochondrial MORFs in yeast (Bayer-Csaszar et al., 2017), implying that MORF proteins play a role in nearly all the editing events via protein interactions. This study found that ZmMORF1 and ZmMORF8 can strongly interact with both bCCP1 and PCW1, and ZmMORF1 also interacts with ZmMORF8 (Figure 9, A–C). The 66 editing sites mediated by bCCP1 overlap entirely with the sites edited by PCW1 (Figure 7B), and bCCP1 interacts with PCW1 (Figure 4, A–C). bCCP1, ZmMORF1, and ZmMORF8 interact with PPR-E EMP7 and PCW1 (Figures 4, A–C, 8, C, D, and 9, A–C), and ZmMORF8 enhanced the interaction between EMP7 and PCW1 in a Y3H assay (Figure 9D). These results suggest that bCCP1, PCW1, ZmMORF8, and/or ZmMORF1 may form a complex to mediate RNA editing in vivo.

Based on these results we propose a model accounting for the C-to-U editing for PPR-E proteins in mitochondria. As shown in Figure 10, a PPR-E protein recognizes and binds to the upstream sequence of the target editing site via the tandem PPR motifs. The PPR-E protein may weakly or transiently interact with PCW1, the trans cytidine deaminase. bCCP1, ZmMORF1, and ZmMORF8 can strongly interact with PCW1 and the PPR-E protein, constituting an active editosome to position PCW1 on the target cytidine. Deamination of the cytidine converts it to uridine in mitochondria (Figure 10). This model may be conserved in PPR-E-mediated RNA C-to-U editing in plants.

Figure 10.

Model of the PPR-E complex in mitochondria. The PPR-E protein lacks cytidine deaminase activity. To function in C-to-U editing of mitochondrial mRNAs, it recruits the trans deaminase PCW1 with the assistance of bCCP1 and the MORF proteins ZmMORF1 and ZmMORF8.

Materials and methods

Plant materials and growth conditions

The bccp1-1 mutant of maize (Z. mays) was isolated from the UniformMu mutagenic population in a nearly isogenic W22 background (McCarty et al., 2005). The bccp1-2 mutant was isolated from the maize ethyl methanesulfonate-induced mutant population in the B73 genetic background (Lu et al., 2018). All maize plants were grown under natural conditions in the Shandong University Experimental Field in Qingdao, Shandong Province. Arabidopsis was grown at 22°C with a 16-h light/8-h dark cycle in a growth room. Tobacco (N. tabacum) plants were grown at 28°C with a 12-h light/12-h dark cycle in a growth room.

Creation of transgenic lines over-expressing bCCP1 and complementation

To create bCCP1 transgenic maize (bCCP1-OE), the full-length coding sequence of the bCCP1 gene was amplified and ligated to the pUNTF vector under the control of the maize ubiquitin promoter. The pUNTF-bCCP1 construct was transformed into inbred KN5585 by callus transformation. bCCP1-OE3 plants were crossed with bCCP1/bccp1-1 plants and selfed in the next generation in a complementation analysis. To over-express bCCP1 in the Arabidopsis (A. thaliana) nuwa mutant, the full-length coding sequence of the bCCP1 gene was amplified and ligated to the pBI121 vector under the control of the 35S promoter. The pBI121-bCCP1 construct was used to transform nuwa heterozygotes with the floral-dip method (Clough and Bent, 1998). The transgenic plants were screened in MS medium containing hygromycin and identified by PCR using primers bCCP1-F5/R5.

Creation of the pcw1 mutants by CRISPR/Cas9

The pcw1 mutants were created using CRISPR/Cas9-based technology. Two sites of PCW1 were targeted by the guiding sequence of 20 bp, 5′-AAATCCAACTCATTCCCGTA-3′ (position at 759 to 778 from the start codon) and 5′-CAGTCTTGGGTTCCACATCG-3′ (position at 145 to 164 from the start codon). The target sequences were ligated to the WMC009 vector and then transformed into inbred KN5585 by callus transformation. Two independent mutants, pcw1-1 and pcw1-2, were isolated, which carry a deletion of 775C and two bases of AT from position 161 from the start codon in PCW1, respectively.

Light microscopy of cytological sections

The bccp1-1, bccp1-2, and pcw1-1 mutant and wild-type kernels were harvested from ears of self-pollinated heterozygotes at 14 DAP (bccp1-1) or 12 DAP (bccp1-2 and pcw1-1). The kernels were cut along the longitudinal axis, and the slices containing embryo and endosperm were fixed, dehydrated, infiltrated, embedded, de-paraffinized and stained as described previously (Liu et al., 2013).

Subcellular localization

The full-length bCCP1 (without stop codon) was first cloned into the pENTR/D-TOPO vector (Thermo Fisher Scientific, Waltham, MA, USA; http://www.thermofisher.com), then subcloned binary vectors to generate the pGWB5-bCCP1-GFP and pBI221-bCCP1-GFP constructs. Agrobacterium EHA105 containing pGWB5-bCCP1-GFP was infiltrated into tobacco epidermal cells. The pBI221-bCCP1-GFP construct was introduced into Arabidopsis protoplasts by polyethylene glycol-mediated transformation. Fluorescence signals were detected 24 h after infiltration under a ZEISS LSM 880 confocal microscope. MitoTracker Red and DAPI were used to stain the mitochondrion and nucleus, respectively. The subcellular localization of PCW1 was carried out similarly by transforming the pBI221-PCW1-GFP construct into Arabidopsis protoplasts.

RNA extraction, RT–PCR, and RT–qPCR

Total RNA was extracted from plant tissues using TRIzol reagent (ThermoFisher Scientific, www.thermofisher.com) and was treated with DNase I (New England Biolabs, Ipswich, MA, USA; www.neb.sg) to remove any contaminating genomic DNA. Single-stranded cDNA was synthesized using random hexamer primers with a Transcriptor First Strand cDNA Synthesis kit (ThermoFisher Scientific). RT–qPCR was carried out using FastStart Essential DNA Green Master (Roche Diagnostics, Basel, Switzerland) and a LightCycler 96 (Roche Diagnostics). Relative expression was calculated using the 2^(−ΔΔCt) formula as described in a previous report (Wang et al., 2019b). ZmActin (GRMZM2G126010) was used to normalize the samples. Each experiment was replicated 3 times. The primers used in the RT–PCR and RT–qPCR analyses are shown in Supplemental Data Set 3.

STS-PCR-seq assay

The embryos and endosperm from wild-type, bccp1, and pcw1 kernels of segregating ears at 12 DAP were used to isolate RNA. Each RNA sample was extracted from the embryo and endosperm sampled from six independent selfed ears for the bccp1-1, bccp1-2, pcw1-1, and pcw1-2 mutants. Two independent RNA samples were used to analyze the editing by STS-PCR-seq. Thus, the editing efficiency at certain sites represents 12 biological replicates (Supplemental Data Set 1). The 35 protein-coding mitochondrial transcripts were amplified from the cDNA (Liu et al., 2013). High-throughput sequencing and data analysis were performed as described previously (Wang et al., 2019b). The threshold for declaring a decreased editing effectiveness was defined as [T/(T + C)% in the mutant − T/(T + C)% in the wild-type] −20% or less at least for one allele or the editing effectiveness is nearly zero in the mutant. The threshold for declaring increased editing effectiveness was defined as [T/(T + C)% in the mutant − T/(T + C)% in the wild-type] ≥20%. The primers used for RT–PCR to amplify the mitochondrial genes are in Supplemental Data Set 3.

Direct sequencing of RT–PCR amplicons

Embryo and endosperm samples were dissected from wild-type, bccp1, and pcw1 kernels at 12 DAP. Each RNA sample was extracted from the embryos, and the endosperm was sampled from six independent selfed ears. An RNA editing analysis was conducted from these samples by directly sequencing the RT–PCR amplicons as described in Liu et al. (2013). The necessary cDNA was obtained as described above and subjected to a series of RT–PCRs directed at the full set of 35 mitochondrial genes (primers given in Supplemental Data Set 3).

Y2H assay

The Y2H assay was performed according to the manual of the Matchmaker Gold Yeast Two-Hybrid System (Takara, Shiga, Japan). The full-length sequence without the transit peptide of bCCP1, PCW1, Emp7, ZmMORF1 (GRMZM2G139441), ZmMORF3 (GRMZM2G054537), ZmMORF9 (GRMZM2G003765), ZmMORF5 (GRMZM2G383540), ZmMORF6 (GRMZM5G808811), and ZmMORF8 (GRMZM2G169384), and the truncated fragments of bCCP1 and PCW1 without the transit peptide (bCCP1^ΔbZIP, bCCP1^ΔC, bCCP1_bZIP, PCW1_CC, PCW1 ^ΔDYW, and PCW1 ^ΔCC-DYW) were cloned into pGBKT7 and pGADT7, respectively. Colony pictures were taken after 3 days at 30°C in –Leu–Trp dropout (DDO), –Leu–Trp–His–Ade dropout (QDO), and QDO+X-α-gal plates.

Yeast three-hybrid assay

The full-length PCW1 cDNAs were ligated into the GAL4 activation domain vector (pGADT7). Emp7 and ZmMORF8 were placed into the pBridge vector that expresses two proteins, a GAL4 DNA-binding domain fusion and an additional protein under the control of a methionine-inducible promoter. The resulting construct, Emp7-pBridge-ZmMORF8, expresses BD-EMP7 and ZmMORF8. The pGADT7-PCW1/EMP7-pBridge-ZmMORF8 set and pGADT7-PCW1/EMP7-pBridge set were co-transformed into the yeast strain Y2H Gold. Yeast growth was determined after incubation in DDO (–Leu–Trp) for 3 days and in –Leu–Trp–His–Met, QDO, and QDO+X-α-gal medium for 9 days at 30°C.

BiFC

The full-length cDNA without stop codons of bCCP1, PCW1, EMP7, ZmMORF1, and ZmMORF8 were cloned into the PUC-SPYNE and PUC-SPYCE vectors, respectively. The BiFC assay was performed as described previously (Walter et al., 2004). The SPR2-cYFP/PPR–SMR1-nYFP set was used as the positive control for Figure 8D. Both SPR2 and PPR–SMR1 are PPR proteins that interact in Nicotiana benthamiana mitochondria in BiFC assays (Cao et al., 2022). The PPR14-nYFP/PPR–SMR1-cYFP set was used as the positive control for Figures 4, C, 8, D, and 9, B. As reported by Wang et al. (2020), both PPR14 and PPR–SMR1 are PPR proteins that interact in Arabidopsis protoplast mitochondria in BiFC assays. After transforming at 22°C for 30 h, the YFP signals were detected under a ZEISS LSM 880 confocal microscope, and MitoTracker Red was used as the mitochondrial marker.

LCI assay

The full-length sequence without stop codons of bCCP1, PCW1, EMP7, ZmMORF1, and ZmMORF8, and bCCP1^ΔC amplified from cDNA of W22 inbred maize were cloned into NLuc and CLuc, respectively. The LCI assay was performed as described previously (Chen et al., 2008). After infiltration at 28°C for 48 h, LUC activity was detected using a live animal and plant imager (Lumazone pylon 2048B).

Phylogenetic analysis

The protein sequences used in the phylogenetic analysis were acquired from the Uniport database. The alignment of protein sequences used in Supplemental Figures S6, S7 and S11A are showed in Supplemental Files S1, S2 and S3, respectively. A phylogenetic tree was produced using the Neighbor-Joining method with the MEGA version 7 software. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the p-distance method, and the units are the number of AA differences per site.

Accession numbers

Sequence data for the maize bCCP1, PCW1, ZmDYW2A, ZmDYW2B, ZmNUWA, ZmGRP23, ZmMORF1, ZmMORF3, ZmMORF4, ZmMORF5, ZmMORF6, and ZmMORF8 genomic loci, cDNA, and protein sequences can be found at the maizegdb database (www.maizegdb.org) with accession numbers bCCP1 (AC218148.2_FG008), PCW1 (GRMZM2G158645), ZmDYW2A (GRMZM2G017821), ZmDYW2B (GRMZM2G073551), ZmNUWA (GRMZM2G074599), ZmGRP23 (GRMZM2G164018), ZmMORF1 (GRMZM2G139441), ZmMORF3 (GRMZM2G054537), ZmMORF9 (GRMZM2G003765), ZmMORF5 (GRMZM2G383540), ZmMORF6 (GRMZM5G808811), and ZmMORF8 (GRMZM2G169384); at the Rice Genome Annotation Project (uga.edu) with accession numbers OsPCW1 (LOC_Os01g53610), OsDYW2 (LOC_Os04g09530); or at National Center for Biotechnology Information (NCBI) under the following accession numbers: bCCP1 (LOC100279827), PCW1 (LOC103651234), ZmDYW2A (LOC100304360), ZmDYW2B (LOC100381568), OsPCW1 (LOC4327881), OsDYW2 (LOC4335071), ZmNUWA (LOC100279583), ZmGRP23 (LOC100191742), ZmMORF1 (LOC100272746), ZmMORF3 (LOC100286015), ZmMORF9 (LOC100284899), ZmMORF5 (LOC100283838), ZmMORF6 (LOC100383758), and ZmMORF8 (LOC100282986).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Linkage analysis of the bccp1-1 mutant (supports Figure 1).

Supplemental Figure S2. Null mutation of bCCP1 severely impairs seed development in maize (supports Figure 1).

Supplemental Figure S3. Allelism test using heterozygous bccp1-1 and bccp1-2 (supports Figure 1).

Supplemental Figure S4. Overexpression of bCCP1 can rescue the empty pericarp kernels in the bccp1-1 mutant (supports Figure 1).

Supplemental Figure S5. Alignment of bCCP1, ZmNUWA, NUWA, GRP23, and ZmGRP23 (supports Figure 2).

Supplemental Figure S6. Nonrooted phylogenetic tree of bCCP1, ZmNUWA, ZmGRP23, NUWA, GRP23, and their orthologs (supports Figure 2).

Supplemental Figure S7. Phylogenetic tree of bCCP1, ZmNUWA, NUWA, and their orthologs (supports Figure 2).

Supplemental Figure S8. bCCP1 cannot rescue the embryo-lethal phenotype of nuwa (supports Figure 2).

Supplemental Figure S9. Part of the severely defective editing sites in bccp1-1 and bccp1-2 (supports Figure 3).

Supplemental Figure S10. Editing at some mitochondrial sites is increased in the bccp1 mutants (supports Figure 3).

Supplemental Figure S11. Evolutionary relationship of atypical PPR–DYW proteins in maize, Oryza sativa and Arabidopsis (supports Figure 4).

Supplemental Figure S12. Interaction tests between bCCP1 and the atypical PPR–DYW proteins (supports Figure 4).

Supplemental Figure S13. Quantification of the BiFC assays (supports Figures 4, 8, and 9).

Supplemental Figure S14. Null mutation of PCW1 leads to the arrest of embryogenesis and severely impaired endosperm development (supports Figure 6).

Supplemental Figure S15. Part of the severely defective editing sites in pcw1-1 (supports Figure 7).

Supplemental Figure S16. Forty-seven mitochondrial sites showed increased editing in the pcw1 mutants (supports Figure 7).

Supplemental Figure S17. Interaction test between PCW1 and ZmMORFs (supports Figure 9).

Supplemental Figure S18. bCCP1-1 may encode a truncated bCCP1 protein without a C-terminal, bCCP1^ΔC (supports Figures 1 and 2).

Supplemental Figure S19. Interaction test between PCW1 and the domains of bCCP1 (supports Figure 4).

Supplemental Figure S20. LCI assay tests for the interactions of bCCP1^ΔC with PCW1, EMP7, ZmMORF1, and ZmMORF8 (supports Figures 4, 8, and 9).

Supplemental Figure S21. Test for the interactions of the domains of PCW1 with bCCP1, ZmMORF1, and ZmMORF8 (supports Figure 9).

Supplemental Figure S22. Expression of ZmAOX2 and ZmAOX3 is significantly induced in the bccp1 and pcw1-1 mutants (supports Figures 1 and 6).

Supplemental Figure S23. Conservative analysis of the gating domain in PCW1 and of the N-terminus of the gating domain in PPR-E+ proteins (supports Figures 5 and 7).

Supplemental Table S1. Segregation ratio of the selfed ears of bccp1-1 heterozygotes (supports Figure 1).

Supplemental File S1. Alignments of protein sequences of bCCP1, ZmNUWA, ZmGRP23, and their orthologs for the phylogenetic tree in Supplemental Figure S6.

Supplemental File S2. Alignments of protein sequences of bCCP1, ZmNUWA, NUWA, and their orthologs for the phylogenetic tree in Supplemental Figure S7.

Supplemental File S3. Alignments of protein sequences of atypical PPR–DYW proteins in maize, Oryza sativa, and Arabidopsis for the phylogenetic tree in Supplemental Figure S11.

Supplemental Data Set 1. Number of reads at each editing site (gene position) for each library.

Supplemental Data Set 2. Editing sites are specifically recognized by PPR-E, PPR-E+, and PPR–DYW proteins.

Supplemental Data Set 3. Primers used for the study.

Supplemental Data Set 4 . Protein sequences of E+ domains in 5107 PPR-E+ proteins.

 

Y.W. and B.C.T. designed the research. Y.W. conducted most of the experiments. H.L. participated in linkage analysis. Z.Q.H. participated in the editing sites analysis. B.M. and L.W. participated in the Y2H analysis. Y.W., Y.Z.Y., Z.H.X., and B.C.T. analyzed the data. Y.W. and B.C.T. wrote the article.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell) is: Bao-Cai Tan (bctan@sdu.edu.cn).