Maize PPR278 Functions in Mitochondrial RNA Splicing and Editing

Pentatricopeptide repeat (PPR) proteins are a large protein family in higher plants and play important roles during seed development. Most reported PPR proteins function in mitochondria. However, some PPR proteins localize to more than one organelle; functional characterization of these proteins remains limited in maize (Zea mays L.). Here, we cloned and analyzed the function of a P-subfamily PPR protein, PPR278. Loss-function of PPR278 led to a lower germination rate and other defects at the seedling stage, as well as smaller kernels compared to the wild type. PPR278 was expressed in all investigated tissues. Furthermore, we determined that PPR278 is involved in the splicing of two mitochondrial transcripts (nad2 intron 4 and nad5 introns 1 and 4), as well as RNA editing of C-to-U sites in 10 mitochondrial transcripts. PPR278 localized to the nucleus, implying that it may function as a transcriptional regulator during seed development. Our data indicate that PPR278 is involved in maize seed development via intron splicing and RNA editing in mitochondria and has potential regulatory roles in the nucleus.


Introduction
Pentatricopeptide repeat (PPR) proteins are widely distributed in plants, with more than 400 members in maize (Zea mays L.) [1], which were also divided into two subfamilies: The P subfamily containing motif P only, and the PLS subfamily harboring the P, L, and S motifs [2,3]. PLS-subfamily members were further classified into PLS, E, E+, and DYW types according to their C-terminal domain [4,5]. PPR proteins have been isolated from species including maize, rice (Oryza sativa L.), and Arabidopsis thaliana L., and functional analysis revealed that they could bind RNA and might play important roles in RNA metabolism, including RNA splicing and editing, cleavage, and maturation [3].

The ppr278-1 Mutants Cause Aberrant Seed Development in Maize
We isolated a kernel mutant from our mutant libraries established by an ethyl methanesulfonate (EMS)-based method, which produced smaller seeds compared to the wild type (WT) ( Figure 1A-D), and named it ppr278-1 based on similar phenotypes with other ppr mutants in maize. Only~5% of mutant kernels could germinate normally on average and grew to the seedling stage ( Figures 1E and S1A,B). However, the vigor of mutant seedlings was notably decreased, and homozygous mutant seedlings did not survive to the reproductive stage ( Figure S1B). The 100-kernel weight of ppr278 mutant plants was significantly lower than that of WT, which might be due to developmental defects in mutant kernels or disruption of seed-filling processes ( Figure 1F,G). The segregation ratio of seeds in mutant ears was 3:1 (normal kernels: mutant kernels, 212:69), implying that the ppr278-1 phenotype is controlled by a single recessive locus ( Figure 1A,B).
significantly lower than that of WT, which might be due to developmental defects in mutant kernels or disruption of seed-filling processes ( Figure 1F,G). The segregation ratio of seeds in mutant ears was 3:1 (normal kernels: mutant kernels, 212:69), implying that the ppr278-1 phenotype is controlled by a single recessive locus ( Figure 1A,B).

PPR278 Encodes a Constitutively Expressed PPR Protein
We used bulked segregate analysis-RNA sequencing (BSR) to isolate PPR278. Endosperm and embryos from M3 generation kernels at 12 days after pollination (DAP) were used to construct RNA-seq libraries of WT and mutant-type pools. High-quality single nucleotide polymorphism (SNP) markers were called by aligning RNA-seq reads to the B73 reference genome (V4). We identified 32 candidate genes including 14 missense variants, 10 synonymous variants, 5 3ʹ untranslated region (UTR) variants, 2 5ʹUTR variants, and 1 stop codon-gained variant (Zm00001d015156) (Table S1). Zm00001d015156 (located on chromosome 5) that had the highest ΔSNP index among missense variants and the stop codon-gained variant [48]; therefore, we determined Zm00001d015156 as the candidate gene corresponding to PPR278. Sequence analysis revealed that Zm00001d015156 had one SNP (C-to-T mutation) in the ppr278-1 mutant. This SNP was located in the 15th exon at +2269 bp from the ATG start codon, causing a premature stop codon in Zm00001d015156 (Figure 2A). To test whether the mutation in Zm00001d015156 causes the ppr278-1 phenotypes, we performed co-segregation analysis by genotyping a larger segregating population (281 kernels) and observed that the SNP in the ppr278-1 mutant co-segregated with the low seed germination phenotype ( Figure S2). To further test the candidate gene, we

PPR278 Encodes a Constitutively Expressed PPR Protein
We used bulked segregate analysis-RNA sequencing (BSR) to isolate PPR278. Endosperm and embryos from M3 generation kernels at 12 days after pollination (DAP) were used to construct RNA-seq libraries of WT and mutant-type pools. High-quality single nucleotide polymorphism (SNP) markers were called by aligning RNA-seq reads to the B73 reference genome (V4). We identified 32 candidate genes including 14 missense variants, 10 synonymous variants, 5 3 untranslated region (UTR) variants, 2 5 UTR variants, and 1 stop codon-gained variant (Zm00001d015156) (Table S1). Zm00001d015156 (located on chromosome 5) that had the highest ∆SNP index among missense variants and the stop codon-gained variant [48]; therefore, we determined Zm00001d015156 as the candidate gene corresponding to PPR278. Sequence analysis revealed that Zm00001d015156 had one SNP (C-to-T mutation) in the ppr278-1 mutant. This SNP was located in the 15th exon at +2269 bp from the ATG start codon, causing a premature stop codon in Zm00001d015156 (Figure 2A). To test whether the mutation in Zm00001d015156 causes the ppr278-1 phenotypes, we performed co-segregation analysis by genotyping a larger segregating population (281 kernels) and observed that the SNP in the ppr278-1 mutant co-segregated with the low seed germination phenotype ( Figure S2). To further test the candidate gene, we screened another mutant (hereafter referred to as ppr278-2) from the mutant libraries, which showed a similar phenotype as the ppr278-1 mutant (Figures S1A and S3). We identified a SNP (C-to-T mutation) at +2404 bp in the 15th exon of Zm00001d015156 in the ppr278-2 mutant, also resulting in a premature stop codon ( Figure 2A). Crossing, self-pollinating, and subsequent genetic analyses confirmed ppr278-1 and ppr278-2 as allelic mutants, revealing that Zm00001d015156 is the causal gene for PPR278 ( Figure 2B). Expression analysis showed that PPR278 was expressed in multiple tissues, with the highest expression level in 6-DAP kernels ( Figure 2D). Thus, we cloned PPR278, encoding a constitutively expressed PPR protein, and confirmed that loss-of-function alleles of PPR278 caused the abnormal phenotype mentioned above. screened another mutant (hereafter referred to as ppr278-2) from the mutant libraries, which showed a similar phenotype as the ppr278-1 mutant (Figures S1A and S3). We identified a SNP (C-to-T mutation) at +2404 bp in the 15th exon of Zm00001d015156 in the ppr278-2 mutant, also resulting in a premature stop codon ( Figure 2A). Crossing, self-pollinating, and subsequent genetic analyses confirmed ppr278-1 and ppr278-2 as allelic mutants, revealing that Zm00001d015156 is the causal gene for PPR278 ( Figure 2B). Expression analysis showed that PPR278 was expressed in multiple tissues, with the highest expression level in 6-DAP kernels ( Figure 2D). Thus, we cloned PPR278, encoding a constitutively expressed PPR protein, and confirmed that loss-of-function alleles of PPR278 caused the abnormal phenotype mentioned above.

PPR278 Belongs to the p-Subfamily and Dually Localizes in Cytoplasm and Nucleus
PPR protein sequences were downloaded from either NCBI (https://www.ncbi.nlm.nih.gov/) (Accessed on 9 March 2022) or PPR energy (https://ppr.plantenergy.uwa.edu.au/) (Accessed on 9 March 2022) databases, and fulllength protein sequences were used for phylogenetic analysis, in which PPR278 was clustered into the P-subfamily ( Figure S4A) [49,50]. The prediction of PPR energy showed that PPR278 and its Arabidopsis homolog AT4G21880.1 grouped into the P-subfamily clade [42]. Further analysis indicated that PPR278 contained 11 p motifs ( Figures 2C, S4B and S5), and amino acid numbers of these motifs ranged from 32 to 37, which differed from typical P-subfamily PPR proteins ( Figure S6).

PPR278 Belongs to the p-Subfamily and Dually Localizes in Cytoplasm and Nucleus
PPR protein sequences were downloaded from either NCBI (https://www.ncbi.nlm.nih. gov/) (Accessed on 9 March 2022) or PPR energy (https://ppr.plantenergy.uwa.edu.au/) (Accessed on 9 March 2022) databases, and full-length protein sequences were used for phylogenetic analysis, in which PPR278 was clustered into the P-subfamily ( Figure S4A) [49,50]. The prediction of PPR energy showed that PPR278 and its Arabidopsis homolog AT4G21880.1 grouped into the P-subfamily clade [42]. Further analysis indicated that PPR278 contained 11 p motifs (Figures 2C, S4B and S5), and amino acid numbers of these motifs ranged from 32 to 37, which differed from typical P-subfamily PPR proteins ( Figure S6).
Most PPR proteins target to mitochondria and/or plastids in plants, whereas only a few are predicted to localize in nuclei [4,5,42]. Based on Csbio (http://www.csbio.sjtu.edu.cn/ bioinf/Cell-PLoc-2/) (Accessed on 9 March 2022) and Cello (http://cello.life.nctu.edu.tw/) (Accessed on 9 March 2022), PPR278 was predicted to be in the nucleus and cytoplasm ( Figure S2C). Subcellular localization experiments using onion (Allium cepa L.) epidermal cells and maize protoplasts revealed that PPR278 targeted to both nuclei and cytoplasm, suggesting that the functions of PPR278 are more complex than those of other PPR proteins targeted to mitochondria or/and plastids ( Figure 3A,B). Moreover, the secondary structure of PPR278 consists of numerous paired anti-parallel alpha helices, as predicted by the AlphaFold Protein Structure Database (http://www.alphafold.ebi.ac.uk) (Accessed on 9 March 2022) ( Figure 3C), which form the helix-turn-helix motifs necessary for RNA binding [5,51]. Therefore, even though PPR278 is a P-subfamily member, it contains 11 noncanonical P-type PPR repeats, which is a special feature. Another feature of PPR278 is that it is double-targeted in mitochondria and the nucleus. Like other PPR proteins, a series of paired anti-parallel alpha helices of PPR278 provide the ability to bind specific target RNA.
Most PPR proteins target to mitochondria and/or plastids in plants, whereas only a few are predicted to localize in nuclei [4,5,42]. Based on Csbio (http://www.csbio.sjtu.edu.cn/bioinf/Cell-PLoc-2/) (Accessed on 9 March 2022) and Cello (http://cello.life.nctu.edu.tw/) (Accessed on 9 March 2022), PPR278 was predicted to be in the nucleus and cytoplasm ( Figure  S2C). Subcellular localization experiments using onion (Allium cepa L.) epidermal cells and maize protoplasts revealed that PPR278 targeted to both nuclei and cytoplasm, suggesting that the functions of PPR278 are more complex than those of other PPR proteins targeted to mitochondria or/and plastids ( Figure 3A,B). Moreover, the secondary structure of PPR278 consists of numerous paired anti-parallel alpha helices, as predicted by the AlphaFold Protein Structure Database (http://www.alphafold.ebi.ac.uk) (Accessed on 9 March 2022) ( Figure 3C), which form the helix-turn-helix motifs necessary for RNA binding [5,51]. Therefore, even though PPR278 is a P-subfamily member, it contains 11 non-canonical P-type PPR repeats, which is a special feature. Another feature of PPR278 is that it is double-targeted in mitochondria and the nucleus. Like other PPR proteins, a series of paired anti-parallel alpha helices of PPR278 provide the ability to bind specific target RNA.  PPR proteins are RNA-binding proteins involved in RNA metabolism, including intron splicing and RNA editing, in mitochondria and plastids [3]. To test whether PPR278 is involved in cis-splicing in mitochondria, we performed semi-quantitative reverse transcription PCR (RT-PCR) analysis between mutant and WT plants for 35 mitochondrial genes using 12-DAP kernels. We observed expression changes in four genes, rps2A, nad5, ccmF N , and apocytochrome b (cob), between ppr278-1 and WT plants ( Figure 4A). Additionally, RT-PCR analysis showed that the cis-splicing efficiency of nad2 intron 4 (1383 bp) and nad5 introns 1 (870 bp) and 4 (952 bp) was affected in the ppr278-1 and ppr278-2 mutants ( Figures 4B,C and S7A). We detected no differences at the transcript level or in cis-splicing for the other mitochondrial genes examined, including 25 tRNAs and 3 rRNAs between ppr278 and WT plants ( Figure S7B). PPR278 is not only involved in mitochondrial gene expression (rps2A, nad5, ccmF N , and cob), but it is also necessary for the cis-splicing of nad2 intron 4 and nad5 introns 1 and 4. marker (bottom panels of (A,B)). Bar = 50 μm (A) and 5 μm (B). (C) Predicted protein structure of PPR278. The purple and red helices indicate the N terminus (N) and C terminus (C) of PPR278, respectively.

PPR278 Is Essential for the Cis-Splicing of nad2 Intron 4 and nad5 Introns 1 and 4
PPR proteins are RNA-binding proteins involved in RNA metabolism, including intron splicing and RNA editing, in mitochondria and plastids [3]. To test whether PPR278 is involved in cis-splicing in mitochondria, we performed semi-quantitative reverse transcription PCR (RT-PCR) analysis between mutant and WT plants for 35 mitochondrial genes using 12-DAP kernels. We observed expression changes in four genes, rps2A, nad5, ccmFN, and apocytochrome b (cob), between ppr278-1 and WT plants ( Figure 4A). Additionally, RT-PCR analysis showed that the cis-splicing efficiency of nad2 intron 4 (1383 bp) and nad5 introns 1 (870 bp) and 4 (952 bp) was affected in the ppr278-1 and ppr278-2 mutants ( Figures 4B,C and S7A). We detected no differences at the transcript level or in cis-splicing for the other mitochondrial genes examined, including 25 tRNAs and 3 rRNAs between ppr278 and WT plants ( Figure S7B). PPR278 is not only involved in mitochondrial gene expression (rps2A, nad5, ccmFN, and cob), but it is also necessary for the cis-splicing of nad2 intron 4 and nad5 introns 1 and 4.

PPR278
Is Required for C-to-U RNA Editing in 10 Mitochondrial Transcripts RNA editing is a major function of PPR proteins in mitochondria, and nearly 500 Cto-U RNA editing sites in mitochondrion-encoded transcripts were reported in the maize inbred line B73 [35]. Therefore, we examined the editing profiles of the 35 mitochondrion-encoded transcripts in the ppr278 mutants and identified 94 C-to-U RNA editing events (Table 1, Figures S8-S10). We analyzed the C-to-U RNA editing efficiency by directly comparing editing profiles in WT and ppr278 plants and observed decreased efficiency of 67 C-to-U RNA editing sites in seven mitochondrial transcripts and increased editing efficiency in one editing site in rps2B-550 in the ppr278 mutant (Table 1, Figures S8-S10). However, we could not determine the editing efficiency of the remaining 26 C-to-U editing sites in atp8 and nad3 via the peak of C and/or U (Figures 5A and S9A). To understand how atp8 and nad3 are affected in the ppr278 mutant, we sequenced~40 independent clones from WT and ppr278 plants and analyzed editing efficiency (T/(T + C)%). The T/(T + C) ratio was reduced among all editing sites in atp8 and nad3 in the ppr278 mutant ( Figures 5B and S9B). These results indicated that PPR278 is required for editing 94 C-to-U sites of mitochondrial transcripts related to mitochondrial complexes I, IV, and V, cytochrome c biogenesis, and ribosomal proteins ( Figure S11A).
To further investigate the effects of aberrant C-to-U RNA editing on protein translation, we investigated the coding sequence and corresponding amino acid sequences of all 35 mitochondrion-encoded transcripts in the ppr278 mutant. Nearly 83.16% of C-to-U RNA editing events caused amino acid substitutions ( Figure S11A). Notably, the replacement of leucine accounted for more than 40% of all editing events, ranking highest in silenced and non-silenced sites ( Figure S11B) [35]. We also discovered that editing failed at 37 bases distributed among atp1, atp4, and ccmF N in the ppr278-1 and ppr278-2 mutants (Table 1). We analyzed the failed editing at these bases in the ppr278 mutants and observed that editing at all bases except for ccmF N -417 and atp1-30 caused amino acid changes. The other failed editing sites in ccmF N resulted in amino acid substitutions at positions 11, 9, 4, 4, 3, 2, and 1, corresponding to Leu-to-Pro, Leu-to-Ser, Try-to-Arg, Phe-to-Ser, Ser-to-Pro, Cys-to-Arg, and Phe-to-Leu, respectively ( Figure S10, Table 1). ccmF in Escherichia coli, the homolog of maize ccmF N , encodes a large protein with 11 transmembrane domains [52]. We used TMHMM (http://services.healthtech.dtu.dk/service.php?TMHMM-2.0) (Accessed on 9 March 2022) to analyze the transmembrane domains of ccmF N and determined that the lack of editing in ccmF N disrupted the 1st, 4th, 5th, and 6th transmembrane domains in the ppr278 mutant ( Figure S10C). We also analyzed the amino acid conservation of C-to-U editing sites of atp8, nad3, and ccmF N , which indicated that these sites were relatively conserved in terms of amino acids change ( Figures 5, S9C and S10D). In total, PPR278 is essential for C-to-U RNA editing in 10 mitochondrial transcripts, including 37 abolished sites in atp1, atp4, and ccmF N . Loss of function of PPR278 lead to a series of amino acid substitution events due to failed C-to-U RNA editing.

Mutation of PPR278 Alters the Expression of Mitochondrial Function-Related Genes
Considering the nuclear localization of PPR278 and its implication in nuclear gene regulation, we performed RNA-seq analysis using 18-DAP endosperms and embryos from ppr278-1 and WT plants. The materials were collected from segregating ears with two biological replicates ( Figure 6A). Those with a fold change and p-value above the threshold (fold change > 2 and p-value < 0.05) were identified as differentially expressed genes (DEGs). We identified 2466 DEGs between ppr278-1 and WT plants, including 1542 up-regulated DEGs and 924 down-regulated DEGs ( Figure 6B, Tables S3 and S4).

PPR278 Functions in RNA Cis-Splicing and RNA Editing
P-subfamily PPR proteins are generally involved in RNA splicing [4]. However, three PPR proteins of the P-subfamily function in C-to-U RNA editing in Arabidopsis. For example, PPR596 influences editing efficiency at partial editing sites (rps3eU64RW, rps3eU603FF, rps3eU887SL, rps3eU1344SS, rps3eU1352PL, rps3eU1470SS, rps3eU1534RC, rps3eU1571AV, rps3eU1580SF, rps3eU1598SL, Pseudo-rps14eU99, and Pseudo-rps14Eu194) in mitochondrial transcript rps3 and rps14, while another gene P-type protein, PPR-MODULATING EDITING (PPME), is responsible for C-to-U RNA editing at both sites nad1-898 and nad1-937 [53][54][55]. Here, we characterized PPR278 as a splicing factor that participates in intron splicing of nad2 intron 4 and nad5 introns 1 and 4. In addition, PPR278 was also recognized as an editing factor that is responsible for 94 sites of C-to-U RNA editing in mitochondrial transcripts.
PPR278 is necessary for intron splicing of nad2 intron 4 and nad5 introns 1 and 4, which further highlights the role of P-subfamily PPR proteins in intron splicing. Multiple PPR proteins and other splicing factors may participate in splicing the same intron. For instance, nad2 intron 4 cis-splicing was also abnormal in emp16 and emp12 mutants in maize [9,17]. ZmSMK9 is involved in the cis-splicing of nad5 introns 1 and 4 [18]. Furthermore, EMP25, PPR101, and PPR231 are required for splicing nad5 intron 1 in maize. Thus, mutations in these PPR proteins affect the cis-splicing of mitochondrial nad5 introns [18,21,22]. Similarly, EMP11, DEK2, and EMP8 are associated with nad1 intron splicing [12][13][14][15][16]. In addition, EMP10 and DEK37 are necessary for cis-splicing of nad2 intron 1 [11,56]. These findings indicate that, in its function as a spicing factor, PPR278 may function in spliceosome complexes in mitochondria, possibly in complexes with these other factors that also affect splicing of the same introns.
In addition to its splicing role-and unlike previously reported P-type PPR proteins-PPR278 functions in C-to-U RNA editing in genes associated with other mitochondrial complexes besides complex III ( Figure S11A), particularly in ccmF N with 30 abolished C-to-U RNA editing sites in the ppr278 mutant ( Figure S10, Table 1). Failed editing of ccmF N abolished the formation of several transmembrane domains ( Figure S10C), especially the formation of the 5th transmembrane domain, which is important for protein conformation [32]. ccmF N -302 and ccmF N -1553 are essential for C-to-U editing and the biogenesis of cytochrome c ( Figure S10A) [31,32,36]. Additionally, a conserved WWD domain (G [2X] W [2 or 4X] WG [2X] WXWD) consisting of two W-to-R substitutions in sites ccmF N -1357 and ccmF N -1375 is required for accessibility for heme b attachment to cytochromes c1 and c [26,32]. Another conserved histidine (H) in site ccmF N -1342, called PHis2 involved in heme binding, was also fully unedited in the ppr278 mutant [57,58] (Figure S10B). Moreover, atp1-1292 and atp4-59 sites are important for the assembly of mitochondrial complex V [31,32,35,36]. Thus, loss of PPR278 function led to 37 abolished C-to-U RNA editing sites and an abnormal mitochondrial complex in maize (Table 1).
In general, P-subfamily members primarily function in splicing and PLS-subfamily members primarily function in editing. PPR278 is the first P-subfamily member determined to be involved in intron splicing of mitochondrial transcripts and C-to-U RNA editing in maize. However, several other PPR proteins act simultaneously as splicing and editing factors in plants. The E-type PPR protein DEK55 plays a crucial role in RNA editing at multiple sites as well as in the splicing of nad1 and nad4 introns in maize, and the DYW-type PPR protein BLX (Baili Xi) controlling mitochondrial RNA editing and splicing is essential for seed development in Arabidopsis [15,16,19,59]. These discoveries have expanded our knowledge about PPR proteins: The function of a PPR protein is not dependent on its classification category. There is a distinguishable divergence function in P-subfamily and PLS-subfamily members, and PPR278 clustered in the P-family by phylogenetic analysis ( Figure S4A). The irregular motif p and classification of PPR278 provide a plausible explanation for why PPR278 has both splicing and editing functions in mitochondria. Notably, no C-to-U editing sites were detected in nad2 and nad5 cDNA sequences in the ppr278 mutant, suggesting that nad2 intron 4 and nad5 introns 1 and 4 have no effect on the editing of the exon sequence. This implied that PPR278 might conduct C-to-U RNA editing and intron splicing independently [9].

PPR278 Is Involved in Regulation of Nuclear Gene Expression
Several nucleus-localized PPR proteins are involved in gene regulation [4,5,42]. These nucleus-localized PPR proteins (SOAR1, GRP23, PNM1, OsNPPR1, and OsNPPR3) interact with transcription factors and/or RNA polymerases to regulate the transcription and/or mRNA processing of nuclear genes associated with mitochondrial function or localized to mitochondria, and likely function as potential regulators of gene expression during embryogenesis in plants [43][44][45][46][47]. In maize, DEK43 targets to nuclei, but its functions in nuclei remain largely unknown [19]. In our study, PPR278 localized to nuclei ( Figure 3A,B), and RNA-seq analysis showed that the expression of many genes was significantly changed in the ppr278-1 mutant ( Figure 6B).
Genes related to post-embryonic development were dramatically down-and upregulated in ppr278 mutants ( Figure 6C). Moreover, genes associated with plant and seed development were widely down-regulated in the ppr278-1 mutant. This down-regulation might explain the lower germination rate and seedling-lethal phenotype of the ppr278 mutant. Notably, GO terms GO:0006334 (nucleosome assembly essential for a series of energy-consuming biological processes [60]) and GO:0045735 (nutrient reservoir activity associated with kernel development and filling) were also significantly enriched in the dek2, dek10, dek35, dek37, and dek41 mutants, characterized by defective kernels and abnormal mitochondrial function [10][11][12][13]61]. The down-regulated genes in ppr278-1 plants were enriched for both kinds of these GO terms. In addition, 7 and 23 DEGs were highly associated with the GO terms 'spliceosome' and 'mitochondrial respiratory chain', respectively, and exhibited extensive up-regulation ( Figure 6C, Table S4). This enrichment suggests that the mitochondrial oxidation respiratory chain was severely disrupted in ppr278 mutants. It remains to be tested whether the altered expression of nucleus-encoded genes arises from PPR278 binding their transcripts directly in the nucleus or if it is a consequence of arrested mitochondrial function. Based on our results, we conclude that PPR278 has a precise regulatory role in the expression of mitochondrial and nuclear genes and might function in transcription and processing of nuclear mRNA.

Plant Materials
The maize (Zea mays L.) ppr278-1 and ppr278-2 mutants were obtained from our EMSbased mutant library. In brief, the mature pollen of inbred line B73 was treated with 0.015% ethyl methanesulfonate solution (vol/vol) in a 1:10 pollen:solution ratio for 30 min. The resulting M1 plants were self-pollinated, and M2 seeds showing a Mendelian segregation ratio were selected. M2 plants were further self-pollinated to obtain M3 progenies for bulked pools. The 12-DAP embryos and endosperms from 20 WT and 20 mutant-type kernels from segregating M3 ears were used for bulked segregant analysis. The plants were grown in the experimental field at Shangzhuang Experimental Base of China Agricultural University under natural conditions (Beijing, China).

Bulked Segregant Analysis
The 12-DAP embryos and endosperms samples were ground in liquid nitrogen, and total RNA was extracted with TRIzol Reagent (Thermo Fisher SCIENTIFIC, 15596018, Waltham, MA, USA) according to the manufacturer's instructions. The cDNA library was prepared with VAHTS Stranded mRNA-seq Library Prep Kit for Illumina V2 (Vazyme, NR612, Nanjing, China) and sequenced on the Illumina X-ten platform. Raw reads were aligned to the B73 reference genome (V4) using HISAT2. The alignment files were converted to BAM format and sorted using Samtools. SNP calling, including C-to-T and G-to-A mutations, was performed by Samtools and Bcftools using unique reads. High-confidence SNPs were selected using the following criteria: (1) the SNP was supported by at least five reads, and (2) the SNP was detected in both the mutant pool and its corresponding WT pool. Functional annotation of SNPs was performed using SnpEff to screen SNPs between the two RNA pools. The SNP index was listed in Table S1.

Candidate Gene Analysis and Validation
We identified 14 genes with missense mutations, 10 genes with synonymous mutations, 5 genes with 3 UTR mutations, 2 genes with 5 UTR mutations, and 1 gene with stop codongained mutation (Zm00001d015156) in WT and mutant pools from our bulked segregant analysis. Then the ∆SNP index (SNP index difference between the mutant bulk and WT for each SNP) was calculated for each SNP. Zm00001d015156 was considered as candidate gene because of its mutant type and the highest SNP index among these genes. To test the degree of linkage between the candidate gene Zm00001d015156 and the mutant phenotype in the M3 segregating population, we designed primers PPR278-1F/R and PPR278-2F/R closely linked with the mutant sites (Table S5). Next, we used heterozygous ppr278-1 and ppr278-2 plants for reciprocal pollination followed by genetic analysis using the kernels from the pollinated ears ( Figure 2B).

Subcellular Localization of PPR278
The full-length PPR278 coding sequence without the stop codon was inserted into the vectors pSuper1300::eGFP and pSPYNE-35S::eGFP. Approximately 1 µg of the recombinant plasmid PPR278::GFP was transformed into onion (Allium cepa L.) epidermal cells and maize protoplasts. After culture for 20 h at 28 • C, the fluorescence signals were detected by a LSM710 confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany). AtAHL22-mCherry was used as a nuclear marker [62].

Phylogenetic Analysis
Sequences of PPR278 and its orthologous proteins were downloaded from either NCBI (Bethesda, MD, USA) or PPR energy databases (Perth, Australia) [49,50]. The full-length protein sequences were used to construct a phylogenetic tree with the neighbor-joining method using Mega-X software.

RT-PCR and qRT-PCR
Total RNA was treated with RQ1-free DNase I (Promega, M610A, Madison, WI, USA) to completely remove DNA contamination. The cDNA was synthesized using oligo (dT) primers (Promega, C1181, Madison, WI, USA). Quantitative reverse transcription PCR (qRT-PCR) was performed using TB Green Premix Ex Taq (Takara, RR420A, Toyobo, Osaka, Japan) with an ABI 7500 system (Thermo Fisher SCIENTIFIC, Waltham, MA, USA), and relative gene expression was calculated with the 2 −∆∆Ct method as described previously [63]. The data were normalized using ZmActin as a control. The experiments were replicated two times, and the primers used in the qRT-PCR (PPR278-764F and PPR278-1003R) are listed in Table S5.

Analysis of RNA Splicing by PPR278
Total RNA was extracted and reverse transcribed with random primers as described above. The expression differences of maize mitochondrial transcripts between WT and ppr278-1 plants were compared by PCR using 18s RNA as the reference gene. The primers used for mitochondrial genes were described previously [31]. The expression levels in WT and ppr278-1 plants were normalized using ZmActin. The PCR reaction buffer and DNA polymerase were from Mei5bio (MF743-10, Beijing, China). To further analyze RNA splicing, we chose the primers covering group II introns as described previously [9].

Analysis of RNA Editing by PPR278
Mitochondrial RNA was reverse transcribed with random primers as described above. The full-length sequences of 35 mitochondrial transcripts in ppr278-1 and ppr278-2 kernels at 12-DAP were reverse transcribed and directly sequenced as described previously [31]. The C-to-U RNA editing status of mitochondrial transcripts from WT and ppr278-1 plants was compared. The editing sites identified in the ppr278-1 mutant were confirmed in the ppr278-2 mutant. The editing efficiency of atp8 and nad3 in WT, ppr278-1, and ppr278-2 plants was estimated by cloning the atp8 and nad3 coding sequences into vector B-zero (TransGen Biotech pEASY-Blunt Zero Cloning Kit, CB501-01, Beijing, China) and sequencing more than 40 independent colonies for each gene.

RNA-Sequencing Analysis
Total RNA of 18-DAP endosperm and embryo (4 µg) tissues was extracted as described above, and two ppr278-1 or WT biological samples were collected for construction of the RNA-sequencing library. The cDNA library was performed according to Hieff NGS MaxUp II Dual-mode mRNA Library Prep Kit for Illumina (Yeasen, H9001360, Beijing, China) and sequenced by the Illumina NovaSeq platform for 150-nucleotide paired-end reads. Raw reads were aligned to the B73 reference genome (V4) using HISAT2. Data were normalized as fragments per kilobase of exon per million fragments mapped (FPKM), since the sensitivity of RNA-seq depends on the transcript length. The significant DEGs were filtered with the criterion: fold change > 2 and p-value < 0.05.