Genetic and phenotypic analysis of the defective kernel 55-1 (dek55-1) mutant
A mutant with a defective kernel phenotype was isolated from an ethylmethanesulfonate-induced maize B73 background population and was subsequently named dek55-1. The dek55-1 kernels segregated from self-pollinated progeny of dek55-1/+ heterozygotes at a 1:3 mendelian ratio (Fig. 1a, Additional file 1: Table S1). These results suggested that dek55-1, which exhibits a recessive phenotype, is caused by a monogenic mutation, which was confirmed in other populations generated from dek55-1/+ heterozygotes crossed with C733 or S162 inbred lines (Additional file 1: Table S1).
The dek55-1 kernels were smaller and presented a whitish pericarp, and they could be distinguished from the wild-type (WT) kernels at 15 days after pollination (DAP) (Fig. 1a). At maturity, the dek55-1 kernels were much smaller and shrivelled (Fig. 1b, c). To further determine the mutant phenotype, both WT and dek55-1 kernels were longitudinally sliced at 15 DAP. Compared to the WT kernels, the mutant kernels had a small, soft endosperm. (Fig. 1d, e). Furthermore, compared with the WT kernels, the dek55-1 kernels contained a smaller mature embryo and a reduced proportion of hard endosperm (Fig. 1f, g). In addition, the weight of dek55-1 kernel was approximately 70% lower than that of WT kernels (Fig. 1h). No dek55-1 seeds (0/100) germinated under field conditions, implying that embryo arrest is lethal in the mutants.
To further investigate the developmental structure of dek55-1 kernels, we examined the kernel tissue structure of the WT and dek55-1 mutant at 12 and 18 DAP (Fig. 1i-l). At 12 DAP, the dek55-1 embryo had only a small scutellum whose development was arrested at the coleoptile stage and a large interspace between the endosperm and the seed coat. In contrast, the WT embryo contained a visible coleoptile, a shoot apical meristem, a scutellum, and two leaf primordia, and the kernel was filled with endosperm cells (Fig. 1i, j). At 18 DAP, the WT embryo had developed complete structures, including four leaf primordia, a shoot apical meristem, and a clearly visible root apical meristem (Fig. 1k), while the dek55-1 embryos presented only a single leaf primordium (Fig. 1l). In addition, fewer starch grains accumulated in the dek55-1 endosperm cells than in the WT endosperm cells at this stage (Fig. 1k, l), and a cavity was observed in the dek55-1 endosperm (Fig. 1l). Taken together, these results indicate that developmental defects in the embryo and endosperm had occurred in the dek55-1 mutant.
Map-based cloning of DEK55
To identify the DEK55 gene, we applied the classic map-based cloning strategy to identify filial 2 (F2) mutant kernels, which segregated from a self-pollinated F1 hybrid ear. Four genomic DNA pools (10 mutant kernels per pool) and the DNA of both of the parents were used to identify the chromosomal location of the DEK55 gene. Six simple sequence repeat (SSR) markers on chromosome 5 were strongly correlated with defective kernel phenotypes, suggesting that the candidate gene may be on chromosome 5. Further analysis showed that the DEK55 gene is located between umc1705 and umc2302 on chromosome 5 (Fig. 2a). One thousand eight hundred and sixty-eight mutant kernels from the F2 population were genotyped to narrow the gene location by the use of six polymorphic molecular markers. The DEK55 gene was ultimately located on an approximately 1.29 Mb region between molecular marker 3 (M3) and M4 (Fig. 2a). There are 25 putative protein-coding genes in this region (http://ensembl.gramene.org/Zea_mays/Info/Index). To identify the mutated genes, the genomic DNA of 25 candidate genes was amplified and sequenced. Sequence alignment revealed a single-nucleotide polymorphism in the E-subgroup PPR protein gene (Zm00001d014471). In the dek55-1 mutant, nucleotide C was replaced with nucleotide T at +449 bp, resulting in the substitution of the amino acid serine (Ser) with phenylalanine (Phe). However, no change in the mRNA expression level was observed (Fig. 2a-d). To validate our results, we obtained a new mutant, dek55-2, from the maize ethylmethanesulfonate-induced mutant database [34]. The dek55-2 mutant showed a single-nucleotide mutation (G to A) at +729 bp (Fig. 2b), which led to a truncated protein (Fig. 2d). The mutant dek55-2 also produced defective kernels with a small white pericarp (Fig. 2e). An allelic test between dek55-1 and dek55-2 heterozygotes revealed that normal and mutant kernels segregated at the expected 3:1 ratio (normal/mutant; 450/143; P=0.62) in the F1 ear (Fig. 2e). For a control, all the kernels from the ear that were crossed between the dek55-2 heterozygote and WT were normal (Fig. 2e). Taken together, these results indicate that the mutation in the Zm00001d014471 PPR gene was responsible for the defective kernel phenotype, so the annotated gene was designated DEK55.
DEK55 is a mitochondrial E-subgroup PPR protein
Sequence alignment demonstrated that the DEK55 gene is a 1893 bp long ORF with no introns. Moreover, DEK55 encodes a 630 amino acid protein containing 13 PPR motifs and an E domain at the carboxy-terminal end (Fig. 2b-d and Additional file 1: Fig. S1). Mutations in dek55-1 and dek55-2 were located in the third and fifth PPR motifs, respectively (Fig. 2d). The mutation in dek55-2 resulted in a truncated DEK55 protein missing the last eight PPR motifs and the E domain.
To examine the subcellular localization of DEK55, a p35S:DEK55-enhanced green fluorescent protein (EGFP) vector was constructed and transformed into maize protoplasts. The fluorescent signal of DEK55-EGFP overlapped with that of MitoTracker (a mitochondrion-specific dye) (Fig. 3a), suggesting that, in maize, DEK55 is a mitochondrial PPR protein (Fig. 3a). In addition, expression analysis in various maize tissues demonstrated that DEK55 is relatively highly expressed in the roots, anthers, and ears, with relatively low expression in the stems, leaves, silk, tassels, and kernels (Fig. 3b).
DEK55 is involved in the C-to-U editing of 14 transcripts at multiple sites
PPR proteins usually participate in modifying organelle transcripts [10]. It has been reported that E-subgroup PPRs are involved in the C-to-U editing of mitochondrial pre-mRNAs [14, 32, 33]. To explore whether DEK55 is involved in this processing, the transcript levels of 35 maize mitochondrial genes that encode functional proteins were analysed in both WT and dek55-1 kernels. RNA editing of these transcripts in the dek55 (dek55-1 and dek55-2) and WT (WT-1 and WT-2) kernels was detected via the strand- and transcript-specific RNA sequencing (STS-PCRseq) strategy [35]. The sequencing reads were mapped to the 35 mitochondrial gene transcripts, and 482 C-to-U RNA editing sites were examined in the WT and dek55 kernels (Additional file 2: Table S1). The results revealed that, compared with that of these RNA editing sites between the WT and dek55 kernels (Additional file 2: Table S2), the C-to-U editing percentage of 31 editing sites in 14 transcripts (atp1, atp8, ccmFc, ccmFn, cob, mat-r, nad3, nad4, nad6, nad7, rps12-ct, rps12, rps13, and rps3) was significantly altered in the dek55 kernels (Fig. 4, Additional file 2: Tables S2-S4), whereas the editing percentage of 24 sites decreased (Fig. 4a) and that of seven sites increased in the dek55 kernels compared with WT kernels (Fig. 4b, Additional file 2: Tables S2 and S4). The editing efficiency at the atp1-1490, ccmFn-287, mat-r-1877, and rps13-56 sites dramatically decreased in the dek55-1 and dek55-2 kernels, and the editing percentage in the dek55 mutants was more than 50% lower than that in the WT kernels (Fig. 4a, Additional file 2: Table S3). Direct sequencing of reverse transcription-polymerase chain reaction (RT-PCR) products of the atp1-1490, ccmFn-287, mat-r-1877 and rps13-56 sites also indicated that the editing efficiency was significantly reduced in the dek55 kernels at these RNA editing sites (Fig. 4c). Deficient C-to-U RNA editing led to altered amino acid residues in dek55 (Fig. 4c). Moreover, at the atp8-123 site, the editing efficiency of only the dek55-2 kernels (5%) was more than 50% lower than that of the WT kernels, and at nad4-77 sites, the editing efficiency of only the dek55-1 kernels (24.2%) was more than 50% lower than that of the WT kernels (Fig. 4a). Taken together, the above results indicated that DEK55 is required for RNA editing at multiple sites, especially the atp1-1490, ccmFn-287, mat-r-1877, and rps13-56 sites.
DEK55 is essential for the trans-splicing of nad1 introns 1 and 4 and the cis-splicing of nad4 intron 1
The transcript levels of 35 maize mitochondrial genes were examined, and the results showed that the transcript levels of nad1 and nad4 were significantly downregulated in the dek55 mutants (Fig. 5a). The genomic DNA of nad1 contains four group II introns, and with the exception of the 2nd intron, all are trans-splicing introns. (Fig. 5c). The genomic DNA of nad4 has three cis-splicing introns (Fig. 5d) [13, 36]. The full maturation of nad1 and nad4 transcripts requires complete intron splicing. We therefore further analysed the intron splicing efficiency of nad1, nad4, and other genes in the dek55-1 and WT kernels via quantitative reverse transcription-polymerase chain reaction (qRT-PCR). Compared with that in the WT kernels, the splicing efficiency of the first and fourth introns of nad1 and the first intron of nad4 in the dek55-1 mutant kernels decreased (Fig. 5b). Furthermore, we amplified each intron and full transcript of nad1 and nad4 via RT-PCR (Fig. 5c, d). The transcript abundance of nad1 exons 1-2 and exons 4-5 and the full-length DNA fragments significantly decreased (Fig. 5c). RT-PCR could not amplify the intronic DNA fragments (1F+2R, 3F+4R, 4F+5R) in the dek55 and WT kernels because the 1st, 3rd, and 4th introns of nad1 are trans-spliced. (Fig. 5c). The unspliced 2nd intronic fragments of nad1 in the dek55 mutant kernels were similar to those in the WT kernels (Fig. 5c). The abundance of nad4 spliced exons 1-2 and full-length DNA fragments significantly decreased, and the abundance of nad4 unspliced intron 1 transcripts significantly increased (Fig. 5d). Our findings suggest that the significant decrease in abundance of nad4 and nad1 transcript in the dek55 mutant kernels was caused by the abnormal splicing of nad4 intron 1, nad1 intron 1, and intron 4 (Fig. 5a-d). Therefore, DEK55 is necessary for the trans-splicing of the two nad1 introns (1st and 4th) and cis-splicing of the first nad4 intron in maize.
The dek55-1 mutant exhibits reduced complex I activity and increased alternative respiratory pathway activity
Four genes, i.e., nad1, nad4, nad3, and nad6, encode the subunits of complex I NAD1, NAD4, NAD3, and NAD6, respectively [36]. The rps13 gene encodes a ribosomal protein, and atp1 and atp8 encode the ATPase subunit 1 and subunit 8 subunits of ATP synthase F1, respectively [36]. Defects in the posttranscriptional processing of these genes may impair the biosynthesis of mitochondrial complexes [17, 37-39]. We performed blue native polyacrylamide gel electrophoresis (BN-PAGE) and an in-gel NADH dehydrogenase activity assay to investigate the accumulation level and activity of mitochondrial complexes in WT and dek55-1 endosperm. BN-PAGE showed that the abundance of complex I and supercomplex I+III2 in the dek55-1 mutant significantly decreased (Fig. 6a). However, no significant differences for complex V were observed between the WT endosperm and dek55-1 endosperm (Fig. 6a). Furthermore, the activity of complexes I and I+III2 was reduced in the dek55-1 mutant (Fig. 6b). Taken together, these results indicate that defects in mitochondrial transcript splicing and/or editing might affect the abundance and activity of mitochondrial complex I.
The mitochondrial respiratory chain in plants includes the cytochrome c and alternative oxidase (AOX) pathways [40]. When the main cytochrome c pathway is blocked, AOX activity can be increased to compensate for the respiratory pathway [41]. In dek55-1, the functions of complex I were abolished (Fig. 6a, b). Thus, we performed qRT-PCR to detect the expression levels of AOX genes in WT and dek55-1 kernels, and the results showed a 512-fold increase in the expression level of the AOX2 gene in the dek55-1 mutant kernels compared to the WT kernels. (Fig. 6c). Collectively, our results indicate that the respiratory pathway is severely blocked in dek55-1 mitochondria.
DEK55 interacts with ZmMORF1 and ZmMORF8 in yeast
Previous studies have shown that MORFs directly interact with PPR proteins and play a role in RNA editing at numerous editing sites [42, 43]. In the present study, DEK55 was found to be responsible for 31 RNA editing events in maize, so we speculated that DEK55 might interact with MORFs to form an editing complex involved in RNA editing in maize. Thus, we used MORFs in Arabidopsis as bait to search for putative MORFs in maize; seven putative MORFs were identified in maize (Fig. 7a). A yeast two-hybrid assay was performed to screen for MORFs that interact with DEK55, and the results indicated that DEK55 can interact with ZmMORF1 and ZmMORF8 in yeast (Fig. 7b).