The Zea mays Mutator-insertion allele of dek47 (dek47-1) was isolated from the UniformMu population (McCarty et al., 2005). The second allele of dek47 (dek47-2) was obtained from the Maize EMS-induced Mutant Database population (Maize EMS-induced Mutant Database)¹ (Lu et al., 2018). Dek47-1 (±) heterozygotes were identified by PCR with gene-specific primer Dek47-R2 and Mutator-specific Mu-TIR8 primer. The A. thaliana T-DNA insertion lines rug3-1 (SALK_092071) and rug3-2 (GABI_466G01) were obtained from the Arabidopsis Biological Resource Center. The maize plants were grown in the experimental field under natural conditions in Qingdao and Sanya, China. Arabidopsis and tobacco (N. benthamiana) plants were grown in pots under 22°C with a 16-h photoperiod and 25°C with a 12-h photoperiod, respectively.

For light microscopy, the immature kernels of dek47-1 and WT at 8 and 14 days after pollination (DAP) were sectioned from segregating ears for paraffin sections. Sections of the maize kernels were fixed in 4% (w/v) paraformaldehyde, dehydrated in ethanol, infiltrated and embedded in paraffin, de-paraffinized, stained with Safranin O, and imaged with a stereo microscope (Carl-Zeiss, Jena, Germany) as described previously (Zhang et al., 2013).

The full-length coding sequence of Dek47 was amplified using high-fidelity DNA polymerase KOD plus Neo (TOYOBO) and cloned into vector pENTR™/D-TOPO^® (pENTR^™ Directional TOPO^® Cloning Kits, Invitrogen), performed LR recombination reaction (Gateway^® LR Clonase^™ II Enzyme Mix, Invitrogen) with the binary vector pGWB5. For subcellular localization, DEK47-GFP was transiently expressed in tobacco epidermal cells using MitoTracker Red CMXRos (Invitrogen) as a marker of mitochondria with a working concentration of 100 nm, as described previously (Liu et al., 2013). Fluorescence signals were observed under a ZEISS LSM 880 confocal microscope (Carl-Zeiss, Jena, Germany).

For RNA isolation, total RNA was isolated from the embryo and endosperm of immature kernels of dek47-1 and WT at 12 DAP using the RNeasy Mini Kit (Qiagen, Germany). The RNA was treated with RNase-free DNase I (NEB) to remove potential DNA contaminants. Reverse transcription-PCR (RT-PCR) and Quantitative real-time PCR (qRT-PCR) were performed according to the manufacturer’s instructions (TransGen Company, Beijing, China). RNAs were normalized against the maize Actin gene (GRMZM2G126010) or the Arabidopsis Actin gene ACT2 (AT3G18780). Information of all the primers used in this study is listed in Supplementary Table S1.

The maize and Arabidopsis mitochondrial RNA splicing was analyzed as described previously (Xiu et al., 2016; Weissenberger et al., 2017). This analysis was done with three biological replicates.

Blue native-polyacrylamide gel electrophoresis (BN-PAGE) and detection of NADH dehydrogenase activity were performed basically as described (Meyer et al., 2009). Crude mitochondria from fresh wild-type and dek47-1 kernels were extracted and used for BN-PAGE using a Native-PAGE sample prep kit (Invitrogen) as described previously (Chen et al., 2016). The gel was stained with Coomassie brilliant blue (CBB) 250 and incubated in the reaction buffer (0.14 mM NADH, 1.22 mM NBT, and 0.1 M Tris–HCl, pH 7.4) to detect the NADH dehydrogenase activity. 130 μg of crude mitochondrial membrane protein extracts from wild-type (WT) and dek47-1 mutant maize kernels was loading in BN-PAGE gels.

Different quantities (2, 4, 8, and 8 μg) of crude mitochondrial membrane protein extracts from the developing kernels at 11 DAP were separated by 12.5% SDS-PAGE. After transferring to a nitrocellulose membrane (0.45 mm; Millipore), the membrane was incubated with primary antibodies using polyclonal rabbit antibodies against Nad9, Cyt_C, Cyt_C1, Cox2, ATPase, and AOX as described previously (Liu et al., 2020a). The HRP-conjugated goat anti-rabbit antibody (Abcam) was used as the secondary antibody. Signals were visualized on X-ray films (Kodak, Tokyo, Japan) using the ECL reagents (Pierce, Thermo Fisher Scientific, Waltham, MA, United States) according to the manufacturer’s instructions.

The respiration rates were determined according to the previous report with modifications (Wang et al., 2015). Respiration rates were indicated by O₂ consumption of the fresh maize kernels at 11 DAP using a Clarke-type Chlorolab II oxidative electrode (Hansatech).

To test the genetic complementation of the Arabidopsis rug3 with maize Dek47, the protein coding region of Dek47 was cloned in pENTR/D-TOPO and transferred to the binary vector pGWB2 using the gateway system. This placed the Dek47 gene under the CaMV35S promoter. This construct was delivered to the Agrobacterium strain EHA105 and transformed into the Arabidopsis rug3-1 and rug3-2 mutants.

The experiments were performed by using the Yeastmaker Yeast Transformation System 2 (Clontech). The Dek47 sequence with a truncation of the N-terminal 1–40 amino acids was cloned into the pGADT7-AD and pGBKT7-BD vectors used as prey or bait, respectively. These constructs were co-transfected into Y2H Gold strain and spotted onto the DDO (SD/−Leu/−Trp) medium and QDO (SD/−Ade/-His/−Leu/−Trp) medium.

The coding sequences of DEK47, PPR14, PPR20, PPR-SMR1, Zm-mTERF15, and Zm-mCSF1 were cloned into pCAMBIA1300-nLUC [N-terminal luciferase (NLUC)] and pCAMBIA1300-cLUC [C-terminal luciferase (CLUC)], respectively, according to a previous report (Chen et al., 2008). The 4-week-old tobacco leaves were infiltrated with Agrobacterium strain EHA105 containing different combinations of NLUC and CLUC fusion constructs. The combination of ZmMORF8-NLUC and ZmMORF1-CLUC was used as a positive control (Liu et al., 2020b). The luciferase signals were imaged using the Lumazone FA Pylon2048B system. The experiments were carried out with at least three replicates.

To reveal the function of the RCC1-like proteins in maize, we analyzed the Mu insertional mutants in RCC1-like genes in the UniformMu population in the W22 genetic background (McCarty et al., 2005). A Mu insertion line (UFMu-06802) in an RCC1-like gene (GRMZM2G114748) was chosen for further study because its progeny segregated defective kernels (dek), named dek47-1. As shown in Figures 1A,B, developing dek47-1 kernels were white and watery at 17 DAP and collapsed at maturity. The ratio between the dek47-1 and wild-type (WT) kernels in the self-pollinated ears was approximately 1:3 (Figures 1A,B), suggesting that dek47-1 is a monogenic recessive mutation. The dek47-1 kernels remained smaller and whiter than its siblings at 30 DAP (Figures 1C,D).

Maize embryo development is characterized by four stages, pre-embryo, transition, coleoptilar, and late embryogenesis. Endosperm development is characterized by four stages, coenocyte, cellularization, differentiation, and maturation (Olsen, 2001). Sectioning of the mature seeds showed that the dek47-1 kernel contained an abnormal embryo and a small endosperm (Figures 1E,F). The comparison between the mutant and WT in the same ear indicated that the embryo and endosperm development was inhibited in dek47-1. When the wild-type embryo developed into the coleoptilar stage at 8 DAP (Figures 1G,K), the dek47-1 embryo reached the pre-embryo stage (Figures 1H,L). When the WT embryo reached the late embryogenesis stage (Figure 1I), the dek47-1 embryo appeared to just enter the early coleoptilar stage with a scutellum structure and partial shoot or root meristems (Figures 1J,M). At 30 DAP, the dek47-1 embryo remains at the coleoptilar stage with abnormal cell differentiation (Figures 1E,F). The endosperm development in dek47-1 was delayed as well, and the inhibition of endosperm development in dek47-1 appeared throughout the whole development process, which is difficult to pinpoint to a specific stage. These results indicate that the embryo development is arrested at the coleoptilar stage and the endosperm development is delayed in dek47-1.

To determine whether the dek phenotype is indeed caused by mutation in GRMZM2G114748, we performed linkage analysis. The presence of the Mu insertion was verified by PCR with the Mu specific primer Mu-TIR8- and GRMZM2G114748-specific primers, and the homozygote was determined by a lack of wild-type GRMZM2G114748 and a presence of the Mu insertion. The results showed that the Mu insertion in GRMZM2G114748 is tightly linked to the dek47 phenotype in a population of 48 plants (Supplementary Figure S1). Sequencing result confirmed that the Dek47-1 mutant carries a Mu insertion in GRMZM2G114748 at 439 bp downstream from the translation start codon ATG (Figure 2A). To genetically confirm that GRMZM2G114748 is the causal gene, an independent allele was isolated from the ethyl methane sulfonate (EMS)-mutagenized B73 population (Lu et al., 2018), designated dek47-2. dek47-2 contains an EMS-induced G-to-A mutation at +858 bp downstream of the ATG start codon, causing an alteration from TGG (Trp/W) to TGA (stop codon). This mutation terminates the GRMZM2G114748 protein at the 286th amino acid residue, causing a truncation of 146 amino acid residues (Figure 2A). Similar to dek47-1, the selfed dek47-2/+ heterozygotes segregated a quarter of dek kernels with a phenotype similar to dek47-1 (Supplementary Figure S2A). The crossed ears between dek47-1/+ and dek47-2/+ heterozygotes produced ~25% dek kernels (Supplementary Figure S2B). Thus, we conclude that GRMZM2G114748 is the causal gene for the dek47 phenotype, hereafter referred to as Dek47.

Dek47 contains six exons and five introns, encoding a 45.44-kDa protein with 432 amino acid residues (Figure 2A). Protein domain analysis showed that DEK47 contains seven tandem RCC1 repeats using Prosite² (Figure 2A). Based on the classification by Hadjebi et al. (2008), DEK47 is a typical RCC1 subgroup protein with a character that the RCC1 repeats make up almost the entire protein. RT-PCR analyses showed that no wild-type Dek47 transcript can be detected in dek47-1, indicating that Dek47 is not expressed in dek47-1 (Supplementary Figure S3A), suggesting that dek47-1 is probably a null mutation. Quantitative real-time PCR (qRT-PCR) results showed that the transcript of Dek47 was detected in all tested tissues with high levels of expression in shoots, leaves, and silks, and low in roots, flowers, and kernels (Supplementary Figure S3B). These data indicate that Dek47 is a constitutive gene that may play an important role throughout maize growth and development, rather than a seed specific gene.

DEK47 is predicted to have a putative mitochondrial localization signal by TargetP.³ To determine its subcellular localization, the full-length Dek47 was fused to the N-terminus of GFP and the DEK47-GFP fusion was placed under the control of the CaMV 35S promoter in pGWB5. The fusion was transiently expressed in tobacco leaf epidermal cells via the Agrobacterium-mediated transformation. Strong green fluorescence signals were detected in mitochondria, which were merged with the red fluorescence signals of mitochondria stained by the MitoTracker (Figure 2B), indicating that DEK47 is localized in mitochondria.

To evaluate whether the mutation of DEK47 affects the mitochondrial respiration, three mitochondrial respiratory rates were measured, the total respiratory (V_t), cytochrome respiratory capacity (V_cyt), and alternative respiratory capacity (V_alt). Compared with WT, V_t was sharply decreased to a low level in dek47-1 (Table 1). The ratio of V_cyt/V_t was markedly decreased and V_alt/V_t strikingly increased in dek47-1, implying that the cytochrome respiration pathway is impaired and the alternative respiratory pathway (AOX) is activated as a result of the Dek47 mutation.

To determine whether the AOX is enhanced in dek47-1, we detected the protein abundance of AOX via Western blotting using the specific antibody against AOX. The results showed that the abundance of AOX was drastically increased in dek47-1, comparing with the nearly undetectable level of AOX protein in the WT (Figure 3C). There are three AOX genes: AOX1, AOX2, and AOX3 in maize (Karpova et al., 2002). RT-PCR results showed that all three AOX genes are expressed at undetectable levels in WT, whereas the expression of AOX2 was dramatically increased in the dek47 alleles (Supplementary Figure S4A). qRT-PCR analysis provided consistent results (Supplementary Figure S4B), indicating that the AOX pathway is enhanced in dek47.

Dysfunction of mitochondrial respiration can be caused by the defective electron transfer that involves mitochondrial complexes I to IV (Acin-Perez et al., 2008; Millar et al., 2011; Lobo-Jarne and Ugalde, 2018). To determine whether the mitochondrial complex is defective, we employed the BN-PAGE gel analysis and analyzed the complexes by using crude mitochondria isolated from the WT and dek47-1. The CBB staining showed that the complex I was markedly reduced, whereas the complexes III and V were increased in dek47-1 (Figure 3A). In-gel NBT-NADH activity staining results showed that the NADH dehydrogenase activity of complex I was dramatically reduced in the dek47-1 mutant as the activity of the super CI+CIII₂ complex was nearly undetectable and CI a fraction of the WT activity (Figure 3B). These data indicate that the mutation of Dek47 affects the assembly and activity of complex I.

To further assess the effects of the loss of Dek47 function on mitochondrial complexes, Western blot was performed to detect the abundance of the components of the mitochondrial complexes using antibodies against Nad9 (complex I), Cyt_C (cytochrome c), Cyt_C1 (complex III), Cox2 (complex IV), and ATPase (complex V). As shown in Figure 3C, the level of Cyt_C showed no difference between WT and dek47-1, the level of Nad9 was slightly decreased in dek47-1, whereas the levels of Cyt_C1, Cox2, and ATPase were increased in dek47-1, suggesting that the deficiency of complex I may promote an accumulation of complexes III, IV, and V in response.

The deficiency of mitochondrial complex I can be caused by a deficiency of the complex I components or the assembly factors (Klodmann et al., 2010). To investigate whether the mitochondrial complex components are affected, we analyzed the transcripts of 35 mitochondrion-encoded genes by RT-PCR and qRT-PCR analysis in the WT and dek47-1 kernels at 12 DAP. Results showed that the expression levels of the 34 mitochondrial genes were comparable between the WT and dek47-1, except for nad2 which was barely detectable in the mutant (Figures 4A,B).

The maize mitochondrial nad2 gene contains 5 exons and 4 introns. Primers were designed to anchor on the exons across each intron (Figure 5A). RT-PCR results indicated that the splicing of all four nad2 introns was reduced in dek47-1 with a substantial decrease in intron 3 (Figure 5B), suggesting that Dek47 is required for splicing of all four introns of nad2 transcript. To evaluate the impact on other introns, we compared the splicing efficiency of the mitochondrial 22 group II introns between the WT and dek47 mutants by qRT-PCR (Figure 5C). Results showed that the splicing efficiency of the four nad2 introns was decreased in dek47-1 compared with that in WT (Figure 5C), whereas the splicing of other introns appeared normal. These results are consistent with the RT-PCR results, suggesting that Dek47 is indeed required for the splicing of the four introns of nad2 transcript.

A previous study showed that RUG3 protein functions in the splicing of nad2 introns 2 and intron 3 in Arabidopsis mitochondria (Kuhn et al., 2011). BLAST analysis indicates that DEK47 shares a 63% sequence identity with AT5G60870 (RUG3) from Arabidopsis (Supplementary Figure S5). To address the functional relationship between DEK47 and RUG3, Dek47 was expressed in the rug3-1 and rug3-2 homozygotes to test whether DEK47 could rescue the Arabidopsis rug3 mutant phenotype. Two Arabidopsis T-DNA insertion lines rug3-1 (SALK_092071) and rug3-2 (GABI_466G01) were obtained from the Arabidopsis Biological Resource Center (Figure 6A). The two alleles of rug3 are defective in the intron splicing of nad2 and show developmentally delayed phenotypes (Kuhn et al., 2011). The genotypes of rug3 mutants and complemented lines Dek47/rug3-1 (Com1) and Dek47/rug3-2 (Com2) were confirmed by PCR analysis (Supplementary Figure S6). The rug3-1 and rug3-2 homozygous plants that carried the Dek47 transgene in Com1 and Com2 exhibited normal growth and development comparing with the wild-type (Figure 6A). qRT-PCR results revealed that the splicing efficiency of nad2 introns was largely restored in the complemented plants Com1 and Com2 (Figure 6B). These data suggest that DEK47 can complement RUG3 in Arabidopsis and is likely the ortholog of RUG3 in maize.

Previous studies reported that the splicing factors Zm-mTERF15, Zm-mCSF1, PPR14, PPR20, and PPR-SMR1 are involved in the splicing of mitochondrial nad2 intron 3 in maize (Chen et al., 2019; Wang et al., 2020; Yang et al., 2020). Thus, we tested whether DEK47 can physically interact with these proteins by Y2H (Figure 7A). Results showed that co-transformants between DEK47 and these splicing factors did not grow on QDO (SD/−Ade/-His/−Leu/−Trp) medium except the positive control (Figure 7B), suggesting that DEK47 may not directly interact with these splicing factors in the Y2H system. To independently test that result, the luciferase complementation imaging (LCI) assay was performed. DEK47, Zm-mTERF15, Zm-mCSF1, PPR14, PPR20, and PPR-SMR1 were fused to the NLUC and CLUC, respectively. Co-expression of the combination of ZmMORF8-NLUC and ZmMORF1-CLUC was used as a positive control (Liu et al., 2020b). No luciferase activity was detected in the tobacco leaves co-expressing DEK47 and the above splicing factors (Supplementary Figure S7). These results indicate that DEK47 may not physically interact with the related splicing factors.

Co-transformants of DEK47AD/DEK47BD did not grow on QDO medium (Figure 7B), and co-transfection of DEK47-NLUC and DEK47-CLUC failed to produce a visible luciferase activity signal (Supplementary Figure S7), indicating that DEK47 cannot form a dimer or multimer by itself in the Y2H and LCI system.

RCC1 motif is defined as a conserved domain consisted of 51–68 amino acid residues to fold into a bladed beta-propeller in many eukaryotes (Renault et al., 1998). The human RCC1 protein is involved in mitosis, nucleo-cytoplasmic transport, and nuclear-envelope assembly as a signal for chromosome position (Hetzer et al., 2002). In A. thaliana, four RCC1-domain proteins (UVR8, RUG3, TCF1, and SAB1) have been characterized (Brown et al., 2005; Kuhn et al., 2011; Ji et al., 2015, 2019). In maize, 31 putative RCC1-like proteins are identified; however, none of which have been characterized at the molecular level. In this study, we revealed the function of a RCC1 family protein DEK47 and its critical role in kernel development in maize.

Genetic evidence demonstrates that DEK47 loss of function arrests both embryo genesis and endosperm development, giving rise to a dek phenotype in maize (Figure 1). dek47-1 contains a Mu insertion in the second RCC1 motif, and dek47-2 contains a point mutation causing a translation stop codon in the fifth RCC1 motif. Both mutants exhibited the dek phenotype, and the crosses between the two alleles produced the dek phenotype as well. Molecular analyses indicated that the absence of DEK47 reduces the splicing efficiency of the four introns of nad2, especially the splicing of nad2 intron 3 in dek47 alleles (Figure 5C), suggesting that DEK47 is required for the splicing of four nad2 introns in maize mitochondria.

Nad2 is a central subunit of complex I on the mitochondrial membrane and essential to the assembly of complex I (Klodmann et al., 2010; Ligas et al., 2019). Previous studies have reported that the absence of Nad2 results in disassembly and severely reduced activity of complex I, leading to the delayed growth and abnormal seed development. For instance, loss-of-function of Emp16, Emp10, Emp12, Dek37, and PPR20 exhibited severely disassembly and impaired activity of mitochondrial complex I and arrested embryo and endosperm development in maize, leading to embryo lethality caused by the splicing deficiency of nad2 transcript (Xiu et al., 2016; Cai et al., 2017; Dai et al., 2018; Sun et al., 2019; Yang et al., 2020). Similar cases were also occurred in the dek47 mutants, implying the vital roles of DEK47 in mitochondrial complex I biogenesis and seed development in maize.

The OXPHOS pathway is the major respiration pathway in plant cells where electrons from NADH and FADH₂ are transferred via complex I to IV to O₂ to produce H₂O. Along with this electron transfer, protons are pumped out to the inner and outer membrane space. Proton flow from the high concentration space to the low concentration matrix drives mitochondrial complex V (ATPase) to synthesize ATP. Complex I is the entry point of the electron transfer chain. The mutation of DEK47 affects the splicing of the nad2 introns, leading to a deficiency of Nad2 protein and a disassembly of Complex I (Figure 3A). As such, it blocks the OXPHOS pathway and stimulates the expression of the AOX pathway. The deficiency of Nad2 blocks the respiration pathway is found in several mutants, such as emp16 (Xiu et al., 2016), emp8, and emp12 (Sun et al., 2018, 2019).

Alignment analysis indicates that DEK47 protein shares a 63% identity with RUG3 in Arabidopsis (Supplementary Figure S5), suggesting that these two are likely orthologs. At the molecular level, both DEK47 and RUG3 are required for the splicing of mitochondrial nad2 introns 2 and 3 (Kuhn et al., 2011), suggesting that DEK47 and RUG3 are functional orthologs. However, DEK47 is also required for the splicing of nad2 introns 1 and 4, which is not observed in RUG3 (Figure 5), suggesting that DEK47 may have gained additional function in facilitating the splicing of other introns in contrast to RUG3 in Arabidopsis. This functional divergence is common to orthologous pairs between maize and Arabidopsis, such as PPR2263/MEF9 and ZmDEK36/ATDEK36 (Sosso et al., 2012; Wang et al., 2017). PPR2263 functions in the editing of mitochondrial nad5-1,550 and cob-908 sites in maize, whereas MEF9 only participates in the editing at nad5-1550 site in Arabidopsis. Maize DEK36 is only required for mitochondrial RNA editing at multiple sites, whereas AtDEK36 is involved in RNA stabilization except for RNA editing. Thus, functions of orthologous RNA editing factors and splicing factors have diverged to cope with genetic alterations in mitochondria during the evolution of maize and Arabidopsis. In that reasoning, one possible explanation is possible that genetic alteration in the maize nad2 introns 1 and 4 invokes the recruitment of DEK47, but not in the case of Arabidopsis nad2 introns 1 and 4 for RUG3, and another one could be the difference in available splicing factors between the two species.

Previous studies reported that different types of splicing factors are required for intron splicing in plant mitochondria, such as maturases, CRM-domain protein, RNA DEAD helicases, PPR proteins, PORR domain family proteins, and RCC1 proteins (Brown et al., 2014). Among these splicing factors, PPR protein PPR14, PPR-SMR1, and CRM-domain protein Zm-mCSF1 interacts with each other to facilitate the splicing of nad2 intron 3 (Chen et al., 2019; Wang et al., 2020). However, how these factors coordinate to function in intron splicing is not clear. In this study, the splicing efficiency of four nad2 introns is decreased in dek47, whereas the splicing of other mitochondrial introns appears indistinguishable compared with WT (Figure 5C), suggesting that DEK47 is specifically required for the intron splicing of nad2. However, RCC1-like proteins have not been shown to possess the RNA-binding activity. Instead, RCC1-domain proteins is shown to directly interact with proteins and lipids to mediate diverse biological processes in animals (Hadjebi et al., 2008). The RCC1-domain proteins and the WD40-repeat proteins share a similar seven-bladed propeller in tertiary structure (Hadjebi et al., 2008). Surface electrostatics of seven-bladed WD structures is highly conserved. The seven-bladed WD beta-propellers have multiple binding partners and are involved in protein–protein interactions (Valeyev et al., 2008). Based on that, it is reasonable to speculate that the 7 RCC1 domains of DEK47 also form seven-bladed beta-propellers to participate in nad2 intron splicing through interacting with other proteins, such as the PPRs. Previous studies have identified five splicing factors that are involved in the splicing of nad2 intron 3 in maize, including PPR14, PPR20, PPR-SMR1, Zm-mCSF1, and Zm-mTERF15 (Chen et al., 2019; Wang et al., 2020; Yang et al., 2020). However, our protein interaction analyses showed that DEK47 did not physically interact with these splicing factors in the Y2H analysis and LCI assay (Figure 7, Supplementary Figure S7). It is possible that DEK47 may recruit other unknown factors to participate in the splicing of nad2 intron 3, or such interaction is too weak or transient to be detected by the Y2H and LCI assays.

We observed a marked difference in the dependence on complex I for plant growth and development between Arabidopsis and maize. The deficiency of mitochondrial complex I usually leads to embryo lethality in maize, whereas such mutants can survive and set seeds in Arabidopsis. For example, the loss-of-function of EMP16 and DEK47 affects the assembly of complex I, leading to severely aborted embryo and endosperm development in maize (Xiu et al., 2016). But the mutants of otp43, ndufs4, and slo1 with a complete loss in the assembly and activity of complex I in Arabidopsis can survive and set seeds (de Longevialle et al., 2007; Meyer et al., 2009; Sung et al., 2010). It is not clear why such a difference exists between maize and Arabidopsis. One possibility is that the maize embryogenesis is more sensitive to the OXPHOS than that of Arabidopsis. The loss of respiratory complex I occurred several times in unicellular eukaryotes during evolution. In multicellular plant species, the mitochondrial complex I of European mistletoe (Viscum album) is absent and the complex I activity cannot be detected (Maclean et al., 2018; Senkler et al., 2018), implying that complex I is not essential for mitochondrial function in European mistletoe. In plants, the classical OXPHOS electron transport chain (Complexes I, IV, III, and IV) and non-energy conserving alternative electron transport bypasses (alternative oxidase and NAD(P)H dehydrogenases) are present to regulate mitochondrial respiration (Millar et al., 2011). When the function of complex I is impaired, alternative electron transport bypass is enhanced in plants. Consequently, the corresponding Arabidopsis mutants can survive with a low phosphorylation efficiency via the adjustment in cellular metabolism and development (Meyer et al., 2009), whereas the maize mutants cannot as observed in emp8, emp10, and emp16 (Xiu et al., 2016; Cai et al., 2017; Sun et al., 2018). The molecular basis of this difference invites further investigation.