A peptide chain release factor 2a gene regulates maize kernel development by modulating mitochondrial function

Hui Zhng, Yijin Feng, Kunyng Song, Guofng Li, Jio Jin, Jingjing Go, Yongtin Qin,Hongqiu Wng, Jinpeng Cheng, Zonghu Liu, Jihu Tng,b,*, Zhiyun Fu,*

a Key Laboratory of Wheat and Maize Crops Science/Collaborative Innovation Center of Henan Grain Crops/College of Agronomy, Henan Agricultural University, Zhengzhou 450046, Henan, China

b The Shennong Laboratory, Zhengzhou 450002, Henan, China

c Hebi Academy of Agricultural Sciences, Hebi 458030, Henan, China

d Yunnan Tang Jihua Expert Workstation/Zhaotong Academy of Agricultural Sciences, Zhaotong 657000, Yunnan, China

Keywords:Embryo lethality Mitochondrion Peptide release factor 2 Zea mays

ABSTRACT Mitochondrial protein translation that is essential for aerobic energy production includes four essential steps of the mitochondrial ribosome cycle,namely,initiation,elongation,termination of the polypeptide,and ribosome recycling.Translation termination initiates when a stop codon enters the A site of the mitochondrial ribosome where it is recognized by a dedicated peptide release factor (RF).However, RFs and mechanisms involved in translation in plant mitochondria,especially in monocotyledons,remain largely unknown.Here, we identified a crumpled kernel (crk5 allele) mutant, with significantly decreased kernel size, 100-kernel weight, and an embryo-lethal phenotype.The Crk5 allele was isolated using map-based cloning and found to encode a mitochondrial localization RF2a.As it is an ortholog of Arabidopsis mitochondrial RF2a,we named the gene ZmmtRF2a.ZmmtRF2a is missing the 5th–7th exons in the crk5 resulting in deletion of domains containing motifs GGQ and SPF that are essential for release activity of RF,mitochondrial ribosome binding, and stop codon recognition.Western blot and qRT-PCR analyses indicate that the crk5 mutation results in abnormal mitochondrion structure and function.Intriguingly, we observed a feedback loop in the crk5 with up-regulated transcript levels detected for several mitochondrial ribosome and mitochondrial-related components, in particular mitochondrial complexes CI, CIV,and a ribosome assembly related PPR.Together, our data support a crucial role for ZmmtRF2a in regulation of mitochondrial structure and function in maize.

1.Intr oduction

Mitochondria,the powerhouses of eukaryotic cells,are essential semi-autonomous membrane-bound organelles that derive from an alpha-proteobacterial lineage.Disfunctional mitochondria reduce or negate energy supply, resulting in defective human growth by defective mitochondrial translation [1–6] and plant growth associated with mitochondrial transcription [7–10].However,a few genes associated with mitochondrial translation reportedly affect kernel development in maize.For example, DEK66,encoding a ribosomal assembly factor located in mitochondria,possesses GTPase activity and possibly affects mitochondrial protein translation[11].DEK44 encodes mitochondrial ribosomal large subunit protein L9 (RPL) and is directly involves in 5′maturation and translation initiation of rps3 mRNA, which is associated with mitochondrial translation [12].But the underlining regulatory mechanism of mitochondrial translation is largely unknown.

Mitochondria maintain their own protein-synthesis machinery,which is highly coordinated with cytosol protein synthesis machinery and essential for cellular homeostasis[13].Components of plant mitochondrial translational machinery have diverged considerably from their bacterial predecessors.They contain larger ribosomes with many plant-specific proteins and protein factors,in addition to several changes to the standard genetic code including unconventional stop codons [14].Mitochondrial translation termination factors and release factors(RFs)recognize stop codons in the A-site of the small ribosomal subunit to hydrolyze the ester bond between the P-site tRNA and the nascent polypeptide chain in the active site of the large ribosomal subunit, therefore aiding termination of protein synthesis.In bacteria, stop codons UAG and UAA are recognized by RF1, whereas UGA and UAA are read by RF2[15–17].In human,mitochondrial RF1a(mtRF1a)or mtRF1L(mtRF1-Like)releases ribosomes at UAG or UAA stop codons[2,18],whereas mtRF1 was recently proven to recognize the noncanonical stop codons AGA and AGG of cytochrome C oxidase subunit 1(COX1) through a network of interactions between the codon,mtRF1 and ribosomal RNA (rRNA) [19–20].In Arabidopsis, four genes predicted to encode mtRF proteins, namely mtRF1(At2g47020), and mtRF2a, mtRF2b, and mtRF2c, which are like RF2 (At1g56350, At3g57190, and At1g33330) [14,21].Only mtRF2a is predicted to contain the RF motif GGQ, which is essential for ribosome binding and the release activity of RF [22], and the SPF motif which is necessary for recognizing the UAA/UGA stop codons[23].When ribosomes reach to the extremity of non-stop mRNAs,the mtRF2 is speculated to regulate peptide release and ribosome recycling [14].However, the function of mtRF2 in monocots is unclear.

There are also three predicted mtRF2 protein coding-genes in the maize genome, mtRF2a, mtRF2b, and mtRF2c [21].In this study, we characterized a kernel mutant identified from breeding populations and isolated its causal gene Zm00001d018330 by map-based cloning.Functional annotation and subcellular localization experiments revealed that Zm00001d018330 encodes a mitochondrial peptide release factor with high similarity to Arabidopsis mtRF2a.We used western blot, BN-PAGE, transcriptome, and qRT-PCR analyses to investigate whether ZmmtRF2a modulates mitochondrial function.Our results showed that the loss of ZmmtRF2a function alters the transcript and protein levels for several important components of both the mitochondrial transport chain complex and mitochondrial ribosome.The study sheds light on the function of mtRF2 in the mitochondrial translation system.

2.Materials and methods

2.1.Plant materials

A spontaneous crumpled kernel mutant caused by the crk5 allele was identified and isolated from a maize breeding population.The normal kernels from segregated ears were planted, which were self-pollinated and crossed with Zheng58 at the same time.Individuals showed segregated ears were used to obtain F2population.And an F2population of 20,400 individuals was generated.All materials were planted at the research farm of Henan Agricultural University, Zhengzhou, Henan, China (113°42′E, 34°48′N).Three biological replications were used for ear and kernel cytological sectioning and functional validation experiments.These samples were derived from the heterozygous F3individuals at 8, 10, 12, 13, 14,15, 18, 20, and 25 d after pollination (DAP).

2.2.Preparation of sections for cytology

Immature kernels were fixed in a formalin-acetic acid-alcohol(FAA) solution (50% ethanol, 5% acetic acid, and 3.7% formaldehyde) and stored at 4 °C overnight after which they were embedded in paraffin and sectioned into 10 μm slices using a Leica RM2235 microtome(Germany).The slices were stained with toluidine blue(Sinopharm Chemical Reagent Co.,Ltd.)to enhance contrast and visibility of the structures.Sections were observed using a Leica M165FC stereomicroscope(Germany)[24]and by scanning electron microscopy(SEM,ESEM FEI Q45,USA)and electron transmission microscopy (TEM, JEOL Model JEM-1400, Japan).Samples were assayed using methods previously reported by Wu and Messing [25] and Zhang et al.[26].

2.3.Map-based cloning of the Crk5 allele

A total of 1100 SSR markers distributed evenly across the ten chromosomes were used for bulk segregation analysis of the Zheng58 × crk5 F2population.New linked markers were developed according to sequence differences between wild-type (WT)and crk5 to narrow down the location of the candidate gene(Table S1).Critical recombinants were verified by phenotypic assessment of the corresponding F3plants.All primers used in the study are listed in Table S1.

2.4.Measurement of storage reserves

At least 25 endosperms from each genotype were excised and pulverized into fine powder using a crusher for starch extraction.For each sample, 100 mg of flour was used to analyze total starch content using a K-TSTA Megazyme kit (Ireland, K-TSTA-100A Megazyme)[26].Three biological replicates were assessed for each sample.

2.5.RNA extraction and RNA-seq analysis

Total RNA was extracted from kernel samples of homozygous F3WT and crk5 mutant kernels on the segregating ears at 13 days after pollination (DAP) using an RNAprep Pure Plant Kit (Tiangen Biotech Co., Ltd., Beijing, China).One microgram of total RNA was used for first-strand cDNA synthesis using HiScript QRT SuperMix for qPCR (+gDNA wiper) (R223-01, Vazyme Biotech Co., Ltd., Nanjing, Jiangsu, China).Quantitative real-time RT-PCR (qRT-PCR)was performed on a Bio-Rad CFX96 system (Bio-Rad, CA, USA)using the 2× SYBR Green qPCR Premix (Universal) (KS0601-500,Codonx Life Sciences, Beijing, China).The comparative CT method(2–ΔΔCT)was used to calculate relative mRNA levels with the maize ACTIN2 gene used as the internal control.For discrimination of the transcription levels of mitochondrial genes between the WT and crk5 in qRT-PCR, equal amounts of cDNA template were amplified with 35 cycles and 5 μL of PCR products were assayed on 1%agarose gels.Three biological replications were sampled for each genotype.Primers are listed in Table S1.

Strand-specific libraries for RNA-seq were prepared from ribosomal RNA-depleted mRNA using a ScriptSeq Complete Plant kit(Epicentre) and sequenced on an Illumina Nova6000 platform.Clean reads were mapped to the maize B73 reference genome(RefGen_V4) using Bowtie2 (Bowtie2_V2.4.2) [27].Gene expression levels were converted to Fragments per Kilobase Million(FPKM) for each transcript model and differentially expressed genes (DEGs) were selected by the following criteria: |log2(Fold change)| >1 and FDR < 0.05 (false discovery rate, P adjust), calculated by the DEseq2(DEseq2_1.20.0)software[28].Gene Ontology(GO)and Kyoto Encyclopedia of Genes and Genomes(KEGG)analyses of DEGs were implemented using the TopG(https://www.bioconductor.org/packages/release/bioc/html/topGO.html) and KOBAS packages [29], respectively.

2.6.Subcellular localization

The complete coding sequence(CDS)of ZmmtRF2a without stop codon was cloned and ligated into the pRTL2-GFP vector with the enhanced GFP (eGFP) reporter.The resulting 35Spro::ZmmtRF2a:eGFP construct was transiently expressed in maize protoplast as described by Yoo et al.[30].Transfected protoplasts were observed using a Zeiss LSM710 (Carl Zeiss AG, Germany) confocal microscope.The primers are listed in Table S1.

2.7.BN-PAGE and complex activity assay

Mitochondrial proteins were isolated from WT and crk5 endosperms 13 DAP using a method previously described by our laboratory [31].BN-PAGE and the complex I activity assay were performed as described[32]and in-gel activity staining of complex IV was performed as described by Sabar et al.[33].

2.8.Immunoblot analysis

Mitochondrial proteins were separated using SDS-PAGE, then transferred to a polyvinylidene difluoride membrane (0.22 μm;Millipore) and incubated with a 1:2000 dilution of primary rabbit anti-plant mitochondrial protein polyclonal antibodies (Actin,NAD7, NAD9, SDH1, CYC1, COX2, ATP6), then with a secondary Goat anti-Rabbit IgG antibody(Earthox Lot#110491,1:10000 dilution).The membrane was rinsed and developed using an EasySee Western Blot kit (DW101-02, TransGen, China) and the signal was visualized using a Tanon 5200 imager (Tanon, Shanghai,China).

2.9.Phylogenetic analysis

Amino acid sequences of RFs from different species were retrieved from the National Center for Biotechnology Information Nonredundant Protein Sequences Database using a BLASTP search function against the Zm00001d018330 protein sequence and aligned using the online MUSCLE tool (https://www.ebi.ac.uk/Tools/msa/muscle/ [34]).A rooted phylogenetic tree was constructed using the neighbor-joining method fitted in the MEGA7.0 software package[35].The evolutionary distances were computed using Poisson correction analysis.

2.10.Flow cytometry

Endosperm nuclei from 13 DAP WT and crk5 mutant kernels were collected from the same segregated ears and extracted as previously described [36,37].Three biological replicates of the samples underwent flow cytometry analysis using a FACSCelesta flow cytometer (BD Biosciences, San Diego, CA, USA) equipped with an argon ion laser with a wavelength of 488 nm [38].At least 10,000 nuclei were collected for each sample and analyzed using the FlowJo Software (FlowJo, Ashland, OR, USA).

3.Results

3.1.Kernel development was arrested in the crk5 mutant

Fig.1.Phenotypic comparison between wild-type(WT)and crk5 mutant.(A–D)Kernel characteristics of wild type(WT)and crk5 mutant from the same ear.Scale bars,1 cm.(E) Longitudinal sections of WT and crk5 mutant kernels.En, endosperm; Em, embryo.Scale bar, 2 mm.(F) 100-kernel weight of WT and crk5 mutant (n = 3).Values are means±SD,***,P<0.001(Student’s t-test).(G)Starch contents in WT and crk5 mutant endosperms(n=3).Values are means±SD,*,P<0.05(Student’s t-test).(H)Number of starch granules per mm2 between WT and crk5 mutant(n=3).Values are means±SD,***,P<0.001(Student’s t-test).(I,J)Scanning electron microscope(SEM)images of the central regions from mature endosperms of WT (I) and crk5 mutant (J) kernels.Scale bars, 50 μm.(K) Germination of WT and crk5 mutant seeds.Scale bar, 1 cm.

The crk5 mutant, identified as a spontaneous small kernel mutant in a maize breeding population, showed delayed endosperm filling and development compared with wild-type (WT)plants (Fig.S1).Development defects were first detected 8 DAP and increased severity was observed at maturity (Fig.S1).Mature kernels of the crk5 mutant were significantly smaller than those of the WT (Figs.1A–E, S2).This culminated in a 56% reduction in 100-kernel weight for crk5 mutant kernels (Fig.1F).Consistent with reduced kernel size, longitudinal sectioning revealed that crk5 mutant kernels had smaller embryos, less endosperm and a 9% reduction in starch content, compared to the WT (Fig.1G).In contrast to the round and compact matrix of starch granules in WT endosperm, starch granules in mutant kernels were irregular and 22.95% less in the number of starch (Fig.1H–J).This defective embryo and endosperm development was associated with germination failure (Fig.1K).These results indicated that embryo and endosperm development was arrested in the crk5 mutant.Paraffin sections of developing WT and crk5 mutant kernels were examined to further investigate developmental differences.Endosperm development in the mutant at 10 DAP was delayed compared to the WT and there was a large space between the endosperm and pericarp (Fig.2A).At 15 DAP, whole embryo structures and fully developed endosperms were presented in the WT, whereas crk5 mutant kernels developed visible embryos,they lacked discernably differentiated scutella, roots and leaf primordia, and had morphologically abnormal endosperm (Fig.2B–C).Regular and darkly stained aleurone(AL)and more than two layers of the basal endosperm transfer layer (BETL) were observed in WT endosperm(Fig.2D–I), but the mutant endosperm lacked AL structures and critical cell wall ingrowth in BETL cells.Transmission electron microscopy images also revealed a reduction in the number of both starch granules and protein bodies in the crk5 mutant (Fig.2J–M).These observations were consistent with the loose starch granule distribution, reduced kernel size, and lower 100-kernel weight determined in the mutant (Fig.1).

3.2.Map-based cloning of the crk5 allele

To map the crk5 locus, heterozygous individuals were crossed with inbred line Zheng58 due to lethality of the homozygote.The hybrid individuals were self-pollinated to produce a population of F2plants.An ear on an F2plant had normal and mutant kernels in the ratio of 3:1(159:45,χ2=0.94<χ20.05=3.84)and two-thirds of F3individuals from normal kernels showed segregation of kernel type; that is the F2population segregated 167 Crk5/Crk5: 354 Crk5/crk5 (χ22:1= 0.31; P > 0.05).These results indicates that crk5 mutant phenotype was controlled by a single recessive allele.

The Crk5 locus was preliminarily mapped on chromosome 5 by linkage with SSR markers HA-26 and HA-61 based on 419 F2mutant kernels(Table S1).Genotyping of an additional 20,000 F2individuals with four new linked markers further narrowed the locus down to a 112.4-kb physical interval between markers C5 (1 recombinant)and C300(2 recombinants)(Fig.3A).Of Eight protein coding genes annotated in this interval in the B73 reference genome(RefGen_V4, www.maizegdb.org) (Fig.3A) only four,

Fig.2.Cytological characterization of wild-type(WT)and crk5 mutant kernels.(A–I) Paraffin sections of WT and crk5 mutant kernels at 10 DAP (A)and 15 DAP(B–I).Scale bars,1 mm for(A,B),200 μm for(C),100 μm for(D,E),50 μm for(F,G),and 200 μm for(H,I)respectively.En,endosperm;Em,embryo.(J,M)Ultrastructure of 15 DAP starchy endosperms of the WT(J and L)and crk5 mutant(K and M)observed by transmission electron microscopy.Scale bars,10 μm for(J,K)and 2 μm for(L,M).SG,starch granule;PB, protein body.

Fig.3.Map-base cloning of the Crk5 locus.(A)The Crk5 locus was mapped to a ～112.4 kb region on chromosome 5 of the B73 reference genome.Numbers of recombinants and population size are shown under each marker.(B)Schematic of the Crk5 gene structure.The polymorphic site is marked with green inverted triangles.The blue box and bold black line indicate exons and introns,respectively.The gray boxes indicate the untranslated regions(UTRs).(C)RT-PCR detection of the Crk5 locus using primers shown in (B).DNA ladder from the top to the bottom is 2.0, 1.5, 1.0, 0.75, 0.5, 0.25, and 0.1 kb.(D) Schematic diagram of the ZmmtRF2a with conserved domain indicated.

Zm00001d018326, Zm00001d018328, Zm00001d018330, and Zm00001d018333, expressed in endosperm (Table S2).Zm00001d018326 and Zm00001d018328 were first excluded because their EMS and CRISPR-Cas9 mutants failed to show a distinctive kernel phenotype (Fig.S3).Zm00001d018333 was similar in expression and cDNA sequence between the WT and crk5 mutant except for four SNP in HapMap2 (https://ensembl.gramene.org/Zea_mays/Info/Index), indicating that that it was not the causal gene.To find the critical polymorphism, RNA-seq data were compared between the WT and crk5 mutant.Zm00001d018330 in the crk5 mutant lacked expressed transcript reads after the 4th exon (Figs.3B, S4A).Consistent with this finding, primers designed to detect transcript specifically after the 4th exon generated an RT-PCR band for WT but not for crk5 mutant samples (Fig.3C).Zm00001d018330 encodes a protein of 467 amino acids with a peptide chain release factor 2 domain (178–188 aa).Its counterpart in Arabidopsis was predicted to be mitochondrial peptide chain release factor 2a (Fig.3D).Zm00001d018330 in the crk5 mutant encodes a truncated protein lacking 197 amino acid residues at the C-terminal.Together,these results indicate that mutation of Zm00001d018330 was the most likely cause of the crumpled kernel phenotype.Given its homology,we named this gene ZmmtRF2a.

3.3.ZmmtRF2a is constitutively expressed, and its protein is localized to mitochondria

A public database containing maize gene expression profiles indicates constitutive expression of ZmmtRF2a (Zm00001d018330,AC211357.4_FG004) with increased expression detected during early endosperm development (https://maizegdb.org/) [39](Fig.S5).Protein sequence analysis revealed that ZmmtRF2a is highly homologous to bacterial RF2 (E value 5e-117, Identity 48.31%, https://blast.ncbi.nlm.nih.gov/Blast.cgi), and homology with human mtRF1a is higher than with mtRF1(Fig.4A).Phylogenetic analysis confirmed that ZmmtFR2a belongs to the plant RF2 clade, along with the human mitochondrial mtRF1a protein(Fig.4B).ZmmtRF2a contains a GGQ motif required for RF ribosome binding and release activities[40],and an SPF motif that recognizes stop codons [14,23](Fig.4A).These two motifs are absent in the in the crk5 mutant, thus preventing stop codon recognition and RF ribosome release (Figs.3B, S4B).

Although ZmmtRF2a is a homolog of the predicted Arabidopsis mitochondrial peptide release factor RF2a, TargetP and PSORT bioinformatic analysis revealed that ZmmtRF2a lacks mitochondrial transit peptides (https://services.healthtech.dtu.dk/services/TargetP-2.0/[41],https://psort.nibb.ac.jp/).To define the subcellular localization of ZmmtRF2a, we constructed a 35SPro::ZmmtRF2aeGFP vector and transiently expressed it in maize protoplasts.As shown in Fig.4C,the majority of eGFP signals appeared as elliptical or circular cytosolic signals,which did not overlap with chloroplast autofluorescence.By contrast, eGFP signals overlapped with RFP fluorescence from the co-transformed mitochondrial marker ZmNAD9-RFP.These results indicate that ZmmtRF2a localizes to the mitochondria, despite the lack of a predicted mitochondrial targeting signal peptide.

3.4.Loss of ZmmtRF2a activity affects mitochondrial morphology and function

We analyzed the effects of ZmmtRF2a loss on mitochondrial morphology and function to investigate whether ZmmtRF2a acts as a mitochondrial peptide release factor.Transmission electron microscopy revealed distinct differences between samples from the mutant and WT.Mitochondria in the crk5 mutant appeared much more varied in size with evidence of collapse; many mitochondria also appeared vacuolated and lacked a distinct ridge or double-layer membrane structure (Fig.5A).Consistent with the abnormal mitochondria observations results of blue native polyacrylamide gel electrophoresis (BN-PAGE) with consecutive in-gel activity and immunoblotting (Fig.5B–D), indicated altered assembly of mitochondrial respiratory chain complex subunits.We detected a decrease in complexes I and III and a slight increase in complex V in the crk5 mutant compared to the WT (Fig.5B).The decreased level of complex I protein was associated with significantly decreased NADH dehydrogenase activity of complex I(Fig.5C).Furthermore, western blotting showed a substantial reduction in complex IV activity in the crk5 mutant(Fig.5D),especially for Cox2 (complex IV) (Fig.5E).We also observed a high accumulation of SDH1 (complex II) in the crk5 mutant (Fig.5E),probably representing a compensation effect resulting from a reduction in the other complexes.However, there were increased transcript levels of these mitochondrial respiratory chain complex subunit genes in response to the mitochondrial abnormalities in the mutant (Fig.5F).Consistent with its peptide release function,the inconsistency we observed between transcription and protein levels indicated that the ZmmtRF2a mutation influences mitochondrial mRNA translation activity.In response to the defective cytochrome c pathway, plants can activate an alternative oxidase(AOX) pathway to maintain the tricarboxylic acid cycle and electron transport [42].We detected more than 1300-fold and 12-fold up-regulation of AOX2 and AOX3, respectively, in the mutant(Fig.5G).These results suggest that loss of ZmmtRF2a function in the crk5 mutant activates alternative respiration activity to compensate for defects in the mitochondrial oxidative phosphorylation pathway.Taken together, our data show that ZmmtRF2a is required for mitochondrial function and that intact GGQ and SPF motifs are important for ZmmtRF2a function.

3.5.ZmmtRF2a activity is important for endosperm development

To identify downstream targets influenced by ZmmtRF2a activity during kernel development, we performed transcriptome deep sequencing (RNA-seq) for the crk5 mutant and WT kernels at 13 DAP; 7346 of 29,108 expressed genes were identified as differentially expressed genes (DEGs) (|log2(Fold change)| > 1,FDR < 0.05) (Table S3), including 4492 up-regulated DEGs and 2854 down-regulated DEGs (Fig.6A).To identify biological processes most affected by the loss of ZmmtRF2a function, we subjected the DEGs to gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses.Cell cycle (GO:0007049,P = 3.29E-13), nucleotide metabolic process (GO:0009117,P = 1.09E-06), and phosphorylation (GO:0016310, P = 1.25E-06)were three most enriched GO terms for the up-regulated DEGs(Fig.6B), whereas starch biosynthetic process (GO:0019252,P = 6.11E-07) and cellular amino acid biosynthetic process(GO:0008652, P = 7.37E-06) were the top two most enriched GO terms among down-regulated DEGs (Fig.6C).As expected,mitochondrial-related GO terms were also enriched, these included mitochondrial transmembrane transport (GO:1990542,P = 1.27E-27), aerobic respiration (GO:0009060, P = 4.65E-23), tricarboxylic acid cycle(GO:0006099,P=5.05E-18),respiratory electron transport chain (GO:0022904, P = 1.19E-07), mitochondrial respiratory chain complex assembly (GO:0033108, P = 1.22E-07),cellular amino acid metabolic process (GO:0006520, P = 1.25E-07), mitochondrial electron transport, ubiquinol to cytochrome c(GO:0006122, P = 3.13E-07), mitochondrial translation(GO:0032543, P = 7.52E-07), mitochondrial ATP synthesis coupled electron transport (GO:0042775, P = 2.24E-06), respiratory chain complex IV assembly (GO:0008535, P = 2.94E-06), oxidoreduction coenzyme metabolic process (GO:0006733, P = 4.67E-06), alternative respiration (GO:0010230, P = 2.43E-05), and mitochondrial electron transport cytochrome c to oxygen (GO:0006123,P = 8.15E-05) (Fig.6D).

Fig.5.Mitochondrial function is disrupted in crk5 mutant kernels.(A) Ultrastructure of developing endosperms of the WT and crk5 mutant (15 DAP).Mt, mitochondrion.Scale bar,1 μm.(B)Blue native(BN)-PAGE analysis of mitochondrial complexes in crk5 mutant and WT.The BN gel was stained with Coomassie Brilliant Blue(CBB).(C)NADH dehydrogenase activity of mitochondrial complex I in crk5 mutant and WT.DLDH was used as a loading control.(D)In-gel activity assay of mitochondrial complex IV in crk5 mutant and WT.(E) Western blotting of mitochondrial proteins using antibodies against NAD7, NAD9, SDH1, CYC1, COX2, and ATP6.Anti-actin was used as the loading control.(F)qRT-PCR of mitochondrial genome-encoded genes associated with mitochondrial ETC(Electron Transfer Chain)function.(G)qRT-PCR of genes associated with the alternative respiratory pathway,including AOX1,AOX2,and AOX3.ACTIN was used as an internal control(n=3).Values are means±SD(*,P<0.05;**,P<0.01;***,P<0.001 by Student’s t-tests).

Unexpectedly, most up-regulated DEGs included two critical Cyclin A and D for G1/S transition phase and one Cyclin B for G2/M transition phase, indicating feedback regulation on genes that are critical for cell cycle progression(Table S3)[43].Flow cytometry analysis revealed a lower nuclei ploidy proportion for the crk5 mutant compared with the WT,indicating arrest of endoreduplication in the mutant (Fig.6E, F).Transcripts encoding most mitochondrial complex and mitochondrial ribosomal subunits were dramatically increased in crk5 mutant compared with the WT.Taking the mitochondrion-coding genes for example, four of the six mitochondrial DEGs between crk5 mutant and WT were up-regulated,including two complex IV subunit coding genes COX1 and COX2 and mitochondrial ribosomal small subunit (rps) coding genes rps2 and rps7 (Tables S3, S4).Although ribosomal large subunit genes rpl16 and rps12 were both down-regulated in the crk5 mutant, other mitochondrial ribosomal function-related nuclearcoding genes, such as PPR, mitochondrial electron transport chain complex genes, and alternate oxidative respiration-related genes(AOX2 and AOX3) were up-regulated (Tables S3, S4).These results were further validated by qRT-PCR (Fig.5F, G).By contrast, the transcript levels of complex II components, such as SDH2(Zm00001d005663), and genes that regulate cytochrome c oxidase function or involved in amino acid metabolism related to the TCA cycle were significantly down regulated (Table S3).These results indicate that ZmmtRF2a is required for mitochondrial function and ATP energy production.This presumption is consistent with the deformed mitochondrial structure in the crk5 mutant(Fig.5A).

Fig.6.Transcriptomic analysis and flow cytometric assay WT and crk5 mutant endosperm at 13 DAP.(A)Volcano plot for differentially expressed genes(DEGs)between crk5 mutant and WT.(B, C) GO analysis of DEGs that are up- (B) and down-regulated (C) in crk5 mutant compared with WT.(D) GO analysis of mitochondrial-related DEGs between WT and crk5 mutant.(E, F) Flow cytometric assay and quantification of the endoreduplication in endosperm of WT and crk5 mutant at 13 DAP (n = 3).Values are means ± SD, (*, P < 0.05; **, P < 0.01; ***, P < 0.001 by Student’s t-tests).PI-A, area of PI (propidium iodide, fluorescent fuel for the nucleus) fluorescence signal.

Consistent with the powerhouse function of mitochondria, loss of ZmmtRF2a function undoubtedly influences nutrient transport from maternal to filial tissues.As feedback regulation, we found that BETL-specific genes (e.g., MN1, BETLs, MRP1; Fig.S6A;Table S3) were up-regulated in crk5 mutant to perhaps overcome this bottleneck.A similar phenomenon was observed for many mitochondrially targeted nuclear genes, such as Emp4, Dek2,DEK35, MEG1, and ZmTCRR1 (Fig.S6A; Table S3).Several starch and protein biosynthesis genes were also down-regulated in crk5 mutant, such as cyPPDK1, cyPPDK2, BT2, Sh2, O11, O2, NKD2, PBF1,NAC130, NAC128, NKD3, PBF1, TAR3, FL2, and NKD1 (Fig.S6C;Table S3) and 22 of 28 zein genes (Fig.S6B; Table S3).These transcriptional differences likely contributed to the reduced starch content, as well as abnormal starch granules and protein bodies detected in the crk5 mutant (Fig.2).

4.Discussion

4.1.ZmmtRF2a is a mitochondrially targeted protein generated in the nucleus

RF2 is a group of organellar release factors that are localized to the mitochondria (mtRF2, i.e., mtRF2a, mtRF2b, and mtRF2c) or plastids (pRF2a) [21].RF2a has three functionally described structural features: a codon-recognition (CR) domain (with an alpha-5 helix), SPF, and a peptidyl-hydrolase (PTH) domain that contains a universally conserved GGQ motif that provides RF2a with peptide chain release factor activity.The mitochondrial mtRF2a is consistently found in streptophytes (land plants), red algae, dictyosteliida, and some stramenopiles (namely in brown algae, oomycetes and Blastocystis)but is absent in most animals,fungi and excavata,apart from the heterolobosean Naegleria gruberi [21].However,there are no experimental data on its molecular function in plants and subcellular localization in eukaryotes.Here, we verified that ZmmtRF2a is a nuclear-encoded protein that is localized to the mitochondria through a subcellular localization process that does not depend upon a mitochondrial signal peptide (Fig.4C).

4.2.ZmmtRF2a is essential for mitochondrial function

Previous research indicated that regulation of plant mitochondrial gene expression is primarily governed by posttranscriptional mechanisms.However, the available experimental evidence is largely confined to phenomena impacting RNA quality and quantity[44]and regulation of plant mitochondrial translation is poorly understood.In Arabidopsis, the mitochondrial mRNAs transcribed by the nad6 and ccmC genes were shown to be processed upstream of in-frame stop codons by northern hybridization and qRT-PCR.These non-stop mRNAs were verified to be translated and might be related to the presence of multiple release factors of mtRF2 [14].In addition, a nucleus-encoded plastidlocalized RF2 was found to influence the stability of UGAcontaining transcripts in Arabidopsis chloroplasts [45].Human mtRF1 recognizes the two noncanonical stop codons AGA and AGG in COX1,but these noncanonical stop codons are not in maize COX1 (Fig.S7).To explore whether mtRF2a might also modulate the translation of special mitochondrial mRNAs as in Arabidopsis,the CDSs of nad6 and ccmC mRNA from Arabidopsis and maize were subjected to BLAST analysis(Fig.S8).We identified large variations from the 581st nucleotide in the nad6 CDS (Fig.S7A) and from the 711th nucleotide in the ccmC CDS(Fig.S8B).Furthermore,biased stop codon preference was not found in the mitochondrial DEGs between WT and crk5 (Table S4).These results indicate that mtRF2a might recognize conventional stop codons without preference (Figs.4B, S3), therefore differing from reported Arabidopsis mitochondrial and chloroplastial RF2 activities.This inference was supported by western blots, where changed mitochondrial proteins have different conventional stop codons, such as NAD7(UGA), NAD9 (UAA), and COX2 (UAA) (Fig.5E), as well as the six mitochondrial DEGs for COX1, COX2, rps12, rps2, rps7, and rpl16(Fig.5F; Table S3).

4.3.ZmmtRF2a affects kernel development by modulating mitochondrial function

Mitochondrial function is essential for tissue development and cell cycle progression.Mitochondrial defects can result in large changes in plant development, such as those reported for mitochondrial ribosomes [46].For example, a weak allele of the ribosome biogenic factor Reas 1 gene (Dek*) partially inhibits the maturation and output of the ribosome subunit in 60S, inducing pleiotropic cellular responses [47]; loss of function of the putative 50S ribosomal protein L9(DEK44)affects the expression of respiratory chain-related protein-coding genes and cell cycle related genes, producing defective kernels [12].Here, we report that null mutations of ZmmtRF2a also cause up-regulation of respiratory chain-related protein-coding and cell cycle-related genes in response to mitochondrial dysfunction.BETL cells are important for nutrient transport from maternal vascular tissues to the endosperm.They have a high metabolic rate that is supported by numerous, small, spherical mitochondria.Several studies reported mitochondrial dysfunction in mutants with defective kernels and abnormal BETL, such as dek44, smk4, emp602, dek2, and emp4[12,48–51].These defects can also be accompanied by reduced storage reserves and metabolic processes, as observed for the emp602 and dek2 mutants [49,50].Similarly, the crk5 mutation involving ZmmtRF2a also leads in decreased expression of genes involved in the biosynthesis of storage reserves (Fig.S6C;Table S3).Hence, ZmmtRF2a is required for kernel development by modulating correct translation of mRNAs associated with mitochondrial function and for expression of mRNAs involved in other metabolic processes that are reliant on the energy produced by mitochondria.

CRediT authorship contribution statement

Hui Zhang:Conceptualization, Investigation, Data curation,Methodology, Writing – original draft, Writing – review & editing.Yijian Feng:Conceptualization, Investigation, Data curation.Kunyang Song:Investigation,Data curation,Methodology.Guofang Li:Conceptualization, Investigation.Jiao Jin:Conceptualization,Investigation.Jingjing Gao:Investigation.Yongtian Qin:Data curation, Methodology.Hongqiu Wang:Investigation, Data curation.Jinpeng Cheng:Investigation, Data curation.Zonghua Liu:Supervision, Funding acquisition.Jihua Tang:Conceptualization,Visualization, Supervision, Funding acquisition, Writing – review& editing.Zhiyuan Fu:Conceptualization, Data curation, Writing– original draft, Writing – review & editing, Supervision, Funding acquisition, Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the National Natural Science Foundation of China(31971893,U2004144),the Key Technologies R&D Program of Henan Province (232102111080, 212102110043), and Academician Expert Workstation (202305AF150082).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2023.10.006.