Mitochondrial localization of ORF346 causes pollen abortion in alloplasmic male sterility

Shifei Sng,Hongto Cheng,Mengyu Ho,Bingli Ding,Desheng Mei,Hui Wng,Wenxing Wng,Ji Liu,Li Fu,Kede Liu,Qiong Hu,*

a Oil Crops Research Institute of Chinese Academy of Agricultural Sciences/Key Laboratory for Biological Sciences and Genetic Improvement of Oil Crops,Ministry of Agriculture and Rural Affairs,Wuhan 430062,Hubei,China

b National Key Laboratory of Crop Genetic Improvement,Huazhong Agricultural University,Wuhan 430070,Hubei,China

Keywords:

A B S T R A C T The Nsa cytoplasmic male sterility(CMS)system confers stable male sterility and offers great potential for production of hybrid seeds in oilseed rape.However,genes responsible for male sterility in Nsa CMS have not been identified.By mitochondrial genome sequencing of Nsa CMS and its maintainer line,we identified in an Nsa CMS line several chimeric genes encoding hypothetical proteins harboring transmembrane domains.One novel chimeric gene orf346 showed high identity with cox1 at the 5′terminal region and was co-transcribed with nad3 and rps12 genes.Transgenic plants of orf346 fused with or without mitochondrial targeting peptide conferred complete male sterility in Arabidopsis.ORF346 was mitochondrion-localized.Expression of orf346 in Escherichia coli inhibited bacterial growth,with excessive accumulation of reactive oxygen species and decreased ATP content.These results reveal a link between the newly identified mitochondrial gene orf346 and the abortion of Nsa CMS.Inadequate energy supply and excessive accumulation of reactive oxygen species may account for pollen abortion in Nsa CMS plants.

1.Introduction

Cytoplasmic male sterility(CMS)is maternally inherited and results in inability to produce functional pollen[1].Besides its utility for generating hybrid seed in crops,CMS has also emerged as a system for studying anther development and cytoplasm–nucleus interactions[2,3].Previous studies[4–6]generally confirmed that cytoplasmic male sterility is associated with abnormal DNA recombination in the mitochondrial genome.CMS may be derived spontaneously from natural mutation,as are Pol CMS and Shan 2A CMS,two systems widely used in rapeseed(Brassica napus)hybrid production in China.Some types of CMS have been produced by protoplast fusion.For example,Nsa CMS and SaNa-1A CMS were generated from somatic hybrids ofB.napuswithSinapis arvensisandS.alba,respectively[4–6].Owing to the stable expression of male sterility under diverse environmental conditions,Nsa CMS system has the potential to increase hybrid seed purity during hybrid seed production.Restorer lines of Nsa CMS conferred resistance to stem rot(Sclerotinia sclerotiorum)and pod shattering[7].Identifying male-sterility genes will shed light on the mechanism of Nsa CMS and ensure wider application of this system.

InBrassica,one of the genera suitable for investigating the mechanism of CMS,several types of CMS have been identified.CMS genes includingorf224,orf138,orf222,orf263,andorf288were identified as being responsible for Pol CMS,Ogu CMS,Nap CMS,tour CMS,and Hau CMS,respectively,inBrassicaspecies[8–12].These functionally characterized genes provide clues for cloning novel CMS-associated genes in other CMS systems.Most CMSassociated genes are co-transcribed with mitochondrial genes encoding ATPase,cytochrome oxidase,or ribosomal proteins[13,14].In the energy-deficiency model[2],a CMS protein causes mitochondrial deficiencies,such that energy requirements during male reproductive development are not met.Orf224was cotranscribed with theatp6gene,generating 2.2-kb and 1.9-kb transcripts in the Pol CMS system[9].In Ogura CMS,geneorf138was co-transcribed withatp8and associated with the male-sterile phenotype[15,16].Orf222,showing 79% sequence similarity to Polorf224and co-transcribed with the third exon of thenad5gene(nad5c).orf139was proposed[10]to be associated with Nap CMS.Orf263was associated with tour CMS and co-transcribed withatp6[13].Orf288was responsible for the male sterility of Hau CMS inB.junceaand shown to be co-transcribed withatp6[11].A genomic insertion ofatpAupstream oforf220was detected in aB.junceaCMS line and a transgenic study showed thatorf220fused with mitochondrial localization signal caused male sterility[17,18].Orf108associated with CMS inB.junceacarryingMoricandia arvensiscytoplasm was co-transcribed with theatpAgene[14].Many CMS-associated proteins showed cytotoxic effects on bacterial(Escherichia coli)growth,including ORF288 of Hau CMS[11]and ORF138 of Ogura CMS[19].

Nsa CMS was identified as a novel alloplasmic male-sterility system derived from somatic hybridization ofB.napuscv.Zhongshuang 4 andS.arvensisvar.Yeyou 18[20].The complete organelle genomes ofS.arvensis,Nsa CMS,and its isonuclear maintainer line were sequenced[21,22].Comparison of the mitochondrial genomes revealed the source of Nsa CMS cytoplasm and identified the most likely candidate genes responsible for Nsa male sterility.

To ascertain the responsible gene and the underlying mechanisms for Nsa CMS,we characterized the Nsa CMS-associated candidate genes.One chimeric mitochondrial geneorf346was identified as a best candidate gene responsible for the male sterility of Nsa CMS.Further expression pattern analyses,transgenic plants and biochemical analyses all conformed thatorf346is the gene responsible for male sterility.

2.Materials and methods

2.1.Plant materials

Zhongshuang 4(ZS4)and Zhongshuang 6(ZS6)areB.napuscultivars developed at the Oil Crops Research Institute,Chinese Academy of Agricultural Sciences(OCRI-CAAS)in Wuhan.Nsa CMS and its restorer lines NR1 were produced by somatic hybridization between Zhongshuang 4 andS.arvensiscv.Yeyou 18[4].All plants were grown at OCRI-CAAS.

2.2.Semi-quantitative RT-PCR and RT-qPCR

Total RNA was isolated with an RNAprep pure kit(Tiangen,Beijing,China)according to the manufacturer’s protocol.First-strand cDNA synthesis was performed with Hiscript II QRT Super-mix kit(Vazyme,Nanjing,China)for reverse transcription reaction(RT-PCR)according to the manufacturer’s instructions.The circularized RNA(CR)-RT-PCR method[23]was used to determine the 5′and 3′mRNA termini oforf309andorf346.T4 RNA ligase(Thermo Scientific,Waltham,USA)was used to link 5′and 3′mRNA termini.After circularized RNA was generated,the first strand of cDNA was obtained by reverse-transcribing the RNA with a reverse kit(Vazyme,Nanjing,China)according to the manufacturer’s instructions.Primers orf309CRF/R and orf346CRF/R were used to amplify the cDNA,and the products were ligated to a pTOPO vector(Solarbio,Beijing,China)for sequencing.For semi-quantitative RTPCR,the reaction was conducted using the following program:3 min at 95 °C,33 cycles of 30 s at 95 °C,30 s at 57 °C,and 45 s at 72°C.The expression level of theBnactingene(BnaC02g00690D)was used to standardize the RNA sample for each RT-PCR[24].RTqPCR was performed with a CFX96 Real Time System(Bio-Rad,Hercules,CA,USA)using SYBR Green mix(Transgen Biotech,Beijing).The reactions were performed using the following program:95 °C for 30 s,40 cycles of 5 s at 95 °C,and 30 s at 60 °C.Primers used for semi-quantitative RT-PCR and RT-qPCR are listed in Table S1.

2.3.Northern blot analysis

Total RNA of Nsa CMS and the maintainer line was used for RNA gel blotting as described previously[11]with some modifications.A 20-μg aliquot of RNA of each sample was separated on 1.5%denatured agarose gel with 4-morpholinepropanesulfonic acid(MOPS)(Solarbio,Beijing,China)buffer containing 2% formaldehyde.The RNA was then transferred onto Hybond N+membranes(Amersham,Loughborough,Leicestershire,UK)by capillary blotting and then crosslinked to the membranes by UV irradiation.Pre-hybridization was performed at 42 °C for 4 h and then probe was added.Probe was labeled with digoxin(DIG)according to the manufacturer’s protocol for the DIG High Prime Lab/Detection Kit I(Roche,Switzerland).After several washes,membranes were exposed to colorimetric solution with nitro-blue-tetrazolium/5-br omo-4-chloro-3-indolyl phosphate(NBT/BCIP)(Roche,Basel,Switzerland)and photographed with a gel imaging system,GelDoc XR(Bio-rad,Hercules,CA,USA).

2.4.Expression of orf346 in E.coli and Western blotting

Coding regions of full-length and transmembrane/non-trans membrane structural domains oforf346were amplified from cDNA of the Nsa CMS line by PCR using gene-specific primers.PCR products were cloned into the pET28a vector(kindly provided by Prof.Jinxiong Shen,Huazhong Agricultural University),which harbors a His tag.The expression plasmids were transformed intoEscherichia colistrain Rossatta.Protein expression was induced at 18 °C after addition of 1 mmol L-1isopropyl-b-D-thiogalactopyranoside(IPTG)when the optical density(OD)ofE.coliculture was about 0.6.After induction for 1 h,the OD of the sample was measured at 600 nm with a spectrophotometer every 1 h.

Western blotting was performed as described by Ding et al.[25].Briefly,equal amounts of crude protein extracted fromE.colicells were separated on a 12%(w/v)SDS-PAGE gel and transferred to polyvinylidene fluoride(PVDF)membrane(Millipore,MA,USA).Then the membrane was incubated in blocking buffer(1%BSA,0.05% Tween-20,500 mmol L-1NaCl,and 20 mmol L-1Tris-HCl at pH 7.5)for 1 h.After addition of primary antibody(at 1:1000 dilution),the membrane was incubated for 3 h at room temperature.After washing with phosphate-buffered saline with Tween-20(PBST)for three times(10 min each time),blots were incubated with secondary antibody(goat anti-rabbit,1:3000 dilution)for 1.5 h at room temperature and washed three times with PBST.An antibody against the His tag was purchased from a commercial company(TransGen Biotech,Beijing,China).

2.5.Assessment of intracellular reactive oxygen species levels

The oxidant-sensitive probe 2′,7′-dichlorodihydrofluorescein diacetate(H2DCFDA[Beyotime,Shanghai,China])was used to measure intracellular oxidation levels inE.coli.After culturing for 10 h at 18°C,the OD600 of recombinant cells ofE.colibacterial solution were adjusted to 1.5,and 2 mL of bacterial solution was collected and re-suspended in 1 mL Luria-Bertani(LB)buffer.Then,1 μg mL-1H2DCFDA was added,followed by incubation for 20 min with shaking at 37°C.After brief washing of the cells with LB buffer three times,images ofE.coliwere acquired with a confocal microscope.H2DCFDA-stained bacterial cells emitted bright green fluorescence under the confocal laser scanning microscope and were observed in at least three visual fields.

2.6.Assessment of ATP content in E.coli cells

ATP content was measured using the luciferin–luciferase method according to ATP detection kit instructions(Beyotime).After culture for 5 h at 18 °C, the concentration ofE.colicells was adjusted to 1.5 of OD600. Then 2 mL of cells were used to detect ATP content according to the kit instructions. Luminescence inten-sity was measured with a GloMax Explore luminometer (Promega, WI,USA). A standard curve was prepared based on ATP concentra-tions from 0.1 nmol L-1to 10 μmol L-1.

2.7.Vector construction and genetic transformation of Arabidopsis and B.napus

TheAP3promoter was inserted into a pCAMBIA2300 vector.TheRfo1-165encoding mitochondrial targeting peptide(Genbank accession:LOC108827691)was fused with the open reading frame oforf346,and the fragment was inserted into the modified pCAMBIA2300 vector.The construct was then introduced intoAgrobacterium tumefaciens(GV3101).Genetic transformation ofArabidopsisandB.napuscv.Zhongshuang 6 mediated byA.tumefacienswas performed as described[26].

2.8.Transient expression and subcellular localization of ORF346

Subcellular localization of ORF346 was predicted by online software PredSL(http://aias.biol.uoa.gr/PredSL/)and SignalP 3.0(http://www.cbs.dtu.dk/services/SignalP-3.0/).The coding region oforf346without stop codon was amplified from cDNA of Nsa CMS line by PCR using gene-specific primers.The sequences were then fused to the coding region of green fluorescent protein(GFP)driven by the35Spromoter in the pM999 vector(kindly provided by Dr.Meng Yuan,Huazhong Agricultural University).Plasmids were transformed intoArabidopsisprotoplasts by polyethylene glycol treatment[27].After 12 h to 16 h following transformation,fluorescence signal was observed with a confocal microscope.Mitotracker Red CMX-Ros staining solution(Molecular Probes,Invitrogen)was added to mark the mitochondria of protoplasts before confocal observation.

3.Results

3.1.Screening candidate sterility genes by sequence and expression comparison between Nsa CMS and ZS4

Sequenced comparison of mitochondrial genomes between Nsa CMS line and itsiso-nuclear maintainer line ZS4 showed that the Nsa CMS mitochondrial genome was more collinear with that ofS.arvensisthan with that of ZS4(Fig.1A,B).Sixteen open reading frames(ORFs)were identified as specific to Nsa CMS or harboring sequence variation between Nsa CMS and its maintainer line.Prediction with TMHMM software(http://www.cbs.dtu.dk/services/TMHMM-2.0/)indicated that 11 ORFs,orf106,108,110,115,130,170,192,224,257,309and346,were transmembrane proteins.Among these genes,onlyorf224(Nsaorf224),309and346were found to be chimeric.The chimeric structure ofNsaorf224contained partial sequences of theatp8gene.Bothorf309andorf346showed partial sequence identity with thecox1gene(Fig.1C).

Expression differences of the 11ORFs encoding proteins with transmembrane domains in the flower buds of Nsa CMS,the two parental lines(ZS4 andS.arvensis),and the restorer line were examined.No expression was detected fororf108,orf110,ororf257in any of the lines.The expression levels oforf106,orf170,andNsaorf224did not differ among the four lines(Fig.1D).Fiveorfs:orf130,orf115,orf192,orf309,andorf346were not expressed in the maintainer line,but were expressed in the other three lines.Among these fiveorfs,the expression level oforf346was significantly reduced in the restorer andS.arvensiscompared to that in the Nsa CMS line,whereas the other fourorfsshowed no significant change either in the restorer orS.arvensis.

3.2.orf346 was co-transcribed with nad3 and rps12 genes

We performed Northern blot assay of the Nsa CMS candidate genesNsaorf224,orf309,andorf346having chimeric structure and encoding proteins with transmembrane domains,to assess their co-transcription status.Full-length coding regions of the three genes were used as probes to detect the expression fragment pattern in Nsa CMS and maintainer(ZS4)lines.Northern blotting revealed no fragment pattern difference for theNsaorf224probe between Nsa CMS and ZS4,with two fragments in both lines at two different developmental stages(Fig.2A,left).One specific fragment was found in the male-sterile line whenorf309was used as a probe,whereas no hybridization to ZS4 RNA samples occurred(Fig.2A,middle),indicating thatorf309was not co-transcribed with any functional mitochondrial genes.Two fragments of different sizes were observed in the Nsa CMS line for theorf346probe,whereas only one faint fragment was detected in the maintainer line ZS4(Fig.2A,Right).As there was a homologous ORF(orf322)having 91% sequence similarity withorf346in the maintainer line[22],we speculated that this faint band was due to no specific binding of the probe toorf322in the maintainer line.CR-RT-PCR assays were performed to assess the co-transcription of genes with Nsa CMS candidate genes.Sequencing of the CR-RT-PCR product amplified withorf309specific primers showed that no gene was co-transcribed withorf309,in agreement with the Northern blot result.Orf346was found to be co-transcribed withNADH dehydrogenase subunit 3(nad3)andsmall ribosomal subunit protein 12(rps12),and the starting and terminating sites of co-transcripts were determined from the sequence information of the CR-RTPCR product amplified withorf346-specific primers(Fig.2B,C).Primers F3/R3 covering all coding regions oforf346,nad3,andrps12were also designed to verify this co-transcription result.A clear fragment with expected size was amplified from the male sterile line but not from the maintainer line,confirming the cotranscription oforf346withnad3andrps12(Fig.2C).Nsaorf224was also found to be co-transcribed withcytochrome c biogenesis B(ccmB),also in agreement with the Northern blot result.

3.3.Expression differences of genes involved in co-transcription and respiration

Asorf346was co-transcribed withnad3andrps12,we examined the expression ofnad3,rps12andorf346/nad3/rps12complex in the flower buds of Nsa CMS,the two parental lines(ZS4 andS.arvensis)and a restorer.The transcripts ofnad3andrps12were significantly enriched in the Nsa CMS line compared with ZS4,with only a very weak fragment in ZS4 but a bright fragment in Nsa CMS for each of the genes(Fig.S1).There was no significant expression difference ofnad3orrps12among Nsa CMS,S.arvensis,and the restorer.As expected,the transcript of the co-transcription complexorf346/-nad3/rps12was not detected in ZS4,but was detected in all of the other three lines.We speculated that the enrichment of transcripts ofnad3andrps12detected by RT-PCR was caused by the transcription of theorf346/nad3/rps12complex in both Nsa CMS and the fertileS.arvensisand restorer lines.However,theorf346/-nad3/rps12transcript was markedly reduced inS.arvensisand restorer lines in comparison with the Nsa CMS line,in agreement with the expression result fororf346.

Expression analysis of 34 functional mitochondrial genes revealed that 11 genes were down-regulated by sterility genes,with significantly reduced expression level in Nsa CMS line compared to the maintainer line,especially at the early stage of bud development(Fig.2D).

Fig.1.Comparison ofmale-sterility candidate genes.(A)Comparison of syntenic region of Nsa CMS mitotype with that of its maintainer line.(B)Comparison of syntenic region of Nsa CMS mitotype with that of Yeyou 18.(C)Chimeric structure of three candidate genes.(D)Expression-level identification of genes with sequence differences between Nsa CMS and its maintainer line.

3.4.Functional analysis of orf346 by gene transformation

To verify the association of candidate geneorf346with male abortion of Nsa CMS, the gene was fused with the mitochondrial targeting pre-sequence (MTS) of Ogu CMS restorer geneRfo(from 1 to 165) under the control of theArabidopsisAPETALA3(AP3) promoter (Fig. 3A). ＞90% (58/64) of transgenic plants overexpressingorf346fused with MTS failed to produce pollen and developed abnormal flowers (Fig. 3C). The pistil of the wild type was surrounded by sepals and stamens, whereas flowers from malesterile transformants lacked stamens and petals (Fig. 3C). Two of these transgenic lines were propagated, yielding 43 T2generation plants, 20 male-sterile and 23 male-fertile. These frequencies were coincident with 1:1 segregation (P=0.647＞0.05, χ2=0.938 ＜χ20.05=3.842). A vector containingorf346without MTS under the con-trol ofAP3promoter was constructed (Fig. 3B). After transformation, 95% (20/21) of the transgenic plants were pollenaborted, yielding 11 male-sterile and 12 male-fertile plants in the T2 generation. These results showed thatorf346was the gene that caused male sterility in Nsa CMS, and that even if it did not contain a MTS, it could induce pollen abortion inArabidopsis.

3.5.Mitochondrial localization of ORF346

Transmembrane prediction with TMHMM showed that ORF346 was a transmembrane protein.However,no mitochondrialtargeted signal was detected after subcellular localization prediction by PredSL and SignalP 3.0.To ascertain whether ORF346 was able to bind to mitochondria,full-lengthorf346fused with gene sequence encoding green fluorescent protein(GFP)was transiently expressed inArabidopsisprotoplasts using the PEG-mediated method.The mitochondria were stained with mitotracker Red CMX-Ros dye before confocal observation.GFP fluorescence matched well with mitotracker dye signal(Fig.4A),suggesting that ORF346 was a mitochondrially located protein.

3.6.Growth of E.coli was repressed by expression of orf346

To determine whether ORF346 was cytotoxic to bacterial growth,we cloned its coding sequence into the expression region of the pET-28a vector,and induced its expression with IPTG inE.coli.E.colicells transformed with pET-28a empty vector grew well with or without IPTG induction(Fig.4B).Growth rate ofE.coliwas strongly repressed when pET-28a-orf346 was induced at 18 °C(Fig.4B).However,expression of ORF346 protein was not observed when the gel was stained with Coomassie blue.We performed Western blot assay to detect the protein with 6×Histag antibody.One fragment of approximately 40 kDa was clearly visible in the total protein sample extracted from the pET-28aorf346 transformedE.coliafter induction by IPTG(Fig.4C).Samples extracted from empty plasmid(pET-28a)or pET-28a-orf346 transformedE.coliwithout IPTG induction did not display signal.

Almost all sterility genes contain a segment of hydrophobic structure with similarity to those of toxin proteins.Using the online software TMHMM server 2.0(http://www.cbs.dtu.dk/services/TMHMM/),three transmembrane regions were identified in the N terminus(1–107 AA)of ORF346(Fig.4D).To further identify the cytotoxic region oforf346gene,two truncated versions oforf346were cloned and expressed using the pET28a vector.One contained the sequence encoding the three transmembrane regions of 1–107 AA,and the other contained the rest of theorf346gene.The growth ofE.coliwas normal when the expressed protein lacked the transmembrane domains(1–107AA).The growth ofE.coliwas significantly inhibited when the expressed protein included only the transmembrane domains(Fig.4E,F).This result indicated that the toxic region of ORF346 was located in the transmembrane structure region.

Fig.2.Structural and co-transcription analysis of the candidate gene orf346 for Nsa CMS.(A)Transcript profiles of orf224,orf309,and orf346 in Nsa CMS and ZS4 were assayed with RNA gel blot analysis.(B)Identification of the 5′-and 3′-termini of the orf346 transcript from Nsa CMS by circularized RNA(CR)-reverse transcriptase(RT)-polymerase chain reaction(PCR).The orf346 and the co-transcribed genes are indicated with blue and green.Numbers are those of different clones from PCR products.(C)The primers for CR-RT-PCR and expression analysis,and initiation and termination sites of orf346 transcripts.Expression of orf346 was detected in Nsa CMS and ZS4 by primer F3/R3.(D)Transcription of many genes/ORFs involved in energy production was suppressed in Nsa CMS compared with that in ZS4.-1 and-2 refer to buds with diameter＜2.5 mm and＞2.5 mm,respectively.

3.7.ORF346 causes reactive oxygen species(ROS)accumulation and ATP reduction

ROS and ATP content in ORF346-transformed cells inE.coliwas monitored.The fluorescence intensity of H2DCFDA in pET28a-346-transformed cells was markedly increased after IPTG induction for 10 h(Fig.5A).However,the fluorescence of the pET-28a transformants was barely visible(Fig.5A).ATP quantification assay showed that there was a large(59.8%)decrease of ATP content in cell cultures transformed with pET28a-346 and induced by IPTG,compared with the pET28a-transformed cells or pET28a-346-transformed cells without IPTG induction(Fig.5B).Thus,expression oforf346could cause excessive ROS accumulation and inhibit ATP production.

4.Discussion

4.1.orf346 is the functional gene controlling pollen abortion in Nsa CMS

Fig.3.Transgenic plants with orf346 in Arabidopsis.(A,B)Schematic illustration of constructs used for transformation of Arabidopsis.orf346 was fused with and without Rfo encoding signaling peptide and driven by Arabidopsis AP3 promoter.(C)Expression of orf346 in transgenic Arabidopsis plants.Male-sterile line(S),full fertility(F)and wild type(WT).(D)Phenotype analysis of orf346 fused with MTS transformants in Arabidopsis.Scale bar,5 mm.

Fig.4.Subcellular localization and function of ORF346 in Escherichia coli employing pET28a of ORF346.(A)Subcellular localization of ORF346.Scale bars,10 μm.(B)Influence of orf346 expression on growth of E.coli cells induced by IPTG.The expression vector pET-28a was used as a control.(C)Detection of ORF346 in cell extracts of E.coli.(D)Truncated ORF346 with different transmembrane domains:(a)full-length protein sequence with three transmembrane domains;(b)truncated ORF346 with transmembrane domains;(c)truncated ORF346 without transmembrane domain;(E)Expression effect on E.coli growth of truncated ORF346 with transmembrane domains;(F)Expression effect on E.coli growth of truncated ORF346 without transmembrane domain.

Fig.5.ROS and ATP detection in E.coli cells.(A)Detection of ROS production in E.coli grown in IPTG-containing medium after confocal observation.After culture for 6 h with or without IPTG,transformed cells were stained with H2DCFDA and observed under a confocal microscope.Scale bars,100 μm.(B)ATP content in transformed E.coli cells.

We infer thatorf346is responsible for the male sterility of Nsa CMS inB.napus.Expression oforf346in mitochondria ofA.thalianaresulted in male-sterile plants.Using the anther-specific promoterAP3to driveorf346has caused complete pollen abortion inArabidopsis,together with the disruption of the development of petals and stamens.The male-sterile phenotype was found to be an inherited trait,and co-segregated with theorf346gene in T2progeny.Anther-specific promoters such asAP3,TA29,andLat52,were usually used previously to verify the function of sterility genes[11,18,28–33].TheAP3promoter was used to drive some CMS candidate genes includingorf288of Hau CMS,orf220of CMS-orf220 fromB.juncea[11,18],orf129of I-12 CMS from wild beet[28]andorf147of A4 CMS from pigeon pea[29]for transformation intoArabidopsisor tobacco,and their role in controlling male sterility was confirmed.TheTA29promoter drivingorf507of Peterson CMS in pepper[30],S-pcfof in Petunia CMS-pcf[31],andLat52promoter drivingorf239of common bean Sprite CMS[32],T-urf13of maize T CMS[33]have also been applied to cause male-sterile phenotypes in tobacco,petunia,or pepper for investigating the function of sterility genes.However,we obtained no transgenic plants with pollen abortion inB.napusbyAP3-drivenorf346transformation.TheAP3promoter also failed to drive the expression oforf288of Hau CMS to produce a pollen-aborted phenotype inB.napus[11].We speculate that theAP3promoter fromArabidopsisis not able to drive the expression of sterility genes in rapeseed.

Down-regulation oforf346inS.arvensisand the restorer is likely due to the interaction between fertility restoration gene andorf346.As the expression profile of CMS-associated genes is known to be regulated by fertility-restorer genes[34],we infer thatorf346is associated with Nsa CMS inB.napus.

4.2.orf346 is a chimeric gene and co-transcribed with nad3 and rps3

The Nsa CMS geneorf346identified in our study contains a fragment of thecox1gene.Almost all cytoplasmic male-sterility genes identified so far are chimeric ORFs,consisting of sequences of essential mitochondrial genes,includingcox(cytochrome oxidase),atp,nad,and other mitochondrial genes[2].Rice Honglian CMS geneorfH79[35]and wild beet Owen CMS geneorf129contain a sequence identical to part of thecox2gene[28].Radishorf463 for Don CMS is chimeric withcox1[36].orfH522for sunflower PET1 CMS[37],orf222for Nap CMS andorf224for Pol CMS[10,38]inB.napusare all chimeric withatp8.Maize S CMS geneorf77contains three segments derived fromatp9[39].In Pepper Peterson CMS,the chimericorf456encodes a protein with a segment of ATP6[40].It has been speculated[11]that CMS genes consisting of portions of mitochondrial genes competitively interact with mtETC complexes,resulting in a deficiency of energy supply.The expression ofcox1in Nsa CMS line at early flower development stage was markedly reduced relative to that in the maintainer line,suggesting that the chimeric nature of this CMS gene might be involved in the regulation of sterility.However,there is no direct evidence that the expression of genes chimeric to a CMS gene is necessary for pollen abortion in CMS.

Many CMS genes are co-transcribed with known mitochondrial genes,includingatpandnadgenes.For example,in Brassicas,orf224of Pol CMS inB.napus[9],orf288of Hau CMS[11]andorf263of tour CMS[13]inB.junceawere all co-transcribed with theatp6gene.orf222of Nap CMS inB.napusis co-transcribed with thenad5gene[10].orf108ofM.arvensisCMS inB.junceais cotranscribed with theatpAgene[14].orf346was shown to be cotranscribed withnad3andrpl12genes in the present study.It was reported[41]that the genomic region containingnad3andrps12was highly conserved evolutionarily and these two genes were co-transcribed in multiple crops.The tract upstream of this region has been frequently found to be recombined.For example,orf206was located upstream to this region and co-transcribed withnad3andrps12in radish[42].The newPetuniaCMS geneS-pcfwas also located upstream of this region,and co-transcribed withnad3andrps12[43].The gene structure of Nsa CMSorf346was different from that oforf322,an ORF with 91% sequence similarity and located at the same position in the maintainer line[22],indicating thatorf346arose from genomic recombination or rearrangement in this region during species evolution.The enrichment ofnad3andrps12as well asorf346/nad3/rps12transcripts in Nsa CMS may be associated with regulation of sterility at the post-transcriptional level.

Fig.6.A schematic diagram of the abortion mechanism of Nsa CMS.ORF346 may cause disturbance of ATP and ROS in pollen development and eventually lead to pollen abortion.

Many genes involved in respiration pathways,such asnad,rps,andcox,were suppressed in the Nsa CMS line.Mitochondria are essential for cellular energy production including ATP synthesis and electron transport,which are required for plant growth and development[44].Pollen development requires substantial energy.Accordingly,disturbances of genes for energy production in the mitochondrial genome can lead to defects of male fertility[2].A set of ribosomal proteins are components of rRNA molecules[45].Plant mutants of ribosomal proteins in mitochondria affect many aspects of plant development including leaf morphogenesis,embryogenesis,and formation of reproductive tissues[46].Thus,disturbance of genes involved in mitochondrial respiration and integrity can also account for the male sterility of Nsa CMS.

4.3.ORF346 is a cytotoxic membrane protein

Many characterized CMS proteins are toxic to the growth ofE.coli,including maize T CMS URF13,sunflower PET1 CMS ORF522,radish Ogu CMS ORF138,rice BT CMS ORF79,and WA CMS WA352,Hau CMS ORF288[2].Indeed,the CMS protein was speculated[2]to directly kill cells in the cytotoxicity model.However,almost all toxin experiments have been performed in prokaryotic systems such asE.coliand there is no direct evidence of CMS protein causing cytotoxicity in plant anther cells.The mechanism underlying the association of CMS toxic proteins with male sterility is still unknown.The toxicity of WA CMS protein toE.coliwas dependent on a transmembrane protein,but transgenic plants with truncated WA352 lacking a transmembrane region also displayed male sterility[46].The transmembrane region of ORF288 was toxic toE.colibut could not cause male sterility in plants,and ORF288 protein lacking the cytotoxic regions also caused male sterility[47].In our study,the Nsa CMS-associated protein ORF346 was toxic to bacterial cells,and its toxicity was caused by the transmembrane region.Whether the toxicity of ORF346 is the factor causing pollen abortion in Nsa CMS awaits further investigation.

4.4.Nsa CMS is associated with energy deficiency and ROS over accumulation

Previous studies[48–50]suggested that CMS is associated with an imbalance between production and scavenging of ROS in rice,cotton,and other species.One explanation for male sterility is that the CMS protein causes a deficiency in mitochondrial energy such that it cannot meet the requirements of male reproductive development[2,51,52].

The mitochondrion is a source of ROS and may be involved in oxidative stress response and trigger cell death[53].If ROS produced in cells cannot be removed effectively,accumulated ROS may lead to cell death[54].Mitochondrially derived ROS is a regulator of apoptotic cell death in oat and other species[55].In the mitochondrial genome,most CMS-associated genes are chimeric and co-transcribed with genes that encode mitochondrial subunits of respiratory enzymes involved in ROS release or ATP production.Thus,CMS is considered to be associated with the imbalance between production and scavenging of ROS[48].Excess ROS is accumulated in anthers of rice CMS lines[56,57].Excessive accumulations of ROS and malondialdehyde(MDA)in cells are both considered[48]to be inducible factors for cell apoptosis.

The structural features of CMS proteins provide a possible basis for the energy-deficiency model,and expression oforf346inE.coliconfirmed that it hindered the growth ofE.coli,along with the surplus ROS and decreased ATP content in transformed cells.These findings support the notion that energy production and ROS are essential for pollen development(Fig.6).

5.Conclusions

ORF346 caused accumulation of ROS and decrease of ATP,providing an evident association between the newly identified mitochondrial CMS geneorf346and pollen abortion of Nsa CMS.Expression oforf346inB.napusNsa CMS may cause disturbance of ATP production and ROS accumulation during pollen development.Pollen abortion in Nsa CMS plants may be caused by inadequate energy supply and excess ROS accumulation.

CRediT authorship contribution statement

Qiong Huconceived the experiments.Shifei Sang,Hongtao Cheng,Mengyu Hao,Bingli Ding and Desheng Meidesigned and performed the major experiments.Hui Wang,Wenxiang Wang,Jia Liawere responsible for field experiment management,Li Fuparticipated project management,Hongtao Cheng,Shifei Sang,Qiong Hu and Kede Liuwrote the manuscript.

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

We thank Jinxiong Shen,Huazhong Agricultural University,for valuable comments and suggestions.This study was supported by the National Key Research and Development Program of China(2016YFD0101300),the Natural Science Foundation of China(30871553),the Fundamental Research Funds for Central Nonprofit Scientific Institution(1610172017005),the Agricultural Science and Technology Innovation Program of CAAS(Group No.118),the Hubei Agricultural Science and Technology Innovation Center(201620000001048),and the China Agriculture Research System(CARS-12).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.01.008.