Rfd1, A Restorer to the Aegilops Juvenalis Cytoplasm, Functions in Fertility Restoration of Wheat Cytoplasmic Male Sterility

Cytoplasmic male sterility (CMS) is a crucial means for the utilization of heterosis, which is of great signi�cance for improving the yield and quality of hybrids. Currently, fertility restorationhas been extensively investigated in crops, but fertility restoration of CMS wheat with Aegilops Juvenalis cytoplasm are poorly understood.Here, a backcross population BC 1 F 1 derived from a cross between male sterile line Ju706A, its maintainer line 706B, and restorer line LK783, was used to map Rfd1locus by bulked segregant analysis and wheat 660K SNP genotyping.In which Ju706A displayedcomplete male sterility,and its fertility can be restored by LK783 with a pair of dominant genes Rfd1Rfd1. The locus was located to 2.4-Mb region on chromosome 1BS by markersAX-174254104 and AX-111201011.Combined with transcriptomic analysis and quantitative real-time PCR assay,TraesCS1B02G197400LC, the most likely candidate gene for Rfd1, encoded a pectinesterase which was localized in the cell wall, and was highly expressed in fertile anthers.The silencing of Rfd1 resulted in decreased fertility and heterogeneous expression of Rfd1 promoted pollen germination and affected the vegetative growth.It implies that Rfd1 is required for anther or pollen development and male fertility in CMS wheat with Aegilops Juvenalis cytoplasm. Further, a 7 bp deletion in Ju706A was employed to develop speci�c marker Xnwafu1 for molecular marker-assisted selection of restorers. Thestudy providesa new understanding for exploring the fertility restoration mechanism of CMS.


Introduction
Heterosis, or hybrid vigor is widespread in the biological world and plays an important role in improving wheat yield.Male sterility, as an effective and economical approach, is of signi cance in the utilization of heterosis (Huang et al. 1990).According to the pattern of inheritance, male sterility can be divided into cytoplasmic male sterility (CMS) and genic male sterility (GMS) (Han et al. 2018).CMS manifested as maternal inheritance, relating to the function of mitochondrial genes, and its fertility can be restored by restorer genes (Hanson and Bentolila 2004;Horn 2006;Tan et al. 2011;Feng et al. 2015).In contrast, GMS is controlled by dominant or recessive nuclear genes, which is involved in development of male gametes, affecting male fertility (Tang et al. 2006; Ke et al. 2010;Fox et al. 2017).CMS is easier to identify maintainer lines compared with GMS, and widely used in many crops, including wheat.
CMS with different types containing T -type, P-type, K -type, V -type, Q -type, S-type, and D 2 -type has been extensively researched in wheat (Tsunewaki et al. 1976;Panayotov 1980;Zhang et al. 1987Zhang et al. , 1989;;Mohr 1993; Murai and Tsunewaki 1993;Sun 2001).Among them, D 2 -CMS with Ae. juvenalis cytoplasm (designed as Ju-CMS) possessing the characteristics of strong recovery and strong resistance to wheat powdery mildew has great potential value in improving the production of hybrid wheat (Zhang et al. 2002).However, the systematic research of CMS line with Ae.Juvenalis cytoplasm remains unclear.
Parallel with the discovery of CMS line in wheat, fertility restorer genes have also been studied.In T-CMS, eleven major fertility restorer (Rf) genes, named Rf1-Rf11 in turn, have been successfully identi ed on chromosomes 1A, 7D, 1B, 6B, 6D, 5D, 7B, 5A, 4D, 4A, and 5B (Liu et al. 1999).In particular, Rf1 and Rf3 as fertility-restorer genes bind to the mitochondrial orf279 transcript and induce cleavage, preventing expression of the CMS trait (Melonek et al. 2021).In K-CMS, the fertility gene Rfk1 is located on chromosomes 1BS and the distance is 34.4 ± 3.8cM between the Rfk1 and the centromere through the analysis of the telomeres in Chinese Spring wheat (Tsunewaki 1978;Hamawaki et al. 1980).Furthermore, fertility restorer genes, Rfv1, Rfm1 and Rfn1 in Aegilops kotschyi, Ae. mutica and Ae.uniaristata cytoplasm, respectively, may share the same locus on chromosomes 1BS, the locus is designated Rf multi (Tsunewaki 2015).Recently, the Rfk1 is delimited within physical intervals of 26.0 Mb by BSR-Seq (chen et al. 2021).Although the phenomenon of D 2 -CMS was studied (Murai and Tsunewaki 1995;Zhang and Huang 1997), there are few studies on mapping and cloning of restoring genes for D 2 -CMS, especially for the cytoplasmic male sterility of Ae. juvenalis.
Pectin is one of the components of plant cell wall and plays an indispensable role in plant growth and development.Pectinesterase (PME) affects the balance of plasticity and mechanical strength that is essential for cell wall and pollen tube growth by modulating the pectin esteri cation-deesteri cation degree (Jiang et al. 2005; Tian et al. 2006).Studies have found that plant cell swelling is caused by the partial autolysis of the cell wall promoted by PME (Richard and Noat 1986).In tomato, Pmeu1, which encodes a salt-dependent PME, plays a crucial role in cell wall strength (Gaffe et al. 1997).In Chinese cabbage, BcPME37c, which is highly expressed in the stamens and its protein is localized in the cell wall, is remarkably important in the formation of the pollen wall (Xiong et al. 2019).In tobacco, exogenous PME can thicken the apical cell wall to inhibit pollen tube growth (Bosch et al., 2005a).PME by interacting with pectin methylesterase inhibitors (PMEI) to maintain cell wall plasticity and regulation pollen development.( In this study, Ju-CMS line Ju706A and restoring line LK783 were used to construct backcrossing population Ju706A//LK783/706B and the restorer Rfd1 was mapped by combining BSA with wheat 660K SNP array genotyping.TraesCS1B02G197400LC was identi ed as a strong candidate gene for Rfd1 by RNA-Seq and qRT-PCR analysis.Furthermore, barley stripe mosaic virus-induced gene silencing (BSMV-VIGS) and gene overexpression experiments were performed to verify the function of Rfd1.Finally, Xnwafu1 molecular marker was developed for marker-assisted selection.Our work lays a certain theoretical foundation for understanding the fertility restoration mechanism of cytoplasmic male sterility in wheat.Moreover, it also provides solid theoretical and technical support for the production practice of hybrid wheat.

Plant materials
Male sterile line Ju706A, maintainer line 706B and the restorer line LK783 and Ju783R were used as research materials.Ju706A is a Ju-CMS line with Aegilops Juvenalis cytoplasm.706B has the cytoplasm of common wheat and the same nuclear background as Ju706A.LK783 is a wheat line with strong fertility recovery ability against Ae.Juvenalis cytoplasm (Liu et al. 2002).These materials were used to construct genetic mapping populations.In May 2017, the hybrid combination LK783/706B was formulated and the progeny BC 1 F 1 (Ju706A //LK783/706B) population was obtained in May 2019.
Finally, a total of 506 plants were obtained and were used to ne map of restorer gene.Ju783R, a restorer line with Aegilops Juvenalis cytoplasm and the same nuclear background as LK783, carries the restorer Rfd1Rfd1.Ju783R was used for BSMV-VIGS, grown in an arti cial climate box under long-day conditions (14 h: 10 h day: night at 25°C).Moreover, 203 restorer lines, 41 maintainer lines and 293 materials from different regions of China were used for molecular marker-assisted selection.All the above materials were grown in Yangling (108° E, 34°15′ N), China.
Colombian ecotype (Col-0) Arabidopsis thaliana and transgenic plants were used for gene overexpression analysis, grown in an arti cial climate box under long-day conditions (16 h: 8 h day: night at 22°C).Nicotiana benthamiana was used for subcellular localization, planted in an arti cial climate box under long-day conditions (16 h: 8 h day: night at 26°C).

Phenotype identi cation
During the Tns, the orets and anthers were collected to observe morphological characteristics in stereomicroscope (Motic, K4000).Pollen grains were stained with 1% I 2 -KI solution, DAPI solution, aniline blue solution and photographed by stereomicroscope (Motic, K4000).In addition, anthers from male sterile and fertile plants were collected and immersed into glutaraldehyde solution, SEM and TEM were performed as described by (Liu et al. 2018).SEM and TEM images were collected using the scanning electron microscopy (JSM-6360LV, JEOL) and transmission electron microscope (JEM-1230, JEOL), respectively.

DNA extraction and Bulked segregant analysis
Total genomic DNA of Ju706A, LK783 and BC 1 F 1 individuals was extracted from fresh leaves using cetyltrimethyl ammonium bromide (CTAB) protocol with some modi cations (Han et al. 2015).According to the result of the fertility survey in the BC 1 F 1 segregation population, sterile gene pool (SP) and fertile gene pool (FP) were constructed using the bulked segregant analysis (BSA) method.20 extremely sterile plants with 0% seed setting rate were selected to form SP and 20 extremely fertile plants with seed setting rate more than 85% were selected to form FP. The SP, FP and two parents were used to screen polymorphic molecular markers and acquire sequencing data via 660K SNP arrays.

660K SNP array genotyping and identi cation of candidate SNPs
The two bulked DNA pools and the two parents were genotyped by 660K SNP arrays at Capitalbio Technology Co. (Beijing; http://www.capitalbiotech.com).The original scan results of the SNP array were genotyped using AAS from Affymetrix or GenomeStudio from Illumina.Based on the genome-wide SNP genotyping results, quality control of SNPs was performed to obtain high-quality SNPs.Dish QC was used for quality control and samples with dish QC higher than 0.82 were used for subsequent analysis.
The fertility restorer gene was mapped based on high-quality SNPs and BSA methods.Moreover, the proportion of differential SNPs in each chromosome was calculated after quality control.The experiment was performed in two biological replicates.

SSR and SNP primers design
Primers for SSR markers were designed using the Primer Server tool available through the Triticeae Multiomics Center (http://202.194.139.32/PrimerServer/) and SNP markers were designed with BatchPrimer3 v1.0 (https://probes.pw.usda.gov/batchprimer3/)(Table S3).Before screening the markers in BC 1 F 1 populations, the SSR markers and SNP markers were used to identify polymorphisms between two parents and two pools.The PCR ampli cation conditions were as follows: 94°C for 5 min; followed by 35 cycles of 94°C for 50 s, 60°C for 50 s, 72°C for 1min; and 72°C for 10 min.The PCR products were tested on polyacrylamide gel electrophoresis (PAGE) (Han et al. 2015) and silver stained (Liang et al. 2014).According to the obtained polymorphic markers and the recombinant plants, the genetic linkage map was drawn using Joinmap4.0software with the maximum likelihood method.

Prediction of the candidate gene and sequence analysis
After identifying the polymorphic markers closely related to Rfd1, functional annotation of interval genes were obtained from JBrowse tool available through the Triticeae Multi-omics Center (http://202.194.139.32/jbrowse/).To acquire the full-length coding sequences of Rfd1, the speci c primers were designed (Table S3) and Rfd1 was cloned in Ju706A and LK783 using KOD DNA Polymerase (TOYOBO Co., Osaka, Japan).The puri ed PCR product was ligated into the PBM16A vector (Biomed Gene Technology Co., Beijing, China) and transformed into Escherichia coli.Subsequently, positive independent clones were picked and sequenced in TsingKe Biotechnology Co., Ltd (Beijing; http://www.tsingke.net).The sequence alignment was analyzed using DNAMAN software.

RNA extraction and expression analysis of candidate genes
Total RNA was extracted using RNA Extraction kit (Tiangen, Beijing, China) and rst-strand cDNA was prepared using a PrimeScript First-strand cDNA Synthesis Kit (Roche, Switzerland).In order to determine expression level of candidate genes, qRT-PCR with 2 × Real Star Green Power Mixture (Gene-star Biosolutions Co., Beijing, China) was detected in QuantStudioTM Real-Time PCR System (Applied Biosystems, USA).Sequence-speci c primers for qRT-PCR were designed using Triticeae Multi-omics Center (http://202.194.139.32/PrimerServer/)(Table S3).The qRT-PCR ampli cation conditions were as follows: 95°C for 30 s; followed by 45 cycles of 95°C for 15 s, 60°C for 30 s.The actin1 (GenBank: GQ 339766.1)for wheat was used as an internal control.The relative expression levels of the target genes were calculated using the 2 −ΔΔCT method and each sample was performed in three technical replicates.

Subcellular localization
To determine the subcellular localization of Rfd1, The CDS of Rfd1 without the stop codon was ampli ed and inserted into transient expression vector pCAMBIA1302 with a green uorescence protein (GFP) coding region to generate 35S:Rfd1-GFP, which was driven by the cauli ower mosaic virus (CaMV) 35S promoter.The 35S: GFP (control), 35S: Rfd1-GFP and cell membrane marker PIP2A-mCherry were transformed into the Agrobacterium tumefaciens GV3101 and then injected into the leaves of Nicotiana benthamiana plants (Sparkes et al. 2006).For plasmolysis treatment, 0.8 mol/L mannitol solution was used in this study.Fluorescence signals were visualized using a IX83-167 FV1200 (Japan) confocal laser-scanning microscopy.

Plasmid construction for BSMV-VIGS and wheat infection
In order to verify the function of Rfd1, BSMV-VIGS experiment was carried out.The cDNA fragment (217bp) of Rfd1 was selected to construct BSMV-VIGS vector and ampli ed in Ju783R by speci c primers (Table S3).Viral vector construction and viral RNA transcription in vitro were as described (Yang et al. 2021).Transcripts in vitro including α/β/γ, α/β/γ-TaPDS and α/β/γ-Rfd1 were mixed and diluted with FED solution to obtain three infective solutions: BSMV: 000, BSMV: TaPDS and BSMV: Rfd1.When the ag leaves were fully picked out, BSMV infective solution was applied to the ag leaf and the next leaf of wheat.After the infection, the phenotypes of leaves and anthers were observed by stereomicroscope (Motic, K4000).Total RNA was extracted from the anthers of uninfected, BSMV: 000infected and BSMV: Rfd1-infected plants, and then transformed into cDNA for qRT-PCR.The determination of pectin content adopted the carbazole colorimetric method and the speci c method referred to Yu et al. (1996).After the wheat matures, the seed setting rate was investigated: seed setting rate (%) = (number of seeds per spike/spikelet number per spike)×100.

Construction of overexpression vector and Arabidopsis transformation
The rst-strand cDNA of LK783 were used for PCR ampli cation and primers used for overexpression analysis were designed by Triticeae Multi-omics Center (http://202.194.139.32/PrimerServer/)(Table S3).Subsequently, the ampli cation product was cloned into pCAMBIA1302 by NcoI and SpeI digestion and homologous recombinase ligation to generate pCAMBIA1302: Rfd1.The overexpression vector pCAMBIA1302: Rfd1 was introduced into GV3101 (Agrobacterium tumefaciens) and transformation of Arabidopsis thaliana by the method of oral dip (Clough and Bent, 1998).
Seeds of the three Rfd1-overexpressing lines and wild-type were germinated on MS basal medium (0.22g MS powder + 0.75g sucrose + 0.8g agar) in growth chamber.Half of the seedlings were transplanted into the soil, and the other half were collected for qRT-PCR.Relative expression levels for Rfd1 in three transgenic lines were normalized using the actin2 (GenBank: NC 003074.8)for Arabidopsis as standard.Morphological characteristics and alexander staining of pollen grains were observed by stereomicroscope (Motic, K4000).

Statistical analysis
The data of all the above experiments were counted in the excel table, Student's t-test in the GraphPad Prism version 8.0.1 for Windows (GraphPad Software, San Diego, California USA, www.graphpad.com)were used to perform all statistical analyses.The values were designated as the means ± SD of three biological replicates per measurement.

Results
Ju706A, a male sterile line with Aegilops Juvenalis cytoplasm, performs complete male sterility with defective anther and pollen development To observe the phenotypic characteristics of male sterile line Ju706A, the restorer line LK783 and their F 1 , the orets and anthers from them were collected at the trinucleate stage (Tns).There was no signi cant difference in pistil between LK783 and Ju706A (Fig. 1a, d).However, compared with LK783, Ju706A displayed complete male sterility with indehisced anthers and aborted pollen grains, which were not dyed su ciently with I 2 -KI (Fig. 1b-c, e-f).In addition, the fertility of F 1 was normal when pollinated with LK783 as pollen ponors, indicating that the fertility of Ju706A can be restored by LK783 (Fig. 1g-i).
To further investigate the cytological differences between LK783 and Ju706A, SEM analysis was performed at the Tns.The result largely consistent with the above phenotypic characteristics, the Ju706A produced thin and shriveled anthers comparing with LK783 (Fig. 1j, q).In contrast to the neat and parallel cutin polymers on the outer epidermis of LK783, the outer epidermis of Ju706A was wrinkled and irregular in shape (Fig. 1k, l, r, s).The inner epidermis of both LK783 and Ju706A contained Ubisch bodies, but the Ubisch bodies arrangement of Ju706A were sparse and messy (Fig. 1m, n, t, u).Moreover, microspores of LK783 showed large and plump features, whereas Ju706A showed a small, shrunken morphology (Fig. 1o, p, v, w).These results indicated the anthers and pollen development was disrupted in Ju706A.

Rfd1 is mapped to 2.4-Mb interval by BSA and 660K SNP array genotyping
To investigate the inheritance pattern of fertility restorer genes, Ju706A and LK783 were separately used as the female and male parents to construct BC 1 F 1 populations.In segregation populations, there were a total of 506 plants: 232 sterile plants and 274 fertile plants.According to the χ 2 test, the BC 1 F 1 population separation ratio was 1: 1 (χ 2 = 3.48 < χ 2 0.05,1 = 3.84) (Table S1), thereby indicating that the fertility restorer character is controlled by single dominant gene locus (designed as Rfd1) from LK783.To map Rfd1, 326 SSR primers evenly distributed on 21 chromosomes of wheat were used to detect polymorphism in the two parents and two DNA pools (extremely sterile mixture pools and extremely fertile mixture pools).Consequently, nine polymorphic SSR primers (Xgwm18, Xbarc137, Xwmc694, Xbarc061, Xcfd59, Xgpw1143, Xgpw4331, Xgpw3010, Xgpw4346) located on chromosome 1B were identi ed and used to detect in 180 sterile individuals.The results showed that Rfd1 was delimited in a 14.8 cM region between Xgwm18 and Xgpw1143 markers, which were closely linked to the Rfd1 on 1BS (Fig. 2c, Table S2).
To further develop new polymorphic markers and narrow the target region, wheat 660K SNP array genotyping was carried out.In total, 23,086 differential SNPs (Used SNPs) were detected in the two parents and two DNA pools (Fig. 2a).Especially, 6,287 differential SNPs (27%) were located on chromosome 1B, which was the highest proportion among 21 chromosomes of wheat (Fig. 2b).Furthermore, according to the results of primary mapping, 50 pairs of SNP markers on chromosome arm 1BS were selected to con rm the ne mapping interval.Six markers (AX-94768879, AX-95117169, AX-174254104, AX-111201011, AX-94793363, AX-111281507) were found to exhibit polymorphisms both in parents and two DNA pools.Subsequently, polymorphic SNP markers were used to detect the recombinant plants among 232 sterile individuals.The linkage analysis results showed that Rfd1was narrowed to an approximately 2.4-Mb interval (134421108-136845581) between the SNP markers AX-174254104 and AX-111201011 (Fig. 2d, Table S2).Detection of partial polymorphic markers in sterile and fertile plants were shown in Fig. S1.

TraesCS1B02G197400LC is a strong candidate gene for Rfd1
According to the physical distance of the linkage markers, 19 genes were predicted to be located between AX-174254104 and AX-111201011 markers using the Triticeae Multi-omics Center (http://202.194.139.32/jbrowse/) (Fig. 2e, Table S4).In addition, 1,356 differentially expressed genes (DEGs) were ltered based on the RNA-Seq data of Ju706A and LK783.Four DEGs, comprising TraesCS1B02G197400LC, TraesCS1B02G115300, TraesCS1B02G115400, TraesCS1B02G115500, were screened in the candidate interval (Fig. 3a, Table S4).To identify candidate genes for Rfd1, qRT-PCR assay of these four DEGs was performed using RNA from different tissues (root, stem, leaves, anthers) and different developmental stages of anthers (Euns, Luns, Bns, and Tns).The results showed that TraesCS1B02G115300, TraesCS1B02G115400, and TraesCS1B02G115500 had similar expression trends, which were expressed in various tissues and the highest in roots.However, TraesCS1B02G197400LC, a gene encoding a pectinesterase, was the highest expressed in anthers at the Tns (Fig. 3b-f).
To understand the sequence differences between Ju706A and LK783, the full-length encoding sequence of TraesCS1B02G197400LC was isolated from Ju706A and LK783.It is characterized by 1350bp with only one exon and encodes 449 amino acids, a 7-bp deletion (CGGGGGC) was occurred at the 1176 bp in Ju706A, resulting in a frameshift mutation at the 393th amino acid (Fig. S2, Fig. S3).Taken together, TraesCS1B02G197400LC was suggested as a strong candidate gene for Rfd1.

Rfd1 is localized in the cell wall
To determine the subcellular localization of Rfd1, the 35S: GFP (as the control) and 35S: Rfd1-GFP were constructed (Fig. 4a) and transiently expressed in the leaves of N. benthamiana.The results indicated the 35S: GFP signals were expressed throughout the cell (Fig. 4b), while 35S: Rfd1-GFP signals were observed in the cell membrane or cell wall (Fig. 4c).To further verify the localization of Rfd1, Agrobacterium tumefaciens containing cell membrane marker PIP2A-mCherry and 35S: Rfd1-GFP was co-in ltrated into tobacco leaves.The co-expressed results showed that the GFP signals of 35S: Rfd1-GFP overlapped with the RFP signals of membrane marker PIP2A-mCherry (Fig. 4d).In the plasmolyzed cells, 35S: Rfd1-GFP signals were observed in the cell wall and PIP2A-mCherry signals in the cell membrane (Fig. 4e).These results indicated that the Rfd1 is localized in the cell wall.

Silencing of Rfd1 leads to decreased fertility
To identify the function of Rfd1 in wheat, BSMV-VIGS assay was performed on the restorer line Ju783R carrying the restorer genes Rfd1Rfd1.When the ag leaves of Ju783R were fully picked out, ag leaves were inoculated with BSMV: 000, BSMV: TaPDS and BSMV: Rfd1, respectively.The results showed that leaves of the uninfected plants were normal dark green, the leaves of BSMV: 000 plants appeared stripes and the leaves of BSMV: TaPDS plants displayed photobleaching, which indicated that the silencing of gene was effective.In addition, severely streaked leaves emerged in BSMV: Rfd1-infected plants (Fig. 5a).Compared with the normal orets and anthers in control, including uninfected, BSMV: 000 and BSMV: TaPDS plants, BSMV: Rfd1 plants exhibited skinny orets and indehisced anthers (Fig. 5b, Fig. S4a).Furthermore, to determine the silencing effect of Rfd1, qRT-PCR assay and seed setting rate investigation were performed.Compared with the control plants, the expression of Rfd1 was remarkably reduced in BSMV: Rfd1 plants (Fig. 5f).The seed setting rate reached more than 80% in the control plants, while there was less than 40% in BSMV: Rfd1 plants (Fig. 5c, g).In addition, due to the decreased expression of Rfd1, the normal degradation of pectin was hindered, so that the pectin content of BSMV: Rfd1 plants was higher than that of the control (Fig. 5h).
To investigate the causal factor responsible for the decreased seed-setting rate of the BSMV: Rfd1 plants, the pollen vitality was observed.The results showed that the pollen grains of the control plants were completely fertile.However, the pollen grains of BSMV: Rfd1 plants showed incomplete staining (Fig. 5d), displayed reduced pollen germination rate (Fig. 5e, i), contained two abnormal round sperm nuclei and a vague vegetative nucleus (Fig. S4b), and presented small and shrunken morphology (Fig. S4c).Therefore, the above results indicated that the silencing of Rfd1 affected pollen vitality and effectively reduced fertility.

Silencing of Rfd1 exhibits abnormal development of anthers and microspores
To further explore the causes of male sterility, SEM and TEM observations were performed at the Tns.SEM showed that the anthers of the control plants were large and dehisced, and the outer epidermis of the anthers was highly de ned with relatively glossy (Fig. S5a-b, d-g).By contrast, the anthers of BSMV: Rfd1 plants failed to split and the outer epidermis was scattered and wrinkled.(Fig. S5c, h, i).Moreover, The Ubisch bodies were relatively sparsely arranged on the inner epidermis and the microspores were deformed and shrunken in the BSMV: Rfd1 plants compared with control plants (Fig. S5j-u).
Additionally, TEM was applied to observe the dynamic process of anthers development from the Euns to Tns.At the Euns, there were no remarkable morphological differences in tapetum and microspore development between BSMV: 000 and BSMV: Rfd1 plants.At the Luns, the tapetum layer of BSMV: 000 plants began to degrade, while BSMV: Rfd1 plants were less degraded compared with BSMV: 000 plants.At the Bns, the tapetum layer continued to degrade, but the tapetum layer of BSMV: Rfd1 plants was still thicker than that of BSMV: 000 plants.Until the Tns, the tapetum layer of BSMV: 000 plants was almost completely degraded with leaving only cell residues and the pollen grains were released from the ruptured pollen wall; whilst BSMV: Rfd1 plants had a visible tapetum layer and the anther locule remained closed with no pollen release (Fig. 6a, b, Fig. S6a, b).Simultaneously, development of microspores and microspore walls were also observed.The microspore walls contained the intine layer and the exine layer including tectum, baculum and nexine.At the Euns, the microspores and microspore walls development in the BSMV: 000 plants was equivalent to that of the BSMV: Rfd1 plants.From the Luns and Tns, The microspores of BSMV: Rfd1 plants were shrunk and the exine of the anthers was arranged irregularly accompanied by abnormal accumulation of sporopollenin compared with BSMV: 000 plants.(Fig. 6c, d, Fig. S6c, d).Taken together, electron microscopy results suggested that Rfd1 is required for anther development, tapetum degradation and microspore wall formation.

Heterogeneous expression of Rfd1 promoted pollen germination in Arabidopsis thaliana
To further explore the function of Rfd1, an overexpression vector was constructed and three independent transformants (OE-Rfd1-3, OE-Rfd1-8 and OE-Rfd1-17) were obtained (Fig. 7a, b, c).To observe pollen activity, alexander staining and pollen germination in vitro were performed.The results showed that pollen grains of both wild type plants and overexpressed lines could be fully stained with alexander solution (Fig. 7e), but in the pollen germination experiment, the pollen germination rate was 64.6% in wild type plants, while the results in overexpressed lines were signi cantly increased from 72-84.1% (Fig. 7d,   f).
In addition, compared with the wild type plants, the overexpressed lines showed delayed mossy and owering, and the number of rosette leaf increased signi cantly (Fig. S7a, c).At 40 days, overexpressed plants were observed to have more in orescence side branches (2)(3)(4) and tended to grow simultaneously with the main stem (Fig. S7b, d).These results indicated that heterogeneous expression of Rfd1 affected the vegetative growth and promoted pollen germination in Arabidopsis thaliana.

Development of molecular markers
Based on the 7 bp deletion of TraesCS1B02G197400LC in Ju706A, an InDel marker, Xnwafu1, was developed for molecular marker-assisted selection.Here, 180 sterile plants of BC 1 F 1 population, 203 restorer lines and 41 maintainer lines were used to verify the accuracy of Xnwafu1 marker.The results of polyacrylamide gel electrophoresis showed that there were no products in the sterile plants and maintainer lines, but the restorer lines had products at 260 bp.The Xnwafu1 can identify all sterile plants (100%) in BC 1 F 1 populations, 40 maintainer lines (97.6%) and 190 restorer lines (93.4%).Furthermore, 293 materials from different regions of China were used for fertility identi cations and marker detection assays.According to results of fertility identi cations (seed setting rate of F 1 is higher than 60% was considered fertile) and Xnwafu1 marker test, the identi cation rate of the marker was 70.3% (Table S5, Fig. S8).Hence, Xnwafu1 can be used as a good molecular marker for molecular marker-assisted selection of restorers.

Discussion
BSA combined with SNP array genotyping to accelerate gene mapping Bulked segregant analysis (BSA) is an important method to identify genetic variants related to target genes and BSA combined with RNA sequence (RNA-seq) has been an e cient gene mapping technology which can accurately locate the target gene ( ).In this study, we used the wheat 660K genotyping technology combined with the BSA to develop SNP markers and narrow the restorer gene to the 2.4-Mb interval.Thus, the availability of the next-generation technologies and high-throughput SNP genotyping platforms greatly advanced gene mapping and cloning in wheat.
TraesCS1B02G197400LC encodes a PME, which may be involved in pollen wall formation and pollen tube germination TraesCS1B02G197400LC encodes PME, which is an important enzyme in plant cell wall development.In plants, the abnormal development of the pollen wall can lead to male sterility.In Arabidopsis, the ms2 mutant displays defective exine layer and anomalous sporopollen accumulation process, leading to male sterility (Aarts et al.1997).The mutation of DEX1 exhibits delayed the exine formation, damaged cell membrane structure and disorganized sporopollenin accumulation, resulting in male sterility (Paxson- Sowders et al.1997;2001).The mutant ms33 shows advanced inner wall formation and declined fertility (Fei and Sawhney 2001).In this study, the silencing of Rfd1 exhibits skinny and shrunken pollen grains, scattered and irregular exine, and aberrant sporopollenin accumulation, causing the pollen wall defect.
In previous studies, it is found that the apical region of pollen tube is almost composed of a pectin network structure.Most of the esteri ed pectin exist in the apical region of the pollen tube to maintain the growth of the pollen tube, while the deesteri ed pectin is more concentrated in the area far away from the apical (Bosch et al. 2005b;Qun et al. 2007).Therefore, PME was also reported to participate in pollen development, pollen tube germination and growth.In Arabidopsis thaliana genome, there are 67 PME genes, of which at least 18 genes are highly expressed in pollen (Pina et al. 2005).In particular, AtPPME1and VGD1 encoding a PME play an important role in pollen tube growth (Jiang et  It can be seen from the above that the diversity of fertility restorer genes may be due to the complex fertility restoration mechanism of CMS. Furthermore, there is a single locus on 1BS, Rf multi , responsible for fertility restoration to Ae. Kotschyi, Ae.Mutica and Ae.Uniaristata cytoplasm (Tsunewaki 2015).However, the restorer Rfk1 (TraesCS1B02G197400LC) located on 1BS chromosome against Ae.Kotschyi cytoplasm is a dominant gene locus derived from LK783, which differs from Rf multi (Chen et al. 2021).In this study, Rfd1 was mapped to a physical region of 2.4-Mb containing 19 candidate genes on chromosome 1BS.Furthermore, it was found that only TraesCS1B02G197400LC was highly expressed in anthers by RNA-Seq and qRT-PCR analysis.Coincidentally, we found that TraesCS1B02G197400LC was strongly expressed in the anther of LK783, but was weakly expressed in KTP116A and Ju706A (Fig. S9a, b).Nucleotide sequence alignment showed that both KTP116A and Ju706A had a 7 bp deletion (Fig. S9c, d).In addition, the molecular marker Xnwafu-4 can also distinguish between LK783 and Ju706A (Fig. S9e).Thereby suggesting that Ju706A and KTP116A may have the same restorer locus derived from LK783.

Conclusion
In conclusion, fertility restorer locus for Aegilops Juvenalis cytoplasm in wheat, Rfd1, was mapped to physical intervals of 2.4-Mb on chromosome 1BS by BSA and 660K SNP array genotyping in BC 1 F 1 population.TraesCS1B02G197400LC encoding a pectinesterase was identi ed as a candidate gene for Rfd1, which was strongly expressed in the anthers and its protein was localized in the cell wall.Downregulation of Rfd1 reduced pollen activity and affected the normal development of anther and microspores, leading to male sterility.Moreover, heterologous overexpression of Rfd1 increased pollen vigor and affected vegetative growth.The research provided a good foundation for further study of fertility restoration mechanism in cytoplasmic male sterility with Aegilops Juvenalis cytoplasm.

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download.

Figure 2 Fine
Figure 2

Figure 3 Expression
Figure 3

Figure 7 Identi
Figure 7 FigureS1.docxFigureS2.docxFigureS3.docxFigureS4.docxFigureS5.docxFigureS6.docx FigureS7.docxFigureS8.docxFigureS9.docxTableS1.docxTableS2.xlsxTableS4.xlsxTableS5.xlsx Abe et al. 2012; Trick et al. 2012; Nestler et al. 2014; Ramirez-Gonzalez et al. 2015; Garcia et al. 2016; Singh et al. 2016; Du et al. 2017; Li et al. 2018; Sun et al.2018; Wang et al. 2018).Besides, BSA coupled with resequencing has been a rapid way to identify many important genes (Takagi H et al. 2013; Fletcher RS et al. 2016; Zhang et al. 2017; Tan et al. 2018; Wei et al. 2019).In recent years, the combination of BSA and SNP array genotyping technique has become more and more mature, and become a widely used technology in gene mapping.Wheat 9K array (Cavanagh et al. 2013), 90K array (Wang et al. 2014) and 660K array (Jia and Zhao.2016) are used for gene mapping in wheat.Especially, wheat 660K array is the chip with the highest SNP density and it has been widely used in marker development (Jin et al. 2016; Li et al. 2017; Wu et al.2018; Zhai et al.2018 (Praveen et al. 2015)2)10)l.2006bage,antisensetransgenicplants of BcMF3 display pollen abortion and the decreased germination rate(Wang et al. 2004;Liu et al. 2006).In maize, ZmGa1P encoding a pollen-speci c PME regulates pollen tube growth(Zhang et al. 2018).Moreover, the coordinated action of PME and PMEI participates in pollen germination and pollen tube growth(Shi and Yang et al. 2010).In this study, pollen germination experiments were conducted on wheat and Arabidopsis thaliana, the pollen grains of Rfd1 silencing plants in wheat hardly germinated, while overexpress of Rfd1 in Arabidopsis thaliana promoted pollen tube germination and growth.Therefore, Rfd1 may have an indelible role in pollen wall formation and pollen development.Con rming function of Rfd1 based on BSMV-Virus induced gene silencing Virus induced gene silencing (VIGS) is a rapid and high-throughput technology for identifying gene functions.Recently, BSMV-VIGS has been widely used in the functional research of genes related to plant growth and development, stress resistance and disease resistance.BSMV-VIGS is the rst time to establish and induce the silencing of Lr21, Hsp90, RAR1 and SGT1 in wheat, identifying that Lr21 has the function of wheat stripe rust resistance (Sco eld et al.2005).The function of PDS and GFP in Dasypyrum villosum is successfully silenced by BSMV-VIGS (Wang et al. 2009).BSMV-VIGS system is also successfully used in barley (Hein et al. 2005), Brachypodium distachyon (Demircan and Akkaya 2010), Arabidopsis thaliana (Manmathan et al. 2013), maize (Jarugula et al. 2018) and wheat (Yang et al. 2017; Han et al. 2021).In this study, to verify whether TraesCS1B02G197400LC was the candidate of Rfd1, BSMV-VIGS experiment was carried out.The results showed that the leaves of Rfd1 silencing plants appeared white streaks, and the anthers were not dehisced.TheI 2 -KI staining, DAPI staining, aniline blue staining and pollen germination experiment all showed that the pollen grains were partially sterile.In addition, the seed setting rate and the gene expression level of Rfd1 silencing plants were decreased compared with the control plants.These results demonstrated that the silencing of Rfd1can lead to partial loss of function and decreased fertility.Fertility restoration in Ju706A is controlled by Rfd1 derived from LK783In recent years, fertility restorer genes for CMS have been mapped and cloned in various crops, such as sugar beet(Matsuhira et al. 2012), maize(Feng et al.2015), onion(Kim et al.2015), rice(Huang et   al.2015), sorghum(Praveen et al. 2015).Apart from Rf2 (encoded a mitochondrial glycine-rich protein) and Rf17 (encoded a mitochondrial protein) in rice, Rf2 (encoded a mitochondrial aldehyde dehydrogenase) in corn, Rf1 (encoded a protein resembling OMA1) in sugar beet, the rest of the fertility restorer genes encode pentatricopeptide repeat (PPR) protein.