Fine mapping of the dominant female sterility gene and novel model of hybrid seed production in cabbage (Brassica oleracea L. var. capitata)

Abstract

Hybrid breeding based on male sterility requires the removal of male parents, which is time- and labor-intensive; however, the use of female sterile male parent can solve this problem. In the offspring of distant hybridization between Brassica oleracea and Brassica napus, we obtained a mutant, 5GH12-279, which not only fails to generate gynoecium (thereby causing female sterility) but also has serrated leaves that could be used as a phenotypic marker in seedling screening. Genetic analysis revealed that this trait was controlled by a single dominant gene. Further analysis revealed that Bo2g005230, an orthologous gene of LATE MERISTEM IDENTITY1 (LMI1) in Arabidopsis, was predicted as the candidate gene and was renamed BoLMI1c. Sequence analysis revealed that homoeologous exchange (HE) occurred within the BoLMI1c gene body of 5GH12-279, which resulted in the generation of a novel fusion transcript. Two pairs of primers, N5230-1F/1R and N5230-2F/2R, were designed and successfully used for the identification of different genotypes of BoLMI1c. Transcriptome analysis revealed that BoLMI1c orchestrates the expression of several related biological processes and transcription factors. Furthermore, we found that self-pollination with mason bees produced no seeds in 5GH12-279, whereas the near-isogenic line 5GH12-170 produced seeds that were normal. Therefore, a new labor-saving hybrid seed production system with no need to remove the male parents, which is especially important for mechanized harvest in the future, has been proposed. Our study provides a valuable source of dominant female sterility and suggests the potential utilization of the female sterile line in hybrid breeding for mechanized harvest.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00122-025-05002-x.

Introduction

In plants, the ovary, style and stigma compose single or compound pistils, which together constitute the gynoecium (Bollier et al. 2018). During sexual reproduction in plants, pollen grains from stamens germinate on stigmas of the gynoecium, penetrate the pistil style with elongated pollen tubes, pass through the micropyle and enter the ovule to achieve fertilization. The characteristics of the gynoecium, including the stigma exsertion rate, stigma length, stigma breadth, and pistil style length, can greatly affect the effectiveness of pollination and subsequently influence yield; thus, the absence of the gynoecium can have devastating impacts on hybrid seed production (Liu et al. 2015; Zhou et al. 2017; He et al. 2019).

The gynoecium development process is highly dynamic and requires complex gene regulatory networks (Zúñiga-Mayo et al. 2019; Ballester et al. 2021). The establishment of the gynoecium involves organ identity genes that determine the whorl formation pattern in flowers and meristem identity genes that stop floral meristem activity and achieve floral meristem termination (Saddic et al. 2006; Zúñiga-Mayo et al. 2019). In Arabidopsis thaliana, known meristem identity genes trigger the formation of the first flower and contribute to establishing and maintaining floral meristem identity. These genes include LEAFY (LFY), APETALA 1 (AP1), TERMINAL FLOWER 1 (TFL1), LATE MERISTEM IDENTITY 1 (LMI1), CAULIFLOWER (CAL), FRUITFULL (FUL), CRABS CLAW (CRC), WUSCHEL (WUS), WUSCHEL RELATED HOMEOBOX 1 (WOX1), CLAVATA 1 (CLV1), SHORT VEGETATIVE PHASE (SVP), AGAMOUS-LIKE 24 (AGL24), etc (Saddic et al. 2006; Grandi et al. 2012; Zúñiga-Mayo et al. 2019; Yao et al. 2024; Cui et al. 2024). Mutations in these genes cause very dramatic changes in meristem identity, such as increased or decreased numbers of floral organs and enlarged or diminished floral meristems. LMI1 is a homeodomain leucine zipper (HD-Zip) transcription factor (TF) that belongs to the HD-Zip I subfamily and functions as a meristem identity regulator (Wang et al. 2021). LMI1 is a positive regulator of AGL24 and SVP and a target of LEAFY; the combination of LMI1 and LEAFY activates CAL to influence meristem identity (Saddic et al. 2006; Wang et al. 2021). In addition, the LMI1 gene also has additional LFY-independent roles in leaf morphogenesis and bract formation.

Dominant genes that control the disappearance of the entire gynoecium and subsequently cause female sterility have not been reported in plants. Some female sterility genes have been reported, mainly in rice, and all of the mutations that cause female sterility are recessive (Mao et al. 2021; Li et al. 2022; Wang et al. 2024a, b; Huang et al. 2024). For example, Li et al. (2022) identified a thermosensitive female sterile mutation in rice, a spontaneous single-nucleotide polymorphism (SNP) mutation in the TFS1 gene, and elucidated how the mutation controls female sterility‒fertility conversion through its response to temperature. A novel mutation (A1106G) in the MEL2 gene resulted in female sterility in rice (Wang et al. 2024a, b). In cucumber, Cheng et al. (2022) reported that CsSPT and CsALC function in female fertility and carpel fusion by regulating transmitting tract (TT) differentiation.

Female sterile mutants are unable to deposit pollen on the stigma, disrupting seed setting and generating economic and production losses. However, if female sterile mutants are used as male parents and male sterile individuals are used as female parents in a hybrid production system, these mutants can also have great commercial value. All the seeds produced in this type of hybrid production system are hybrids that are needed because neither parent alone is able to produce seeds; therefore, the production of these hybrid seeds can greatly reduce the labor required for planting and harvesting via mechanized management (Qu et al. 2012; He et al. 2019). Increasing our understanding of mutations in the female sterility gene and advancing the utilization of the female sterility gene for hybrid breeding will be beneficial for mechanized management in the breeding process in the future.

In this study, we identified a female sterile mutant, 5GH12-279, in cabbage. The female sterility trait was found to have a rare dominant inheritance pattern, as shown in the F₁ and BC₁F₁ populations. Bulked segregant analysis sequencing (BSA-seq) and map-based cloning revealed that the candidate gene was located in the 108-kb interval on chromosome C02. Bo2g005230 was predicted as the candidate gene on the basis of gene function annotation, expression analysis and sequence variation analysis. Furthermore, InDel-based primers were developed for early screening of the dominant female sterility gene, and the mutant was preliminarily applied in cabbage breeding. These findings provide valuable dominant female sterile sources and new perspectives for the development of cabbage hybrids at a lower cost and with greater efficiency.

Materials and methods

Plant materials and growth conditions

5GH12-279 is a spontaneous mutant with a female sterile phenotype that was identified in the BC₅ generation of the cabbage Ogura cytoplasmic male sterility (CMS) fertility restorers. Cabbage Ogura CMS fertility restorer lines were created by distant hybridization between Brassica oleracea and Brassica napus and multigenerational backcrossing in our previous study to introduce the Rfo fertility restorer gene from Brassica napus into Brassica oleracea (Yu et al. 2016, 2017, 2020; Ren et al. 2020). The Ogura CMS fertility restorer line 5GH12 is a cross combination from the BC₅ generation. Through molecular marker-assisted screening (MAS), we identified five Rfo-positive individuals within this line. Unexpectedly, two of these individuals (5GH12-84 and 5GH12-279) exhibited a female sterile phenotype, while the remaining three (5GH12-170, 5GH12-340, and 5GH12-369) displayed normal fertility. We attribute this phenotypic segregation to a mutation that likely occurred via homologous exchange during earlier generations of distant hybridization. Both female sterile individuals (5GH12-84 and 5GH12-279) are heterozygous for this mutation. 11-192 is a cabbage inbred line with normal gynoecium. 11-192 was crossed with 5GH12-279 to obtain the F₁ population and then the F₁ generation was self-crossed to obtain the F₂ population. Female sterile individuals from the F₁ population were crossed with 11-192 to generate the BC₁F₁ population. All the plant materials used in this study were grown in a greenhouse under long-day conditions (16 h of light and 8 h of dark) at the Institute of Vegetables and Flowers, Chinese Academy of Agriculture Sciences (IVFCAAS, Beijing, China). When female sterile mutants and any individuals with the female sterile phenotype in our study were used as female parents or needed to self-pollinate, the plants were grown in a greenhouse (at a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C, both with 16 h of light and 8 h of dark).

Paraffin sectioning and SEM analysis

Flower buds from 5GH12-279 and 5GH12-170 at different stages were collected and immediately fixed with FAA. Flower bud samples were embedded in paraffin, transversely sectioned into 4-μm slices with a Leica microtome and stained with 0.25% (w/v) toluidine blue O. The slides were viewed with an Olympus BX51 microscope and photographed with a digital camera.

Flower buds from 5GH12-279 and 5GH12-170 were fixed in electron microscope fixative at 4 °C overnight. The samples were washed three times for 10 min each time with 0.1% phosphoric acid buffer, dehydrated in a graded ethanol series, and substituted with 100% ethanol. A critical-point drier was used to dry the samples twice, once for 15 min and once for 20 min. The samples were dissected under a microscope, sputter coated with platinum, and viewed with a scanning electron microscope.

BSA-seq and fine mapping of the female sterility gene

The 11-192 individuals were used as female parents in crosses with 5GH12-279 to generate the F₁ population. Thirty flower buds from normal gynoecium-phenotype individuals and thirty flower buds from female sterile individuals of the BC₁F₁ population were randomly selected to construct two DNA pools. High-quality genomic DNA from the two bulks and two parental lines (11-192 and 5GH12-279) was extracted using a Plant Genomic DNA Kit (Tiangen, Beijing, China) and subsequently used for constructing Illumina libraries via the NEBNext® Ultra™ II DNA Library Prep Kit for Illumina®. Pools were sequenced on an Illumina HiSeq with 150 bp paired-end reads. The sequencing depth of the parents was 20x, and that of the two extreme pools was 30x. The resulting data were filtered, and clean data were obtained with > 85% of bases scoring Q30 and above, which were aligned to the cabbage reference genome TO1000 (https://plants.ensembl.org/Brassica_oleracea/Info/Index) (Parkin et al. 2014), and the SNP index and Δ(SNP index) were calculated. InDel markers based on 11–192 and 5GH12-279 resequencing data were designed in the candidate region obtained via BSA using Primer 3 to genotype the individuals among the populations. A total of 674 individuals from F₁ and 1454 individuals from the BC₁F₁ population were analyzed via markers with polymorphisms between 11–192 and 5GH12-279 for fine-mapping of the female sterility gene. The primers used in this study are shown in Supplementary Table 1.

Candidate gene analysis

To identify the female sterility gene and determine the nucleotide variations between the wild-type cabbage and mutated female sterile cabbage, the promoter sequences, gene sequences, and cds sequences of all the genes located within the candidate region of 5GH12-170 and 5GH12-279 were analyzed, amplified, and sequenced. Regions up to 2 kb upstream of the transcription start site (TSS) of genes were defined as promoter regions. The reference genome we used was the cabbage reference genome TO1000 (https://plants.ensembl.org/Brassica_oleracea/Info/Index).

Phylogenetic analysis of BoLMI1c

The protein sequence of BoLMI1c was used as a query to search for homologs via BLASTP with an expected threshold of 0.05 within the protein database of the National Center for Biotechnology Information (NCBI) database. The matching proteins (with a query coverage rate > 95% and percent identity > 80%) were selected. Four homologs of LMI1 in A. thaliana, Bo2g005230 (BoLMI1c), Bo3g002560, Bo9g181710 (BoLMI1a), and Bo9g181720 (BoLMI1b), were obtained, and homologous protein sequences were utilized for phylogenetic analysis via the neighbor‒joining (NJ) method with 1000 bootstrap replicates using MEGA 7.0 software (Kumar et al. 2016). Subsequently, aesthetic enhancements were applied through the ITOL website (Letunic and Bork 2021).

qRT‒PCR analysis and RNA-seq

Total RNA was extracted using the TIANGEN RNAprep Pure Plant Kit (Tiangen, Beijing, China) according to the manufacturer’s instructions. First-strand cDNA was synthesized using the HiScript III 1st Strand cDNA Synthesis Kit (Vazyme, Nanjing, China). qRT-PCR of different organs (roots, stems, leaves, sepals and petals, anthers, buds, pistils, and siliques) from the wild-type cabbage Xiwang was performed with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) on a CFX96 Real-Time System (Bio-Rad, USA). Three replicates were performed. Relative expression levels of the genes were calculated via the 2^−ΔΔCt method (Livak and Schmittgen 2001). B. oleracea actin was used as the internal reference gene. The qRT-PCR primers used are shown in Table S1. RNA-seq analysis of buds from 5GH12-170 and 5GH12-279 was performed at Biomarker Technologies Co., Ltd. (Beijing, China). Three replicates of each sample were performed for RNA-seq analysis. Differential expression analysis was performed using DESeq2 (version 1.30.1). Genes with a fold change greater than 2 and an adjusted P value less than 0.05 were defined as the differentially expressed genes (DEGs). RNA-Seq libraries from each sample were generated simultaneously to mitigate any batch effects.

Results

Characterization of the female sterile mutant in cabbage

A female sterile mutant, 5GH12-279, was found in cabbage Ogura cytoplasmic male sterility (CMS) fertility restorer lines derived from distant hybridization between Brassica oleracea and Brassica napus in our previous study (Yu et al. 2016, 2017, 2020; Ren et al. 2020). Compared with the 5GH12-170 (near-isogenic line of 5GH12-279, normal gynoecium), the 5GH12-279 mutant presented a phenotype of female sterility, crumpled petals, which result in the failure of sepals to enclose buds fully, serrated leaves, and a typical floral homeotic defect in the inflorescence (Fig. 1A). Only female sterile individuals have serrated leaves, which implies that this trait could be treated as a phenotypic marker in the seedling screening of female sterile individuals. Gynoecium initiation and outgrowth can be restored at a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C, but this restoration sometimes results in the coexistence of florals without gynoecium and florals with normal gynoecium on one plant, as well as the appearance of outward-turning, cracked and even curved styles (Fig. 1B). Scanning electron microscopy and transverse section analysis revealed that the 5GH12-279 mutant failed to form gynoecium, but the stamens developed normally (Fig. 1C and D). The mature pollen grains of the mutant 5GH12-279 can be stained with Alexander stain solution. The gynoecium of wild-type plants can be successfully pollinated with pollen from the mutant 5GH12-279, which also indicates that the pollens of the mutant 5GH12-279 are viable. These results suggest that the mutated gene is involved in gynoecium initiation.

Fig. 1.

Identification of the female sterile mutant in cabbage. A Morphological comparison of 5GH12-170 (top) and the female sterile cabbage mutant 5GH12-279 (bottom) plants. Bars = 4 mm. B Growth of the gynoecium in 5GH12-279 under a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C. a Flowers without gynoecium and flowers with restored gynoecium are present on one plant. b The style of the pistil occasionally bends. c The style of the pistil occasionally cracked. d Gynecia with cracked and bent styles grow together with the restored normal gynoecium. Bars = 1 cm. C Scanning electron micrographs of pistil and stamen from 5GH12-170 (top) and mutant 5GH12-279 (bottom). D Transverse sections of young pistils and anthers from 5GH12-170 (top) and 5GH12-279 (bottom) plants were stained with 0.25% (w/v) toluidine blue O. Bars = 50 μm

Fine mapping of the female sterility gene

Since 5GH12-279 was derived from the backcrossed population 5GH12 where the female sterile phenotype segregated, we infer that the mutated gene responsible for female sterility in 5GH12-279 is heterozygous. To identify the mutated female sterility gene, the 5GH12-279 mutant (female sterile phenotype, P2) was crossed with 11-192 (normal gynoecium, P1). The F₁ population comprised 674 individuals, with 353 exhibiting a female sterile phenotype and 321 exhibiting a normal gynoecium phenotype (1:1 ratio according to a chi-square test) (Table 1). The BC₁F₁ population contained 1454 individuals, with 747 female sterile individuals and 707 normal individuals, and the segregation ratio was confirmed to be 1:1 via a chi-square test (Table 1). These results indicate that the female sterile phenotype is controlled by a single dominant nuclear gene.

Table 1.

Chi-square (χ²) goodness-of-fit test ratios of female sterile phenotype segregation in the F₁ and BC₁F₁ populations

Populations	Total plant number	Number of stigma-less plants	Number of stigma-normal plants	Expected ratio	χ2	P value
F1	674	353	321	1:1	1.52	0.22
BC1F1	1454	747	707	1:1	1.10	0.29
F2	231	172	59	3:1	0.04	0.85

χ² > χ² _0.05 = 3.84 was considered significant

A P value < 0.05 was considered significant

^*Stigma-less plants and normal stigma plants were evaluated at the flowering stage via visual inspection

BSA-seq was subsequently performed to preliminarily map the mutated female sterility gene. After the raw data were filtered, 101.32 Gb of clean data were obtained with a Q30 ratio ≥ 93.64%. The number of mapped reads was 66,170,859 for the N (female sterility) pool and 81,279,201 for the Y (Normal) pool, accounting for 85.40% and 86.59%, respectively, of the total number of reads (Supplementary Table 2). Overall, 1,886,823 variants, including 1,592,057 SNPs and 294,766 InDels, were obtained. The region with the highest peak (P < 0.01), which contains 1.65 Mb (0-1.65 Mb) on chromosome C02 according to the cabbage reference genome TO1000, was confirmed as the candidate interval associated with the mutated female sterility gene (Fig. 2A). To fine map the mutated female sterility gene, ten InDel markers within the 1.65-Mb candidate region with polymorphisms between the parents (11–192 and 5GH12-279) were developed and used to analyze 674 individuals from the F₁ population and 1454 individuals from the BC₁F₁ population. A linkage map of the ten InDel markers was constructed (Supplementary Table 1 and Fig. 2B). The InDel markers PA07 and PA127 were closely associated with the mutated female sterility gene. Given the location of the InDel markers PA07 and PA127 in the reference genome, the mutated female sterility gene was ultimately mapped to a 108-kb region (C02: 721,128 bp-828,966 bp) (Fig. 2B).

Fig. 2.

Fine mapping of the female sterility gene. A Plot of the Δ(SNP index) values obtained from the two bulk samples. The top red line indicates the threshold line (99% confidence level). The x-axis represents the position of the nine chromosomes, and the y-axis represents the Δ (SNP index) value. The gray lines indicate a p value threshold of 0.05. B Linkage map of the female sterility gene. The InDel markers PA07 and PA127 were closely linked to the female sterility gene

Analysis of candidate genes in the 108-kb candidate region associated with the female sterile phenotype

Nineteen genes were located in this region according to the TO1000 reference genome (Fig. 2B and Table 2). According to the comparative genomic annotation of A. thaliana (TAIR), three genes, Bo2g004170, Bo2g005230 and Bo2g005260, are strongly related to the formation of floral organs (Table 2). Bo2g004170 (BoPTL) is a homolog of the AT5G03680 gene in Arabidopsis and encodes a trihelix TF involved in limiting the lateral growth of floral organs and influencing petal and sepal formation. Bo2g005260 (BoTFL1) is a homolog of the AT5G03840 gene in Arabidopsis that controls inflorescence meristem identity and is involved in the floral initiation process. Bo2g005230 (BoLMI1c) is a homolog of the AT5G03790 (LMI1) gene in Arabidopsis, which encodes an HD-Zip I meristem identity regulator that acts primarily downstream of the LFY gene. The LMI1 gene also plays a role in leaf morphogenesis, which is consistent with the phenotypic characterization of the mutant 5GH12-279.

Table 2.

The 19 putative gene models in the target mapping region

Code	Gene ID	Start (C02)	End (C02)	Homologous gene in A. thaliana	Annotation
1	Bo2g004170	733,819	736,689	AT5G03680	This gene encodes a trihelix transcription factor whose expression is limited to margins of floral and vegetative organs
2	Bo2g004180	763,757	764,259	–
3	Bo2g004190	765,823	766,486	AT4G00416	Protein containing methyl-CpG-binding domain. Has sequence similarity to that of human MBD proteins
4	Bo2g005190	771,305	772,164	–
5	Bo2g005200	773,309	776,305	–
6	Bo2g005210	777,056	777,256	–
7	Bo2g005220	788,179	789,693	AT2G18960	Encodes a plasma membrane proton ATPase
8	Bo2g005230	792,200	794,718	AT5G03790	Encodes a homeodomain leucine zipper class I (HD-Zip I) meristem identity regulator that acts together with LFY to induce CAL expression
9	Bo2g005240	795,336	797,969	AT5G03800	Encodes a protein with a large central domain of 14 internal pentatricopeptide motifs (some degenerate) arranged in tandem. Mutations in this locus result in embryo lethality
10	Bo2g005250	798,460	800,084	AT5G03810	GDSL-motif esterase/acyltransferase/lipase
11	Bo2g005260	802,837	803,991	AT5G03840	Controls inflorescence meristem identity. Involved in the floral initiation process
12	Bo2g005270	807,029	807,552	AT5G03850	Nucleic acid-binding, OB-fold-like protein
13	Bo2g005280	808,471	810,939	AT5G03860	Encodes a protein with malate synthase activity
14	Bo2g005290	811,778	812,836	AT5G03870	Glutaredoxin family protein
15	Bo2g005300	814,996	817,594	AT5G03900	Iron–sulfur cluster biosynthesis family protein
16	Bo2g005310	818,376	819,302	AT5G03905	Iron–sulfur cluster biosynthesis family protein
17	Bo2g005320	819,721	822,615	AT5G03940	Mutant has yellow first leaves; chloroplast signal recognition particle subunit
18	Bo2g005330	823,002	824,507	–
19	Bo2g005340	825,907	827,411	–

The expression levels of all 19 genes in this 108-kb candidate region were determined via qRT‒PCR, and the results revealed significant changes in the expression levels of six genes: Bo2g005230, Bo2g005260, Bo2g005270, Bo2g005280, Bo2g005290, and Bo2g005310 (Fig. 3). Notably, the expression levels of the abovementioned functionally related genes, Bo2g005230 and Bo2g005260, were downregulated by 2.3-fold and 2.6-fold, respectively, in 5GH12-279 (Fig. 3).

Fig. 3.

Expression patterns of 19 candidate genes between 5GH12-170 (Y) and 5GH12-279 (N) determined via qRT-PCR. The ACTIN gene was used as a reference gene. The error bars represent the standard errors of four biological replicates. Three genes, Bo2g004170, Bo2g005230 and Bo2g005260, which are strongly related to the formation of floral organs and are located in the fine-mapping interval, are highlighted in red. Asterisks represent significant differences (ns means p > 0.05, *means p < 0.05, **means p < 0.01, ***means p < 0.001, ****means p < 0.0001)

Combined with the results of the gene functional analysis and gene expression analysis, the above seven genes (Bo2g004170, Bo2g005230, Bo2g005260, Bo2g005270, Bo2g005280, Bo2g005290, and Bo2g005310) were selected for cloning and sequencing. No sequence variations in the coding sequences or promoter sequences of Bo2g004170, Bo2g005260, Bo2g005270, Bo2g005280, Bo2g005290, and Bo2g005310 were detected between the female sterile mutant 5GH12-279 and the near-isogenic line 5GH12-170. However, homoeologous exchange (HE) between Brassica oleracea and Brassica napus was detected within the Bo2g005230 gene body of 5GH12-279 (Fig. 4A). The newly formed BoLMI1c gene sequence in 5GH12-279 is shown in the Supplementary material, and the predicted CDS is shown in Fig. 4B. According to the predicted gene structure, a new CDS of the mutated BoLMI1c gene was successfully amplified (Supplementary material). This HE may strongly reduce the transcription levels of original Bo2g005230 and generate novel fusion transcripts and new protein fusion variants, which results in phenotypes of crumpled petals, serrated leaves and female sterility.

Fig. 4.

Characteristics of BoLMI1c. A Homologous exchange was detected within the BoLMI1c gene body. Alignment of the newly formed BoLMI1c genome sequence with the B. oleracea reference genome (a) and the B. napus reference genome (b). The BLAST hits are indicated by thick lines, and the gaps in the alignment are connected by thin lines. The color indicates the identity, with stronger hits in black and weaker hits in pink. B Original gene structure of Bolmi1c (a) and the predicted gene structure of BoLMI1c (b). CDSf: first (starting with the start codon) coding segment. CDSi: internal (internal exon) coding segment. CDSI: last (ending with stop codon) coding segment. CDSo: this gene contains only one coding exon. PolA: terminal polyA signal. TSS: transcription start site. C Phylogenetic analysis of BoLMI1c and its homologous genes. The phylogenetic tree is primarily divided into three branches, which are indicted by different colored bars. Tree scale = 0.1. D Relative expression levels of BoLMI1c. E Abundance of Bo2g005230 transcripts in the original three exons from the transcriptome data. Y: 5GH12-170. N: 5GH12-279. Each sample was performed in three replications (Y1, Y2, and Y3 and N1, N2, and N3)

On the basis of the gene function annotations, gene expression levels and sequence variant analysis, we speculate that Bo2g005230 is likely to control the female sterile phenotype in cabbage and was renamed Bo2g005230 BoLMI1c. Phylogenetic analysis and expression pattern analysis were subsequently performed. A phylogenetic tree of the BoLMI1c protein with corresponding homologs from other species was constructed to analyze the evolutionary relationships. The results revealed that BoLMI1c was conserved within cruciferous crops and closely related to Brassica cretica-KAF2564855.1 and Brassica napus-CAF1878764.1 (Fig. 4C). To analyze the expression patterns of BoLMI1c, qRT‒PCR was performed on different tissues of the wild-type cabbage Xiwang. BoLMI1c was expressed mainly in anthers and buds, and its expression level was low in roots (Fig. 4D).

Transcriptomic changes in the female sterile mutant

To further investigate the key responsive genes and pathways involved in the female sterile phenotype, we performed transcriptome analysis of buds from 5GH12-279 and 5GH12-170. Principal component analysis (PCA) revealed two separate clusters, indicating obvious differences between 5GH12-279 and 5GH12-170 (Fig. 5A). Transcriptome analysis revealed upregulated expression of 1,191 genes and downregulated expression of 1,375 genes in 5GH12-279 compared with those in 5GH12-170 (Fig. 5B and C). Only two genes, Bo2g005230 (BoLMI1c) and Bo2g005280, in the 108-kb candidate region were significantly differentially expressed (Fig. 5D). Sequence analysis revealed no variations in the promoter, full-length genome sequence or coding regions of Bo2g005280 between 5GH12-279 and 5GH12-170, and this gene was functionally unrelated, which further confirmed that Bo2g005230 (BoLMI1c) was the candidate gene. Moreover, high-throughput sequencing data from the transcriptome were used to measure the presence and abundance of Bo2g005230 transcripts between 5GH12-279 and 5GH12-170. A robust decrease in the Bo2g005230 transcript in the second and third exons was detected (Fig. 4E), which was consistent with our predicted results.

Fig. 5.

Transcriptome analysis between 5GH12-279 (N) and 5GH12-170 (Y). A PCA of all samples. B The number of upregulated and downregulated DEGs. C Volcano plot showing differentially expressed genes. D Gene expression pattern and DEG analysis for 19 genes in the candidate interval. a Heatmap of the 19 genes identified within the candidate gene region. b Differential expression of Bo2g005230 in the transcriptome. c Differential expression of Bo2g005280 in the transcriptome. E The top 20 significantly enriched KEGG terms. F KEGG pathway analysis of differentially expressed genes. G GO enrichment analysis of differentially expressed genes. H The top 20 significantly enriched GO terms

The 2566 DEGs were further subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analyses. Among the top 20 significantly enriched pathways, 278 DEGs were significantly enriched in ‘metabolic pathways’, and 158 DEGs were enriched in ‘biosynthesis of secondary metabolites’ (Fig. 5E). The top five metabolic pathways were ‘global and overview maps’, ‘carbohydrate metabolism’, ‘biosynthesis of other secondary metabolites’, ‘amino acid metabolism’ and ‘lipid metabolism’ (Fig. 5F). GO analysis revealed ‘metabolic process’, ‘cellular process’, ‘biological regulation’, ‘regulation of biological process’ and ‘response to stimulus’ as the top five significantly enriched biological processes (Fig. 5G). Additionally, several gynoecium initiation-related biological processes, such as ‘pectin metabolic process’, ‘galacturonan metabolic process’, ‘pectin catabolic process’, ‘cellular component assembly involved in morphogenesis’, and ‘sporopollenin biosynthetic process’, were significantly enriched (Fig. 5H).

Next, we analyzed all DEGs annotated as putative TFs. According to the PlantTFDB, a total of 4,691 TF genes in the cabbage genome were identified, 243 of which were significantly differentially expressed between 5GH12-279 and 5GH12-170. Among the 243 differentially expressed TF genes, 30 belong to the ERF family, 21 belong to the bHLH family, 21 belong to the NAC family, 21 belong to the WRAY family, and 17 belong to the MYB family. Many TF genes involved in flower organ development were among the top 20 significantly differentially expressed TF genes (Supplementary Table 3). Some TF genes required for proper floral transition are upregulated during flower organ development, such as Bo7g087070 (BoMIF2), which is involved in the regulation of floral meristem termination; Bo8g049940 (BoWRKY28), which is involved in the suppression of megasporocyte cell fate; and Bo6g095400 (BoMYB105), which is involved in meristem initiation and maintenance. In addition to BoLMI1c, the TF genes included Bo8g077160 (BoNPC6), which is involved in gametophyte development; Bo5g044980 (BoREM17a) and Bo5g044970 (BoREM17b), which can promote floral transition; Bo3g023970 (BoREM20), which can affect embryo sac differentiation; and Bo1g008140 (BoMYB52), which is involved in the degree of pectin methyl esterification and the regulation of its activity, were also downregulated. These results indicate that these TF genes may be involved in the regulation of the transcriptional network associated with gynoecium initiation and the female sterile phenotype.

Development of a co-dominant marker of BoLMI1c

Furthermore, the female sterile phenotype of some serrated-leaved cabbage varieties cannot be reliably identified based on leaf morphology alone. To overcome this limitation and facilitate the application of this female sterility gene across diverse varieties, a co-dominant marker N5230, consisting of two pairs of primers, was developed on the basis of the newly formed gene sequence of BoLMI1c and wild type gene Bolmi1c (Table 3 and Supplementary material). The primers N5230-1F/1R amplifed a 2247 bp BoLMI1c-specific band in female sterile individuals (Figure S1). N5230-1F/1R was subsequently used for genotyping the segregating populations, and the results revealed cosegregation between the amplified bands and the female sterile phenotype across all the recombinant plants (144 normal gynoecium individuals (aa) and 144 female sterile individuals (Aa) randomly selected from the F₁ and BC₁F₁ populations). The primers N5230-2F/2R amplifed a 1078 bp Bolmi1c-specific band (Figure S1). Therefore, the homozygous and heterozygous genotypes of BoLMI1c gene can be confirmed by using the above two pairs of primers. A total of 231 individuals of the F₂ population were screened, leading to the identification of 56 individuals harboring the homozygous female sterile locus, 116 individuals harboring the heterozygous female sterile locus and 59 individuals containing the homozygous Bolmi1c locus, conforming to a 1:2:1 Mendelian separation ratio via a chi-square test (Figure S1 and Table 1). These results indicate that the co-dominant marker N5230 can be used for the selection of female sterile cabbages and distinguish the genotypes of female sterility individuals.

Table 3.

Sequence of the marker N5230

Primer name	Primer sequence (5′–3′)	Length of amplified fragment	Locus
N5230-1F	ATGGAATGGTCAATGACAAGCAACG	2247 bp	BoLMI1c-specific
N5230-1R	TTAAAACCAAAAGTTGCACTCTTC
N5230-2F	ATATATATACTGTTCATCCCGCGC	1078 bp	Bolmi1c-specific
N5230-2R	TGTTGCATTGATCAATACCGTAAA

Construction of a new double-sterility hybrid seed production system (cytoplasmic male sterile [CMS] female parent × dominant female sterile [DFS] male parent) for heterosis utilization

To guarantee that the mutant is available for practical applications, we set out to determine whether 5GH12-279 mutants can produce seeds. The results revealed that self-pollination with mason bees produced no seeds in the mutant 5GH12-279 under a normal pollination environment, whereas the near-isogenic line 5GH12-170 produced seeds that were normal. In addition, we successfully obtained self-pollinated progeny at a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C.

Therefore, we propose a double-sterility (DS) system based on the female sterile mutant identified in this study (Fig. 6). The three implementation steps are as follows:.

Fig. 6.

A new double-sterility system (cytoplasmic male sterile [CMS] female parent × dominant female sterile [DFS] male parent) was constructed for heterosis utilization

Step 1: Development of heterozygous female sterile lines. The mutated female sterility gene from 5GH12-279 is introgressed into an elite line through 5-7 generations of constant backcrossing. Phenotypic screening and molecular markers are used to select key individuals exhibiting female sterile phenotype and possessing genetic backgrounds similar to elite lines.

Step 2: Generation and large-scale propagation of homozygous female sterile lines. Heterozygous female-sterile lines obtained in Step 1 produce functional gynoecia under a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C. Under these conditions, These heterozygous female sterile lines are subsequently self-pollinated to produce progeny. Homozygous female sterile lines are identified using the co-dominant marker N5230. For large-scale propagation of homozygous female-sterile lines, self-pollination is facilitated by pollinators in protected fields.

Step 3: Hybrid seed production. A novel model of hybrid seed production in cabbage is established, in which hybrid seeds are obtained by crossing the CMS line (female parent, a different elite line) with the homozygous female sterile line obtained from Step 2 (male parent). Mixed-seeding and mixed-harvesting with fully mechanized management is implemented.

Discussion

Distant hybridization is a powerful source and HE is a crucial mechanism for generating variations (Wang et al. 2024a, b; Yang et al. 2021). In interspecific hybrids, HEs are highly enriched within gene bodies and are likely to cause changes in genome structure and gene expression, as well as physiological and phenotypic consequences (Gaeta et al. 2007; Hurgobin et al. 2018; Lloyd et al. 2018; Zhang et al. 2020a, b). For example, in tobacco, allelic variation in a QTL region potentially due to HE-induced duplication-deletion events was observed to be correlated with partial resistance to P. nicotianae (Shi et al. 2022). The mutant 5GH12-279 was found in the backcross population of the Ogura CMS fertility restorers, which were obtained from distant hybridization between Brassica oleracea and Brassica napus. HE occurs within the BoLMI1c gene body; this variation may strongly reduce the transcription levels of Bolmi1c and result in the formation of new proteins and a new phenotype of female sterility. This study provides a prototypical case of the gene mutation created by HE.

LMI1 has been reported to be involved in many important biological processes in different species, including the regulation of leaf margins, promotion of leaf complexity, formation of leaf shape and leaf lobes, regulation of leaf organ proportions, facilitation of oil gland development and essential oil facilitation biosynthesis, asymmetric bending of lateral petals, elaborate petal development and specialized character formation (Sicard et al. 2014; Vlad et al. 2014; Andres et al. 2017; Vuolo et al. 2018; Chang et al. 2019; Zhang et al. 2020a, b; Kierzkowski et al. 2019; Wang et al. 2021; Wang et al. 2024a, b; Zhang et al. 2024). LMI1 not only plays a role in leaf morphogenesis but also functions as a meristem identity regulator (Saddic et al. 2006; Wang et al. 2021). Cabbage has four homologs of LMI1, namely, Bo2g005230 (BoLMI1c) on chromosome C02 (C02), Bo3g002560 on C03, and Bo9g181710 (BoLMI1a) and Bo9g181720 (BoLMI1b) on C09. Among Brassica species, BoLMI1a (Bo9g181710) in Brassica oleracea and BnA10.LMI1 (BnaA10g26330D) in Brassica napus are both likely involved in the formation of unlobed leaf phenotype (Hu et al. 2018; Zhang et al. 2021). In our study, the 5GH12-279 mutant presented a phenotype of female sterility, crumpled petals, and serrated leaves. The phenotype of the mutant in cabbage was similar to that in Arabidopsis, which further verified the reliability of the candidate gene BoLMI1c. Although we identified the strong candidate gene BoLMI1c, further exploration and analysis of the target interval via transgenic and gene editing approaches would offer more convincing evidence.

However, how the mutated BoLMI1c gene acts in a dominant manner to induce female sterility is still worth discussing. HE usually affects many genes, causing extensive copy number variations, allele substitutions, and subsequent dosage changes (Sun et al. 2024). A 2.833-kb fragment insertion in the first intron of BnFLC.A2 generates a novel loss-of-function allele, Bnflc.a2, which has been introgressed into chromosome C2 via HE. This A02-C02 HE event generated an early-flowering phenotype in rapeseed (Chen et al. 2018). In tobacco, HE causes the substitution of low-expressing SAR8.2 genes with high-expressing homologues, resulting in enhanced resistance to soil-borne pathogens (Shi et al. 2022). In this study, we present evidence that HE contributes to phenotypic variation in 5GH12-279 through map-cloning, gene expression, and RNA-seq analyses. This newly formed CDS probably encodes a functional protein that causes the female sterile phenotype. Transcription factors have functions as repressors or activators (Wang et al. 2010, 2022; Yu et al. 2013). Some TFs even have a dual role in transcription, acting, in a context-dependent manner, both as a repressor and an activator (Ne et al. 2022). In maize, specific TFs cause male sterility by inhibiting anther/pollen development or activating sterility-related proteins (An et al. 2020; Xiao et al. 2020). Thus, the transcription factor BoLMI1c likely disrupts gynoecium development by altering expression of key regulatory genes, ultimately leading to female sterility. LMI1-like genes have recently been proposed as evolutionary hotspots for leaf shape diversity in model plants and are also involved in the regulation of inflorescence meristem indeterminacy. It is likely that the BoLMI1c gene regulates its phenotypic variation through different pathways and mechanisms.

In this study, we found that the 5GH12-279 mutant can grow normal gynoecium and floral organs under proper conditions (at a stable low temperature of 20 °C ± 2 °C or a stable high temperature of 32 °C ± 2 °C, both with 16 h of light and 8 h of night). On the one hand, this feature enables us to achieve self-pollination and obtain seeds of the female sterile male parent that can be used for the new double-sterility hybrid seed production system. On the other hand, since a daily mean temperature of 20–25 °C is considered the optimum temperature for cabbage pollination, female sterile individuals are sometimes susceptible to the external environment, which may lead to the appearance of gynoecia at an inappropriate time. We still lack an accurate understanding of the conditions under which the female sterility‒fertility conversion of our mutant can be achieved. However, Bai et al. (2021) created the thermosensitive male sterility line in rice via a temperature-sensitive mutation in receptor kinase and found that it was feasible to identify thermo-sensitive genic male sterility (TGMS) lines with lower sterility onset temperatures suitable for agronomic applications within the transgenic population. In our study, the sensitivity of individuals to temperature was different. Therefore, extensive creation and screening of insensitive individuals is a feasible way to solve the above problems. Besides, the characterisation of fertility can be changed by only one SNP. A point mutation in ARGONAUTE7 (AGO7) can confers complete female sterility under regular/high temperature and partial female fertility under low temperature in rice (Li et al. 2022). A natural allele of OsMS1 confers thermosensitive recessive male sterility in rice (Wu et al. 2022), whereas a deletion of merely 1 bp in the promoter can turn the Ms-cd1 gene into the dominant male sterility gene in Brassica oleracea (Han et al. 2023). More studies on regulatory elements and molecular mechanisms will be conducted to determine the precise control of female sterility‒fertility conversion and the emergence of gynoecia without being affected by external environmental conditions. In the future, the thermosensitive characteristic of the mutant 5GH12-279 can also be changed by means of transgenic or gene editing.

In hybrid seed production, the removal of male parents is essential. This process currently relies predominantly on manual labor, consuming an average of two person-days per hectare. Owing to the narrow row spacing and interrow distance, the removal of male parents is not amenable to mechanized operation. On the other hand, manual removal significantly elevates the risk of omission, thereby increasing the likelihood of self-pollinated seeds from the male parents being mixed into hybrids. Mechanized management represents a future trend in hybrid seed production. Male sterility is one of the traits most extensively studied in plants because it can be effective applied in heterosis utilization and hybrid breeding (Han et al. 2023). A large increase in crop yield, a decrease in seed production cost and higher seed purity and hybrid rates have been achieved in part by producing F₁ hybrids via a male sterility system (Ren et al. 2020; Yu et al. 2020; Han et al. 2023). However, importantly, mixed-seeding and mixed-harvesting hybrid seed production in mass amounts by machines cannot be achieved with this approach because the male parent possesses functional pistils that can also produce seeds. Therefore, a male parent with normal male fertility but female sterility is considered ideal for mixed-seeding and mixed-harvesting production via mechanized management, which can enable fully mechanized hybrid breeding (Maruyama et al. 1991; Wang et al. 2024a, b). However, the generation, propagation and preservation of female sterile lines are the most restricting factors in current applications (Wang et al. 2024a, b). The female sterile mutant 5GH12-279 and the dominant female sterility gene identified in our study constitute excellent tools for addressing this restriction.

In the double-sterility (DS) system, hybrids are produced using homozygous female sterile lines, so the linked serrated leaf trait does not segregate. The presence of serrated leaves does not impair hybrid performance, such as yield or marketability. Moreover, this trait can serve as a phenotypic marker for hybrid identification. However, its applicability is limited since many cabbage varieties naturally possess serrated leaves. Therefore, the development of a co-dominant molecular marker to distinguish BoLMI1c genotypes provides a more robust strategy. The marker N5230 developed here enables wider utilization of the female sterility gene across diverse cabbage varieties and breeding programs.

To date, the female sterility genes that have been reported and utilized are mainly from rice, and all of the mutations are recessive (Mao et al. 2021; Li et al. 2022; Wang et al. 2024a, b; Huang et al. 2024). This study is the first report of a dominant female sterility gene in plants. Although the environmental conditions for the conversion of the female sterile status of the mutant are complicated and not thoroughly understood, the successful acquisition of selfed progeny confirmed the feasibility of the new DS system.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution statements

YZ conceived and designed the experiments. WR, JS and JL performed the experiments and analyzed the data. WR. wrote the manuscript. YZ and XH revised the manuscript. YW, LY, MZ, HL, YW, and JJ provided valuable suggestions on the manuscript. All the authors have read and approved the manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (32172578), the earmarked fund for the Modern Agro-Industry Technology Research System, China (CARS-23) and the Science and Technology Innovation Program of the Chinese Academy of Agricultural Sciences (CAASASTIP-IVFCAAS).

Data availability statement

All the data generated or analyzed in this study are included in this published article and its supplementary information files. The transcriptome data and BSA data have been deposited in the NCBI Sequence Read Archive (SRA) database under BioProject IDs: PRJNA1193781 and PRJNA1193861.

Declarations

Conflict of interest

The authors declare that they have no competing interests.