Introduction

Flowering is the most significant feature of angiosperms. In this process, plants are transformed from vegetative to reproductive growth. Life cycle alternation between the diploid and haploid phases is complete in flower organs when the gametophyte is fertilized and becomes a sporophyte. In general, flowers are composed of four types of organs: sepals, petals, stamens, and carpels. The morphology of these structures may vary among species, but most carpels consist of a stigma, a style, and an ovary containing ovules (Kuusk et al. 2002). Historically, plant male sterility has been the focus of plant breeders. However, with the development of biotechnology, female sterility has been attracting increasing attention. Female sterility is classified into three main categories: (a) flowers with incomplete or no female organs; (b) flowers with abnormal ovule development or no megasporocytes; and (c) flowers with normal ovules that fail to develop into mature seeds after pollination (Li et al. 2006). The sterile phenotype might be attributed to changes to the sporophyte or gametophyte. Female-sterile mutants with malformed ovaries, ovules, embryo sacs, or megaspore mother cells have been induced using various molecular biology techniques (Sundaresan and Alandetesaez 2010; Yang et al. 2016; Liu et al. 2017; Cao et al. 2018). Female sterility facilitates the investigation of pistil and ovule development and the mechanisms underlying their genetic regulation (Bencivenga et al. 2011; Tedder et al. 2015; Wei et al. 2015; Zhou et al. 2016; Singh et al. 2017).

The pistil is located in the innermost floral whorl. It generates the embryo sac, functional megaspores, and mature seeds. The traditional ABC model of flower development classified several major function class genes that are specifically identified in each whorl. C-class genes alone control the carpels (Coen and Meyerowitz 1991; Pinyopich et al. 2003). High-throughput sequencing and molecular biology have identified many important genes in female-sterile mutants of Arabidopsis and other species (Tedder et al. 2015; Chen et al. 2017; Liu et al. 2017, 2018). Members of the homeotic MADS-box genes have been identified as key regulators controlling the orchestration of female reproductive organs during flower determination and development (Liljegren et al. 2000; Kater et al. 2006; Kaufmann et al. 2009; Hu et al. 2011). AGAMOUS (AG) is a key gene for plant sexual reproduction. AG is spatially restricted to the third and fourth whorls and functions redundantly with APETALA2 (AP2), LEUNIG (LUG), CURLY LEAF (CLF), and STERILE APETALA (SAP). An AG-independent carpel development pathway has been identified (Bowman et al. 1989; Ito et al. 2007). Two redundant genes, SHATTERPROOF1 (SHP1) and SHP2, are closely related to AG and specify valve margin identity (Liljegren et al. 2000). HECATE (HEC) cooperates with SPATULA (SPT) to regulate the female reproductive tract (Gremski et al. 2007). Functional exploration of ETTIN demonstrated that auxin involved in the polarity determinants of gynoecium (Nemhauser et al. 2000; Pagnussat et al. 2009). In tomato, the down-regulation of ARF6 and ARF8 targeted by miR167 leads to defects in floral development and female sterility (Liu et al. 2014). A spontaneous frame-shift mutation (fst) in rice does not produce any activated embryos and endosperms with completely sterile females. Anatomical analysis showed that the abortion occurred in the sporophytic tissues including the integuments, fertilized pre-embryos, and endosperms (Lee et al. 2013). Arabidopsis YABBY and KANADI genes mutation causes amorphous or arrested integument growth (Eshed et al. 2001; McAbee et al. 2006). Examination of complex pistil structures is expected to provide information on their morphological development and gene regulatory networks.

Megasporocytes are derived from the differentiation of hypodermal cells in the ovule primordium. The sporocyte subsequently undergoes meiosis, giving rise to megaspores (Scott et al. 2004). A defect in any of these processes leads to the occurrence of plant sterility. Many important genes that play vital roles in ovule development have been identified. SPOROCYTELESS/NOZZLE (SPL/NZZ) of Arabidopsis is vital for the differentiation of megasporocytes and has complex regulatory mechanisms for sporogenesis (Ito et al. 2004; Wei et al. 2015). The WUSCHEL (WUS) homeobox gene facilitates the coordination of behaviors of different cell populations during ovule development (Groß-Hardt et al. 2002). In Arabidopsis female-sterile mutant, aintegumenta (ant), integuments were not observed and megasporogenesis was blocked at the tetrad stage (Elliott et al. 1996; Klucher et al. 1996). In sterile apetala (sap) mutants of Arabidopsis, megasporogenesis is arrested in ovules during meiotic division, with the tetrad and embryo sacs being entirely absent (Byzova et al. 1999).

The size and shape characteristics of mature plant organs vary among species and determine plant yield and biomass. The growth of plant organs is determined by cell proliferation and cell expansion (Sugimoto-Shirasu and Roberts 2003). These two processes are integrated. Subtle signals regulate the specific spatial and temporal patterns in organs to control their final size. Genetic and molecular mechanisms should be elucidated because they are vital for plant organ development. Certain factors regulating cell proliferation have been identified, including AUXIN-REGULATED GENE INVOLVED IN ORGAN SIZE (ARGOS), AINTEGUMENTA (ANT), CINCINNATA (CIN), GROWTH-REGULATING FACTORS (GRFs), GRF-INTERACTING FACTORS (GIFs), and KLUH/CYP78A5 (Krizek 1999; Hu et al. 2003; Kim et al. 2003; Nath et al. 2003; Kim and Kende 2004; Anastasiou et al. 2007). ROTUNDIFOLIA3 (ROT3), ARGOS-LIKE (ARL), BIGPETAL (BPE), ROOT PHOTOTROPISM2 (RPT2a), SMALL AUXIN UP RNA (SAUR), MEDIATOR25/PHYTOCHROME AND FLOWERING TIME 1 (MED25/PFT1), and KUODA1 (KUA1) (Kim et al. 1998; Hu et al. 2006; Szécsi et al. 2006; Lee et al. 2011; Spartz et al. 2012; Xu and Li 2012; Lu et al. 2014) were identified controlling organ growth by regulating cell expansion. In Arabidopsis, the ubiquitin receptor DA1 was found to regulate organ size by repressing cell proliferation (Li et al. 2008). Further, the mutant of STERILE APETALA (SAP)/SUPPRESSOR OF DA1 (SOD3) was found to suppress the da1-1 phenotype. The biological function of SAP in controlling organ size by regulating cell proliferation was demonstrated by its interaction with KIX8/9 and PPDs (Wang et al. 2016; Li et al. 2018).

Ethyl methanesulfonate (EMS) is widely used to construct artificial mutant libraries because it is easy to handle and induces mutations at a high frequency (Serrat et al. 2014). Numerous EMS-induced mutant populations have been created in Brassica, with many important genes being identified. In oilseed rape EMS-induced mutant populations, FAE1 was identified controlling erucic acid synthesis in the seed, as well as major genes involved in the sinapine biosynthesis pathway (Wang et al. 2008; Harloff et al. 2012). The dyad mutant of Arabidopsis was screened from EMS-mutagenized M2 plants to identify female-sterile mutants (Siddiqi et al. 2000). It was mapped at the lower end of chromosome 5 with PCR-based simple sequence length polymorphism (SSLP) markers and F2 populations. A maize rotten ear (rte) mutant that shows distinct defects in vegetative and reproductive development was identified in a series of EMS-mutagenized populations (Chatterjee et al. 2014). Positional cloning revealed that rte encodes a boron efflux transporter BOR1. Histology and microscopy observations between rte and wild-type plants revealed the structural importance of boron in the formation of reproductive organs. Mutants exhibiting sterility traits identified in EMS-mutagenized populations provide an excellent opportunity to explore the growth and development of plant reproductive organs. EMS mutagenesis usually causes random single-point G/C-to-A/T transitions (Sikora et al. 2011). It supports competent survival rates and forward and reverse genetic analyses.

Chinese cabbage is widely cultivated in East Asia. In the present study, a female-sterile mutant (fsm) was obtained from the Chinese cabbage DH line ‘FT’ in the process of inducing mutations with EMS (Huang et al. 2017). Simple sequence repeats (SSR) and insertion/deletion (Indels) markers were used to build a high-density genetic linkage map. Whole-genome re-sequencing was performed to detect variations in the target mapping region. A single-nucleotide polymorphism (SNP) that prematurely terminates amino acid coding in BraA04001030 was confirmed. Therefore, BraA04001030 was presumed as a candidate gene for fsm. This study facilitates further mutant gene cloning of fsm and contributes toward elucidating the molecular mechanisms underlying female sterility in Chinese cabbage.

Materials and methods

Plant materials

A doubled haploid (DH) line of Chinese cabbage, ‘FT,’ was used as the wild type for isolated microspore culture and ethyl methanesulfonate (EMS) mutagenesis. A female-sterile mutant (fsm) was identified from the populations. Cabbage Sprouts (Brassica campestris ssp. chinensis L.) DH line ‘701,’ a material with distant relationship with Chinese cabbage, was used as the female parent to construct the segregating population.

Stigma surface observations by scanning electron microscopy (SEM)

Buds of ‘FT’ and fsm were randomly selected for self-pollination during flower opening. The pollinated stigmata were collected after 4 h. Samples were fixed in 2% glutaraldehyde at 4 °C overnight and then dehydrated with a graded ethanol dilution series (soak the sample in 50% alcohol solution, 75% alcohol solution, and 90% alcohol solution for about 5 min). Then, the samples were mounted on stubs and sputter-coated with gold. All samples were observed under a scanning electron microscope (Hitachi TM3030; Hitachi Ltd., Chiyoda, Tokyo, Japan).

Fluorescence staining of pollen tubes

Aniline blue staining was used to visualize the pollen tube growing within the pistil as described previously but with modifications (Chen et al. 2007). The day before flowering, stamens were removed from the bud. At the flower-opening stage, the pistils of the ‘FT’ and fsm flowers were self-pollinated. Pistils were collected 4 h after pollination and fixed in FAA solution (50% ethanol, 5% acetic acid, and 3.7% formaldehyde) at 4 °C overnight. They were then rinsed three times with distilled water and softened with 8 M NaOH for 12 h. Then, they were rinsed three times in 0.1 M K3PO4 (pH 8.0) and stained in 0.05% (w/v) aniline blue solution for 12 h in the dark. The pistils were observed and photographed under a fluorescence microscope (Axio Observer A1; Carl Zeiss AG, Oberkochen, Germany).

Construction of the mapping population

Classic map-based cloning methods were used to locate mutant genes. ‘701’ was the designated female parent, and fsm was the male parent to construct the segregating population. The F1 plants were self-pollinated, and F2 plants with the mutant phenotype were used for linkage analysis and gene mapping. Plants were selected by visual examination for female sterility or fertility at flowering. The parental lines and the F2 mapping population were grown in 2015 at the greenhouse of Shenyang Agriculture University, Liaoning Sheng, China.

DNA extraction, marker development, and PCR amplification

For the parents and F2, total genomic DNA was extracted from fresh leaves using a modified version of the cetyltrimethylammonium bromide (CTAB) method and stored at − 30 °C. Polymerase chain reactions (PCR) were performed using a final reaction mixture volume of 10 μL. The PCR products were separated on 5% denaturing polyacrylamide gels at 220 V for 1.5 h and then subjected to silver staining.

A total of 749 SSR markers were used to identify polymorphisms between the two parents. They were obtained from the Brassica rapa Genome Sequencing Project (BrGSP) and were well distributed on ten linkage groups of the Brassica A genome. It was confirmed that the markers were closely linked to the fsm gene. Thus, new markers were designed in Primer Premier V. 5.0 (PREMIER Biosoft International, Palo Alto, CA, USA) based on the reference genome sequence in the BRAD database (http://brassicadb.org/brad/). Newly developed SSR and Indel primers were synthesized by GENEWIZ, Inc., Tianjin, China.

Linkage analysis and genetic map construction

Tightly linked markers were first identified by screening 300 F2 mapping populations. Larger F2 populations and increasingly closely linked markers distributed on both sides of the mutant gene were subsequently obtained. A genetic linkage map between the molecular markers and the mutant gene was constructed according to the recombination frequencies of the polymorphic markers with MapMaker V. 3.0 (Whitehead Institute, Cambridge, MA, USA) (Lander et al. 1987). Map distances (centimorgans; cM) were calculated with the mapping function of Kosambi (1943).

Variation detection by whole-genome re-sequencing

Total genomic DNA was extracted from ‘FT’ and fsm as described in the previous section. The purity and quality were validated with Qubit and NanoDrop ND-1000 (NanoDrop Technologies LLC, Wilmington, DE, USA). Next, DNA-seq libraries were generated with a TruSeq Library Construction Kit (Illumina, Inc., San Diego, CA, USA). DNA samples were broken into 350-bp fragments by Covaris sonication. The fragments were blunt end-repaired, A-tailed, and ligated with full-length adapters for PCR amplification. The PCR products were purified (Agencourt AMPure XP; Beckman Coulter Inc., Brea, CA, USA) and sequenced on an Illumina HiSeq 2500 (Illumina, Inc., San Diego, CA, USA). The Brassica rapa reference genome was downloaded from the BRAD database. After quality control, clean data from two samples were aligned to the reference genome with Burrows-Wheeler Aligner (BWA) software (parameter: mem –t 4 –t 32 -M). Repetitive sequences were removed using SAMTOOLS (parameter: rmdup). SNP detection and annotation were performed with SAMTOOLS (parameter: mpileup –m 2 –F 0.002 –d 1000) and ANNOVAR, respectively.

Candidate gene prediction

To screen for candidate fsm, all annotations in the target region were identified in the Brassica database. Primers were designed to amplify the candidate genes according to the sequence information available in the Brassica database. The PCR amplification products were purified, cloned into the pGEM®-T Easy Vector (Promega, Madison, WI, USA), and transformed into the TOP10 competent cells (CWBIO, Beijing, China). Xgal-IPTG was added to the medium, and positive clones were selected by the blue-white spot. Recombinant plasmids were sequenced by GENEWIZ, Inc. (Tianjin, China). Multiple sequence alignments were performed in DNAMAN V. 6.0 (Lynnon LLC, San Ramon, CA, USA).

Quantitative real-time PCR (qRT-PCR) validation

Total RNA was extracted from buds and pistil of ‘FT’ and fsm using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). The cDNA was synthesized using oligo (dT) primers and a FastQuant First-Strand cDNA Synthesis Kit (Tiangen, Beijing, China). The primer pair for the candidate gene BraA04001030 (F: 5′-ACGCTACTGGGTCGGACTTTT-3′; R: 5′-ATAACAGCTTCTGGATCGGTGAG-3′) was used for qRT-PCR. Actin was used as an internal control (Huang et al. 2017). The reaction system was carried out with Real PreMix Plus (SYBR Green) on a Bio-Rad IQ5 real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA). The final reaction mixture volume was 20 μL, and the following program was run: pre-incubation at 95 °C for 5 min; 45 cycles of 95 °C for 30 s; 60 °C for 30 s; 72 °C for 15 s; and a final extension step of 72 °C for 30 min. Each cDNA sample was prepared in three biological replicates. The relative expression was calculated using the 2−ΔΔCt method (Livak and Schmittgen 2001) and analyzed in Bio-Rad IQ5 Manager (Bio-Rad Laboratories, Hercules, CA, USA).

Results

Characterization of the fsm mutant

The morphological traits of the floral organs in fsm exhibited clear variation in the sepals, petals, stamens, and ovary (Fig. 1a–g). The ovaries of fsm were short and thin (Fig. 1g), and the seed pods failed to expand, even after artificial pollination. When crossed as a female, fsm was completely abortive, but it was fertile as a pollen donor. The organ size of leaves and flowers was noticeably smaller. The leaves of fsm were comparatively smaller than those of ‘FT’ at the seedling stage, and also the leafy heads (Fig. 1h, i). In a previous study, we demonstrated that the female abortion of fsm was caused by the abnormal development of the ovules (Huang et al. 2017). Genetic investigation showed that fsm was controlled by a single recessive nuclear gene.

Fig. 1
figure 1

Floral organ morphology of ‘FT’ and fsm. a Inflorescences of wild-type ‘FT’ (left) and mutant fsm (right); b single flower of ‘FT.’ c single flower of fsm. d sepals of ‘FT’ (left) and fsm (right); e petals of ‘FT’ (left) and fsm (right); f stamens of ‘FT’ (left) and fsm (right); g pistils of ‘FT’ (left) and fsm (right); h ‘FT’ (left) and fsm (right) at the seedling stage; and i leafy heads of ‘FT’ (left) and fsm (right). Scale bar: 1 mm

Full size image

Stigma observations

Female sterility of plants is mainly attributed to imperfect ovary structure or the dysfunction of embryo sac and female gametes. During the flowering day, the ovaries of fsm were wizened and failed to elongate, even under artificial pollination. To investigate whether female sterility occurs during pollen–pistil interaction, we evaluated the pollen tube growth and papillar cell viability.

At the flower-opening stage, the papillae of ‘FT’ resembled normal rounded bilobed domes (Fig. 2a). The carpel apex was incompletely healed and left a groove or notch in the middle of the fsm stigma. The area covered by stigma mastoid cells was flat in fsm (Fig. 2b). The papillar cells of fsm were abnormally elongated. The papillar cells were oblate and large compared to those of ‘FT’ (Fig. 2c, d). Four hours after pollination, the pollen grains adhered to the stigmata of both ‘FT’ and fsm. They then hydrated and germinated on the papillar cells. Pollen tubes extended and penetrated the papillar cells (Fig. 2e, f). However, there were relatively fewer pollen grains adhering to the papillae in fsm. Although there were some morphological changes in the stigma cells, the function of the hydrated pollen grains was normal.

Fig. 2
figure 2

Morphological observations of papillae under a scanning electron microscope (SEM) and aniline blue staining of pollen tube germination. a Papillae of ‘FT’; b papillae of fsm; c phenotype of unpollinated papillar cell of ‘FT’; d unpollinated papillar cell of fsm; e at 4 h after pollination, pollen tubes (Pt) emerged and penetrated ‘FT’ papillar cells; f At 4 h after pollination, pollen tubes (Pt) emerged and penetrate fsm papillae cells; g and staining of pollen tubes in the style of ‘FT.’ h Staining of pollen tubes in the style of fsm. i Pollen tubes penetrated embryo sacs in ‘FT.’ j Pollen tubes penetrated the embryo sacs in fsm. PT pollen tubes. Red arrows indicate pollen tubes. Scale bar: 20 μm

Full size image

Pollen tube development

From the results of the pollen tube aniline blue staining of the pistil, after pollination, the pollen tube of the style in female-sterile mutant fsm was normal, and the sperm cells could penetrate the stigma and style (Fig. 2g, i). A more detailed observation revealed that the pollen tube also developed normally in the stalk portion of the ovule (Fig. 2j).

Preliminary mapping of the fsm gene

A map-based cloning strategy was adopted to isolate the mutant gene of fsm. A total of 176 pairs of SSR markers were uniformly distributed among ten linkage groups of the Brassica A genome and were used to detect polymorphisms between fsm and ‘701.’ Of these, 67 markers exhibited polymorphism. Thirty homozygous recessive F2 individuals were selected to determine whether they were linked with the mutant loci. SSR marker cnu-m439a on chromosome A04 was linked to the target gene. It was used for screening in other homozygous recessive F2 individuals for linkage analysis and gene mapping. The linkage analysis indicated that cnu-m439a was linked to fsm at a genetic distance of 1.16 cM (Fig. 3a).

Fig. 3
figure 3

Genetic and physical maps of the fsm gene and candidate gene analysis. a Linkage map of chromosome 4 constructed using 3108 sterile F2 plants. The fsm gene was mapped to the region between SSR-R82 and SSR-g17. Numbers are genetic distances between adjacent markers; b fine mapping of the fsm gene. The fsm gene was restricted to a region between Indel-I2 and Indel-I7. The number of recombinants between the markers and fsm is indicated under the linkage map. Numbers are genetic distances in centimorgans (cM) between adjacent markers; c candidate region of the fsm locus annotated in the Brassica database. The fsm gene was narrowed down to a 1376-kb region containing 107 predicted genes. Detailed information on these genes is presented in Table S2

Full size image

Markers near cnu-m439a on chromosome A04 were developed for polymorphism detection and linkage analysis. The polymorphic molecular marker SSR-R39 was identified on the other side of the target locus. Its recombinants differed from cnu-m439a. The new markers SSR-g37, SSR-g17, SSR-R82, and SSR-g7 were located between cnu-m439a and SSR-R39 and were linked to the mutant loci (Fig. 3b, Table S1). Statistical analyses of the recombinants of these markers were conducted for linkage analysis and gene mapping. The results indicated that SSR-R39, SSR-g37, and SSR-g17 were located on the opposite side of fsm relative to SSR-R82, SSR-g7, and cnu-m439a. Recombination frequency data demonstrated that SSR-R82 and SSR-g17 were tightly linked to the mutant locus at genetic distances of 0.08 cM and 0.11 cM, respectively (Fig. 3a).

Fine mapping of fsm gene

New SSR and Indel markers were developed in the interval between SSR-R82 and SSR-g17 to map the mutant gene of fsm more precisely. SSR-g13, Indel-I2, and Indel-I7 exhibited polymorphisms and were tightly linked to fsm (Table S1). According to the number of recombinants, the target locus was between Indel-I2 and Indel-I7 at genetic distances of 0.05 cM and 0.06 cM, respectively (Fig. 3b). Therefore, the mutant gene was located in a 0.11-cM region on the A04 linkage group (Fig. 3c). When the genetic map was aligned to the B. rapa genome, it was found that the physical distance between Indel-I2 and Indel-I7 was 1376.69 kb and there were 107 genes in the target region (Fig. 3c, Table S2).

In traditional map-based cloning, molecular marker linkage analysis is based on the recombination rate between the marker and the target locus. Further increases of the mapped population failed to reveal more linkage markers to narrow the positioning interval. Next, whole-genome re-sequencing was conducted to detect variations between ‘FT’ and fsm. After filtering, sequencing quality was established. It was found that sample data volume was adequate, GC distribution was normal, and the database was successfully sequenced. Exonic, homozygous, and non-synonymously encoded mutation sites in the fsm were filtered out of the mutation sites detected by genomic re-sequencing.

Candidate gene prediction

In the final mapping region between Indel-I2 and Indel-I7, 12 possible SNPs exist in fsm and were distributed on six genes (Table S3). Corresponding primers were designed to verify the sequence based on the locations of the 12 SNP sites. Sequencing alignment results confirmed that a single-nucleotide C-to-A (TCG/TAG) transition occurred in BraA04001030, causing the premature occurrence of a termination codon (Fig. 4b). Four pairs of primers were designed along the full length of BraA04001030 to sequence the gene in fsm and ‘FT.’ Alignment analysis showed that only one single-nucleotide polymorphism (SNP) was found in fsm. It was consistent with the re-sequencing results. Genotype analysis of seven recombinants between the two most closely linked markers indicated that the female-sterile phenotype cosegregated with this SNP (Fig. S1). BraA04001030 encodes a homologue of the STERILE APETALA (SAP) transcriptional regulator. In Arabidopsis thaliana, SAP regulates flower and ovule development (Byzova et al. 1999). Therefore, we selected BraA04001030 as the most likely candidate gene for fsm. The genomic sequence of BraA04001030 is 1131 bp in length and has only one exon, which encodes a protein of 376 amino acid residues. The SNP (C-to-A) was located on the of 599th base and caused the termination of amino acid coding at the 200th residue (S → U) in BraA04001030 of the fsm mutant (Fig. 4b). The amino acid sequence alignment showed a high similarity (75.00%) with AtSAP (AT5G35770). Proteins that share significant homology with SAP are found in Brassica and other species. A phylogenetic tree was constructed showing the kinship between these species (Fig. S2).

Fig. 4
figure 4

Gene structure and alignment analysis of encoded amino acid sequences between ‘FT’ and fsm. a Gene structure of BraA04001030 and the site of SNP. The arrows indicate the primers of qRT-PCR; b alignment of amino acid sequences between ‘FT’ and fsm. Red rectangle: position where encoded amino acid was prematurely terminated

Full size image

Analysis of gene expression

The expression levels of BraA04001030 in fsm and ‘FT’ were examined by qRT-PCR. Gene expression levels in the inflorescence and pistils of fsm resembled those of the wild type (Fig. 5). The single-nucleotide polymorphism (SNP) that occurs in BraA04001030 did not cause differences in expression between ‘FT’ and fsm.

Fig. 5
figure 5

Expression analysis of BraA04001030 in ‘FT’ and fsm by qRT-PCR. a Pistil; b inflorescence

Full size image

Discussion

The female organs of plants are highly complex. The use of female-sterile mutants could provide more insights into the mechanisms regulating flower development (Tedder et al. 2015). The female-sterile mutant (fsm) of Chinese cabbage is ideal for studying ovary and ovule development. In the current study, we showed that the growth of stigma papillar cells and pollen tubes was normal in the mutant fsm, based on pistil aniline blue staining and stigma scanning. Combined with previous observations of morphological characteristics in fsm ovaries (Huang et al. 2017), we consider that the pistil pollination process in fsm was not affected, but that sterility may due to the production of megasporocytes or the mitosis of functional megaspores. We performed genetic mapping of the fsm gene using a large F2 population and a number of closely linked markers. Furthermore, genomic re-sequencing was conducted to detect possible variations in the target region. As a result, BraA04001030 was predicted as the candidate gene for fsm. The female-sterile mutant of Chinese cabbage could be used as a basis for studies exploring the development process and molecular regulation mechanisms of pistils and ovules in Chinese cabbage.

Identification of the variant site of the mutant could be used to elucidate the occurrence of the mutant phenotype at the molecular level. Genetic mapping demonstrated that the mutant gene of fsm was anchored between the two most tightly linked markers on chromosome A04, namely Indel-I2 and Indel-I7. There was still a large area in the positioning interval, and numerous genes remained. No other effective polymorphic markers were found in this interval, and the candidate gene of fsm could not be more accurately characterized. When the location information was aligned to the B. rapa genome, it was found that the target region was in the centromeric region of chromosome A04 (Cheng et al. 2013). The low recombination rate and high frequency of repetitive sequences were associated with this region (Cheng et al. 2002). The heterochromatin structure and domains near the centromeres are highly complex. Consequently, further candidate gene identification was hindered in this region. Therefore, it remains a substantial challenge to clone, sequence, and assemble the genomic components in this region. In this study, whole-genome re-sequencing between ‘FT’ and fsm was conducted to detect variation in the final positioning interval. PCR amplification and Sanger sequencing were used for SNP validation since they are the gold standards for this purpose (Kang et al. 2015). A single-nucleotide polymorphism (SNP) was identified inside BraA04001030. This gene is orthologous to Arabidopsis thaliana STERILE APETALA (AtSAP) and plays vital roles in the regulation of flower and ovule development (Byzova et al. 1999). The SNP in BraA04001030 causes premature stop coding, which may alter SAP-encoded protein activity and significantly influence vital traits. The relative expression of SAP in ‘FT’ and fsm was equivalent. This observation was consistent with SAP expression levels determined for Arabidopsis sod3-1 (Wang et al. 2016). Single-nucleotide variation in a given gene might not necessarily alter gene expression. The abnormal function of SAP in fsm was manifested mainly in amino acid coding.

The reduced-fertility mutant sap of Arabidopsis was identified in a transposable element system, and its phenotype resembled that of ap2 (Byzova et al. 1999). The ovules of ap2-1 appeared normal. The sap single mutant showed defects in megasporogenesis. The genetic interaction between SAP and AP2 was analyzed. There was no embryo sac in the sap/ap2-1 mutant, and megaspores within the dyad or tetrad were not observed. Analysis of the pattern of expression of sap ap2 double mutants and sap or ap2 single mutants revealed that both genes function synergistically in the inflorescence meristem but were not coregulated. Megasporogenesis abortion occurs earlier in sap ap2 when compared to single mutant sap. Therefore, both genes complement each other during the initiation of megasporogenesis. In our previous and current studies, the infertility of fsm was confirmed by the abortion of the ovule, which has the same phenotype as the SAP mutant in Arabidopsis. AGAMOUS (AG) was restricted by APETALA2 (AP2) to the inner two floral whorls (Drews et al. 1991). In Arabidopsis thaliana, the sap ag-1 double mutant exhibit unique floral architecture and axillary flower bud features. The phenotype of the double mutant indicates that SAP genetically interacts with AG to maintain the identity of the floral meristem. In Chinese cabbage, fsm also exhibits multiple axillary flower buds and similar floral architecture with sap. It remains to determine whether SAP genetically interacts with AG and AP2 during the flower development of Chinese cabbage.

In addition, the fsm mutant of Chinese cabbage exhibited variation in organ size. The leaves, inflorescence, and flower organs of fsm were smaller than those of the wild-type ‘FT.’ The phenotype was consistent with that of the sap mutant in Arabidopsis (Wang et al. 2016). It was confirmed that SAP promotes meristemoid cell proliferation in Arabidopsis. In fsm, the organ size variation may be limited by cell proliferation. In Arabidopsis, the ubiquitin receptor DA1 acts synergistically with the E3 ubiquitin ligases DA2 and ENHANCER OF DA1 (EOD1)/BIG BROTHER to regulate cell proliferation and control organ growth (Xia et al. 2013). The ubiquitin pathway regulates plant organ size (Du et al. 2014). Protein sequencing shows that SAP participates as an F-box protein in the SKP1/Cullin/F-box (SCF) E3 ubiquitin ligase complex (Wang et al. 2016). SCFs repress signaling in several phytohormone pathways that regulate plant growth and development (Gray et al. 2001). Recently, it was reported that SAP targets PEAPODs (PPDs) and KIX8/9 proteins for degradation for controlling organ size (Wang et al. 2016; Li et al. 2018). Chinese cabbage is an important leafy crop. Exploring the molecular mechanism underlying plant organ size is expected to provide an important resource for breeding. Identification of mutant genes might help explain mutant phenotypes at the molecular level.

The accessibility of information on the Chinese cabbage genome has greatly facilitated map-based cloning. This study is the first to identify SAP as the candidate gene for female sterility in Chinese cabbage. Detailed mapping of the genetic interaction networks will help elucidate the ontogeny. Further exploration is required to characterize the molecular mechanisms underlying organ size and female reproductive development in Chinese cabbage.

Author contribution statement

HF, WJL, SNH, and ZYL conceived and designed the study. WJL and TXL, CT, YHW performed the experiments. WJL wrote the paper. WJL, SNH, and HF reviewed and edited the manuscript. All authors read and approved the manuscript.