Shungshung Xin,Hongli Dong,Yixin Cui,Yilin Liu,Guifu Tin,Nnxi Deng,Hufng Wn,Zhi Liu,Xiorong Li,Wei Qin,b,*

a College of Agronomy and Biotechnology,Southwest University,Chongqing 400715,China

b Engineering Research Center of South Upland Agriculture,Ministry of Education,Chongqing 400715,China

c College of Pharmaceutical Sciences and Chinese Medicine,Southwest University,Chongqing 400715,China

Keywords:Adenine phosphoribosyltransferase 5 Brassica napus QTL mapping RNA-seq Seed number per silique

ABSTRACT Seed number per silique(SNPS)is one of seed yield components in rapeseed,but its genetic mechanism remains elusive.Here a double haploid(DH)population derived from a hybrid between female 6Q006 with 35–40 SNPS and male 6W26 with 10–15 SNPS was investigated for SNPS in the year 2017,2018,2019 and 2021,and genotyped with Brassica 60K Illumina Infinium SNP array.An overlapping major QTL(qSNPS.C09)explaining 51.50%of phenotypic variance on average was narrowed to a 0.90 Mb region from 44.87 Mb to 45.77 Mb on chromosome C09 by BSA-seq.Subsequently,two DEGs in this interval were detected between extreme individuals in DH and F2 populations by transcriptome sequencing at 7 and 14 days after pollination siliques.Of which,BnaC09g45400D encoded an adenine phosphoribosyltransferase 5(APT5)has a 48-bp InDel variation in the promoter of two parents.Candidate gene association analysis showed that this InDel variation was associated with SNPS in a nature population of rapeseed,where 54 accessions carrying the same haplotype as parent 6Q006 had higher SNPS than 103 accessions carrying the same haplotype as parent 6W26.Collectively,the findings are helpful for rapeseed molecular breeding of SNPS,and provide new insight into the genetic and molecular mechanism of SNPS in rapeseed.

1.Introduction

Food shortage is one of the most serious global problems in this century.The world’s population is expected to increase from 7.7 billion at present to 9.7 billion in 2050,before reaching a peak of nearly 11 billion by the end of the century(https://www.un.org/en/un75/shifting-demographics).Meanwhile,agricultural land has been lost due to urbanization and other consequences of unsustainable land management in recent decades.The United Nations Food and Agricultural Organization(FAO)estimates that moderate or severe food insecurity affects＞30%of the world population in 2020(https://www.fao.org/state-of-food-security-nutrition/en/).

The intake of edible oil has increased consistently in dietary over the past 15 years[1].Rapeseed(Brassica napus L.)consisting of high-quality fatty acid components and antioxidants,is the second largest edible oil crop after soybean in the world.Rapeseed is annually grown on a total of 34 million hectares of land in China,Europe,Canada,and Australia,producing 75 million tons seeds(https://www.fao.org/faostat/en/#data/QCL/visualize).Therefore,developing high-yielding varieties has always been one of the most important goals of rapeseed breeding to compensate for alternate competing demands,including rapid population growth,reduction of available land,and consequent changes in consumption patterns.

Many agronomically important traits,including yield,are governed by a few genetic loci known as quantitative trait loci(QTL).Seed number per silique(SNPS)is one component of seed yield per plant together with seed weight and silique number per plant in rapeseed,and positively correlates with yield[2].More than 100 QTL for SNPS have been reported in rapeseed[3],but only a few quantitative trait genes(QTG)have been identified for SNPS[4,5].The corresponding molecular mechanism underlying SNPS remains largely unknown in rapeseed.

SNPS is determined by many factors during ovule and seed development,including the number of ovules per ovary during ovule initiation,the ratio of fertile ovules,the ratio of successfully fertilized ovules,and the proportion of fertilized ovules that developed into seeds[4].Number of genes have been identified to control SNPS in plants[6,7].For example,mutations in multiple pollen-expressed Leu-rich repeat extensin(lrx)genes cause severe defects in pollen germination and pollen tube growth,resulting in a reduced seed set[8].Loss-of-function of Baili Xi(BLX)causes an increased level of reactive oxygen species in mitochondria and cytosol,and results in lethal embryo[9].Mutants in the imprinted PICKLE RELATED 2 gene(PKR2)suppress seed abortion of fertilization independent seed class mutants and paternal excess interploidy crosses in Arabidopsis[10].

Next-generation sequencing technologies have enabled sequencing of large numbers of plants at low cost,providing opportunities to analyze the genetic architecture of complex traits.By combining QTL mapping and dynamic RNA-seq,here we identified a candidate gene BnaC09.APT5,which exhibited differential expression pattern between high and low SNPS individuals at 14 days after pollination in rapeseed.The function of this gene was confirmed in rapeseed by candidate gene association analysis.This study not only provides a favorable gene of SNPS for breeding high-yield rapeseed,but also suggests an alternative strategy to discover causal QTG by integrating genetic and RNA-seq data.

2.Materials and methods

2.1.Field experiment and phenotype investigation

A double haploid(DH)population consisting of 91 individual lines was developed via microspore culture from a hybrid between female 6Q006 with 35–40 seeds per silique and male 6W26 with 10–15 seeds per silique.The DH lines and parents were planted in a randomized complete block design with three replications in the field at Southwest University,Chongqing(29.9°N,106.4°E),China,in four years(2017–2019 and 2021).The plot was composed of three rows,including 30 individuals,with plant density of ten plants each row with length of 2 m spacing 0.3 m.The field management followed the standard agriculture practice.

Seed number per silique was determined by the average performance of five plants in the middle of plot,where 10 welldeveloped siliques were detached in the main inflorescence of each plant to investigate seed number at maturity.

2.2.Data analysis

Analysis of variance(ANOVA)and Pearson’s correlation coefficient analysis were performed by using GLM and CORR program in SAS V9.3(SAS Institute Inc.,Cary,NC,USA),respectively.The broad-sense heritability was calculated as h2=σ2G/(σ2G+σ2GE/n+σ2e/nr),where σ2G,σ2GE,σ2eare the variations of genotypic,the interaction of the genotype by environment and error,respectively.n and r are the numbers of environments and repetitions,respectively[11].

2.3.Genetic linkage group construction and QTL mapping

The genomic DNA of DH lines and two parents were isolated from fresh leaves and genotyped using Brassica 60K Illumina Infinium SNP array[12].After filtering those SNPs with no polymorphism,missing rate＞25% and minimum allelic frequency(MAF)＜5%,the remaining SNPs were used for linkage map construction using JoinMap 4.0 software.QTL mapping was performed using the composite interval mapping(CIM)model in WinQTL Cartographer 2.5 software at a threshold LOD=3.0 which determined by performing 1000 permutation tests at a significant level of P＜0.05.

To narrow down the QTL interval,two DH groups,‘H-DH’group consisting of six DH lines of high SNPS and‘L-DH’group consisting of six DH lines of low SNPS in DH population,were constructed.The DNA of each line from same group was bulked equally and employed for re-sequencing together with two parents according to the standard protocol provided by Illumina.The clean reads obtained from both parents and two DNA bulks were aligned against the rapeseed reference genome‘‘Bra_napus_v2.0”(ZS11,BioProject:PRJNA237736)using BWA software[13].The SNP and InDel sites were detected by GATK software package[14].

Heterozygous and non-polymorphic sites between two parents,and the sites with read depth＜10 between two groups were filtered out.SNP/InDel-index at the remaining SNP and InDel loci in each group was calculated with the proportion of reads harboring the variant that is different from the reference sequence according to the method of Takagi[15].A sliding window analysis with a 1-Mb window size and 10-kb step size was employed to plot Δ(SNP/InDel-index),which was the subtraction of SNP/InDel-index between‘L-DH’and‘H-DH’.The significance threshold of the test statistic for a QTL was calculated from 1000 random resamplings based on a 1% experiment-wise error.

2.4.Transcriptome analysis and candidate gene mining

Considering obvious occurrence of seed abortion at 7–14 days after pollination(DAP)in parent 6W26,but not in parent 6Q006,we screened transcriptome profiles at 7 and 14 DAP siliques in two sets of bulk pools.Except for the set of‘H-DH’and‘L-DH’groups in DH population,the other set is composed of‘H-F2’bulked from 8 individuals with high SNPS and‘L-F2’bulked from 6 individuals with low SNPS in F2population.The eight RNA-seq libraries of‘H-DH’,‘L-DH’,‘H-F2’and‘L-F2’at 7 and 14 DAP siliques were sequenced on the Illumina HiSeq 4000 platform at BioMarker Technologies Co.,Ltd.(Beijing,China).Clean reads were mapped to rapeseed reference genome‘‘Bra_napus_v2.0”(ZS11,BioProject:PRJNA237736)using HISAT2 software[16].Differentially expressed genes(DEGs)were detected using EBseq software package[17],filtered with the following requirements:false discovery rate(FDR,Benjamini-Hochberg multiple test correction)＜0.001 and absolute fold change＞2.Gene ontology(GO)enrichment was performed using Metascape[18],to identify significantly enriched terms among DEGs(P＜0.01),containing at least three genes in each term.

2.5.Reverse transcription PCR and real-time quantitative PCR

Total RNA was extracted from siliques using SteadyPure Plant RNA Extraction Kit(Accurate Biotechnology Co.,Ltd.,Changsha,Hunan)and cDNA was synthesized using 5×All-in-One RT Master-Mix(abm).qRT-PCR was performed with the BlasTaq 2×qPCR MasterMix(abm)in a CFX96TM Real-Time qPCR Detection System(Bio-Rad Laboratories,Hercules,CA,USA)according to the manufacturer’s instructions.The primers were designed using Primer 5 software and listed in Table S6.BnActin7 gene was used as the internal control for relative expression gene patterns analysis of rapeseed.The reactions were performed from three biological replicates(two technical replicates per biological sample).Gene expression level was calculated using the 2-ΔΔCTmethod.

2.6.Candidate gene association analysis

A natural population composing of 157 rapeseed accessions was investigated for SNPS across four years(2015,2016,2018,2019)in a previous study[3].To test association of candidate gene adenine phosphoribosyltransferase 5(APT5)with SNPS in rapeseed,those accessions were genotyped using a InDel marker in the candidate gene(Forward primer:5′-GGTTCATAAATTCTGTGTGCCT-3′,Reverse primer:5′-GATAATTCAAAGTGGAGAAGGGT-3′)and haplotype effects were analyzed with a two-sample t-test.

3.Results

3.1.Morphological and genetic characterizations of seed number per silique

An obvious difference in seed number per silique(SNPS)was found between two parents,6Q006 with around 35–40 seeds per silique but 6W26 with around 10–15 seeds per silique at maturity(Fig.1a).The observation of silique dissection during seed development showed that parent 6W26 sharply decreased by 40%for SNPS from 7 days after pollination(DAP)to 14 DAP,unlike parent 6Q006 with little alteration of SNPS(Fig.1b).The SNPS of the reciprocal hybrids between 6Q006 and 6W26 were in the middle of those of two parents with 21.06±1.76 and 22.72±1.45 seeds per silique,and displayed no difference between reciprocal F1lines,indicating that SNPS is partially dominant(Fig.1a).

To investigate the genetic mechanism underlying SNPS,a DH population developed by microspore culture of the F1between 6Q006 as female and 6W26 as male,was investigated for SNPS at maturity together with two parents and F1across four years(2017–2019,2021).An obvious bimodal distribution for SNPS in DH population was detected with average of 18.34,19.41,17.97 and 21.06 in the year 2017,2018,2019 and 2021,respectively(Fig.1c–f).The results of Chi-Squared test showed that it conformed to a 1:1 separation ratio(χ2=1.89,0.17,1.53,2.29 for the year 2017–2019 and 2021,respectively)in DH population,indicating that SNPS is controlled by a main locus.

Analysis of variance(ANOVA)for SNPS in DH population showed significant differences among genotypes,environments,and the interactions between genotype and environment(P＜0.0001),but no significant difference among repetitions(P=0.063)(Table 1).A moderately high broad-sense heritability(89.01%)was detected for SNPS in the DH population across 4 years(Table 1).This was in accordance with the significant correlation of SNPS between years(r=0.78 on average).

3.2.High-density SNP map construction and QTL mapping

Ninety-one DH lines were genotyped by the Brassica 60K SNP array together with two parents.After filtering the SNPs with no polymorphism,missing rate＞25%and minimum allelic frequency(MAF)＜5%,total of 7706 SNPs corresponding to 1611 bins were successfully mapped into 19 chromosomes of B.napus using Join-Map 4.0 software.In detail,3176(41.21%)SNPs were localized to the A subgenome with average marker density of 0.87 cM,while 4530(58.79%)SNPs were mapped to the C subgenome with average marker density of 1.16 cM(Fig.2;Table S1).

The linkage map was subsequently employed for QTL mapping using phenotypic data from four years(2017–2019,2021).A total of 9 QTL for SNPS were identified across four years at the LOD threshold of 3.0,explaining 3.33%–60.64% of the phenotypic variance(Fig.2;Table S2).Based on the physical positions of the flanking markers for the identified QTL on the reference genome(B.napus_v2.0),the QTL were dispersed on chromosomes A01,A03,A06,A09,C01 and C09(Table S2).QTL on chromosomes A01,A03,A06,A09,C01 are environment-specific QTL that could only be detected in only one year with relative minor phenotypic variance(3.33%–8.72%).Only one overlapping major QTL(qSNPS.C09)showed a major effect on SNPS and was significant in all four environments with a mean LOD value of 21.53.qSNPS.C09 was mapped into a～1.0 cM interval flanked by markers M1546 and M1550(Fig.2)explaining 60.64%,56.56%,60.01% and 28.77% of phenotypic variance in the year 2017,2018,2019 and 2021,respectively,indicating a stable QTL of large effect on SNPS.The allele from parent 6Q006 with high SNPS contributes positive effect on SNPS in qSNPS.C09.The overlapping QTL interval was corresponded to the region of reference genome of rapeseed‘Darmor_v5’from 43.47 Mb to 45.86 Mb on chromosome C09,containing 407 genes.

To narrow down the interval of qSNPS.C09,the genomic DNA from‘H-DH’group with average SNPS of 32.97 and‘L-DH’group with average SNPS of 9.18 were sequenced together with two parents using the Illumina HiSeq 4000 platform(Fig.S1).A total of 6 Gb clean reads were produced for each sample with average of 15×depth coverage of the rapeseed reference genome(B.napus_v2.0)(Table S3).After filtering,94.82% of reads with high quality were successfully mapped into the reference genome.A total of 3,899,019 polymorphic loci were detected between the parental lines across the whole genome,including 5288 SNPs and 2602 InDels on chromosome C09.An extremely significantΔ(SNP/InDel-index)peak above the critical threshold level(P＜0.01)was located in the interval of qSNPS.C09,and thus qSNPS.C09 was narrowed to a 0.90 Mb region in the reference genome‘Darmor_v5’from 44.87 to 45.77 Mb on chromosome C09 harboring 169 genes(Fig.3).

3.3.RNA-seq analysis and candidate gene mining

Considering that an obvious seed abortion occurred between 7 DAP and 14 DAP in parental line 6W26,but not in parental line 6Q006(Fig.1b),we speculated that the causal gene and its related genes might exhibit differential expression pattern at 7 DAP and 14 DAP siliques between the extreme individuals in the segregation populations,e.g.,DH and F2.A set of‘H-F2’group from 8 F2individuals with average SNPS of 31.03 and‘L-F2’group from 6 F2individuals with average SNPS of 12.18(Fig.S1)were employed to RNA-seq using the Illumina HiSeq 4000 platform at 7 and 14 DAP siliques together with‘H-DH’and‘L-DH’groups,producing average of 5.5 Gb clean data for each sample.Total of 48,582 genes with significant changes in expression level between 7 and 14 DAP siliques were identified,including 11,586(3548 up-regulated and 8038 down-regulated),11,438(3723 up-regulated and 7715 down-regulated),11,724(4445 up-regulated and 7729 down-regulated),and 13,834(6437 up-regulated and 7397 down-regulated)DEGs in‘H-F2’,‘L-F2’,‘H-DH’and‘L-DH’groups,respectively(Table S4).Venn diagram showed that 202 upregulated(H_up)and 384 down-regulated(H_down)DEGs overlapped in two‘high SNPS’groups(‘H-F2’and‘H-DH’),while 211 up-regulated(L_up)and 290 down-regulated(L_down)DEGs overlapped in two‘low SNPS’groups(‘L-F2’and‘L-DH’)(Fig.4a).Those overlapped genes were assigned to 20 terms in Gene Ontology(GO)biological processes(Table S5).The top 10 GO enriched terms included cytokinin-activated signaling pathway,cell wall organization or biogenesis,response to brassinosteroid,response to hypoxia,response to auxin,secondary metabolic process,positive regulation of cellular biosynthetic process,external encapsulating structure organization,response to light stimulus,primary alcohol metabolic process(Fig.4b;Table S5).Considering that the cytokinin-activated signaling pathway was the most significant term among GO enriched terms,harboring 14 DEGs(Fig.4c),we chose eight DEGs(BnaC05.BRX,BnaA05.RR19,BnaC05.ABCG14,BnaC02.AHP4,BnaA06.ARR6,BnaA01.ARR7,BnaA05.RR3,BnaA07.ARR15)to investigate their expression at 7 and 14 DAP siliques by qRT-PCR analysis.As expected,all of genes exhibited same expression tendency as RNA-seq(Fig.S2).

Table 1 Analysis of variance for seed number per silique in DH population.

Fig.1.The phenotypic variation of seed number per silique in two parents and DH population.(a)Silique performances of the parents and their reciprocal crossing F1 hybrids.Scale bar,5 mm.(b)Observation for seed number per silique at 7 DAP(up)and 14 DAP(down)between parental line 6Q006(left)and 6W26(right).Scale bars,1 mm.(c–f)Frequency distribution of seed number per silique at maturity in DH population across 4 years.DAP,days after pollination.

Fig.2.Genetic linkage map and QTL mapping for seed number per silique(SNPS)in DH population.The genetic map was constructed with 7706 SNP markers.The consensus QTL for SNPS were marked with different colors.

Among all of the DEGs,only two DEGs BnaC09g44400D and BnaC09g45400D were located in the interval of qSNPS.C09.We amplified the full length sequences including the promoter region of the two genes from two parents,and found no sequence variance for BnaC09g44400D,but a 48-bp InDel variation(5′-TCACA ACTAC TTTCT TCTAC AAGTG GCGGT GAACA GAAAA AGACA GAG-3′)in the promoter of BnaC09g45400D,ranging from-840 to-793 bp from the start code ATG(Fig.5a).As expected,qRT-PCR analysis showed that BnaC09g45400D gene has no significant difference for the gene expression at 7 DAP siliques between two parents,but showed higher expression level at 14 DAP siliques in parent 6Q006 than in parent 6W26(P＜0.05)(Fig.5b).These results are in accordance with the observation that SNPS decreases sharply in 6W26,but not in 6Q006 from 7 DAP to 14 DAP.Therefore,we considered BnaC09g45400D as candidate gene underlying qSNPS.C09.

Using the protein sequence of BnaC09g45400D as a query,we identified 5 orthologs in Arabidopsis(Fig.S3).They share a conserved domain of adenine phosphoribosyltransferase,which catalyzes cytokinins from nuclear bases to nucleotides[19].Among which,At5g11160(AtAPT5)exhibit the highest homology with BnaC09g45400D(95.8% protein similarity)(Fig.S3).Thus,we named BnaC09g45400D as BnaC09.APT5.

Fig.3.Distribution ofΔSNP/InDel-index on chromosome C09.The horizontal axis represents physical distance,and the vertical axis representsΔ(SNP/InDel-index)value.Red curve is the result of sliding window analysis with a 1-Mb window size and 10-kb step size,red dotted lines are threshold value(0.73).

Fig.4.Gene expression pattern analysis and functional analysis of DEGs between high and low SNPS groups in DH and F2 at 7 and 14 DAP siliques.(a)Venn diagram of upand down-regulated DEGs in‘H-DH’,‘L-DH’,‘H-F2’,‘L-F2’groups.Numbers of DEGs in different databases are shown.(b)Top 10 enriched biological processes of Gene Ontology.(c)Expression heatmap for DEGs involved in cytokinin-activated signaling pathway.

3.4.Candidate gene association analysis

To test whether the 48-bp variance in the promoter of BnaC09.APT5 is associated with SNPS,we developed a pair of InDel primers(Table S6)to specifically amplify the fragment harboring 48-bp variance in DH population.49% of DH lines(45/91)with the same genotype as 6Q006(without 48-bp)exhibited higher SNPS than 51% of DH lines(46/91)with the same genotype as 6W26(with 48-bp)(P＜0.001)(Fig.5c).It indicated that the 48-bp sequence variation co-separated with SNPS in DH population.

We also used this InDel marker to genotype 157 rapeseed accessions,which were investigated across four years for SNPS in the previous study[3].As expected,this InDel variation was detected in this nature population,where 54 accessions with the same haplotype(A-48bp)as parent 6Q006 had higher SNPS than the 103 accessions with the same haplotype(B+48bp)as parent 6W26 across four years(P＜0.001)(Fig.5d).This further indicates that BnaC09.APT5 plays an important role in controlling SNPS in rapeseed.

Fig.5.Candidate gene BnaC09g45400D analysis in rapeseed.(a)Diagram of sequence differences of BnaC09g45400D between the two parents.The position of a 48-bp InDel variance is marked with triangle.(b)The expression patterns of BnaC09g45400D at 7 and 14 days after pollination siliques of two parents.(c)Co-separation analysis of functional marker for SNPS in DH population.(d)Haplotype analysis in nature population of rapeseed across four years.SNPS,seed number per silique;*,P＜0.05;***,P＜0.001(t-test).

4.Discussion

Seed number per silique is one of the three components of seed yield per plant in rapeseed,however,its genetic mechanism remains largely unknown.In this study,a candidate gene BnaC09.APT5 for seed number per silique was identified by combining QTL mapping and dynamic transcriptomic analysis.A 48-bp insertion in the promoter region of BnaC09.APT5 associated with SNPS in rapeseed.Our research provides a new insight into genetic mechanism of seed number per silique,and the functional marker developed according to the variances within BnaC09.APT5 will be helpful for molecular breeding in rapeseed.

4.1.Discovery of causal gene by combining QTL mapping with dynamic transcriptome sequencing

Most quantitative trait loci in crop species have been identified with QTL mapping and genome-wide association study(GWAS).However,a few causal genes underlying QTL were identified by the traditional map-based cloning approaches due to low resolution of QTL mapping and GWAS,especially in plants with long linkage disequilibrium decay distances.The availability of high-quality reference genome sequences and cheap deep sequencing technologies have accelerated gene function annotation and QTL mapping in model plants and crops by integration of omics data and genetics information.

The seed number per silique is finely regulated during seed development.In this study,seed abortion at 7–14 DAP siliques occurred in parent 6W26,but not in 6Q006,suggesting that causal gene and related genes exhibit differential expression patterns between two groups with extreme high and low SNPS.By subtraction of DEGs between high and low SNPS groups at 7 and 14 DAP siliques,we pinpointed a candidate gene in the interval of QTL.Its function in SNPS was verified by candidate gene association analysis in rapeseed.Our data suggest that the integrative analysis of multidimensional data provides alternative strategies to analyze the genetic architecture of complex traits.

4.2.BnaC09.APT5 controls seed number per silique possibly through cytokinin pathway

Seed development is a complex process involved in ovule initiation,fertilization,filling and ripening,which is influenced by many factors,including abiotic stresses in vitro[20,21],and transcriptional regulators and plant hormones in vivo[4,22–26].Among hormones,cytokinin has a central function in positively regulating reproductive meristem activity[6,27–29].Reduced expression of Gn1a,a gene of cytokinin oxidase/dehydrogenase(OsCKX2),causes cytokinin accumulation in inflorescence meristems and increases the number of reproductive organs,resulting in enhanced grain yield[27,30].Double mutant of the ckx3 and ckx5 increase 77%seed set per silique and 55%seed yield compared with the wild type in Arabidopsis[31].

There are five isoforms of Adenine Phosphoribosyl Transferase(APT)encoded in the Arabidopsis genome,among them,APT1 and APT3 may affect seed number.APT1 catalyzes cytokinins from nuclear bases to nucleotides,and an Arabidopsis thaliana mutant lacking APT1,is partially male sterile due to defects soon after meiosis,resulting in the reduction of pollen abortion[19].APT3 transcripts were detected in the ovary,particularly in the septum region(later also in the ovule),throughout microsporogenesis with higher levels detected after the tetrad stage(https://www.arabidopsis.org/servlets/TairObject?type=publication&id=501710086),indicating a possible role in ovule development.AtAPT5 is a member of APTs with a high homology with AtAPT1[19].In the present study,a homolog of AtAPT5,BnaC09.APT5 located in the interval of major QTL,exhibited different expression pattern between two parents at 7–14 DAP siliques,and associated with SNPS in natural population of rapeseed.GO analysis revealed that DEGs between‘High’and‘Low’SNPS groups were enriched in the cytokinin-activated signaling pathway as the most significant term.Those findings suggest BnaC09.APT5 may control seed number per silique possibly through cytokinin-related pathway in rapeseed.However,more details should be required for understanding the molecular mechanism.In conclusion,this study provides a candidate gene for rapeseed molecular breeding.The application of the elite genes like BnaC09.APT5 will in a sense alleviate the pressure of insufficient edible oil supply caused by the increase of population.

CRediT authorship contribution statement

Shuangshuang Xin:Conceptualization,Writing–original draft,Formal analysis,Investigation,Visualization.Hongli Dong:Conceptualization,Writing–original draft,Formal analysis,Investigation,Visualization.Yixin Cui:Methodology,Software,Data curation,Investigation.Yilin Liu:Methodology,Software,Data curation,Investigation.Guifu Tian:Methodology,Software,Data curation,Investigation.Nanxi Deng:Methodology,Formal analysis,Investigation.Huafang Wan:Methodology,Writing–review&editing.Zhi Liu:Methodology,Writing–review&editing.Xiaorong Li:Methodology,Writing–review & editing.Wei Qian:Conceptualization,Project administration,Funding acquisition,Writing–review & editing.

Declaration of competing interest

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

Acknowledgments

This work was supported by the National Basic Research Program of China(2015CB150201),the Natural Science Foundation of Chongqing(cstc2019jcyj-bshX0055,cstc2019jcyj-zdxmX0012,and cstc2020jcyj-msxmX0461).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.07.012.