Identification Of QTNs And Candidate Genes For Seed Size And Shape Traits During Domestication And Improvement In Soybean

Soybean(Glycine max(L.)Merr.,2n = 2× = 40)is economically imperative food and oilseed crop worldwide.Artificial selection during soybean domestication and improvement results in substantial phenotypic divergence between wild and landrace and between landrace and bred soybeans.Therefore,mining the genes that control seed size and shape traits has great significance to understand the evolution of soybean.Seed size and shape traits were different between wild and landrace soybeans and between landrace and bred soybeans.However,limited information is available about the quantitative trait nucleotides(QTNs)and candidate genes related to domestication and improvement traits,such as seed size,seed shape,seed development,and seed weight.To address this issue,in this study 286 soybean accessions(14 wild,153 landrace,and 119 bred soybeans),genotyped by 106,013 high-density SNPs and phenotyped by seven seed size and shape traits in 10 environments,were used to conduct a selective sweep and multi-locus genome-wide association studies(GWAS)to identify domesticated and improved QTNs and their candidate genes for seed size and shape traits in soybean.Firstly,the SNP information in 286 soybean accessions was analyzed by the XP-CLR approach to identify selective sweeps.Secondly,the SNP markers were associated by six multi-locus GWAS approaches with seed size and shape traits to detect significant QTNs,these QTNs were compared with the above-mentioned selective sweeps,along with those in previous studies,to obtain the domesticated and improved QTNs for these traits.Thirdly,domestication and improvement genes were identified within the interval of 100 kb of each domesticated and improved QTN.The high expression and differential expression analysis across soybean varieties in early and mid-maturation stages of seed development were conducted to select candidate domestication genes(CDGs)and candidate improvement genes(CIGs).To confirm these candidate genes,allele frequency differences in wild,landrace,and bred soybeans were investigated,and LOF function and UTR variants were mined.In addition,these CDGs and CIGs were used to explain the differences of seed size and shape traits between wild and landrace soybeans and between landrace and bred soybeans.Finally,we extracted the elite alleles of QTNs for seed size and shape traits identified in this study to predict the best cross combinations.The main findings are as follows:1.Analysis of variance(ANOVA)showed significant differences between wild,landrace,and bred soybeans,indicating the presence of an evolutionary effect in soybean.Further,orthogonal contrast analysis revealed significant differences of seed size and shape traits between wild and landrace soybeans(F-value 5.94 to 214.29)and between landrace and bred soybeans(F-value 273.49 to 1684.26).Using polygenic and residual variances,broad-sense heritabilities for seed size and shape traits ranged from 60.36 to98.17(%),implying the existence of genetic variation in 286 soybean accessions.2.The phenotypic and BLUP values for the above-mentioned seven traits in 286 soybean accessions were associated by six multi-locus GWAS models with 54,950 SNP markers.As a result,a total of 1748 and 331 significant QTNs(LOD > 3)were detected.Of 1748 QTNs,412,403,418,291,260,278,and 303 were significantly associated with SL,SW,ST,100-SW,SLT,SLW,and SWT,respectively,and their LOD scores and PVE values ranged from 3.00~38.72 and 1.16E-09~23.38(%),respectively.Among 331 BLUP QTNs,65,75,74,71,50,45,and 38 were significantly associated with the abovementioned traits,respectively,and their LOD scores and PVE values ranged from3.00~26.26 and 5.25E-07~16.05(%),respectively.Among all the above 1799 QTNs,1382 were known,and 417 were new.3.All the above-mentioned 1799 QTNs were compared with 952 candidate domestication regions(CDRs)and 538 candidate improvement regions(CIRs),and 245 and 187 QTNs(or clusters)were found to be in CDRs and CIRs,respectively.Around the245 and 187 QTNs,there were a total of 5376 genes in selective sweeps.To obtain reliable candidates,transcriptome datasets at seven seed developmental stages were used to determine high expression genes at the R4-R7 seed developmental stages.As a result,there were 2513 high expression genes.4.Using the genotypic datasets of 302 soybean accessions in previous study,SNPs with significant allele frequency differences between wild and landrace soybeans and between landrace and bred soybeans were captured in 1602 out of 1667 potentially domesticated and 790 out of 1383 potentially improved genes,respectively,and these SNPs were annotated as LOF and UTR variants.The genes with UTR variants were subjected to differential expression analysis using gene expression levels of two wild,two landraces,and two bred soybeans at 15,25,35,and 55 days after flowering.A total of1227 potentially domesticated and 361 potentially improved genes were identified after merging differentially expressed genes(DEGs)and genes with LOF variants.In furtheranalysis,a total of 73 CDGs and 36 CIGs were found by RNA-seq analysis,KEGG enrichment analysis,and mi RNA targets to be associated with seed size-related traits.Among 73 CDGs and 36 CIGs,12 CDGs and 12 CIGs had known seed-seize/weight function in soybean,and 61 CDGs and 24 were new.SNP variants in CDGs were further verified by the reference genomes of two wild(W05 and PI483463)and one landrace(Williams 82)soybeans,whereas SNP variants in CIGs were confirmed by the reference genomes of one landrace(Williams 82)and one cultivar(Zhonghuang 13)soybeans.5.The seed trait differences between wild and landrace soybeans and between landrace and bred soybeans were explained by domesticated and improved genes in the following aspects: 1)The alleles of domesticated and improved QTNs located near CDGs and CIGs showed significant differences in 100-SW.The proportion of SW increasing alleles of domesticated QTNs was significantly increased in landrace soybean compared to wild soybean.Similarly,frequencies of SW increasing alleles of improved QTNs were significantly increased from landrace to bred soybeans.2)LOF mutations in CDGs and CIGs involved substitutions that result in functionally different amino acids;these mutations can lead to alteration or loss of protein function,which might be involved in seed trait differences.Sixteen CDGs and eight improved genes had LOF mutations.Similarly,the allele frequency of SNPs in UTRs of 46 CDGs and 19 CIGs showed significant differences between wild and landrace soybeans and between landrace and bred soybeans,respectively,which may affect transcription.Among these,11 CDGs and three CIGs were also targeted by mi RNAs,and these mi RNAs may be involved in posttranscriptional regulation of gene expression.3)The homologous genes of CDGs and CIGs were involved in seed size/weight and development in Arabidopsis;however,one CDGs(Glyma20g34360)and two CIGs(Glyma02g47030 and Glyma05g33340)encoded unknown protein but these genes in this study were identified as candidate genes for these seed traits.6.The numbers of elite alleles for SL,SW,ST,100-SW,SLT,SLW,and SWT were101,108,106,92,76,55,and 50,respectively.Among 286 soybean accessions,the number of elite alleles for the above seven traits ranged 8~97,13~89,12~97,10~72,5~46,4~35,and 5~32,respectively.Based on the above-mentioned elite allele information,the best five cross combinations for each trait were predicted.For example,Deqingxiangzhudou × Yafanzaodou may be available to improve seed size traits,and Jindou 3 × Heihe 54 may be available to improve seed size and shape traits.This study provides a framework for the systematic identification of domesticated and improved genes in soybean,as well as elite alleles of these genes.Thus,this study provides useful information for trait genetic foundation,marker-assisted selection,cross combination prediction,and functional genomics for seed size and shape traits.