| Soybean oil traits are related to nutrition,and seed size traits are not only related to yield breeding targets but also a quality parameter.These are domesticated traits.Therefore,the study of these traits has theoretical and practical significance.Although soybean seed oil traits are main objectives for geneticists to decipher genetic architecture of traits.However,most of the previous studies based on the relationship between traits(or metabolites)and genes(two-dimensional),rather than on the threedimensional network relationship between gene-metabolite-oil traits,meanwhile.In the light of previous studies,only few pleiotropic and epistatic genes have been reported.Therefore,a systematic understanding of metabolites of soybean seed oil composition,and elucidating the genetic network relationship of oil properties will have a positive effect on the improvement of soybean traits.In this study,we analyzed the genetic network relationship between oil-related traits,metabolites and genes in combination with genome-wide association study,metabolomic,transcriptomic and molecular biology.Firstly,6 seed oilrelated traits from 286 soybeans in 2014,2015,2016 and 52 oil-related metabolites from 214 soybeans in 2015 together with genotypic data were used to do pGWAS and mGWAS,along with additive-by-additive epistatic effect and G × E interaction effect analysis to identify the most likely candidate genes involved in the oil composition and metabolites.Then,the MCP,SCAD and t-test were used to detect the significant relation between seed oil-related traits and metabolites,combining the transcriptional expression analysis of candidate genes involved in the oil composition and metabolites and were found main genes involved in oil synthesis.Afterthat,we constructed a genetic network between gene-metabolite-oil trait,to validate these results,five high-oil and five low-oil accessions were used to conduct hypothesis testing for each node(gene,metabolite or trait)in the above sub-networks,and the protein interaction between main genes in the network was verified by luciferase complementary image analysis(LCI).Finally,6 oil-related and 4 seed size-related traits in 2011,2012,2014 and 2015 from 286 soybeans,and 5 TAGs form 398 RILs together with genotype data,were used for single-locus GWAS,multi-locus GWAS and linkage analysis to find the association between SNP in the 20000～21000 Kb region on chromosome 13(Chr13).Further,we used these QTNs to find the candidate genes associated with seed oiland seed size-related traits,and analyzed the genetic variants of candidate genes.The transgenic experiment was used to verify the function of candidate genes related to seed oiland seed size-related traits.So,statistical analysis,gene molecular structure analysis and LCI experiment were used to verify the interaction between the two QTNs or genes.Histocytological comparison of over-expression and RNA interference soybean seeds were used to analyze the cytological reasons in the changes of seed oil content and size.The main results of genetic network between gene-metabolite-oil trait were as follows:1.A total of 334 significant QTNs were found in the pGWAS,associated with oil traits in 2014,2015 and 2016.Among them,56,46,50,68,75,and 39 QTNs associated with palmitic,stearic,oleic,linoleic,linolenic and oil concentrations,respectively.The numbers of QTNs detected in metabolic traits ranged from 46 to 75,these significant loci show large effects to explain the metabolites variation with an average level of 5.51%.Interestingly,we identified 1001 significant QTNs in mGWA correlated with 52 oil-related metabolites;262,116,66,111,96 and 350 QTNs were found associated with 9 organic acids,5 soybean isoflavone,6 PE,6 PI,6 PC and 20 amino acids,respectively.The number of QTNs was detected,on average 12～37 QTNs in each metabolite trait and an average variation for each QTN was 6.63%.In the QTL × E analysis(QEs)in oil related traits,26 significant QEs were detected.Among them,4,8,10 and 4 QEs(P<0.001)were identified in oil content,linolenic acid,palmitic acid and stearic acid,respectively.In the QEs significantly associated with linolenic acid,we identified a candidate gene GmCds1 near the loci Chr18-4720420(p=6.5e-4)site(63.8 kb)is associated with phosphatidylglycerol biosynthesis in the soybean metabolic pathway database.In the epistatic effect analysis,26 pairs additive-by-additive interaction effects(QQs)were found;3,5,9,4,3 and 2 significant QQs(LRT≥ 9.21)were identified in oil content,linolenic,linoleic,oleic,palmitic and stearic acid,respectively.In the QQs significantly associated with linolenic acid,oil synthesis related genes GmPDAT and GmDAGAT1 were found near the Chr13-20532852 bp and Chr13-20704034 bp sites(LRT≥20.08).In the MCP,SCAD and t-test results,31 significant correlations between oil traits and metabolites were found.2.The candidate genes were found within 100 kb window from each QTN that was associated with oil traits and metabolites.A total of 284 and 279 candidate genes significantly associated with oil-related traits and metabolites were identified through comparative genome analysis,KEGG and differential expression analysis,and 43 candidate genes were found common between them.The expression levels of 12 genes in cultivated soybeans were significantly higher than that of wild soybeans,such as,GmFATA2,GmFATB1a,GmPDHC,GmTIM,GmGPDH,GmLACS2,GmPLA2A,GmDGAT1/GmDAGAT1,GmPDAT,GmPgs1,GmPLDa1 and GmLPEAT2.In the above candidate genes,28 main genes were found in the oil synthesis,amino acid and TCA cycle 3 primary metabolic networks,19 of them are the main candidate genes related to oil synthesis.At present,GmFATB1a,GmLPEAT2,GmPLDa1 and GmDAGAT1 are function-known genes.The functions of other genes are inferred by comparative genomics and expression profile data analysis.Comparative genomics analysis showed that GmSDH1,GmMDH,GmFUM1 and GmIDH-V involved in the TCA cycle processes and GmAGT,GmP5C1,GmHMT2 and GmAtbcat involved in soybean amino acid synthesis processes,respectively.3.Using the significant correlations between oil traits,metabolites and candidate genes associated with them,a total of 133 3-dimension gene-metabolite-oil traits networks were constructed.Among these sub-networks,24 were found to be involved in known KEGG metabolic pathways(https://www.kegg.jp/kegg/pathway.html),e.g.pyruvate-GmPDATGmFATA2-oil content and 109 were newly identified.To support the results of this study,three pieces of evidences are provided here.First,part of the nodes in the above 60 subnetworks have literature reports by molecular biology studies;second,20,51,46 and 16 subnetworks were found to have one,two,three,and four significant nodes,respectively,between five high-oil and five low-oil soybean accessions;lastly,the interaction between GmPDAT and GmDAGTA,GmFATA2 were detected by LCI.Using those networs,GmPDAT,GmAGT and GmACP4 reveal the genetic relationships between pyruvate and three major nutrients,and GmPDAT,GmZF351 and GmPgs1 reveal the genetic relationship between amino acids and seed oil content.Moreover,GmCds1,along with average temperature in July and the ainfall from June to September,influence seed oil content across years.From the analysis above,it can be seen that GmDAGATl and GmPDAT are related to soybean oil-related traits,and the function of AtPDAT1 in Arabidopsis is inconsistent.To this end,we further studied the oil-related genes on chromosome 13 and verified them through transgenic experiments.The main results are as follows:The main results in identifying pleiotropic and epistatic gene for seed oil-and sizerelated traits were as follows:4.With both univariate and multivariate analyses,in the region of 20000～21000 Kp on Chr 13,Chr13-20532852 bp locus was found significantly associated with linolenic acid(P=8.28e-09～1.58e-06).And Chr13-20682820-20719806 bp loci were associated with oleic acid,linolenic acid,linoleic acid,seed thickness,seed length,seed width,and 100-seed weight(P=6.08e-10～7.17e-06,LOD=2.62～4.33).Within the range of about 100 kb for each of the two loci(20532852 and 20704079 bp),there were 34 and 33 genes,respectively.Among these genes,GmDGAT1 and GmPDAT were closely linked to loci Chr13-20532852(38.4 kb)and Chr13-20704079 bp(8.7kb),and these were candidate genes of them.To further confirm the two candidate genes,first,we analyzed the variants of GmPDAT and GmDGAT1 across wild and landrace soybeans,and found five and four SNPs,respectively,and with four and one SNPs were further confirmed in 302 soybean accessions.Second,linkage analysis of 5 TAGs traits was carried out,and significant QTLs near these two genes were detected.5.The Real-time quantitative PCR analysis of GmPDAT transgenic soybean showed that,the expression level of GmPDAT was 5.93～6.81,0.80～1.29,and 2.68 in OX,RNAi and wild-type(WT,W82)plants,respectively.Obviously,the expression level was found significantly higher in OX than in WT(P=1.66e-04～3.26e-03)plants,and the expression level was significantly lower in RNAi than in WT plants(P=5.38e-06～8.24e-05).Through the measurement of the seed oil-and seed size-related traits(SL,SW,ST and 100SW,we identified GmPDAT gene positive plants in T2.Interestingly,we found that seed oil contents in OX,RNAi and WT plants were 194.10～222.53,140.03～160.89 and 177.86 mg/g,respectively;seed linoleic acids were 106.99～120.56,72.26～91.02 and 95.93 mg/g,respectively;seed linolenic acid were 14.90～18.64,9.40～11.98,and 13.02 mg/g,respectively.In addition,we found seed size traits among OX,RNAi and WT plants:100SW were 19.75～19.97,17.54～18.15 and 18.46 g,respectively;SL were 7.82～8.12,7.24～7.27 and 7.32 mm,respectively;SW were 6.87～7.05,6.48～6.54 and 6.59 mm,respectively;ST were 5.47～5.64,5.02～5.38 and 5.54 mm,respectively.Seed oil content,seed linoleic acid,seed linolenic acid,100SW,SL and SW in OX significantly larger than that in WT plants.Seed oil content,seed linoleic acid,seed linolenic acid and ST in RNAi significantly smaller than that in WT plants.The cell slice analysis of outer integument cell size in the seeds of WT,RNAi and GmPDAT overexpressing plants,we found that the cell lengths and widths in the mature seeds of OX-30 transgenic plants increased 15.7%and 9.0%than WT plants,and increased 14.7%and 22.0%than WT plants in the premature seeds(45-DAF),respectively.The cell lengths and widths in the mature seeds of RNAi-16 transgenic plants decreased 25.2%and 23.5%than WT plants,and 9.0%and 12.8%in the premature seeds(45-DAF),respectively.The cell sizes of OX plants were significantly larger than those of WT seeds.Moreover,we also compared the oil bodies(OBs)of OX,WT and RNAi seeds using transmission electron microscopy.OX seed cells contained apparently bigger OBs than WT and RNAi,and the seed cells in RNAi-16 appear to contain the smallest OBs.By linkage analysis for TAGs,we found that four QTLs(Chr13-20606048-Chr1320775650 bp)with r2 from 3.27%to 10.15%were around the gene GmPDAT.Those results showed that GmPDAT may participate in the control of TAGs synthesis to increase the OBs of soybean seeds,and increases the size of soybean seed cells in order to store more oil contents.In addition,we found some signs of the protein interactions between GmPDAT and GmDAGAT1.First,the PEPIS program was used to detect the epistatic effect between the above two loci(Chr13-20532852 and Chr 13-20704079)for seed oil-and size-related traits collected in 2011,2012,2014 and 2015.The significant epistatic effects were detected and responsible for the variation in linolenic acid(LRT=24.37,2014;LRT=11.52,2012),and seed thickness(LRT=10.80,201 1).Second,the protein interactions between candidate genes were analyzed by online website STRING(https://string-db.org//).and found the protein interaction signals between them(the statistic value was 0.75,which was significantly larger than the medium confidence of 0.40).Lsatly,there was a strong interaction signal found between GmPDAT and GmDAGAT1 validated by LCI.Then,the transgenic experiments did not detect the interaction between the two genes.This study provides a new way in clarifying the genetic network of soybean oil synthesis,and also providing a new method for identifying polypotency and epistasis genes. |
