QTL Mapping And Candidate Gene Identification For Agronomic And Yield-related Traits In Foxtail Millet(Setaria Italica)

Foxtail millet has developed into a model genetical system for deciphering the plant architecture,C₄ photosynthesis,nutritional properties and abiotic stress of cereal crops due to its short life cycle,small stature,self fertilization,high reproductive coefficient,C₄attribute,abiotic tolerance,small genome（～430Mb）,diploid,rich nutrition and other superior characteristics.During the domestication process of foxtail millet from its wild ancestor,the morphological architecture of the plant has undergone tremendous changes.In particular,the selection of the vegetative and inflorescence branch is extremely strong.And the vegetative branch（tiller,axillary branching）and inflorescence structure directly determine the biomass and grain yield of foxtail millet.However,their studies on molecular quantitative genetics are still rare.Thus,molecular genetic studies of agronomic and yield-related traits are necessary for dissection of molecular regulation mechanism and breeding application in foxtail millet and its relatives.In the present study,a recombinant inbred line population was constructed from a cross between Yugu1 and Longgu7.The two parents and all lines in the population were resequenced by using the Illunima platform.And SNP were obtained by alignment with reference genome.A high-density genetic map was constructed.Combined with the phenotypic data,QTL for 17 agronomic and yield-related traits was preliminarily mapped across multi-environments.Furthermore,using（Yugu1×Setaria viridis H1）F₂,F_2:3,F_2:4 populations,QTL（q TN5.2）for tiller on chromosome 5 and QTL（q AB9.1）for axillary branching on chromosome 9 were finely mapped.In addition,the F₂ secondary population was established from a cross between RIL123 and Longgu7 to finely map QTL（q GWP9.1）for grain weight per plant on chromosome 9.Finally,combining with the results of QTL fine mapping,the candidate genes were identified by the transcriptome analysis of different development stages of panicles between Yugu1 and Longgu 7 and tillering and axillary bud differentiation stages between Yugu1 and Setaria viridis H1.The main results are as follows:1.Construction of a high-density genetic mapBased on a Yugu1×Longgu7 RIL population and genome-wide resequencing data,a linkage map harboring 2297 bin markers generated from 1325599 SNPs and 167818In Dels on Hidden Markov model（HMM）and 74 SSR markers was constructed,spanning 1315.1 c M with an average distance of 0.56 c M.The number of markers on chromosomes in the map is for 195-372,the genetic distance is for 117.73-192.42c M,and there are 17 gaps larger than 5c M.2.QTL mapping for 17 agronomic and yield-related traitsA total of 221 QTL for 17 agronomic and yield related traits measured on（Yugu1×Longgu7）RIL population across 14 environments were identified,explaining 5.5～36%of the phenotypic variation,of which 62 QTLs were detected in at least three environments,including the most stable q LMS6.1 mapped on nine environments.One hundred-thirty QTL with overlapping intervals formed 22 QTL clusters.Of these,the increasing phenotypic alleles of 72 QTL were from Longgu7,and the increasing phenotypic alleles of 149 QTL were from Yugu1.3.Fine mapping of QTL for tiller,axillary branching traitsQTL（q TN5.2）for tiller was located in the 37265590-40247648bp interval on chromosome 5 based on the results of the preliminary mapping for tiller on（Yugu1×Longgu7）RIL population and the bulk segregation analysis（BSA）between B1（no tillers and no branches）and B3（multiple tillers and no branches）constructed on the（Yugu1×Setaria viridis H1）population.And 53 In Dels and 4 SSR were developed to detect 889 individual genotypes from（Yugu 1×Setaria viridis H1）F₂population.Then,a high-density linkage map was constructed,covering 10.7c M with the average genetic distance of 0.19c M on chromosome 5.Combined with the phenotypic values of tiller on F₂,F_2:3,F_2:4 popolations and the genotype data of plants with chromatid exchange in F_2:4population,q TN5.2 was finely mapped to the 0.57c M,with a physical distance of 255kb（39551535-39807014bp）,a LOD value of 16.72-34.62,an increase of phenotypic value of 0.53-0.72,and an explanation of phenotypic variation of 8.2-13.2%.The favorable allele was derived from Setaria viridis H1.QTL（q AB9.1）for axillary branching was finely mapped in the4350040-10448047bp interval on chromosome 9 according to the results of BSA analysis between B2（multiple tillers and multiple branches）and B3（multiple tillers and no branches）constructed on（Yugu1×Setaria viridis H1）F_2:3 population.And 43In Dels and 19 SSR markers were developed in the interval.Then,889 individual plants on（Yugu 1×Setaria viridis H1）F₂ population were genotyped,and a high-density linkage map covering 33.5c M with an average distance of 0.54c M was constructed.Combined with the phenotypic values of axillary branching on F₂,F_2:3,F_2:4 populations and the genotype data of plants with chromosome recombination on F_2:4population,q AB9.1 was finely mapped to the 0.33c M interval,with a physical length of 521kb（7465265-7986643bp）,a LOD value of 5.37-20.64,and an increased phenotypic value of 0.33-2.39,which explained 4.3-10.1%of the phenotypic variation,and the favorable allele came from Setaria viridis H1.4.Candidate genes for tiller,axillary branching traitsIn accordance with the fine mapping for tiller and axillary branching and transcriptome data,it is shown that among the 45 genes in the q TN5.2 interval on chromosome 5,Seita.5G356600 and Seita.5G357400 had significantly different expression between Yugu1 and Setaria viridis H1 at the tillering stage.And there had a single base non synonymous variation in the coding region of Seita.5G356600.The two genes were predicted to be candidate genes for tiller q TN5.2.In the axillary bud differentiation stage,8 of the 60 genes covered in the q AB9.1 interval,Seita.9G120600,Seita.9G122600,Seita.9G122700,Seita.9G123300,Seita.9G124100,Seita.9G125000,Seita.9G125300 and Seita.9G125400 had significant differences in expression,and Seita.9G122600,Seita.9G122600 and Seita.9G122700,Seita.9G124100 and Seita.9G125300 had non synonymous variations in the coding regions,which were predicted to be candidate genes for q AB9.1.5.Fine mapping of QTL for grain weight related traitsRIL123,a recombinant inbred line with favorable alleles derived from Yugu1 in q GWP9.1 region,was selected to cross with Longgu7 to construct F₂ secondary mapping population on the basis of the preliminary mapping of QTL（q GWP9.1）for grain weight per plant in RIL population.And 35 In Dels and 3 SSR were developed within the QTL（q GWP9.1 and q PWP9.1）regions to genotype 1287 individual plants.A high-density linkage map with a genetic distance of 45.65c M was constructed.Combined with the phenotypic data for panicle weight per plant,grain weight per plant,grain number per plant and 1000-grain weight in F₂ lines,the QTL for panicle weight per plant,grain weight per plant and grain number per plant were finely mapped to the1.89-1.96Mb and 2.91-3.16Mb by inclusive composite interval mapping.LOD values were 9.70 and 7.99 and 9.97,respectively,explaining 3.90%and 3.25%and 3.93%of phenotypic variation.The additive effects were 1.68g,1.36g and 478.5,respectively.And the favorable alleles were all from RIL123.6.Candidate genes for grain weight related traitsFour genes,Seita.9G034700 at booting stage,Seita.9G034700 and Seita.9G035000at flowering stage,Seita.9G034600 and Seita.9G035800 at filling stage,were significantly different expression between Yugu1 and Longgu7 and were identified as candidate genes for QTL（q GWP9.1 and q PWP9.1）on chromosome 9 according to the fine mapping results of panicle weight per plant,grain weight per plant and grain number per plant combined with the transcriptome anlysis.And five genes,Seita.9G054000 at the young panicle stage,Seita.9G050900,Seita.9G051900 and Seita.9G054500 at booting stage,Seita.9G051900 at flowering stage,and Seita.9G050900,Seita.9G052200 at filling stage were significantly different expression between Yugu1 and Longgu7 and were preliminarily identified as candidate genes for q GNP9.1.