| Cadmium(Cd)contaminations in soils lead to numerous Cd accumulation on crop plant.Accumulation of Cd poses a significant risk to human health as it is ingested via the food chain.Maize is one of the world’s most important food crops and has a strong ability for uptake and transport Cd.It is of great significance to strengthen the breeding of maize varieties with low Cd accumulation and to analyze and control the genetic rules of Cd accumulation.In this study,a set of association population with rich genetic diversity and a ten-generation intermated B73 × Mo17(IBMSyn10)double haploid(DH)population were used as experimental materials.Genome-wide association study(GWAS)and linkage mapping were conducted to identify the genetic rule for controlling the Cd accumulation in leaves and kernels in maize under different Cd conditions.In addition,bioinformatic tools were used to analyze the phylogenetic relationship,gene structure,and motif component of ZmHMAs,and finally we conducted the candidate-gene association analysis to identify favorable alleles of ZmHMA3 for controlling Cd accumulation in leaves of maize.Furthermore,selection breeding is an essential tool of variety breeding and estimation of breeding value is the core of selection breeding.Genome prediction was used to explore the application of whole genome selection in maize Cd accumulation and assess predictive accuracy on different models and population size.Combining with the results of GWAS,we examined the effects of prediction accuracy on significant SNPs and whole SNPs.Main results of this study are as follows:(1)Two hundred and sixty-nine,diverse accessions were evaluated for Cd accumulation of leaves under low-Cd and middle-Cd conditions at the seedling stage,and Cd accumulation of leaves and kernels under the high-Cd condition at the maturing stage.Analysis of variance(ANOVA)indicated that diverse accessions differed significantly in Cd concentration of leaves,while significant genotype by environment interactions was observed.With the level of Cd contamination in soil increasing,Cd concentration in leaves significantly increased.Correlation analysis showed there was positive relationship between kernel Cd concentration and leaf Cd concentration.In addition,there was a remarkable difference in Cd accumulation of maize among different subgroups.For example,the mean value of Cd concentration of leaves and kernels in the tropical germplasm was significantly lower than that in the stiff stalk(SS)and non-stiff stalk(NSS)in different environments.A total of 5 lines of tropical germplasm with low-Cd accumulation were selected by principal component analysis(PCA)and phenotype cluster in this study.The inbred lines of low Cd accumulation would provide the new germplasm for breeding of the variety with low-Cd accumulation and the genetic basis of Cd accumulation.(2)Combining a set of 43737 high quality SNPs(Single Nucleotide Polymorphism),a model Q+K(Population structure and Kinship matrix)was applied to identify association SNPs for Cd concentration of leaves and kernels.At P value less 1.18×10-6,10 SNPs,5SNPs and 62 SNPs located on 5 of the 10 chromosomes were significantly associated with leaf Cd concentration under low-Cd,middle-Cd and high-Cd conditions,respectively.Meanwhile,the average phenotypic variation was 20.07%,20.57% and 15.80%,respectively.Among significant SNPs,10 SNPs in a single region located on chromosome between PZE-102116673(155,082,333 bp)and PZE-102121893(168,349,525 bp).Five SNPs associated with kernel Cd concentration were located on chromosome 1,4 and 6were significantly associated with kernel Cd concentration.Each locus explained from17.63% to 20.24% of phenotypic variations on average.(3)Quantitative Trait Locus(QTL)for leaf Cd accumulation was detected using IBMSyn10 DH population with an ultra-high density binmap.Composite interval mapping was performed using the threshold of LOD ≥ 3.84.Five QTLs for leaf Cd concentration was mapped and located on chromosomes 2,5,7,8,and 9.Noteworthy,a major QTL named q LCd2 was identified,explained 41.24% and 38.41% of the phenotypic variation in2015 and 2016,respectively.(4)Based on the results of GWAS and QTL mapping,a major and co-located QTL on chromosome 2 was associated with Cd concentration in leaves of maize.According to the gene functional annotations,GRMZM2G455491 and GRMZM2G175576,simultaneously encoding a cadmium/zinc-transporting ATPase and belonging to HMA family,might control the Cd accumulation in maize leaves.(5)HMA family is one of the currently known heavy metal transport which plays an important role in controlling the metal transfer and accumulation in plant.A total of 12 ZmHMA genes were identified in the maize genome in this study.On the basis of phylogenetic tree topology and domain composition,HMA proteins could be subdivided into six clusters.ZmHMA1-ZmHMA4 belong to the Zn/Co/Cd/Pb subgroup and ZmHMA5-ZmHMA12 belong to the Cu/Ag subgroup phylogenetically.Expression analyses shown that the expression level of ZmHMA family genes varied greatly in different tissues and had the space-time difference.ZmHMA4(GRMZM2G455491)was not expressed or exhibited at an extremely low level in different tissues collected under a non-limiting condition,except anther tissue.The greatest Cd-inducible gene response was observed in ZmHMA3(GRMZM2G175576)of roots,stems and leaves,which was up regulated more than 5-fold,16-fold,and 2 fold after 24 h of Cd treatment compared to normal growing conditions,respectively.(6)Candidate-gene association analysis between extracted phylogenetic sites in ZmHMA3 and Cd leaf concentration of maize was conducted using Q+K model(population structure matrix and kinship matrix).Twenty and three polymorphic sites were associated with Cd concentration under various soil environments(p < 0.01).Furthermore,4 SNPs and 2 In Dels sites located on the CDS region and caused the amino acid substitutions and insertions or deletions.Subcellular Localization indicated that ZmHMA3 protein was primarily localized on the plasma membrane.(7)Models of genomic prediction(Bayes A,Bayes Lasso,Bayes B,RR-BLUP,Gaussian kernel and Exponential kernel)were used to predict the breeding value for Cd accumulation of cross population of maize.A certain proportion(50%,70% and 90%)of maize lines randomly selected from the associate population were used to as a training population to predict the phenotype of the remaining 25% of the associates(cross population).The study found that the prediction accuracy of RR-BLUP,Gaussian kernel and Exponential kernel had remarkably higher of the accuracy which predicted by Bayes A,Bayes Lasso,and Bayes B.Furthermore,the prediction accuracy had increased with the size of the training population increase.When the size reached 70% of the associate population(200 accesses),the fastest-growing tendency was observed.In addition,accuracy of different phenotype predicted by the same model was also was also obsessively differed.For all traits,the prediction of LCd15-Field and LCd16-Field was the best.In addition,combing the GWAS results,the prediction accuracy was increased obviously by using a set of significant SNPs(p < 0.001).Accuracy of LCd15-Low,LCd16-Middle,LCd16-Field,and KCd15-Field was improved by an average of 13.97%. |
