Analysis of the genetic architecture of maize kernel size traits by combined linkage and association mapping

Abstract

Kernel size-related traits are the most direct traits correlating with grain yield. The genetic basis of three kernel traits of maize, kernel length (KL), kernel width (KW) and kernel thickness (KT), was investigated in an association panel and a biparental population. A total of 21 single nucleotide polymorphisms (SNPs) were detected to be most significantly (P < 2.25 × 10^-6 ) associated with these three traits in the association panel under four environments. Furthermore, 50 quantitative trait loci (QTL) controlling these traits were detected in seven environments in the intermated B73 × Mo17 (IBM) Syn10 doubled haploid (DH) population, of which eight were repetitively identified in at least three environments. Combining the two mapping populations revealed that 56 SNPs (P < 1 × 10^-3 ) fell within 18 of the QTL confidence intervals. According to the top significant SNPs, stable-effect SNPs and the co-localized SNPs by association analysis and linkage mapping, a total of 73 candidate genes were identified, regulating seed development. Additionally, seven miRNAs were found to situate within the linkage disequilibrium (LD) regions of the co-localized SNPs, of which zma-miR164e was demonstrated to cleave the mRNAs of Arabidopsis CUC1, CUC2 and NAC6 in vitro. Overexpression of zma-miR164e resulted in the down-regulation of these genes above and the failure of seed formation in Arabidopsis pods, with the increased branch number. These findings provide insights into the mechanism of seed development and the improvement of molecular marker-assisted selection (MAS) for high-yield breeding in maize.

Keywords: QTL mapping; co-localization; functional genes; genome-wide association study; kernel size; maize.

Figure 1

Phenotypes of kernel size traits and variations of kernel size between two parental lines of the IBM Syn 10 DH population. Phenotypes of KL, KW and KT illustrated with 10 kernels of the two parental lines in IBM Syn 10 DH population. Bar = 1 cm.

Figure 2

Manhattan plots of the association analysis for KL in four environments by FarmCPU. Manhattan plot of KL on 10 chromosomes for the association analysis across four environments by FarmCPU. The dotted red line indicates the significance threshold of P‐value 1 × 10⁻⁴. The significant SNPs are labelled with red dots. ⑤, distribution of SNP markers on 10 chromosomes in association pool, the colour represents the density of the SNP markers; ①–④, represent different environments: ①, 2016 Jinghong; ②, 2016 Hongya; ③, 2016 Ya'an; ④, BLUP. Stable‐effect SNPs co‐detected in a multi‐environment are shown in orange rectangle‐shaped boxes.

Figure 3

QTL on 10 chromosomes for three kernel size traits across seven environments. Circos graph displaying integrated QTL on 10 chromosomes for kernel size traits across seven environments. The innermost ring black strip represents bin markers on 10 chromosomes of maize; the outermost ring scale indicates the physical location on the 10 chromosomes. Seven circles from the outside to the inside represent seven environments, E1b–E7b, respectively. For each environment, different colours represent different traits: green, KL; blue, KW; red, KT. The fan‐shaped region between the two markers (highlighted with the black line) represents each pleiotropic QTL. The stable QTL co‐detected in a multi‐environment are shown in orange rectangle‐shaped boxes.

Figure 4

Expression of zma‐miR164e results in the failure of seed formation in Arabidopsis. (a) Morphological characterization of expressing zma‐miR164e (OE) transgenic plants in Arabidopsis thaliana. Representative plants from T1‐positive transgenic lines are shown: T‐1, T‐2 and T‐3. Wild type (WT) as the control. Bar = 2 cm. (b) The flowers and pods of the OE and WT of Arabidopsis. For OE, flower has no petals (d and e), pod has no seed (f); however, flower (a and b) and pod (c) of WT were normal development. Bar = 400um. (c) The mature pod of OE was significantly smaller than that of WT. Bar = 400 μm. (d) RT‐PCR analysis of the expression levels of zma‐miR164e in the mixed samples of inflorescence and buds of WT and the transgenic Arabidopsis plants. zma‐miR164e was no expression in WT, but expressed in the transgenic plants T‐1, T‐2 and T‐3. U6 was used as loading control. (e) The phenotypic values of pod length (mm). The pod length of the transgenic Arabidopsis plants was significantly shorter than WT (error bars indicate ± SD. t‐test; ***, P < 0.001). (f) qRT‐PCR analysis of CUC1, CUC2 and NAC6 in the mixed samples of inflorescence and buds of WT and the transgenic Arabidopsis plants. ß‐Tubulin was used as internal reference to calculate the relative expression of target genes using the formula log10[2^{−(Ct target gene − Ct ß‐Tubulin)}]. qRT‐PCR data represent the average of three biological replicates. (Error bars indicate ± SD, t‐test; ***P < 0.001; **P < 0.01).

Figure 5

Zma‐miR164e‐directed cleavege in Arabidopsis CUC1 mRNA decreases the accumulation of the CUC1 protein. (a) eGFP:CUC1 (OD _600 nm = 0.6) was transiently expressed alone or co‐expressed with ath‐miR164e (OD _600 nm = 0.3/0.6/0.9) in tobacco leaf cells. The eGFP:CUC1 protein accumulation decreased with increasing ath‐miR164a concentrations. The result served as positive control for this experiment. (b) Binding sites of ath‐miR164a and CUC1. (c) The eGFP:CUC1 (OD _600 nm = 0.6) was transiently expressed alone or co‐expressed with zma‐miR164e (OD _600 nm = 0.3/0.6/0.9) in tobacco leaf cells. The eGFP:CUC1 protein accumulation decreased with increasing zma‐miR164e concentrations. d, Binding sites of zma‐miR164e and CUC1. (e) That zma‐miR164e cannot suppress the protein accumulation of eGFP:CUC1m whose binding site sequence has synonymous mutations. (f) synonymously mutated sequence of binding sites in CUC1. (h) and (i) eGFP intensity change of eGFP:CUC1 and eGFP:CUC1m with the increase in zma‐miR164e concentration. (g) eGFP intensity change of eGFP:CUC1 with increasing ath‐miR164a concentrations. The data of each sample were the average of the randomly detected 10 nuclei. Same results were obtained in three independent experiments. (Error bars indicate ± SD, t‐test; ***P < 0.001).