Dissecting repulsion linkage in the dwarfing gene Dw3 region for sorghum plant height provides insights into heterosis

Abstract

Heterosis is a main contributor to yield increase in many crop species. Different mechanisms have been proposed for heterosis: dominance, overdominance, epistasis, epigenetics, and protein metabolite changes. However, only limited examples of molecular dissection and validation of these mechanisms are available. Here, we present an example of discovery and validation of heterosis generated by a combination of repulsion linkage and dominance. Using a recombinant inbred line population, a separate quantitative trait locus (QTL) for plant height (qHT7.1) was identified near the genomic region harboring the known auxin transporter Dw3 gene. With two loci having repulsion linkage between two inbreds, heterosis in the hybrid can appear as a single locus with an overdominance mode of inheritance (i.e., pseudo-overdominance). Individually, alleles conferring taller plant height exhibited complete dominance over alleles conferring shorter height. Detailed analyses of different height components demonstrated that qHT7.1 affects both the upper and lower parts of the plant, whereas Dw3 affects only the part below the flag leaf. Computer simulations show that repulsion linkage could influence QTL detection and estimation of effect in segregating populations. Guided by findings in linkage mapping, a genome-wide association study of plant height with a sorghum diversity panel pinpointed genomic regions underlying the trait variation, including Dw1, Dw2, Dw3, Dw4, and qHT7.1. Multilocus mixed model analysis confirmed the advantage of complex trait dissection using an integrated approach. Besides identifying a specific genetic example of heterosis, our research indicated that integrated molecular dissection of complex traits in different population types can enable plant breeders to fine tune the breeding process for crop production.

Keywords: genome-wide association studies; heterosis; plant height; pseudo-overdominance; repulsion linkage.

Fig. 1.

Plant height measurements and crossing design. (A) Total plant height, height at the base of the panicle, height at the flag leaf, and preflag leaf height were measured directly (solid arrows). The flag leaf-to-apex interval, flag leaf-to-base of panicle interval, preflag leaf-to-flag leaf interval, and preflag leaf-to-base of panicle interval were calculated (dotted arrows). qHT7.1 affects all eight plant height components, whereas the effect of Dw3 is detected for six components. (B) Crossing design for hybrid development. Eight inbreds were selected by their genotypes at qHT7.1 (AA or aa), Dw3 (BB or bb), and Dw1 (CC or cc) to make the crosses. The four on the left have the dw1dw1 background, and the four on the right have the Dw1Dw1 background. Two specific crosses (black lines) are expected to generate F₁s showing heterosis in plant height. For all other crosses (gray lines), the F₁ is expected to have a height similar to that of the second inbred parent indicated by the arrow.

Fig. S1.

Distributions of plant height across five environments [Puerto Rico 2011 (PR11), Puerto Rico 2012 (PR12), Kansas 2011 (KS11), Kansas 2012 (KS12), and Iowa 2013 (IA13)] and the distribution of BLUP values based on the five environments (Combined). The red line is the plant height of Tx430, and the blue line is the plant height of P898012.

Fig. S2.

Histograms of plant height BLUP values in five environments and plant height BLUP values by combing all five environments.

Fig. S3.

Linkage map built with 1,756 SNP markers. The positions of the three plant height QTL are indicated by the blue bars.

Fig. 2.

Linkage mapping of plant height. (A–C, Left) Diagrams defining Results from composite interval mapping for total plant height (A), Flag leaf height (B), and Flag leaf-to-apex interval (C). (Center and Right) Results from composite interval mapping. (Center, Upper) The logarithm of the odds (LOD) score profile with the permutation threshold indicated by the horizontal line. (Center, Lower) The additive effect (a) with the Tx430 allele as the reference. (Right, Upper) The LOD score profile for enlarged chromosome 7 region. (Right, Lower) The additive effect for enlarged chromosome 7 region.

Fig. S4.

Mapping results for eight components of plant height. For each trait, the upper two panels are results from composite interval mapping, and the lower two panels are results from a single-marker scan. (Left) Results across 10 chromosomes. (Right) Results for chromosome 7. a, additive effect; LOD, logarithm of the odds.

Fig. S4.

Mapping results for eight components of plant height. For each trait, the upper two panels are results from composite interval mapping, and the lower two panels are results from a single-marker scan. (Left) Results across 10 chromosomes. (Right) Results for chromosome 7. a, additive effect; LOD, logarithm of the odds.

Fig. S4.

Mapping results for eight components of plant height. For each trait, the upper two panels are results from composite interval mapping, and the lower two panels are results from a single-marker scan. (Left) Results across 10 chromosomes. (Right) Results for chromosome 7. a, additive effect; LOD, logarithm of the odds.

Fig. S4.

Mapping results for eight components of plant height. For each trait, the upper two panels are results from composite interval mapping, and the lower two panels are results from a single-marker scan. (Left) Results across 10 chromosomes. (Right) Results for chromosome 7. a, additive effect; LOD, logarithm of the odds.

Fig. S5.

Gel image of the two Dw3 alleles. L, 1 Kb Plus DNA Ladder (Life Technologies). Lanes 1–4, PCR fragments from Tx430, P898012, RIL88 (shortest entry), and RIL197 (tallest entry), respectively. The fragments from Tx430 and RIL88 are 882 bp longer than the fragments from P898012 and RIL197.

Fig. 3.

Genome-wide association mapping of plant height. (A) Initial genome scan. (B) Second genome scan with a model including Dw1 and Dw2 as covariates. (C) Genome scan using the MLMM. (Left) Genome-wide results. (Right) Enlarged chromosome 7 region. The horizontal line in each subsection is the Bonferroni-corrected significance threshold. The positions of Dw1, Dw2, Dw3, qHT7.1, and the possible position for Dw4 are indicated. The red points in C are the significant SNPs (entering the final model as covariates) identified by MLMM.

Fig. 4.

Plant height of parental lines and corresponding F₁ hybrids. (A and B) Heterosis caused by repulsion linkage of qHT7.1 and Dw3, under different backgrounds of Dw1. (C and D) The tall allele of qHT7.1 shows complete dominance over the short allele of qHT7.1 under different backgrounds of Dw3 and Dw1. In each picture, the seed parent is on the left, and the pollen parent is on the right.

Fig. 5.

Effect of repulsion linkage on QTL detection in a simulated F₂ population with 200 individuals. (A–C) Heritability of 0.1. (D–F) Heritability of 0.3. (B and E) The d/a ratios when one QTL was detected. (C and F) The d/a ratios when two QTL were detected. For all d/a ratios, each point is the median value from 5,000 runs, and missing points indicate that those scenarios were not observed.

Fig. S6.

Effect of repulsion linkage on QTL detection in a simulated F₂ population with 500 individuals. (A–C) Heritability of 0.1. (D–F) Heritability of 0.3. (B and E) The d/a ratios when one QTL was detected. (C and F) The d/a ratios when two QTL were detected. Each point is the median value from 5,000 runs, and missing points indicate that those scenarios were not observed.

Fig. 6.

Standardized plant height for the 16 genotype combinations in the sorghum association panel ordered by their genotypes at Dw1, Dw2, Dw3, and qHT7.1. The width of each box in the boxplot is proportional to the square root of the number of accessions in each group. Orange rectangles represent homozygous dominant genotypes, and blue rectangles represent homozygous recessive genotypes. No accessions were found in the seventh genotype combination.