Genome-Wide Association Reveals Trait Loci for Seed Glucosinolate Accumulation in Indian Mustard (Brassica juncea L.)

Abstract

Glucosinolates (GSLs) are sulphur- and nitrogen-containing secondary metabolites implicated in the fitness of Brassicaceae and appreciated for their pungency and health-conferring properties. In Indian mustard (Brassica juncea L.), GSL content and composition are seed-quality-determining traits affecting its economic value. Depending on the end use, i.e., condiment or oil, different GSL levels constitute breeding targets. The genetic control of GSL accumulation in Indian mustard, however, is poorly understood, and current knowledge of GSL biosynthesis and regulation is largely based on Arabidopsis thaliana. A genome-wide association study was carried out to dissect the genetic architecture of total GSL content and the content of two major GSLs, sinigrin and gluconapin, in a diverse panel of 158 Indian mustard lines, which broadly grouped into a South Asia cluster and outside-South-Asia cluster. Using 14,125 single-nucleotide polymorphisms (SNPs) as genotyping input, seven distinct significant associations were discovered for total GSL content, eight associations for sinigrin content and 19 for gluconapin. Close homologues of known GSL structural and regulatory genes were identified as candidate genes in proximity to peak SNPs. Our results provide a comprehensive map of the genetic control of GLS biosynthesis in Indian mustard, including priority targets for further investigation and molecular marker development.

Keywords: Brassica juncea; genome-wide association studies; glucosinolates (GSL); seed quality.

Figure 1

(a) Cluster analysis based on genetic distance using an UPGMA tree with branches coloured based on geographical origin: India and Pakistan and rest of South Asia (blue), rest of Asia, Europe, North America, Africa and Australia (green) and unknown origin (yellow). (b) Population structure as depicted by a sorted bar plot of ancestry proportions for K = 2–4, inferred with ADMIXTURE. (c) Ten-fold cross-validation error of ADMIXTURE analyses of K = 1 to 12. (d) Principal component analysis (PCA) coloured based on cluster assignment (threshold of 70% membership probability) at K = 2 in ADMIXTURE. Orange triangles used for cluster 1, purple squares for cluster 2 and green dots for admixture cluster.

Figure 2

Distribution of raw mean values of (a) total GSLs, (b) gluconapin and (c) sinigrin, reflecting the ADMIXTURE cluster assignment at K = 2 of each accession (orange for cluster 1, purple for cluster 2 and green for admixture cluster). Correlations using log-transformed values of (d) gluconapin and total GSLs and (e) sinigrin and total GSLs in the full diversity panel. Correlation using log-transformed values of (f) gluconapin and total GSLs in cluster 1 and (g) sinigrin and total GSLs in cluster 2. Orange used triangles for cluster 1, purple squares for cluster 2 and green dots for admixture cluster.

Figure 3

Manhattan and q-q plots for the GWAS of (a) total GSLs, (b) sinigrin and (c) gluconapin using BLINK (purple dots) and FarmCPU (orange dots) models. The horizontal line represents significance threshold at 5% after Bonferroni multiple test correction (−log10 (p) = 5.45).