Genome-wide association analysis reveals new targets for carotenoid biofortification in maize

Abstract

Genome-wide association analysis in CIMMYT's association panel revealed new favorable native genomic variations in/nearby important genes such as hydroxylases and CCD1 that have potential for carotenoid biofortification in maize. Genome-wide association studies (GWAS) have been used extensively to identify allelic variation for genes controlling important agronomic and nutritional traits in plants. Provitamin A (proVA) enhancing alleles of lycopene epsilon cyclase (LCYE) and β-carotene hydroxylase 1 (CRTRB1), previously identified through candidate-gene based GWAS, are currently used in CIMMYT's maize breeding program. The objective of this study was to identify genes or genomic regions controlling variation for carotenoid concentrations in grain for CIMMYT's carotenoid association mapping panel of 380 inbred maize lines, using high-density genome-wide platforms with ~476,000 SNP markers. Population structure effects were minimized by adjustments using principal components and kinship matrix with mixed models. Genome-wide linkage disequilibrium (LD) analysis indicated faster LD decay (3.9 kb; r (2) = 0.1) than commonly reported for temperate germplasm, and therefore the possibility of achieving higher mapping resolution with our mostly tropical diversity panel. GWAS for various carotenoids identified CRTRB1, LCYE and other key genes or genomic regions that govern rate-critical steps in the upstream pathway, such as DXS1, GGPS1, and GGPS2 that are known to play important roles in the accumulation of precursor isoprenoids as well as downstream genes HYD5, CCD1, and ZEP1, which are involved in hydroxylation and carotenoid degradation. SNPs at or near all of these regions were identified and may be useful target regions for carotenoid biofortification breeding efforts in maize; for example a genomic region on chromosome 2 explained ~16% of the phenotypic variance for β-carotene independently of CRTRB1, and a variant of CCD1 that resulted in reduced β-cryptoxanthin degradation was found in lines that have previously been observed to have low proVA degradation rates.

Fig. 1

Principal coordinate analysis plot based on the GBS + 55 K data, color-labeled based on the K-means clustering results. Black, red, and green color represent group 1 (tropical heterotic group B lines), 2 (provitamin A source lines), and 3 (tropical heterotic group A lines), respectively

Fig. 2

GWAS manhattan plots using the mixed linear (G+Q+K) model and the 55 K + GBS combined genotype data. QPM quality protein maize (binary phenotype), ZEA zeaxanthin, L:Z lutein:zeaxanthin ratio, BCX β-cryptoxanthin, BC1 β-carotene, BC2 β-carotene with the S10_135911532 marker as an additional covariate in the model. All carotenoids’ phenotypic values (y) were transformed to log₁₀(y + 1) prior to analyses