Elsa Matthus, Lin-Bo Wu, Yoshiaki Ueda, Stefanie Höller, Mathias Becker, Michael Frei

Theoretical and Applied Genetics, October 2015, Volume 128, Issue 10, pp 2085-2098

http://link.springer.com/article/10.1007/s00122-015-2569-y

Key message

A genome-wide association study in rice yielded loci and candidate genes associated with tolerance to iron toxicity, and revealed biochemical mechanisms associated with tolerance in contrasting haplotypes.

Abstract

Iron toxicity is a major nutrient disorder affecting rice. Therefore, understanding the genetic and physiological mechanisms associated with iron toxicity tolerance is crucial in adaptive breeding and biofortification. We conducted a genome-wide association study (GWAS) by exposing a population of 329 accessions representing all subgroups of rice to ferrous iron stress (1000 ppm, 5 days). Expression patterns and sequence polymorphisms of candidate genes were investigated, and physiological hypotheses related to candidate loci were tested using a subset of contrasting haplotypes. Both iron including and excluding tolerant genotypes were observed, and shoot iron concentrations explained around 15.5 % of the variation in foliar symptom formation. GWAS for seven traits yielded 20 SNP markers exceeding a significance threshold of −log₁₀ P > 4.0, which represented 18 distinct loci. One locus mapped for foliar symptom formation on chromosome 1 contained two putative glutathione-S-transferases, which were strongly expressed under iron stress and showed sequence polymorphisms in complete linkage disequilibrium with the most significant SNP. Contrasting haplotypes for this locus showed significant differences in dehydroascorbate reductase activity, which affected the plants’ redox status under iron stress. We conclude that maintaining foliar redox homeostasis under iron stress represented an important tolerance mechanism associated with a locus identified through GWAS.

Fig. 3  Association mapping results for shoot Fe concentration after 5 days of Fe treatment (1000 ppm). (A) Frequency distribution of Fe con-centrations in the association panel. (B) q–q plots comparing expected and observed Pvalues for marker–trait associations. (C) Manhattan plot displaying P values for each marker according to mixed linear model analysis. Blue dotsindicate the top 50 SNPs (−log10P values ranging from 2.80 to 4.15) and red dots indicate SNP exceeding the significance level of −log10P > 4.0. (D) Linkage block analysis of the candi-date locus on chromosome 2; a 0.63 Mb region spanning from 34.21 to 34.84 Mb is shown; triangles framed in black indicate linkage blocks determined as stated in “Materials and methods”. (E) Average Fe concentrations of lines representing different alleles at the highly significantly (−log10 P > 4.0) associated SNP marker (id2015632). Data bars show mean values and standard errors; bars not sharing the same letterare significantly different at P < 0.05 by t test (color figure online).