The genetic architecture of amino acids dissection by association and linkage analysis in maize

Abstract

Amino acids are both constituents of proteins, providing the essential nutrition for humans and animals, and signalling molecules regulating the growth and development of plants. Most cultivars of maize are deficient in essential amino acids such as lysine and tryptophan. Here, we measured the levels of 17 different total amino acids, and created 48 derived traits in mature kernels from a maize diversity inbred collection and three recombinant inbred line (RIL) populations. By GWAS, 247 and 281 significant loci were identified in two different environments, 5.1 and 4.4 loci for each trait, explaining 7.44% and 7.90% phenotypic variation for each locus in average, respectively. By linkage mapping, 89, 150 and 165 QTLs were identified in B73/By804, Kui3/B77 and Zong3/Yu87-1 RIL populations, 2.0, 2.7 and 2.8 QTLs for each trait, explaining 13.6%, 16.4% and 21.4% phenotypic variation for each QTL in average, respectively. It implies that the genetic architecture of amino acids is relative simple and controlled by limited loci. About 43.2% of the loci identified by GWAS were verified by expression QTL, and 17 loci overlapped with mapped QTLs in the three RIL populations. GRMZM2G015534, GRMZM2G143008 and one QTL were further validated using molecular approaches. The amino acid biosynthetic and catabolic pathways were reconstructed on the basis of candidate genes proposed in this study. Our results provide insights into the genetic basis of amino acid biosynthesis in maize kernels and may facilitate marker-based breeding for quality protein maize.

Keywords: GWAS; Quality Protein Maize (QPM); amino acid; co-expression network; linkage mapping; metabolism.

Figure 1

Functional category annotations for 308 candidate genes and their respective percentages identified via GWAS as significantly associated with amino acid traits in maize kernels.

Figure 2

Chromosomal distribution of amino acids loci and QTLs identified in this study. QTL regions (represented by the confidence interval for linkage mapping and the 100 kb up‐ and downstream of the lead SNP for association mapping) across the maize genome responsible for amino acid levels from the different populations are shown as midnight blue (BB), green (AM1), cyan (AM2) gold (KB) and red (ZY) boxes, respectively. The class represents different amino acid families. AT, pyruvate‐derived amino acid family related traits; ATT, aspartate‐derived amino acid family related traits; BCAA, branched‐chain amino acid family related traits; GT, glutamate‐derived amino acid family related traits; PT, phenylalanine‐derived amino acid family related traits; ST, serine‐derived amino acid family related traits; His, histidine family related traits. The x‐axis indicates the genetic positions across the maize genome in Mb. Heatmap under the x‐axis illustrates the density of amino acid loci and QTLs across the genome. The red arrows show the QTL hotspots. The detailed information of all detected loci and QTLs is shown in Tables S3 and S4. Amino acid traits from different derived families are marked by distinct colours as shown on the right.

Figure 3

A maize amino acids network involving key genes identified in this study by GWAS. The different colours represent the different amino acids families. The purple, sky‐blue, red, brown, dark green, orange lines represent the metabolism pathway of pyruvate‐derived, glutamate‐derived, aspartate‐derived, serine‐derived, Histidine, phenylalanine‐derived amino acids, respectively. The blue lines represent the TCA cycle. Candidate genes identified in this study by GWAS are shown in the respective pathway. KARI, Ketol‐acid reductoisomerase; GHMT, Glycine hydroxymethyltransferase; SPT, Serine palmitoyltransferase; PGDH, Phosphoglycerate dehydrogenase; IDH, Isocitrate dehydrogenase; TS, Tryptophan synthase; TGTA, Tryptophan Glutamate transaminase; DSOR, Disulphide oxidoreductase; MODH, 3‐methyl‐2‐oxobutanoate dehydrogenase; IVD, Isovaleryl‐CoA dehydrogenase; DHDPR, Dihydrodipicolinate reductase; TD, L‐threonine 3‐dehydrogenase; AS, Asparagine synthase; AI, Asparaginase; SK, Shikimate kinase; PAL, Phenylalanine ammonia‐lyase; GAL, Glutamate‐ammonia ligase; NNAT, Nicotianamine aminotransferase; ACO, Aconitate hydratase; SSADH, Succinate semialdehyde dehydrogenase; DHQS, 3‐dehydroquinate synthase; TyrDC, Tyrosine decarboxylase; PRA‐CH, Phosphoribosyl‐AMP cyclohydrolase.

Figure 4

A co‐expression network of the amino acids metabolism. The red nodes represent the 14 candidate genes from GWAS. The yellow nodes represent the co‐expressed genes overlapping with candidate genes of GWAS. The green nodes represent that genes directly or indirectly related to amino acids metabolism. The blue nodes represent the transcription factors. 1, GRMZM2G147191; 2, GRMZM2G009808; 3, GRMZM2G119482; 4, GRMZM2G178826; 5, GRMZM2G010202; 6, GRMZM5G829778; 7, GRMZM2G081886; 8, GRMZM2G090241; 9, GRMZM2G082214; 10, GRMZM2G161868; 11, GRMZM2G169593; 12, GRMZM2G006480; 13, GRMZM2G127308; 14, GRMZM2G036464.

Figure 5

GWAS for Lys/Total with significant SNP‐trait association in this study. (a) Manhattan plot displaying the GWAS result of the Lys/Total level. (b) Regional association plot for locus O2. The SNPs in the promoter and gene body of O2 were shown in red. (c) Gene structure of O2. Filled black boxes represent exons, and filled white ones represent UTRs. (d) A representation of the pairwise r ² value among all polymorphic sites in O2, where the colour of each box corresponds to the r ² value according to the legend. (e) Manhattan plot shows the association between expression level of O2 and genomewide SNPs. Significant signals are mapped to SNPs within O2, indicating a cis transcriptional regulation of this gene. (f) Plot of correlation between the Lys/Total level (red) and the normalized expression level (sky blue) of the O2. The r value is based on a Pearson correlation coefficient. The P value is calculated using the Student's‐t test. (g) Box plot for Lys/Total level (red) and expression of O2 (sky blue).

Figure 6

Validation of association analysis using QTL Interval and progeny test. (a) LOD curves of QTL mapping for level of Lys/Total in maize kernels on chromosome 7. (b) Bin map of a heterogeneous inbred family with a heterozygous region on chromosome 7. (c) Progeny test using four progeny families derived from the residual heterozygous line. (d) Scatterplot of association results between SNPs in the confidence interval and the level of Lys/Total. Association analysis was performed using the mixed linear model controlling for the population structure (Q) and kinship (K). (e) The candidate genes of 400 kb in the confidence interval. G1 to G5 represent GRMZM2G138727, GRMZM2G565441, GRMZM2G138976, GRMZM5G873335 and GRMZM2G446625, respectively. *** and *** indicate significant correction between the Lys/Total and the normalized expression levels of candidate genes at P < 0.01 and P < 0.001. (f) Box plot for Lys/Total (red) and expression of GRMZM2G138727 (skyblue) based on duplication (D) and no duplication (ND). (g) Box plot for Lys/Total (red) and expression of GRMZM2G138727 (skyblue) based on B73 (GAT) and By804 (TAT) like haplotype. (h) Box plot for Lys/Total and expression of GRMZM2G138727 based on B73 (GAT) and By804 (TAT) like haplotype within no duplication (ND).

Figure 7

GWAS for Leu/Total with significant SNP‐trait association in this study. (a) Manhattan plot displaying the GWAS result of the Leu/Total level. (b) Regional association plot for locus Acetolactate synthase (GRMZM2G143008). (c) Gene structure of ALS. (d) Manhattan plot shows the association between expression level of GRMZM2G143008 and genomewide SNPs. (e) Plot of correlation between the Leu/Total level (red) and the normalized expression level of the Acetolactate synthase gene (skyblue). The r value is based on a Pearson correlation coefficient. The P value is calculated using the Student's‐t test. (f) The relative expression of GRMZM2G143008 in transgenic and non‐transgenic plants. ZH11 was DNA as the positive control, and plasmid was the over‐expression construct as the negative control. (g) Bar plot for amino acid traits in rice transgenic lines relative to wide type.