Linkage and association analyses reveal that hub genes in energy-flow and lipid biosynthesis pathways form a cluster in upland cotton

Abstract

Upland cotton is an important allotetraploid crop that provides both natural fiber for the textile industry and edible vegetable oil for the food or feed industry. To better understand the genetic mechanism that regulates the biosynthesis of storage oil in cottonseed, we identified the genes harbored in the major quantitative trait loci/nucleotides (QTLs/QTNs) of kernel oil content (KOC) in cottonseed via both multiple linkage analyses and genome-wide association studies (GWAS). In 'CCRI70' RILs, six stable QTLs were simultaneously identified by linkage analysis of CHIP and SLAF-seq strategies. In '0-153' RILs, eight stable QTLs were detected by consensus linkage analysis integrating multiple strategies. In the natural panel, thirteen and eight loci were associated across multiple environments with two algorithms of GWAS. Within the confidence interval of a major common QTL on chromosome 3, six genes were identified as participating in the interaction network highly correlated with cottonseed KOC. Further observations of gene differential expression showed that four of the genes, LtnD, PGK, LPLAT1, and PAH2, formed hub genes and two of them, FER and RAV1, formed the key genes in the interaction network. Sequence variations in the coding regions of LtnD, FER, PGK, LPLAT1, and PAH2 genes may support their regulatory effects on oil accumulation in mature cottonseed. Taken together, clustering of the hub genes in the lipid biosynthesis interaction network provides new insights to understanding the mechanism of fatty acid biosynthesis and TAG assembly and to further genetic improvement projects for the KOC in cottonseeds.

Keywords: Gene cluster; Hub gene; Interaction network; Kernel oil content; Quantitative trait loci; Upland cotton.

Fig. 1

Phenotypic statistics of kernel oil content (KOC) of CCRI70 RILs and their parental lines, and correlations of the natural panel and CCRI70 RILs across different environments. A. Phenotypic performance of KOC of CCRI70 RILs (including highest, lowest, and average values) and their parental lines. B. Correlation analysis of the natural panel across five environments. C. Correlation analysis of CCRI70 RILs across 14 environments and their BLUE values. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Distribution of quantitative trait loci (QTLs) across the genome of cotton and the additive effects of stable QTLs. A. Distribution of QTLs of CCRI70 RILs identified by the CHIP strategy. B. Distribution of QTLs of CCRI70 RILs identified by the SLAF-seq strategy. C. Distribution of QTLs of 0–153 RILs. D. Additive effects of stable QTLs of CCRI70 RILs identified by the CHIP strategy. E. Additive effects of stable QTLs of CCRI70 RILs identified by the SLAF-seq strategy. F. Additive effects of stable QTLs of 0–153 RILs. The red bars indicate the number of QTLs that are stably epressed in at least three environments, and the green bars indicate the number of QTLs that are only expressed in one or two environments. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Manhattan map and the quantile–quantile (QQ) chart in the genome-wide association studies (GWAS) of the natural panel. A. Manhattan map of GLM model. B. QQ chart of GLM model. C. Manhattan map of MLM model. D. QQ chart of MLM model.

Fig. 4

Bioinformatics studies of the candidate genes harbored in stable QTLs. A. The gene numbers harbored in the stable QTLs in the CHIP and SLAF-seq strategies of CCRI70 RILs. B. Comparison of the genes in the stable QTLs of CCRI70 RILs, 0–153 RILs, and the natural panel. C. gene ontology annotation (GO) term enrichment analysis of the candidate genes. D. The first 20 significantly enriched GO terms. E. The primary and secondary level annotaions of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway of the candidate genes. F. The first 20 significantly enriched KEGG pathways. G. The number of genes related to oil metabolism in GO term enrichment and KEGG pathway annotation after referring to the genes related to oil metabolism in Arabidopsis.

Fig. 5

Analysis of the protein interaction network of candidate genes under the molecular complexity detection (MCODE) algorithm.

Fig. 6

Candidate gene identification in a major quantative trait locus (QTL) on chromosome 3 (A03). A. The genetic position and significance of the common QTL identified both in CCRI70 and 0–153 RILs. B. Physical position of the marker interval of common QTL. C. Haplotypes of the overlapping region of common QTL, where the hub genes were identified. D. Arrangement of the hub genes on the physical maps of the different Gossypium genome assemblies. E. SNP differences identified in the hub genes between sGK156 and 901–001, the two parental lines of CCRI70 RILs. F. SNPs identified in the hub genes in the natural panel. G. The gene expression levels (FPKM in log2(FPKM + 1)) in different organs, tissues, and developmental stages of fiber and ovule.

Fig. 7

A working model of the hub genes LPLAT1 and PAH2 in the Kennedy pathway. This model was modified based on Li-Beisson et al. , Wang et al. , Zhao et al. , and Zhu et al. .