Integrating QTL mapping with transcriptome analysis mined candidate genes of growth stages in castor (Ricinus communis L.)

Abstract

Background: The growth stages largely determine the crop yield, while little is known about their genetic mechanisms in castor. In this study, the QTL mapping and candidate gene mining of growth stages were conducted using populations F₂ and BC₁, combining with differential expression analysis and weighted gene co-expression network analysis (WGCNA). The traits studied included the emergence date (ED), the budding date of primary spike (PSBD), the flowering date of primary spike (PSFD), the maturation date of primary spike (PSMD), and the maturation date of primary branch spike (PBSMD).

Results: A total of 20 QTLs conferring four traits (ED, PSBD, PSFD and PBSMD) were identified in the F₂ population, with a phenotypic variation explained (PVE) of single QTL ranged from 0.24 to 25.46%. Five QTLs underlying PSMD and PBSMD were identified in the BC₁ population, with a PVE of single QTL ranged from 4.74 to 10.82%. To our surprise, almost all the identified QTLs were clustered within two marker intervals, the RCM1769-RCM1838 on linkage group 6 and RCM950-RCM917 on linkage group 3. Subsequently, 473 open reading frames (ORFs) were searched out within these two clusters and 110 differentially expressed genes (DEGs) were screened out from these ORFs by the comparative transcriptome clean data (a total of 140.86 G) at the budding date, the initial flowering date and the full flowering date between parental racemes. With these DEGs, five distinct gene co-expression modules were generated using WGCNA. Showing significant differential expression between parents, four candidate genes (LOC8261128, LOC8278994, LOC8281165 and LOC8259049) in module MEturquoise, were recognized and were annotated as RcSYN3, RcNTT, RcGG3 and RcSAUR76 respectively. This finding implies their potential role in regulating the growth stages of castor.

Conclusion: In this study, numerous QTLs conferring growth stages were detected and four candidate genes were mined, which need to be functionally validated. The results provided a new insight into the genetic structure of ED, PSBD, PSFD, PSMD and PBSMD, offered the candidate genes and molecular markers for their improvement as well in castor.

Keywords: Castor (Ricinus communis L.); Differential expression analysis; Growth stages; QTL analysis; WGCNA.

Fig. 1

Frequency distribution and correlation analysis in populations F₂ (a) and BC₁ (b). For correlation analysis, reflecting the strength of the correlation by the size of the shaded area and are colored from red (coefficient = 1) to cyan (coefficient = -1); * and ** refer to significant and extremely significant correlation respectively

Fig. 2

Distribution map of QTLs conferring growth stages in castor. For physical mapping, the black lines are the CIM mapping results and the red lines are that in ICIM

Fig. 3

Differential gene expression analysis. (a) Up and down regulated DEGs in racemes of both parents at different stages within two QTL clusters; (b) Venn diagram of union DEGs in racemes of both parents within two QTL clusters

Fig. 4

WGCNA of the DEGs expression matrix. (a) Gene-based co-expression network analysis dendrogram; (b) The correlation heat map between five modules and three stages. Each row represents a module labeled with the same color as in (a), the number in each cell represents the correlation coefficient and shows with color, the p-value of the corresponding module-trait is exhibited in parentheses; (c) The expression pattern diagram of DEGs in MEturquoise module. The upper part shows the clustering heatmap of DEGs, with high expression in red and low expression in green, and the lower part shows the expression patterns of DEGs in different assay samples

Fig. 5

Differential expression of candidate genes controlling growth stages in 9048 and 16–201. (a) Correlation between qRT-PCR and RNA-seq data; (b-e) Relative expression levels of 4 predicted candidate genes. One-way ANOVA was performed, and ** indicates significance level at 0.01