Dynamic transcriptome landscape of maize embryo and endosperm development

Abstract

Maize (Zea mays) is an excellent cereal model for research on seed development because of its relatively large size for both embryo and endosperm. Despite the importance of seed in agriculture, the genome-wide transcriptome pattern throughout seed development has not been well characterized. Using high-throughput RNA sequencing, we developed a spatiotemporal transcriptome atlas of B73 maize seed development based on 53 samples from fertilization to maturity for embryo, endosperm, and whole seed tissues. A total of 26,105 genes were found to be involved in programming seed development, including 1,614 transcription factors. Global comparisons of gene expression highlighted the fundamental transcriptomic reprogramming and the phases of development. Coexpression analysis provided further insight into the dynamic reprogramming of the transcriptome by revealing functional transitions during maturation. Combined with the published nonseed high-throughput RNA sequencing data, we identified 91 transcription factors and 1,167 other seed-specific genes, which should help elucidate key mechanisms and regulatory networks that underlie seed development. In addition, correlation of gene expression with the pattern of DNA methylation revealed that hypomethylation of the gene body region should be an important factor for the expressional activation of seed-specific genes, especially for extremely highly expressed genes such as zeins. This study provides a valuable resource for understanding the genetic control of seed development of monocotyledon plants.

Figure 1.

Overview of the time series maize seed samples used for RNA-seq analysis. The photographs show the changes in maize embryo, endosperm, and whole seed during development. The 53 samples shown here were used to generate RNA-seq libraries. Bar = 5 mm.

Figure 2.

Analysis of global gene expression among different samples. A, Venn diagram of the 26,105 genes detected among embryo, endosperm, and whole seed. B, Number of genes expressed in each of the samples. C, Comparison of expression levels of genes detected in embryo and endosperm tissues.

Figure 3.

Global transcriptome relationships among different stages and tissues. A, PCA of the RNA-seq data for the 53 seed samples shows five distinct groups: I for embryo (light red), II for endosperm (light blue), and III to V for early (III), middle (IV), and late (V) whole seed (light purple). B to D, Cluster dendrogram showing global transcriptome relationships among time series samples of embryo (B), endosperm (C), and whole seed (D). The y axis measures the degree of variance (see the “Materials and Methods”). The bottom row indicates the developmental phases according to the cluster dendrogram of the time series data. au, Approximately unbiased.

Figure 4.

Coexpression modules. A to C, Expression patterns of coexpression modules of embryo (A), endosperm (B), and whole seed (C), ordered according to the sample time points of their peak expression. For each gene, the RPKM value normalized by the maximum value of all RPKM values of the gene over all time points is shown.

Figure 5.

Distribution and enrichment of genes, tissue-specific genes, TFs, and tissue-specific TFs in coexpression modules of embryo and endosperm. A and B, Bars indicate the percentage of all detected genes (green), tissue-specific genes (blue), TFs (red), and tissue-specific TFs (purple) observed in a coexpression module (C) or in the development phase (Total) relative to the total number of each group detected across samples for embryo (A) and endosperm (B). The number of genes represented by the percentage is shown on the right y axis. Enrichment for tissue-specific genes and TFs was evaluated with Fisher’s exact test based on the number of genes observed in each coexpression module, whereas enrichment for tissue-specific TFs was evaluated based on the number of TFs observed in each coexpression module. Asterisks represent significant enrichment at a false discovery rate ≤ 0.05.

Figure 6.

Analysis of highly expressed genes in the endosperm. A, The distribution of the 100 most highly expressed genes in the endosperm ordered by mean expression in different modules. B, The dynamic transcript levels of different zein gene family members in the endosperm as reflected by their percentage among all detected gene transcript levels. C, Heat map showing RPKM values of 35 zein genes in the different development stages of the endosperm. +, Having intact coding regions; −, with premature_stop; N, no; Y, yes.