Inferring the choreography of parental genomes during fertilization from ultralarge-scale whole-transcriptome analysis

Abstract

Fertilization precisely choreographs parental genomes by using gamete-derived cellular factors and activating genome regulatory programs. However, the mechanism remains elusive owing to the technical difficulties of preparing large numbers of high-quality preimplantation cells. Here, we collected >14 × 10(4) high-quality mouse metaphase II oocytes and used these to establish detailed transcriptional profiles for four early embryo stages and parthenogenetic development. By combining these profiles with other public resources, we found evidence that gene silencing appeared to be mediated in part by noncoding RNAs and that this was a prerequisite for post-fertilization development. Notably, we identified 817 genes that were differentially expressed in embryos after fertilization compared with parthenotes. The regulation of these genes was distinctly different from those expressed in parthenotes, suggesting functional specialization of particular transcription factors prior to first cell cleavage. We identified five transcription factors that were potentially necessary for developmental progression: Foxd1, Nkx2-5, Sox18, Myod1, and Runx1. Our very large-scale whole-transcriptome profile of early mouse embryos yielded a novel and valuable resource for studies in developmental biology and stem cell research. The database is available at http://dbtmee.hgc.jp.

Keywords: development; fertilization; parthenogenesis; transcription factor; transcriptome.

Figure 1.

Systematic representation of transcriptome analysis and summary of mapped read distributions. (A) Developmental stages targeted by the RNA-seq assay. (B) Distribution of uniquely mapped reads along mouse genomic features. (C,D) Histograms showing the number of RefSeq exons and introns and the coverage levels. (E) Histogram showing the number of intronic ncRNAs (coverage >0.6) and their proportion among the total detected introns shown in D. Averaged numbers across replicates are shown. (Oo) Oocyte.

Figure 2.

Summary of transcriptome discovery. (A) The histogram shows the total number of mRNAs and ncRNAs discovered in the present study. (B) Cluster analysis for the major ncRNAs shows dramatic expression changes between 1C and 2C, except for miRNAs, whose small size precluded analysis. (C) ncRNAs are more actively transcribed than mRNAs. Relatively few RNAs are specifically transcribed in p1C and p4C embryos compared with normal development. (D) Our RNA-seq assay discovered 5364 mRNAs that are not detected with microarrays. (FPKM) Fragments per kilobase of exon per million mapped reads, a unit of digital gene expression level; (Oo) oocyte.

Figure 3.

Gene expression changes and significant parthenogenetic biological processes. (A) Differential gene expression in all of the possible pairs. (B–E) Representative GO:BP terms enriched in parthenotes (fold change > 1.5). (Oo) Oocyte.

Figure 4.

Temporal gene expression patterns and their functional associations. (A,B) Hierarchical clustering of normalized expression levels results in 25 expression patterns. (C) Representative GO terms enriched in each pattern (P < 0.01) showing that minor ZGA is involved in broad developmental processes. (Oo) Oocyte.

Figure 5.

Analysis of TF-binding motif enrichment. Heat map showing enriched TF-binding motifs in core promoters (P < 0.01). (Nsc) Nascent; (Spr) sperm.

Figure 6.

Inference of gene regulatory network. (A) Network including 53,229 directed edges among 1897 nodes (Supplemental Fig. S6 in high resolution). The edge direction is from 108 TFs (triangles) to target genes (circles). Dark-gray triangles represent TFs pooled at oocytes. The inset emphasizes the highest CC of non-TF genes in each category. (B) Overall distribution of CCs. (C) Distribution of targeted genes by 108 TFs. Color corresponds to the gene categories shown in Table 3. (Coeff.) Coefficient.