Parental methylome reprogramming in human uniparental blastocysts reveals germline memory transition

Abstract

Uniparental embryos derived from only the mother (gynogenetic [GG]) or the father (androgenetic [AG]) are unique models for studying genomic imprinting and parental contributions to embryonic development. Human parthenogenetic embryos can be obtained following artificial activation of unfertilized oocytes, but the production of AG embryos by injection of two sperm into one denucleated oocyte leads to an extra centriole, resulting in multipolar spindles, abnormal cell division, and developmental defects. Here, we improved androgenote production by transferring the male pronucleus from one zygote into another haploid androgenote to prevent extra centrioles and successfully generated human diploid AG embryos capable of developing into blastocysts with an identifiable inner cell mass (ICM) and trophectoderm (TE). The GG embryos were also generated. The zygotic genome was successfully activated in both the AG and GG embryos. DNA methylome analysis showed that the GG blastocysts partially retain the oocyte transcription-dependent methylation pattern, whereas the AG blastocyst methylome showed more extensive demethylation. The methylation states of most known imprinted differentially methylated regions (DMRs) were recapitulated in the AG and GG blastocysts. Novel candidate imprinted DMRs were also identified. The production of uniparental human embryos followed by transcriptome and methylome analysis is valuable for identifying parental contributions and epigenome memory transitions during early human development.

Figure 1.

Generation of human diploid AG and GG embryos. (A) Schematic illustration of human diploid Bi-P, GG, and AG embryo generation. Briefly, following ICSI, two oocytes were fertilized. Four to six hours later, female pronuclei were identified based on their proximity to the second (2nd) polar bodies and smaller size compared with male pronuclei. The female pronucleus from individual zygotes was removed to generate haploid androgenotes. The male pronucleus from the donor androgenote was then extracted and mixed with Sendai virus and transferred into the recipient androgenote, leading to the formation of a diploid androgenote for subsequent development into a blastocyst. GG embryos were produced with the same method. (B) The development of Bi-P and uniparental embryos to blastocysts after pronuclei transfer. The uniparental embryos developed blastocysts with identifiable TE and ICM, the same as the Bi-P embryos in vitro. (C) The development ratios of Bi-P, GG, and AG embryos.

Figure 2.

The gene expression of uniparental blastocysts. (A) Clustering RNA-seq of human GV, MI, MII, two-cell, eight-cell, diploid AG/GG/Bi-P, and primed hES cells. (B) Plots showing the oocyte-specific (left) and ZGA-related (right) gene expression levels in Bi-P, AG, and GG blastocysts. The criteria of the “oocyte-specific genes” were oocyte-specific genes identified by selecting those expressed at the oocyte stage (FPKM ≥ 5) but not expressed or expressed at low levels in the embryo stage (FPKM ≤ 1) in both mRNA-seq and total RNA-seq. On the other hand, only the genes that were expressed (FPKM ≥ 5) in the post-ZGA stages but not expressed in the oocyte (FPKM ≤ 1) were identified as ZGA genes. (C) Bar plot showing the expression levels of the marker genes CDX2, GATA6, POU5F1, and NANOG in MII oocytes and Bi-P (two-cell eight-cell and blastocysts), AG, and GG blastocysts. (D) Venn diagrams show the DEGs in the uniparental versus biparental groups. Only the genes expressed at one stage with FPKM ≥ 5 and at least twofold changes between two groups and with a P-value generated by DESeq2 (Love et al. 2014) less than 0.05 were selected as DEGs. (E) Venn diagram showing the comparison of DEGs in our study and previously published studies (Zhang et al. 2019). Only the genes expressed at one stage with FPKM ≥ 5 and at least twofold changes between AG and GG and with a P-value generated by DESeq2 (Love et al. 2014) less than 0.05 were selected as DEGs in our study. (Bi-P) Biparental embryo, (AG) androgenetic, (GG) gynogenetic.

Figure 3.

DNA methylome of diploid AG, GG, and Bi-P blastocysts. (A) Global CpG methylation in the AG, Bi-P, GG embryos, and gametes. Note that the AG blastocysts exhibited the lowest level of global DNA methylation, whereas the sperm showed the highest level, as expected. The sperm and oocyte DNA methylation data were used in the Okae et al. (2014) study. (B) Principal component analysis showed a genome-wide DNA methylation relationship among three types of blastocysts and germ cells. (C) Nonsupervised cluster analysis of the DNA methylome from different blastocysts and gametes. (D) The metaplot showing DNA methylation around gene bodies in the AG, Bi-P, GG blastocysts, and gametes. (E) The relationship between TSS (left) and gene body (right) DNA methylation and gene expression of the Bi-P blastocysts and the uniparental blastocysts. (F) Heatmaps showing DEGs in the AG and GG groups and associated gene promoter methylation levels. Only the genes expressed at one stage with FPKM ≥ 5 and at least twofold changes between two groups and with a P-value generated by DESeq2 (Love et al. 2014) less than 0.05 were selected as DEGs. (Bi-P) Biparental embryo, (AG) androgenetic, (GG) gynogenetic.

Figure 4.

DNA methylation of known imprinted regions in the AG, GG, and Bi-P blastocysts. (A) UCSC snapshots showing the previously known DMRs located in PEG3, SNURF, IGF2, and H19. Our results showed that germline DMRs survived demethylation during early embryo development, indicating that our diploid AG and GG blastocysts are suitable models to screen for putative DMRs. The sperm and oocyte DNA methylation data are from Okae et al. (2014). (B) Heatmaps showing the DNA methylation levels of known imprinted DMRs and the related gene expression. DNA methylation in two-, four-, and eight-cell embryos published previously was included as a control (Leng et al. 2019). All replicates for the AG, GG, and Bi-P blastocysts were pooled. These known imprinted DMRs were clarified as previously reported (Okae et al. 2014). (Bi-P) Biparental embryo, (AG) androgenetic, (GG) gynogenetic.

Figure 5.

Putative germline regions with differential DNA methylation. (A) UCSC snapshots showing the novel DMRs. (B) Heatmaps showing the DNA methylation levels for newly identified DMRs and the related gene expression. DNA methylation in two-, four-, and eight-cell embryos published previously was included as a control. The DMRs were firstly identified based on a pairwise comparison between AG and GG blastocysts as previously described (Zhang et al. 2018). Only those DMRs with changes in CG methylation levels between sperm and oocyte greater than 0.1 (e.g., for maternal DMRs, the oocyte showed higher methylation levels than the sperm) were taken as allelic putative DMRs (Methods). One hundred five sperm-specific DMRs and 506 oocyte-specific DMRs were detected. (C) Correlation between mCG difference and nearest gene expression difference between AG and GG blastocysts. (D) GO analysis for the nearest genes of maternal and paternal DMRs.