Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2015 Jun 4;161(6):1453-67.
doi: 10.1016/j.cell.2015.04.053.

A Unique Gene Regulatory Network Resets the Human Germline Epigenome for Development

Affiliations
Free PMC article

A Unique Gene Regulatory Network Resets the Human Germline Epigenome for Development

Walfred W C Tang et al. Cell. .
Free PMC article

Abstract

Resetting of the epigenome in human primordial germ cells (hPGCs) is critical for development. We show that the transcriptional program of hPGCs is distinct from that in mice, with co-expression of somatic specifiers and naive pluripotency genes TFCP2L1 and KLF4. This unique gene regulatory network, established by SOX17 and BLIMP1, drives comprehensive germline DNA demethylation by repressing DNA methylation pathways and activating TET-mediated hydroxymethylation. Base-resolution methylome analysis reveals progressive DNA demethylation to basal levels in week 5-7 in vivo hPGCs. Concurrently, hPGCs undergo chromatin reorganization, X reactivation, and imprint erasure. Despite global hypomethylation, evolutionarily young and potentially hazardous retroelements, like SVA, remain methylated. Remarkably, some loci associated with metabolic and neurological disorders are also resistant to DNA demethylation, revealing potential for transgenerational epigenetic inheritance that may have phenotypic consequences. We provide comprehensive insight on early human germline transcriptional network and epigenetic reprogramming that subsequently impacts human development and disease.

Figures

None
Figure 1
Figure 1
Developmental Timeline and Isolation of a Pure Population of hPGCs (A) Developmental timelines of human and mouse PGCs based on embryological landmarks of germ cell development. Notable epigenetic changes in mPGCs are depicted as colored bars. Blue arrow line indicates developmental ages of human embryos covered in the current study. (B) Isolation of hPGCs from Wk7 female embryonic gonads by FACS with cell-surface markers TNAP and c-KIT. Mesonephros is used as a negative control. The purity of hPGCs was tested by alkaline phosphatase staining (bottom). See also Figure S1.
Figure 2
Figure 2
RNA-Seq Reveals Unique Transcriptional States of hPGCs (A) Hierarchical clustering of gene expression profiles. Biological replicates of Wk5.5–Wk9 male (M) and female (F) hPGCs, gonadal somatic cells (soma), and conventional H9 ESCs are shown. Note that only one Wk5.5 hPGC sample was available for RNA-seq. (B) Principal component analysis (PCA) of gene expression in hPGC samples. Arrow line indicates developmental progression along PC2 and PC1. (C) Heatmap showing mean expression of representative genes in human samples. Differentially expressed genes between day 4 wild-type (WT) and BLIMP1 mutant hPGCLCs [log2(fold change)>2, p < 0.05] are highlighted. Note that mutant cells lack BLIMP1 protein as determined by immunofluorescence, but frame-shifted mutant transcripts are detected. “Endo,” endoderm; “Meso,” mesoderm; “TE,” trophectoderm. (D) Expression of key genes in human and mouse PGCs. Mean expression in biological replicates of hPGCLCs, Wk7–Wk9 hPGCs (gonadal), E7.5 (early), and E11.5–12.5 (gonadal) mPGCs are shown. (E) Immunofluorescence of TFCP2L1, KLF4, and TEAD4 on Wk7 female genital ridge cryosections. hPGCs are counterstained by BLIMP1 or SOX17. Scale bars, 20 μm. (F) Schematic illustrating the unique transcriptome of hPGCs. (G) Volcano plot showing differentially expressed genes between day 4 BLIMP1 mutant and wild-type (WT) hPGCLCs [log2(fold change)>2, p < 0.05]. See also Figure S2 and Table S1.
Figure 3
Figure 3
DNA Demethylation and Chromatin Reorganization in the Human Germline (A) 5mC and 5hmC levels at ICRs of H19 and GNAS ICRs and promoters of DAZL and DDX4 in hPGCLCs and surrounding soma determined by Glu-qPCR. Day (D) 0 represents 4i ESCs. Data are represented as mean ± SEM of two biological replicates. (B and D) Immunofluorescence analysis for (B) 5-methylcytosine (5mC) and (D) 5-hydroxymethylcytosine (5hmC) on human embryo cryosections. hPGCs are counterstained by TFAP2C or OCT4. Scale bars, 20 μm. (C) Fluorescence intensity of indicated epigenetic modifications in hPGCs and surrounding soma (corresponds to images in Figures 3B, 3D, 3H–3J, S3E, and S3F). Around 20–800 hPGCs and >200 soma are used for quantification at each time point. (p < 0.01 between hPGC and soma; Wilcoxon signed-rank test.) (E) Immunofluorescence analysis for UHRF1, TET1, and TET2 on genital ridge. Arrowheads indicate Ki-67-positive (proliferating) hPGCs, which are UHRF1 negative. Scale bars, 20 μm. (F) Fluorescence intensity of epigenetic modifiers in Wk7 female hPGCs and surrounding soma (corresponds to images in Figures 3E, S3A, and S3B). Sample sizes are indicated. (p < 0.0001; Wilcoxon signed-rank test.) (G) Expression of epigenetic modifiers in biological replicates of day 4 wild-type (WT) and BLIMP1 mutant hPGCLCs by RNA-seq. (H–J) Immunofluorescence analysis for (H) H3K27me3, (I) H3K9me2, (J) H3K9me3, and HP1α/ MacroH2A2. Arrowheads in (H) indicate H3K27me3 foci in somatic cells. Scale bars, 20 μm. (K) Magnified immunofluorescence images (corresponds to yellow boxes in Figures 3B, 3D, and 3J) showing hPGCs with enrichment of 5mC, H3K9me3, and MacroH2A2 at chromocenters (arrowheads), which is not the case for 5hmC. Scale bars, 5 μm. See also Figure S3.
Figure 4
Figure 4
Comprehensive DNA Demethylation in hPGCs Revealed by Base-Resolution BS-Seq (A) Violin plots showing distribution of CpG methylation levels in overlapped 1 kb genomic tiles of conventional H9 ESCs, Wk7 female gonadal somatic cells (soma), Wk5.5–Wk9 female and male hPGCs, sperm, and ICM. Common tiles with a minimum of 5 CpGs and at least 20% of the total CpGs covered by at least 5× in each sample are considered. These thresholds were applied to all subsequent methylation analyses unless stated otherwise. Due to low coverage, ICM only has ∼42% of common tiles fulfilling the above criteria. Black point indicates median. (B) Averaged CpG methylation level profiles of all genes from 5 kb upstream (−) of transcription start sites (TSSs), through scaled gene bodies to 5 kb downstream (+) of transcription end sites (TES). (C) Density plots illustrating DNA methylation dynamics of 1 kb tiles between indicated pairs of samples. Color intensity indicates tile counts within each bin, whereas blue regression lines show trends of methylation changes. (D) K-means clustering of repeat-free tiles into seven dynamic groups. Gray tiles of ICM do not pass coverage thresholds and are therefore not shown. Tiles in cluster 6 retain partial methylation in hPGCs. (E) Violin plots showing distribution of CpG methylation levels in human and mouse PGCs at conserved repeat-free 1 kb tiles. (F) Unsupervised hierarchical clustering of methylation levels of conserved repeat-free 1 kb tiles in human and mouse PGCs. See also Figure S4 and Table S2.
Figure 5
Figure 5
Imprint Erasure and Regulation of Gene Expression by Promoter Methylation (A) DNA demethylation dynamics of ICRs in hPGCs. Published MII oocyte RRBS data are included. Gray boxes indicate ICRs that do not pass the minimum coverage thresholds. (B) DNA demethylation dynamics of CGI-containing X chromosome promoters that are partially methylated (30%–70% methylation) in both ESCs and female gonadal somatic cells. Color key is shared with (A). (C) Scatterplot of differential gene expression and difference in promoter methylation between Wk7 female hPGCs and ESCs. Genes with >20% promoter methylation and log2(read counts) > 3 in either samples are shown. Colored points represent differentially expressed genes (log2(fold change)>2 and p < 0.05). Genes upregulated in hPGCs with promoter demethylation are highlighted in the purple box. (D) GO biological processes and SMART protein domain enrichment of genes highlighted in purple box of (C). (E) Promoter methylation and expression levels of representative germ-cell-related and KRAB-ZFP genes. (F) UCSC genome browser screenshots of CpG methylation at promoter of representative genes. Each vertical line represents one CpG site (≥5×). (G) Scatterplot of differential gene expression and difference in promoter methylation of KRAB-ZFPs between Wk7 hPGCs and ESCs. KRAB-ZFPs with >20% promoter methylation and log2(read counts) > 3 in either samples are shown. Colored points represent differentially expressed genes. (H) Unsupervised hierarchical clustering of KRAB-ZFPs expressions. Purple box indicates KRAB-ZFPs highly expressed in hPGCs. Asterisks indicate genes with promoter demethylation in hPGCs compared to ESCs. See also Figure S5 and Table S3.
Figure 6
Figure 6
Demethylation and Transcriptional Dynamics of Retrotransposons (A) Violin plots showing distribution of CpG methylation in major human repetitive elements classes and families. Common repeat loci with a minimum of 5 CpGs (≥5× coverage) in each sample are used. This threshold is used for all repeat methylation analyses. (B) Average DNA methylation and expression of notable human repeat families. For each family, color keys from blue to red indicate evolutionarily older to younger retrotransposons. (C) Scatterplot of differential expression and difference in methylation levels of potentially active retrotransposon loci between Wk9 female hPGCs and ESCs. Red points indicate differentially expressed repeat loci [log2(fold change)>2, p < 0.05 and log2(read counts) > 1], and blue points indicate repeat loci with low expression [log2(read counts < 0)]. Only loci that are close to full length are shown (AluY and AluYa-Yk > 268 bp; L1HS > 6,000 bp; SVA > 1,600 bp). See also Figure S6.
Figure 7
Figure 7
DNA Demethylation Escapees as Candidates for Transgenerational Epigenetic Inheritance (A) Genomic feature distribution of repeat-poor (<10% overlap with repeats) and repeat-rich (≥10% overlap) escapees. (B) Enrichment analysis of tissue-specific expression (UniProt UP_TISSUE) of 2,092 protein-coding genes with escapees in their gene bodies. (C) Human diseases and traits associated with genes with repeat-poor and repeat-free escapees in their gene bodies. (D) Enrichment of epigenetic modifications in repeat-poor escapees. (E) Enrichment of ZFP57 motif among repeat-poor escapees and methylated retrotransposons (with >700 copies). (F) K-means clustering of DNA methylation levels of Wk7-9 hPGCs common repeat-poor escapees across germline development. (G) UCSC genome browser screenshots of two representative genes with escapees (repeat free). Escapee region of SRRM2 is conserved between human and mouse, whereas that of CSNK1D overlaps with CGI at an alternative promoter. (H) Venn diagram showing overlap of homologous genes with repeat-free escapees in their gene bodies between human and mouse. (I) Schematic showing dynamics of preimplantation and germline epigenetic reprogramming in humans. hPGCs undergo the most comprehensive wave of global DNA demethylation, which reaches a minimum of ∼5% CpG methylation at weeks 7–9. Some single copy and repeat loci remain methylated and are candidates for transgenerational epigenetic inheritance. Dotted line indicates postulated methylation dynamics. See also Figure S7.
Figure S1
Figure S1
Isolation of hPGCs by FACS for RNA-Seq and BS-Seq, Related to Figure 1 (A) Morphology of Wk5.5-Wk9 human genital ridges. The ridge first appears as a thin coelomic epithelium (yellow arrow) adjacent to the mesonephros (white arrowhead) at Wk5.5 and subsequently develops into distinct genital ridges (Wk6-9). Sexual differentiation begins at Wk6 and the morphological differences of male (M) and female (F) ridges become apparent at Wk8-9. Scale bars, 1 mm. (B) Isolation of hPGCs from Wk5.5 to Wk9 genital ridges by FACS with cell surface markers TNAP and c-KIT. TNAP-high and c-KIT-high population was collected for alkaline phosphatase staining and consistently showed > 97% of AP positive hPGCs (red). (C) Details of in vivo and in vitro samples collected for RNA-Seq and BS-seq. Asterisks indicate RNA-Seq samples published in a previous study (Irie et al., 2015). (D) Number of hPGCs sorted per embryo at Wk5.5 to Wk9.
Figure S2
Figure S2
Transcriptional Dynamics of Human and Mouse PGCs, Related to Figure 2 (A) Two-dimensional PCA plot (PC2 against PC1) and gene loading plot. Arrow line indicates developmental progression of hPGCs from Wk5.5 to Wk9. (B) Gene ontology biological processes enrichment heatmap of upregulated genes [log2(fold change)>2, p < 0.05] in hPGCs in comparison to indicated samples. (C) Correlation of transcription profiles of human and mouse PGCs of various stages. The comparison included hPGCs, hPGCLCs (Irie et al., 2015) and mPGCs of various stages (Magnúsdóttir et al., 2013). Only genes expressed in both human and mouse cells (log2(read counts) > 3) but absent in soma were used for analysis. Expression levels were averaged over biological replicates. (D) Co-expression network analysis of human and mouse PGCs. Grey indicates co-expressed genes; red indicates genes upregulated in humans; blue indicates genes upregulated in mice. (E) K-means clustering of individual HERVH loci RNA expression into 6 dynamics.
Figure S3
Figure S3
Global Epigenetic Dynamics in hPGCs, Related to Figure 3 (A–C) Immunofluorescence analysis for (A) UHRF1 and DNMT1; (B) DNMT3A and DNMT3B and (C) TET1 on human embryo cryosections. Arrowheads in (A) indicate examples of Ki-67-positive (proliferating) hPGCs which are UHRF1-negative. Conventional ESCs [4th row in (B)], which strongly expresses DNMT3B, is used as a positive control in contrast to the absence of DNMT3B signals in Wk7 F ridges. Scale bars, 20 μm. (D) Expression of epigenetic modifiers in various human samples by RNA-Seq. (E and F) Immunofluorescence analysis for (E) H3K9me2; (F) H3K9me3 and HP1α/ MacroH2A2 on human embryo cryosections. Scale bars, 20 μm.
Figure S4
Figure S4
Comprehensive DNA Demethylation Revealed by Base-Resolution BS-Seq, Related to Figure 4 (A) Median CpG methylation levels of tiles in individual biological replicates of Wk5.5-Wk9 female and male hPGCs, ESC and soma. For soma and hPGCs, each replicate refers to cells collected from one individual embryo. Uncommon tiles with a minimum of 5 CpGs and at least 20% of the total CpGs covered by at least 5 reads in each sample are used. (B) Distribution of CpG methylation levels in CpG islands (CGI), low-, intermediate-, and high-CpG density promoters (LCP, ICP and HCP respectively), exons, introns, intergenic regions and active enhancers of selected somatic tissues of NIH Roadmap Epigenomics Project. (C) Distribution of methylation levels in tiles with various CpG densities. Note that tiles of Wk7-Wk9 hPGCs are globally demethylated regardless of CpG density. (D) Distribution of methylation levels of repeat-containing and repeat-free 1 kb tiles. Note that a larger proportion of repeat-containing tiles retains partial methylation compared to repeat-free tiles.
Figure S5
Figure S5
X Chromosome Demethylation and Regulation of Gene Expression by Promoter Methylation, Related to Figure 5 (A) UCSC genome browser screenshots of CpG methylation levels over whole chromosome 1 (left panel) and chromosome X (right panel) (B) Hexagonal bin plots showing promoter demethylation in hPGCs is globally uncoupled with gene expression. Differential gene expression are plotted against difference in promoter methylation between Wk7 female hPGCs and ESC (left panel) or gonadal somatic cells (right panel). Color intensity indicates genes counts within each bin. (C) Scatter plot of differential gene expression and difference in promoter methylation of KRAB-ZFPs between soma and ESCs. Interpro KRAB-ZFPs with > 20% promoter methylation and log2(read counts)>3 in either samples are shown. Colored points represent differentially expressed genes (log2(fold change)>2 and p < 0.05). (D) Venn diagram showing overlap of upregulated KRAB-ZFPs in Wk7 F hPGCs or soma (refer to Figures 5G and S5C) against ESCs.
Figure S6
Figure S6
Demethylation and Transcriptional Dynamics of Retrotransposons, Related to Figure 6 (A) Pie charts showing distribution of major repeat subfamilies that retain CpG methylation (> 30%) in the indicated samples. For hPGCs, only repeat loci that are commonly methylated in week 7-9 are considered. Value under each pie chart indicates the number and percentage of repeats that are methylated. Note that methylated L1, ERV1 and SVA repeats are overrepresented in hPGCs. (B) Distribution of RNA expression levels of major repeat families. Note that no global upregulation of repeats is observed in hPGCs upon genome-wide demethylation, except for SVA which showed a larger fraction of expressed loci at Wk7 and Wk9. (C) Correlation of SVA subfamilies expression and methylation. “r” represents Pearson correlation coefficient. Note that moderate negative correlation of methylation and expression is observed in week 7 female hPGC. Linear regression is illustrated as blue line.
Figure S7
Figure S7
DNA Demethylation Escapees Are Candidates of Epigenetic Inheritance, Related to Figure 7 (A) Distribution of retrotransposon subfamilies in repeat-rich escapees. Note that there is a bias toward the most abundant and active subfamilies, including AluY and L1PAs. (B) UCSC genome browser screenshots of two representative escapees in hPGCs. A region with retained DNA methylation of DSPP is conserved between human and mouse and intersects with low complexity simple repeats. TACC2 has an escapee region across its promoter, which is not conserved in the mouse. (C) DAVID enrichment analysis of GO biological processes and KEGG pathways of 2,092 protein-coding genes with repeat-poor escapees in their gene bodies. Top-enriched terms are shown for each category, redundant GO terms were removed. (D) Scatter plot of inter-individual DNA methylation variation and disease association of repeat-poor escapees. Shown are 2,892 (out of 7,071) high-confidence repeat-poor escapees that have at least 40% DNA methylation in hPGCs of one of the individuals, at least 10 CpGs and at least 20% of the total CpGs in the region covered by 5X in all four individuals. Pearson correlation coefficients of escapee methylation levels between individuals are indicated. (E) Coefficient of variation of methylation levels for repeat-rich, repeat-poor and enhancers-associated escapees, which have at least 40% DNA methylation in one of the individuals, at least 10 CpGs and at least 20% of the total CpGs in the region covered by 5X in all four individuals. (F) Transcription factor binding sites enrichment of repeat-poor escapees. NCBI GEO accession number of each dataset are shown. (G) DAVID enrichment analysis of SMART protein domains of 2,092 protein-coding genes with repeat-poor escapees in their gene bodies. (H) Enrichment of ZFP98 motif among repeat-poor escapees and methylated retrotransposons (with > 700 copies). (I) DNA methylation dynamics in gametes, blastocyst and hPGCs of repeat-poor escapees that intersect with annotated enhancers and CGIs. Methylation levels of CGI and enhancer escapees were in general amplified in gametes, especially for enhancer regions, which were fully re-methylated in sperm. However, these regions showed very similar partial methylation levels in blastocyst following resetting of DNA methylation in early embryos.

Comment in

Similar articles

See all similar articles

Cited by 133 articles

See all "Cited by" articles

References

    1. Bendsen E., Byskov A.G., Andersen C.Y., Westergaard L.G. Number of germ cells and somatic cells in human fetal ovaries during the first weeks after sex differentiation. Hum. Reprod. 2006;21:30–35. - PubMed
    1. Burns K.H., Boeke J.D. Human transposon tectonics. Cell. 2012;149:740–752. - PMC - PubMed
    1. Chuva de Sousa Lopes S.M., Hayashi K., Shovlin T.C., Mifsud W., Surani M.A., McLaren A. X chromosome activity in mouse XX primordial germ cells. PLoS Genet. 2008;4:e30. - PMC - PubMed
    1. Dawlaty M.M., Breiling A., Le T., Raddatz G., Barrasa M.I., Cheng A.W., Gao Q., Powell B.E., Li Z., Xu M. Combined deficiency of Tet1 and Tet2 causes epigenetic abnormalities but is compatible with postnatal development. Dev. Cell. 2013;24:310–323. - PMC - PubMed
    1. Gkountela S., Li Z., Vincent J.J., Zhang K.X., Chen A., Pellegrini M., Clark A.T. The ontogeny of cKIT+ human primordial germ cells proves to be a resource for human germ line reprogramming, imprint erasure and in vitro differentiation. Nat. Cell Biol. 2013;15:113–122. - PMC - PubMed

Publication types

Feedback