Contribution of epigenetic landscapes and transcription factors to X-chromosome reactivation in the inner cell mass

Abstract

X-chromosome inactivation is established during early development. In mice, transcriptional repression of the paternal X-chromosome (Xp) and enrichment in epigenetic marks such as H3K27me3 is achieved by the early blastocyst stage. X-chromosome inactivation is then reversed in the inner cell mass. The mechanisms underlying Xp reactivation remain enigmatic. Using in vivo single-cell approaches (allele-specific RNAseq, nascent RNA-fluorescent in situ hybridization and immunofluorescence), we show here that different genes are reactivated at different stages, with more slowly reactivated genes tending to be enriched in H3meK27. We further show that in UTX H3K27 histone demethylase mutant embryos, these genes are even more slowly reactivated, suggesting that these genes carry an epigenetic memory that may be actively lost. On the other hand, expression of rapidly reactivated genes may be driven by transcription factors. Thus, some X-linked genes have minimal epigenetic memory in the inner cell mass, whereas others may require active erasure of chromatin marks.

Fig. 1

Xist RNA and H3K27me3 profiles in the ICM cells of early and mid blastocysts. a Examples of individual ICM of early (E3.5) and mid (E4.0) pre-implantation stage embryos (photographs right panel, scale bar 20 µm) analysed by immunolabelling with antibodies against H3K27 tri-methylation (red) combined with Xist RNA-FISH (green). For each stage, an intact ICM (scale bar, 10 μm) and an enlarged nucleus (scale bar, 5 μm) are shown (IF/RNA-FISH). The cells below the white line illustrate the cluster of cells that have lost Xist RNA coating and H3K27me3 enrichment on the Xp and are presumably the epiblast. b Proportion of ICM cells showing enrichment of H3K27me3 on the Xist RNA-coated X-chromosome in early and mid blastocyst stages are presented as mean (right panel). Below the graph, the total cell number analysed is indicated, followed by the number of female embryos analysed in brackets. ICM inner cell mass, RNA-FISH RNA-fluorescent in situ hybridization, IF immunofluorescence

Fig. 2

Xist RNA, X-linked gene expression and H3K27me3 profiles in the ICM cells of early to late blastocyst stage embryos. a Examples of individual ICM analysed by immunolabelling with antibodies against H3K27 tri-methylation (grey scale) and combined with RNA-FISH for Xist RNA (green) and primary transcription from the X-linked genes (red), together with representative nucleus are shown (scale bar, 10 μm). b Schematic representation of the X-chromosome showing the location of the loci analysed in a and c. Atp6ap2 gene is known to escape XCI in 60–80% of blastocyst cells and used as a control of the experiment. c Percentage (mean) of cells showing biallelic expression for X-linked genes in ICM of independent early (E3.5), mid (E4.0) and late (E4.5) blastocyst stage embryos. d Examples of individual ICM analysed by immunolabelling with NANOG (grey scale), combined with RNA-FISH for Xist (green) and X-linked genes (Atp7a and Kif4) (red) at early (E3.5) and mid (E3.75) blastocyst stage embryos. For each stage, an intact ICM (IF/RNA-FISH) and enlarged nuclei (white squares) are shown. Dotted lines indicate the position of NANOG-positive cells (scale bar, 10 μm). e Proportion (mean) of NANOG-positive ICM cells showing different Xist and X-linked gene expression patterns at early (E3.5) and mid (E3.75) blastocyst stage embryos. Below the graph, the total cell number analysed is indicated, followed by the total number of female embryos analysed in brackets. f Proportion (mean) of NANOG-negative ICM cells showing different Xist and X-linked gene expression patterns at early (E3.5) and mid (E3.75) blastocyst stage embryos. Below the graph, the total cell number analysed is indicated, followed by the total number of female embryos analysed in brackets

Fig. 3

Single-cell RNAseq reveals loss of heterogeneity in the E4.0 mid ICM compared to early E3.5 ICM. Principal component analysis (PCA) based on scRNAseq data from trophectoderm (E3.5), early (E3.5, 10–25 cells per ICM) and mid (E4.0, 20–40 cells per ICM). ICM cells on the 1000 most variable genes (a) and on published pluripotency and differentiation candidate genes (n = 23, list in d) (b). Different stages are designed by different colours. n = 14, 23 and 5 cells, respectively, for E3.5 ICM, E4.0 ICM and E3.5 TE (details of each single cell is listed in Supplementary Data 1). c Hierarchical clustering (top) and Pearson distance (bottom) of pluripotency and lineage genes (listed in d) expression variation in E3.5 and E4.0 single cells, based on Pearson’s correlation. Cells were clustered by lineage (TE, PrE and Epi), then by stage. n = 42 single-cell samples. d Level of expression of the 23 candidate genes involved in pluripotency and lineage differentiation in the 42 single-cell samples and used to classify cells according to their lineage are shown. Cells were ordered according to the hierarchical clustering in c. TE trophectoderm, PrE primitive endoderm, ICM inner cell mass, Epi epiblast

Fig. 4

Different X-linked gene reactivation behaviours in the ICM. The mean of allele-specific expression ratios for each informative and expressed X-linked gene in E3.5 (trophectoderm and ICM) and E4.0 (primitive endoderm and epiblast) female C57BL/6JxCAST/EiJ embryos are represented as heatmaps, with strictly maternal expression (ratio ≤0.15) in red and strictly paternal expression (ratio ≥0.85) in blue. Colour gradients are used in between these two values as shown in the key. Genes are ordered by genomic position (left) or by timing of reactivation (right). Further information is provided in Supplementary Data 2 and Methods section. Black, red and grey arrows are, respectively, highlighting example of early-, later-reactivated genes and escapees. As expected, Xist RNA is paternally expressed in the trophectoderm cells. Ogt and Yipf6 genes display similar paternal expression in the trophectoderm, escape imprinted XCI, and show random monoallelic expression and CAST/EiJ bias, respectively, (Supplementary Fig. 1). n = 116 genes, NA, data not available (below threshold)

Fig. 5

Link between Xist expression, epigenetic landscapes and Xp reactivation. a Anti-correlation is shown between the level of Xist expression and the number of biallelically/reactivated and informative X-linked genes in scRNAseq (Spearman correlation). Male E3.5 single cells have been added and used as control for Xist expression and X-linked gene parental expression. Genes with level of expression as (RPRT <1) are considered as non-expressed in our samples. b Enrichment of H3K27me3 and H3K4me3 on paternal X-chromosome obtained from Zheng et al. shows significant differences (by Wilcoxon test) between early and escapee reactivation-timing classes compared to late and very late. Low cell chromatin immunoprecipitation sequencing (ChIPseq) have been performed with ICM cells of pre-implantation embryos (pooled between E3.5–E4.0) after immunosurgery of the ICM. Activated genes show an excess of H3K4me3 and repressed ones an enrichment of H3K27me3. Xist is highlighted with an orange arrow. Early vs. late (p = 2.29 × 10⁻⁴ for H3K27me3 and p = 1.63 × 10⁻³ for H3K4me3) and very late (p = 2.51 × 10⁻² for H3K27me3 and p = 3.95 × 10⁻⁴ for H3K4me3) and escapees vs. late (p = 1.95 × 10⁻⁶ for H3K27me3 and p = 2.09 × 10⁻⁷ for H3K4me3) and very late (p = 7.33 × 10⁻³ for H3K27me3 and p = 6.73 × 10⁻⁸ for H3K4me3) by Wilcoxon test

Fig. 6

Representation of Xp reactivation and different X-linked gene behaviours. Scheme of imprinted XCI, followed by reactivation in the ICM of the blastocyst. Xp silencing is triggered by the long non-coding Xist RNA, followed by H3K27me3 recruitment. At the early blastocyst stage (E3.5), imprinted Xp is maintained in TE, when some genes are already showing reactivation in the ICM, independently of Xist (early-reactivated genes). Those early genes are lowly enriched in H3K27me3 marks and highly enriched in H3K4me3 on their paternal allele compared to the later reactivated ones. A few hours later, when ICM cells are divided into PrE and Epi cells, Xp reactivation appears to be nearly complete only in the future epiblast cells, (based on loss of Xist coating and H3K27me3). In PrE, some early-reactivated genes have already been silenced again. This suggests a fluctuation of early-reactivated genes with different epigenetic memory requirements between early- and late-reactivated genes. TE trophectoderm, PrE primitive endoderm, ICM inner cell mass, Epi epiblast

Fig. 7

UTX is required for proper reactivation of late-reactivated genes. a Conditional Utx allele: FD = Flp and Dre-recombined conditional allele. Recombination of the third exon of Utx by Cre expression gives a knockout FDC allele. b Individual Utx control and mutant female ICMs analysed by immunolabelling with H3K27me3 (grey), combined with Xist (green) and Kif4 (red) RNA-FISH in late blastocysts (scale bar 20 µm). Cells below the white line illustrate the cluster of cells that have lost Xist coating and H3K27me3 enrichment on the Xp and are presumably the epiblast. Enlarged nuclei are shown for not reactivated (1, 3) and reactivated cells (2, 4) (scale bar 5 µm). c Mean of ICM cells showing enrichment of H3K27me3 on the Xist RNA-coated X-chromosome from E4.5 control (Utx ^FD/FD), heterozygous (Utx ^FDC/+ ) and mutant (Utx ^FDC/FDC) female blastocysts. Below the graph, the total cell number analysed is indicated, followed by the number of females in brackets. p value <0.0057 (**), <0.0018 (**) and <0.021 (*) between control and heterozygous vs. mutant, respectively, for H3K27me3-Xist-negative cells, H3K27me3-Xist-positive cells and H3K27me3-positive, Xist-negative cells, by two-sided Dunn’s test. d Example of Utx mutant female ICM analysed by immunolabelling with H3K27me3 (grey), combined with Xist RNA (green) at late (E4.5) blastocyst stage. Red and blue arrows pointed cells, respectively, with both Xist and H3K27me3 enrichment and only H3K27me3 enrichment on the Xp (scale bar 10 µm). e Percentage (mean) of cells showing biallelic expression for X-linked genes in ICM of E4.5 control (Utx ^FD/FD ) and Utx mutant (Utx ^FDC/FDC) embryos. Kif4, Rnf12 and Pdha1 are late- or very late-reactivated genes, when Abcb7 and Atrx are early-reactivated genes (Supplementary Data 2). f Control (Utx ^FD/FD ) and mutant (Utx ^FDC/FDC) E4.5 blastocysts analysed by immunofluorescence against NANOG (green) and GATA6 (red). DAPI is in dark blue (scale bar 20 µm). Percentage of positive cells for Nanog, Gata6 or both have been summarized as the mean in the graph and total cell number analysed is indicated below, followed by the number of females in brackets. Non-significant (n.s.) by Kruskal–Wallis test. FDC Flp, Dre and Cre-recombined knockout allele, MW Mann–Whitney non-parametric test