Diverse epigenetic mechanisms maintain parental imprints within the embryonic and extraembryonic lineages

Abstract

Genomic imprinting and X chromosome inactivation (XCI) require epigenetic mechanisms to encode allele-specific expression, but how these specific tasks are accomplished at single loci or across chromosomal scales remains incompletely understood. Here, we systematically disrupt essential epigenetic pathways within polymorphic embryos in order to examine canonical and non-canonical genomic imprinting as well as XCI. We find that DNA methylation and Polycomb group repressors are indispensable for autosomal imprinting, albeit at distinct gene sets. Moreover, the extraembryonic ectoderm relies on a broader spectrum of imprinting mechanisms, including non-canonical targeting of maternal endogenous retrovirus (ERV)-driven promoters by the H3K9 methyltransferase G9a. We further identify Polycomb-dependent and -independent gene clusters on the imprinted X chromosome, which appear to reflect distinct domains of Xist-mediated suppression. From our data, we assemble a comprehensive inventory of the epigenetic pathways that maintain parent-specific imprinting in eutherian mammals, including an expanded view of the placental lineage.

Keywords: DNA methylation; X chromosome; epigenetic regulators; imprinting; placenta; scRNA-seq.

Figure 1. An inventory of parental-specific expression from single mouse embryos

(A) Simplified schematic of our experimental system to obtain parent-specific expression landscapes (imprinting and X inactivation). E6.5 epiblast (light grey) and extraembryonic ectoderm (ExE, dark grey) are isolated from F1 reciprocal crosses (n = 11; 7 BDF1xCAST, 4m/3f and 4 CASTxBDF1, 1m/3f) and subjected to RNA-seq. Colors are used in figures throughout the manuscript to highlight the tissue of origin. (B-C) Imprinted genes identified in the E6.5 Epiblast and ExE (red: maternally expressed, blue: paternally expressed) using a median imprinted score cutoff and allelic ratios of 1 and 0.25, respectively (dashed lines). Corresponding heatmaps show the allelic ratios in the forward (BDF1xCAST) and reverse (CASTxBDF1) crosses. Allelic ratios are adjusted from an initial range of 0 to 1 such that 0 corresponds to equivalent expression between both alleles: 0 = Biallelic, 0.5 = 100% expressed from one allele (MAT (maternal), PAT (paternal), BDF1, or CAST). Previously uncharacterized imprinted genes are indicated in white, including maternal-specific expression of Brachyury (T) within the ExE. A more detailed explanation of the allelic ratio calculation is provided in the Methods. Asterisk indicates known non-canonical imprinting genes (see Table S1 sheet E) (D) Overlap of imprinted genes between E6.5 Epiblast and ExE (red: maternally expressed, blue: paternal expressed). (E) Scatter plot showing the allelic ratios for X-linked genes in ExE between forward (BDF1xCAST) and reverse (CASTxBDF1) crosses. Maternally expressed genes (red), XCI escaper genes (green), and strain-specific escape from the CAST (brown) and BDF1 (black) inactive X chromosome are indicated. The dashed line indicates the 0.25 allelic ratio threshold used to determine escaper genes. (F) Chromosomal overview of genes that maintain biallelic expression on the Xi (“escapers”) including their shared or strain-specific status. Asterisk indicates escaper genes that function as chromatin modifiers.

Figure 2. Epigenetic regulation of autosomal and X chromosome-specific imprinting

(A) Schematic overview for our strategy to assign roles for selected epigenetic regulators to parent-specific gene expression. Target epigenetic regulators are disrupted by injection of Cas9 and sgRNAs into hybrid F1 (BDF1xCAST) zygotes (ΔDnmt1 n = 3, ΔG9a-GLP n = 9, ΔRnf2 n = 8, ΔEed n = 10). E6.5 Epiblast (light grey) and extraembryonic ectoderm (ExE, dark grey) were collected from each selected embryo. Regulator colors are used throughout the rest of the manuscript: Dnmt1, black; G9a and GLP, orange, gene name Ehmt1 and Ehmt2; PRC1 member Rnf2, light turquoise; and PRC2 member Eed, turquoise. (B) Left: Violin plots of the median allelic ratios of autosomal genes from WT and regulator mutants in Epiblast. Maternal and paternal expressed imprinted genes are indicated with red and blue dots, respectively. Right: Venn diagram shows the intersection of each epigenetic regulators’ contribution to imprint status. An epigenetic regulator was counted as relevant for silencing imprinted genes if the change in allelic ratio between WT and KO was ≥ 20%. PRC1 and PRC2 were summarized as PRC by using the higher delta. In Epiblast, DNA methylation-dependent imprinting is most frequent. Regulator independent imprinted genes have a delta allelic ratio < 20% in all disrupted regulators (see Figure S3B). Imprinted genes with less than two informative allelic ratio values in any regulator disruption data set are not shown. (C) As in (B) for ExE. Extraembryonic imprints appear to depend on a more diverse set of regulators. (D) Left: Violin plots displaying the median allelic ratio of X-linked genes from WT and regulator mutants in female ExE. The blue dot highlights the allelic ratio of the lncRNA Xist. Successful extraembryonic XCI depends on PRC2 and, to a lesser degree, PRC1. Notably, paternal Xist expression is largely stable in all regulator mutants examined (Figure S2H). Right: Venn diagram for imprinted XCI as shown for autosomal imprinting in (B) and (C). An epigenetic regulator was counted as relevant for silencing X-linked genes if the delta allelic ratio change between WT and KO for any regulator was ≥ 20%. Regulator independent X-linked genes have a delta allelic ratio < 20% in all disrupted regulators.

Figure 3. G9a controls non-canonical imprinting at endogenous retrovirus containing promoters

(A) Allelic ratio of imprinted genes and the corresponding changes between wildtype (BDF1xCAST) and regulator disrupted embryos within the Epiblast (left) and ExE (right) lineage. Heatmap ranked by the allelic ratio change between WT and ΔDnmt1. For ExE, boxes highlight DNA methylation-independent, non-canonical expressed genes (Asterisk indicates previous description in the literature) and lncRNA directed PRC targets. Notably, ExE-specific imprinting is most apparent for a set of paternally-expressed, G9a-GLP controlled loci that only weakly depend on PRCs. Imprinted genes with less than two informative allelic ratio values in any experimental data set are not shown. (B) Flow diagram outlining changes in the imprinted landscape between Epiblast and ExE for maternally (top) and paternally (bottom) expressed genes. (C) Genome browser tracks for E6.5 ExE WGBS and RNA-seq data that cover two non-canonical G9a dependent imprinted clusters Jade1 (top) and Slc38a4 (bottom). Boxes highlight G9a-dependent hypomethylated DMRs (Overlapping ERV LTRs are indicated). (D) Identified ExE DMRs using WT and ΔG9a WGBS data (1kb window, n = 3,691, |delta cutoff| ≥ 20%). Pie chart showing the proportion of hypo- and hypermethylated DMRs (top). Feature enrichment of the identified DMRs over background was calculated for intergenic, genic (±1kb of TSS), and different repeat classes using the Fisher’s exact test (bottom).

Figure 4. Polycomb-based repression is critical to maintain the imprinted X chromosome

(A) Pie chart showing the proportion of regulator dependent or independent genes for imprinted XCI. A gene was called “regulator dependent” if the allelic ratio changes between WT and KO by ≥ 20% for any regulator. UpSet plot shows the intersection between the four disrupted regulator members: ΔDnmt1, ΔG9a-GLP, ΔRnf2, and ΔEed. The inset Venn diagram highlights the high overlap between ΔEed and ΔRnf2 dependent X-linked genes. Genes that escape imprinted XCI were excluded from this analysis. (B) Regulator-independent X-linked genes are organized into distinct spatial clusters. Allelic ratios of maternally expressed X-linked genes and the corresponding changes between wildtype (BDF1xCAST) and regulator disrupted ExE lineages are shown (X-linked gene matrix ranked by the genomic position). Regulator-independent X-linked genes are indicated. A gene was called “regulator independent” if the allelic ratio changes between WT and KO is < 20% for every regulator. (C) Regulator independent regions occur in domains with high Xist enrichment and repressive chromatin. 1Mb windows summarizing (average) paternal X expression changes between WT and PRC, for Xist enrichment over input (RAP) (Engreitz et al., 2013) and for H3K27me3 and H2AK119ub enrichment on the inactive X (Żylicz et al., 2019) across the entire X-chromosome.

Figure 5. Three distinct epigenetic mechanisms for controlling parent-specific gene regulation

(A) Illustration of identified mechanisms for parental-specific gene regulation: Imprinted control region (ICR), ICR-directed lncRNA deployment, and non-canonical G9a dominant. (B) Ideograms of mouse chromosomes show the position of epiblast-specific imprinted genes. Pie charts on the top highlight the proportion of the allelic ratio change between WT and ΔDnmt1, ΔG9a-GLP, and ΔPRC samples. The circle size denotes the combined delta change. The symbols below each imprinted region highlight the mechanisms defined in (A). (C) Ideograms as in (B) for the ExE lineage. Question marks (?) highlight speculative mechanisms as informed by our data and the literature. One region includes the solo imprinted gene Glant6, which seems to depend equally on all three regulators. The second region harbors the two maternally expressed genes (T and Pnldc1) that are in proximity to the Igf2r region and thus likely targets of the paternally expressed lncRNA Airn. (D) Summary of X-linked regulator dependent and independent genes, as well as genes that escape the process of imprinted XCI. The model on the right illustrates the mechanism for how the inactive X is maintained in a silent state by Xist and Polycomb.