Transcription of MERVL retrotransposons is required for preimplantation embryo development

Abstract

Zygotic genome activation (ZGA) is a critical postfertilization step that promotes totipotency and allows different cell fates to emerge in the developing embryo. MERVL (murine endogenous retrovirus-L) is transiently upregulated at the two-cell stage during ZGA. Although MERVL expression is widely used as a marker of totipotency, the role of this retrotransposon in mouse embryogenesis remains elusive. Here, we show that full-length MERVL transcripts, but not encoded retroviral proteins, are essential for accurate regulation of the host transcriptome and chromatin state during preimplantation development. Both knockdown and CRISPRi-based repression of MERVL result in embryonic lethality due to defects in differentiation and genomic stability. Furthermore, transcriptome and epigenome analysis revealed that loss of MERVL transcripts led to retention of an accessible chromatin state at, and aberrant expression of, a subset of two-cell-specific genes. Taken together, our results suggest a model in which an endogenous retrovirus plays a key role in regulating host cell fate potential.

Fig. 1. MERVL RNA exhibits dynamic nuclear-cytoplasmic expression during early stages of mouse preimplantation development.

a, Schematic of mouse preimplantation development. Totipotency is restricted to early-stage development (that is, zygote and two-cell stages). Blastomeres gradually transition to a pluripotent state from the four-cell stage onward and develop into a blastocyst consisting of inner cell mass (ICM) and trophectoderm (TE) before implantation in the uterus at 4.5-days postcoitum (dpc). b,c, Violin plots showing the log₂-transformed reads per kilobase of exon per million reads mapped (log₂RPKM) values of MERVL-int (b) and its LTR promoter, MT2_Mm (c) during preimplantation development. Each plot encompasses box plot; central bars represent medians, box edges indicate 50% of data points and the whiskers show 90% of data points. d, Representative images of smFISH for MERVL RNA with 4,6-diamidino-2-phenylindole (DAPI) counterstain during preimplantation development, from four independent experiments. Scale bars, 20 µm. ♀, female pronucleus (PN); ♂, male PN. e, Representative images of immunofluorescence staining for MERVL-Gag protein with DAPI counterstain during preimplantation development, from six independent experiments. Scale bars, 20 µm. ♀, female PN; ♂, male PN. Data for panels in b and c are available as source data. Source data

Fig. 2. MERVL plays a critical role in early lineage specification and maintaining genomic stability of preimplantation embryos.

a, Schematic of full-length MERVL, indicating the positions of ASOs. b, Representative images of smFISH for MERVL RNA (green) with DAPI counterstain (gray) in control and MERVL-KD embryos at early and late two-cell stages, from six independent experiments (top). Representative images of immunofluorescence staining for MERVL-Gag protein (pink) with DAPI counterstain (gray) in control and MERVL-KD embryos at late two-cell stage, from three independent experiments (bottom). Scale bars, 20 µm. c, Representative phase-contrast images of 4.5 dpc blastocysts in control and MERVL-KD, from four independent experiments. Scale bars, 100 µm. d, Percentage of embryos by stages of development in control (n = 84) and MERVL-KD (n = 104). NS, not significant; **P < 0.01; ***P < 0.001, chi-square test. e, Expressions of ICM- and TE-associated genes, as measured by qRT-PCR in control and MERVL-KD morula at 3.5 dpc. Bars show means with standard error of the mean (s.e.m.). Dots represent biological replicates (n = 6). *P < 0.05, **P < 0.01, ***P < 0.001, two-tailed unpaired t-tests. f, Representative images of immunofluorescence staining for OCT4 and CDX2 with DAPI counterstain in control and MERVL-KD morula at 3.5 dpc, from three independent experiments. Scale bars, 20 µm. g,h, Dot plots showing the total number of blastomere per embryo and relative intensity for OCT4 and CDX2 per nucleus in control and MERVL-KD morula at 3.5 dpc, normalized to DAPI signal. Central bars represent medians, and the top and bottom lines encompass 50% of the data points. ***P < 0.001, two-tailed unpaired t-tests. i, Representative images of immunofluorescence staining for E-cadherin (E-Cad, green) with DAPI counterstain (gray) in control and MERVL-KD morula at 3.5 dpc, from two independent experiments (left). Arrows and arrowheads indicate nuclear deformation and micronuclei. Scale bars, 20 µm. Bar chart showing the percentage of abnormal nuclear morphology at morula stage (right). *P < 0.05, chi-square test. j, Representative images of immunofluorescence staining for cleaved caspase-3 (cCasp3, green) with DAPI counterstain (gray) in control and MERVL-KD morula at 3.5 dpc, from three independent experiments (left). Scale bars, 20 µm. Bar chart showing the percentage of apoptotic cells per embryo at morula stage (right). ***P < 0.001, chi-square test. Data for panels in d, e and g–j are available as source data. Source data

Fig. 3. Retroviral proteins and trans-acting MERVL RNA are dispensable for preimplantation development.

a, Schematic of the distinct mechanisms underlying ASO- versus siRNA-mediated RNA targeting. b Representative images of smFISH for MERVL RNA (green) with DAPI counterstain (gray) in control and MERVL-KD embryos at early and late two-cell stages, from four independent experiments (top). Representative images of immunofluorescence staining for MERVL-Gag protein (pink) with DAPI (gray) counterstain in control and MERVL-KD embryos at late two-cell stage, from three independent experiments (bottom). Scale bars, 20 µm. c, Representative phase-contrast images of 4.5 dpc blastocysts in control and MERVL-KD by siRNA, from three independent experiments. Scale bars, 100 µm. d, Percentage of embryos by stages of development in control (n = 73) and MERVL-KD (n = 59). *P < 0.05, chi-square test. e, Schematic of experimental procedure for in trans rescuing MERVL-KD by ASOs. f, Representative images of smFISH for MERVL RNA with DAPI counterstain in early and late two-cell stage embryos in each experimental condition, from four independent experiments. Scale bars, 20 µm. g, Representative phase-contrast images of 4.5 dpc blastocysts in each experimental condition, from two independent experiments. Scale bars, 100 µm. h, Percentage of embryos by stages of development in each experimental condition (n = 78 for control, n = 80 for MERVL-KD and n = 132 for in trans rescue). *P < 0.05; **P < 0.01; ***P < 0.001, chi-square test. i, Schematic of experimental procedure for CRISPRi targeting the MERVL sequence. j, Representative images of smFISH for MERVL RNA (green) with DAPI counterstain (gray) in GFPi control (CRISPRi targeting GFP) and MERVLi (CRISPRi targeting MERVL) embryos at early and late two-cell stages, from three independent experiments (top). Representative images of immunofluorescence staining for MERVL-Gag protein (pink) with DAPI counterstain (gray) in GFPi control and MERVLi embryos at late two-cell stage, from three independent experiments (bottom). Scale bars, 20 µm. k, Representative phase-contrast images of 4.5 dpc blastocysts upon GFPi control and MERVLi condition, from 3 independent experiments. Scale bars, 100 µm. l Percentage of embryos by stages of development upon GFPi control (n = 149) and MERVLi (n = 137). ***P < 0.001, chi-square test. Data for panels in d, h and l are available as source data. Source data

Fig. 4. Ablation of MERVL impacts the expression of a subset of two-cell genes through preimplantation development.

a, Schematic of timing for minor and major ZGA (adapted from Abe et al.). b, Representative images of EU incorporation assay (green) with Hoechst 33342 counterstain (gray) in control and MERVL-KD two-cell stage embryos, from three independent experiments. Scale bars, 20 µm. c, The relative intensity of EU per nucleus, normalized to Hoechst 33342 signal. NS, not significant by two-tailed unpaired t-tests. d, Bidimensional principal-component (PC) analysis of gene expression profiles in control and MERVL-KD embryos through preimplantation development. e, RNA-seq differential gene expression analysis: MERVL-KD versus control embryos obtained at two-cell, four-cell and eight-cell stages; 938, 745 and 277 genes evinced significant changes in expression in MERVL-KD embryos (blue circle, fold change ≥ |2|, P value adjusted (Padj) < 0.05, binomial test with Benjamini–Hochberg correction). f, Stacked bar chart shows the number of upregulated (up)- and downregulated (down)-DEGs. The populations of maternally inherited RNA (as defined in DBTMEE v2; refs. ^,) and others (zygotically expressed genes) are highlighted in red and blue. g, Predicted factors that are upstream of all up-DEGs upon MERVL-KD, assessed by the ChEA database. ChIP-seq, chromatin immunoprecipitation with sequencing; HaCaT, cultured human keratinocyte; MEF, mouse embryonic fibroblast; MESC, mouse embryonic stem cell; LNCaP, lymph node carcinoma of the prostate. h, Bubble plot showing overlap between all DEGs in MERVL-KD embryos with the list of two-cell (2C) genes and DBTMEE v2 transcriptome categories. The bubble plot sizes show the –log₁₀[P values] derived from a hypergeometric test. i, Track views show RNA-seq signals in control and MERVL-KD embryos, on two representative 2C gene loci (as defined in Macfarlan et al.). The y-axis represents normalized tag counts for total RNA-seq in each sample. Refgene, refseq gene in NCBI and UCSC. j, Representative images of immunofluorescence staining for phosphorylated p53 on serine 15 (p-S15-p53, pink) with DAPI counterstain (gray) in control and ASO-mediated MERVL-KD embryos, from three independent experiments. Scale bars, 20 µm. k, Dot plots showing the relative intensity for p-S15-p53 per nucleus in control and MERVL-KD embryos, normalized to DAPI signal. Central bars represent medians, the top and bottom lines encompass 50% of the data points. *P < 0.05 and ***P < 0.001, two-tailed unpaired t-tests. Data for panels in c–h and k are available as source data. Source data

Fig. 5. MERVL-KD embryos retain two-cell-like chromatin accessibility even at mid-preimplantation stage.

a, Genome-wide correlation of ATAC-seq signals by stages of development between control and MERVL-KD embryos. Enrichment levels per 10-kb bin are shown in log₂RPKM values. The Pearson correlation coefficient values (R) indicate the similarity between control and MERVL-KD embryos. b, Average tag density plots of ATAC-seq enrichment around MERVL-int copies (±5 kb) by stages of development in control and MERVL-KD embryos. RPM, reads per million. **Padj < 0.01, Mann–Whitney U test with Bonferroni correction. c, Average tag density plots of ATAC-seq enrichment around TSS (±1 kb) of 2C genes by stages of development in control and MERVL-KD embryos. **Padj < 0.01, Mann–Whitney U test with Bonferroni correction. d, Track view shows ATAC-seq signals in control and MERVL-KD embryos at two-cell, four-cell and eight-cell stages, on representative 2C gene locus (as defined in Macfarlan et al.). The y-axis represents normalized tag counts for ATAC-seq in each sample. The flanking region around TSS is highlighted in red. e, Heatmaps showing ATAC-seq signals across all peak regions ±1 kb in control embryos at two-cell, four-cell, and eight-cell stages. Each peak was ordered by k-means clustering of ATAC-seq signal and yielded five clusters. Cluster 1–2, increased accessibility during preimplantation development (defined as category A); cluster 3–5, were transiently accessible at two-cell stage and reduced accessibility during preimplantation development (defined as category B). f, Stacked bar chart shows ATAC-seq peak distributions across genomic entities (intergenic, intron, exon and promoter) in each cluster. g, GO analysis of genes adjacent to ATAC-seq peaks of cluster 1–2 (category A) and cluster 3–5 (category B). The pluripotent-related and 2C genes were obviously enriched in gene set adjacent to category A and category B, respectively. h, Average tag density plots of ATAC-seq enrichment around peak regions (±1 kb) in each cluster by stages of development in control and MERVL-KD embryos. *Padj < 0.05, **Padj < 0.01, ***Padj < 0.001, Mann–Whitney U test with Bonferroni correction. Data for panels in a, f and g are available as source data. Source data

Fig. 6. Intergenic transcription from the adjacent regions to MERVL was interfered with by MERVL-KD.

a, Box plot showing the RPKM values for annotated genes, grouped by their distance to ASO-targeted full-length MERVL. Central bars represent medians, the boxes encompass 50% of the data points and the whiskers indicate 90% of the data points. NS, not significant, two-tailed unpaired t-tests. b, Box plot showing the RPKM values for adjacent transcripts grouped by their distance to ASO-targeted full-length MERVL. Central bars represent medians, the boxes encompass 50% of the data points and the whiskers indicate 90% of the data points. **P < 0.01, ***P < 0.001, two-tailed unpaired t-tests. c, Schematic of procedure for transcript calling with groHMM, a two-state HMM-based algorithm. aTAR, annotated transcriptionally active regions (blue rectangles); uTAR, unannotated intergenic transcriptionally active region (two-headed black arrow). d, RNA-seq differential expression analysis of uTARs between control and MERVL-KD two-cell stage embryos. MA plot showing differentially expressed (DE) uTARs between control and MERVL-KD embryos. DE uTARs were defined with a Padj < 0.01 (binomial test with Benjamini–Hochberg correction) and shown in blue circles (n = 3,499). e, Number of upregulated (up)- and downregulated (down)-uTARs in MERVL-KD two-cell stage embryos from d. Each transcriptional direction in the DE uTARs is shown in blue (plus strand) and red (minus strand). f, Track views show RNA-seq signals in control and MERVL-KD two-cell stage embryos on two representative uTARs. DE uTARs are highlighted in blue. The y-axis represents normalized tag counts for total RNA-seq in each sample. g, Distributions of relative distances of DE uTARs (red) and all called TARs (black) to MERVL loci. h, Average tag density plots of ATAC-seq enrichment around DE uTARs (±5 kb) in control and MERVL-KD two-cell stage embryos. RPM, reads per million. ***Padj < 0.001, Mann–Whitney U-test with Bonferroni correction. i, Model: MERVL-mediated totipotent-to-pluripotent transition during preimplantation development. MERVL and adjacent intergenic regions are expressed in a zygote and two-cell-stage-specific manner and suppresses a subset of 2C gene expressions through an unknown mechanism to facilitate the totipotent-to-pluripotent transition. KD of MERVL transcript compromises these processes and results in retention of two-cell-like states and embryonic lethality. Data for panels a, b, d, e and g are available as source data. Source data

Extended Data Fig. 1. Design and validation of ASOs targeting MERVL.

(a) Schematic of MERVL indicating the positions of ASOs. Intact MERVL encodes the retroviral proteins; Gag (Group-specific antigens, comprised of MA, matrix; CA, capsid proteins) and Pol (Polymerase, comprised of RT, reverse transcriptase; INT, integrase; dUTPase, dUTP phosphatase). Below whisker plot showing alignment quality in each interspersed genomic MERVL copy (adapted from Dfam: https://dfam.org/family/DF0003918/seed). (b) Pie chart indicates the populations of full length (≥ 5001 bp) and truncated (1–5000 bp) MERVL copies across the mouse genome. (c) Chromosome maps showing the distribution of MERVL copies that are targeted by ASOs throughout the mouse genome. Each colored rectangle (blue for ASO-#1, yellow for ASO-#2, and red for ASO-#3) indicates targeted MERVL copy. (d) Histogram showing the distributions of ASO-targeted (magenta) and untargeted (green) MERVL copies. The abundance of each length of MERVL element was tallied with a given count. The proportion of full-length MERVL (>5 kb) was highlighted in red. (e) Schematic for ASO-mediated KD against MERVL in ESC harboring a doxycycline (Dox)-inducible Dux transgene (termed ESC^DUX). TRE, tetracycline responsive element; rtTA, reverse tetracycline responsive transcriptional activator. (f) Expression levels of MERVL mRNA measured by qRT-PCR in ESC^DUXs nucleofected with each individual MERVL-targeting ASO and scrambled ASO. Relative expression is quantified with ΔΔCt method and normalized to Gapdh expression. Bars show means with s.e.m. Dots represent biological replicates (n = 3 samples). (g) Expression level of MERVL mRNA measured by qRT-PCR in scrambled control and ASOs (Mixed)-mediated MERVL-KD ESC^DUXs in the presence or absence of Dox. Relative expression is quantified with ΔΔCt method and normalized to Gapdh expression. Bars show means with s.e.m. Dots represent biological replicates (n = 4 samples). (h) Expressions of MERVL retroviral protein (Gag), measured by western blotting in scrambled control and ASOs (Mixed)-mediated MERVL-KD ESC^DUXs in the presence or absence of Dox. β-Tubulin was used as a loading control. Representative blot image is from 3 independent experiments. Data for panels in b-d and f-h are available as source data. Source data

Extended Data Fig. 2. The KD effectiveness of each ASO against MERVL in preimplantation embryos.

(a) Representative images of smFISH for MERVL RNA (green) with DAPI counterstain (gray) in two-cell stage embryos from each experimental condition, from 3 independent experiments. Zygotes were injected with scrambled ASO, MERVL ASOs or the corresponding SOs. Scale bar: 20 µm. (b) Representative phase-contrast images of 4.5 dpc blastocysts from each experimental condition, from 2 independent experiments. Scale bar: 100 µm. (c) Percentage of embryos by developmental stage at 4.5 dpc in each experimental condition. The number of embryos in each experimental condition is shown in the bottom. Data for panels in c are available as source data. Source data

Extended Data Fig. 3. CasRx-mediated MERVL-KD and its effects on preimplantation development.

(a) Schematic of experimental procedure for CasRx-mediated KD of MERVL. (b) Representative images of smFISH and immunofluorescence staining for MERVL RNA (top) and MERVL-Gag protein (bottom) with DAPI counterstain in untargeted control and MERVL-KD two-cell stage embryos at 1.5 dpc, from 3 independent experiments. Scale bar: 20 µm. (c) Representative phase-contrast images of 4.5 dpc blastocysts upon untargeted control and MERVL-KD conditions, from 3 independent experiments. Scale bar: 100 µm. (d) Percentage of embryos by stages of development, upon untargeted control (n = 69) and MERVL-KD (n = 70) conditions. N.S., not significant; ***P < 0.001, chi-square test. Data for panel in d is available as source data. Source data

Extended Data Fig. 4. Total RNA-seq analysis of control and MERVL-KD embryos at two-cell, four-cell, and eight-cell stages.

(a) Schematic of sample collection for total RNA-seq analysis in scrambled control and ASO-mediated MERVL-KD embryos. We supplied the apparently healthiest embryos upon MERVL-KD for preparation of a total RNA-seq library. (b) Scatter plots showing the reproducibility between biological replicates in total RNA-seq data. Pearson correlation values (R) are shown. (c) GO analysis of DEGs in MERVL-KD embryos, assessed by the Enrichr. (d) Heatmap showing hierarchical clustering of expression patterns of all 2C genes (as defined in 12) in control and MERVL-KD embryos at two-cell, four-cell and eight-cell stages. Expression level of each gene is shown in Z-score, calculated by subtracting the mean expression value and dividing by standard deviation. Data for panels in c and d are available as source data. Source data

Extended Data Fig. 5. MERVL-KD only dysregulates a few types of transposable element.

(a) Dot plots showing RPKM values for MERVL-int and MT2_Mm in scrambled control and MERVL-KD embryos at two-cell, four-cell, and eight-cell stages. Central bars represent medians, the top and bottom lines encompass 50% of the data points. (b) MA plots show differentially expressed (DE) repetitive elements (annotated in RepeatMasker) between scrambled control and MERVL-KD embryos. DE repeats were defined as those with a |FoldChange| ≥ 2 and Padj < 0.05 (binomial test with Benjamini–Hochberg correction) and shown in red circles. MERVL-int and MT2_Mm were included in downregulated repeats. (c) Violin plots indicate CPM values for chimeric fusion transcripts in scrambled control and MERVL-KD two-cell stage embryos. Each plot encompasses box plot; central bars represent medians, box edges indicate 50% of data points, and the whiskers show 90% of data points. N.S., not significant, two-tailed unpaired t-tests. Data for panels in a-c are available as source data. Source data

Extended Data Fig. 6. RNA-seq analysis upon CRISPRi induction shows that MERVLi embryos resemble ASO-mediated MERVL-KD embryos.

(a) Unsupervised hierarchical cluster of all annotated transcript profiles (n = 55,254) from RNA-seq data in embryos from different groups and developmental stages. Each dendrogram leaf represents an RNA-seq sample and the y-axis shows the distance based on Pearson correlation between each pair of samples. (b) Heatmap showing pairwise Pearson correlation coefficient of gene expression between each pair of samples. (c) Dot plots showing RPKM values for MERVL-int and MT2_Mm in GFPi control and MERVLi embryos. Central bars represent medians, the top and bottom lines encompass 50% of the data points. (d) RNA-seq differential gene expression analysis: MERVLi versus GFPi control embryos obtained at two-cell, four-cell and eight-cell stages. 872, 535 and 273 genes evinced significant changes in expression in MERVLi embryos (blue circles, Padj < 0.05; binomial test with Benjamini–Hochberg correction). (e) Bubble plot showing overlap between all DEGs in MERVLi embryos with the list of 2C genes and DBTMEE v2 transcriptome categories. The bubble plot sizes show the -log₁₀[P values] derived from a hypergeometric test. (f) Track views show RNA-seq signals in GFPi control and MERVLi embryos, on a representative 2C gene locus (as defined in 12). The y-axis represents normalized tag counts for total RNA-seq in each sample. Data for panels in a-e are available as source data. Source data

Extended Data Fig. 7. miniATAC-seq analysis of control and MERVL-KD embryos at two-cell, four-cell, and eight-cell stages.

(a) Schematic of sample collection for miniATAC-seq analysis in scrambled control and ASO-mediated MERVL-KD embryos. We supplied apparently healthiest embryos upon MERVL-KD for preparation of miniATAC-seq library. (b) Scatter plots showing the reproducibility between biological replicates in miniATAC-seq data. Pearson correlation values (R) are shown. (c) Track view of ATAC-seq read enrichments at a representative chromosomal position in all biological replicates. The y-axis represents normalized tag counts for ATAC-seq in each sample.