Suppression of ERK signalling promotes pluripotent epiblast in the human blastocyst

Abstract

Studies in the mouse demonstrate the importance of fibroblast growth factor (FGF) and extra-cellular receptor tyrosine kinase (ERK) in specification of embryo-fated epiblast and yolk-sac-fated hypoblast cells from uncommitted inner cell mass (ICM) cells prior to implantation. Molecular mechanisms regulating specification of early lineages in human development are comparatively unclear. Here we show that exogenous FGF stimulation leads to expanded hypoblast molecular marker expression, at the expense of the epiblast. Conversely, we show that specifically inhibiting ERK activity leads to expansion of epiblast cells functionally capable of giving rise to naïve human pluripotent stem cells. Single-cell transcriptomic analysis indicates that these epiblast cells downregulate FGF signalling and maintain molecular markers of the epiblast. Our functional study demonstrates the molecular mechanisms governing ICM specification in human development, whereby segregation of the epiblast and hypoblast lineages occurs during maturation of the mammalian embryo in an ERK signal-dependent manner.

Fig. 1. Exogenous FGF is sufficient to drive human hypoblast specification.

a Confocal images of Day 6.5 human embryos immunofluorescently labelled for lineage markers NANOG (epiblast), GATA4 (hypoblast) and GATA3 (trophectoderm), and stained for nuclear DAPI. Human embryos were cultured in Control medium, or medium supplemented with increasing concentrations of FGF and Heparin as indicated from Day 5 for 36 h. Total cell number (c) indicated. Scale bars 20 µm. b−c Boxplots showing the number of (b) hypoblast (GATA4 + NANOG-GATA3-) and (c) epiblast (NANOG + GATA4-GATA3-) cells in human embryos cultured in increasing concentrations of FGF and Heparin. Boxplots represent the interquartile (IQR range), with the median shown as a central line; whiskers extend to lowest or highest value within 1.5 * IQR; values for each embryo are shown as individual points. Control n = 11, 250 ng/ml n = 8, 500 ng/ml n = 9, 750 ng/ml n = 9. d Stacked bar charts showing the mean proportion of epiblast and hypoblast in the ICMs per embryo in each treatment group. Control n = 11, 250 ng/ml n = 8, 500 ng/ml n = 9, 750 ng/ml n = 9. Two-tailed t-test, n.s. not significant * p < 0.05, ** p < 0.01, *** p < 0.001. Source data are available on Github.

Fig. 2. Suppression of ERK signalling blocks hypoblast formation in the human blastocyst.

a Confocal images of Day 5 − 6.5 human embryos immunofluorescently labelled for phosphorylated (p)-ERK, lineage markers SOX2 (epiblast), OTX2 (hypoblast), and stained for nuclear DAPI. Cyan arrow indicates a hypoblast cell with high pERK levels. Scale bars 20 µm. b Violin plots showing the cytoplasmic fluorescence intensity of pERK in each lineage in embryos shown in (a). Boxplots represent the interquartile (IQR range), with the median shown as a central line; whiskers extend to lowest or highest value within 1.5 * IQR. Epiblast n = 83 cells, hypoblast n = 41 cells, trophectoderm n = 1535 cells; n = 10 embryos. c Violin plots showing the cytoplasmic fluorescence intensity of pERK over time in embryos shown in (a). Boxplots represent the interquartile (IQR range), with the median shown as a central line; whiskers extend to lowest or highest value within 1.5 * IQR. Day 5 n = 313 cells; 3 embryos, Day 6 n = 542 cells; 3 embryos, Day 6.5 n = 804 cells; 4 embryos. d Confocal images of Day 6.5 human embryos immunofluorescently labelled for lineage markers NANOG (epiblast), GATA4 (hypoblast) and GATA3 (trophectoderm) and stained for nuclear DAPI. Human embryos cultured with or without ERKi (5 µm Ulixertinib) from Day 5 for 36 h. Total cell number (c) indicated. Scale bars 20 µm. e, f Boxplots showing the number of (e) hypoblast (GATA4 + ) and (f) epiblast (NANOG + ) cells in human embryos cultured with and without ERKi. Boxplots represent the interquartile (IQR range), with the median shown as a central line; whiskers extend to lowest or highest value within 1.5 * IQR; values for each embryo are shown as individual points. Control n = 8, ERKi n = 10. g Stacked bar charts showing the mean proportion of epiblast and hypoblast in the ICMs per embryo in each treatment group. Control n = 8, ERKi n = 10. Two-tailed t-test, ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001. Source data are available on Github.

Fig. 3. ERK inhibition alters lineage specification and maintains naïve pluripotency characteristics in human embryos.

a UMAP of single cells from Day 6.5 human embryos cultured in ERKi (orange) or control medium (black). b UMAP coloured according to lineage identity. c UMAP coloured by expression of epiblast or hypoblast marker genes. Scales show log transformed normalised expression values. d Volcano plot with genes significantly differentially expressed between ERKi treated and control cells within the epiblast cluster in (a). Significantly genes (DESeq2 padj <0.05 and log2FC > 1.5) upregulated (orange) or downregulated (purple) in ERKi vs control. Adjusted p-values were calculated using the two-sided Benjamini–Hochberg method to control for FDR across multiple tests. e Hierarchical clustering and heatmap of selected significantly differentially expressed genes between ERKi treated and control cells within the epiblast cluster. Normalized expression shown as z-score. f, g UMAP integrating this study with a compiled human embryo reference dataset, coloured according to 10.1101/2024.02.01.578414 (f) lineage prediction using a deep learning-model and (g) treatment group from this study Control (black) and ERKi (orange). Reference data set is under control conditions. h Stacked bar chart showing lineage prediction of cells from this study using a deep learning-model in control and ERKi conditions. Source data are available on Github.

Fig. 4. Suppression of ERK signalling promotes pluripotent epiblast identity.

a Confocal images of Day 6.5 human embryos immunofluorescently labelled for SOX2 and KLF17 (epiblast) and stained for nuclear DAPI. Scale bars 20 µm. b Table showing outcome of naïve hESC derivation from Control and ERKi cultured embryos in PXGL medium (Control PXGL; ERKi PXGL). c Phase-contrast image of an ERKi PXGL naïve hESC line. d, e Confocal images of an ERKi PXGL naïve hESC line, immunofluorescently labelled for naïve (TFAP2C, SUSD2, KLF4, KLF17) and core (SOX2, OCT4) pluripotency markers and stained for nuclear DAPI. Scale bars 20 µm. Representative images Control n = 2, ERKi n = 5. f Principal component analysis following RNA-seq of hESC derived in this study and previously published primed and naïve hESC lines. Per-gene variance was modelled, and significantly variable genes (DESeq2 padj < 0.05) were used in the loading. Adjusted p-values were calculated using the two-sided Benjamini–Hochberg method to control for FDR across multiple tests. g Heatmap showing the normalised median expression of selected pluripotency genes in previously published hESC and from this study. h Dot plots showing the relative expression of human pre-implantation^– epiblast-enriched genes (% Epi-like identity) for individual hESC lines. Horizontal bars indicate median gene expression in each group. RNA-seq analysis (f-h) of published primed (n = 48) or naïve (n = 55) hESC^,–,, and hESC derived in this study (Control PXGL n = 2, ERKi PXGL n = 5, Control UXGL n = 1, ERKi UXGL n = 2). i Principal components analysis of DNA methylome in published blastocyst, primed hESC and naïve hESC, and hESC derived in this study. j Beanplots showing distribution of DNA methylation for published blastocyst, primed hESC, naive hESC, and hESC derived in this study. Whole-genome bisulphite sequencing analysis (i, j) of published blastocyst (n = 1), primed hESC (n = 2), and naïve hESC (n = 2)^– and hESC derived in this study: Control in PXGL (n = 2), ERKi in PXGL (n = 3), Control in UXGL (n = 1) and ERKi in UXGL (n = 1). DNA Methylation is quantified using autosomal 100-CpG windows. Source data are available on Github.