Pluripotent state transitions coordinate morphogenesis in mouse and human embryos

Abstract

The foundations of mammalian development lie in a cluster of embryonic epiblast stem cells. In response to extracellular matrix signalling, these cells undergo epithelialization and create an apical surface in contact with a cavity, a fundamental event for all subsequent development. Concomitantly, epiblast cells transit through distinct pluripotent states, before lineage commitment at gastrulation. These pluripotent states have been characterized at the molecular level, but their biological importance remains unclear. Here we show that exit from an unrestricted naive pluripotent state is required for epiblast epithelialization and generation of the pro-amniotic cavity in mouse embryos. Embryonic stem cells locked in the naive state are able to initiate polarization but fail to undergo lumenogenesis. Mechanistically, exit from naive pluripotency activates an Oct4-governed transcriptional program that results in expression of glycosylated sialomucin proteins and the vesicle tethering and fusion events of lumenogenesis. Similarly, exit of epiblasts from naive pluripotency in cultured human post-implantation embryos triggers amniotic cavity formation and developmental progression. Our results add tissue-level architecture as a new criterion for the characterization of different pluripotent states, and show the relevance of transitions between these states during development of the mammalian embryo.

Extended Data Figure 1. Epiblast gene expression patterns at peri-implantation stages.

a, CAG–GFP+ embryos recovered at peri-implantation stages, before and after dissection of the epiblast. n = 4 E4.5, 4 E4.75 and 4 E5.0 embryos. b–e, For all the collected samples we computed the total number of reads (b), the fraction of reads mapped to endogenous genes (c), the number of genes with more than 10 reads per million (RPM) (d) and the fraction of reads mapped to mitochondrial genes (e). f, A PCA of the metrics shown in b–e. The sample coloured in grey (E4.5_S1) is characterized by a very low fraction of mapped reads and number of genes detected and was therefore removed from all downstream analysis. g, Hierarchical clustering of the E4.5–E4.75 samples that passed the quality check (black) and samples from the epiblast (red) and primitive endoderm (blue) at E4.5 from the dataset of ref. . One sample (E4.5_S3, grey) clusters with the primitive endoderm, whereas all others cluster with the epiblast. E4.5_S3 was therefore excluded from further analysis. h, PCA of our samples and samples from ref. . i, log fold change in pluripotency marker genes between E5.0 and E4.5–E4.75 samples from our dataset (x axis) and log fold change between E5.5 and E4.5 epiblast samples from ref. (y axis). Squares, circles and triangles mark genes differentially expressed in both datasets, only in our dataset or only in the dataset of ref. respectively. The naive pluripotency genes significantly downregulated at E5.5 were highly enriched among genes downregulated at E5.0 (P = 7 x 10⁻⁸, Fisher’s exact test), whereas the enrichment of upregulated early post-implantation factors was only marginally significant (P = 0.03, Fisher’s exact test). j, Bright field images of cells derived from embryos cultured in IVC1 (0/6 mES cell colonies from IVC1 embryos and 4/6 mES cell colonies from IVC1 + 2i/LIF embryos). Scale bars, 50μ m. k, Immunostaining of mES cells derived from embryos cultured in IVC1 + 2i/LIF for 24 h. Scale bar, 10 μ m. l, Experimental set-up. A, analysis. m, Immunostaining of mouse embryos cultured as indicated in l. Dotted lines indicate the epiblast and the arrow points to the pro-amniotic cavity. Scale bars, 20 μm. n, Lumen formation in embryos from m. n = 12 IVC1 control and 15 IVC1 experimental embryos. χ2 test, * P = 0.0137.

Extended Data Figure 2. Dynamics of polarization, lumenogenesis and naive pluripotency exit in mES cells.

a, Experimental set-up. A, analysis. b, c, Immunostaining of mES cells in 3D Matrigel (b, top, c). Red long arrows (b) indicate the position used to plot intensity profiles (b, bottom). Yellow arrows indicate polarization and white arrows lumen formation. Scale bars, 10 μm. d, Quantification of lumen formation in cells from b and c. n = 21 (2 cells), 22 (4 cells), 20 (8 cells) and 21 (> 10 cells) spheroids. χ2 test, * * * P < 0.0001. e, Fluorescence intensity in cells from f. Raw fluorescence intensity values were normalized to the fluorescence intensity of reporter cells cultured in gelatin + 2i/LIF throughout the timelapse video. n = 55 Nanog–YFP, 92 Δ PE–Oct4–GFP and 63 Rex1::GFPd2 spheroids. AU, arbitrary units. f, Time-lapse frames of naive reporter mES cells cultured in Matrigel without 2i/LIF. mT: membrane Tomato. Time is indicated as h:min. Dotted lines indicate the outline of the spheroids at the end of the experiment. Scale bars, 20 μm. g–o, Expression of pluripotency genes in mES cells cultured in gelatin or Matrigel, with or without 2i/LIF. n = 5 (24 h), 3 (36 h), 3 (48 h) independent samples for Nanog, Esrrb and Otx2; n = 5 (24 h), 3 (36 h), 4 (48 h) independent samples for Fgf5; n = 4 (24 h), 3 (36 h), 3 (48 h) independent samples for Klf4 and Klf2; n = 4 (24 h), 3 (36 h), 4 (48 h) independent samples for Zfp42; and n = 3 (24 h), 3 (36 h), 3 (48 h) independent samples for Tbx3 and Pou5f1. Unpaired Student’s t-test, NS, not significant. ND, not detected. p, q, Immunostaining of mES cells cultured in 3D Matrigel. Scale bars, 10 μm. M, Matrigel.

Extended Data Figure 3. mES cells initiate polarization in naïve conditions in a 3D Matrigel culture.

a, Experimental set-up. A, analysis. b, Immunostaining of mES cells cultured as indicated in a. Arrows indicate lumens. Scale bars, 5 μm. c, Lumen formation in cells from b. n = 31 (2 cells, gelatin + 2i/LIF 48 h, Matrigel − 2i/LIF 24 h), 30 (4 cells, gelatin + 2i/LIF 48 h, Matrigel − 2i/LIF 24 h), 31 (2 cells, gelatin − 2i/LIF 48 h, Matrigel − 2i/LIF 24 h) and 31 (4 cells, gelatin − 2i/LIF 48 h, Matrigel − 2i/LIF 24 h) spheroids. χ2 test, * * * P < 0.0001. d, Internuclear distance in cells from Fig. 2b. n = 31 (+ 2i/LIF) and 30 (− 2i/LIF) spheroids. Mann–Whitney U-test; NS, not significant. e–g, Immunostaining of mES cells cultured in Matrigel with or without 2i/LIF. Squares indicate magnified regions (e). Scale bars, 10 μm (e and f) and 5 μm (g). h–j, Representative frames from time-lapse experiments using different naive reporter mES cell lines, cultured in either gelatin or Matrigel in the presence of 2i/LIF. Time is indicated as h:min. Scale bars, 50 μ m. k–m, Intensity of Nanog–YFP, Δ PE–Oct4–GFP and Rex1::GFPd2 mES cells analysed by time-lapse microscopy from h–j. AU, arbitrary units. n = 65 Nanog–YFP, 55 Δ PE–Oct4–GFP and 43 Rex1::GFPd2 mES cell colonies (gelatin); and n = 61 Nanog–YFP, 84 Δ PE–Oct4–GFP and 61 Rex1::GFPd2 mES cell spheroids (Matrigel). n, Experimental set-up. o, Immunostaining of mES cells cultured as indicated in n. Squares indicate magnified regions. Dotted lines demarcate the border of the colonies. The nucleus–centrosome angle (α) measured in p is indicated. Scale bars, 10 μm. p, Histogram of the angle between the centrosome–nucleus axis and the basal–apical axis of cells in the border of the colonies shown in o. Data are shown as relative frequencies (%). n = 76 (Matrigel + 2i/LIF cultured in gelatin), 61 (Matrigel − 2i/LIF cultured in gelatin), 62 (Matrigel + 2i/LIF) and 60 (Matrigel − 2i/LIF) centrosomes. Kruskal–Wallis test; * * * P < 0.0001; NS, not significant. q, Immunostaining of control and Dgcr8 knockout mES cells (with or without 2i/LIF). Scale bars, 5 μm. r, Angle between the nucleus–centrosome axis and the nuclear–nuclear axis in cells from q. Each dot represents an individual centrosome. n = 60 control + 2i/LIF, 62 control − 2i/LIF, 56 Dgcr8 knockout + 2i/LIF and 58 Dgcr8 knockout − 2i/LIF centrosomes. Kruskal–Wallis test; NS, not significant. s, Internuclear distance in cells from q. n = 30 control + 2i/LIF, 31 control − 2i/LIF, 28 Dgcr8 knockout + 2i/LIF and 29 Dgcr8 knockout −2i/LIF spheroids. Kruskal–Wallis test. t, Nanog intensity in cells from q. n = 30 control + 2i/LIF, 31 control − 2i/LIF, 28 Dgcr8 knockout + 2i/LIF and 30 Dgcr8 knockout − 2i/LIF spheroids. Kruskal–Wallis test; * * * P < 0.0001. u, Immunostaining of mES cells cultured in Matrigel with or without Gö6983. Scale bars, 10 μ m. v, Internuclear distance in cells from Fig. 2i. n = 26 (+ Gö6983) and 28 (− Gö6983) spheroids. Mann–Whitney U-test; NS, not significant. G, gelatin; M, Matrigel.

Extended Data Figure 4. 2i/LIF inhibits lumen formation in mES cells.

a, Immunostaining of mES cells cultured in Matrigel. Scale bars, 10 μm. b, Immunostaining of Rex1::GFPd2 mES cells cultured in Matrigel. Scale bars, 10 μm. c, Oct4, Nanog and Rex1::GFPd2 fluorescence intensity in cells from a and b. n = 30 spheroids per condition (Nanog); n = 21 (+ 2i/LIF) and 20 (− 2i/LIF) spheroids (Oct4); and 21 (+ 2i/LIF) and 20 (− 2i/LIF) spheroids (Rex1::GFPd2). Unpaired Student’s t-test; * * * P < 0.0001; NS, not significant. d, e, Electron microscopy images of mES cells cultured in Matrigel with or without 2i/LIF. Arrows indicate lumens. a, Golgi apparatus; b, basolateral cell–cell adhesion sites; c, tight junctions. Scale bars, 2 μm (d) and 500 nm (e). f, g, Immunostaining of mES cells cultured in Matrigel. Squares indicate magnified regions (g). Arrows indicate inner non-polarized cells. Scale bars, 10 μ m. h, Polarization in cells from g and Fig. 3c. n = 27 (+ 2i/LIF 48 h), 26 (− 2i/LIF 48 h), 24 (+ 2i/LIF 72 h) and 24 (− 2i/LIF 72 h) spheroids. χ2 test; * * * P < 0.0001. i, Experimental set-up. A, analysis. j, Immunostaining of mES cells cultured as indicated in i. Arrow points to lumen. Scale bars, 10 μm. k, Lumen formation in cells from j. n = 20 spheroids per condition. χ2 test; NS, not significant. l, Nanog intensity in cells from j. n = 20 spheroids per condition. ANOVA; * * * P < 0.0001. m, Immunostaining of mES cells cultured in Matrigel with different combinations of inhibitors. Arrow points to lumen. Scale bars, 10 μm. n, Lumen formation in cells from m. n = 20 spheroids per condition. χ2 test; * * * P < 0.0001. o, Nanog intensity in cells from m. n = 20 spheroids per condition. ANOVA; * * * P < 0.0001. p, Immunostaining of mES cells cultured in Matrigel with a single inhibitor or supplement. 2i/LIF was replaced by a single inhibitor or supplement 48 h before plating the cells in Matrigel. Arrows indicate lumens. Scale bars, 10 μm. q, Nanog intensity in cells from p as a function of their ability to undergo lumenogenesis. n = 30 (+ 2i/LIF control), n = 50 (+ LIF), n = 48 (+ GSK3i), n = 51 (+ MEKi) and n = 39 (− 2i/LIF) spheroids. Mann–Whitney U-test; * * P = 0.0055; * * * P < 0.0001; NS, not significant. M, Matrigel.

Extended Data Figure 5. Exit from naive pluripotency is required for lumenogenesis in mES cells.

a, Immunostaining of control and Dgcr8 knockout mES cells (with or without 2i/LIF). b, Lumen formation in cells from a. n = 20 (control + 2i/LIF), 20 (control −2i/LIF), 20 (Dgcr8 knockout + 2i/LIF) and 27 (Dgcr8 knockout − 2i/LIF) spheroids. χ2 test; * P = 0.024. c, Nanog intensity in cells from a. n = 20 (control + 2i/LIF), 20 (control − 2i/LIF), 20 (Dgcr8 knockout + 2i/LIF) and 27 (Dgcr8 knockout − 2i/LIF) spheroids. ANOVA; * * * P < 0.0001. d, Nanog intensity in Dgcr8 knockout mES cells after 72 h of 3D Matrigel culture (a), as a function of their ability to undergo lumenogenesis. n = 6 (lumen) and 21 (no lumen) spheroids. Unpaired Student’s t-test; * P = 0.0208. e, Immunostaining of mES cells cultured in Matrigel with or without Gö6983. f, Lumen formation in cells from e. n = 30 spheroids per condition. χ2 test; * * * P < 0.0001. g, Immunostaining of Rex1::GFPd2 mES cells cultured in Matrigel with or without Gö6983. h, Nanog, Rex1::GFPd2 and Oct4 intensity in cells from e and g. n = 30 spheroids per condition (Nanog); 41 (+ Gö6983) and 40 (− Gö6983) spheroids (Oct4 and Rex1::GFPd2). Mann–Whitney U-test; * * P = 0.0002; * * * P < 0.0001. i, Experimental set-up. A, analysis. j, Immunostaining of mES cells cultured as indicated in i. The percentage of structures showing the representative phenotype is indicated. k, Nanog intensity in cells from j as a function of their ability to undergo lumenogenesis. n = 13 (no lumen) and 20 (lumen) spheroids. Unpaired Student’s t-test; * * * P < 0.0001. l, Immunostaining of mES cells cultured as indicated in i. The percentage of structures showing the representative phenotype is indicated. m, Nanog intensity in cells from l as a function of their ability to undergo lumenogenesis. n = 14 (no lumen) and 17 (lumen) spheroids. Unpaired Student’s t-test; * * * P < 0.0001. n, Immunostaining of DOX-inducible Nanog mES cells. o, Lumen formation in cells from n. n = 30 (+ 2i/LIF/DOX), 30 (+ 2i/LIF), 31 (+ DOX) and 28 (− 2i/LIF) spheroids. χ2 test, NS, not significant. p, Nanog intensity in cells from n. n = 30 (+ 2i/LIF/DOX), 30 (+ 2i/LIF), 31 (+ DOX) and 28 (− 2i/LIF) spheroids. Kruskal–Wallis test; * * * P < 0.0001. q, mRNA levels of Nanog and Otx2 in DOX-inducible Nanog mES cells. n = 5 (Nanog) and 3 (Otx2) independent samples per group. Unpaired Student’s t-test; * * P = 0.0044; NS, not significant. r, mRNA levels of naive pluripotency genes in DOX-inducible Nanog mES cells. n = 4 independent samples per group. Unpaired Student’s t-test; * * P = 0.0075; NS, not significant. s, Immunostaining of control and Nanog knockdown mES cells. t, Lumen formation in cells from s. n = 20 (control siRNA) and 42 (Nanog siRNA) spheroids per condition. χ2 test; NS, not significant. u, Nanog and Rex1::GFPd2 intensity in cells from s. n = 20 (control siRNA) and 42 (Nanog siRNA) spheroids. v, Correlation between Nanog and Rex1::GFPd2 levels in cells from s. n = 20 (control siRNA) and 42 (Nanog siRNA) spheroids. w, Otx2 and Oct4 intensity in Oct4 knockdown mES cells. n = 19 (lumen) and 28 (no lumen) spheroids. x, Immunostaining of control and Oct4 knockdown mES cells. Scale bars, 10 μm (a, e, g, l, n, s, x) and 5 μm (j); arrows indicate lumens.

Extended Data Figure 6. Sialomucins are required for mES cell lumenogenesis.

a, mRNA levels of Podxl in mES cells. n = 5 (24 h), 3 (36 h) and 4 (48 h) independent samples. Unpaired Student’s t-test; * P = 0.0395 (gelatin 24 h); * P = 0.0481 (Matrigel 24 h); * * P = 0.0038 (gelatin 36 h); * P = 0.0126 (Matrigel 36 h); * * P = 0.0075 (gelatin 48 h); * P = 0.0139 (Matrigel 48 h). b, Immunostaining of mES cells cultured in 3D Matrigel. Arrow indicates lumen. Scale bars, 10 μm. c, Immunostaining of control and Dgcr8 knockout mES cells cultured in Matrigel with or without 2i/LIF. Arrow indicates lumen. Scale bars, 10 μm. d, Immunostaining of mES cells cultured in Matrigel with or without Gö6983. Arrow indicates lumen. Scale bars, 10 μm. e, Immunostaining of control and Podxl knockdown mES cells. A binarized image of the Podxl channel was used to determine presence or absence of Podxl. The percentages indicate the proportion of spheroids with or without Podxl in the Podxl siRNA group. Arrows indicate lumens. Scale bars, 10 μm. f, Lumen formation in cells from e. n = 43 (control), 45 (Podxl siRNA with Podxl (Podxl siRNA-treated cells that did not show a reduction in Podxl expression)) and 35 (Podxl siRNA without Podxl (Podxl siRNA-treated cells that showed a reduction in Podxl expression)) spheroids. χ2 test; * * * P < 0.0001. g, mRNA levels of early post-implantation factors in control and Podxl knockdown mES cells 24 h after removal of 2i/LIF. n = 3 independent samples per group. Student’s t-test; NS, not significant. h, mRNA levels of naive pluripotency genes in control and Podxl knockdown mES cells 24 h after removal of 2i/LIF. n = 3 independent samples per group. Student’s t-test; NS, not significant. i, Immunostaining of control and Podxl knockdown mES cells cultured in Matrigel. Arrows indicate polarization. Scale bars, 5 μm. j, Centrosome positions in cells from i. Each dot represents an individual centrosome. n = 40 (control siRNA) and 36 (Podxl siRNA) centrosomes. Mann–Whitney U-test; NS, not significant. k, Internuclear distance in cells from i. n = 20 (control siRNA) and 18 (Podxl siRNA) spheroids. Unpaired Student’s t-test; NS, not significant. l, Immunostaining of Podxl heterozygous (HET) and knockout mES cells. One knockout clone is shown as an example. Arrows indicate lumens. Scale bars, 10 μm. m, Lumen formation in cells from l. n = 30 (HET), 33 (clone 1) and 20 (clone 2) spheroids. χ2 test; NS, not significant. n, Immunostaining of control and Podxl knockdown mES cells with or without GFP–Cd34 overexpression. Arrows indicate lumens. Scale bars, 10 μm. o, Lumen formation in cells from n. n = 37 (control siRNA), 29 (Podxl siRNA), 33 (control siRNA + GFP–Cd34) and 31 (Podxl siRNA + GFP–Cd34) spheroids. χ2 test; * * * P < 0.0001; NS, not significant. p, mRNA levels of sialomucin proteins in control and Podxl knockdown mES cells (with or without GFP–Cd34 overexpression) 48 h after removal of 2i/LIF. n = 4 (without GFP–Podxl) and 6 (with GFP–Cd34) independent samples per group. Unpaired Student’s t-test; * P = 0.0145; * * * P = 0.0001. G, gelatin; M, Matrigel.

Extended Data Figure 7. Otx2 and Oct4 induce Podxl expression upon naive pluripotency exit.

a, mRNA levels of Fgf5 and Podxl in wild-type and Otx2 knockout mES cells at different time points after removal of 2i/LIF. n = 2 independent samples per group. ANOVA; * * P = 0.0092; * * * P < 0.0001. b, c, Immunostaining of wild-type and Otx2 knockout mES cells cultured in Matrigel. d, Lumen formation in cells from b and c. n = 46 (wild-type 48 h), 34 (wild-type 72 h), 36 (Otx2 knockout 48 h) and 35 (Otx2 knockout 72 h) spheroids. χ2 test; * * * P < 0.0001. e, Immunostaining of Otx2 knockout and Otx2-knockout tetON-Otx2 mES cells in the presence or absence of DOX (DOX addition triggers Otx2 overexpression). f, Lumen formation in cells from e and in wild-type mES cells cultured without 2i/LIF and with or without Otx2 overexpression. n = 19 (wildtype), 20 (wild-type tetON-Otx2 − DOX), 20 (wild-type tetON-Otx2 +DOX), 20 (Otx2 knockout), 20 (Otx2 knockout tetON-Otx2 − DOX) and 23 (Otx2 knockout tetON-Otx2 + DOX) spheroids. χ2 test; * * * P < 0.0001. g, Immunostaining of wild-type tetON-Otx2 mES cells overexpressing Otx2 (addition of DOX). h, Lumen formation in cells from g (and their corresponding controls) and in Otx2 knockout mES cells cultured with 2i/LIF and with or without Otx2 overexpression. n = 20 (wild-type), 20 (wild-type tetON-Otx2 − DOX), 50 (wild-type tetON-Otx2 + DOX), 19 (Otx2 knockout), 21 (Otx2 knockout tetON-Otx2 –DOX) and 20 (Otx2 knockout tetON-Otx2 + DOX) spheroids. χ2 test; * P = 0.0196; * * * P < 0.0001. i, j, Nanog (i) and Otx2 (j) intensity in cells from g as a function of their ability to undergo lumogenesis. n = 25 spheroids per condition. Mann–Whitney U-test; * * * P < 0.0001. k, mRNA levels of naive pluripotency genes in wild-type and Otx2 knockout mES cells at different time points after removal of 2i/LIF. n = 2 independent samples per group. ANOVA; * P = 0.02; * * P = 0.0045; * * * P = 0.0002 (Klf4); * * * P = 0.0001 (Tbx3). l, Immunostaining of wild-type and Otx2 knockout mES cells overexpressing GFP–Podxl. m, Lumen formation in cells from l. n = 66 (wild-type) and 61 (Otx2 knockout) spheroids. χ2 test; NS, not significant. n, ChIP–seq analysis of Oct4, Otx2 and H3K27ac in mES cells (top) and after 48 h of transition into epiblast-like cells (EpiLCs, bottom). Shaded area indicates an enhancer within the first intron of Podxl that is activated during cell fate transition. o, Immunostaining of GFP–Podxl overexpressing mES cells. p, Lumen formation in cells from o. n = 30 (+ 2i/LIF clone 1), 32 (− 2i/LIF clone 1), 33 (+ 2i/LIF clone 2) and 34 (− 2i/LIF clone 2) spheroids. χ2 test; * * * P < 0.0001. Scale bars, 10 μm; arrows indicate lumens; (b, c, e, g, l, o). M, Matrigel.

Extended Data Figure 8. Cgn is induced upon naive pluripotency exit and mediates Rab11 vesicle tethering to the apical membrane.

a, mRNA levels of Cgn in mES cells. n = 5 (24 h), 3 (36 h) and 4 (48 h) independent samples per group. Unpaired Student’s t-test; * P = 0.0431 (gelatin 24 h); * P = 0.0347 (Matrigel 24 h); * * * P = 0.0003 (gelatin 36 h), * P = 0.0126 (Matrigel 36 h); * * P = 0.0075 (gelatin 48 h); * P = 0.0139 (Matrigel 48 h). b, c, Immunostaining of mES cells cultured in Matrigel with or without 2i/LIF. Squares indicate magnified regions. d, mRNA levels of naive pluripotency genes and early post-implantation factors in control and Cgn knockdown mES cells. n = 3 independent samples per group. Unpaired Student’s t-test; * * * P < 0.0001; NS, not significant. e, Immunostaining of Rex1::GFPd2 mES cells in control and Cgn knockdown mES cells. Scale bars, 10 μm (b, c, e). Matrigel (M).

Extended Data Figure 9. Characterization of pluripotency gene expression and epiblast morphogenesis in post-implantation human embryos.

a, Immunostaining of day 9–10 human embryos. Dotted lines indicate the epiblast and the arrow indicates the amniotic cavity. Scale bars, 50 μm and 10 μm (magnified areas). b, NANOG intensity in embryos from a. n = 4 (IVC control with lumen), 3 (IVC control without lumen) and 6 (IVC +5i/LAF) human embryos. Mann–Whitney U-test; * P = 0.0381. c, Three-dimensional reconstruction of the epiblast (based on the KLF17 staining) and the amniotic cavity (based on the PODXL staining) of embryos from Fig. 5b. d, Immunostaining of day 9–10 human embryos. n = 2 embryos per group. Scale bars, 50 μm. e, Immunostaining of day 9–10 human embryos. Scale bars, 50 μm. f, Number of apoptotic cells per embryo in embryos from e. n = 3 (IVC control) and 5 (IVC + 5i/LAF) embryos per condition. Mann–Whitney U-test; * P = 0.0179.

Extended Data Figure 10. Exit from naive pluripotency is required for lumenogenesis in hES cells.

a, Immunostaining of WIBR3 Δ PE–OCT4–GFP cells cultured in primed FBS/KSR/bFGF2 medium or N2B27 2i/hLIF with or without DOX. DOX addition induces the expression of NANOG and KLF2. Scale bars, 20 μm. b, c, mRNA levels of pluripotency genes in cells from a. n = 3 independent samples per condition. ANOVA; * * * P < 0.0001 (NANOG and KLF2); * * * P = 0.0001 (TFCP2L1); * * P = 0.0013 (DNMT3L). d, Immunostaining of WIBR3 Δ PE–OCT4–GFP cells cultured in Matrigel in primed or naive conditions. Arrows indicate lumens. Scale bars, 20 μ m. e, Lumen formation in cells from d. n = 38 (2i/hLIF + DOX), 33 (2i/hLIF − DOX), 34 (FBS/KSR/bFGF2) and 20 (mTESR) spheroids. χ2 test; * * * P < 0.0001. f–h, mRNA levels of pluripotency genes and early post-implantation factors in cells from i. n = 4 (NANOG, KLF2, TFCP2L1, DNMT3L, KLF4 and PODXL) and 3 (KLF17) independent samples per group. Kruskal–Wallis test; * * * P = 0.0002 (NANOG); * * * P < 0.0001 (KLF2, DNMT3L, KLF4, PODXL); * * P = 0.0058 (TFCP2L1); * P = 0.0205 (KLF17). i, Immunostaining of WIBR3 Δ PE–OCT4–GFP cells cultured in primed mTESR medium or naive conditions (2i/hLIF + DOX, 5i/LAF and RSet). Scale bars, 10 μm. j, Immunostaining of hES cells cultured in 3D Matrigel with different naive and primed conditions. Arrow indicates lumen. Scale bars, 10 μm. k, Lumen formation in cells from j. n = 31 (2i/hLIF + DOX), 27 (5i/LAF), 33 (RSet) and 20 (mTESR) spheroids. χ2 test; * * P = 0.0019; * * * P < 0.0001. l, Immunostaining of hES cells cultured in 3D Matrigel with different naive and primed conditions. Arrows indicate polarized centrosomes. Scale bars, 5 μm. m, Angle between the nucleus–centrosome axis and the nuclear–nuclear axis in cells from l. Each dot represents an individual centrosome. n = 53 (2i/LIF + DOX), 51 (5i/LAF), 40 (RSet) and 42 (mTESR) centrosomes. Kruskal–Wallis test; NS, not significant. n, Immunostaining of naive hES cells cultured in 3D Matrigel with mTESR. The initial naive conditions in which the cells were cultured are indicated. Arrows indicate lumens. Scale bars, 10 μm. o, Lumen formation in cells from n. n = 30 (2i/hLIF + DOX 48 h), 30 (2i/hLIF + DOX 72 h), 30 (5i/LAF 48 h), 29 (5i/LAF 72 h), 30 (RSet 48 h) and 30 (RSet 72 h). M, Matrigel.

Figure 1. Epiblast gene expression at peri-implantation.

a, Immunostaining of mouse embryos (top). Dotted lines indicate the epiblast; arrowheads indicate polarized Podxl; asterisks indicate Podxl in the primitive endoderm; white long arrows show positions used to plot intensity profiles (bottom). b, Principal component analysis of all samples by all expressed genes. Numbers in brackets indicate percentage of variance. c, Heat map showing expression of core (black), naive (green) and post-implantation (purple) genes. Genes with significantly high (*) or low (#) expression in E4.5–E4.75 compared to E5.0 cells are indicated. P<0.1. d, Lumen formation in embryos from f. n = 19 (IVC1) and 15 (IVC1 +2i/LIF) embryos. χ² test; ***P = 0.0007. e, Nanog intensity in embryos from f. n = 19 (IVC1) and 15 (IVC1 +2i/LIF) embryos. Unpaired Student’s t-test; ***P < 0.0001. f, Immunostaining of in vitro cultured mouse embryos. Dotted lines indicate the epiblast. Arrow indicates lumen. Scale bars, 20 μm (a, f).

Figure 2. Naive mES cells initiate polarization in Matrigel.

a, Experimental set-up. A, analysis. b, Immunostaining of mES cells cultured as indicated in a. Scale bars, 5 μm. c, Centrosome positions in cells from b. n = 60 (+2i/LIF) and 62 (−2i/LIF) centrosomes. Unpaired Student’s t-test; NS, not significant. d, Nanog intensity in cells from b. n = 30 (+2i/LIF) and 31 (−2i/LIF) spheroids. Unpaired Student’s t-test; ***P < 0.0001. e, Immunostaining of mES cells cultured as indicated in a. Squares denote magnified regions. Scale bars, 10 μm. f, Experimental set-up. g, Immunostaining of mES cells cultured as indicated in f. Scale bars, 50 μm. h, mES cell–embryo chimaeras generated using H2B–GFP mES cells cultured as indicated in f. Scale bars, 40 μm. i, Immunostaining of mES cells. Scale bars, 5 μm. j, Centrosome positions in cells from i. n = 52 (+Gö6983) and 56 (−Gö6983) centrosomes. Mann–Whitney U-test; NS, not significant. Arrows denote polarization. M, Matrigel.

Figure 3. Naive pluripotency exit is required for lumenogenesis.

a, Experimental set-up. A, analysis. b, c, Immunostaining of mES cells cultured as indicated in a. Squares denote magnified regions and the arrow in c non-polarized cells. d, Lumen formation in cells from b and c. n = 40 (+2i/LIF 48 h), 40 (−2i/LIF 48 h), 21 (+2i/LIF 72 h) and 22 (−2i/LIF 72 h) spheroids. χ² test; ***P < 0.0001. e, Immunostaining of control and Dgcr8 knockout (KO) mES cells. Arrow indicates lumen. f, Lumen formation in cells from e. n = 29 (control +2i/LIF), 31 (control −2i/LIF), 31 (Dgcr8 knockout +2i/LIF) and 31 (Dgcr8 knockout −2i/LIF) spheroids. χ² test; NS, not significant. g, Nanog intensity in cells from e. n = 20 (control +2i/LIF), 20 (control −2i/LIF), 19 (Dgcr8 knockout +2i/LIF) and 21 (Dgcr8 knockout −2i/LIF) spheroids. Kruskal–Wallis test; ***P < 0.0001; *P < 0.05, NS, not significant. h, Immunostaining of control and Oct4 knockdown mES cells. Arrows indicate lumens. i, Lumen formation in cells from h. n = 39 (control), 53 (Oct4 siRNA with Oct4) and 32 (Oct4 siRNA without Oct4) spheroids. χ test; ***P < 0.0001. Scale bars, 10 μm (b, c, e, h). M, Matrigel.

Figure 4. Vesicle fusion downstream of naive pluripotency exit.

a, Immunostaining of GFP–Rab11a and GFP–Rab11a(S25N) overexpressing mES cells. Arrows indicate lumens. Scale bars, 10 μm. WT, wild-type. b, Lumen formation in cells from a. n = 35 (Rab11a wild-type 48 h), 60 (Rab11a wild-type 72 h), 39 (Rab11a(S25N) 48 h) and 29 (Rab11a(S25N) 72 h) spheroids. χ² test; ***P < 0.0001. c, Immunostaining of heterozygous (HET) and Podxl knockout (KO) mES cells. Arrow indicates lumen. Scale bars, 10 μm. d, Lumen formation in cells from c. n = 34 (HET), 32 (clone 1), 30 (clone 2) and 21 (clone 3) spheroids. χ² test; *P = 0.0245; **P = 0.0019; ***P < 0.0001. e, Immunostaining of control and protamine sulfate-treated mES cells. Arrow indicates lumen. Scale bars, 10 μm. f, Lumen formation in cells from e. n = 41 (control) and 40 (protamine sulfate) spheroids. χ² test; ***P < 0.0001. g, Immunostaining of GFP–Podxl overexpressing mES cells. Squares denote magnified regions. Scale bars, 10 μm. M, Matrigel. h, Immunostaining of mouse embryos. Dotted lines indicate the epiblast, arrows indicate polarized Cgn. Scale bars, 20 μm. i, Immunostaining of control and Cgn knockdown mES cells. Arrow indicates lumen. Scale bars, 10 μm. j, Lumen formation in cells from i. n = 40 (control) and 39 (Cgn siRNA) spheroids. χ² test; ***P < 0.0001.

Figure 5. Naive pluripotency exit direct lumenogenesis in human embryos and hES cells.

a, Immunostaining of day 6–7 human embryos. Squares denote magnified regions, yellow arrowhead indicates the epiblast and white arrowhead indicates the primitive endoderm. Scale bars, 30 μm. b, Immunostaining of day 9–10 human embryos. Dotted lines indicate the epiblast; arrowhead indicates the amniotic cavity; white long arrows show the position used to plot intensity profiles (d). Scale bars, 50 μm and 10 μm (magnified areas). c, Lumen formation in embryos from b. n=9 (IVC control) and 11 (IVC +5i/LAF). χ² test; *P = 0.0134. d, PODXL intensity profiles in embryos from b. e, KLF17 intensity in embryos from b. n = 4 (control with lumen), 5 (control without lumen) and 11 (IVC +5i/LAF) embryos. Unpaired Student’s t-test; *P = 0.0366. f, Immunostaining of hES cells. Arrow indicates lumen. Scale bars, 10 μm. g, Lumen formation in cells from f. n = 31 (2i/hLIF +DOX), 30 (5i/LAF), 30 (RSet) and 30 (mTESR) spheroids. χ² test; ***P < 0.0001. h, Model summarizing the findings of the study.