Membrane Tension Gates ERK-Mediated Regulation of Pluripotent Cell Fate

1The integration of mechanics with cell signaling has a crucial impact on cell function. A number of studies have highlighted the importance of sensing the mechanical properties of the extracellular environment in developmental morphogenesis, immunity, tumorigenesis, and wound healing. The influence of extracellular mechanical cues is also emerging as a key player in tissue homeostasis and stem cell function and fate choice. Most of these studies have focused on how the mechanics of the environment, in particular substrate stiffness, affects cell function.

2At the same time, several studies suggest that changes in cell shape correlate with changes in signaling and fate. For example, it was shown that serum response factor signaling is regulated by spreading in epidermal stem cells, and Stat3 signaling is linked to spreading in embryonic stem cells. It was also shown that the extent of spreading may affect fate choice in mesenchymal stromal cells. Importantly, cell shape is controlled by forces exerted by the cytoskeleton on the cell surface; thus, it is a direct readout of intrinsic cell mechanics (subsequently cell mechanics, reviewed in and ). Therefore, cell mechanics and substrate stiffness could both affect signaling. However, cell and substrate mechanics are often coupled. As a result, studying the effects of cell mechanics on cell function is challenging, and the mechanisms by which the intrinsic mechanical properties of a cell could affect signaling remain elusive.

3An excellent model system to investigate mechanical signaling in the context of a well-defined fate transition is embryonic stem cells (ESCs). Mouse ESCs possess the ability to produce all tissues, a property called pluripotency. These cells can be maintained indefinitely in a “naive” state and ushered through distinct phases of pluripotency, including the highly characterized first stage of differentiation, exit from naive pluripotency. The signaling requirements of exit from naive pluripotency are well understood; the process has been shown to be primarily driven by fibroblast growth factor/extracellular signal-regulated kinase (FGF/ERK) signaling. Interestingly, it was recently shown that ESCs are not sensitive to external mechanical cues, yet ESCs display a striking shape change, from round to spread, as they exit naive pluripotency, indicative of changes in cell mechanics. Thus, mouse ESCs constitute a particularly appropriate model system to dissect the role of cell mechanics in signaling in a well-defined fate transition.

4Here, we show that the shape change observed in early ESC differentiation is driven by a reduction in effective membrane tension. We demonstrate that this change in membrane tension is regulated by the key pluripotency factor β-catenin. We further show that the decrease in membrane tension accompanying early differentiation leads to the increased endocytosis of FGF signaling components. This results in the increased activity of ERK, which is necessary for exit from naive pluripotency. Our study unveils a key mechanism whereby changes in cellular mechanics regulate the signaling that drives fate transitions.

5To study the interplay between cell shape and cell fate change, we asked whether cell shape correlates with fate during the early stages of exit from naive pluripotency. ESCs were cultured on gelatin-coated plates and maintained in N2B27 medium supplemented with the mitogen-activated protein kinase kinase (MEK)/ERK inhibitor PD0325901, the glycogen synthase kinase 3 (GSK3) inhibitor/β-catenin agonist CHIRON (CHIR), and leukemia inhibitory factor (the culture medium known as 2i+L). Exit from naive pluripotency was initiated by withdrawing the inhibitors and culturing cells in N2B27 medium alone. After 24 h in N2B27 (T24), cells displayed a mixture of shapes, with most cells already flat and spread, and some still round and in colonies (Figures 1A and S1A). This heterogeneity is consistent with previously described asynchrony in exit from naive pluripotency. To identify a potential correlation with cell fate, we fixed the cells and, in cells of different shapes, quantified the expression of Nanog, a marker of naive pluripotency, and Otx2, a transcription factor upregulated in cells that have exited naive pluripotency. We found that Nanog levels were significantly higher and Otx2 significantly lower in round cells as compared to spread cells (Figures 1B and 1C), thus establishing a correlation between cell shape and exit from naive pluripotency.

6We next characterized the dynamics of cell spreading during exit from naive pluripotency. To do this, we captured time-lapse images of ESCs stably expressing LifeAct-GFP to visualize filamentous actin (Video S1). We observed that, before spreading, the cells displayed a phase of intense blebbing (Figures S1B–1D). Notably, blebbing is often a sign of low membrane-to-cortex attachment, a key mechanical parameter controlling the apparent tension of the plasma membrane. Apparent membrane tension (subsequently referred to as “membrane tension”) depends on the in-plane tension of the lipid bilayer and on membrane-to-cortex attachment, and is a measure of the resistance of the plasma membrane to deformations. As such, it is a major regulator of cell shape. Furthermore, in our scanning electron microscopy (SEM) images T48 cells displayed more surface membrane folds and reservoirs compared to naive cells (Figures 1A and S1A), suggesting differences in membrane tension.

7We next asked whether exit from naive pluripotency is associated with a change in membrane tension. To address this, we used a tether pulling assay using optical tweezers (Figures 1D, 1E, and S1E; Video S2), in which the force exerted on the bead by the membrane tether (the “trap force”) is a direct readout of membrane tension. We found that ESCs significantly decrease their membrane tension as they exit naive pluripotency (Figure 1E). Moreover, at T24, when cell populations display mixed morphologies, round cells had a significantly higher membrane tension than blebbing and spread cells. This is consistent with previous studies showing that cell spreading is facilitated by decreased membrane tension. Therefore, we conclude that cell spreading during exit from naive pluripotency occurs concomitantly with a decrease in plasma membrane tension.

8Membrane tension has been shown to be regulated largely by linkers between the plasma membrane and the underlying actomyosin cortex, such as ezrin-radixin-moesin (ERM) and myosin I proteins. To narrow the list of potential regulators of membrane tension, we checked the expression of genes encoding for linker proteins in ESCs and in their in vivo counterpart, the pre-implantation epiblast, using a published dataset. We found that ERM proteins, and in particular Ezrin, were up to 10 times more expressed than myosin I proteins. We thus focused on ERM proteins, which are activated by phosphorylation. At the population level, we found that the level of phosphorylated ERM (pERM) was sharply decreased after 2i+L removal (Figures 1F and S1F). We confirmed these results using immunostaining of T16 cells, which showed that spread cells have lower levels of pERM than round cells (Figures S1G–S1J).

9We then investigated which pluripotency-regulating signaling pathway is primarily responsible for the decrease in pERM upon removal of ESC media. We reduced the medium to the minimal signaling environment necessary to maintain naive pluripotency (2i). We then separately removed PD0325901 (PD03) and CHIR from 2i to study the effects of MEK/ERK activation and GSK3b activation, respectively. We found that, while PD03 removal did not lead to a pERM and decrease, CHIR removal resulted in a rapid and significant decrease in pERM and decrease in membrane tension (Figures S2A–S2C), pointing to a role for GSK3b signaling in regulating pERM levels. Given that increased GSK3b activation leads to β-catenin degradation, and that β-catenin is partly localized at the plasma membrane, we asked whether β-catenin depletion would lead to a decrease in pERM and subsequent decrease in membrane tension. To address this question, we measured pERM levels and membrane tension in β-catenin knockout (KO) ESCs and found that both were significantly lower compared to wild-type (WT) ESCs (Figures 2A and 2B). These results suggest that GSK3b-driven β-catenin degradation mediates a decrease in pERM, which in turn controls membrane tension and cell shape during exit from naive pluripotency.

10We next investigated the link between β-catenin and ERM phosphorylation. Given that there is a known link between RhoA activity and the cadherin-catenin complexes and that active RhoA is one of the key activators of ERMs, we speculated that RhoA may connect β-catenin to ERM. To test this hypothesis, we measured RhoA activity during exit from naive pluripotency using an active fluorescence resonance energy transfer (FRET) reporter and an active RhoA pull-down assay and found that ESCs displayed a significant decrease in RhoA activity during exit from naive pluripotency (Figures 2C–2F and S2D). Furthermore, our results show that the levels of active RhoA are significantly lower in β-catenin KO cells compared to WT ESCs (Figures 2E and 2F).

11We then used a doxycycline-inducible ESC line expressing a phosphomimetic, constitutively active form of RhoA to probe whether higher RhoA activity does indeed lead to higher ERM phosphorylation, and thus higher membrane tension in ESCs. After doxycycline induction, we found that inducible, constitutively active RhoA (iRhoA_CA) cells showed high levels of ERM phosphorylation and subsequently maintained high membrane tension during exit from naive pluripotency (Figures 2G and 2H). Our findings indicate that β-catenin degradation by GSK3 leads to a decrease in RhoA activity, which drives a subsequent decrease in ERM phosphorylation and membrane tension, suggesting that RhoA activity represents a key link between β-catenin and ERM regulation.

12We then asked whether the membrane tension decrease plays a role in controlling exit from naive pluripotency. We used both the iRhoA_CA cells and a doxycycline-inducible ESC line expressing a phosphomimetic, constitutively active form of ezrin (iEZR_CA). After induction, iEZR_CA cells showed high EZR_CA expression and no growth defects (Figures S3A and S3B). After withdrawal of the naive factors (2i+L), iEZR_CA and iRhoA_CA cells did not spread (Figures 3A and S3C), even at T48, and they maintained high membrane tension similar to ESCs (Figures 2H and 3B). We also found that, compared to controls, iEZR_CA and iRhoA_CA cells at T24 maintained significantly higher levels of Nanog and failed to efficiently upregulate Otx2 (Figures 3C, 3D, S3C, and S3D). A clonogenicity assay indicated that unlike control T48 WT cells, iEZR_CA and iRhoA_CA cells placed back in 2i+L after 48 h in N2B27 were able to survive and form naive colonies with the same efficiency as naive ESCs (Figures 3E and S3E; see Method Details). To confirm the early differentiation phenotype, we performed RNA sequencing (RNA-seq) comparing iEZR_CA and WT ESC, T24, and T48 cells. We found that iEZR_CA cells showed a significant impairment both in loss of naive pluripotency markers and subsequent expression of early implantation markers (Figures 3F, 3G, and S3I; Table S1). These results indicate that maintaining a high membrane tension in pluripotent cells significantly inhibits their ability to exit naive pluripotency.

13Furthermore, treating cells with methyl-β-cyclodextrin, a cholesterol-sequestering compound that increases membrane tension (; Figures S3J and S3K), delayed exit from naive pluripotency (Figure S3L). Conversely, treating cells with NSC 668394, an ezrin inhibitor that decreases membrane tension (; Figures S3J and S3K), caused cells to exit naive pluripotency more efficiently than control cells (Figure S3L). Moreover, a recent study showed that using a purely mechanical linker between actin and the plasma membrane to increase effective membrane tension also blocks exit from naive pluripotency (, this issue of Cell Stem Cell).

14To assess the developmental relevance of these findings, we used a gastruloid assay (adapted from ) in which we compared the ability of iEZR_CA to form gastruloids at different stages (Figures 3H–3K). We found that iEZR_CA gastruloids displayed clear morphological defects, failing to elongate and adopt the characteristic morphology of properly progressing gastruloids with distinct multiaxial organization. The lack of elongation suggests major developmental defects in these gastruloids. We also performed morula injection of iEZR_CA and control cells at the 8-cell stage and maintained embryos in culture for 2 or 3 days, resulting in embryos that are the equivalent of developmental stages embryonic days (E)4.5 and E5.5, respectively. We consistently observed a higher level of the naive pluripotency marker Nanog in the iEZR_CA compared to native and control cells (Figures S4A and S4B), thus suggesting that high membrane tension also leads to a delay in differentiation in the embryo.

15Our data strongly suggest that a pERM-controlled decrease in plasma membrane tension is essential for cell spreading and for exit from naive pluripotency.

16To clarify the relative importance of cell shape change and membrane tension, we used micropatterning to control the extent of cell spreading with disks of different diameters (small: 25–50 μm or large: 100 μm). We seeded ESCs at equal densities onto both small and large micropatterns in N2B27 only. We found that in contrast to cells plated on large patterns or non-patterned surfaces, spatially constrained T24 cells on small patterns did not change shape and maintained a rounded morphology similar to ESCs (Figures 4A and 4B). Spatially constrained cells at T24 maintained high levels of Nanog, suggesting defects in early differentiation (Figures 4B, 4C, and S4C). However, spatially constrained cells also maintained a high membrane tension (Figure 4D). Strikingly, the exit from naive pluripotency on small micropatterns, where cells cannot spread, was rescued by knockdown of ERM (Figures 4E and S4D–S4G) or knockdown of myosin I family proteins (Figures S4H and S4I), treatments previously shown to reduce membrane tension. Together, these experiments point to the existence of positive feedback loops between membrane tension decrease and cell spreading, and indicate that it is the decrease in membrane tension, rather than cell spreading, that is required for efficient exit from naive pluripotency.

17To further test whether membrane tension regulates exit from naive pluripotency directly or as a result of cell spreading, we plated ESCs on either laminin or E-cadherin, where they display spread morphologies even in 2i+L (Figure 4F). Despite the fact that for cells on laminin and E-cadherin exit from pluripotency was associated with a much less pronounced spreading increase than for cells on gelatin, we still observed a similar drop in membrane tension during exit from pluripotency (Figure 4G). This further indicates that it is the decrease in membrane tension, and not the spreading directly, that is needed for efficient dismantling of naive pluripotency.

18We then asked how decreased membrane tension facilitates exit from the ESC state. We speculated that the facilitation may be through endocytosis, which is a major regulator of signaling events and has been shown to be regulated by membrane tension. We assessed global endocytosis levels using a fluid phase uptake assay, in which we cultured cells in the presence of either fluorescent dextran or a pH-sensitive fluorescent dextran (Figure 5A). Using this assay, we found that T24 cells displayed a significant increase in endocytosis levels once spread (Figures 5B, 5C, S5A, and S5B). We next interfered with this increase using drugs known to inhibit endocytosis (dynasore, pitstop2, and chlorpromazine hydrochloride). We found that WT cells treated with these drugs failed to increase their endocytosis rates during exit from naive pluripotency (Figure 5D). We next assessed whether cells with endocytosis defects could exit naive pluripotency normally and found that these cells maintained a higher level of Nanog and a lower level of Otx2 expression compared to controls (Figures 5E–5G). These observations suggest that endocytosis regulates exit from naive pluripotency.

19We next examined the link between the observed differentiation-induced decrease in membrane tension and endocytosis. We showed that endocytosis was considerably reduced in cells expressing iEZR_CA (Figures 5H and 5I) and that it was increased in β-catenin KO ESCs (Figure S5C), suggesting that high membrane tension antagonizes endocytosis, which is consistent with previous findings in other systems. To test whether enhanced endocytosis could overcome the block in differentiation imposed by EZR_CA, we overexpressed Rab5a, a master regulator of early endosomes, in iEZR_CA cells. The elevation of Rab5a levels has been shown to promote endocytosis. We found that overexpressing Rab5a increased endocytosis levels in iEZR_CA cells compared to non-transfected controls, both in ESCs and T24 cells without leading to growth defects (Figures 5H, 5I, and S5D) and without lowering membrane tension (Figure S5E). Notably, Rab5a expression rescued the early differentiation defect phenotype in iEZR_CA cells, restoring Nanog downregulation (Figures 5J and 5K) and decreasing the number of colonies surviving in a clonogenicity assay, which is indicative of more efficient exit from naive pluripotency (Figure S5F). Furthermore, transfection with Rab5a also enabled the downregulation of Nanog protein in spatially constrained T24 cells (Figure S5G). These results show that increasing endocytosis levels via Rab5a overexpression allows ESCs to exit naive pluripotency notwithstanding a high membrane tension.

20Finally, we asked whether a specific signaling pathway related to the regulation of naive pluripotency was affected by the changes in cell mechanics and endocytosis that we observed. The most obvious candidate was FGF/ERK, as it is key for exit from naive pluripotency, and its EGF/ERK counterpart been shown to be regulated by endocytosis in cancer cells. We thus quantified pERK levels during exit from naive pluripotency and observed that at all time points ERK activity was highly attenuated in the iEZR_CA cells (Figures 6A, 6B, S6A, and S6B). We also confirmed a global reduction in mitogen-activated protein kinase (MAPK)/ERK signaling in iEZR_CA cells by RNA-seq analysis of known transcriptional targets (Figure S6C). These results strongly suggest that maintaining high membrane tension during exit from naive pluripotency prevents ERK activation. To further test the role of ERK activity in the early differentiation impairment of the iEZR_CA cells, we treated iEZR_CA cells with the p90 ribosomal S6 kinase (RSK) inhibitor BI-D1870, which acts as an ERK activator. We found that this treatment was sufficient to rescue the early differentiation defects of iEZR_CA (Figures 6C and 6D), which we further confirmed using a clonogenicity assay (Figure S6D).

21The internalization of receptors by endocytosis has been shown to influence the levels of pERK activity. Thus, we measured the internalization of FGF receptor 1 (FGFR1), which is upstream of pERK in ESCs, iEZR_CA, and control cells using immunofluorescence. We found that iEZR_CA cells had less internalized FGFR1 compared to controls (Figures S6E and S6F), despite expressing similar protein levels (Figure S6G). We were able to increase the levels of internalized FGFR1 in iEZR_CA by overexpressing Rab5a (Figures S6E and S6F). We then specifically probed pERK activity at endosomes using a recently described FRET sensor reporting on ERK activity at early endosomes (Figure 6E; ). Using live imaging, we found that pERK at endosomes sharply increased as cells spread (Figures 6E–6G and S6H; Video S3). We confirmed this with measurements in fixed cells, finding that spread T24 cells showed a higher FRET ratio compared to round T24 cells. Notably, we observed a lower FRET ratio in T24 cells, where endocytosis rates were reduced chemically with dynasore or by increasing membrane tension via iEZR_CA expression (Figure 6H). Furthermore, increasing endocytosis independently of membrane tension by expressing Rab5a in iEZR_CA cells, resulted in a FRET ratio similar to spread control cells at T24 (Figure 6H). Thus, our results indicate that ERK activity, which helps disassemble naive pluripotency, is induced by membrane tension-regulated endocytosis in cells exiting pluripotency (Figure 6I).

22There is extensive interplay between substrate mechanics, cell mechanics, and cell function. Moreover, cell shape, which is a readout of cell mechanics, is correlated with cell fate. Because cell shape, cell mechanics, and cell fate are deeply intertwined, it is challenging to disentangle them to understand specifically how cell mechanics mediates signaling. To probe the role of cell mechanics in cell fate transitions, we focused on exit from naive pluripotency in ESCs. Interestingly, we found that it is the change in cell surface mechanics, not the cell shape, that regulates early differentiation. We found that membrane tension decreases as ESCs exit naive pluripotency, and prohibiting ESCs from spreading inhibited both membrane tension reduction and early differentiation. Decreasing membrane tension in constrained cells was sufficient to rescue the differentiation defect. These results suggest an important feedback between cell shape and cell surface mechanics during the early differentiation of ESCs but indicate that here it is cell surface mechanics that constitutes the primary mediator of cell fate transitions.

23Our results also suggest a previously unknown synergy between essential pluripotency regulators β-catenin and FGF/ERK. This synergy demonstrates clear feedback between cell signaling and surface mechanics. We show that β-catenin regulates membrane tension, which in turn gates downstream signaling activity by regulating endocytosis. We speculate that β-catenin regulates membrane tension by modulating RhoA activity via the cadherin-catenin complex, which has been shown to regulate various Rho GTPases. In this model, upon the initiation of early differentiation, degradation of β-catenin by GSK3 leads to a decrease in RhoA activity and subsequent decrease in ERM phosphorylation/membrane tension. Ultimately, our results suggest a role for β-catenin in maintaining naive pluripotency by preserving high membrane tension, which would thus suppress endocytic FGF/ERK signaling in naive cells.

24Previous studies have pointed to correlations between mechanical changes and endocytosis, with links to developmental processes. Furthermore, osmotic changes, which among other activities affect membrane tension, have been linked with the endocytosis of receptors, leading to the inhibition of signaling pathways during transdifferentiation of muscle progenitors. In contrast, our study demonstrates that in ESCs, FGF signaling is activated by increased endocytosis as a result of changes in cell surface mechanics. By unveiling a specific pathway activated as a result of a change in membrane tension, our study draws a direct link between cell mechanics, endocytic signaling, and fate choice.

25Our findings point to plasma membrane tension as a central mediator of cell signaling. As the plasma membrane defines the boundary between the inside and outside of the cells, it is an ideal cellular element to integrate environmental cues and mechanical changes into changes in cell signaling. Furthermore, we speculate that observed effects of substrate mechanics on stem cell fate regulation in other systems may be due in part to the resulting changes in membrane tension and downstream endocytosis rate. Going forward, it will be important to assess whether and how the stiffness of the microenvironment can affect membrane tension. Given that many morphogenetic processes in development are accompanied by significant shape changes, it will also be interesting to further explore this connection in vivo. Our work unveils a mechanism directly connecting a specific change in cell mechanics to the activation of a key biochemical signaling pathway. The interplay between mechanics, membrane tension, and endocytic signaling is likely to be an important regulator of cell fate in various developmental and pathological contexts.

26A limitation of our study is the lack of quantitative three-dimensional (3D) cell shape measurements. In our study, we classified cells by shape using two categories: round and spread. Quantitative cell shape measurement may allow us to identify an intermediary state between round and spread. Also, as described in this study and others, differentiation is an asynchronous process. As such, the bulk assays (e.g., Western blots, RNA-seq) used in our investigation may lead to averaging out transient mixed populations.

27Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kevin J Chalut, kc370@cam.ac.uk.

28This study did not generate new unique reagents

29The accession number for the RNA sequencing data reported in this paper is GEO: GSE159433 (accessible through: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE159433)

30ES-E14TG2a (referred as WT) directly derived from mouse embryos were a kind gift from the Austin Smith’s laboratory (Cambridge Stem Cell Institute). The LifeAct GFP cell line is from. These cells have been generated and kindly given to us by Ian Rosewell (Francis Crick Institute, under Holger Gerhardts project license). Dox-inducible IRES mCherry Ezrin_CA & RhoA_CA ES cells were kindly given to us by Ayaka Yanagida from the Chalut and Nichols’s lab. The EZR_CA was engineered using a PiggyBac system with E14 ES cells. The H2B-mCherry cells were kindly given to us by Tim Lohoff from the Nichol and Reik’s lab.The β-Catenin Knockout cell line was kindly given to us by the lab of Austin Smith and has been described in.

31Mice used were intercrosses of CD1 (Charles River). All embryos used in this study were obtained from natural mating. Embryo staging was based on the assumption that, on average, mating occurred at midnight so that at midday, the embryos were assigned E0.5. Embryos were flushed at the relevant stages from oviduct (eight-cell stage embryos) using flushing and holding a medium (M2, Sigma). The sex of embryos and the ages of mice using mating were not concerned in this study. This research has been regulated under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 following ethical review by the University of Cambridge Animal Welfare and Ethical Review Body (AWERB). Use of animals in this project was approved by the ethical review committee for the University of Cambridge, and relevant Home Office licenses (Project license No. P76777883) are in place.

32Mouse embryonic stem cells were cultured on Falcon flasks coated with 0.1% gelatine. On occasion, laminin (Sigma-Aldrich #11243217001) coated dishes were used instead of gelatine. ES cells used were E14TG2a. Cells were cultured in N2B27+2i+LIF (2i+L) at 37°C with 7% CO2. Cells were passaged every other day using Accutase (Sigma-Aldrich, #A6964) and regularly tested for mycoplasma. Culture medium was changed every 24h. The culture medium was made using DMEM/F-12, 1:1 mixture (Sigma-Aldrich, #D6421-6), Neurobasal medium (Life technologies #21103-049), 2.2 mM L-Glutamin, B27 (Life technologies #12587010), 3 μM Chiron (Cambridge Bioscience #CAY13122), 1 μM PD0325901 (Sigma-Aldrich #PZ0162), 20 ng/ml of LIF (Merck Millipore # ESG1107), 50 mM β-Mercapto-ethanol, 12.5 ng.mL-1 Insulin zinc (Sigma-Aldrich #I9278) and home-made N2. The 200 X home-made N2 was made using 0.791 mg.mL-1 Apotransferrin (Sigma-Aldrich #T1147), 1.688 mg.mL-1 Putrescine (Sigma-Aldrich #P5780), 3 μM Sodium Selenite (Sigma-Aldrich #S5261), 2.08 μg.mL-1 Progesterone (Sigma-Aldrich #P8783), 8.8% BSA. Exit from naive pluripotency was triggered by passaging ~500 000 cells and seeding them in N2B27 only (N2B27) on T25 gelatine coated flasks.

33Transfections of plasmids were performed using 5 μg of plasmid and 3.5 μL of Lipofectamine 2000 (ThermoFischer Scientific #11668019), incubated in 250 μL OptiMEM for 5 minutes, then mixed and incubated at room temperature for 30 minutes, and added to cells passaged onto Ibidi dishes with polymer coverslip bottom for observations (IBIDI Scientific, 81156).

34Due to the lack of fluorescent tag, transfection efficiency of Rab5a was assessed by performing immunofluorescence against Rab5a of un-transfected and transfected cells. Average fluorescence level was determined in transfected and untransfected cells by measuring average fluorescence intensity using Fiji. Cells with an average fluorescence higher than 20% of the average of untrasnfected cells were counted as being positively transfected. Using this method, we estimate Rab5a transfection efficiency to be ~72%. siRNA transfections were performed using 30 μM of siRNA and 5 μL of Lipofectamine RNAiMAX (ThermoFischer Scientific #13778075) incubated in 250 μL OptiMEM for 5 minutes, then mixed and incubated at room temperature for 30 minutes and added to cells passaged onto micropatterns.

35Trap force measurements were performed using a home-built optical tweezer using a 4W 1064 nm Laser Quantum Ventus with a 100x oil immersion objective (NA 1.30, CFI Plan Fluor DLL, Nikon) on an inverted microscope (Nikon Eclipse TE2000-U) equipped with a motorized stage (PRIOR Proscan). The optical tweezer was calibrated following. Measurements were performed using concanavalin-A coated (50 μg/ml) carboxyl latex beads (1.9μm diameter, Thermo Fisher C37278). Beads were incubated on a shaker with concanavalin-A for one hour prior to the experiment. Bead position was recorded every 90 ms in bright field prior and during tether formation. The trap force was calculated based on the calibration of the trap and the bead position using a home-made Fiji plugin (HDB) (typical values for the trap stiffness were k~0.130 pN/nm.)

36Cells were seeded in 6 well dishes in different conditions (N2B27 or 2i+L) and cultured for different periods of time, as indicated. RNA was extracted using a QIAGEN (205310) kit. Extracted RNA cDNA was generated from the RNA according to kit’s instructions (Thermo Fischer Scientific #4368814). For the RT-qPCR, pre-designed primers (see table of primers used), and the SYBR Green Master Mix (QIAGEN; 204141) were combined according to the kit’s instructions. qPCR was performed on Bio-Rad CFX qPCR.

37Cells were plated into 6-well plates in different conditions (with or without 2i+L), as indicated. Proteins were extracted on ice using Laemmli buffer (Bio-Rad #1610747). Samples were put at 95°C for 5 min and then sonicated on ice. Isolated whole protein content was measured using the Thermo Scientific Pierce 660nm Protein assay (#22660) using a BSA gradient kit (Bio-Rad; 500-0206) and 20 μg of protein extract was loaded onto either NuPAGE 4%–12% Bis-Tris protein gels (Thermo Fischer Scientific #NP0321BOX) or Novex WedgeWell 14% Tris-Glycine Mini Gels (Thermo Fischer Scientific #XP00140BOX) depending on the size of the protein of interest. Gels were run at 125 V for 2 h. Proteins were transferred for 70 minutes at 70 V to a nitrocellulose membrane (Thermo Fischer Scientific #88024). Membranes were blocked for 90 min using Odyssey blocking buffer (Li-Cor #927-50003). Primary antibodies were added at the proper dilution Table S1 in either TBS-T + 5% milk or 5% BSA and incubated overnight at 4 degrees. Membranes were washed three times for 15 min in TBS with TWEEN. Infrared species-appropriate secondary antibodies were then added Table S1 for one hour. Membranes were washed three times with PBS-T for 15 min and imaged on the Li-Cor Odyssey scanner. Bands were quantified using LI-COR Biosciences Image Studio Lite.

38Cells were fixed in IBIDI dishes (IBIDI Scientific, 81156) with 4% formaldehyde. Cells were permeabilized during 10 min in PBS+0.1% Triton-X. Blocking was then done using with 2% FBS 2% BSA in PBS with 0.1% Triton for 45 min. Cells were then incubated for 1h30min with the primary antibody diluted at the appropriate concentration in the same buffer as used for blocking. Cells were then washed 3 times 5 min with PBS+0.1% Triton. Secondary antibodies were added for 1 h diluted at 1:800 in the same buffer as used for blocking and primary antibody incubation. Cells were washed with PBS + 0.1% Triton 3 times for 5 min. Finally, cells were incubated for 5 min with PBS+DAPI before a final rinse in PBS.

39Library preparation was done by in-house facility us NEBNext® Multiplex Oligos for Illumina. Paired-end sequencing was performed using the Novaseq sequencing platform, yielding 517 million reads (single-lane). Reads were trimmed and adapters removed using FastQC and TrimGalore (Babraham Bioinformatics). Mouse genome build GRCm38/mm10 was used to align reads with GSNAP version 2015-09-29. Genes were annotated using Ensembl release 81 and read counts were quantified using HTSeq. Differential expression analysis was computed using DESeq2 a design containing an interaction term (Time point:CellType).

40For gene ontology analysis, the significant interaction terms at 24hrs and 48hrs were selected (i.e., differently regulated between iEZR_CA cells and wild-type celld during exit from naive pluripotency), using thresholds of base mean > 5 and an absolute log2 fold change > 1. DAVID 6.8 was used to compute the statistical enrichment of Gene Ontology terms. For each category (KEGG Pathway and Cellular Component), the list of enriched terms was filtered using Benjamini-Hochberg false-discovery rate correction (p_adj < 0.1). Additionally, only terms with fold enrichment > = 2 and at least 1% of annotated genes regulated were selected. The top 5 (ordered on increasing adjusted p value) are plotted for each of the four functional annotations category.

41Highly variable genes were determined by averaging, for each gene, the intra-group variance (3 replicates per group, 6 experimental groups). The intra-group (or technical) variance was then compared to the distribution of variance computed over 3 randomly selected samples (inter-group variance). Highly variable genes were those for which the technical variance was significantly lower than the inter-group variance.

42Glass coverslips (Marienfekd GmBH) were sonicated in 1 M HCl for 5-15 minutes, washed and plasma-treated for 30 s for glass passivation. Then 0.1 mg/mL of PLL-g-PEG (PLL(20)-g[3.5]-PEG(2)/Atto663 from SuSoS) solution was added to the coverslips for 30 min at room temperature. Chrome MASK from Delta Mask was used. The mask was illuminated with UV light for 5 min on the chrome side, then the PEG-coverslips were added on the chrome side and illuminated for 6 min to burn the PEG in the shape of the desired pattern (circles in our case). Coverslips were then dried and stored. Coverslips were rehydrated before use.

43For immunofluorescence assays, we seeded the cells onto the patterns (coated with laminin, see above for details) in 2i+L at a density allowing for the formation of small colonies, occupying the whole available space on the small circles while still having free space on the bigger circles. Cells were then washed three times in PBS and placed in N2B27 only without 2i+L. After 24h in N2B27, cells were fixed and immunostained.

44ES cells (five to eight cells per embryo) were injected into eight-cells embryos via a laser-generated perforation in the zona pellucida using XYClone (Hamilton Thorne Biosciences). Injected embryos were cultured in N2B27 with or without 1 μg/ml Dox for two-days for eight-cells injection or one-day for blastocyst injection, an equivalent of E4.5 blastocysts at 37°C and 5% CO2.

45Embryo and cell manipulations were carried out under a dissecting microscope (Leica Microsystems). The zona pellucida was removed using acid Tyrode’s solution (Sigma) at E4.0. E4.5 blastocysts were subjected to immunosurgery as previously described. In brief, blastocysts were incubated for 45-60 minutes in a 1:5 dilution of anti-mouse rabbit serum (Sigma) in N2B27, washed in N2B27 and further incubated for 30-60 minutes in a 1:5 dilution of rat serum (in-house) in N2B27 for the complement reaction. The ICM was subsequently cleaned from residual trophectoderm with a narrowly fitting glass pipette. Isolated ICMs were culture in N2B27 at 37°C and 5% CO2 with or without 1 μg/ml Dox an equivalent to E5.5 in which PrE lineage cells surround the matured EPI.

46RhoA pulldown assay was performed using the Cytoskeleton Inc pull down kit (#BK036) and carefully following the provided instructions. In a nutshell, the assay is using beads coated with the Rho binding domain of Rhotekin, a rho effector protein which bind specifically to the GTP-bound RhoA. His tagged RhoA protein and GTPγS were used as controls.

47Embryos or one-day cultured isolate ICMs were fixed with 4% paraformaldehyde (PFA; Thermo Fisher Scientific) in Phosphate buffered saline (PBS; Sigma) at room temperature for 15 minutes. Then, the samples were rinsed in PBS containing 3 mg/ml polyvinylpyrrolidone (PBS/PVP; Sigma), permeabilised with PBS/PVP containing 0.25% Triton X-100 for 30 minutes. Blocking was performed with a buffer comprising PBS containing 0.1% bovine serum albumin (BSA; Sigma), 0.01%Twenn20 (Sigma) and 2% donkey serum at 4°C for 2-3 hours. Primary antibodies (See Table S1) were diluted in blocking buffer, and samples were incubated in the appropriate antibody solution at 4°C overnight. They were rinsed three times in blocking buffer for 15 minutes ~each. Secondary antibodies were diluted in blocking buffer with 500 ng/ml 4’,6-diamidino-2-phenylindole (DAPI; Invitrogen) and samples were incubated in the appropriate antibody solution at room temperature for one hour in the dark. They were rinsed three times in blocking buffer for 15 minutes ~each. Whole staining process was performed on a microwell mini Tray (Nunc). Embryos or cultured isolated ICMs were put in a small drop of blocking buffer on poly-D-lysine (Sigma) coated glass-bottom dishes (MatTek) covered with mineral oil (Sigma), and their images were acquired using a Leica TCS SP5 confocal microscope.

48Gastruloids produced as described before. Briefly, mouse ES cells were detached from gelatin-coated vessels using accutase for 4 min (Sigma, A6964). The iEZR_CA ES cells were cultured with doxycycline for 24 h or 48 h before aggregation and during the following protocol. Detached cells were then washed once with warm N2B27 and twice with warm PBS (Sigma, D8537). In between the washes, the cells were pelleted for 3 min at 170 x g. Subsequently, the cells were resuspended to fresh N2B27 and the number of cells counted using an automated cell counter (Merck, Scepter 2.0, PHCC20040). The cell suspension was diluted so that 40 ul of suspension had 300 cells. 40 ul of suspension was then transferred to low attachment 96 well round-bottomed plates (Sigma, M3562) for 48 h of aggregation. Next, 150 ul of fresh N2B27 medium with 3 uM CHRI99021 (Cambridge Bioscience, CAY13122) was pipetted to the wells for 24 hours. After which 150 ul of fresh N2B27 medium was changed to the wells daily until 120 hours.

49WT and iEZR_CA organoids were imaged using a cell culture microscope (Olympus, CKX53SF) and 10x air objective (Olympus, NA 0.25) mounted with a digital camera (Canon, EOS 250D). The resulting images were analyzed with ImageJ Fiji distribution. Briefly, the images were opened as an image stack and converted to 8-bit grayscale images. Next, the grayscale images were thresholded manually to best segment the organoid outlines. The resulted binary masks were then measured using Fiji build in analyze particles function. “Feret’s diameter” and “shape descriptors” were recorded and analyzed.

50Cells were seeded on IBIDI dishes (IBIDI Scientific, 81156) a few hours before imaging. Cells were then incubated for 10 min in N2B27 or 2i+L with 10 μg of either pH-Rhodo Dextran (pHrodo Red Dextran, 10,000 MW, for Endocytosis, Thermofischer Scientific) or Alexa Fluor Dextran (Dextran, Alexa Fluor 647; 10,000 MW, Thermofischer Scientific). Cells were then rinsed immediately prior to imaging. Each dish was imaged for 30 min.

51For quantifications with either pH Rhodo or Regular Dextran, total intensity inside the cell was measured with Fiji by using a sum z projection. The signal was corrected by taking a ROI of the background of a similar size as of the measured cells. The background signal was then subtracted from the signal from cell. Only single cells or groups of cells with clear boundaries were quantified. CellMask Deep Red (Invitrogen # C10046) was used to identify individual cell boundaries. Experiments were always performed side by side with cells in 2i+LIF as control. Data are normalized to cells in 2i+L.

52Imaging was performed on an inverted confocal microscope (Olympus FV1200), at 37°C with 5% CO2 for live imaging, taking z stacks 10 μm in total height with a 1 μm step using a 63x objective. For FRET quantifications, we used a home-written (HDB) custom Macro in Fiji FRET and donor channel signals were corrected for background.

53For the FRET sensor for RhoA activation, a sum-projection was performed before FRET quantification. We then measured average ratio in each cell by using three circular ROIs (of 1 μm diameter) in the region of positive FRET ratio (thus at the cell cytoplasm or cortex and not in the nucleus for example). We then averaged the obtained data for each cell.

54For the FRET sensor for ERK activation in the endosome, vesicles were semi-automatically identified and average fluorescence was quantified in each vesicle across all the channels. Small Z maximum projections (over 3 z slices of 1 μm) were quantified in order to ensure capture entire endocytotic vesicles. We then calculated the ratio of values obtained in donor and FRET channels to obtain FRET ratio following and methods.

55To test for speed and efficiency of exit from pluripotency, replating assays were performed. For each experimental condition, ~400 000 cells were plated onto 6-well dishes coated with 0.1% gelatin, in N2B27 to induce exit from naive pluripotency. After 48h in N2B27, the cells were resuspended and counted. A specific number of cells (typically 400) was then replated in N2B27+2i+L onto a 12-well plate. After 4 days, the number of colonies was manually counted; as only naive cells survive in 2i+L, this assay quantifies the efficiency of pluripotency exit. As a positive control, cells cultured in 2i+L were replated in 2i+L. Occasionally replated cells were tested for naive pluripotency using a Blue-color AP staining kit to add another control that replated cells were indeed naive pluripotent (Cambridge Bioscience #AP100B-1).

56Samples were fixed with glutaraldehyde 1%. Preparation for scanning electron microscopy was performed as in without membrane extraction. Jeol 7401 - high resolution Field Emission Scanning Electron Microscope was used for imaging.

57Rab5a and FRET plasmids were provided by Giorgio Scita’s lab

58Ezr_GFP and EZR_CA_GFP were provided by Guillaume Charras

59FRET RhoA sensor was obtained from Adgene (#12150) and is characterized in )

60Live imaging was performed on an inverted confocal microscope (Olympus FV1200), at 37°C with 5% CO2, taking z stacks 10 μm in total height with a 1 μm step using a 63x objective and using low laser power. Fix imaging was performed on an inverted confocal microscope (Leica TCS SP5) taking z stacks 20 μm in total height with a 1 μm step using a 63x objective. All image analysis were performed using Fiji.

61To measure Nanog and Otx2 levels, we used DAPI to locate the cell nuclei, identified the nuclei mid plane and make a ROI of the nucleus. We then used this ROI to measure average Nanog or Otx2 intensity per nucleus, thus controlling for cell/nuclear size in measuring transcription factor expression.

62For pERM measurements by IF, we used a sum-projection. We then identified ROIs in the cell (using the cortical actin identified via the phalloidin channel (Alexa Fluor 568 Phalloidin (ThermoFischer Scientific # A12380)) as boundary for the ROI) and measured the integrated signal (the mean intensity signal∗number of pixels in the selection) to compensate for differences in cell size. Additionally, we corrected the measured values by subtracting background using the same ROI as for the cell but in a region with no cell present. Data were normalized to round cells. signal was used to determine cell boundaries and to classify cells according to shape based on visual inspection.

63For Fgfr1 quantifications, average fluorescence was measured in each cell using CellMask Deep Red (Invitrogen # C10046) as a marker to locate individual cells. Average intensity was then manually measured in the midplane of the cells by taking a ROI following the cell boundary.

64For #blebs in Figure S1, the number of blebs was manually counted at every time point (every 20 minutes) across a z stack using LifeAct and brightfield as guides to identify individual blebs (note that only actin filled blebs could be observed with this method).

65For all statistical analysis, PRISM 7 (Graphpad software, Inc) was used. Statistical details can be found in the legend of each figure. D’agostino and pearson test was used to test for normality. N represents number of independent biological replicates. Pooled independent experiments are used in dot plots.