Data reporting
No statistical methods were used to predetermine sample size. The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.
Mouse embryo recovery
Mice were maintained in accordance with national and international guidelines. All experimentation involving animal subjects was approved by the Institutional Animal Care and Use Committee at Yale School of Medicine and conducted following the approved animal handling protocol (protocol no: 2023-20352). All experimental mice were maintained in specific pathogen-free conditions on a 12 h:12 h light:dark cycle temperature-controlled facility (between 20–26 °C and humidity 30–70%) with free access to water and food, and used from 6 weeks of age. All mice were bred to a mixed CD1 albino background. V. Greco provided the H2B-GFP:Tcf-LEF51 reporter mouse line.
To collect in vivo embryos at gastrulation stage, 5- to 7-week-old female CD1 mice were naturally mated with 12- to 24-week-old male CD1 mice and sacrificed 6.5, 6.75, 7.0 or 7.25 days post coitum. Uteri were recovered and embryos were dissected from deciduae in DMEM medium containing 5% FBS, 20 mM HEPES, and 1× Penicillin-Streptomycin (Gibco) warmed up to 37°. The sex of the embryos was not determined.
Embryo culture at gastrulation
Embryos were cultured in a 1:1 ratio of DMEM F/12 (Gibco) to rat serum (Envigo), with 1× Glutamax (Gibco), 0.2× Penicillin-Streptomycin (Gibco), and 0.2× MEM NEAA (Gibco) in 37 °C at 20% O2 and 5% CO2. Medium was equilibrated in 5% CO2 for ≥15 min prior to culture.
Tetraploid complementation
Mouse embryos at the pre-implantation 2-cell stage (E1.5) were recovered in KSOM medium by flushing the oviduct, from 5- to 6-week-old CD1 females that were superovulated by injection of 10 IU pregnant mare serum gonadotropin (ProSpec) followed by 10 IU human chorionic gonadotropin (Sigma) after 48 h and were mated with CD1 males. Two-cell-stage embryos were then fused to induce tetraploidy using BTX Embryo Manipulation Electro Cell Fusion System. Fused embryos were transferred to advanced KSOM (Sigma) covered with mineral oil (FUJIFILM) and cultured to blastocyst stage until embryonic stem cell injection. Each host tetraploid blastocyst was injected with 10–15 mouse ES cells, the injected blastocysts were transferred into uterus of the prepared day 2.5 CD1 pseudopregnant surrogates (6 weeks old) which were plugged by vasectomized CD1 male mice. These mouse embryonic stem cell-derived embryos were collected at the certain developmental timeline for subsequent experiments by uterine dissection (as described above).
Metabolism and signal modulation experiments
Chemical inhibitors were administered as follows except where otherwise stated: 2 mM 2-DG52 (Santa Cruz Biotechnology), 20 µM BrPA (Santa Cruz Biotechnology), 10 µM PD0325901 (StemCell Technologies), 10 µM SU5402 (StemCell Technologies), 200 µM galloflavin53 (Cayman Chemical), 100 µM oligomycin54 A (Santa Cruz Biotechnology), 5 µM 6-AM55 (Cayman Chemical), 10 µM YZ98 (Cayman Chemical), 5 µM shikonin56 (Cayman Chemical) and 5 µM azaserine57 (Cayman Chemical or Santa Cruz Biotechnology). Embryos were treated for 4 h, 7 h, 12 h or 18 h. Mesoderm explants were treated for 3 h, 12 h or 16 h. Stem cells were treated for the duration of the experiment as indicated in the figure captions. The proteoglycan sulfation inhibitor NaClO3 (Sigma) was applied at 20 mM. FGF2, FGF4 and FGF8 (all Peprotech) were applied at 50 ng ml−1.
Nutrient-sparse and rescue experiments
Nutrient-sparse medium was prepared using an Advanced DMEM/F-12 medium devoid of d-glucose, l-serine, l-glutamine and sodium pyruvate (Caisson Laboratories) with the addition of 25 mM sodium bicarbonate (Sigma), 0.2× penicillin-streptomycin (Gibco) and 0.2× MEM NEAA (Gibco). This was supplemented with the following nutrients, according to each experimental condition: 17 mM glucose (Gibco), 0.5 mM sodium pyruvate (Gibco), and/or 1× Glutamax (Gibco). For galactose rescue experiments 20 mM galactose (Sigma) supplemented to the culture medium. For GlcNAc rescue experiments 1 mM or 2 mM GlcNAc (Sigma) supplemented to the culture medium. All embryos were cultured in a 1:1 ratio of customized DMEM and rat serum (Envigo).
Cell culture, gastruloids and in vitro mesoderm-directed-differentiation assay
All mouse embryonic stem cells were cultured at 37 °C in 20% O2 and 5% CO2 and passaged once they had reached 80% confluency. Cells were routinely tested for mycoplasma contamination by PCR. Mouse embryonic stem cells were cultured on gelatinized tissue-culture–grade plates in FBS-containing DMEM medium with 2i/LIF (1 μM MEK inhibitor PD0325901, 3 μM GSK-3 inhibitor CHIR99021 and 10 ng ml−1 LIF). FBS-containing DMEM (Gibco) medium comprised of: 18% inactivated FBS (Gibco), 1.2× penicillin-streptomycin (Gibco), 1.2× Glutamax (Gibco), 1.2× MEM NEAA (Gibco), 1.2 mM sodium pyruvate (Gibco), and 120 µM 2-mercaptoethanol (Gibco). The ERK-KTR mouse embryonic stem cells used in this study were sourced from Simon et al.34 (JAX 035333).
For in vitro mouse gastruloid experiments, we followed the protocol as described20,21. Gastruloids were treated with indicated drugs culture day between 3 and 4 and fixed for subsequent HCR staining or qPCR assay, as described in Dingare et al.7.
For in vitro mesoderm-directed differentiation experiments, we followed the protocol as described58, with some modifications. Cells were passaged (day 0) after reaching 80% confluency into 6-well culture plates or 8-well µ slides (Ibidi) at a density of 20,000 per cm2 in FBS-containing DMEM medium with 10 ng ml−1 Fgf2 (R&D Systems). Cells were treated at day 1 with 10 ng ml−1 FGF2, at day 2 with 10 ng ml−1 FGF2 and 5 µM CHIR99021 (StemCell Technologies), and at day 3 and 4 with 5 µM CHIR99021. Drug treatments added to the culture at least 12 h after FGF2 treatment was started on day 1, or on when cells transitioned to mesoderm state on day 3.
In situ mesoderm explant assay
For mesoderm isolation from E7.25 mouse embryos, we followed a modified protocol as described59. Embryos were collected as described above. Extraembryonic tissue proximal to the amnion was removed, and the cup-shaped embryo was transferred to a dissociation medium consisting of 0.5% Trypsin EDTA (Gibco) and 2.5% Pancreatin (Thermo Scientific Chemicals) for 15 min at 4 °C. Embryos were washed with collection medium (described earlier) and mesoderm tissue was dissected using insect pins (Roboz) attached to syringes. Explants were transferred to fibronectin-coated 18-well µ slides (Ibidi) plates and left to adhere for 3–4 h in medium containing a 1:4 ratio of rat serum to embryo culture medium (described earlier), prior to downstream experiments and/or chemical inhibitor treatments. For live imaging of migration dynamics, explants were imaged every 5 min up to 3 h. For immunofluorescence staining, explants were cultured up to 17 h and fixed in PBS containing 4% paraformaldehyde (PFA) for 20 min. For RNA sequencing, explants were cultured to 27 h then scraped off with sterile insect pins and flash-frozen.
Glucose-uptake assay
Mouse embryos were collected at early streak, mid streak and late streak stages of gastrulation and cultured with 1 mM 2-NBDG60 (Cayman Chemicals) for 2 h. For multiphoton microscopy, embryos were immediately live-imaged. For confocal microscopy, embryos were fixed in PBS containing 4% PFA for 15 min then immediately imaged after a 5 min wash in PBS-T (PBS with 0.05% Tween-20).
DQ gelatin assay
Cells underwent mesoderm-directed differentiation on 8-well µ slides (Ibidi) as previously described. On day 2.5 when cells are at the EpiSC stage, cells were cultured with differentiation medium that included 50 µg ml−1 DQ gelatin (Invitrogen), then were fixed on day 3.5 in PBS containing 4% PFA for 20 min, protected from light. Cells were incubated in blocking buffer (described below) with DAPI (overnight at 4 °C or 20 min room temperature) prior to imaging.
Invadopodia assay
18-well µ slides (Ibidi) were prepared with FITC–Fibronectin (Sigma-Aldrich) and 0.1% gelatin following the manufacturer’s 2-day protocol and protected from light. Explants were cultured on these plates for 16 h prior to fixation in PBS containing 4% PFA for 20 min. Explants were incubated in blocking buffer (described below) with DAPI (overnight at 4 °C or 20 min room temperature) prior to imaging.
Labelling active mitochondria
Mouse gastrulas were collected at ES stage and cultured with either 200 nM MitoTracker Deep Red or 100 nM tetramethylrhodamine, methyl ester, perchlorate (TMRM). Embryos were cultured in the respective dyes for 30 min and then imaged in a humidified chamber with 37 °C and 5% CO2 according to the same parameters described below for live imaging.
Immunofluorescence staining
Samples were fixed in PBS containing 4% PFA for 20-45 min room temperature, or in methanol for 20 min at 4 °C where stated. After fixation, samples were washed twice with PBS-T (PBS with 0.05% Tween-20) and permeabilized in PBS with 1 mM glycine and 0.3% Triton X-100 for 20–60 min at room temperature. Primary antibody incubations took place overnight at 4 °C in blocking buffer (PBS containing 10% fetal bovine serum (FBS), 10% Tween-20). Samples were washed twice with PBS-T prior to secondary antibody incubations at 4 °C in blocking buffer. On the final day, samples were washed twice with PBS-T, then transferred into PBS-A droplets (PBS with 0.75% Bovine Albumin Fraction V (Gibco)) and covered with mineral oil (Sigma-Aldrich) in 35 mm glass-bottom dishes (MatTek) before confocal imaging. All PBS-T washes were done for 10 min room temperature, and all room temperature incubations or washes took place on a rocking platform. All antibodies used in this study are listed in Supplementary Table 1.
Image data acquisition and processing
Samples were imaged with the Leica STELLARIS 5 microscope using a HC PL APO CS2 40×/1.10 or a HC FLUOTAR L 25×/0.95 W VISIR 0.17 water objectives, a z-spacing of 0.75 µm to 5 µm and appropriate laser and filters for Alexa 405, Alexa 488, Alexa 546 and Alexa 633 or combinations thereof. To correct for fluorescence decay along the z axis during embryo imaging, ‘z-compensation by AOTF and PMT’ was defined in a control embryo and applied across all experimental conditions during that imaging session, so that changes in laser power and gain across the z-stack were equivalent across conditions, and only adjusted to each embryo’s size. Raw data were processed using open-source image analysis software Fiji/ImageJ2 2.9.0 or AIVIA 10.5.1 AI Image Analysis Software and assembled in Photoshop 2021 22.3.1 (Adobe) or Illustrator 2024 28.6 (Adobe). Transverse views were generated from 1 µm z-spaced images using ImageJ’s orthogonal viewer. Digital quantifications and immunofluorescence signal intensity graphs were obtained using plot profile measurements in Image J and visualized in GraphPad Prism10.2.3. software.
Time-lapse live imaging
Confocal time-lapse imaging of embryos, mesoderm explants, and mesoderm-differentiated cell cultures were performed using Leica STELLARIS 5 microscope using a 25× (HC FLUOTAR L 25×/0.95 W VISIR 0.17) or 40× (HC PL APO CS2 40×/1.10) water objective and appropriate laser/filters for Alexa 488, Alexa 546 and Alexa 633 or combinations thereof. Samples were imaged under a humidified chamber with 37 °C and 5% CO2. Explants were imaged at 5 min intervals in 2 µm z-spaced planes for up to 4 h on pre-treated ibidi dishes (Ibidi). Images were processed using AIVIA, described below under ‘Image analysis’.
Multiphoton live imaging for NAD(P)H autofluorescence
Multicolour two-photon microscopy was used for live-embryo imaging of NAD(P)H dynamics. Embryos kept in ex vivo culture medium (described earlier) with the addition of 20 mM HEPES. Images were acquired with LA Vision TriM Scope II (LaVision Biotec) laser scanning microscope equipped with a Chameleon Vision II and Discovery ultrafast lasers (Coherent) for different wavelengths imaged sequentially after each z-section15. Wavelengths of 750 nm were used for NAD(P)H and 940 nm for H2B-GFP:Tcf-LEF or ERK-KTRmClover reporter. Although excitation ranges of NAD(P)H and GFP overlapped, their emission was separated by band pass filters: blue range (425–475 nm; NAD(P)H) and green range (500–550 nm; FAD). Exclusion of nuclear GFP signal (in H2B-GFP:Tcf-LEF embryos) from NAD(P)H channel validated separation of emission signals of NAD(P)H and GFP. Embryos were imaged using a 40× water immersion lens (Nikon; NA 1.15) at 400 Hz with pixel size of 0.2 µm or 0.3 µm and a z-step of 1 or 1.5 µm. For optimal signal-to-noise ratio, Line averaging of 2 was done for all NAD(P)H images. The fluorescence detected from the reduced metabolites through this method captures both NADH and NADPH (thus called NAD(P)H)61. It should be noted that the intracellular concentrations of the non-phosphorylated NADH and NAD+ is much higher than NAD(P)H and NADP+ (ref. 62). We further validated that NAD(P)H fluorescence signal in the embryos closely co-localized and followed 2-NBDG uptake.
Image analysis
All embryos were positioned exactly along the A–P axis (refer to figures) during imaging for easier quantifications.
Angle of GLUT expression in the epiblast
Mid-embryo sagittal sections of embryos stained for GLUT1 and GLUT3 and with DAPI were used for glucose-uptake quantification in Fiji/ImageJ2. For each embryo, the angle vertex was allocated at the proximal-most boundary between epiblast and extraembryonic ectoderm, at the mid-point between posterior-most (0°) and anterior-most (180°) epiblast. Two angle values were calculated for every embryo (‘GLUT start’ and ‘GLUT end’) to capture the range of observable GLUT expression in the epiblast, along with the primitive streak distal length of the embryo to allocate Theiler staging (ES, MS or LS, as described above). ‘GLUT start’ and ‘GLUT end’ means were calculated for each stage and visualized in a rose diagram with RStudio.
Primitive streak distal elongation percentage
Mid-embryo sagittal sections of DAPI-stained embryos were used for primitive streak distal elongation quantification in Fiji/ImageJ2. For each embryo, the 0% cut-off was marked by anterior epiblast morphology, the 100% cut-off was assigned to the distal-most epiblast cell, and ‘PS’ (primitive streak) was marked at the distal-most point where the primitive streak morphology extends (Fig. 2b). Measurements were obtained for the vertical distance between 0% and 100% and between 0% and ‘PS’, so that each embryo’s primitive streak elongation percentage (0-to-PS divided by 0-to-100) is adjusted to its size. Thus, a control embryo’s primitive streak distal elongation percentage of 95% (Fig. 2b) can be interpreted as ‘this embryo has elongated its primitive streak to 95% of its epiblast height’. Theiler stages could be assigned with this measurement: ≤50 for ES; between 50 and 100 for MS; >100 for LS.
Basement membrane breakdown
Mid-embryo sagittal sections of laminin and DAPI-stained embryos were used for basement membrane calculations in Fiji/ImageJ2. For each embryo, the 0% cut-off was marked by anterior epiblast morphology, the 100% cut-off was assigned to the distal-most epiblast cell, and ‘intact BM’ was marked at the distal-most point where laminin was still intact (Fig. 2d). Measurements were obtained for the vertical distance between 0% and 100% and between 0% and ‘BM’, so that each embryo’s basement membrane breakdown percentage (0-to-BM divided by 0-to-100) is adjusted to its size. Thus, a control embryo’s basement membrane breakdown readout of 96% (Fig. 2d) can be interpreted as ‘this embryo has broken down 96% of its vertical basement membrane length’.
FITC–fibronectin invadopodia assay
z-sections of DAPI-stained mesoderm explants imaged with a 40× objective were used for fibronectin perforation calculations, with ≥3 images captured per explant. For every image, cell number and perforation number were calculated. Invadapodia degradation is represented as a percentage (perforation number divided by cell number).
ERK-KTR nuclear:cytoplasmic ratio quantification
Sagittal sections of ERK-KTR embryos were used to quantify ERK activity. Manual segmentations were drawn for each cell in Fiji/ImageJ2 to delineate nuclear and cytoplasmic areas, using ERK-KTR and brightfield channels to verify cell morphologies (Extended Data Fig. 8b). For every cell, the following measurements were quantified: nuclear area na, cytoplasmic area ca (including the nucleus), nuclear ERK-KTR intensity ni, and nuclear-subtracted cytoplasmic ERK-KTR intensity ci. To calculate the nuclear:cytoplasmic ratio, the average nuclear intensity (ni/na) was divided by the average cytoplasmic intensity (ci/(ca − na)), such that a ratio ≥1 indicates no ERK activity.
Proliferation quantification
Live-imaged videos captured with a 40× objective were used for proliferation counts of TCF-LEF reporter mesoderm explants, quantified manually in Fiji/ImageJ2. For each explant, an overall cell count for ‘starting population’ was calculated on the first frame, and a ‘cell division’ event (TCF-LEF telophase observation) was also assigned by careful frame-to-frame assessment over the course of the video. A proliferation index (cell division number/starting population) was then obtained for each explant, such that a highest index of 1 can be interpreted as ‘every cell at the beginning of the video has divided by the end of the video’. This was then adjusted to the total frame length of the video so that explants across different experimental replicates and video lengths could be compared. Thus, a control explant with a readout of 16.5% (Fig. 4b) can be interpreted as ‘16.5% of the mesoderm cells at the beginning of the video have divided by the end of the video’.
AIVIA-based image analysis
AIVIA 10.5.1 AI image analysis software was used to examine mesoderm explant migration dynamics. Nuclear segmentations were generated using a ‘cell-tracking’ recipe, applying a pixel classifier that was trained on TCF-LEF nuclear fluorescent channels from videos of each treatment group in every experimental replicate (Fig. 4b). Parameter values were modified between rounds of pixel classifier ‘previewing’ and training to ensure nuclear segmentation accuracy. Every track was manually examined against the brightfield channel to verify detection accuracy. Incorrect lineages were corrected in the ‘track editor’ or discarded from the final dataset, and parent and daughter tracks were treated independently. Every track measurement of interest was exported to Excel and normalized to its detection length (‘first frame’ subtracted from ‘last frame’) so that lineages of different detection lengths could be compared. Thus, all data points are plotted in µm min−1 or µm s−1.
Directed-differentiated mesoderm cell counting
Imaris 10.0.1 software was used to count T-positive, total, and pyknotic cells. Nuclear segmentations were generated using the ‘Surfaces’ function. Parameters such as background subtraction and morphological splitting were adjusted using the Creation Wizard tool to identify the optimal algorithm settings. Pyknotic cells were counted in the DAPI channel using an increased threshold for background subtraction and a smaller diameter for segmentation than when counting healthy cells using the DAPI signal. Surfaces were examined relative to the original staining throughout creation and in their final state to ensure accuracy.
Western blot
Protein was extracted using RIPA buffer (Thermo Scientific 89900). The reagents were supplemented with 1× protease inhibitor (Thermo Fisher Scientific, 87786) and 1× phosphatase inhibitor (Phos Stop Roche 4906845001) to prevent protein degradation and dephosphorylation during extraction. The protein concentrations were quantified using Pierce BCA. protein assay kit (ThermoFisher, 23225) following the manufacturer’s instructions. The protein was denatured at 95–100 °C for 5 min in 1× NuPAGE LDS sample buffer (ThermoFisher, NP0007) containing 10% v/v β-mercaptoethanol (Millipore Sigma, M6250). The protein electrophoresis was performed using 4–20% precast polyacrylamide gels (Bio-Rad, 4568095) in 1× Tris/Glycine/SDS buffer (diluted from Bio-Rad, 1610732) at 110 V for 1 h for all proteins. The proteins were transferred from the gel onto a nitrocellulose membrane (ThermoFisher, IB301032) by wet transfer for 1 h at 100 V using the Bio-Rad Mini Trans-Blot Cell following the manufacturer’s instructions. Wet transfer buffer was 20% methanol, 200 mM glycine, and 250 mM Tris. The membrane was blocked in 5% milk in TBST for 1 h at room temperature with gentle shaking. Primary antibodies were incubated with the membrane at recommended concentrations at either 1 h room temperature or 4 °C overnight with gentle shaking. The membrane was washed three times in TBST for 5 min each, followed by incubation with horseradish peroxidase-conjugated secondary antibodies diluted at 1:5,000 ratio in 5% non-fat milk in TBST for 1 h at room temperature. The blots were then washed three times in TBST for 5 min each. Blots were treated with SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher, 34094) for 2–5 min at room temperature. The blots were imaged using a CCD camera-based imager ProteinSimple FluorChem E system.
qPCR with reverse transcription
Total RNA was extracted from cells using an RNeasy Micro Kit per the manufacturer’s instructions (Qiagen). cDNA synthesis was performed with 1 μg of total RNA using a High-Capacity cDNA Reverse Transcription Kit according to the manufacturer’s instructions (Applied Biosystems). The amounts of mRNA were measured using the PowerUp SYBR Green PCR Master Mix (Applied Biosystems). Relative levels of transcript expression were assessed by the ΔΔCt method, with Gapdh as an endogenous control. For primers used in qPCR with reverse transcription, see Supplementary Table 2.
RNA-sequencing sample collection and data analysis
Total RNA was extracted from mesoderm explants using a PicoPure RNA Isolation Kit per the manufacturer’s instructions (Applied Biosystems). RNA samples were submitted to the Yale Center for Genome Analysis for quality assessment, library preparation and sequencing. Each RNA sample contained explants from 2 embryos, and 2 samples were submitted per condition. Paired-end reads were aligned using Star (v2.7.9a) to the mouse genome (GRCm38) using the Ensembl transcriptome (release 109)63. Analysis of differential gene expression was performed using DESeq2 (v1.40.1). To identify differentially regulated genes between samples for downstream analyses, we selected genes with log2-transformed fold change greater than 0.7 or less than −0.7 and an adjusted P value < 0.01. Gene ontology analysis was performed using topGO (v2.52.0) using the ‘weight01’ algorithm for Fisher’s exact tests on R studio (version 2023.12.1 + 402). KEGG pathway analysis was performed using DAVID (https://david.ncifcrf.gov/tools.jsp)64,65.
Reanalysis of single-cell sequencing data
Previously published single-cell RNA sequencing data from gastrulating mouse embryos were accessed using MouseGastrulationData (v1.12.0) (https://github.com/MarioniLab/MouseGastrulationData). Data were subset to include only ‘epiblast’, ‘primitive streak’ and ‘nascent mesoderm’ cell states from all samples staged E6.5 to E8.5 (samples 1–10, 12– 20 and 23– 37; samples 11, 21 and 22 were omitted from our analysis because their staging was ambiguous, for example, a sample labelled as ‘mixed gastrulation’). Counts were log-normalized using Seurat (v. 4.3.0)66,67,68,69,70. Cells were first randomly down-sampled for visualization purposes to 400 cells per cell type. Pseudotime and trajectory interference analyses were performed using Slingshot (v2.8.0)71 for principle curve calculation; SingleCellExperiment (v1.22.0)72 and scater (v.1.28.0)73 were used for visualization of gene expression over pseudotime. The principle curve was then traced over the first two principal components to infer pseudo-temporal organization.
Statistics and reproducibility
Statistical tests were performed on GraphPad Prism 10.2.3 software. For comparison of two groups, two-sided unpaired t-tests were applied. For comparisons of three or more groups, one-way ANOVAs were used, followed by Dunnett’s multiple comparison tests for comparisons to control or Tukey’s multiple comparison tests for comparisons between all groups, unless otherwise stated. P values are displayed on figure panels or in legends. All the experiments were performed at least in three biological replicates unless specifically described in the methods and the figure legends. Figure legends indicate the number of embryos, cells, explants or stem cell structures and, when relevant, the number of experiments performed for each analysis. Statistical power calculations were not used to determine sample size.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
