a) Immunofluorescence stainings for pERK and TBXT and their radial intensity profiles in 2D gastruloids at different times. Error bars in graphs are standard deviations over N=4 colonies. b,c) Effect FGF receptor inhibition (FGFRi, b) or MEK inhibition (MEKi, c) on pERK and TBXT in colonies fixed at 42h after BMP treatment. Inhibition for either 30 minutes at 41.5h (rounded to 42 in figure) or throughout differentiation. d) BMP only control and FGFRi inhibition in cells expressing doxycycline-inducible SOScat with and without addition of doxycycline. e-f) Cell numbers (e) and radial profiles of TBXT positive cells (f) for conditions in (d). Error bars represent standard deviation. Statistical significance was assessed using one-way ANOVA followed by Tukey’s HSD test for pairwise comparisons. g) PS marker TBXT, pERK and pluripotency marker SOX2 in Nodal knockout cells with inhibitors of Wnt secretion (WntSeci) and BMP receptors (BMPRi) with or without MEKi or FGFRi. All scale bars 50um.

a) Schematic of the FGF/ERK signaling pathway. b) Distribution of pERK levels in TBXT+ cells versus ISL1+ and other cells based on data and thresholds in Supp. Fig. 2b. c) Expression of key markers in PHATE39 projection of scRNA-seq data for 42h colony. d) Cell type annotation. PS-LC: primitive streak-like cells, NasMe: nascent mesoderm, AE-LC: amniotic ectoderm-like cells: PGC-LC: primordial germ cell-like cells, endo: definitive endoderm. e) Expression of canonical FGF ligands (log transformed and smoothened). f) Comparison of FGF ligand expression in 2D human gastruloids to human, monkey, and mouse embryo, normalized within each type of sample (see methods). g-i) Expression of several genes involved in FGF signaling with strong differential expression between PS-like cells or nascent mesoderm and pluripotent cells, split by gene category.

a) phospho-FGFR1 stain with and without 1h FGFRi treatment at 41h, radial intensity profile on the right. b) FGFR1 with and without 1h FGFRi treatment at 41h, radial intensity profile on the right. Error bars in (b,c) represent standard deviation over N=4 colonies. c) qPCR data for relative expression of FGF receptor 1 isoforms IIIb, IIIc and total FGFR1 in pluripotent cells (pluri), gastruloids at 42h (MP), and directly differentiated PS-like cells (dPS). Error bars represent standard deviation of technical triplicates. d) Absolute FGFR1 isoform expression from RNA-seq for anterior/mid primitive streak and pluripotent cells from Loh 201638 (Loh_ APS/MPS/HESC) as well as the pluripotent and PS-LC clusters from Fig.2. All scale bars 50 um.

a) pERK and pFGFR1 response of pluripotent colonies in standard culture grown in FGF-free (E6) medium for 24 hrs and then treated with FGF2 for 30 minutes. b,c) Fluorescent dextran (b) and pERK (c) in micropatterned pluripotent colony 24h after media change containing both FGF2 and dextran. d) Radial intensity profile of dextran and pERK for conditions in b,c), averaged over 2 colonies. e) Cross-section of 2D gastruloid shows FGF receptor 1 predominantly below the tight junction marked by ZO-1. White box marks are magnified in inset e’. f) pERK and pFGFR1 response in transwell experiment with cell treated with FGF either apically or basally. The illustrations in this panel were created using BioRender.com/v73i147. g) ERK signaling in scratched colonies with or without FGFRi. h) Quantification of pERK intensity as a function of distance from the scratch. Error bars represent standard deviation over colonies. i) ERK signaling stains at different times after scratching. j) Quantification of pERK intensity as a function of distance from the scratch at different times (left) and peak pERK level over time (right). Error bars represent standard deviation over N=4 colonies. Scale bars 50 micron.

a) Pattern at 42h with exogenous FGF2 in the culture medium for different amounts of time. b) Pseudotime analysis of FGFs and TBXT, TBX6 expression. c-e) FGF4 (c), FGF8 (d), and FGF17 (e) FISH co-stained for TBXT and pERK at 30h or 32h and 40h or 42h after BMP4 treatment. f) Radial intensity corresponding to (c-e). g-i) FGF4 (g), FGF8 (h), and FGF17 (i) FISH co-stained for TBXT and pERK in control siRNA versus FGF4 siRNA (g), FGF8 siRNA (h) and FGF17 siRNA (i) with corresponding radial intensity profiles. j) TBX6 and SOX17 expression in control shRNA versus FGF4 and FGF17 shRNA. Error bars in all panels except (b) represent standard deviation over N=4 colonies. Error bars in (b) represent the 95% confidence interval of the generalized additive model (see methods).

Graphical summary.

Exogenous FGF2 acts on the colony edge and is prevented from reaching the basal FGFR1 receptors elsewhere by tight junctions. Primitive streak cells express FGF4 and FGF17 and these FGFs are associated with high ERK activity. FGF8 is expressed further inside the colony.

Cell signaling reagents.

Primary antibodies used for immunofluorescence.

Secondary antibodies.

Marker genes for cell fate annotation

qPCR primers

Probes for Fluorescent In Situ Hybridization

a) Stains and quantification for pERK and TBXT in smaller colonies, diameter from left to right: 500um, 350um, 250um. b) Staining for three primitive streak markers with or with FGFRi or MEKi. c) DAPI stain corresponding to Fig. 1d. d) Radial intensity profile for TBXT corresponding to Fig. 1d. e) BrdU and cleaved Cas3 stainings after MEK or FGFR inhibition. f) Quantification of conditions in e. g) Quantification of conditions in Fig. 1g for five images each, N indicates total number of cells. h) Low density differentiation with and without MEK and FGFR inhibition. i) pERK and TBXT stains and quantification for different doses of MEK inhibitor shows low doses effectively block differentiation with minimal impact on cell number.

a) PHATE projection of scRNA-seq data colored for sample. b,c) Expression of canonical FGF ligands across different cell types in 2D gastruloids (b) and human CS7 embryo (c). Expression across clusters is normalized by row, while total expression is shown in a separate column to the directly to left. The leftmost column indicates differential expression between PS-like and pluripotent cells. The color scale of (differential) expression is cut off for better contrast. d) Threshold-based annotation (left) for human, monkey, and mouse embryos used in interspecies comparison (Fig. 2f), compared to original annotation (right). e) Interspecies comparison as in Fig. 2f but for nascent mesoderm and PS-LC + nascent mesoderm. f) Bulk RNA-seq data by Loh et al38 for expression of different FGFs in directed differentiation to mid primitive streak (MPS) or anterior primitive streak (APS) relative the human embryonic stem cells (hESCs). g) Expression of genes involved in FGF signaling.

a) DAPI image corresponding to Fig. 3a. b-c) pFGFR1 (b) or pERK (c) and TBXT stains with and without continuous MEKi treatment. d) DAPI image corresponding to Fig. 3b. e) Design of qPCR primers to detect FGFR1 isoforms, diagram adapted from Eswarakumar et al72. f) Primer sequence and properties. g) Amplicon sizes match prediction. Scale bars 50um.

a,b) Heparan sulfate (HS) stain (a) and radial intensity profile (b). Scale bar 50um. Error bands represent standard deviation over N=4 colonies.

a) DAPI image corresponding to Fig. 5a. b) Pseudotime dynamics from Fig. 5b with each gene normalized individually. c) Correlations between FGFs and TBXT, TBX6 colored for clusters, with cluster colors matching Fig. 2d. d-f) DAPI images and overlays including DAPI corresponding to Fig. 5c-e, respectively. g) Combined FISH for FGF4 and FGF8. h) Scatterplots of FGF4, FGF8, and FGF17 mRNA versus TBXT colored for cell density.

a) Additional channels for the colony shown in Fig. 5g. b) FGF4 shRNA in ESI17 cells. c-d) FGF4 siRNA (c) and shRNA (d) in PGP1 cells. e-f) additional channels for Fig. 5h-i. g) FGF17 shRNA in ESI17 cells. h-i) FGF17 siRNA (h) and shRNA (i) in PGP1 cells. j) additional channels for the TBX6 ctrl and FGF17 shRNA in Fig. 5j. k) FGF4 + FG17 shRNA double knockdown (stains from multiple colonies).