Single-cell trajectory analysis for studying PP and anterior Endo specification.

(A) URD differentiation tree of zebrafish embryos from blastula stage to somitogenesis stage74. Cells are colored according to developmental stages. Gray dotted frame indicates the ectoderm lineages and black dotted frame indicates the mesendoderm lineages. Green line shows the pharyngeal endoderm lineage (anterior endoderm, roman number I), red line shows the prechordal plate lineage (roman number II) and yellow line indicates their progenitors. (B) Simplified schematic showing the patterns of endoderm and axial mesoderm of zebrafish embryos at 75% epiboly. Dotted frame indicates the enlarged view of dorsal anterior region, showing that anterior Endo sporadically locates near PP. (C) URD branchpoint plots of the PP and anterior Endo development from scRNA-seq data of zebrafish embryos74, showing pseudotime (x-axis) and random walk visitation preference from pharyngeal endoderm to prechordal plate domains (y-axis). Cells are colored by developmental stages (left) and expression of Endo markers (sox32, sox17 and pax1b; right top) and PP markers (gsc, frzb and he1a; right bottom). (D) Force-directed layout of PP and anterior Endo cells of zebrafish embryos from blastula stage to somitogenesis stage. Cells are colored by Palantir27 pseudotime (top left), differentiation potential (top right) and branch probabilities of PP (bottom left) and anterior Endo (bottom right). (E) Branch probabilities of five cells randomly selected from progenitor cells. Bars are colored by cell types. (F) URD differentiation tree of Nodal explants from blastula stage to the end of gastrulation. Gray dotted frame indicates the ectoderm lineages and black dotted frame indicates the mesendoderm lineages. The green, red and yellow dots indicate the Endo cells (anterior Endo), PP cells and the progenitor cells respectively, and the gray dots indicate other cells. (G-H) URD branchpoint plots of the PP and Endo development from scRNA-seq data of Nodal explants showing pseudotime (y-axis) and random walk visitation preference from PP to Endo domains (x-axis). Cells are colored by developmental stages (G) and expression of Endo markers (sox32, sox17 and nkx2.7; top) and PP markers (gsc, frzb and he1a; bottom) (H). (I) Force-directed layout of PP and anterior Endo cells of Nodal explants from blastula stage to the end of gastrulation. Cells are colored by the levels of Palantir27 pseudotime (top left), differentiation potential (top right) and branch probabilities of PP (bottom left) and anterior Endo (bottom right).

Live imaging analyses for Tg(gsc:EGFP;sox17:DsRed) embryos.

(A) Time series of 3D reconstruction of a representative Tg(gsc:EGFP;sox17:DsRed) embryo. (B) Schematic diagram indicating the focused region in the Tg(gsc:EGFP;sox17:DsRed) embryo. (C and D) Temporal profiles of sox17 (RFP) & gsc (GFP) (C) and the ratio between RFP intensity and GFP intensity (D) in the highlighted cell from (A). (E) Double-color RNA-fluorescence in situ hybridization (FISH) of sox17 and gsc in wild-type embryos at 6 hpf (top) and 7.5 hpf (bottom). (N >= 30, n = 5/5, 6 hpf; N >= 30, n = 5/5, 7.5 hpf; N: embryos were used in the experiment; n: embryos were imaged, expression observed/total imaged) (F) Scatter plots and inferred linear regression comparing the relative intensities of sox17 and gsc in wild-type embryos at 6 hpf (top) and 7.5 hpf (bottom). Each FISH experiment was performed for at least 3 independent replicates (technical replicates). Scale bar: 30 µm (A) and 50 µm (E).

Nodal-Lefty regulatory loops are involved in cell fate separation between anterior Endo and PP.

(A and B) Dot plots visualizing ligand-receptor interactions between PP and Endo. The interactions from PP (ligand) to Endo (receptor) and from Endo (ligand) to PP (receptor) are plotted respectively in A and B. (C) Schematic diagram illustrating PP (red) and Endo (green) in zebrafish embryos at 6 hpf. (D) Simplified schematic showing the interaction network between Nodal and Lefty. (E) 3D reconstruction of representative confocal-scanned images showing pSmad2 levels in zebrafish embryonic animal pole stimulated by Nodal injection at 128-cell stage (WT, N >= 10, n = 3/4; ndr1kd, N >= 10, n = 3/4; lft1ko, N >= 10, n = 4/4; N: embryos were used in the experiment; n: embryos were imaged, observed/total imaged). (F) Comparisons of pSmad2 levels in wild-type embryos, lft1 mutants and ndr1 morphants at 6 hpf (quantified from E). Statistical differences between two samples were evaluated by Student’s t-test. *indicates P-value < 0.05, **indicates P-value < 0.01 and ***indicates P-value < 0.001. (G) Schematic diagram showing the experimental workflow of single-cell RNA sequencing of Nodal explants constructed from lft1 mutants and ndr1 morphants at 6 hpf. (H) Overall UMAP plot of integrated single-cell datasets of Nodal explants generated from wild-type embryos (left), ndr1 morphants (middle) and lft1 mutants (right). (I) Histograms showing the ratios of the cell proportions of endoderm against EVL (top) and prechordal plate (bottom) against EVL. Per indicates percentage. As EVL cells can be specified spontaneously and are Nodal-independent75, here the proportions of prechordal plate and endoderm were adjusted by using the proportions of EVL cells as an internal control. (J) Venn plot showing the selection of Nodal direct target genes used for defining Nodal score (see Methods). (K) Dot plot visualizing Nodal score in wild-type, ndr1-morphant and lft1-mutant Nodal explants. Dot size indicates the percentage of cells within the cell group; the color indicates the average Nodal score levels. Scale bar: 50 µm (E).

Single-cell trajectory tree analyses of PP and anterior Endo reveal that chromatin organization is involved in the separation of these two lineages.

(A) Pseudotime trajectory analysis of PP and anterior Endo using integrated single-cell datasets of wild-type (left), ndr1-morphant (middle) and lft1-mutant (right) Nodal explants. Cells are colored by cell types (top) and pseudotime levels (bottom). (B) Heatmaps representing clusters of genes that co-vary along the pseudotime during the separation of PP and anterior Endo from common progenitors. Top GO terms enriched by the genes in each cluster are listed with their corresponding adjusted P-values. Representative GO terms of each cluster are highlighted in different colors. (C) Diagram of gene set selection for GO analysis. As a control of batch effects, EVL differentially expressed (DE) genes are removed from mesoderm PP DE genes. (D and E) GO enrichment analysis of mesoderm PP up-regulated genes in lft1 mutants (D) and down-regulated genes in ndr1 morphants (E). Redundancy of enriched GO terms are removed, and each GO term contains at least 10 transcripts. Note that only the top 16 enriched GO terms identified in lft1 mutant are plotted for a visualization purpose. (F and G) GSEA of “chromatin organization” term on PP cells from lft1 mutants (F) and ndr1 morphants (G) respectively. P-adjust indicates Benjamini-Hochberg adjusted p-value. (H-M) SWI/SNF complexes involved in regulating cell fate separation between the PP and anterior Endo. (H) HCR co-staining of gsc, frzb and sox32 in wild-type embryos (top, N >= 40, n = 5/5, N: embryos were used in the experiment, n: embryos were imaged, expression observed/total imaged), as well as those treated with 1 µM (middle, N >= 40, n = 4/5, N: embryos were used in the experiment, n: embryos were imaged, expression observed/total imaged) and 5 µM (bottom, n = 5/6, expression observed/total imaged) AU15330. Regions framed by dotted white lines were used to quantify the cell number of anterior Endo. (I) Box plot showing the cell numbers of anterior Endo in (H). (J and K) Volinplot indicating the expression levels of srcap in PP and Endo cells within both embryos (J) and Nodal explants (K). (L) HCR co-staining of gsc, frzb and sox32 in wild-type embryos and embryos with srcap knockdown. Embryos at 7.5 hpf (Ctrl, N >= 30, n = 6/6; srcap MO, N >= 30, n = 4/5; N: embryos were used in the experiment; n: embryos were imaged, expression observed/total imaged) were evaluated. (M) Box plot showing the cell numbers of anterior Endo in (L). Statistical differences between two samples were evaluated by Student’s t-test (I and M). *indicates P-value < 0.05; NS indicates P-value >=0.05. Each HCR experiment was performed for at least 3 independent replicates (technical replicates). Scale bar: 50 µm (H and L).

Investigating the mechanisms of PP and Endo separation through integrative single-cell RNA-seq and ATAC-seq analysis.

(A) Schematic diagram showing the experimental workflow of single-cell multi-omics of 6 hpf zebrafish embryos. (B) UMAPs based on scATAC-Seq (left), scRNA-Seq (middle) and co-embedding scRNA-seq and scATAC-seq (right) datasets. Cells are colored by different cell types. (C) UMAP plot of the PP and Endo cells. Cells are colored by cell types (left) and the expression of gsc (middle) and sox32 (right). (D) Volcano plot showing the differentially expressed genes in PP and Endo based on RNA levels of single-cell multi-omics datasets. (E-F) Track plot showing chromatin accessibility of each cell type on the gene locus of gsc (E) and ripply1 (F). (G) Heatmap showing the chromatin accessibility level of genes with differential chromatin accessibility in PP and Endo. Red and green box indicates genes with higher chromatin accessbility levels in PP and Endo cells respectively. (H and I) Dot plot (H) and ridge plot (I) showing the level of Nodal score in PP and Endo. (J) Box plots showing the expression profiles of sox17 (Endo), gsc (PP), ripply1 (PP) and krt4 (EVL) in our previous bulk RNA-seq datasets of Nodal explants. (K) Box plots (left) and bar plot (right) showing the expression levels of SOX17 and GSC in the human embryonic stem cells (hESCs) treated with different dosages of Activin protein (see Methods). The values were calculated as the mean ± SEM (K).

Exploring the mechanisms underlying the role of Ripply1 in the regulation of cell fate separation between PP and anterior Endo.

(A) Schematic diagram illustrating the workflow for investigating the role of gsc and ripply1 in mesendoderm cell fate separation through loss-of-function studies. (B) Schematic presentation of gsc and ripply1 mutations and the sequencing results of each genotype. The mutations are a 10-bp deletion and a 1-bp insertion in gsc and ripply1 coding sequences respectively. Texts in red indicate coding sequences. Sanger sequencing results display at the bottom, showcasing wild-type (top), heterozygous mutant (middle) and homozygous mutant (bottom) embryos of gsc and ripply1 mutations. (C) HCR co-staining of sox32 and frzb in the embryos of Tg(gsc+/+;ripply1+/+) (WT), Tg(gsc+/+;ripply1-/-) (ripply1 mutant) and Tg(gsc-/-;ripply1-/-) (double mutant) at 8 hpf (Here embryos of three different genotypes are shown. Other genetypes are shown in Figure S10A.) [Tg(gsc+/+;ripply1+/+), n = 4/5; Tg(gsc+/+;ripply1-/-), n = 4/6; Tg(gsc-/-;ripply1-/-), n = 5/5; expression observed/total imaged]. (D) Box plot showing the quantification of the numbers of anterior Endo cells identified in (C). (E and F) Assessing the role of ripply1 in anterior Endo specification. EGFP driven by sox17 promotor (E) and WISH of sox32 (F) are used to indicate the Endo cells. Embryos with gsc overexpression are used as a positive control. The statistics of observed/total sampled embryos were shown in bottom left of the panels (E and F). (G) Schematic diagram showing the strategy of conducting CUT&Tag experiment for ripply1. (H) Metaplots and heatmaps for CUT&Tag signals over R1_rep1 binding peaks (n=72,683) in wild-type and HA-ripply1-injected embryos. Two replicates were performed. (I) Dot plot showing the enriched GO terms using the genes that annotated from the differentially enriched peaks. (J and K) Screenshot of R1_con (technical control, without primary antibody, see Methods), WT (ddH2O-injected) and R1 (HA-ripply1-injected) CUT&Tag peaks around sox17 (J) and sox32 (K) loci. Each HCR and ISH experiment was performed for at least 3 independent replicates (technical replicates). Statistical differences between two samples were evaluated by Student’s t-test (D). Scale bar: 50 µm (C) and 200 µm (E and F).

Model underlying the separation of PP and anterior Endo.

An outline depicts PP and anterior Endo specification. Nodal signaling drives a differential chromatin states between PP and anterior Endo through epigenetic regulators. And then, key transcriptional factors, such as gsc and ripply1, are differentially expressed in these two lineages and suppress anterior Endo specification by directly binding to the cis-elements of sox17 and sox32.