Identification and Single-Cell Functional Characterization of an Endodermally Biased Pluripotent Substate in Human Embryonic Stem Cells

Abstract

Human embryonic stem cells (hESCs) display substantial heterogeneity in gene expression, implying the existence of discrete substates within the stem cell compartment. To determine whether these substates impact fate decisions of hESCs we used a GFP reporter line to investigate the properties of fractions of putative undifferentiated cells defined by their differential expression of the endoderm transcription factor, GATA6, together with the hESC surface marker, SSEA3. By single-cell cloning, we confirmed that substates characterized by expression of GATA6 and SSEA3 include pluripotent stem cells capable of long-term self-renewal. When clonal stem cell colonies were formed from GATA6-positive and GATA6-negative cells, more of those derived from GATA6-positive cells contained spontaneously differentiated endoderm cells than similar colonies derived from the GATA6-negative cells. We characterized these discrete cellular states using single-cell transcriptomic analysis, identifying a potential role for SOX17 in the establishment of the endoderm-biased stem cell state.

Keywords: GATA6; differentiation bias; human embryonic stem cell heterogeneity; lineage priming.

Graphical abstract

Figure 1

GATA6 Is Expressed in a Small Subset of hESCs (A) Representative FACS plot of the Shef4 GATA6-GFP reporter line S4G6 A3 cultured in KO/SR and MEF conditions. Black peak represents the unmodified parental Shef4 control line, and red, the Shef4 GATA6-GFP reporter line. (B) Representative FACS plot of SSEA3 vs GATA6 expression. Left panels show gating controls P3X (above) and TRA-1-85 (below) on the Shef4 parental line. Right panel shows the identification of distinct cell populations: SSEA3 high, GATA6 negative (3+/6−); SSEA3 high, GATA6 low (3+/6L); SSEA3 high, GATA6 high (3+/6H); SSEA3 negative GATA6 high (3−/6+), of the GATA6 reporter line. (C) Representative FACS plots of additional stem cell surface markers, SSEA3, TRA-1-81 or SSEA4 vs GATA6 expression with the same controls as (B).

Figure 2

Gene Expression Profiles of Fractions 3+/6−, 3+/6L, 3+/6H, and 3−/6+ (A and B) qPCR using an Applied Biosystems pluripotency TaqMan array on each cell fraction. Hierarchical clustering using Spearman's rank correlation showed strong segregation of genes into two groups: stem cell-associated (group 1) (A) and lineage-associated genes (group 2) (B) with respective gene names. Colormap indicates level of expression of 1/Δ-CT values standardized by row. (C) Boxplot analysis of average gene expression of lineage-specific genes in each cell fraction grouped by specific germ layer. ^∗Kruskal-Wallis statistical test results are indicated for p values <0.05.

Figure 3

High GATA6 Expression Results in a Reduced Cloning Efficiency (A) Percentage cloning efficiency of each cell fraction (3+/6−, 3+/6L, 3+/6H, and 3−/6+) using OCT4 (left) and SOX2 (right) as markers for the stem cell state. Sorted fractions were plated as single cells at clonogenic density in KO/SR and MEF conditions. Cloning efficiency was calculated by dividing the number of OCT4-positive (left) or SOX2-positive (right) colonies by starting seed density. Error bars represent SD of three biological experiments. Student’s t test was used to determine significance (OCT4 graph: 3+/6− to 3+/6H *p = 0.0017, 3+/6− to 3−/6+ ^∗∗p = 0.0001, SOX2 graph; 3+/6− to 3−/6H ^∗p = 0.0019, 3+/6− to 3−/6+ ^∗p = 0.0002). (B) Proportion of OCT4-positive (OCT4[+]) cells in OCT4-positive colonies derived from single cells from fractions 3+/6−, 3+/6L, 3+/6H, and 3−/6+. Positive colonies include one or more OCT4(+) cells. Counts are shown as bar plots (blue) with superimposed estimated nonparametric distribution (red). (C) Kullback-Leibler symmetric divergence between OCT4-associated distributions shown in (B). This measure increases with reduced similarity between distributions; zero indicates identical distributions. (D) Proportion of SOX2(+) (SOX2-positive) cells in SOX2-positive colonies derived from single cells from fractions 3+/6−, 3+/6L, 3+/6H, and 3−/6+. Positive colonies include one or more SOX2(+) cells. Counts are shown as bar plots (blue) with superimposed estimated nonparametric distribution (red). (E) Kullback-Leibler symmetric divergence between SOX2-associated distributions shown in (D).

Figure 4

Stable, Long-Term Self-Renewing hESC Subclones Can Be Derived from GATA6-Expressing Cells (A) Flow cytometric analysis of subclones derived from the 3+/6−, 3+/6L, and 3+/6H fractions. Unsorted represents the unsorted cells of the reporter line. P3X was used as a negative control, and markers SSEA3 (red), TRA-1-81 (orange), and SSEA4 (blue) were used to identify stem cells. FACS plots show one clone from each fraction, which is representative of four clones analyzed from each fraction. (B) qPCR analysis of two subclones from each fraction for core stem cell transcription factors, shown as Delta-CT normalized to beta-actin; error bars are the SD from three technical repeats. Red bar represents the reporter line, and individual 3+/6−, 3+/6L, and 3+/6H subclones are shown by green, orange, and blue bars, respectively. (C) qPCR for lineage-specific markers of each germ layer in unsorted (red) and two subclones from each fraction. Bar color as in (B), showing Delta-CT normalized to beta-actin with three technical repeats. (D) qPCR of day 10 EBs from unsorted (red), and subclones from 3+/6− (green), 3+/6L (orange), and 3+/6H (blue) fractions for genes specifying mesoderm (left panel) and ectoderm (right panel) to demonstrate pluripotency of the lines. Data shown as fold change against undifferentiated cells from the same starting population. Error bars are SD of three technical repeats. (E) Flow cytometric analysis of reporter line (top left) and 3+/6− (top right), 3+/6L (bottom left), and 3+/6H (bottom right) subclones for SSEA3 versus GATA6 expression 5–8 passages after initial single-cell seeding. Gates were set using P3X and Shef4 parental line as SSEA3 and GFP negative controls respectively. Plot shows one clone representative of four clones analyzed from each fraction.

Figure 5

High GATA6 Expression Results in Endoderm Differentiation Bias at Population and Single-Cell Level (A) qPCR of differentiating cells from the 3+/6L (red), 3+/6H (green), 3-/6+ (blue) fractions in a non-directed EB differentiation assay, shown as fold change against differentiating cells from 3+/6− fraction, for genes expressed in endoderm (top left), mesoderm (top right), and ectoderm (bottom left). Beta-actin was the normalizing gene. Error bars represent three biological replicates. (B) Representative images of colonies derived from 3+/6−, 3+/6L, 3+/6H, and 3−/6+ fractions. Images were taken at ×10 magnification on an InCell Analyzer 2000 and automated quantitative analysis performed using developer toolbox software. The same algorithms were used for each technical and biological repeat and the process was automated to eliminate human bias. (C) Quantification of colony types from 3+/6−, 3+/6L, 3+/6H, and 3−/6+ fractions showing the percentage of colonies per fraction with colony phenotype shown in (B) from three biological repeats. (D) Percentage of colonies containing OCT4 and SOX17 (top graph) or OCT4 and GATA4 (bottom graph) positive cells only. Significance was calculated using t test of three biological replicates and stars represent degree of significance (^∗p < 0.05). Numbers for each fraction: 3+/6− = 83, 3+/6L = 103, 3+/6H = 122, 3−/6+ = 66. (E) Histogram showing the distribution of SOX17(+) cells in OCT4-positive colonies resulting from single cells from 3+/6−, 3+/6L, 3+/6H, and 3−/6+ fractions. Positive colonies include at least two OCT4(+) cells. Counts are shown as a bar plot (blue) with superimposed estimated nonparametric distribution (red). (F) Kullback-Leibler symmetric divergence between SOX17-associated distributions in OCT4-positive colonies. This measure increases with reduced similarity between distributions; zero indicates identical distributions.

Figure 6

Single-Cell Transcriptomic Analysis of Endodermally Biased hESCs (A) tSNE analysis of the four cellular subsets representing 13 putative clusters separated according to gene expression per single cells. Single cells are represented by individual dots and are colored according to cell fraction library. Numbers represent cluster number assigned arbitrarily. (B) Heatmap of the top 30 most differentially expressed genes between the 13 individual putative clusters, as described in Figure 6A. Color scheme is based on Z score distribution from −2 (blue) to +2 (red). Right margin color bars represent gene sets specific to each cluster. Left margin color bars represent top Gene Ontology terms of the top 30 most differentially expressed genes for each cluster. (C) Heatmap of the average expression of genes typically associated with later developmental processes including heart/skeletal (top) and hepatic (lower) lineages across the 13 clusters. Color scheme is based on the averaged normalized expression of each gene from no expression (yellow) to expression (blue). (D) Heatmap of the average expression of genes associated with both the stem cells and early differentiating cells across all the individual 3+/6H clusters and the 3+/6L cluster. Color scheme is based on the averaged normalized expression of each gene from no expression (yellow) to expression (blue). (E) Scatter dot plot to show the mean, upper, and lower limit expression of SOX17 between cluster 6 of the 3+/6L and cluster 13 of the 3+/6H fractions. Student's t test was used to determine statistical significance of ^∗∗∗∗p > 0.0001.