BMP-treated human embryonic stem cells transcriptionally resemble amnion cells in the monkey embryo

Abstract

Human embryonic stem cells (hESCs) possess an immense potential to generate clinically relevant cell types and unveil mechanisms underlying early human development. However, using hESCs for discovery or translation requires accurately identifying differentiated cell types through comparison with their in vivo counterparts. Here, we set out to determine the identity of much debated BMP-treated hESCs by comparing their transcriptome to recently published single cell transcriptomic data from early human embryos ( Xiang et al., 2020). Our analyses reveal several discrepancies in the published human embryo dataset, including misclassification of putative amnion, intermediate and inner cell mass cells. These misclassifications primarily resulted from similarities in pseudogene expression, highlighting the need to carefully consider gene lists when making comparisons between cell types. In the absence of a relevant human dataset, we utilized the recently published single cell transcriptome of the early post implantation monkey embryo to discern the identity of BMP-treated hESCs. Our results suggest that BMP-treated hESCs are transcriptionally more similar to amnion cells than trophectoderm cells in the monkey embryo. Together with prior studies, this result indicates that hESCs possess a unique ability to form mature trophectoderm subtypes via an amnion-like transcriptional state. This article has an associated First Person interview with the first author of the paper.

Keywords: Amnion; BMP4; Differentiation; Extra-embryonic mesoder; Human embryonic stem cells; ScRNA-seq; Trophectoderm.

Fig. 1.

‘AME cells express trophoblast lineage specific genes. (A) Heatmap showing Pearson correlation coefficients for expression of known lineage markers corresponding to human trophectoderm, epiblast and primitive endoderm lineages (12 genes represented in Fig. S1C), between individual ‘AME cells and average expression of same genes in other indicated cell types. (B–E) Box plots showing expression of indicated genes in indicated cell types.

Fig. 2.

‘AME cells are transcriptionally more similar to monkey trophectoderm-derived than monkey amnion cells. (A) Heatmap showing Pearson correlation coefficients of average expression of variable genes in the monkey embryo [(C,E); CV>1 across 1453 monkey cells; 2440 genes] in indicated cell types. (B) Heatmap showing Pearson correlation coefficients for expression of variable genes in the monkey embryo between individual human AME cells and monkey cell type averages. Genes used in A and B are the same as those in Fig. 2D. (C) Box plots showing expression of indicated genes in indicated lineages. The symbols represent species corresponding to the two datasets (human and monkey).

Fig. 3.

Pseudogenes lead to placement of ‘AME cells in the PSA cluster. (A,C) Histogram of coefficient of variation of expressed genes (genes with FPKM>1 in at least 50% cells of a given lineage) across all cells. Principal component analyses (PCA) of 555 cells with variable expressed genes. Variability is defined by a threshold in CV values. The threshold CV and the number of genes that cross the threshold are indicated above the graph. The percentage in x/y labels represents the % of variance in the data explained by each PC. Read counts calculated as log2(fpkm+1) were used for the PCA. Cells are color-coded by the lineages assigned in Xiang et al. (2020). In C, the genes were filtered to include only protein coding genes, prior to determining variable expressed genes. (B) Contribution of different gene biotypes to the first two principal components. Normalized PC coefficient=cumulative PC coefficient for a biotype/cumulative PC coefficient for all biotypes. Cumulative PC coefficient is calculated as the sum of the absolute PC1 coefficient of all genes in that biotype. The two biotypes that contribute the most are shown. (D) PCA plots in C plotted for a subset of cells corresponding to embryonic day 12 and 14.

Fig. 4.

‘AME cells contain a mix of EVT, STB and ambiguous cells. (A) Histogram of coefficient of variation of expressed genes in D12, D14 CTB, EVT, STB and ‘AME cells. (B) PCA of variable genes (CV>0.5) across cells in D12, D14 CTB, EVT, STB and ‘AME cells. Methodology and axis labels same as in Fig. 3. (C,D) Scatter plots showing expression of indicated genes in D12,14 CTB, EVT, STB and ‘AME cells. (E,G) Heatmap showing Pearson correlation coefficients for expression of variable genes [E: genes used for PCA in Fig. 4B; G: genes used for PCA in Fig. 3C (CV 0.5)] between individual ‘AME cells and indicated cell type averages. (F,H) Heatmap showing Pearson correlation coefficients for expression of variable genes [F: genes used for PCA in Fig. 4B; H: genes used for PCA in Fig. 3C (CV 0.5)] between individual ‘AME cells. (I) ‘AME cells reclassified as EVT, STB and ambiguous cells.

Fig. 5.

BMP-treated hESCs transcriptionally resemble monkey early post-implantation amnion cells. (A) Pearson correlation coefficients between indicated samples for 571 lineage-specific genes determined from in vivo monkey embryo data. (B) Venn diagram for differentially upregulated genes in indicated samples compared to the epiblast. Amnion refers to samples labeled as E-AM, trophectoderm to TE derived and epiblast to postE-epiblast in A. (C) Pearson correlation coefficients between indicated samples for 560 lineage-specific genes determined from in vivo monkey embryo data.

Fig. 6.

‘Intermediate cells are mislabeled extra embryonic mesoderm cells. (A,C) Heatmap showing Pearson correlation coefficients for expression of known lineage markers in human embryo (A) or variable genes in the monkey embryo (C) between individual ‘intermediate cells and indicated cell type averages. Genes in C are same as those in Fig. 2D. (B,D) Box plots showing expression of indicated genes in indicated lineages.

Fig. 7.

‘ICM cells are mislabeled CTB cells. (A) Heatmap showing Pearson correlation coefficients for expression of known lineage markers between individual ‘ICM cells and indicated cell type averages. (B–D) Box plots showing expression of indicated genes in indicated lineages.