Derivation of functional trophoblast stem cells from primed human pluripotent stem cells

Abstract

Trophoblast stem cells (TSCs) have recently been derived from human embryos and early-first-trimester placenta; however, aside from ethical challenges, the unknown disease potential of these cells limits their scientific utility. We have previously established a bone morphogetic protein 4 (BMP4)-based two-step protocol for differentiation of primed human pluripotent stem cells (hPSCs) into functional trophoblasts; however, those trophoblasts could not be maintained in a self-renewing TSC-like state. Here, we use the first step from this protocol, followed by a switch to newly developed TSC medium, to derive bona fide TSCs. We show that these cells resemble placenta- and naive hPSC-derived TSCs, based on their transcriptome as well as their in vitro and in vivo differentiation potential. We conclude that primed hPSCs can be used to generate functional TSCs through a simple protocol, which can be applied to a widely available set of existing hPSCs, including induced pluripotent stem cells, derived from patients with known birth outcomes.

Keywords: cytotrophoblast; naive pluripotent stem cells; placenta; primed pluripotent stem cells; trophoblast stem cells.

Graphical abstract

Figure 1

Protocol for conversion of primed hPSCs into TSCs (A) Protocol schematic. (B) Morphology of H9 hESC line as undifferentiated (day 0), following 4 days of BMP4/IWP2 treatment (day 4), and after 5 passages in our modified Okae medium for TSCs (H9-TSCs). Bar: 312.5 μm. (C) Flow-cytometric analysis of H9-derived TSCs for EGFR and ITGA6. (D) Immunofluorescence staining of H9-TSCs for GATA3 and KRT7. Bar: 124.5 μm. Data in (B–D) are representative of n = 5 independent experiments. See also Figure S1.

Figure 2

Characterization of hPSC-derived TSCs (A) Principal-component analysis of hPSCs (undifferentiated/day 0), H9 hESCs treated with BMP4/IWP2 for 1–4 days, hPSC-derived TSCs after 6–8 passages in the modified Okae TSC medium (TSC), and primary (placenta-derived) TSCs (lines 1048 and 1049). Each dot on the principal-component analysis (PCA) represents a sample from an independent experiment (n = 2 for H1 and iPSCs, day 0 and TSC; n = 3 for all other samples). Sample clusters from k-means clustering, marked by circles, show that hPSC-derived TSCs cluster together with primary (placenta-derived) hTSCs. (B) Heatmaps of undifferentiated hPSCs and primary and hPSC-derived TSCs showing genes that are either upregulated (839 genes) or downregulated (779 genes) in primary TSCs compared with undifferentiated hPSCs. GSEA shows that hPSC-derived TSCs were enriched in primary TSC-associated genes (NES 2.44, padj < 0.004) but not in (undifferentiated) hPSC-associated genes. A few TSC-associated genes are noted in the heatmap. (C) qPCR of the indicated CTB markers in undifferentiated primary TSC (1049) compared with hPSC (H9)-TSCs. Data were normalized to L19 and shown as fold change over undifferentiated 1049 (D0 = day 0) and represent mean ± SD for n = 3 independent experiments. ^∗p < 0.05; ^∗∗p < 0.01 by Student’s t test. (D) DNA methylation surrounding the ELF5 promoter in undifferentiated (U) H9 ESCs, TSCs derived from both H9 and a human dermal fibroblast (HDF)-derived iPSCs, an umbilical-cord-derived mesenchymal stem cell line (1754), and a primary (placenta-derived) TSC line (1049). Each line represents a distinct sequenced clone (n = 3 to 9). (E) Flow-cytometric analysis of primary (1049) and hPSC (H9)-derived TSCs for HLA-A2 and HLA-Bw6 with (gray) and without (purple) valproic acid (VPA) in the culture medium. Data are representative of 2 independent experiments. (F) Heatmap of amnion-specific markers (based on a more stringent analysis by Seetheram et al. in this issue of Stem Cell Reports of the Roost et al., 2015 dataset) in undifferentiated hPSCs and primary (1048 and 1049 TSC) and hPSC-derived TSCs, as well as amnion (AM). GSEA showed that neither primary nor hPSC-derived TSCs were enriched in amnion-specific genes (NES = -0.94 with padj = 0.561 and NES = -1.06 with padj = 0.354, respectively). See also Figure S2.

Figure 3

In vitro and in vivo differentiation potential of hPSC-derived TSC (A) Morphology, lineage-specific gene expression, and hCG secretion of primary (1049) and hPSC (H9)-derived TSCs differentiated into syncytiotrophoblast (STB). Bar: 312.5 μm. qPCR data were normalized to L19 and shown as fold change over undifferentiated (day 0/D0) 1049 TSCs. hCG secretion was normalized to ng of DNA. Both qPCR and ELISA data represent mean ± SD for n = 3 independent experiments. ^∗p < 0.05; ^∗∗p < 0.01 by Student’s t test. (B) Morphology, lineage-specific gene expression, and flow-cytometric analysis for surface HLA-G expression of primary (1049) and hPSC (H9)-derived TSCs differentiated into extravillous trophoblast (EVT). Bar: 312.5 μm. qPCR data were normalized to L19 and shown as fold change over undifferentiated (day 0/D0) 1049 TSCs and represent mean ± SD for n = 3 independent experiments. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001 by Student’s t test. Flow-cytometric data are representative of 3 independent experiments. (C and D) Tumors generated 10 days following injection of primary (1049) and hPSC (H9)-derived TSCs into NOD-SCID mice (representative of n = 2 independent experiments). (C) H&E staining shows the tumor cells invading through muscle (1049-TSC tumor) or forming a tumor with a necrotic center (H9-TSC), both characteristic of human trophoblastic tumors. Scale bars: 250 μm for low-power (left) and 50 μm for high-power (right) images. (D) Immunohistochemical staining of the same tumors using antibodies against EGFR, hCG, and HLAG shows positively stained cells (brown) in the TSC-derived lesions. Scale bars: 50 μm.

Figure 4

Transition from primed pluripotency to TSCs involves a trophectoderm-like intermediate (A) Morphology of H9-ESCs at days 0, 2, and 4 following treatment with BMP4/IWP2. At the end of 4 days, over 94% of the cells express EGFR by flow-cytometric analysis. Data are representative of n = 5 independent experiments. Bar: 312.5 μm. (B) qPCR of H9-ESCs at days 0, 2, and 4 following BMP4/IWP2 treatment, for markers of pluripotency (NANOG and POU5F1), naive pluripotency (DNMT3L, KLF17, DPPA3, and DPPA5), trophectoderm (ENPEP, TACSTD2, NR2F2, and CDX2), and cytotrophoblast (CTB) (TP63, VGLL1, GATA2, GATA3, TFAP2C). Data were normalized to L19 and shown as fold change over undifferentiated H9-ESCs (D0 = day 0). ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001 by Student’s t test. (C) Morphology and EGFR expression of BMP4/IWP2-treated H9-ESCs passaged in modified Okae TSC medium. During this time, the cells initially lose EGFR expression and slowly re-gain it after at least 5 passages. Data are representative of n = 3 independent experiments. Bar: 312.5 μm. (D) PCA of cells transitioning from primed pluripotency (hPSCs) through BMP4/IWP2 induction then undergoing adaptation through 6 passages (p0–p6) in modified Okae TSC medium. Arrow points to the passage in which the hPSC-derived cells are farthest away from primary TSC. Each dot on the PCA represents a sample from an independent experiment (n = 2 for H1 and iPSC, day 0 and TSC; n = 3 for all H9 samples). (E) Median expression of genes in clusters 1 and 2 from Click-clustering analysis Figures S3C and S3D) of H9-ESCs at day 4 of BMP4/IWP2 treatment and across TSC medium adaptation (10 sample groups indicated by red box in D). See also Figure S3.

Figure 5

Comparison of trophectoderm and TSC derived from naive and primed hPSCs (A) PCA of our cells, combined with those published by et al., 2021, including, from left to right, (1) undifferentiated naive and primed hPSCs (day 0) and their early trophoblast derivatives (days 1–2 in our BMP4/IWP2 treatment, and day 1 of Io et al.’s induction); (2) naive and primed hPSCs induced into a trophectoderm/TE-like fate (days 3–4 in our BMP4/IWP2 treatment, and days 2–3 of Io et al.’s induction); (3) primary (placenta-derived) TSCs (1048 and 1049 lines), our primed hPSC-derived TSCs (H9-TSCs at passage 6, H1-TSCs at passage 7, iPSC-TSCs at passage 8), and Io et al.’s naive hPSC-derived TSCs (naive cytotrophoblast [N-CT] at passages 3–15); and (4) primary CTB (CTB-1E include our preps from 10 different 5- to 8-week-gestation placentae; other four samples include two 9-week- and two 11-week-gestation CTB preps from Io et al., 2021). For hPSC lines, each dot represents a sample from an independent experiment (n = 2 for H1 and iPSCs, day 0 and TSCs, and all samples from Io et al., 2021; n = 3 for all H9 samples); for CTB samples, each dot represents a biological replicate. Clusters from K-means clustering shown in gray dashed circles. (B) Dendrogram of the same samples, showing primed and naive hPSCs and their early derivatives clustering separately (left) from primary CTB and hPSC-derived TSCs (naive and primed), as well as hPSC-derived TE (naive and primed). (C) Single-cell RNA-seq analysis of H9-ESCs at day 0, at 12 and 24 h, and at 4 days post BMP4/IWP2 treatment, as well as H9-TSCs, separated into 4 clusters by Seurat. Feature plots of pluripotency-, TE-, and CTB-associated markers show induction of TE- (and not naive pluripotency-) associated genes, prior to finally transitioning into TSC state, with expression of CTB-associated markers. (D) UCell analysis using epiblast (EPI) and TE-specific genes from single-cell RNA-seq of extended culture human embryo dataset by Zhou et al. (2019). Note clusters 0 (H9-12 and 24 h) and 1 (undifferentiated H9) are enriched in EPI genes, while clusters 2 (H9-TSC) and 3 (H9 at 4 days post BMP4/IWP2 treatment) are enriched in TE genes. (E) Analysis by PlacentaCellEnrich tool of transcriptomic signatures of BMP4/IWP2-treated hPSCs, primary TSCs, and hPSC-derived TSCs compared with undifferentiated hPSCs. Note similar enrichment proportions for STB-, extravillous trophoblast (EVT)-, and CTB-associated genes across all 3 comparisons. See also Figure S4.