SOX17 is a critical specifier of human primordial germ cell fate

Abstract

Specification of primordial germ cells (PGCs) marks the beginning of the totipotent state. However, without a tractable experimental model, the mechanism of human PGC (hPGC) specification remains unclear. Here, we demonstrate specification of hPGC-like cells (hPGCLCs) from germline competent pluripotent stem cells. The characteristics of hPGCLCs are consistent with the embryonic hPGCs and a germline seminoma that share a CD38 cell-surface marker, which collectively defines likely progression of the early human germline. Remarkably, SOX17 is the key regulator of hPGC-like fate, whereas BLIMP1 represses endodermal and other somatic genes during specification of hPGCLCs. Notable mechanistic differences between mouse and human PGC specification could be attributed to their divergent embryonic development and pluripotent states, which might affect other early cell-fate decisions. We have established a foundation for future studies on resetting of the epigenome in hPGCLCs and hPGCs for totipotency and the transmission of genetic and epigenetic information.

Graphical abstract

Figure 1

Specification of hPGCLCs from Human Embryonic Stem Cells (A) Schematic protocol for hPGCLCs specification from hESCs. (B) Development of day 1–7 embryoids derived from WIS2-NANOS3-mCherry hESCs. Top row: images of embryoids. Bottom row: FACS analysis of the dissociated embryoids with anti-TNAP-Alexa Fluor 647 and NANOS3-mCherry to detect hPGCLCs. (C) Expression analysis by RT-qPCR of TNAP-positive 4i hESCs (hESC TNAP+), TNAP/NANOS3-mCherry-positive hPGCLCs (TNAP+N3+), and the remaining cells (TNAP-N3-) of day 4 embryoids (D4 embryoid). Relative expression levels are shown with normalization to β−ACTIN. Error bars indicate mean ± SD from three independent biological replicates. (D) Immunofluorescence of a day 4 embryoid showing coexpression of NANOS3-mCherry, NANOG, and OCT4 in hPGCLCs. Scale bar, 66 μm.

Figure 2

hPGCLC Shares Transcriptional Profile with Human Embryonic PGCs and TCam-2 Seminoma (A) Unsupervised hierarchical clustering (UHC) of gene expression in 4i hESC, preinduced cells (Pre-induced), day 4 hPGCLCs (hPGCLC), gonadal hPGC, TCam-2, and gonadal somatic cell (Soma). RNA-seq was performed on two biological replicates (#1 and #2) for each cell type. (B) PCA of RNA-seq data. Arrowline indicates potential germline progression from 4i hESC to hPGCLC and gonadal hPGC. (C) Heat map of gene expression of key PGC-associated genes (early and late) and of pluripotency, mesoderm, endoderm, and gonadal somatic (Soma) markers. (D) Venn diagram illustrates common and differentially expressed genes. Significantly upregulated genes in hPGCLC, gonadal hPGC, and TCam-2 (with log2 (fold change) > 3 and adjusted p value < 0.05 versus gonadal Soma, respectively) were compared. Representative genes that were exclusive to each category are indicated. Text boxes indicate gene ontology biological processes (BP) terms that were significantly enriched as indicated by p values. Asterisk denotes custom categories absent from BP annotation. (E–H) Immunofluorescence analysis for (E) BLIMP1, (F) PRDM14, (G) SOX2, and (H) SOX17 on 4i hESCs (top row), day 4 hPGCLC embryoids (second row), human week 7 male gonad (third row), and TCam-2 (bottom row). Samples were counterstained with TFAP2C or OCT4 to identify hPGCLCs in embryoids and hPGCs in embryonic gonad. Arrows indicate cytoplasmic enrichment of PRDM14 (F). Scale bars, 70 μm.

Figure 3

CD38 Expression in Human Germ-Cell-Related Cells and Epigenetic Changes in hPGCLCs (A) FACS analysis of NANOS3-mCherry and CD38 on WIS2-NANOS3-mCherry cell line cultured in 4i medium and on day 4 and 5 embryoids following hPGCLC induction. Ratios of CD38 low and high expression in the NANOS3-mCherry-positive cells are indicated. (B) FACS histogram of CD38 low and high populations in TCam-2. (C) FACS analysis of CD38 and TNAP on genital ridges isolated from a week 6 human embryo. (D) Expression analysis by RT-qPCR for FACS-sorted TNAP-positive 4i hESCs (TNAP+ hESC) and CD38 low or high/NANOS3-mCherry day 5 hPGCLCs. Relative expression levels are shown with normalization to β−ACTIN. Error bars indicate mean ± SD from two independent biological replicates. (E and F) Immunofluorescence analysis for 5hmC (E) and TET1 (F) on day 4 embryoids cryosection. OCT4 or BLIMP1 were used to identify hPGCLCs (highlighted). Scale bars, 50 μm. (G) Quantification of immunofluorescence intensity of various epigenetic marks/modifiers in hPGCLCs and somatic neighbors in day 1–4 embryoids (see also Figures S3A–S3C). For UHRF1, only KI-67-positive (proliferating) cells were used for quantification. Numbers below each box denotes number of cells analyzed. Black central line represents the median, boxes and whiskers represent the 25^th and 75^th, and 2.5^th and 97.5^th percentiles, respectively. Wilcoxon signed-rank test was used to test for statistical significance. #p < 0.05; ^∗p < 0.0001.

Figure 4

Sequential Expression of Germ-Cell-Related Transcription Factors in Single Cells during hPGCLC Specification (A and B) Immunofluorscence analysis for (A) BLIMP1, SOX17, and TFAP2C and (B) BLIMP1 and T in cryosections of day 1–8 embryoids after hPGCLC induction. Bottom row in (B) shows high exposure (digital) image of T, indicating low but specific expression in hPGCLC. SOX17-positive or BLIMP1-positive cells are highlighted. Scale bars, 50 μm. (C) Percentage of SOX17-positive (+) cells in day 1–8 embryoids that were also TFAP2C+ or BLIMP1+. Corresponds to data in Figure 4A. (D) Percentage of BLIMP1-positive (+) cells in day 1–8 embryoids that were TFAP2C+, NANOG+, or OCT4+. Corresponds to data in Figures 4A, S4A, and S4B. (E) Summary model for dynamics of hPGCLC specification in embryoids. SOX17-positive cells are first scattered in day 1 embryoids. They gain expression of BLIMP1, TFAP2C, and NANOG sequentially and form a cluster from day 2 onward until the formation of nascent hPGCLC.

Figure 5

Role of BLIMP1 in hPGCLC Specification (A) Western blot analysis of BLIMP1 and SOX17 in TNAP-positive (TNAP+) cells sorted from wild-type (WT) and BLIMP1 knockout (BLIMP1 KO) day 4 embryoids after hPGCLC induction. TUBULIN was used as loading control. (B) FACS analysis of TNAP and NANOS3-mCherry on WT and BLIMP1 knockout (BLIMP1 KO) day 4 embryoids. (C) Immunofluorscence for OCT4 and SOX17 in cryosections of WT and BLIMP1 KO day 4 and 8 embryoids. OCT4-positive cells are highlighted. Scale bar, 50 μm. (D) Expression analysis by RT-qPCR for WT TNAP/NANOS3-mCherry double-positive cells (WT; TNAP+N3+) and BLIMP1 KO TNAP single-positive cells (BLIMP1 KO; TNAP+) sorted from day 4 embryoids. Relative expression levels are shown with normalization to β−ACTIN. Error bars indicate mean ± SD from two independent biological replicates.

Figure 6

Role of SOX17 in hPGCLC Specification (A) Western blot analysis of SOX17 expression of WT day 4 TNAP/NANOS3-mCherry-positive hPGCLCs (WT, TNAP+N3+), and whole SOX17 knockout day 4 embryoids. TUBULIN was used as loading control. (B) FACS analysis of TNAP and NANOS3-mCherry on WT and SOX17 KO day 4 embryoids. (C) RT-qPCR analysis of TNAP/NANOS3-mCherry FACS-sorted WT double-negative (TNAP-N3-) or -positive (TNAP+N3+) cells sorted from day 4 embryoids and whole SOX17 KO embryoids (SOX17 KO). Relative expression levels are shown with normalization to β−ACTIN. Error bars indicate mean ± SD from two independent biological replicates. (D) Immunofluorescence of day 4 embryoids derived from WT, SOX17 knockout (SOX17 KO), and from 1 to 1 mixture of WT and SOX17 KO 4i hESCs. The number of NANOS3-mCherry+ cells with or without SOX17 expression is shown. Quantification was based on seven to nine confocal images from four independent embryoids of each condition. Scale bars, 50 μm. (E and F) FACS analysis of TNAP and CD38 on day 5 embryoids derived from SOX17 knockout 4i hESCs containing SOX17 fusion construct with human glucocorticoid receptor ligand-binding domain (SOX17 KO+ SOX17 GR). Embryoids were derived in the presence (E) or absence (F) of cytokines with (Dex+) or without (Dex−) addition of dexamethasone. (G) RT-qPCR analysis of day 5 hPGCLC derived from WT and SOX17 KO (S17KO) and SOX17 KO + SOX17-GR (S17KO+S17GR) hESCs with (+) or without (−) dexamethasone (Dex) and in the presence (+) or absence (−) of cytokines. FACS-sorted NANOS3-mCherry/TNAP double-positive cells or whole embryoids (for S17KO) were used. Relative expression levels are shown with normalization to GAPDH. Error bars indicate mean ± SD from two biological replicates. (H) Model for establishment of hPGC transcription network by SOX17 and BLIMP1. SOX17 induces germ cell genes and, potentially, endoderm gene. Expression of BLIMP1, downstream of SOX17, suppresses endodermal genes, as well as mesodermal genes. As a result, the SOX17-BLIMP1 axis initiates hPGC program from competent cells upon induction by BMP signaling. The hPGC specification gene network is abrogated in the absence of SOX17 or BLIMP1.

Figure 7

Induction and Isolation of hPGCLCs from Competent hiPSCs/hESCs (A) FACS analysis of TNAP and NANOS3-mCherry (top) and TNAP and CD38 (bottom) on day 4 embryoids induced from 4i hESCs after preinduction (left), directly without preinduction (middle) or from conventional hESCs (right, Conv hESC). (B) FACS analysis of TNAP and CD38 in 4i hiPSCs (top) and day 4 embryoids derived from 4i hiPSCs after direct induction (bottom). (C) Expression analysis by RT-qPCR on TNAP-positive hiPSCs (iPSC TNAP+), TNAP/CD38 double-negative (TNAP−CD38−) population and TNAP/CD38 double-positive population (TNAP+CD38+) on day 4 after hPGCLC induction. Relative expression levels are shown with normalization to β−ACTIN. Error bars indicate mean ± SD from two independent biological replicates. (D) Overview of human germline development. hESCs in 4i reversibly attains competence for germ cell fate. Exposure of 4i cells to cytokines containing BMPs results in strong induction of hPGCLCs following expression of SOX17-BLIMP1, which are among the key regulators of germ cell fate. SOX17 and BLIMP1 are detected in in vivo gonadal hPGC and TCam-2 seminoma, indicating a likely progression of early human germ cell lineage. CD38, a cell-surface glycoprotein, is shared by all cells with germ cell characteristics, but not by hESC. Loss of SOX17 or BLIMP1 abrogates hPGCLC specification.

Figure S1

Generation of NANOS3-mCherry Reporter Knockin hESC Lines and hPGCLC Differentiation, Related to Figure 1 (A) Targeting strategy of generation of NANOS3-mCherry knock-in reporter hESC lines. P2A-mCherry sequence in frame with the last exon of the human NANOS3 gene was inserted. We have generated plasmids encoding TALEN molecules specific to the region covering NANOS3 stop codon. Scissors indicate TALEN cutting site. Southern blot was performed with 5′ external probe (5′ probe) and the BglII restriction enzyme sites. Correct targeting and loop-out of resistance cassette was conducted in three independent human ESC lines (WIS2, LIS1 and WIBR3). (B) Number of the NANOS3-mCherry/TNAP positive cells per embryoid during PGCLC induction with BMP2, human LIF, SCF and EGF (BMP2+L/S/E) or BMP2 alone. (C) FACS analysis for NANOS3-mCherry and TNAP positive population using WIS2-NANOS3-mCherry cell line after 4 days of hPGCLC induction by BMP2 (500 ng/ml), BMP4 (500 ng/ml) or BMP2 (250 ng/ml) and BMP4 (250 ng/ml) together with human LIF, SCF, EGF and ROCK inhibitor. Numbers show the percentage of the TNAP/NANOS3 double positive population in the boxes. (D) FACS analysis for NANOS3-mCherry and TNAP positive population using WIS2-NANOS3-mCherry cell line after 3 days of hPGCLC induction with BMP2 (50, 250 and 500 ng/ml) or without BMP2 (0) in the presence of ROCK inhibitor. (E) FACS analysis of NANOS3-mCherry and TNAP positive population from WIS2-NANOS3-mCherry cell line with PGCLC induction with BMP2 alone from day 1 to day 4. (F) Immunofluorescence of TFAP2C and NANOS3-mCherry, and OCT4 and BLIMP1 on day 4 embryoids. Scale bar = 70 μm.

Figure S2

Global Transcriptome Analysis of hPGCLC, Related to Figure 2 (A) Pearson correlation heat map of gene expression (log₂(normalized read counts)) in various samples with unsupervised hierarchical clustering. The color key indicates the correlation coefficient. (B) Two-dimensional principal components analysis of gene expression (PC3 against PC2) of the indicated samples (left panel). A corresponding loadings plot indicates the weight of various genes on PC2 and PC3 (right panel). Gene names are color-coded to illustrate association with pluripotency (black), early germ cell development (green), late germ cell development (red) and gonadal somatic cell development (blue). Arrowline indicates potential germline progression from 4i hESC via hPGCLC to gonadal hPGC.

Figure S3

Immunofluorescence for Epigenetic Modifications and Modifiers in Embryoids, Related to Figure 3 (A–C) Immunofluorescence analysis was carried out for (A) 5-methylcytosine (5mC); (B) DNMT3A; (C) UHRF1 on day 4 embryoids. TFAP2C, OCT4 or BLIMP1 were used to counterstain for hPGCLCs. hPGCLCs were highlighted by white dashed lines. White arrowheads showed examples of KI-67-positive PGCLCs while yellow arrows showed the KI-67-negative PGCLCs (C). Scale bars = 50 μm. (D) Expression level of methylation related epigenetic modifiers from RNA-Seq dataset. Mean normalized read counts from two biological replicates were shown. (E) Immunofluoresence of PRMT5 on cryosections of embryoids collected at day 1, 2, 4 and 8 post-hPGCLC induction. hPGCLCs were counterstained with OCT4 as highlighted. Scale bar = 70 μm.

Figure S4

Sequential Expression of Germ-Cell-Related Genes during PGCLC Specification, Related to Figure 4 (A–C) Immunofluoresence of (A) PRDM14 and NANOG; (B) OCT4; and (C) NANOS3-mCherry on cryosections of day 1-8. hPGCLC were counterstained with BLIMP1 or OCT4 as highlighted. Arrowheads indicate enrichment of PRDM14 in cytoplasm. Scale bars = 70 μm.

Figure S5

Generation of BLIMP1 KO in WIS2-NANOS3-mCherry hESC Line, Related to Figure 5 (A) Targeting strategy of BLIMP1 knockout in hESC with the designated guide RNA (gRNA) and the resulting deleted sequences. (B) Immunofluorescence of SOX17, NANOS3-mCherry and OCT4 (upper panel); and BLIMP1, NANOG and TFAP2C on wild-type (WT) and BLIMP1 KO day 4 embryoids. Scale bars = 70 μm.

Figure S6

Generation of SOX17 KO in WIS2-NANOS3-mCherry hESC Line, Related to Figure 6 (A) Expression analysis by RT-qPCR of pluripotency and germ cell genes after knock-down of SOX17 in TCam-2. Two independent, miRs against SOX17 (S17 miR #1 and #2) were used for the knockdown with a scramble miR as control. Error bars are mean ± SD. Relative expression levels are shown with normalization to GAPDH. Representative data were shown from two independent biological replicates. (B) Targeting strategy of SOX17 knockout in hESC with the designated guide RNA (gRNA) and the resulting deleted sequences. (C) Immunofluorescence of SOX17, BLIMP1 and TFAP2C (left panel); and TFAP2C and NANOS3-mCherry on wild-type (WT) and SOX17 KO day 4 embryoids. Scale bars = 70 μm (D) FACS analysis of day 4 embryoids derived from wild-type (WT), SOX17 knockout (SOX17 KO) and 1 to 1 mixture of WT and SOX17 KO cells (WT + SOX17 KO).

Figure S7

Comparison of hESC Transcriptomes under Conventional or 4i Culture Conditions, Related to Figure 7 and Table S2 (A) Scatter plot showing global gene expression levels (mean log₂(normalized read counts) of two replicates) between WIS2 hESC cultured under 4i and conventional condition. Each dot represents one gene. Differentially expressed genes (log₂(fold change) > 2 and adjusted p value < 0.05) were presented as red dots (upregulated in 4i condition) or blue dots (upregulated in conventional condition). (B) Gene ontology (GO) term enrichment analysis of upregulated genes in 4i WIS2 hESC (upper panel) and conventional WIS2 hESC (lower panel). Top 15 GO biological process terms that were enriched in each condition were shown (DAVID GOTERM_BP_FAT with gene count > = 5 followed by GO Trimming to reduce term redundancy). (C) Heat map showing expression of representative pluripotency and mesodermal genes expression in WIS2 hESCs cultured under 4i or conventional conditions (Conv). RNASeq data from two biological replicates were shown (#1 and #2). Asterisk indicates differential expression with statistical significance (log₂(fold change) > 2 and adjusted p value < 0.05). (D) Immunofluorescence analysis of T and OCT4 on WIS2 hESCs cultured under conventional (Conv hESC) and 4i condition (4i hESC).