Cell culture

Cynomolgus monkey female (XX) and male (XY) embryonic stem cells (ESCs) were cultured on mitomycin C-inactivated mouse embryonic fibroblasts (MEFs) as previously described [10]. Briefly, ESC colonies were cultured in Nutrient Mixture F-12 (DMEM/F12, Invitrogen) medium containing 20% KOSR (Invitrogen) supplemented with 4 ng/mL bFGF (Chemicon), 1% nonessential amino acids (Invitrogen), 0.1 mM β-mercaptoethanol (Invitrogen), and 1% penicillin–streptomycin-L-glutamate (Invitrogen). For 4i-state conversion, ESC colonies were dissociated by collagenase IV and reseeded on MEFs with a 4i conversion medium as previously reported [11]. The 4i conversion medium comprised ESC culture medium supplemented with 10 ng/mL human LIF (PeproTech), CHIR99021(3 µM) (Tocris), PD0325901 (0.5 µM) (R&D Systems), SP600125 (10 µM) (Selleck), and SB203580 (10 µM) (Selleck). The medium was changed daily. All chemicals were from Sigma-Aldrich unless otherwise stated.

Monkey PGCLCs differentiation

To induce PGCLC differentiation, 4i-state cells were dissociated with TrypLE Express (GIBCO), then the cells were seeded into ultra-low cell attachment U-bottom 96-well plates (Corning) at a density of 3000 cells/well in 200 µL PGCLC differentiation medium containing Glasgow's MEM (GMEM, GIBCO), 15% KOSR, 0.1 mM nonessential amino acids, 0.1 mM β-mercaptoethanol, 1% penicillin–streptomycin-L-glutamate, 1 mM sodium pyruvate, 250 ng/mL BMP4 (R&D Systems), 250 ng/mL BMP2 (PeproTech), 1 µg/mL human LIF (PeproTech), 100 ng/mL SCF (PeproTech), 50 ng/mL EGF (PeproTech), and 10 µM ROCK inhibitor, Y-27632 (Selleck). The medium was changed daily.

Immunofluorescence

EBs were harvested on day 2 and day 4 and fixed in 4% paraformaldehyde for 30 min at 25°C. Then the EBs were rinsed thrice with phosphate-buffered saline (PBS) and dehydrated with sucrose solutions of increasing concentrations of 15%, 20%, and 30% (wt/v) each for 2 h. Next, EBs were embedded in optimal cutting temperature compound (Tissue-Tek OCT, Sakura Finetek), frozen, and stored at –80°C. Before immunofluorescence staining, samples were prepared as cryosections with an 8-µm thickness on pretreated glass slides (CITOGLAS) and air-dried for 1 h. Subsequently, the slides were washed with PBS thrice for 5 min each. Then, the slides were permeabilized in PBST (PBS with 0.2% Triton X-100) for 30 min, washed with PBS thrice for 5 min each, and blocked in 3% (w/v) bovine serum albumin in PBS for 1 h. Slides were then incubated with primary antibodies overnight at 4°C. Subsequently, the slides were washed in PBST thrice for 10 min each. Then, Alexa Fluorophore (488, 594, and/or 647)-conjugated secondary antibodies were incubated with the slides in the dark at 25°C for at least 1 h. Images were taken using a Leica TCS SP8 confocal microscope (Leica). After wash, fluorescence-conjugated secondary antibodies, and 4′,6-diamidino-2-phenylindole (DAPI) were incubated with the slides in dark at 25°C for 2 h. Images were taken using Leica TCS SP8 confocal microscope (Leica). In vitro cultured cynomolgus monkey embryos [12] and 4i condition or MEF cells were included as positive or negative controls, respectively. For antibodies used, see  Table S1 (table_s1_ioac133.doc).

Quantification of the proportion of PGCLCs within EB during differentiation

Cryosections from day 2 (three EBs) and day 4 EBs (three EBs) were stained with BLIMP1 TFAP2C, OCT4, and T antibodies. Further, 500 cells within day 2 or 4 cryosections were randomly selected and the proportion of BLIMP1, TFAP2C and double-positive cells or OCT4, T, and double-positive cells were calculated. The experiments were repeated thrice.

Quantification of the Ki67-positive PGCLCs and 5hmc levels of PGCLCs

For quantification of the Ki67 positive PGCLC cells, cells cultured in 4i condition and cryosections from day 2 (three EBs) and day 4 EBs (three EBs) were stained with BLIMP1, TFAP2C, and Ki67 antibodies. Further, 500 cells within 4i condition day 2, and day 4 cryosections were randomly selected and the proportion of SOX17, TFAP2C, Ki67, and triple-positive cells were calculated. Notably, the proportion of Ki67 of cells in the 4i condition was 100% although SOX17 and TFAP2C were not expressed. The experiments were repeated thrice.

For quantification of the 5hmc levels of BLIMP1/SOX17 double-positive PGCLCs, the mean fluorescence intensity (MFI) values were collected using the mask channel option on 5hmC and DAPI channels. Relative intensity values for individual nuclei were calculated by dividing the MFI of 5hmC by the MFI of DAPI.

Quantitative real-time polymerase chain reaction

Total RNA was extracted using TRIzol reagent (Invitrogen). Complementary DNA (cDNA) synthesis was performed using 1 µg of total RNA with HiScript II 1st Strand cDNA Synthesis Kit (Vazyme). Quantitative polymerase chain reaction (qPCR) analysis was performed according to the manufacturer's instructions and monitored with SYBR green PCR master mix [FastStart Universal SYBR Green Master (Rox); Roche] using the Applied Biosystems QuantStudio 7 Real-Time PCR Detection System (Thermo Fisher). PPIA/ß-ACTIN was used as a control for normalization. Fold-change in mRNA expression level was normalized to the cycle threshold using naïve cells. The primers used are listed in Table 1.

Table 1.

Primers used for qPCR

The scRNA-seq library preparation and sequencing

The scRNA-seq libraries were generated using the Chromium Single Cell 3′Reagent Kit (10× Genomics). Briefly, a single-cell suspension (500 cells/µL in PBS) was mixed with real-time quantitative (RT-PCR) master mix and loaded together with Single Cell 3′Gel Beads and Partitioning Oil into a Single Cell 3′Chip (10× Genomics) according to the manufacturer's instructions. RNA transcripts from single cells were uniquely barcoded and reverse transcribed within droplets. cDNA molecules were preamplified and pooled followed by library construction. All libraries were quantified on a 2100 Bioanalyzer (Agilent) with RT-PCR and then subjected to 150-bp paired-end sequencing on an Illumina Xten platform (sequenced by Novogene).

The scRNA-seq data processing

Sequencing data were first processed using the 10× Genomics Cell Ranger pipeline ( https://www.10xgenomics.com) and sequences were aligned using STAR [13] against the Macaca fascicularis reference genome (ensemble release-91). Subsequently, unique molecular identifier (UMI) matrices from each sample were merged in R and imported to Seurat [14]. Cells with less than 700 detected genes and more than 4000 genes, or less than 700 UMIs and more than 30 000 UMIs were excluded from further analysis, as well as cells with more than 10% of UMIs mapped to mitochondrial genes. Next, the raw UMI matrix was normalized by LogNormalize and scaled by ScaleData. Highly variable genes were identified from these cells using FindVariableGenes with the following parameters: x.low.cutoff = 0.0125, x.high.cutoff = 3, and y.cutoff = 0.5.

Principal component analysis was then performed, and the top 10 principal components were used for t-distributed stochastic neighbor embedding (t-SNE) and unsupervised clustering. Clusters with fewer than 200 cells were omitted from further analysis. Differentially expressed genes of each cluster were produced by FindAllMarkers. Cell type identifications were based on known genetic markers. Gene Ontology (GO) analysis was performed with ClusterProfiler [15], and GO terms with P-value less than 0.05 were considered statistically significant.

Single-cell-pseudotime analysis

For pseudotime analysis, the UMI matrix calculated in Seurat was passed into Monocle2 [16], a computational method that performs lineage trajectory reconstruction based on scRNA-seq data, and highly variable genes identified by Seurat were used as ordering genes for trajectory calculation. After trajectory reconstruction, we used BEAM to identify genes with branch-dependent expression and used plot_multiple_branches_heatmap to visualize these genes along the branches.

Statistical analysis

The results were presented as mean ± standard error of the mean. Significance was assessed using unpaired t-tests (two-tailed); *P<0.05, **P<0.01, and ***P<0.001; ns, nonsignificant. Statistical analyses are described in detail in the figure legends for each panel.

Data availability

All sequencing data were deposited at the NCBI GEO under accession number GSE139432.

PGCLC differentiation from male and female monkey ESCs

We detected expression of OTX2 ( Figure S1A (fig_s1_ioac133.pdf)), a marker of human formative PSCs [17, 18] in monkey PSCs cultured in a previously reported 4i-based medium and cells showed higher correlation coefficients to human formative PSCs [17, 18] than naïve PSCs [19] ( Figure S1B (fig_s1_ioac133.pdf)), indicating the capability of PGCLC specification, male and female monkey ESCs were cultured in 4i conversion medium for at least two passages as previously reported before PGCLC induction (9). The cells cultured in the 4i conversion state formed dome-shaped colonies (Figure 1A and  Figure S1C (fig_s1_ioac133.pdf)) and expressed pluripotent markers, such as OCT4, SOX2, and NANOG. These cells also expressed primate-specific PSC surface markers, such as TRA-1-81 (Figure 1B and  Figure S1D (fig_s1_ioac133.pdf)). Then, 3000 4i-state ESCs were used to form EBs in low-attachment 96-well plates for 4 days. Cell aggregates with clear boundaries could be observed on day 2, and EBs had gradually grown up on day 4 with increased diameters ( Figure S1E (fig_s1_ioac133.pdf)). RT-qPCR analysis confirmed PGCLC differentiation. As shown in Figure 1D, after 2 days of induction, EBs began to express conserved and primate-specific PGC markers, such as PRDM1 (also known as BLIMP1), TFAP2C, KIT, DDX4 (also known as VASA), and SOX17. On day 4, the expression levels of these PGC markers peaked, although levels of SOX17 and PRDM1 were slightly decreased, whereas OCT4 expression was relatively stable throughout differentiation (Figure 1D). Immunof luorescence staining of PGC markers showed that the ratios of BLIMP1/TFAP2C and SOX17/BLIMP1 double-positive cells increased during the 4-day differentiation [male: 1.29% (n = 620) to 20.21% (n = 876) and 26.71% (n = 307) to 33.80% (n = 639), female: 6.57% (n = 624) to 15.44% (n = 1036) and 25.59% (n = 758) to 31.38% (n = 835), respectively] (Figure 1E and  Figure S2A (fig_s2_ioac133.pdf)). Furthermore, the positive ratios of other PGC makers, such as OCT4 (POU5F1) and T also increased during the differentiation of male and female cells ( Figure S2B (fig_s2_ioac133.pdf)). In addition, we also found a significant increase in 5-hydroxymethylcytosine (5hmC) in monkey PGCLCs (mPGCLCs) during differentiation, which is consistent with a study in humans ( Figure S2C (fig_s2_ioac133.pdf)). The PGCLCs induction in our modified approach was reproducible as the percentage of SOX17/BLIMP1/TFAP2C triple-positive PGCLCs increased from day 2 to day 4 within male and female EBs (average 6.9% in day 2 male EBs, 12.49% in day 4 EBs, 10.06% in day 2 female EBs, and 18.17% in day 4 female EBs) (Figure 1E). Taken together, these data indicated that PGCLCs were induced from monkey PSCs after induction in a 4i medium.

Figure 1.

PGCLC differentiation of naïve-state monkey ESCs. (A) Representative bright-field images showing colony morphologies of female primed and 4i-state ESCs. Scale bar = 50 µm. (B) Representative immunofluorescence images of female 4i-state ESCs stained with anti-OCT4 and anti-NANOG antibodies. Scale bar = 50 µm. All experiments were performed in triplicates, unless otherwise stated. (C) Development of female EBs on days 2 and 4. Scale bar = 50 µm. All experiments were performed in triplicates, unless otherwise stated. (D) Dynamics of the expression of key genes during PGCLC induction by real-time PCR. Expression is normalized to ß-ACTIN or PPIA. Fold change was calculated relative to the expression levels of each gene in 4i-state cells, which was given a value of 1.0. All experiments were performed in triplicates, unless otherwise stated. (E) Immunostaining of BLIMP1/TFAP2C and BLIMP1/SOX17 expression in days 2 and 4 female EBs (left panel). Percentage of BLIMP1/TFAP2C and BLIMP1/SOX17 cells in day 2 and day 4 female EBs (middle panel). Scale bar = 50 µm. Percentage of SOX17/BLIMP1/TFAP2C triple-positive PGCLCs within days 2 and 4 EBs (right panel, shown is mean and SEM). ns, not significant; *P < 0.05. n, independent experiments, at least four cell lines were used.

Dissecting the cell types within EBs during PGCLC differentiation

As cell types within EBs during PGCLC specification from PSCs are elusive, we next examined the cell types within EBs. The 4i-state cells and EBs (96 EBs from day 2 and 96 EBs from day 4) were dissociated into single cells and subjected to high throughput droplet-based scRNA-seq analysis as in previous reports [20, 21]. After quality filtration, we obtained transcriptomic profiles of 38 193 cells from six samples (female and male 4i-state cells, and day 2 and day 4 EBs ( Table S2 (table_s2_ioac133.docx)). The average number of cells detected was 6478 (range from 4200 to 9005), and genes detected were 2442 (range from 746 to 3997), with a mean coverage of 10 472 reads (UMI) per cell. Also, the average percentage of reads mapping to mitochondrial genes (percent. mito) was 2.3% (range from 0% to 9.9%), suggesting our cells are with high cell viability ( Figure S3A (fig_s3_ioac133.pdf)).

Then, the t-SNE analysis was performed and eight clusters of cells were identified (Figure 2A). Based on the expression of marker genes, the identity and percentage of each cluster were investigated (Figure 2B and  Figure S3B (fig_s3_ioac133.pdf)). Cluster 1 was a hepatocyte with an expression of AFP and F2, two well-known markers of hepatocytes [22–24]. Cluster 2 was trophoblast, as cells in cluster 2 expressed trophoblast markers, such as KRT7 and WFDC2 [25]. Cluster 3 was identified as 4i cells according to the expression of the pluripotent markers such as SOX2 and OCT4 (POU5F1) [26]. Cluster 4 was differentiating (Differ.) cells within the 4i population, as they expressed the differentiated cell markers, such as STMN2 and FOXA2 [27, 28]. Interestingly, we identified two types of PGCLCs (cluster 5 and cluster 6): PGC1 (cluster 5) and PGC2 (cluster 6). Both of these cell types expressed PGC markers, such as TFAP2C and POU5F1 (OCT4) [6, 8], whereas PGC2 cells expressed genes involved in epithelial-to-mesenchymal transition (EMT), such as MMP9, SNAI2 (also known as SLUG), and TGFβ1 [29–31]. Cluster 7 was identified as neural progenitor cells (NPCs), as they expressed SOX2 and PCSK1N, the two well-known NPCs makers [32]. Finally, the last identified cell type (cluster 8) was mesenchymal cells (cluster 8), which expressed COL3A1 and COL1A1 [33]. In addition, the levels of previously reported G1/S- and G2/M-specific genes [34, 35] were analyzed in cells within EBs. Interestingly, PGCLCs were relatively more quiescent than the 4i-state cells, and there was no difference between PGC1 and PGC2 cells ( Figure S3C (fig_s3_ioac133.pdf)).

To verify cell identities, the differentially expressed genes of each cluster of cells were further identified and GO analysis was performed (Figure 2C and  Table S3 (table_s3_ioac133.xlsx)). Genes involved in cellular homeostasis and blood coagulation were enriched in hepatocytes (cluster 1). Trophoblast cells (cluster 2) were enriched for specific genes involved in tissue morphogenesis, etc. The 4i-state cells (cluster 3) specific genes were enriched in GO terms such as RNA processing process. Differ. cells (cluster 4) expressed genes related to tissue morphogenesis, epithelium development, and embryonic morphogenesis. PGCLCs (clusters 5 and 6) expressed genes involved in cellular modified amino acid metabolic process, mitochondrial membrane organization, cell migration, etc. Genes belonging to the nervous system, head, and tissue development were enriched in NPCs (cluster 7). Finally, mesenchymal cells (cluster 8) expressed genes enriched in GO terms for tissue development, stem cell differentiation, and embryo development. Overall, our findings revealed the cellular heterogeneity within EBs and suggested the trophoblast, endoderm (hepatocyte, cluster 1), mesoderm (mesenchymal cells, cluster 8), and PGCLC lineage were the major cell types in the differentiation system.

EBs as a niche for monkey PGCLC specification

As discussed earlier, we identified eight clusters of cells during PGCLC differentiation. To reconstruct the developmental trajectory of all cell types, pseudotime trajectory analysis was applied following a previously described method [36] (Figure 3A). We found that all cells formed a Y-like trajectory: 4i-state PSCs occupied the one branch of the pseudotime trajectory, and endoderm (hepatocyte) and mesoderm (mesenchymal cell) lineages were the right and left branches, respectively (Figure 3B). We also found that PGC1 cells were located with 4i-state PSCs in pseudotime trajectory, whereas part of the PGC2 cells was located in the mesoderm lineage branch (Figure 3B), suggesting PGC1 and PGC2 might be early- and late-stage PGCs, respectively. Expression patterns of early- and late-stage PGC marker genes [37] also confirmed this speculation (Figure 3C). As a small portion of 4i cells (named 4i-overlap) were located with PGC6, we investigated the differentially expressed genes between PGC6 and 4i-overlap cells. As small numbers of 4i cells (named 4i-overlap) were located with PGC2, we investigated the differentially expressed genes between PGC6 and 4i-overlap cells. We observed that 80 genes were upregulated in 4i-overlap cells, such as EEF1A1, HMGA1, and EEF1G, suggesting rapid proliferation of these cells ( Figure S3D (fig_s3_ioac133.pdf)). Collectively, these findings imply that although small numbers of 4i cells overlap with PGC2 in pseudotime trajectory, they display distinct transcriptome profiles.

We next determined the gene expression dynamics along the site pseudotime trajectory. As shown in Figure 3D, five clusters of gene expression patterns (here, “here” referred to as cluster 1–5) from prebranch (4i-state PSCs and PGC1) to the endoderm (hepatocyte) or mesoderm (mesenchymal cell and part of PCG2 cells) lineages were observed. Genes enriched in clusters 1 and 2 were highly expressed in the endoderm lineage, and genes belonging to clusters 3 and 5 were specifically expressed in prebranch cells (4i-state PSCs and PGC1) and the mesoderm lineage. Finally, cluster 4 represented genes enriched in the mesoderm lineage. We then performed GO analysis to reveal the functional enrichment of these clusters. For example, mRNA catabolic process (clusters 3 and 5), and ossification were enriched in prebranch, and genes belonging to lipid catabolic process, cell cycle regulation, and response to nutrient levels were enriched in other branches (clusters 1, 2, and 4) ( Figure S3E (fig_s3_ioac133.pdf)). Collectively, these findings revealed the unique developmental trajectory of cells within EBs.

As extra-embryonic mesenchyme (EXMC), amnion, and cytotrophoblasts cells formed a niche to induce PGC specification from amnion in vivo through secretion of BMP4, BMP2, and WNT3A [28], we wondered whether EBs also provided a niche for PGCLC specification as in vivo. We determined gene expression levels of the classical signaling pathways (NOTCH, WNT, TGF, and BMP) within EBs (Figure 3E). We found that WNT2, TGFβ1, and TGFβ2 were highly expressed in mesenchymal cells, hepatocytes expressed BMP2 and WNT5A, and trophoblast cells expressed BMP4 and WNT3. Moreover, the expression of downstream genes of the WNT and TGFβ superfamilies were upregulated in PGC1 and PGC2 cells, for example, RUVBL1, CTBP2, CSNK2A1, and ID1–4 (Figure 3E). The expression of BMP2, BMP4, WNT3A, and WNT5A in day 4 EBs was validated by immunostaining ( Figure S3F (fig_s3_ioac133.pdf)).

Figure 2.

Single cell transcriptomic profiles and cell type identification. (A) t-SNE plots of male and female 4i-state cells, days 2 and 4 EB cells. Cells are colored by identified cell types. (B) A bubble plot showing expression of cluster-specific genes. The color of the dot represents the average expression and the size represents the percentage of cells expressing that gene. (C) Heatmaps showing the differential gene expression patterns of each cluster. Representative genes and key gene ontology enrichments are shown. (D) t-SNE plots of male and female 4i-state cells, day 2 and day 4 EB cells. Cells are colored by differentiation time points (days 2 and 4).

Next, the expression of hepatocyte (HNF4A), trophoblast (CK7), and PGCLC (SOX17) markers within EB was detected by immunofluorescence staining. We observed that on day 4 the majority of SOX17- and HNF4A-positive cells were located on the outer surface of EBs and were enclosed by trophoblast cells (CK7-positive) (Figure 3F). Moreover, SNAI2 expression was also confirmed in part of the SOX17-positive population (Figure 3F), consistent with our transcriptomics data that PGC2 had migratory characteristics and expressed EMT genes. Taken together, these findings indicated that cells within EBs formed an autocrine niche to induce PGCLC differentiation by secreting growth factors important for PGCLC specification.

Figure 3.

Reconstruction of the niche for PGCLC induction. (A) Pseudotime analysis of the differentiation trajectory and cell types was plotted in the pseudotime trajectory (B). (C) Expression of early- and late-stage PGC markers in PGC1 and PGC2. The color key from purple to yellow indicates low to high expression, respectively. (D) Heatmap showing the gene expression dynamics during PGCLC differentiation. Genes are clustered according to expression patterns, and cell types are colored. (E) Dot plots showing expression of members of the WNT, TGFβ, and FGF signaling pathways in identified cell types. (F) Representative immunofluorescence images of EB cryosections stained with anti-SOX17, anti-HNF4A, anti-CK7, and anti-SNAI2 antibodies. Arrow indicates SOX17- and HNF4A-positive cells were enclosed by trophoblast cells (CK7 positive) and arrowhead indicates SNAI2 and SOX17 double-positive cells. Scale bar= 50 µm. All experiments were performed in triplicates unless otherwise stated.

Figure 4.

Transcriptional differences between male and female PGCLCs. (A) t-SNE plots of PGCLCs. Cells are colored by differentiation time points (days 2 and 4). (B) t-SNE plots of PGCLCs. Cells are colored by cell types (M, male or F, female). (C) GO term enrichment analysis of genes enriched in male and female PGCLCs. (D) Quantification of cells (n = 1000 cells) with expression of SOX17, TFAP2C, and Ki67, respectively (black line) or both (red line) in 4i-state and in EB cryosections. Scale bar= 50 µm. All experiments were performed in triplicates unless otherwise stated.

Transcriptional landscape of monkey male and female PGCLCs specification in vitro

We next determined the transcriptomic landscape of PGCLCs during in vitro differentiation. As the majority of PGCLCs emerged on day 4 (Figure 2D and Figure 4A), we wondered whether there were differences in the timing of PGCLC specification between male and female cells. As shown in Figure 4B, besides a portion of female germ cells emerging on day 2, the majority of male PGCLC differentiation occurred on day 4, suggesting that the differentiation rate of male cells might be slower than that of female cells. We then sought to determine the developmental stage (early or late) of PGCLCs. We found that the majority of female PGCLCs were early-stage (PGC1) while male PGCLCs were both early- and late-stage (PGC1 and PGC2) ( Figure S4A (fig_s4_ioac133.pdf)).

Next, to dissect the differences in gene expression patterns between male and female PGCLCs, differentially expressed genes were identified in male and female PGCLCs. GO analysis indicated that genes enriched in female PGCLCs were related to translation and peptide synthesis, while those in male PGCLCs were enriched in the GO terms regulation of binding and response to hypoxia (Figure 4C and  Table S4 (table_s4_ioac133.xlsx)), implying that male PGCLCs might be migrating, as hypoxia promotes the migration of murine germline stem cells [38]. As the ratio of PGCLCs decreased beyond day 4 of differentiation in both humans [6], we next determined whether the proliferation of PGCLCs was impaired during prolonged differentiation. As shown in Figure 4D, the percentage of Ki67-positive PGCLCs was decreased, suggesting mitotic arrest might be a reason for the loss of PGCLCs during differentiation.

Developmental trajectory of PGCLCs

We showed that the population and differentiation of PGCLCs were highly asynchronous and heterogeneous. To reconstitute the PGCLC developmental trajectory, we dissected subpopulations of PGCLCs using t-SNE analysis. Based on the t-SNE analysis, PGCLCs could be clearly grouped into six clusters: IGF2-expressing PGCs (cluster 1), HPGD-expressing PGCs (cluster 2), PDPN-and IGTA6 expressing PGCs (cluster 3), YWHAZ-expressing PGCs (cluster 4), CD24-expressing PGCs (cluster 5), and KRT18-expressing PGCs (cluster 6) (Figure 5A and B). Also, we verified that PDPN and IGTA6 were expressed in PGCLCs by immunostaining on day 4 EBs ( Figure S4B (fig_s4_ioac133.pdf)). Pseudotime analysis was then used to investigate the cell fate transitions of PGCLCs. Further, PGCLCs formed a “U-like”structure, with cluster 5 (CD24-expressing PGCs) and cluster 6 (KRT18-expressing PGCs) occupying the beginning heads and other PGCLCs at the tail and sub-branch at the turn (Figure 5C). Thus, pseudotime analysis delineated the stepwise developmental path of PGCLC differentiation in vitro.

We next examined gene expression dynamics of all branches within the pseudotime trajectory (states 1–3; clusters 1–7). We found three types of gene expression patterns: genes enriched in prebranch (state 2; clusters 5 and 6) and other states (states 1 and 3; clusters 1–4, and 7) PGCLCs, respectively (Figure 5D). According to GO analysis, translation- and peptide biosynthetic-related genes were enriched in prebranches. Response to biotic stimulus, microtubule-based processes, and ribosome biogenesis-related genes were enriched in other branches (states 1 and 3) ( Figure S4C (fig_s4_ioac133.pdf)).

To evaluate whether our monkey PGCLC induction system could be used to study primates (humans and monkey) conserved or monkey-specific mechanisms underlying PGC specification, we compared marker gene expression patterns between in vivo human PGCs (hPGCs) [35] and in vitro mPGCLCs. We found that expression patterns of a portion of PGC markers were conserved between hPGCs and mPGCLCs, such as TFAP2C and POU5F1, while the expression patterns of another portion of genes were species-specific, such as NANOS3 (Figure 5E).