Reconstitution of the cellular niche requirements for primordial germ cell-like cell progression in humans
Yolanda W. Chang, Marjolein Trimp, Talia van der Helm, Ioannis Moustakas, Albert Blanch-Asensio, Arend W. Overeem, Susana M. Chuva de Sousa Lopes
TL;DR
Researchers found that human primordial germ cell-like cells can mature and avoid dedifferentiation when cultured in specific human cellular niches, such as reconstituted fetal ovaries or amnion-like cell aggregates.
Contribution
This study demonstrates that human-specific cellular niches can support the progression of hPGCLCs without relying on mouse-derived cells.
Findings
hPGCLCs in reconstituted fetal ovaries upregulated germ cell markers and initiated meiosis.
hPGCLCs co-cultured with amnion-like cells showed less dedifferentiation and acquired migratory characteristics.
Stem cell factor (SCF) was essential for hPGCLC survival in both niche environments.
Abstract
Human primordial germ cell-like cells (hPGCLCs) can be specified from human-induced pluripotent stem cells (hiPSCs), offering a valuable model for human germ cell development. However, further maturation steps of hPGCLCs rely on mouse feeders, or co-culture with mouse gonadal somatic cells. Exposure of hPGCLCs to human cellular niche has not been attempted. Here, we co-cultured female hPGCLCs in two distinct niche compartments. In reconstituted fetal ovary (rOv) culture, human fetal germ cells proliferated and initiated meiosis, while hPGCLCs upregulated gonadal germ cell markers such as DDX4. Additionally, hPGCLCs were supported and matured into migratory hPGCLCs in 3D co-culture with amnion-like cells (AMLCs). Compared to rOv, hPGCLCs in PGCLC/AMLC aggregates were less prone to dedifferentiate. In both niches, stem cell factor (SCF) was crucial for the survival of hPGCLCs. Together,…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPluripotent Stem Cells Research · Sperm and Testicular Function · Telomeres, Telomerase, and Senescence
Introduction
Female fertility can be determined by many factors, including the size and quality of the oocyte reserve established during embryonic development. At around week 2 (WD2) of development (Carnegie Stage CS6), a founder population known as primordial germ cells (PGCs) emerges in either the posterior epiblast or the amnion of the peri-implantation embryo (Hertig et al., 1958; Kobayashi and Surani, 2018; Popovic et al., 2019). These PGCs migrate through the hindgut endoderm and mesentery, reaching the genital ridge by WD5 (CS12–13) (Gomes Fernandes et al., 2018). Upon colonizing the developing gonads, PGCs upregulate gonadal germ cell markers, such as deleted in azoospermia-like (DAZL) and DEAD box protein 4 (DDX4) progressing to (pre-meiotic) oogonia (Anderson et al., 2007; Guo et al., 2015; Heeren et al., 2016; Nicholls et al., 2019). In females, from WD11 onwards, some oogonia initiate meiosis in response to retinoic acid (RA) (Childs et al., 2011; Heeren et al., 2016; Kurilo, 1981; Le Bouffant et al., 2010). These meiotic oogonia progress to bona fide oocytes by entering dormancy at the diplotene stage (dictyate arrest) and forming primordial follicles, structures that contain one oocyte associated with one layer of flat granulosa cells inside a surrounding basal membrane (Czukiewska and Chuva de Sousa Lopes, 2022; Heeren et al., 2015).
In vitro, human PGC-like cells (hPGCLCs) can be derived from human-induced pluripotent stem cells (hiPSCs) (Irie et al., 2015; Sasaki et al., 2015). Recently, we also reported that hPGCLCs can be generated from hiPSCs alongside amnion-like cells (AMLCs) using a basal membrane extract (BMEx) overlay culture (Overeem et al., 2023). Transcriptomics analysis revealed that hPGCLCs derived using these methods resemble pre-migratory PGCs, suggesting that differentiating hPGCLCs further into oocytes is still a significant challenge (Irie et al., 2023; Overeem et al., 2023). Additionally, isolated hPGCLCs are prone to de-differentiation during culture (Kobayashi et al., 2022; Murase et al., 2020). To propagate hPGCLCs while maintaining their germ cell identity, current approaches require culturing on feeder layers with elimination of de-differentiated cells using FACS sorting, or using feeder-conditioned media (Kobayashi et al., 2022; Murase et al., 2020).
Yamashiro and colleagues have demonstrated that female hPGCLCs can differentiate into DDX4+ oogonia, when aggregated with isolated somatic cells from mouse embryonic day (E)12.5 ovaries, forming xenogeneic reconstituted ovaries (xrOvaries), and co-cultured for over 70 days (Yamashiro et al., 2018). This indicated the importance of the gonadal somatic environment for the maturation of hPGCLCs. However, it remains unknown whether hPGCLCs would undergo more efficient or further maturation when exposed to the human gonadal somatic niche, instead of the mouse gonadal somatic niche.
In this study, we demonstrated that the gonadal niche provided by second trimester human fetal ovaries supported the maturation of hPGCLCs in reconstituted fetal ovaries (rOvs), although most hPGCLCs dedifferentiated. To provide the hPGCLCs with a more developmentally matched cellular niche, we established a 3D co-culture system with amniotic cells (AMLC) and that enabled an effective transition to a migratory state, with minimal dedifferentiation. After 17 days in this culture system, hPGCLCs upregulated migratory PGC markers such as DMRT1, CDH5, and CXCR4 (Gomes Fernandes M. et al., 2018; Irie et al., 2023), matching non-human primate cynomolgus monkey CS9-CS11 PGCs transcriptionally (Zhai et al., 2022). Using migratory hPGCLCs may prove beneficial for the further optimization of successful and robust protocols for in vitro gametogenesis in humans.
Results
Human fetal germ cells proliferate and enter meiosis in reconstituted ovary aggregates
rOvs were generated by digesting second trimester human fetal ovaries (N = 4 WD16-WD19) into single cells and reaggregated in either U-bottom or V-bottom low-attachment 96-well plates for 2 days. This reaggregation method resulted in uniformly sized 3D aggregates, showing adequate integration of hPGCs in the fetal somatic niche. The 3D rOvs were then embedded in 1.5% agarose droplets and cultured further until day 12 or 22 (Figures 1A and 1B). Immunofluorescence of rOv sections revealed an increased number of POU5F1+ PGCs, DDX4+ premeiotic germ cells and DDX4+SYCP3+ meiotic germ cells over time (Figures 1C and 1D). The percentage of germ cells increased from 1.9%–7.9% on day 2 (D2) to 4.6%–20% on day 12 (D12) (Figure 1D). Comparing the proliferating (Ki67+) germ cells in WD17 ovaries to D12 rOvs, we observed higher percentages of proliferating PGCs (Ki67+ and POU5F1+) and pre-meiotic germ cells (Ki67+ and DDX4+POU5F1-) in D12 rOvs (Figures S1A and S1B).Figure 1. Maturation of fetal germ cells in rOvs(A) Schematic representing aggregation (days 0–2) and further culture (days 2–12 or 22) of rOvs.(B) Bright field images of rOvs at day 2 (D2) in a V-bottom 96-well and day 12 (D12) of subsequent culture in an agarose droplet. Scale bars, 750 μm.(C) Immunofluorescence for POU5F1, DDX4, and SYCP3 in histological sections of rOvs after D2 reaggregation and D12 culture. Scale bars, 50 μm.(D) Percentage of PGCs (POU5F1+DDX4+), pre-meiotic germ cells (DDX4+SYCP3-), and meiotic germ cells (SYCP3+DDX4+) on D2 and D12 per total cells per rOv (n = 2–4 rOvs per condition). Data are presented as average ± standard deviation.(E) Immunofluorescence for FOXL2, TP63, and DDX4; and COL4, NR2F2, and DDX4 in histological sections of WD17 ovary (left) and D12 rOvs (right). White dashed squares are shown magnified right of the image. Scale bars, 50 μm. See also Figure S1.
Next, we assessed the expression of FOXL2 and N2RF2, which mark (pre)granulosa and stromal cells respectively (Figure 1E). We observed that FOXL2+ (pre)granulosa cells and DDX4+ germ cells in the Ovs mostly localized together in a peripheral ring, resembling the organization in the fetal ovary in vivo. In addition, the stromal cells (NR2F2+) clustered together in the rOvs core and deposited collagen fibers (COL4) at the interface with the germ cell compartment, as observed in the fetal ovary in vivo (Figure 1E).
We also investigated whether agarose embedding was critical by comparing the percentage of germ cells in 3D rOvs cultured floating in low attachment 96-well plates and embedded in agarose (Figure S1C). Although comparable results were observed regarding germ cell number and distribution, agarose-cultured rOvs displayed a smoother and rounder 3D shape with evenly distributed germ cells (Figures S1C and S1D). The use of agarose droplets also allowed us to use 24-well plates, which was more practical than 96-wells for long-term culture. In summary, the rOv culture system recreated the germ cell niche, which allowed proliferation of both fetal germ cells and somatic cells and recapitulated meiotic entry.
RA promote meiotic progression up to pachytene stage in rOv culture
We examined the effects of stem cell factor (SCF), Forskolin (FK), ascorbic acid (AA), BMP2, and RA on the proliferation and maturation of fetal germ cells in rOvs (N = 4 WD16-WD19). SCF is known to be involved in PGC migration (Orth et al., 1997) and PGC survival in the gonads (Dolci et al., 1991). Several studies showed that combining SCF with FK increases the proliferation of both human and mouse PGCLCs (Murase et al., 2020; Ohta et al., 2017). In addition, AA acts as an antioxidant during gonadal tissue remodeling and apoptosis (Thomas et al., 2001), whereas RA, produced by the mesonephros, is essential for the initiation of meiosis (Bowles et al., 2006; Le Bouffant et al., 2010). BMP2 and RA together have been shown to induce meiotic entry in mPGCLCs (Miyauchi et al., 2017).
On D12, we obtained similar percentages of PGCs, pre-meiotic germ cells, and total germ cells across all culture conditions tested (Figures 2A and S2A). However, the combination of BMP2, RA and AA significantly improved the percentage of SYCP3+ cells compared to basal media (p = 0.0453, N = 4 fetal ovaries from 4 individual donors) (Figures 2A and S2A). SYCP3+ cells accounted for 9.5% of total germ cells in aRB27 (basal media) rOvs, and 25.4% in BMP2/RA/AA rOvs (Figure 2B). Culturing rOvs to D22 showed that the percentage of germ cells increased across all culture conditions (Figures 2C and 2D). More SYCP3+ cells were found in both RA and BMP2/RA/AA conditions (24% and 21%) compared to other conditions (5%–15%) (Figures 2D and S2B). Given that B27 supplement contains retinol (vitamin A), which can be converted to RA by enzymes present in the fetal gonad (Le Bouffant et al., 2010), we investigated whether fetal germ cells would enter meiosis without the presence of either retinol or RA. We observed the presence of meiotic (SYCP3+) germ cells in aRB27 without retinol or B27, although overall germ cell percentages were lower without B27 (Figure S2C).Figure 2. Effects of RA and SCF on fetal germ cells in rOvs(A) Percentage of POU5F1+ cells, DDX4+SYCP3- cells and SYCP3+ cells per total cells per rOv (N = 4 donors; n = 2–4 rOvs per condition). Data are presented as average ± standard deviation. Paired Student’s t test was used; ^∗^p < 0.05.(B) Percentage of different types of germ cells per rOv (N = 4 donors) from (A). Data are presented as average ± standard deviation.(C) Immunofluorescence for POU5F1, DDX4, and SYCP3 in histological sections of rOvs after D12 and D22 in different culture conditions. Scale bars, 50 μm.(D) Percentage of POU5F1+ cells, DDX4+SYCP3- cells, and SYCP3+ cells per total cells per rOv (n = 2–4 rOvs from WD17, per condition) at D12 and D22 in different culture conditions. Data are presented as average ± standard deviation.(E) Immunofluorescence for SYCP3, γH2AX, and HORMAD1 in histological sections of D22 rOv_RA and WD17 ovary. Merged image (left) and single channels. Scale bars, 50 μm.(F) Percentage of meiotic cells based on expression of SYCP3, γH2AX, HORMAD1 in D22 rOv_RA (n = 6 rOvs from WD17) and WD17 ovaries (N = 2 donors). Data are presented as average ± standard deviation.(G) Representative images of diplotene cells expressing TP63, ZP3, and DDX4 in WD17 fetal ovary and D22 rOv_RA. Merged image (left) and single channels. Scale bars, 10 μm.(H) Immunofluorescence for KIT, POU5F1, and DDX4 in WD17 fetal ovary, D12 rOv and D12 rOv_SCF. Merged image (top) and single channels. Scale bars, 10 μm. See also Figure S2.
We compared the stages of meiotic germ cells in rOvs to WD17 ovaries based on the expression of meiotic markers and DAPI (Figure 2F). Germ cells in leptotene display signs of chromosome condensation and expression of SYCP3, γH2AX and HORMAD1. Chromosome pairing begins in zygotene, with individual chromosome threads becoming more visible by SYCP3 and γH2AX. Due to the synapsis between sister chromatids, germ cells in pachytene are distinguished by concentrated focal spots of γH2AX and downregulation of HORMAD1. Diplotene oocytes express TP63 and ZP3, downregulate SYCP3 and are surrounded by a layer of granulosa cells (Wang and Pepling, 2021; Zickler and Kleckner, 2023). The majority of meiotic germ cells in D22 rOvs were in leptotene (45%) and zygotene (36%) and only 19% were in pachytene. In WD17 ovaries, most germ cells were in zygotene (38%) and pachytene (41%) (Figures 2E and 2F). In WD17 ovaries, around 15% of the meiotic germ cells are in diplotene (TP63+ZP3+), in contrast to cultured rOvs showing very low frequency of diplotene germ cells (Figures 2F and 2G).
SCF-KIT interaction plays an important role in anti-apoptosis and primordial follicle formation (Hoyer et al., 2005; Overeem et al., 2021); however, the addition of SCF did not influence the percentages of germ cells in rOvs (Figures 2A and 2B). In WD17 fetal ovaries, KIT was expressed by POU5F1+ PGCs, as well as POU5F1-DDX4+ germ cells but with lower intensity, whereas in rOvs cultured in both aRB27 and aRB27 with SCF, KIT was lowly expressed by POU5F1+ PGCs (Figures 2H and S2D), suggesting that the SCF-KIT interaction between germ cell and somatic cells in rOvs may be disrupted, resulting in loss of KIT. To conclude, the rOvs mimicked in vivo conditions, with fetal germ cells responding to RA and progressing through meiosis. However, meiotic progression is slower in rOvs as more leptotene cells and fewer diplotene oocytes were observed when compared to WD17 ovaries. As 20–45% of the total germ cells in rOVs remained POU5F1+ PGCs (Figure S2B), we speculated that rOvs would support POU5F1+ hPGCLCs.
The survival of hPGCLCs in rOvs depends on SCF
To investigate whether rOvs could support maturation of hPGCLCs, we generated an hiPSC reporter line harboring POU5F1::GFP and DDX4::tdTomato (PG/DT) endogenous tags (Figure S3A) to distinguish hiPSC-derived germ cells from the rOv germ cells. To coculture hPGCLCs with rOvs, PG/DT hiPSCs were first differentiated into hPGCLCs using our previously described 2D BMEx overlay method (Overeem et al., 2023). On day 5 of differentiation, GFP+ITGA6+ hPGCLCs were isolated by FACS sorting. For the reconstitution of hPGCLC-rOvs, 5,000 hPGCLCs were combined with 30,000 fetal ovary cells and allowed to aggregate per well in ultra-low attachment 96-well V-bottom well plates. After 2 days of aggregation (D2), hPGCLC-rOvs were embedded in agarose droplets and cultured for up to 30 days (Figure 3A). On D2, GFP+ hPGCLCs were detected in the hPGCLC-rOvs by live imaging (Figure 3B). Flow cytometry analysis revealed that on D14, only 1.2% of the hPGCLC-rOvs remained GFP+ when cultured in aRB27, while the addition of SCF increased the percentage of GFP+ cells to 2.3% (Figure 3C). The percentage of GFP+ hPGCLCs increased slightly on D21 and D30 in the SCF and SCF/AA/FK conditions, but dropped below 1% in aRB27 (Figure 3C). Similar results were observed across different biological replicates (N = 4 fetal ovaries from 4 individual donors) (Figure 3D), demonstrating that SCF supplementation promotes long-term survival of hPGCLCs in rOvs. Live imaging of the hPGCLC-rOvs on D30 showed GFP+ hPGCLCs in rOvs cultured in SCF and SCF/AA/FK conditions (Figure 3E). In the SCF/AA/FK condition, an increased amount of DDX4:tdTomato+ cells, also positive for germ cell markers TNAP and ITGA6, were observed (Figure 3F). This was further confirmed by colocalization of GFP (nuclei) and tdTomato (both nuclei and cytoplasm) signals in live-imaged hPGCLC-rOvs (Figure 3G).Figure 3. Maturation of hPGCLCs in the human rOvs(A) Schematic representation of the reconstitution of hPGCLCs with rOvs.(B) Image showing POU5F1::GFP positive hPGCLCs in D2 rOvs, compared to rOvs without hPGCLCs. Scale bars, 100 μm.(C) FACS plots showing the percentage of POU5F1::GFP+ ITGA6+ hPGCLCs in rOvs at D14, D21, and D30 cultured under different conditions.(D) Percentage of POU5F1::GFP+ ITGA6+ hPGCLCs in different culture conditions (N = 4 donors). The rOvs were cultured for 21–25 days. Data are presented as average ± standard deviation.(E) Image showing POU5F1::GFP and DDX4::tdTomato in D30 rOvs cultured under different conditions. Scale bars, 100 μm.(F) FACS plots showing percentage of TNAP and ITGA6 (top) and POU5F1::GFP and DDX4::tdTomato (bottom) expression in D30 rOvs cultured under different conditions.(G) Zoomed image showing POU5F1::GFP and DDX4::tdTomato in D30 rOv_SCF/FK/AA. White arrows point to double-positive cells. Scale bars, 50 μm. (H) FACS plot showing percentage of TUBA1B::GFP and ITGA6 expression in D30 rOv_SCF/FK/AA.(I) Percentage of TUBA1B::GFP+ and ITGA6+ hPGCLCs in D30 rOv_SCF/FK/AA (N = 3 donors).Data are presented as average ± standard deviation.(J) Immunofluorescence for GFP (TUBA1B::GFP), POU5F1 and DDX4 (left) and GFP (TUBA1B::GFP), SOX17 and DAZL (right) in D30 rOv_SCF/FK/AA. Yellow dashed box is magnified right of the image. Scale bars, 50 μm. See also Figure S3.
At reagregation, hPGCLCs initially comprised 14% of the hPGCLC-rOvs (5,000 hPGCLCs and 30,000 fetal ovary cells). However, by D30 less than 3% of the cells in hPGCLC-rOvs were hPGCLCs. This prompted us to investigate whether the hPGCLCs were outcompeted by fetal ovarian cells or had differentiated into other cell types. To trace all hiPSC-derived cells, we generated TUBA1B-tagged hiPSC lines (TUBA1B::GFP and TUBA1B::Scarlet) (Figure S3A) and cocultured 5,000 Tubulin-GFP+ ITGA6+ hPGCLCs with 30,000 fetal ovary cells until D30. We observed that only 1.2% of total cells in hPGCLC-rOvs were GFP+ITGA6+, whereas around 7% of total cells in hPGCLC-rOvs were GFP+ITGA6-, indicating the differentiation of TUBA1B::GFP+ cells into non-PGCLCs (Figures 3H and 3I). This effect was greatly exacerbated by the addition of RA, leading to the majority of the cells in hPGCLC-rOvs being iPSC-derived (Scarlet+), but non-PGCLCs (ITGA6-) (Figure S3B). Lastly, immunofluorescence confirmed that TUBA1B-GFP+ hPGCLCs expressed POU5F1 and SOX17, with some of them upregulating DAZL and DDX4 (Figure 3J). To summarize, we demonstrated that rOvs support the maturation of hPGCLCs to upregulate DDX4 and DAZL even if in low numbers and that unlike fetal germ cells, hPGCLCs in rOvs rely on SCF supplementation for survival.
Addition of SCF to hPGCLC-AMLC aggregates allows long-term maintenance of hPGCLCs
hPGCLCs were shown to resemble pre-migratory PGCs based on transcriptomic studies (Overeem et al., 2023), hence it is possible that the second trimester somatic niche is too mature to induce their further development. Using the BMEx overlay differentiation method (Overeem et al., 2023), hPGCLCs specified alongside AMLCs, mimicking the in vivo environment of nascent PGCs. We hypothesized that hPGCLCs would have a lower tendency to differentiate into other cell types when co-cultured with AMLCs. To test this, hiPSCs were differentiated into hPGCLCs and on day 5 the culture was dissociated, aggregated (3,000 cells/aggrewell) for 2 days (D5D2) and subsequently transferred into Vitrogel-droplets and cultured until day 17 days (D5D17) (Figures 4A and 4B). At D5D17, hPGCLC-AMLCs maintained around 35% hPGCLCs if cultured with SCF/AA/FK, consistent with the initial 35% hPGCLCs at D5, but in contrast to D5 hPGCLCs D5D17 hPGCLCs showed upregulation of PGC markers CD38 and SUSD2 (Figure 4C). Comparing different culture conditions at D5D17, barely any hPGCLCs were observed in the absence of SCF (Figure 4D), suggesting that SCF is also important for PGCLCs in hPGCLC-AMLCs.Figure 4hPGCLC/AMLC aggregates for the maintenance of hPGCLCs(A) Schematic representation of the generation of PGCLC/AMLC aggregates.(B) Brightfield images of hPGCLC/AMLC after 2 days reaggregation (D5D2) and on day 17 (D5D17) in Vitrogel droplets. Scale bars, 150 μm.(C) FACS plots showing expression of EPCAM, ITGA6, SUSD2, and CD38 in D5 hPGCLC and D5D17 hPGCLC/AMLC aggregates cultured with SCF/AA/FK.(D) Percentage of ITGA6+EPCAM+ cells in D5D17 hPGCLC/AMLC aggregates (n = 3 experimental replicates) cultured under different conditions. Data are presented as average ±standard deviation. One-way ANOVA analysis was used; ^∗∗∗∗^p < 0.0001; ^∗∗∗^p < 0.001; ns, not significant.(E) Live image showing POU5F1::GFP and DDX4::tdTomato in D5D17 hPGCLC/AMLC aggregates cultured under different conditions. Scale bars, 100 μm.(F) Wholemount immunofluorescence for SOX17 and POU5F1 (left) and HAND1 and TFAP2A (right) in D5D17 hPGCLC/AMLC aggregates cultured with SCF/AA/FK. Yellow dashed box is magnified under the image. Scale bars, 100 μm. (G) FACS plots showing expression of POU5F1::GFP and ITGA6 (left) and KIT and ITGA6 (right) in D5D17 hPGCLC/AMLC aggregates cultured with SCF/AA/FK.(H) Schematic showing different culture conditions between D3 and D5, before generating D5D17 hPGCLC/AMLC aggregates cultured with SCF/AA/FK (left) and bar graph showing the percentage of ITGA6+EPCAM+ hPGCLC at D5 (n = 3 experimental replicates) cultured under different conditions. Data are presented as average ± standard deviation.(I) qPCR analysis of KIT and POU5F1 expression at D5 (average of n = 3 technical replicates) cultured under different conditions D3-D5.(J) FACS plots showing expression of KIT and ITGA6 (left) and EPCAM and ITGA6 (right) at D5 and D5D17, cultured under different conditions between D3-D5.(K) Immunofluorescence for SOX17, KIT, and POU5F1 at D5 cultured under different conditions D3-D5. Images show merged and single channels. Scale bars, 10 μm.
Using the PG/DT hiPSC lines, we were able to visualize POU5F1:GFP+ hPGCLCs in D5D17 hPGCLC-AMLCs using 3D live imaging in particular when SCF/AA/FK was added (Figure 4E). In those culture conditions, D5D17 hPGCLC-AMLCs contained POU5F1+SOX17+ hPGCLCs and a small amount of HAND1+TFAP2A + AMLCs (Figure 4F). Interestingly, KIT was absent from hPGCLCs in D5D17 hPGCLC-AMLCs cultured with SCF/AA/FK (Figure 4G), even though hPGCLCs depended strongly on its ligand SCF for survival.
Using soluble SCF as KIT ligand can lead to the internalization of KIT (Chen et al., 2017; Miyazawa et al., 1995), therefore, we tested whether omitting SCF resulted in expression of KIT during hPGCLC differentiation. We observed that removing BMP4, LIF, EGF, and SCF from the aRB27 basal medium during D3-5 did not affect the efficiency of the differentiation into ITGA6+EPCAM+hPGCLCs at D5 (Figure 4H) or expression of POU5F1 in hPGCLCs at D5 (Figure 4I), however, not adding SCF resulted in drastic upregulation of KIT/KIT in hPGCLCs at D5 as observed by quantitative PCR (Figure 4I), FACS (Figure 4J), and immunofluorescence (Figure 4K). Surprisingly, this difference in KIT/KIT expression did not affect the survival of hPGCLCs at D5. Moreover, hPGCLC-AMLCs generated after culture D3-5 in the presence or absence of SCF and cultured until D5D17 with SCF/AA/FK yield similar percentages (33%) of hPGCLCs at D5D17 (Figure 4J).
In conclusion, SCF affected the expression of KIT/KIT but not of EPCAM, ITGA6 or POU5F1 in D5 hPGCLCs but was essential for the expression of EPCAM and ITGA6; and maintenance of hPGCLCs identity at D5D17. This suggested that exposure to SCF after specification induced hPGCLCs in hPGCLC-AMLCs to adopt a migratory fate, when SCF/KIT interaction plays an important role.
In hPGCLC-AMLCs aggregates hPGCLCs acquire migratory fate
We differentiated hiPSCs into hPGCLCs for 5 days, using SCF between D3-5, and generated D5D17 hPGCLC-AMLCs using 3 different culture conditions between D5 and D17: aRB27, aRB27+SCF and aRB27+SCF/AA/FK (Figure 5A). Subsequently, we collected D0 iPSCs, D5 hPGCLC culture and D5D17 hPGCLC-AMLCs cultured in the 3 different conditions and performed single-cell RNA sequencing (Figures 5 and S4 and Data S1). In line with our previous report (Overeem et al., 2021), in uniform manifold approximation and projection (UMAP) visualization we obtained 2 clusters of PSCs at D0 (CL1 and CL2, expressing different levels of SOX2, USP44, and ZIC5) and 3 main cell types differentiated at D5, NANOS3+SOX17+TFAP2C + hPGCLCs (CL3), VTCN1+GABRP+TFAP2A + amniotic ectoderm-like cells (AExLCs) (CL8) and HAND1+GATA6+SNAI2+ extraembryonic/amniotic mesoderm-like cells (AMxLCs) (CL7) (Figures 5B–5D, S4A, and S4B and Data S1).Figure 5. Single-cell analysis of D5D17 hPGCLC/AMLC aggregates(A) Schematics showing the generation of D5D17 hPGCLC/AMLC aggregates cultures under different conditions.(B) Uniform manifold approximation and projection (UMAP) showing cells colored by time (in days) (left) and cluster identity (right). PSC, pluripotent stem cells; EC, endothelial cells; AExLC, amniotic ectoderm like cells; AMxLC, amniotic mesoderm like cells.(C) UMAP clustering split per day and treatment condition. Dashed line demarks CL4.(D) UMAP showing expression of selected gene of interest.(E) Violin Plots showing expression of selected gene of interest in CL3 and CL4.(F) UMAP integrating CS8-CS11 cynomolgus monkey embryos and this study, colored by the given monkey cluster identity. PS, primitive streak; APS, anterior primitive streak; DE, definitive endoderm; Meso, mesoderm; Nas, nascent; LP, lateral plate; Inter, intermediate; Caud, caudal; Para, paraxial; Rostr, rostral; Pharyg, pharyngeal; Cardi, cardiac; Umb, umbilical; ExE, extra-embryonic; ys, yolk sac; EC, endothelial cell; BP, blood progenitor; Mac, macrophage; Ery, erythrocyte; NMP, neuro-mesodermal progenitor; ECT, ectoderm; NC, neural crest; FB/MB/HB, forebrain/midbrain/hindbrain; SC, spinal cord; AM, amnion; VE, visceral endoderm; Endo, endoderm.(G) UMAP integrating CS8-CS11 cynomolgus monkey embryos and this study, colored by CS of the monkey embryos (left), culture day in this study (center), and cluster number from this study (right).(H) UMAP showing expression of selected gene of interest. See also Figure S5 and Data S1.
At D5D17, the hPGCLCs formed a separate cluster (CL4) from D5 hPGCLCs (CL3), upregulating migratory PGC markers such as DMRT1 and CDH5 (Figure 5D), expressing key germ cell markers such as NANOS3, KIT, UTF1, CD38, and ITGA6 (Figures 5E and S4B and Data S1) and downregulating pluripotency markers such as LIN28A, SALL3, and DNMT3B, primitive streak markers such as MSX2 and ID genes and epithelial markers such as CDH1 and GJA1 (Figures S4A–S4C and Data S1), suggestive of a migratory fate. As expected, without SCF supplementation hPGCLCs were not efficiently maintained at D5D17 (Figure 5C).
Regarding the identity of other cells present at D5D17, we observed that AExLCs at D5 and D5D17 clustered together (CL8) and that the cluster remained small, whereas AMxLCs expanded into four separate clusters (CL0, CL5, CL6, and CL9) (Figures 5B and 5C). CL9 expressed CD93, PECAM1, CD34, and ICAM2 (Figure S4A and Data S1) suggestive of endothelial cells. CL6 expressed high levels of DLK1, POSTN, LUM, and FRZB; whereas CL0 and CL5 expressed lower levels of those markers, but high expression of GAS2, CPE, and PRRX1 (Figure S4A and Data S1). CL0 and CL5 distinguished largely by expression of proliferation markers, such as MKI67 and TOP2A (Figure 5D and Data S1).
To further identity the newly emerged cell types at D5D17, we integrated our dataset with existing single-cell data from cynomolgus monkey embryos at CS8-CS11 (Zhai et al., 2022) (Figures 5F and 5G) and a human embryo at CS7 (Tyser et al., 2021) (Figure S4D). Overall, D5 cells mostly clustered with CS8-CS9 monkey cells while D5D17 cells clustered with CS11 monkey cells (Figure 5G). Notably, CL4 D5D17 hPGCLCs clustered with monkey CS9-CS11 PGCs (monkey CL3) and human CS7 PGCs, but D5 hPGCLCs did not, suggesting that D5D17 hPGCLCs acquired a bonafide migratory fate. CL8 AExLCs clustered with monkey amniotic (monkey CL35) and surface ectoderm (monkey CL33 + 34) (Figure 5G) and human amnion (Figure S4D). Regarding the AMxLCs clusters, CL6 clustered with the monkey umbilical stalk (monkey CL18), a cluster that is lacking in the human dataset; however, unexpectedly CL0 and CL5 instead of clustering with the monkey extraembryonic mesoderm (monkey CL19) they mapped instead to monkey pharyngeal mesoderm, rostral mesoderm, and lateral plate (LP) mesoderm (monkey CL15, CL14, CL9, respectively), expressing several specific markers (Figures 5G and 5H) and human emergent mesoderm (Figure S4D), suggesting that those cells were still plastic and able to adopt an embryonic fate. In conclusion, our analysis revealed substantial changes in the cell composition of hPGCLC-AMLCs aggregates over time, but in contrast to the gonadal niche this more immature niche together with SCF supplementation proved effective to induce migratory identity in hPGCLCs.
Discussion
In this study, we successfully reconstructed the fetal ovary niche in vitro using human fetal ovarian cells from second trimester. Both fetal germ cells and (pre)granulosa cells proliferated using basal medium without additional growth factors, demonstrating that the somatic cell niche alone is sufficient for germ cell growth. The rOvs closely mimicked in vivo fetal ovaries, with fetal germ cells responding to RA and progressing into meiosis. However, compared to the WD17 ovary, rOvs exhibited significantly more leptotene cells and fewer diplotene oocytes, suggesting that meiotic progression is slower in rOvs. Remarkably, meiosis initiation still occurred in rOvs even without retinol supplementation, suggesting that human meiosis initiation may be intrinsic to the gonads and may not depend on RA signaling from the mesonephros as in mice (Bowles et al., 2006). Previously, Mizuta and colleagues reconstituted gestational week 11 (equivalent to WD9) human fetal ovaries (Mizuta et al., 2022). Instead of embedding the reconstituted ovaries in hydrogel droplets, those were cultured floating using advanced MEM supplemented with 10% fetal bovine serum for up to 14 weeks. Despite differences in culture conditions, we both observed progression of meiotic germ cells and the appearance of ZP3+ diplotene germ cells in culture, with Mizuta reporting approximately 50% of germ cells entering meiosis at week 7 of culture, while we observed around 25% at day 12 with RA/AA/BMP2 supplementation.
We showed that rOvs were suitable to induce hPGCLCs to express DDX4 and DAZL, albeit at low numbers, whereas the majority of hPGCLCs transdifferentiated into non-germline cell types. Trans-differentiation of hPGCLCs could be linked to activation of latent pluripotency, a hallmark feature of PGCs that can lead in the formation of germ cell tumors in vivo (Oosterhuis and Looijenga, 2019). However, ectopically migrating PGCs in zebrafish adopted the cell fate of neighboring somatic cells without forming germ cell tumors (Gross-Thebing et al., 2017). Similarly, disrupting the germ cell transcriptional network in mice also resulted in formation of somatic cells (Weber et al., 2010). Whether the loss of germ cell fate by hPGCLCs in rOvs mimics aspects of germ cell tumor formation in vivo remains to be investigated.
We reasoned that exposing hPGCLCs to the gonadal niche directly was not efficient because the gonadal niche would be too mature to induce their further development, therefore we co-cultured hPGCLCs with the cellular niche of amniotic cells, present at D5 differentiation (Overeem et al., 2023). We showed that when cultured for additional two weeks in the presence of this niche the hPGCLCs were still numerous and acquired a migratory identity, with upregulation of SUSD2 and CDH5, both markers of migratory PGCs (Hwang et al., 2020; Irie et al., 2023). We suggest that this migratory state may be more adequate for further co-culture in rOvs, as the gonadal niche is necessary for germ cells to upregulate DAZL (Hu et al., 2015). This approach eliminated the need for laborious isolation of hPGCLCs via FACS sorting, making hPGCLC-AMLCs aggregates a valuable model for optimizing long-term hPGCLC culture.
By comparing the co-culture of hPGCLCs with two different cellular niches, we highlighted how distinct microenvironments influenced hPGCLC development differently, although in both systems the survival of hPGCLCs after specification was SCF-dependent. In agreement, SCF is crucial for the survival of mouse PGCs (mPGCs) during migration in vivo (McCoshen and McCallion, 1975), as well as in culture (Dolci et al., 1991). In human second trimester fetal ovaries, SCF is expressed by (pre)granulosa cells and KIT is expressed by hPGCs (Overeem et al., 2021), then transiently downregulated in meiotic germ cells (Hoyer et al., 2005; Overeem et al., 2021), and SCF/KIT interaction is later crucial for primordial follicle formation in the fetal ovary (Jin et al., 2005; Jones and Pepling, 2013).
Interestingly, before hPGCLC specification, exposure to SCF seemed to led to the internationalization of KIT without further consequences for hPGCLC survival at D5. The loss of KIT in culture could be a result of receptor internalization in response to soluble SCF (Chen et al., 2017; Miyazawa et al., 1995). We hypothesized that receptor internalization further downregulates KIT expression as part of a negative feedback loop. However, once soluble SCF is removed, surface localization of KIT is restored. The downregulation of KIT in the presence of SCF and re-expression in response to SCF removal has also been reported in hematopoietic stem cells (Necas et al., 2014). Remarkably, reduced KIT expression in newly specified D5 hPGCLCs did not impair their ability to respond to SCF after specification and maintenance of germ cell fate in long-term culture.
In summary, we compared the effect of two different cellular niches in the maturation of hPGCLC, demonstrating a significant difference in the capacity to determine germ cell survival and maturation. The hPGCLC-AMLC 3D culture model proved to be a robust system to induce migratory fate in hPGCLCs, a step that may prove essential for achieving in vitro gametogenesis successfully in humans.
Limitations of the study
Migratory PGCs reach the developing gonad at WD5-6, we speculated that second trimester fetal ovaries may be too mature to induce hPGCLC development, it would be more appropriate to reconstitute hPGCLCs with WD5-6 human fetal ovaries which we rarely have access to. Nevertheless, it has been reported that when PGCLCs were co-cultured with mouse E12.5 fetal ovary somatic cells (equivalent of human WD14, prior to meiotic entry), dedifferentiation of hPGCLCs was also observed (Yamashiro et al., 2020).
Methods
Human samples and ethics statement
All experiments in this study followed the guidelines specified in the Declaration of Helsinki for Medical Research involving Human Subjects. For ethical approval, a letter of no objection was issued by the Medical Ethical Committee of Leiden University Medical Center (B21.054). The human fetal ovary samples were collected from elective abortions (no medical indication), after obtaining informed consent from the donors. In total, 11x human fetal ovaries (from N = 11 individual donors) were used in this study (4x for rOvs reconstitution without PGCLCs, 4x for rOvs reconstitution with PGCLCs, 3x for immunofluorescence). Human fetal ovaries were dissected in 0.9% NaCl solution (Fresenius Kabi). To cryopreserve fetal ovaries, the whole ovary was first incubated in 0.25% Trypsin-EDTA (Thermo Fisher Scientific) at 37°C for 30 min (min), with radial rotation at 250rpm. The trypsin was then removed, and the fetal ovary was washed with 1:1 mix of low glucose DMEM (Thermo Fisher Scientific) and M199 media (Thermo Fisher Scientific). The ovary was then cut into pieces (1mm^3^) and cryopreserved in Bambanker (Nippon Genetics).
Fetal ovary reconstitution
Cryopreserved ovarian pieces were thawed at 37°C, rinsed with DMEM:F12 media (Thermo Fisher Scientific) and digested as previously described (Overeem et al., 2023). Subsequently, 30,000 fetal ovarian cells, either with or without 5,000 FACS-sorted hPGCLCs were cultured in aRB27 basal medium [advanced RPMI1640 (Thermo Fisher Scientific) supplemented with B27 (1:100) (Thermo Fisher Scientific), Glutamax (Thermo Fisher Scientific), MEM Non-Essential Amino Acids (Thermo Fisher Scientific), and Mycozap (Lonza)] plus RevitaCell supplement (Thermo Fisher Scientific), in 96-well ultra-low attachment V-bottom or U-bottom well plates (S-bio). For floating culture, rOvs were left in ultra-low attachment wells with media change every 3 days. For agarose embedding, rOvs were embedded in 1.5% low-melting point agarose (Promega) droplets in aRB27 media with 150 μM AA (Sigma-Aldrich), 100 ng/mL SCF (R&D Systems), 5 μM FK (Biogems), 50 ng/mL BMP2 (R&D Systems), and 1 μM RA (Sigma-Aldrich). For immunofluorescence and histology details, see supplemental methods and Table S1.
hiPSC culture and hPGCLC differentiation
LUMC0099iCTRL04 iPSC line (https://hpscreg.eu/cell-line/LUMCi004-A) was cultured in mTeSR-Plus media (STEMCELL Technologies) supplemented with MycoZap (Lonza), on tissue culture plates coated with Geltrex (Thermo Fisher Scientific) diluted in DMEM/F12 (Thermo Fisher Scientific) at 1% (v/v) concentration. Cells were cultured at 37°C in a humidified normoxic incubator with 5% CO_2_. Routine clump passaging was performed every 4–7 days using ReLeSR at 1:5-1:30 split ratio (Stem Cell Technologies). Routine mycoplasma testing was performed. For generation of endogenous reporter lines, see supplemental methods and Table S2.
hiPSCs were differentiated into PGCLCs as described (Overeem et al., 2023). On day 3 and day 4, the media was switched to aRB27 containing 50 ng/mL SCF or indicated otherwise. Day 5 PGCLC cultures were dissociated by incubating with Accutase at 37°C for 15–20 min and pipetting up and down to break clumps into single cells. For FACS sorting, see supplemental methods and Table S1.
Culture of hPGCLCs in 3D aggregates with AMLCs
Day 5 PGCLC culture were dissociated into single-cell suspension, 9 × 10^5^ cells were added per well of Aggrewell 800 (24-well format) (Stemcell Technologies) in mTeSR-Plus media supplemented with RevitaCell. The plate was centrifuged at 100g for 3 min. On day 2, the aggregates were collected, washed with aRB27 media and 1/10 of the total aggregates were embedded per droplet in 50 μL droplets of 50% diluted 3D high concentration Vitrogel (diluted with Vitrogel dilution solution Type I) (The Well Bioscience). The droplets containing the aggregates were allowed to solidify (inverted plate) at 37°C for 15 min, then aRB27 media with supplements was added gently to the well. Media exchange was performed every 2–3 days for up to 2 weeks. On day 17, the aggregates were recovered by rinsing with DPBS^−/−^ (Thermo Fisher Scientific) and mechanically breaking the droplets. For immunofluorescence, histology, RNA isolation, and quantitative reverse-transcription polymerase chain reaction (RT-PCR) details, see supplemental methods and Table S1.
Imaging and quantification
Images were acquired on either 200 or 500 series Dragonfly spinning disk confocal microscope (Andor). Quantification of the PGC, meiotic germ cells and oogonia was done manually by counting cells on the largest cross-section of the rOvs (n = 1–4 rOvs per condition per experiment). Total cell number present in each cross-section was calculated based on DAPI segmentation using Stardist (Caicedo et al., 2019).
For quantification of KIT expression, tissue sections were immunostained and imaged together using the same settings. In ImageJ (v2.3.0) (Schindelin et al., 2012), gray values (corresponding to the intensity of a pixel in a grayscale image) per image were calculated using “Plot Profile,” the background value was subtracted and the area under the curve was calculated for the total gray value. Finally, mean fluorescence intensity per germ cell was calculated by dividing by the total number of germ cells in the image.
Single-cell RNA sequencing
For details regarding sample preparation for single-cell RNA-sequencing, see supplemental methods. Raw RNA sequencing reads were processed using Cell Ranger (v7.1.0). Reads were aligned to the human reference genome (GRCh38), and gene UMI count matrices were obtained using gene annotation in Cell Ranger reference annotation version GRCh38-2020-A. The count matrix was analyzed using Seurat (v5.3.0) in R (v4.4.3). For details regarding further data analysis and integration with CS7 human embryo data (Tyser et al., 2021) and CS8-CS11 cynomolgus monkey data (Zhai et al., 2022), see supplemental methods.
Statistical analysis
Column plots were generated using either GraphPad Prism 9 software or numpy and matplotlib packages. Statistical analyses were performed using GraphPad Prism 9. Specifically, for Figure 2A paired Student’s t test was used to compare % meiotic germ cells in aRB27 with BMP/RA/AA conditions. For Figure 4D, first the data were confirmed to have normal distribution with Shapiro-wilk test, then one-way ANOVA with Tukey’s multiple comparison test was used for multiple pairwise comparisons.
Resource availability
Lead contact
Further information and requests should be directed to Dr. Susana M. Chuva de Sousa Lopes ([email protected]).
Materials availability
Cell lines generated in this study are available from the lead contact.
Data and code availability
Raw scRNA-seq data from this study have been deposited in Gene Expression Omnibus (GEO). The accesssion number for the data reported in this paper is [Database]: [[GSE309100](GSE309100)]. Data from CS7 human embryo data were retrieved from ArrayExpress [Database]: [E-MTAB-9388] (Tyser et al., 2021); and from CS8-CS11 cynomolgus monkey embryos from [Database]: [[GSE193007](GSE193007)] (Zhai et al., 2022). Code for data analysis is available at https://github.com/chuvalab/PGCLCs (https://doi.org/10.5281/zenodo.17311400).
Acknowledgments
We thank the patients who donated the fetal material used as well as the staff of Gynaikon and Vrelingshuis. We thank Dr. K. Szuhai, Dr. C. Freund, Dr. R. Davis, and members of the Chuva de Sousa Lopes group for discussions, and S. Hillenius and S. Czukiewska for technical assistance. This work was supported by the 10.13039/501100003246Dutch Research Council, the Netherlands (VICI-2018-91819642 to Y.W.C., A.W.O., I.M., and S.M.C.d.S.L.), the Nederlandse organisatie voor gezondheidsonderzoek en zorginnovatie (10.13039/501100001826ZonMw), the Netherlands (PSIDER-10250022120001 to T.V.D.H. and S.M.C.d.S.L.), and the 10.13039/501100009708Novo Nordisk Foundation, Denmark (reNEW NNF21CC0073729 to Y.W.C., A.W.O., M.T., I.M., and S.M.C.d.S.L.).
Author contributions
Conceptualization, Y.W.C., A.W.O., and S.M.C.d.S.L.; methodology, all authors; investigation, all authors; formal analysis, Y.W.C., A.W.O., and S.M.C.d.S.L.; writing, all authors; funding acquisition: S.M.C.d.S.L.; supervision: A.W.O. and S.M.C.d.S.L.; all authors approved the final version of the manuscript.
Declaration of interests
The authors declare no competing interests.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Anderson R.A.Fulton N.Cowan G.Coutts S.Saunders P.T.Conserved and divergent patterns of expression of DAZL, VASA and OCT 4 in the germ cells of the human fetal ovary and testis BMC Dev. Biol.7200713610.1186/1471-213X-7-13618088417 PMC 2211489 · doi ↗ · pubmed ↗
- 2Bowles J.Knight D.Smith C.Wilhelm D.Richman J.Mamiya S.Yashiro K.Chawengsaksophak K.Wilson M.J.Rossant J.Retinoid signaling determines germ cell fate in mice Science 312200659660010.1126/science.112569116574820 · doi ↗ · pubmed ↗
- 3Caicedo J.C.Goodman A.Karhohs K.W.Cimini B.A.Ackerman J.Haghighi M.Heng C.Becker T.Doan M.Mc Quin C.Nucleus segmentation across imaging experiments: the 2018 Data Science Bowl Nat. Methods 1620191247125310.1038/s 41592-019-0612-731636459 PMC 6919559 · doi ↗ · pubmed ↗
- 4Chen D.Liu W.Lukianchikov A.Hancock G.V.Zimmerman J.Lowe M.G.Kim R.Galic Z.Irie N.Surani M.A.Germline competency of human embryonic stem cells depends on eomesodermin Biol. Reprod.97201785086110.1093/biolre/iox 13829091993 PMC 5803789 · doi ↗ · pubmed ↗
- 5Childs A.J.Cowan G.Kinnell H.L.Anderson R.A.Saunders P.T.K.Retinoic Acid signalling and the control of meiotic entry in the human fetal gonad P Lo S One 62011 e 2024910.1371/journal.pone.0020249 PMC 310859421674038 · doi ↗ · pubmed ↗
- 6Czukiewska S.M.Chuva de Sousa Lopes S.M.Fetal germ cell development in humans, a link with infertility Semin. Cell Dev. Biol.1312022586510.1016/j.semcdb.2022.03.03535431137 · doi ↗ · pubmed ↗
- 7Dolci S.Williams D.E.Ernst M.K.Resnick J.L.Brannan C.I.Lock L.F.Lyman S.D.Boswell H.S.Donovan P.J.Requirement for mast cell growth factor for primordial germ cell survival in culture Nature 352199180981110.1038/352809 a 01715518 · doi ↗ · pubmed ↗
- 8Gomes Fernandes M.Bialecka M.Salvatori D.C.F.Chuva de Sousa Lopes S.M.Characterization of migratory primordial germ cells in the aorta-gonad-mesonephros of a 4.5-week-old human embryo: a toolbox to evaluate in vitro early gametogenesis Mol. Hum. Reprod.24201823324310.1093/molehr/gay 01129528446 PMC 6018722 · doi ↗ · pubmed ↗