Report

Generation of human oogonia from induced pluripotent stem cells in vitro

See allHide authors and affiliations

Science  19 Oct 2018:
Vol. 362, Issue 6412, pp. 356-360
DOI: 10.1126/science.aat1674

Reconstituting a human ovary

Human pluripotent stem cells (hPSCs) have been induced into human primordial germ cell–like cells (hPGCLCs) in vitro, the first step toward human in vitro gametogenesis. Yamashiro et al. went a step closer to generating mature gametes by culturing hPSCs with mouse embryonic ovarian somatic cells in xenogeneic reconstituted ovaries (see the Perspective by Gill and Peters). Over a period of 4 months, hPGCLCs underwent hallmark epigenetic reprogramming and differentiated progressively into cells closely resembling human oogonia, an immediate embryonic precursor for human oocytes. This study creates opportunities for human germ cell research and provides a foundation for human in vitro gametogenesis.

Science, this issue p. 356; see also p. 291

Abstract

Human in vitro gametogenesis may transform reproductive medicine. Human pluripotent stem cells (hPSCs) have been induced into primordial germ cell–like cells (hPGCLCs); however, further differentiation to a mature germ cell has not been achieved. Here, we show that hPGCLCs differentiate progressively into oogonia-like cells during a long-term in vitro culture (approximately 4 months) in xenogeneic reconstituted ovaries with mouse embryonic ovarian somatic cells. The hPGCLC-derived oogonia display hallmarks of epigenetic reprogramming—genome-wide DNA demethylation, imprint erasure, and extinguishment of aberrant DNA methylation in hPSCs—and acquire an immediate precursory state for meiotic recombination. Furthermore, the inactive X chromosome shows a progressive demethylation and reactivation, albeit partially. These findings establish the germline competence of hPSCs and provide a critical step toward human in vitro gametogenesis.

The germ-cell lineage arises from primordial germ cells (PGCs) that go through a multistep process to generate spermatozoa or oocytes. Methods for in vitro gametogenesis from pluripotent stem cells (PSCs) would provide a powerful tool with which to explore the mechanism of germ-cell development and its anomalies (1). Mouse PSCs have been induced into PGC-like cells (mPGCLCs), which contribute to spermatogenesis upon transplantation into testes (2, 3) and to oogenesis upon aggregation with embryonic ovarian somatic cells [reconstituted ovaries (rOvaries)], followed by transplantation under ovarian bursa (4) or an appropriate culture (5). Resultant gametes from these procedures generate fertile offspring (25). Furthermore, human PSCs (hPSCs) have been induced into hPGCLCs that bear a gene-expression property of hPGCs just after their specification, opening the possibility for human in vitro gametogenesis (6, 7). However, further differentiation of hPGCLCs has not been successful, and whether hPGCLCs can develop as mature germ cells remains unknown.

At week 2 of development, hPGCs express key transcription factors (TFs) such as SOX17, TFAP2C, and BLIMP1 (also known as PRDM1). Then, at week 5, they migrate to and colonize the embryonic gonads to initiate differentiation into oogonia or gonocytes in embryonic ovaries or testes (8, 9) that express RNA regulators such as DAZL and DDX4 [also known as human VASA homolog (hVH)] (1012). Oogonia and gonocytes are very similar in morphology, gene expression, and epigenetic properties until they start sexual differentiation at week 10 into oocytes through meiotic prophase or fetal spermatogonia (812). A characteristic event in germ-cell development is epigenetic reprogramming, which occurs by week 10 and leads to genome-wide DNA demethylation, imprint erasure, and X reactivation (1012).

We explored whether hPGCLCs can undergo further development in vitro in xenogeneic reconstituted ovaries (xrOvaries) with mouse embryonic ovarian somatic cells (fig. S1A). First, we induced male human-induced PSCs (hiPSCs) with the BLIMP1-tdTomato;TFAP2C-EGFP alleles [585B1 BTAG (XY)] into incipient mesoderm-like cells (iMeLCs) and then into hPGCLCs (7, 13). BLIMP1 and TFAP2C are expressed in oogonia and gonocytes at least until week 10 (1012). We isolated BTAG-positive (BT+AG+) hPGCLCs at day 6 of induction by means of fluorescence-activated cell sorting (FACS) and generated xrOvaries. In agreement with a previous report (5), mPGCLCs differentiated efficiently into primary oocytes and formed secondary follicles after a 21-day culture in rOvaries (fig. S1B). At culture day 7, the xrOvaries exhibited a round and flattened shape, and the BT+AG+ cells were distributed uniformly within the xrOvaries (fig. S1C). Subsequently, the xrOvaries expanded laterally with the formation of cyst-like structures (Fig. 1A). From culture days 21 to 77, the xrOvaries exhibited autofluorescence under fluorescence microscopy (fig. S1D). There were ~2000 BT+AG+ cells per xrOvary at culture day 21 and ~500 at culture day 77 (fig. S1E), indicating that only a fraction of the initial hPGCLCs (~5000) survived in xrOvaries.

Fig. 1 hPGCLC differentiation in xrOvaries.

(A) Bright-field (BF) images and FACS by BTAG of xrOvaries with 585B1 BTAG (XY) hPGCLC-derived cells from culture days (ag) 7 to 77. Scale bars, (A) and (C), 500 μm. (B) Expression of DDX4 (magenta) in AG+ cells (yellow) in xrOvaries at culture day 77, with FOXL2 (cyan) and DAPI (white) staining. The boxed area at left (merged image) is magnified at right. Scale bars, 20 μm. (C) A BF image and FACS by AGVT of xrOvaries at culture day 120 with 1390G3 AGVT (XX) hPGCLC-derived cells.

At culture day 77, the AG+ cells in xrOvaries existed as clusters, were positive for a human mitochondrial antigen, bore faint 4′,6-diamidino-2-phenylindole (DAPI) staining, and were delineated by FOXL2+ mouse granulosa cells and their basement membrane (fig. S2, A to C). The intensity of the DAPI staining in AG+ cells appeared to decline progressively during the culture (fig. S2D). The AG+ cells expressed key TFs for early germ cells (TFAP2C, SOX17, and POU5F1), and some were mitotically active (Ki67+) or in apoptosis (cleaved CASPASE3+) (fig. S2A). At culture day 77, many AG+ cells up-regulated DAZL and DDX4 (Fig. 1B and fig. S2E), suggesting that in xrOvaries, hPGCLCs not only survive as germ cells but also differentiate into oogonia and gonocytes. Accordingly, at culture day 77, electron microscopy revealed the presence of large cells, highly similar to oogonia and gonocytes, with clear cytoplasms with sparsely located mitochondria and round nuclei with loosely packed chromatin and prominent nucleoli (fig. S3) (8, 9).

We next generated female hiPSCs that bear the AG;DDX4/hVH-tdTomato alleles [1390G3 AGVT (XX)] (fig. S4) (13), created xrOvaries, and cultured them up to culture day 120. The female AG+ hPGCLCs in xrOvaries developed similarly to the male hPGCLCs (fig. S5) and up-regulated VT at culture day 77, and the AG+VT+ cells appeared to differentiate into AG-negative (AG) VT+ cells at culture day 120 (Fig. 1C and fig. S5A). We examined the expression of key genes by means of quantitative polymerase chain reaction. Both male and female hPGCLC-derived cells expressed early germ-cell genes (BLIMP1, TFAP2C, SOX17, and NANOS3), core or naïve pluripotency genes (POU5F1, NANOG, TCL1B, and TFCP2L1), and up-regulated genes for oogonia and gonocytes (DAZL and DDX4) from around culture days 35 to 49 onward, and also genes for meiosis (SYCP3 and REC8) (fig. S6). The AGVT+ cells at culture day 120 down-regulated early germ-cell and core or naïve pluripotency genes and up-regulated STRA8, a gene essential for meiosis initiation (fig. S6), suggesting their developmentally advanced character. Consistently, the DDX4+ cells at culture day 120 expressed SYCP3 (SCP3) but not γH2AX, DMC1, and SYCP1 (SCP1) proteins, indicating that they have not yet initiated meiotic recombination (fig. S7).

We analyzed the transcriptomes of these cell types (fig. S8A and table S1). Unsupervised hierarchical clustering (UHC) showed that hPGCLC-derived cells are distinct from hiPSCs and iMeLCs and can be subclassified according to their culture period (fig. S8B). Principle component analysis (PCA) gave a concordant result (fig. S8C). We identified the genes with significantly positive or negative PC1/2 loadings (453 genes) (fig. S8D and table S2). UHC classified them into five major clusters (Fig. 2A, fig. S8D, and table S2): Cluster 1 represents the genes up-regulated upon hPGCLC specification or early in xrOvaries and expressed essentially continuously thereafter. Clusters 2 and 5 signify the genes transiently expressed in early hPGCLC-derived cells or down-regulated early in xrOvaries, and cluster 4 represents the genes down-regulated upon hPGCLC specification (Fig. 2A and table S2). Cluster 3, which shows a progressive and coordinated up-regulation from culture day 35 onwards, signifies genes for oogonia and gonocytes and is enriched in genes with gene ontology (GO) functional terms for male meiosis, fertilization, and Piwi-interacting RNA metabolic process (Fig. 2A and table S2). Using the 453 genes, we compared the gene-expression properties of hPGCLC-derived cells with those of oogonia and gonocytes (12). hPGCLC-derived cells from culture day 35 onward, particularly the culture day 77 BT+AG+ and culture day 120 AG+ cells, exhibited a strong similarity to week 7 and 9 oogonia and gonocytes (Fig. 2B and fig. S10).

Fig. 2 Transcriptome dynamics of hPGCLC-derived cells in xrOvaries.

(A) (Left) UHC and heatmap of the expression in the indicated cells of the 453 signature genes for the transitions of hPGCLC-derived cell properties in xrOvaries (fig. S8, C and D, and table S2). (Right) Representative genes in each cluster and their GO enrichments (table S2). (B and C) Comparison of the expression of (B) the 453 or (C) indicated genes (14) between (B) culture day 77 BT+AG+ and culture day 120 AG+VT+ cells or (C) culture day 120 AGVT+ cells and the indicated human germ cells (12, 14).

The culture day 120 AGVT+ cells exhibited a developmentally advanced character (figs. S6, S7, and S8, B and C). To clarify this point further, we examined the expression of genes that distinguish relevant human fetal germ cells (FGCs) [mitotic (oogonia and gonocytes), retinoic acid (RA)–responsive (female), meiotic (female), and oogenesis (female)] (14) in culture day 120 AGVT+ cells, which intriguingly revealed their similarity to RA-responsive FGCs. They down-regulate genes for early germ cells; further up-regulate DAZL, DDX4, MAEL, and KRBOX1; up-regulate RA- or bone morphogenetic protein (BMP)–responsive genes (STRA8, REC8, or ID1/2/3/4, MSX1/2); yet do not sufficiently up-regulate key meiosis genes (SYCP1, DMC1, SPO11, or PRDM9) (Fig. 2C and fig. S11). Thus, hPGCLC development in xrOvaries reconstitutes human germ-cell development, albeit with protracted kinetics, leading to the generation of oogonia and RA-responsive FGCs, a state responding to the signal for the meiotic entry and in preparation for the meiotic recombination (14).

We determined the genome-wide DNA methylation [5-methylcytosine (5mC)] profiles of hPGCLC-derived cells in xrOvaries by means of whole-genome bisulfite sequencing (WGBS) (fig. S12A and table S3). The genome-wide 5mC levels of hiPSCs and iMeLCs were both ~80% and decreased progressively in hPGCLC-derived cells, reaching ~20% in culture day 77 cells and ~13% in culture day 120 cells (fig. S12B), the levels comparable with that in oogonia and gonocytes at weeks 7 to 10 (11, 12). The demethylation occurred throughout the genome (Fig. 3 and figs. S12C and S13, A and B), and the 5mC distribution profiles of the culture days 77 to 120 cells were very similar to those of oogonia and gonocytes at weeks 7 to 10 (Fig. 3), respectively, but not to those of the blastocysts (15) and naïve human embryonic stem cells (hESCs) (fig. S12D) (16, 17) [and hPGCLCs reported by others (18)]. Thus, hPGCLC-derived cells demethylate their 5mCs in a fashion similar to that of oogonia and gonocytes but not early embryonic cells and their putative in vitro counterparts.

Fig. 3 Genome-wide DNA demethylation in hPGCLC-derived cells in xrOvaries.

Comparisons of the 5mC levels (genome-wide 2-kb windows) by contour representations of scatter plots, combined with histogram representations (top and right of scatter plots), between hiPSCs (top) or ag77 BT+AG+ cells (bottom left) and ag120 AG+VT+ cells (bottom right) and the indicated cell types.

We examined whether hPGCLC-derived cells can erase the parental imprints. Blastocysts (15) and somatic cells (11) exhibited ~50% CpG methylation in the differentially methylated regions (DMRs) of paternally and maternally imprinted genes (Fig. 4A and table S4). By contrast, hiPSCs and primed hESCs (12, 16, 17) exhibited hypermethylation in some DMRs, whereas naïve hESCs (16, 17) showed hypomethylation in nearly all the DMRs (Fig. 4A), indicating that hPSCs misregulate the imprint states. Similar to oogonia and gonocytes, hPGCLC-derived cells progressively erased the parental imprints, including the hypermethylated DMRs of hiPSCs (Fig. 4A). We determined the CpG sequences that bear a hypermethylation in hiPSCs compared with three independent hESC lines (table S5). The hPGCLC-derived cells erased such hiPSC-specific methylation in a nearly complete fashion (fig. S13C). Repeat elements are also demethylated in hPGCLC-derived cells, and the demethylation-resistant repeats in vivo, such as ERVK and SVA (11, 12), showed similar resistance in hPGCLC-derived cells (fig. S13D). We defined the “escapees” that retained relatively high 5mC levels in culture day 77 to 120 cells and the week 7 oogonia and gonocytes (table S6). Essentially, all the escapees in the oogonia and gonocytes (12) were included in those in the hPGCLC-derived cells, which were enriched around SVA, ERVK, and ERV1 (fig. S13E).

Fig. 4 Imprint erasure and X chromosome demethylation in hPGCLC-derived cells in xrOvaries.

(A) Heatmaps showing the 5mC levels in the DMRs of the indicated imprinted genes in the indicated cells. (B) Heatmaps (top) and violin plots (bottom, average; red bars) showing the 5mC levels in the CpG islands on the X chromosomes in the indicated cells.

We explored whether hPGCLC-derived cells can reactivate the inactive X chromosome (Xi). RNA fluorescence in situ hybridization (RNA FISH) revealed that the 1390G3 AGVT hiPSCs (passage 29) lack the expression of XIST from both alleles and show monoallelic expression of the X-linked genes (fig. S14, A to D), indicating that they bear one active X chromosome (Xa) and one Xi without XIST (XiXIST) (19). Accordingly, the promoters of the X-linked genes in 1390G3 AGVT hiPSCs exhibited an intermediate (~50%) 5mC level (Fig. 4B and figs. S14E and S15), suggesting that they are unmethylated on the Xa and nearly fully methylated on the Xi (19). The culture day 120 cells, but not day 6 hPGCLCs, exhibited a partial (~20%) reactivation—biallelic expression—of several X-linked genes but did not reactivate XIST (fig. S14, A to D). Consistently, the promoters of the X-linked genes in culture day 120 cells were moderately demethylated (~20%) (Fig. 4B and figs. S14E and S15). These findings demonstrate that hPGCLC-derived cells in xrOvaries undergo proper epigenetic reprogramming—genome-wide DNA demethylation, imprint erasure, and extinguishment of aberrantly acquired/persisting methylation of hiPSCs—reaching minimum 5mC levels (~13%) comparable with those reported in human germ cells. Nonetheless, the XiXIST state in vitro is more resistant to reprogramming, indicating a distinctive epigenetic mechanism for the Xi in hPSCs, which warrants further investigation.

We have provided evidence that primed hiPSCs (20) are competent to generate germ cells with their hallmark of epigenetic reprogramming. In mouse-mouse rOvaries, ovarian somatic cells, likely granulosa cells, provide timely signals and environments for mPGCLCs to mature into primary oocytes and form secondary follicles (5). By contrast, in xrOvaries, mouse granulosa cells create a permissive environment for hPGCLCs to gradually mature into oogonia. Although the underlying mechanism remains unclear, because mPGCLCs undergo epigenetic reprogramming upon expansion and differentiate into oocytes in response to BMP and RA (21, 22), hPGCLC-derived cell proliferation and signals from mouse granulosa cells (fig. S16) would allow hPGCLC-derived cells to mature into oogonia or RA-responsive FGCs (fig. S17). Because both male and female mPGCLCs enter into meiotic prophase under the same condition (22), male hPGCLCs would also enter into meiotic prophase in xrOvaries. Future studies will include exploring a strategy and the mechanism for the differentiation of hiPSC-induced oogonia into oocytes with meiotic recombination or for the differentiation of hPGCLCs into fetal spermatogonia, which promote human in vitro gametogenesis.

Supplementary Materials

www.sciencemag.org/content/362/6412/356/suppl/DC1

Materials and Methods

Figs. S1 to S17

References (2334)

Tables S1 to S6

References and Notes

Acknowledgments: We thank the members of our laboratory for their helpful input on this study: K. Hayashi for his advice on rOvaries; Y. Nagai, Y. Sakaguchi, and M. Kawasaki of the Saitou Laboratory, J. Oishi of the Sasaki Laboratory, and T. Sato and M. Kabata of the Yamamoto Laboratory for their technical assistance; and the Center for Anatomical, Pathological, and Forensic Medical Research at Kyoto University for the preparation of electron microscopy. We also thank T. Mori for his encouragement and support. Author contributions: C.Y. performed hPGCLC induction and xrOvary formation and analysis. K.S., Y.K., and S.Y. assisted in hPGCLC induction, and Y.I. assisted in xrOvary formation. T.N. and T.Y. contributed to the RNA sequencing (RNA-seq) and WGBS; Y.M., K.S., and H.S. contributed to the WGBS; and Y.Y. contributed to the analyses of RNA-seq and WGBS data. I.O. performed the RNA FISH. Y.K. and M.S. designed the experiments and wrote the manuscript. Funding: This work was supported by a Grant-in-Aid for Specially Promoted Research from the Japan Society for the Promotion of Science (JSPS) (17H06098) to M.S., by a JST-ERATO grant (JPMJER1104) to M.S., by funds from the HEALIOS to K.K. and the Pythias Fund to M.S., and by a Grant-in-Aid for Scientific Research on Innovative Areas and Specially Promoted Research from JSPS to H.S. (25112010 and 18H05214). Competing interests: The authors declare no competing interests. Data and materials availability: The accession numbers for the RNA-seq and WBGS data generated in this study are GSE117101 (GEO) and DRA006618/DRA007077 (DDBJ), respectively.
View Abstract

Stay Connected to Science

Navigate This Article