Reconstitution of the Mouse Germ Cell Specification Pathway in Culture by Pluripotent Stem Cells

Hayashi et al., 2011. Saitou Lab.

In mice, germ cell fate is induced in epiblasts at E6.0 by Bmp4 signaling from extraembryonic ectoderm. The primordial germ cells (PGCs) are established at E7.5 as small cluster of alkaline-phosphate positive cells (AP+) in the extraembryonic mesoderm. Blimp1 and Prdm14 regulate PGC fate. All epiblast cells from E5.5-E6.0 express Blimp1 and Prdm14 in response to Bmp4, than the PGC-like cells derived from epiblasts ex vivo can for functional sperm when transplanted into neonatal testes lacking endogenous germ cells. 

{Summary: germ cell fate: Blimp1 + Prdm4 (in epiblast cells E5.5-E6.0) are upregulated in response to Bmp4 (from extraembryonic ectoderm E6.0) --> PGC cells (in extraembryonic ectoderm E7.5) --> sperm}

Pluripotent cells in mice are comprised of (1) inner cell mass (ICM) of preimplantation blastocyst (E3.5-E4.5) and (2) the epiblast of postimplantation embryos  (E5.5-E6.5). These two populations induce into two pluripotent stem cell populations. The ICM gives rise to embryonic stem cells (ESC) and the epiblast give rise to epiblast stem cells (EpiSCs).

ESCs and EpiSCs show distinct morphology, cytokine dependency, gene expression and epigenetic profiles. ESCs bear  the ground state (naive) pluripotency and can contribute to all linages when introduced into the blastocyst. Whereas, EpiSCs are primed pluripotent and are not able to contribute to chimeras  when introduced into the blastocyst. Interestingly, hESCs are more like mEpiSCs, making it difficult for us to capture the hESCs in nonrodent species. 

There are several attempts to grow sperm in vitro starting with ESCs in both mice and humans based on germ cell markers, but these experiments yielded less than 1% of sperm cells which never produced healthy offspring. In addition, some also demonstrated that EpiSCs can also be potential to generate germ cells in vitro, because its subpopulation expresses Blimp1 under self-renewing process and a minority of them are Stella (Pgc/Dppa3) positive. Stella is a marker for established PGCs. But the PGCs population induced from EpiSCs occurs at low frequency even in the presence of BMP4, and their function in vivo has not been demonstrated. 

Therefore, in this study, they have uncovered the conditions that ESCs and PSC (iPSCs) with naive pluripotency  are induced into pregastrulating epiblast-like cells, which derive into PGC-like cells in mice system to produce viable and fertile offspring. [The reason they did not use EpiSCs to induce PGCs in these experiments because they could not find a suitable conditions to improve previous results in the old studies].

Pregastrulating epiblast-like cells from ESCs: (1)ESCs are cultured in serum-free and feeder-free medium implemented with MAPK inhibitor, GSK inhibitor, and leukemia inhibitor factor LIF to allow these ESCs express a uniform property similar to the ICM/preimplantation epiblast at ground state (E4.5). (2) These ESCs could produce PGCs in a few days after ESCs were introduced into the blastocysts. (3) Continuous presence of Activin A (ActA) and basic fibroblast growth factor (bFGF), ESCs converted into EpiSC-like cells exhibiting similarities to the postimplantation (E6.5) epiblasts!!!!! They reasoned that "ESCs might rapidly differentiate into pregastrulating epiblast-like cells with high competence for the PGC fate under conditions similar to those used to induce EpiSC-like cells."

ESCs were derived from E3.5 blastocysts carrying Blimp1-mVenus and Stella-ECFP (BVSC) transgenes. The higher the %KRS, the higher the maintenance of ESC-like state. Stimulation of ActA and bFGF without KRS resulted in an increased cell death rate. After 2 days cultured in ActA, bFGF, and 1% KRS, even though ESCs are induced into epiblast-like cells (EpiLCs), they quickly die after that, and both ESCs and induced EpiLCs are absent of BVSC expression. Nanog and Blimp1 were down-regulated in EpLCs but continue to express in EpSCs. EpiLCs have properties consistent with pregastrulating epiblasts, but EpiSCs do not.  

PGC-like cell induction from EpiLCs: Next they examined if the EpLCs could be induced into PGC-like cells when culture in conditions that cause induction of epiblast cells into PGCs (GMEM + 15% KRS = GK15, cytokines including BMP4). They induce PGCs using ESCs and d1/2/3 EpiLCs for 2d in GK15 medium added with LIF but no BVSC signal was observed. They then cultured ESCs and d1/2/3 EpiLCs in medium containing BMP4 or BMP4 with LIF. Strong BV induction (40%) was observed in d2 and d3 EpiLCs, but not in ESCs or d1 EpiLCs. d2 EpiLCs are highly competent to express Blimp1 in response to BMP4 and cell growth. 

They next examined if  d2 EpiLCs can produce BVSC+ PGC-like cells when culture for longer periods of time. When cultured in full induction condition (GK15 with BMP4, BMP8b, LIF, stem cell factor [SCF]), epidermal growth factor (EGF), BV+ got improved to 16.9% and BVSC+ was 7.2% on d6. This result demonstrated that BVSC+ PGC-like cells were induced by BMP4 and the maintenance/proliferation of BVSC+ cells was enhanced by LIF and even more robustly by LIF, SCF, BMP8b, and EGF. The BVSC+ cells are AP-positive and remained up to 10d under full induction condition. Then, they compared BVSC induction from d2 EpiLCs wth that from epiblast. The efficiency and dynamics of these two BVSC+ cells from these two cell types were the similar.  These results convinced that d2 EpiLCs bear similar, if not identical, properties to pregastrulating epiblast cells. 

Then they evaluated the gene expression profile of PGC-like cells (PGCLCs) induced from d2 EpiLCs. They found that the gene expression dynamics associated with PGCLC induction  were very similar to those associated with PGC specification. They also noted that on d6, the endogenous stella expression in BV+SC- cells was very similar to that in BVSC+ cells, so BV+SC- cells should be considered established PGCLCs. 

Global transcription profiles of EpiLCs and PGCLCs:
The global transcription profile of PGCLC induction was examined from isolated ESCs, d1/2/3/ EpiLCs, EpiSCs, E7.5 epiblasts, and BVSC+ PGCs at E9.5. Based on PC2 scores, the UHC map of nonamplified samples showed that  EpiLC induction from ESCs was a directional and progressive process. EpiSCs were clustered distantly from the other samples, suggesting their divergence from other cell types. The UHC map of amplified samples showed that d2 EpiLC and PGCLCs clustered most closely with E7.5 epiblast and E9.5 PGCs while EpiSCs were more distantly from other cell types. They concluded that the pathway of PGCLC induction from d2 EpiLCs was parallel to that of E9.5 PGC formation from E5.75 epiblasts. [Moreover, the EpiSC induction from epiblast involves a different pathway]. Overall, they demonstrated that the PGCLC formation from EpiLCs derived from ESCs  is a recapitulation of PGC formation form epiblast.  In addition, EpiSCs have more upregulated genes associated with a variety of organ systems than E5.75 epiblasts and d2 EpiLCs, saying that EpiSCs acquire more developmentally advanced characteristics than E5.75 epiblast and d2 EpiLCs.  

Epigenetic Profiles of the PGCLCs:
They evaluated epigenetic profiles of PGCLCs. In ESC to EpiLC differentiation, H3K9me2 and 5mC levels increased and H3K27me3 decreased; but in EpiLCs to PGCL, H3K9me2 and 5mC levels decreased and H3K27me3 increased. 

They next examined PGLCs for the imprinting states of maternally and paternally  imprinted genes. A global decrease of 5mC with relative maintenance of imprinting in PGCLCs was a characteristic observed in migrating PGCs in vivo. Slow growth and G2 cell cycle arrest are characteristics of migrating PGCs. Therefore, PGCLCs is a reconstitution of PGC formation. 

Spermatogenesis and normal offspring from PGCLCs:
PGCLCs induced after 6d were FACS-sorted for BV+ cells. These cells when being transplanted into seminiferous tubules of W/Wv neonatal mice lacking endogenous germ cells resulted in normal spermatogenesis with thicker tubules that contained spermatozoa with normal morphology. Not sorted PGCLCs when transplanted caused testis to develop teratomas (tumor). The efficiency of colonization of the PGCLCs was similar to that of PGCs in vivo.  [Thin tubules usually contained only Setoli cells]. The fertilized oocytes via ICSC with these spermatozoa derived from PGCLCs. They produce viable and healthy offspring (males and females) with normal placentas and imprinting patterns. 

Identification of surface markers for PGCLCs isolation: 
The surface markers of PGCLCs essential for isolating them from stem cells when there are no transgenic reporters bc these PGCLCs are derived from iPSCs or ESCs. SSEA1-high-and-integrin-b3-high population P1 are 99% BV+, correlating well with BV+ PGCLCs. In contrast,  SSEA1-high-and-integrin-b3-low (P2)  or SSEA1-low-and-integrin-b3-high/low (P3) yielded < 2% of BV+. 

To test if this P1 population from AAG ESCs could contribute to spermatogenesis, they transplanted these cells into seminiferous tubules of W/Wv neonatal mice lacking endogenous germ cells. There were no teratomas in testis and showed normal spermatogenesis with GFP signal by AAG transgenes. When fertilizing oocytes with these spermatozoa via ICSC, they produced healthy, fertile offspring.  Therefore, SSEA1 and integrin-b3 could serve as surface makers to isolate PGCLCs from ESCs without contamination of teratomas and relevant transgenic markers. 

PGCLCs. spermatogenesis, and offspring from iPSCs:
They wanted to explore if germ cell production could be possible using iPSCs. iPCS 20D17, 178B-5, 492B-4 lines were used for the experiments. All three lines express Nanog-EGFP (NG) transgene at ground state and shoed differentiation into EpiLCs with proper morphology and NG downregulation. Upon PGCLCs, the NG level was upregulated as early as d2 and NG+ cells formed clusters around the periphery of the aggregates at day 4/6. 

When using FACS to sort SSEA1 and integrin-b3 at d6, only 20D17 line yield similar sorting pattern with ESC lines, and more than 50% of the population was NG+. Consistent with migrating PGCs, the P1 population contained both NG-high and NG-low cells. The expression gene profile of 20D17 line was comparable with BV+ PGCLCs. They then transplanted the P1 isolated from 20D17 population into seminiferous tubules of W/Wv neonatal mice lacking endogenous germ cells. Normal spermatogenesis was observed and norma, fertile offspring were collected. Therefore, although iPSCs exhibit different induction properties depending on the lines, they can nontheless form PGCLCs with proper function. 

Conclusions:
They induced pregastrulating epiblast-like cells (EpiLCs) from ground state ESCs that were maintain by 2i and LIF under serum free and feeder free condition. EpiLCs induction involves Act1 and bFGF, the same cytokines and 1%KRS that are required for the derivation of EpiCLs. d2 EpiLCs were robustly, but not ESCs, d1/3 EpiLCs, or EpiSCs, induced into the PGCs, so only E5.5-E6.0 serve as an efficient precursor for the PGC fate. EpiLCs can serve as a starting material for the induction of other linages derived from the epiblast. 40% of these induced PGCs acquired Blimp1  at a sufficient level proceed to PGCLCs with appropriate genetic, epigenetic, and cellular properties that are compatible to BV+ PGCs formation from epiblast ex vivo.

The identified of surface markers SSEA1 and integrin-b3 for PGCLCs isolation from iPSCs or ESCs. iPSC 20D17 has the highest capacity for germ line transmission among the three lines they used. It was reasoned that efficiency if germline transmission of iPSCs depends highly on Myc gene, and only 20D17 line out of three was derived from retroviral stable transduction of c-Myc. Spermatogenesis of the PGCLCs derived from the iPSCs depends on the original properties of the iPSC lines.