To obtain embryos, ICR females (Charles River Laboratories) were mated with Oct4-deltaPE-GFP transgenic male mice on the B6D2F1 background (Yoshimizu et al., 1999). Female mice were used between 6 and 25 weeks of age. All the animal experiments were performed under the ethical guidelines of the Kindai University, and animal protocols were reviewed and approved by the Animal Care and Use Committee of Kindai University and the Tohoku University Animal Studies Committee.

Oct4-deltaPE-GFP transgenic male mice were mated with ICR female mice in the afternoon and checked for the presence of vaginal plugs the following day. The day on which a plug was found was E0.5. The developing hindgut encapsulating migrating PGCs in an embryo at E8.75 was surgically isolated with a tungsten needle in cold Dulbecco’s Modified Eagle’s Medium (DMEM_high glucose, L-glutamine, and sodium pyruvate plus) (nacalai tesque, 08458-16) supplemented with 10% Fetal Bovine Serum (FBS) (Gibco, 10270-106) and Penicillin-Streptomycin solution (P/S) (1x) (Wako, 168-23191). An isolated hindgut was dissociated with 0.25% trypsin/1 mM EDTA for 10 min at 37°C and then plated on mitomycin C (MMC) (FUJIFILM, 139-18711)-inactivated STO feeders in single well of a 4-well plate (SPL Life Sciences, SPL-30004) in the medium consisting of 15% FBS-containing DMEM medium supplemented with 10 µM SB239063 (Cayman Chemical, 19142) or 10 µM SB203580 (Cayman Chemical, 13067) or the solvent (Dimethyl Sulfoxide: DMSO, nacalai tesque, 09659-14) as a control, and P/S (1x). The medium was changed every 2 days. After the time described in the graph (12 h and 2–6 days, Figure 1B) in culture, the cultured cells were subjected to alkaline phosphatase (ALP) staining to count ALP-positive surviving PGCs. Images were captured with a microscope (Keyence, BZ-X710).

For the establishment of cell lines derived from PGCs, an isolated hindgut and genital ridge where PGC colonized were dissociated with 0.25% trypsin/1 mM EDTA for 10 min at 37°C and then plated on MMC-inactivated STO feeders in a single well of a 4-well plate in the medium described below (detailed media compositions are described in the Supplementary Table S1). DF15^2iLIF medium was used for EGC derivation, and DF15 plus p38 MAPK inhibitor (DK15^SB23 or DK15^SB20) medium was used for cEGLC derivation (see Supplementary Table S1). Medium was changed every 2 days. At 12–14 days after plating, the appeared Oct4-GFP-positive colony-forming cells were picked up, harvested, and then dissociated using TrypLE (Gibco, 12604-013), and transferred onto freshly prepared MMC-inactivated STO feeders for further cultivation. Established EGC and cEGLC lines were cultured on MMC-STO and passaged every 3–4 days at a split ratio of 1:20 in DK15 medium consisting of 15% Knockout Serum Replacement (KSR, Gibco, 10828028)-containing DMEM instead of FBS for better proliferation (supplemented with LIF for EGC instead of 2i compounds, with SB203580 for cEGLC-20 and with SB239063 for cEGLC-23 at the same concentrations as in the derivation of each cell line) (see Supplementary Table S1).

Established mouse ESC lines in ICR background were cultured on MMC-MEF in DF10^2iLIF (Okamura et al., 2021) (see Supplementary Table S1). Derivation and culture of rsEpiSC lines derived from E6.5 mouse epiblasts in ICR background were performed in chemically defined N2B27 medium supplemented with 20 ng/mL FGF2 (Peprotech, 100-18B) and 2.5 µM IWR-1 (Wnt inhibitor, Cayman Chemical, 13659) as described (Wu et al., 2015) (see Supplementary Table S1).

For inhibition of signal transduction essential for the maintenance of pluripotency in conventional naïve and primed state, chemical compounds as potent and selective inhibitors with JAK inhibitor I (JAKi) (Janus kinase inhibitor, Millipore, 420099), SU5402 (Fibroblast growth factor receptor (FGFR) inhibitor; FUJIFILM, 197–16731), SB431542 (Activin pathway inhibitor, Cayman Chemical, 13031), LY294002 (PI3K inhibitor, Cayman Chemical, 70920) were administered at 1 µM (JAKi), 10 µM (SU5402), 10 µM (SB431542), 10 µM (LY294002), respectively, when the cells were passaged. After 4 days with compounds, the cultured cells were applied for immunofluorescence for OCT4 expression to be considered whether an undifferentiated state.

Dissociated 1 × 10⁶ cells (EGC and cEGLC-23) were plated on 35 mm petri dishes (Falcon, 351008) in “embryoid body formation medium” (EB formation medium, see Supplementary Table S1). After 4 days in culture, images were captured with a microscope (Olympus, CKX-53), and then embryoid bodies were processed for immunohistochemistry for GATA4 expression or for further differentiation. To demonstrate the ability to differentiate into three germ layers, formed embryoid bodies cultured in suspension for 4–5 days were transferred to a 0.2% gelatin-coated well in a single well of a 4-well plate for further attachment culture. After the embryoid bodies attached to the dish in “embryoid body attachment medium” (see Supplementary Table S1), the medium was changed to the media described below for further cultivation for 7 days. For ectoderm differentiation, “ectoderm differentiation medium” (see Supplementary Table S1) was used, and for mesoderm- and endoderm differentiation, “mesoderm and endoderm differentiation medium” (see Supplementary Table S1) was used, and prolonged cultured outgrowths of the attached embryoid body were then processed for immunohistochemistry to confirm the differentiation into three germ layers.

Hindguts encapsulating Oct4-GFP-expressing PGCs were isolated from Oct4-GFP mouse embryos at E8.75 and then plated on MMC-inactivated STO-feeder cells in “organ culture medium” (see Supplementary Table S1) supplemented with or without 10 µM SB239063 in a single well of a 4-well plate for hindgut outgrowth. After 8 days in culture, a hindgut outgrowth was scraped off with a pipette chip, dissociated with 0.25% trypsin/1 mM EDTA for 10 min at 37°C, and replated on MMC-inactivated STO-feeder cells in “organ culture medium” (see Supplementary Table S1) without SB239063. Oct4-GFP-positive surviving PGCs and colony-forming cells were observed under a microscope (Keyence, BZ-X710).

For immunofluorescence studies, cells grown in a 4-well plate or cryosections of tissue were fixed with freshly prepared 4% paraformaldehyde in PBS for 15 min at room temperature and permeabilized/blocked with 1% Triton-X, 1% BSA, and 10% FBS in PBS for 1 h at room temperature. Cells and tissue sections were incubated with primary antibodies in 1% FBS, 1% BSA, and 0.1% Triton-X in PBS overnight at 4°C. The next day, the cells were washed three times in 0.1% Triton/PBS and incubated with fluorescence-labeled IgG (H + L) secondary antibodies (Abcam, goat anti-mouse_Alexa Fluor 488: ab150113, _Alexa Fluor 594: ab150116, goat anti-rabbit_Alexa Fluor 488: ab150077, _Alexa Fluor 594: ab150080, donkey anti-goat_Alexa Fluor 594: ab150132) at 1:500 dilution or incubated with HRP-conjugated goat anti-mouse IgG (H + L) (for brown staining) (Proteintech, SA00001-1) secondary antibody at 1:200 dilution in 1% FBS, 1% BSA, 0.1% Triton-X in PBS for 2 h at room temperature. Cells were washed three times, filled with 0.1% Triton/PBS, and nuclei were counterstained with DAPI (Sigma-Aldrich, D9542) for immunofluorescence. For brown staining, a peroxidase Staining DAB kit (nacalai tesque, 25985-50) was performed according to the manufacturer’s instructions. Primary antibodies used in this study include Anti-phosphorylated-p38 MAPK (p-p38 MAPK, diluted at 1:200; Cell Signaling Technology, 4511S), Anti-OCT4 (C-10) (diluted at 1:300, Santa Cruz, sc-5279), Anti-SOX-2 (E-4) (diluted at 1:300, Santa Cruz, sc-365823), Anti-OTX2 (diluted at 1:200, R&D systems, AF 1979), Anti-β-Tubulin III (diluted at 1:1000, Sigma-Aldrich, T2200), Anti-SMA/Actin/α-Smooth Muscle antibody (diluted at 1:600, Sigma-Aldrich, A5228), Anti-HNF-3β (diluted at 1:1000, Santa Cruz), Anti-GATA4 (G-4) (diluted at 1:250, Santa Cruz, sc-25310). For immunofluorescence analyses using the p-p38 MAPK antibody with migrating PGCs, analyses were performed as follows: Isolated hindguts were fixed in 4% paraformaldehyde in PBS overnight at 4°C, followed by incubation in 10% sucrose/PBS overnight at 4°C and then in 20% sucrose/PBS overnight at 4°C. Fixed tissues were embedded in OCT (Tissue-Tek) for cryosectioning at 10 μm, and staining was performed by immunofluorescence described above. Observation was made under a microscope (Keyence, BZ-X710).

Cultured cells were fixed in 4% paraformaldehyde in PBS for 15 min at room temperature. The cells were washed three times in PBS and then stained with the ALP staining solution. The ALP staining solution was prepared as follows: 5 mg of naphthol AS-MX phosphate disodium salt (Sigma, N5000) was dissolved in 500 µL of N, N-Dimethylformamide, and then 10 mg of Fast Red TR Salt Hemi (Sigma, F8764) or Fast Blue RR salt (Wako, 069-03712) was dissolved in the above solution, and the solution was transferred to 9.5 mL of 0.1 M Tris-HCl (pH 9.5). In addition, the staining solution was filtered through a 0.2 µm filter to remove the formed precipitate and left on the fixed cells for more than 10 min at room temperature for better development. The stained cells were then washed three times in PBS and filled with PBS for observation under a microscope (Keyence, BZ-X710).

2 × 10⁶ cells of EGCs or cEGLCs were resuspended in a 1:1 mixture of 100 µL of ice-cold PBS with 100 µL of Matrigel (Corning, 354234) and injected subcutaneously into male immunnodeficient nude mice (KSN/Slc) at 8–16 weeks of age to form teratomas. After 4 weeks, the mice were sacrificed by cervical dislocation, and teratomas were excised, fixed in 4% paraformaldehyde in PBS, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin-eosin for histological analysis.

To generate chimeric embryos, 10 cells from each EGC or cEGLC-23 line were injected into ICR background blastocysts, which were then transplanted into the uteri of pseudopregnant ICR females.

Total RNA was extracted using RNeasy Mini Kit (QIAGEN, 74104) according to the manufacturer’s instructions. RNAs were reverse transcribed using ReverTra Ace qPCR RT Master Mix (TOYOBO, FSQ-201), and real-time PCR was performed using THUNDERBIRD SYBR qPCR Mix (TOYOBO, QPS-201) in MIC qPCR (bio molecular systems). The relative expression levels of each gene were normalized by Gapdh and calculated by the comparative CT method. Primer sequences are listed in Supplementary Table S2.

All sequenced libraries were aligned to the mouse reference genome (UCSC mm10) using STAR (v2.7.0c) (Dobin et al., 2013). Hierarchical clustering analysis was performed using the R hclust function with Euclidean distance. Multi-dimensional scaling (MDS) analysis to show the variation among RNA-seq samples conducted by plotMDS function. Differential expression gene analysis was performed with edgeR (v3.32.1) (Robinson et al., 2009) package using gene count matrix generated by FeatureCounts (v.1.5.2) (Robinson et al., 2009). Sequencing libraries were prepared using the NEBNext Ultra II Directional RNA Library Prep Kit (NEB) and were sequenced on an Illumina Novaseq platform to generate 150 bp paired-end reads. The adaptors of the obtained raw sequence data were trimmed using Trim Galore (v0.6.3), and quality control using FastQC (v0.11.7). Processed reads were aligned to the mouse reference genome (UCSC mm10) using STAR (v2.7.0c) (Dobin et al., 2013). FeatureCounts v1.5.2 was used to generate the read-counting data for each gene. Differential gene expression analysis was performed using TCC (Sun et al., 2013; Tang et al., 2015). MDS plot analysis and Heatmap clustering were performed using the princomp, and heatmap packages from R.

KEGG pathway analysis was performed using clusterProfiler (Yu et al., 2012), and the connections between genes and biological concepts were displayed using the cnetplot function.

Statistical analysis was performed using one-way and two-way ANOVA for multigroup comparison, when significant, group differences were evaluated using the Student’s t-test. p values <0.05 were considered as statistically significant.

To search for chemical compounds involved in PGC reprogramming, we particularly focused on molecular pathways that suppress pluripotency and teratoma formation in germ cells. Since reprogramming of PGCs could be involved in the onset of self-renewal as pluripotent stem cells, in this screening we took advantage chemical inhibitors that target signaling molecules or epigenetic modifiers that converts migrating PGCs into Oct4-GFP-positive colony-forming cells. We used the developing hindgut as a screening material: firstly because it encapsulates all developing PGCs, which is important to avoid bias due to heterogeneity of PGCs; and secondly because the proliferation of surrounding somatic cells after plating into culture dishes is far less than other tissues encapsulating the developing PGCs such as emerging gonads, thereby resulting in easy detection of colony formation of reprogrammed PGCs. We screened approximately 50 different compounds at the various concentrations to monitor induction of colony-forming cells derived from cultured dissociated hindguts of Oct4-GFP embryos at E8.75. This led to identification of SB239063, a p38 MAPK inhibitor, as a candidate chemical compound (Figure 1A). After day 2 in culture, PGCs positive for ALP activity increased and the number of these cells peaked out. Of note, administration of SB239063 in culture significantly supported the survival or proliferation of PGCs (Figure 1B). Furthermore, ALP- and Oct4-GFP-positive cells formed clusters after 4 days in culture in the presence of SB239063, while Oct4-GFP expressing cells were dispersed in control media (Figures 1B,C). After around 12 days in culture, some colony-forming cells expressing Oct4-GFP appeared only when treated with SB239063, and the colonies subsequently expanded in size (Figure 1D). Assuming that the colony-forming cells were derived from PGCs, activated p38 MAPK protein was likely expressed in migrating PGCs in the developing hindgut and may therefore have some function, particularly in inhibiting reprogramming of PGCs to pluripotent stem cells. Consistent with this notion, phosphorylated p38 MAPK protein was clearly and exclusively detected in a sub-population of migrating Oct4-GFP-expressing PGCs at E8.75, but not in surrounding somatic cells (Figure 1E).

In addition to the E8.75 embryo hindguts, we attempted to establish cEGLCs using emerging gonads encapsulating PGCs. However, no Oct4-GFP-positive colony-forming cells emerged from the dissociated emerging gonads at E11.5 and E12.5 in both males and females, cultured under the same conditions with hindguts (Supplementary Table S3).

SB203580, another p38 MAPK chemical inhibitor, also induced Oct4-GFP-expressing colony-forming cells in the same culture system using isolated hindgut (Supplementary Figure S1A). Picked-up colonies of Oct4-GFP-expressing cells could be maintained long-term by passaging with trypsinization (Figure 2A and Supplementary Figure S1B). This indicated that derivation of pluripotent stem cell lines could be performed under novel conditions using p38 MAPK inhibitors. Cell lines derived with SB239063 or SB203580 were named chemically induced Embryonic Germ-like Cell with SB239063 (cEGLC-23 for abbreviation) or cEGLC-20, respectively.

Surprisingly, cEGLCs-23 and -20 could be maintained long-term as Oct4-GFP-expressing cells without LIF supplementation (Figure 2A and Supplementary Figure S1B), whereas EGCs derived from migrating PGCs continuously required LIF (Figure 2A). EGCs displayed a dome-shaped colony morphology that is a representative characteristic of naïve pluripotency, whereas cEGLCs exhibited a more spread and flattened colony morphology indistinguishable from STO-feeders until 2 days after passage, suggesting that cEGLCs may not possess representative characteristics of naïve pluripotency (Figure 2A). However, even without Y-27632, a ROCK inhibitor, the cloning efficiency in cEGLCs was equivalent to that in EGCs (Figure 2B). In addition, there were no significant differences between EGC and cEGLC lines in proliferation rate (Figure 2C) and expression levels of pluripotent markers in terms of transcription and protein (Figure 2D and Supplementary Figure S1C). The ability of the cEGLC lines to differentiate into three germ layers was confirmed by induction of embryoid body differentiation (Supplementary Figure S2) and teratoma formation (Figure 2E and Supplementary Figure S1D). The significance of our analysis of teratoma histology is that it demonstrates the multipotent nature of cEGLCs and thus the reprogramming of migratory PGCs, which are monopotent. In fact, the teratoma tissue was not a simple histological picture, such as seminomas, yolk sac tumors or even choriocarcinomas, which are observed during tumorigenesis of germ cell origin (Barrios et al., 2013), but a complex structure containing mature and diverse tissue, which confirms the multipotent and reprogrammed status of cEGLCs (Figure 2E).

To compare gene expression profiles of cEGLCs with naïve or primed mouse pluripotent stem cells, we performed RNA-sequencing. Unsupervised MDS plot analysis revealed that cEGLCs were closely clustered with ESCs and EGCs (naïve) but not with EpiLCs (formative), EpiSCs, or rsEpiSCs (primed) (Figure 2F_left panel). Unlike other stem cells, P19 embryonal carcinoma cells (ECCs) derived from teratocarcinoma can proliferate indefinitely and maintain an undifferentiated state without external signals. ECCs were also clearly distinguished from cEGLCs. Hierarchical clustering analysis, however, showed that cEGLCs are moderately different from ESCs and EGCs (Figure 2F_right panel). These results indicate that cEGLC lines display the characteristics of naïve pluripotency like those of ESCs and EGCs despite their distinct colony morphology and moderate differences in gene expression.

To determine specific features of cEGLCs, we examined expression of marker genes and performed functional analysis to characterize the difference between naïve and primed states, given that the colony morphology of cEGLCs was not dome-shaped which is a representative feature of the naïve state (Figure 2A and Supplementary Figure S1B). We found that both cEGLC-23 and -20 lines exhibited characteristics differences from naïve pluripotency, namely, lower intensity of ALP activity (Figure 3A and Supplementary Figure S1E), partial but clearly detected OTX2 protein, a definitive marker of epiblast (Figure 3B and Supplementary Figure S1F), and higher expression levels of epiblast markers (Figure 3C). To assess the differentiation potential of cEGLCs into the primitive endoderm, a representative feature of naïve pluripotency, we performed embryoid body analysis (Figure 3D). cEGLCs-23 cells showed a limited ability to form an outer layer of GATA4-positive primitive endoderm in embryoid bodies, which was in sharp contrast to naïve EGCs. Taken together, these results indicate that cEGLCs closely resemble other naive pluripotent stem cells (ESCs and EGCs) in terms of their broad gene expression profiles (Figure 1F), but they also possess partial features of the epiblast that are not normally seen in naïve pluripotency.

Since cEGLCs partially exhibited epiblast-like features, albeit in a naïve state, we investigated the intracellular signaling pathways involved in maintaining pluripotency. Withdrawal of LIF and inhibition of JAK/STAT3 signaling rapidly induce differentiation in mouse ESCs, reconfirming the requirement of LIF-JAK signaling dependence in the naïve pluripotency (Boeuf et al., 1997; Niwa et al., 1998). Similarly, inhibition of FGF2 and Activin A pathways in primed pluripotent EpiSCs induced cell death and neuroectodermal differentiation (Greber et al., 2010; Wu J. et al., 2015). To determine the signaling pathways for maintaining pluripotency in cEGLCs, we exposed these cells to specific inhibitors of JAK (JAKi), FGF2 (SU5402), or Activin A (SB431542). Four days after inhibition of JAK, but not FGF2 or Activin A, most EGC colonies exhibited loss of OCT4 protein expression and decrease of Oct4-GFP signals (Figure 4A). In contrast, OCT4 levels remained unchanged in cEGLCs (Figure 4A). Considering the possibility of switching of the dependency on signaling pathway during culture, we also checked the effect of combined inhibition of JAK and FGF pathways (Figure 4A). Even after treatment with dual inhibitors, however, cEGLCs showed stable expression of OCT4, indicating that cEGLCs maintain pluripotency independent of conventional signaling pathways such as JAK or FGF2.

To determine the signaling pathways required for pluripotency in cEGLCs, we performed Gene Set Enrichment Analysis (GSEA) of DEGs (Differential Expressed Genes) between the ESCs-EGCs group and cEGLCs group (Figure 4B). Visualization of results with “Cnetplot,” based on the correlation between DEGs and GO terms, indicated that PI3K-Akt signaling pathways, focal adhesion, and ECM-receptor interaction could be highly activated in cEGLCs.

The PI3K (phosphoinositide 3-kinase) signaling pathway, which is activated by growth factors and cytokines including LIF, is important for proliferation, survival, and maintenance of pluripotency in mouse ESCs (Watanabe et al., 2006). To test whether the PI3K pathway is essential for pluripotency in cEGLCs, or whether there is a difference in sensitivity to PI3K pathway inhibition between EGCs and cEGLCs, exposure to a specific inhibitor of PI3K-Akt signaling (LY294002) was performed. Upon treatment with LY294002 for 4 days, most cEGLC-23 colonies lost OCT4 protein and Oct4-GFP expression, whereas EGCs were barely affected (Figure 4C). We speculate that in cEGLCs, activation of the PI3K-Akt pathway as a mechanism for pluripotency may utilize unknown exogenous factors rather than well-known ones such as LIF and FGF2/Activin A (Figure 4A), or may utilize self-activated endogenous mechanisms independent of p38 MAPK inhibition (Supplementary Figure S3).

The increased expression of genes related to focal adhesion and ECM (extracellular matrix)-receptor interaction may correlates with changes in the cell’s adherence to the surrounding matrix (Figure 4B). This can influence the physical morphology of colonies (Figure 2A and Supplementary Figure S1B). Increased focal adhesion can lead to a more spread and flattened colony morphology due to stronger interactions with the ECM (see discussion).

A typical method for classifying naïve and primed pluripotency is chimera formation by blastocyst injection, since injection of primed pluripotent stem cells such as mouse EpiSCs into the blastocyst does not result in the formation of chimeras (Mascetti and Pedersen, 2016). As described above, cEGLCs exhibit naïve pluripotency (Figure 2), along with partial epiblast-like features, including expression of protein (Figure 3B) and certain genes (Figure 3C) and lacking the ability to differentiate into primitive endoderm during embryoid body formation (Figure 3D). Intriguingly, chimera formation assay showed that cEGLCs-23 cultured with 2iLIF formed chimeras, but those cultured with SB239063 (DK15^SB23) did not (Figure 5A), indicating that cEGLCs-23 may have a latent ability to form chimeras and that SB239063 likely blocks this ability.

To confirm the molecular features underlying the ability to form chimeras, we analyzed expression patterns of pluripotent markers and transcriptional profiles of cEGLCs cultured with 2iLIF (cEGLC-23/2iLIF). As shown in Figures 5B–D, cEGLCs cultured with 2iLIF exhibited a loss of epiblast-like features such as OTX2 protein expression (Figure 5B), intense ALP activity (Figure 5C), and transcriptional adaptation into naïve pluripotency representing increased naïve and decreased primed marker genes (Figure 5D). Consistent with these results, analysis of RNA-seq data also suggested that the gene expression profiles of cEGLCs-23 cultured with 2iLIF (cEGLC-23/2iLIF) are closer to naïve pluripotent states as observed in EGCs/ESCs (Figure 5E).

To determine whether p38 MAPK is the molecular signal that suppresses pluripotency and teratoma formation in germ cells, we examined isolated hindguts, in which migrating PGCs are closely associated with the surrounding somatic cells, as an ex vivo organ culture model for teratoma formation. Hindguts encapsulating Oct4-GFP-expressing PGCs were isolated from Oct4-GFP mouse embryos at E8.75, and plated on MMC-inactivated STO-feeder cells without dissociation in the presence or absence of SB239063 (Figure 6A). In the control media, some of the surviving Oct4-GFP-expressing PGCs were dispersed but not clustered in the outgrowth of the plated hindgut, (Figure 6B). In the media supplemented with SB239063, several clusters consisting of cells positive for Oct4-GFP expression were formed in the outgrowth of hindgut after 6–8 days. To determine the acquirement of indefinite proliferative potential, after 8 days in culture, hindgut outgrowths were scraped off by pipette tips, followed by cell dissociation by trypsinization and replating in organ culture media without SB239063. Even when SB239063 was removed, Oct4-GFP-positive colony-forming cells appeared and grew continuously. These results, therefore, suggest that PGCs, even within the surrounding somatic cells, are reprogrammed by suppression of p38 MAPK activity in organ culture, and acquire indefinite proliferative potential, which may be the putative origin of teratoma development. Our culture system, therefore, could be considered as an ex vivo organ culture model to probe teratoma formation.