SSCs spontaneously transform into ES-like state in the modified culture system

We established several mouse SSC lines from neonatal testis using IMDM condition [10] with minor modification (medium 1, see component in Table S1). SSCs isolated from testes of 5 days mice stably grew on MEF feeder layer, and formed grape-like colonies within 2–3 passages (Fig. 1a). They could be passaged in vitro for more than 40 passages and stably expressed SSCs markers such as PLZF, GFRA1 and CDH1 (Fig. 1b-d). Primary SSCs in this medium were passaged every 5–7 days, and after 20 generations could be passaged 3–4 days (in a ratio of 1:3 − 1:5). During in vitro culture, typical SSCs colonies were stably maintained (cellular boundary is relatively clear and colonies are relatively loose) (Figure S1a-b), verified by the capacity of germline reconstitution through testis transplantation (Figure S1c-e). Interestingly, a few of compact colonies (less than 10% of total SSCs colonies) were occasionally observed after long-term culture (usually need more than 25 passages), and this type of colonies became dominant after another 4–5 passages (Fig. 1e). Immunofluorescence staining demonstrated that they expressed NANOG and SOX2 (Fig. 1f-h), reflecting that they were in pluripotent state. We subsequently picked these colonies to MEF feeder and cultured them in ESCs medium. Gradually, colonies became compact with high density, the boundary of each cell could hardly be distinguished (Fig. 1j), which were very similar to typical ESCs colonies (Fig. 1i). The expression of pluripotent markers SSEA1 and NANOG indicated these ES-like cells as pluripotent cells (Fig. 1k-m). Majority of these cells highly expressed OCT4, and only a very small number of cells expressed germline marker MVH (Fig. 1n-p), indicating that most cells have lost the characteristics of germline cells. Subcutaneous injection of these cells into nude mice induced teratoma (Figure S2a-b). Both pluripotent cells generated after long-term culture and ES-like cells derived from transformed cells highly expressed pluripotent markers, Nanog, Sox2, Klf4 and Oct4 (Fig. 1q), and had high alkaline phosphatase activity (Fig. 1r-t), similar to ESCs. Based on these observations, we confirmed that SSCs have transformed into pluripotent state, and named them as germline stem cells derived pluripotent cells (GSPCs). However, expression of very low levels of germline markers, such as Gfrα1, α6-integrin, Plzf and Mvh (Fig. 1q), indicated that a few germ cells were still not transformed, yet.

Fig. 1

Newly isolated SSCs transformed into pluripotent state during long-term culture. a-d Newly isolated SSCs were purified and plated on MEF feeder (a), and were identified with IF staining using antibodies against PLZF (b), GFRA1 (c) and CDH1 (d). e Typical colonies of transformed cells from long-term cultured SSCs were exhibited. f-h Identification of transformed pluripotent cells using dual IF staining of OCT4 (f), SOX2 (g) and DAPI (h). i-j The morphology of ESCs (i) and ES-like cells derived from transformed pluripotent cells (j) on MEF feeder were exhibited. k-p The expression of pluripotent and germline markers (k. SSEA1, l. NANOG, m. merge; n. OCT4, o. MVH, p. merge) in ES-like cells were detected using IF staining. q The expression of germline and pluripotent markers were determined using RT-PCR (M. marker, 1. newly isolated SSCs, 2. GSPCs derived from long-term culture, 3. ES-like cells derived from GSPCs, 4. H2O). r-t The alkaline phosphatase activity was detected in ESCs (r), GSPCs (s) and ES-like cells (t). Scale bar = 20 μm

This phenomenon is consistent with reported observation [7], which is interesting, since no exogenous gene or chemical molecule is required to be introduced in. Germline stem cells are believed to retain the potential of pluripotency [2], and GSPCs occasionally appeared in culture for unknown reasons. Therefore, we proposed that some key factors in the culture medium induced the pluripotency of SSCs during culture in vitro.

Supplementation of EGF/LIF increased the efficiency of spontaneous reprogramming of SSCs

Since spontaneous transformation into pluripotent state was associated with medium components, and previous study revealed that p53 deficiency accelerates transformation [7], we therefore hypothesized that some factors may affect the transformation efficiency. Based on the published papers [7, 10, 12, 16] and our experience, we screened many candidate growth factors, and after several rounds of screening finally found that the addition of 10 ng /ml LIF and 20 ng /ml EGF to medium 1 (medium 2) was able to effectively cultivate the SSCs cultured on fresh MEF feeder layers to transform into pluripotent state. Importantly, primary SSCs are required to be stably maintained on fresh MEF feeder in medium 1 for at least 5 passages (about 35–40 days) to adapt to the in vitro culture condition, and then the medium was replaced with medium 2 containing LIF and EGF. About 3 passages later (around 15 days), around 30% colonies distinct to normal SSCs morphology could be observed (Fig. 2c), which were identical to those derived from long-term culture in medium 1 (Fig. 1e). However, they became dominant colonies after 2–3 passages, which were more efficient than those derived from long-term cultured cells in medium 1 (less than 10% colonies transformed after 25 passages, and took another 4–5 passages to become dominant). The adhesion to both neighbor cells and feeder layers increased, and cell boundary was indistinguishable. After further culture, the majority of cell clusters gradually converted into compact colonies (Fig. 2d), which resembled the epiblast cell colonies. Similar to the GSPCs derived from long-term culture in medium 1, these GSPCs highly expressed pluripotent markers including NANOG and SSEA1, and expressed a very low ratio of MVH and hardly expressed PLZF (Fig. 2e-l). Intensive expression of pluripotent markers, Nanog, Sox2, Klf4 and Oct4, further verified the pluripotent characteristics of GSPCs, whereas weak expression levels of SSC and germline markers (Gfrα1, α6-integrin, Plzf and Mvh) suggested that a few untransformed germ cells were probably remained (Fig. 2m). Both SSCs and GSPCs have normal diploid karyotype (Figure S4a), excluding the fates of meiosis or tumorisation. GSPCs derived in medium 2 also expressed alkaline phosphatase (Fig. 2n), and induced teratoma in nude mice after intraperitoneal injection (Fig. 2o). GSPCs derived in medium 2 were able to be stably maintained in vitro for more than 40 passages with a sharper growth curve, compared to GSPCs derived from long-term culture (Figure S2c), probably due to the faster proliferation rate (1:5 − 1:10 subculture every 48–72 h).

Notably, we found that not only the passages of MEF (less than three passages) were pivotal for SSCs transformation, but also the passages of SSCs remarkably affected the transformation efficiency. Newly isolated SSCs on fresh MEF feeder layer hardly yield GSPCs colony in medium 2 (0 of 12 attempts), whereas SSCs cultured on fresh MEF feeder layer for 5 passages in medium 1, could transform within 3–4 passages after replacement of medium2 with a successful rate of over 70% (10 of 14 attempts), and p8 (8 passages) SSCs cultured in medium 1 transformed even more efficiently within p5 SSCs (12 of 14 attempts). It was worth noting that all attempts of SSCs more than 10 passages in medium 1 transformed into GSPCs within 5 passages in medium 2 (Fig. 2p). These observations indicate that stable maintenance in medium 1 is a prerequisite for SSCs transformation in medium 2, and longer culture in medium 1 leads to higher transformation efficiency, which is in agreement with the conclusion that acquirement of the indefinitely proliferative capacity is important for the self-renewal of pluripotent stem cells [17]. Collectively, the transformation efficiency and transformation time of each condition in study were summarized, demonstrating a higher efficiency than reported transformation systems (Table 1).

Fig. 2

Addition of EGF and LIF effectively accelerated SSCs transformation. a The morphology of newly isolated SSCs maintained on MEF feeder with medium1 after 1 passage was exhibited. b, SSCs colonies on MEF feeder with medium 1 for 4 passages, and medium 1 was replaced with medium 2, were exhibited. c After another 3 passages in medium 2, some GSPCs colonies appeared (red arrow heads). d The representative GSPCs colonies after long-term culture were exhibited. e-l The expression of NANOG, SSEA1 and MVH in stable GSPCs was detected using IF (e. PLZF, f. NANOG, g. DAPI, h. merge; i. SSEA1, j. MVH (arrows indicate a few MVH+ cells), k. DAPI, l. merge). m The expression levels of Gfrα1, α6-integrin, Plzf, Mvh, Nanog, Sox2, Klf4, Oct4 and Gapdh in SSCs and GSPCs were determined using RT-PCR (M. marker, 1. SSCs, 2. GSPCs). n Alkaline phosphatase activity of stable GSPCs derived from SSCs cultured in medium 2 was detected. o GSPCs derived in medium 2 induced teratoma in nude mice as early as 8–10 days, while SSCs cultured in medium 1 could not induce teratoma. p The conversion ratio of SSCs cultured in medium 1 was statistically analyzed. q The strategy to make germline specific GFP mice and isolate GFP labeled SSCs. r Tracing the formation of GFP labeled GSPCs under medium 2 condition. The total cells from mTmGfl/+ mice indiscriminately expressed tomato (left), while SSCs from mTmGfl/fl;Ddx4-Cre+ mice specifically expressed GFP (middle), and when they transformed in medium 2, GFP signal was only observed in GSPCs (right). Scale bar = 20 μm

Table 1 Comparison of the transformation time and efficiency

To exclude the possibility that GSPCs were derived from MEF, fresh MEF were cultured in medium 2 for more than 6 passages, or mitomycin-C treated MEF were cultured with medium 2 for more than 20 days, until cell death, no transformed pluripotent stem cells was observed. Moreover, SSCs from mTmGfl/fl;Ddx4-Cre+ mice, which specifically express GFP in germ cells and express Tomato in somatic cells (Fig. 2q), were used to trace the fate of SSCs during culture. The results exhibited that all the colonies were GFP-labelled (Fig. 2r), excluding the possibility of somatic cells or MEFs reprogramming. Thus, we confirmed that addition of EGF/LIF to our modified IMDM medium remarkably increased the transformation efficiency of SSCs into pluripotent state.

Conversion of GSPCs to ES-like state under standard ESC culture condition

Interestingly, ES-like cells transformed by Shinohara’s system could be maintained in vitro only in standard ESCs medium [7]. Therefore, stable GSPCs (usually need another 6–7 passages in medium 2 from the appearance of GSPCs colonies) were transferred into standard ESC medium (medium 3) on MEF feeder to see what happen. In the new condition, a small number of cells underwent apoptosis within the first 2–3 passages, and simultaneously some compact colonies similar to typical ES colonies formed (Fig. 3a-b). ES-like colonies usually became dominant after another 3–4 passages, and could be stably maintained under this condition (Fig. 3c), with the morphology indistinguishable from ESCs colonies (Fig. 3d). These ES-like cells proliferated rapidly, with an average subculture time of 3 days, identical to those generated using Shinohara’s protocol (Fig. 3e). Moreover, their expression profiles of marker genes (Fig. 3f), karyotype (Figure S4a) and alkaline phosphatase activity (Fig. 3g) were identical to ES-like derived from long-term culture condition. ES-like cells also generated teratomas in nude mice (Fig. 3h), confirming their pluripotency. Intensive expression of SSEA1, NANOG, CDH1 and SOX2 further confirmed their pluripotency identity (Fig. 3i-p). A very low proportion of MVH+ cells were detected in colonies (Figure S3a-c), which probably resulted from a few of untransformed SSCs in the colonies.

To analyze the imprinting pattern of GSPCs and ES-like cells, two paternally imprinted regions (H19 and Meg3 IG regions) and two maternally imprinted regions (Igf2r and Peg10 regions) were examined by combined bisulfite restriction analysis (COBRA). GSPCs possessed a similar imprinting pattern with SSCs, which exhibited a typical paternally methylation status in differentially methylated regions (DMRs), whereas ES-like cells had both paternal and maternal imprinting patterns (Meg3 IG and Igf2r regions) (Fig. 3q). These results implied that GSPCs maintained the paternal imprinting characteristics of SSCs, which were remarkably changed in ES-like cells.

It was worth noting that ES-like colony was not observed after directly replacing SSCs medium with ESCs medium in primary or long-term cultured SSCs, since SSCs failed to survive in ESC medium. This indicates that transformation into GSPCs state probably is an essential step for ES-like formation. Based on these observations, we summarized the schematic procedure of SSCs transformation (Fig. 3r).

Fig. 3

ES medium induced GSPCs to transform into ES-like state. a The morphology of GSPCs cultured in medium 3 for 1 passage was exhibited. b ES-like colonies appeared in GSPCs cultured in medium 3 for 3–4 passages. c ES-like colonies became dominant after a few passages. d The typical ES colonies were exhibited. e The growth curve of ES-like cells was exhibited. f The expression levels of Gfrα1, α6-integrin, Plzf, Mvh, Nanog, Sox2, Klf4, Oct4 and Gapdh were determined using RT-PCR (M. marker, (1) SSCs, (2) GSPCs, (3) ES-like cells, (4) ESCs). g Alkaline phosphatase activity was determined in ES-like cells derived from GSPCs. h ES-like cells derived from GSPCs induced teratoma in nude mice, while SSCs cultured in medium 1 could not induce teratoma. i-p IF staining assays detected the expression of pluripotent markers in ES-like cells derived from GSPCs (i, SSEA1. j, NANOG. k, DAPI. l, merge; m, CDH1. n, SOX2. o, DAPI. p, merge). q COBRA demonstrated the parental imprinting characteristics of SSCs, GSPCs and ES-like cells. r The schematic procedure of SSCs transformation into GSPCs and ES-like was summarized. Scale bar = 20 μm

Comparison of the pluripotency of GSPCs and ES-like cells

Although GSPCs and ES-like cells exhibited characteristics of pluripotent stem cells, it’s not clear whether they were different in hierarchy of pluripotency, or just morphologically distinct induced by culture medium. To determine their pluripotency levels, we first compared the growth condition of GSPCs and ES-like cells. Both of these cells could be stably maintained in vitro for more than 30 passages, and the average subculture time was 2–3 days for GSPCs and ES-like cells. Similar to ESCs, GSPCs and ES-like also highly expressed pluripotent markers, including OCT4, NANOG and SOX2. To further determine the pluripotency of these cells, ESCs, ES-like, GSPCs and SSCs were subcutaneously injected into nude mice to test the efficiency of teratoma formation. Notably, the capacity to generate teratomas of GSPCs and ES-like cells was remarkably higher than that of ESCs. Teratomas could be observed 7–10 days post GSPCs or ES-like cells injection, while ESCs need at least 15–20 days to generate visible teratomas (Table 2). This observation indicated a higher homogeneity of GSPCs and ES-like cells than ESCs [18]. Finally, we tested the potential of blastocyst development. ESCs, GSPCs and ES-like were labeled with GFP and sorted for establishment of GFP + clones in vitro, and were microinjected into 8-cell stage blastocysts (2.5 d.p.c). The results demonstrated that GSPCs, ES-like and ESCs cells could participate in embryo development (Fig. 4Sb-d), further confirming their pluripotency.

Table 2 Comparison of teratoma formation efficiencyComparison of gene expression profiles of SSCs, transforming SSCs, GSPCs and ES-like cells through transcriptomic analysis

To reveal the connection between EGF/LIF signals and SSCs spontaneous reprogramming, transcriptome analysis was conducted. Differential expression genes (DEGs) of SSCs (5 passages in medium 1), intermediate state cells (5 passages in medium 1, and 2 passages in medium 2, referred to as “In”), GSPCs (5 passages in medium 1, and 10 passages in medium 2), and ES-like cells (GSPCs cultured for 10 passages in medium 3) were screened to reveal this relationship (Fig. 4a). The results of gene expression analysis indicated that SSCs and In state cells have similar transcriptomic profiles, which are distinctly different from those of GSPCs and ES-like cells (Fig. 4b-c). This analysis revealed that In state cells might be in the initiating stage of SSCs transformation, and transformed into a completely different cell type (GSPCs) after eight passages in LIF/EGF medium. Furthermore, the gene expression profile partially changed when ESC medium was replaced.

Using these transcriptome data, we further compared the gene expression profiles of GSPCs and ES-like cells with ESCs, the well characterized pluripotent cells. We obtained RNA-seq datasets of wild-type mouse embryonic stem cells from the Gene Expression Omnibus (GEO) database [19] and performed a comparative analysis with our study’s dataset. After eliminating batch effects among the data, principal component analysis (PCA) was conducted on individual samples (Figure S5a). The results revealed distinct clustering of the four groups, with the SSCs group significantly separated from the other three, suggesting a unique expression pattern in the SSCs group. The samples of the ESCs group were positioned between the GSPCs and ES-like cells groups, indicating a close similarity of their expression pattern. Furthermore, the gene expression heatmap among samples (Figure S5b) demonstrated a high overlap in expression patterns between the ESCs and GSPCs groups, with hierarchical clustering showing a closer similarity between ESCs and GSPCs. The sample correlation heatmap (Figure S5c) indicated a significant positive correlation between the ESCs group and ES-like cells and GSPCs groups, with correlation coefficients reaching average values of 0.960 and 0.968, respectively, but a lower correlation with the SSCs group.

Furthermore, we compared the expression levels of some representative genes associated with pluripotency in SSCs, GSPCs, ES-like cells and ESCs (Figure S5d), and observed that typical genes for pluripotency, e.g., Nanog, Klf4, Sox2, Esrrb and Lifr, were remarkably activated in all of these cells, confirming their pluripotent state. However, we also noticed that Bmp4, Mapk15 and Pou5f1 were mainly highly expressed in GSPCs, while many genes in Wnt signaling pathways (Wnt3a, Axin2, Apc, Gsk3b, Tcf7 etc.,) and in MAPK signaling pathway (Mapk3k3, Mapk6, Mapk8, Mapk9 etc.,) were strongly expressed in ES-like cells. Most of these genes were also expressed in ESCs, but the expression levels were lower than in ES-like cells. The gene expression heatmap revealed that, compared to GSPCs, ES-like cells are closer to ESCs in the expression pattern of pluripotency associated genes.

Collectively, these results indicate that gene expression characteristics of GSPCs and ES-like cells are similar to pluripotent stem cells.

Fig. 4

Analysis of transcriptomic characteristics of SSCs, intermediate state cells, GSPCs and ES-like cells. a The illustration of sample collection for RNA-sequencing. b-c Violin Plot (b) and heatmap (c) of DEGs identified from the SSCs, In, GSPCs and ES-like cells. d-e Venn diagrams summarized the up-regulated genes (d) and down-regulated genes (e) in In vs. SSCs and GSPCs vs. SSCs. f The relative expression levels of Btc, Apln, Rac1 and Bcl2 genes in SSCs, In state cells and GSPCs were determined using RT-PCR, and were statistically analyzed, *p < 0.05; **p < 0.01. g KEGG of differential genes in SSCs, In state cells, GSPCs, ES-like cells and shared by two datasets. h Hallmarks of differential genes in SSCs, In state cells, GSPCs and ES-like cells. i The relative expression levels of pluripotency associated genes in SSCs, In state cells, GSPCs and ES-like cells were exhibited. j Representative DEGs in SSCs, In, GSPCs and ES-like cells were selected. Left, fold change. Blue, downregulated genes; red, upregulated genes. Right, false discovery rate (FDR)-adjusted p values determined using DESeq2 (–log10-transformed). * represents a remarkable difference. k The relative expression levels of Plzf, Etv5, Dnmt1 and Ccnd1 genes in SSCs, In state cells, GSPCs and ES-like cells were determined using RT-PCR, and were statistically analyzed, *p < 0.05; **p < 0.01

Analysis of the transcriptomic changes during SSCs transformation

To reveal the molecular events during SSCs transformation, we compared the DEGs of In state cells with SSCs and GSPCs. The Venn diagram further highlighted the similarities and differences of gene expression profiles of SSCs, In state cells, and GSPCs. Notably, the number of differentially expressed genes in In state cells was remarkably less than that in GSPCs (443 up-regulated in In state vs. 6403 up-regulated in GSPCs, 1:14.45; 806 down-regulated in In vs. 6850 down-regulated in GSPCs, 1:8.49) (Fig. 4d and e). Among them, genes that were uniquely differentially expressed in In state cells (266 up-regulated and 226 down-regulated) probably represent key genes for the initiation of transformation (refer to Table S4 and S5). The enhanced expression of EGF and EGFR-associated genes (Areg, Btc, Eps8, and Ereg) [20] in In state cells suggested that EGF might function earlier than LIF signal in SSCs transformation. Additionally, several GTPase or G-protein receptor-associated genes (Apln, Appl2, Arhgap6, Arhgap18, Arhgap22, Dock8, Gpr137b, Stard13, and Vav3) [21,22,23,24,25] were up-regulated in In state cells, while the expression levels of Rgs1 (the activator of GTPase activity) [26] and Arhgap30 (stimulate GTP-hydrolysis on RAC1 to inactivate RAC1 activity) [27] were decreased. These findings indicate the potential role of the G-protein family in the onset of SSCs transformation, with RAC1 being a probable key gene involved in transformation. Our transcriptome analysis revealed that Smad3 and Ltbp1, the regulators for TGF-β activation [28], were up-regulated. These findings are consistent with our previous study, which demonstrated that SMAD3 plays a crucial role in the transformation of SSCs. Furthermore, we observed an increase in the expression of genes associated with cell survival, such as Nabp1 [29], Bcl2 [30], Dclre1b [31], Rrm2b [32], and Styk1 [30], and a decrease in the expression of apoptotic or death-associated genes, including Card9 [33], Dapk1 [34], Klf15 [35], Tnfrsf21 [36], and Usp13 [37]. Moreover, we selectively detected the expression levels of these genes (Btc, Apln, Rac1 and Bcl2) using RT-PCR [Fig. 4f], and confirmed the consistency with expression change of transcriptomic analysis. From these results we concluded that the increased proliferative capacity and anti-apoptotic capacity are essential for SSCs transformation.

We also observed an up-regulation of genes promoting proliferation, such as Arid5a [38], Ccnd1 [39], Foxm1 [40], and Prkca [41], and a down-regulation of cyclin inhibitor Cdkn1c [42], suggesting that increased proliferation is a key feature of SSCs transformation. Additionally, we found that the expression levels of many members in the MAPK and Wnt signaling pathways were significantly altered. For instance, Lgr6, Mapk6, Map3k15, and Mapk13 were up-regulated, while Dact1, Gpc3, Lefty1, Lyn, Ptk2b, Six2, and Wnt5b were down-regulated. These results suggest that the MAPK and Wnt signaling pathways play a role in the initiation of SSCs transformation. Finally, we identified 177 up-regulated genes in In state cells compared to SSCs and GSPCs compared to SSCs, which may be pivotal for the late stage of transformation or GSPCs maintenance. These genes include Brca2, Cbx3, Ccdc18, Ccnb1, Ccne2, Cdc7, Klf4, and Sirt1, which are well-known factors associated with pluripotency.

To gain further insights into the molecular mechanisms underlying SSCs transformation, we analyzed the FPKM values of genes in SSCs, In state cells, GSPCs and ES-like cells (Table S6), and identified enriched gene functions through KEGG analysis (Fig. 4g). Gene categories that were highly expressed in SSCs, such as cytokine-cytokine receptor interaction and chemokine signaling pathway, were down-regulated in In state cells and further decreased in GSPCs, suggesting a change in essential growth factors during the transformation process. We then focused on the potential signaling pathways that may play a pivotal role in the transformation process, and found that the average expression levels of genes belonging to mTOR, ERBB (also known as EGFR), MAPK, GnRH, and insulin signaling pathways were up-regulated in In state cells, indicating their involvement in the initiation of SSCs transformation. In GSPCs, transcriptional activation of signaling pathways related to carbohydrate metabolism, amino acid metabolism, DNA repair, cell cycle, and ribosome, indicated that metabolic changes might be essential in this phase. Furthermore, we observed a significant increase in the average expression levels of genes in Hedgehog, Wnt, and TGF-β signaling pathways in ES-like cells. The compact structure of ES-like cell colonies might be attributed to the up-regulation of genes associated with cell adhesion molecules CAMs and adherens junction.

The molecular characteristics of cells at different stages of transformation were analyzed, and the differentially expressed genes (DEGs) were identified (Fig. 4h). In SSCs, genes associated with Epithelial-mesenchymal transition (EMT), IL6-JAK-STAT signaling, hypoxia, apoptosis, p53 signaling pathway, Notch signaling pathway, and Glycolysis were detected. The expression levels of these genes gradually decreased from In state cells to GSPCs, and to ES-like cells. The PI3K-AKT-mTOR signaling pathway was activated in the intermediate state, indicating its involvement in the initiation of SSCs transformation. In GSPCs, the dominant categories were associated with pluripotency, cell cycle (G2M checkpoint), DNA repair, and spermatogenesis. In ES-like cells, the Wnt and TGF-β signaling pathways were activated, and the up-regulation of genes associated with cell adhesion molecules CAMs and adherens junction likely caused the compact structure of ES-like cells colonies. The expression patterns of the DEGs demonstrated the pivotal signaling pathways of each stage in the transformation processes. Interestingly, the trends of gene expression changes were consistent with observations from p53 knockout model, suggesting that the different transformation models of SSCs (p53 knockout or LIF/EGF stimulation) probably shared similar signaling pathways.

To investigate the key signaling pathways involved in the initiation of SSCs transformation, we compared the expression patterns of pluripotency-associated signaling pathways, including Wnt, Ras, TGF-β, and JAK-STAT signaling pathways (Figure S5). We analyzed the expression changes of genes associated with pluripotency from SSCs to In state cells and found that genes in the Wnt and TGF-β signaling pathways were activated, accompanied by transcriptional activation of pluripotent genes such as Sox2, Nodal, and Esrrb, while the expression levels of genes in the JAK-STAT signaling pathway were down-regulated. Additionally, some genes in the RAS signaling pathway were up-regulated, while others were down-regulated. These observations suggest that Wnt, TGF-β, and RAS signaling pathways may be involved in the initiation of SSCs transformation. Although JAK-STAT is essential for SSCs proliferation [43], it becomes dispensable for the initiation of SSCs transformation, GSPCs, and ES-like maintenance, suggesting that other signaling pathways, such as Wnt and TGF-β signaling pathways, may promote proliferation activity of GSPCs and ES-like cells.

Furthermore, the expression levels of genes associated with pluripotency were analyzed to seek the clues of transformation. The expression difference among SSCs, transforming cells (In state) and transformed cells (GSPCs, ESL) was significant (Fig. 4 g-i). Although SSCs and In state cells exhibited similar expression profiles of these pluripotent genes, we noticed that several genes were specifically activated in In state, including Nras, Klf4, Kras, GSK3β, Myc, Akt3, Smad1, Smad3 and Smad4 (Fig. 4i), implying they were pivotal genes for the initiation of SSCs transformation. Notably, Myc and Klf4 were highly expressed in In state, and remarkably down-regulated in GSPCs, indicating that they are essential for SSCs transformation, but dispensable for GSPCs maintenance. Since Myc [44] and Klf4 [45] have been identified as the targets of RAS, and RAS is the downstream molecule of EGF signaling pathway [46], there is a possibility that Myc and Klf4 were activated by Nras and Kras through EGF-RAS-MEK signaling pathway in the initiation of SSCs transformation. Combined with the KEGG results showing that EGFR and MAPK signaling pathways were predominantly activated in In state (Fig. 4g), we proposed that EGF-RAS-MAPK and PI3K-AKT signaling pathways were decisive for the initiation of SSCs transformation. Although the expression levels of Smad1, Smad3 and Smad4 were remarkably enhanced in In state, our previous study revealed that activation of SMAD3 promoted SSCs transformation in the late stage, rather than driving the initiation of SSCs transformation, and revealed Nanog as a direct target of SMAD3 in SSCs [14]. Therefore, we hypothesized that enhanced expression levels of Smad3 and Smad4 were a prerequisite for the transition from In state to GSPCs. Notably, Myc is also a target of the FGF signaling pathway [47], and both SSCs medium (Medium 1) and GSPCs medium (Medium 2) contain 10 ng/ml bFGF. However, Myc was not activated