Wnt5a-β-catenin-TCF7L2 Positively Regulates the Formation of PGC
The molecular mechanism regulating PGC formation was evaluated using KEGG pathway analysis of development-related DEGs (Differentially expressed genes) in ESCs and PGC (SRR3720923 and SRR3720924) (Supplementary Fig. 1A,B). Wnt signaling was also activated during PGC formation (Fig. 1A; Supplementary Fig. 1C), in addition to enrichment of TGF signaling genes21. Wnt signaling–related genes were also significantly activated in an in vitro PGC induction model24 (Supplementary Fig. 1D). These results indicate that Wnt signaling is involved in PGC formation.
We determined that Wnt5a, β-catenin, and TCF7L2 are the key Wnt signaling molecules in the formation of chicken germ stem cells (Supplementary Fig. 1C,D)25. To further demonstrate the involvement of Wnt5a/β-catenin/TCF7L2 signaling, we conducted Wnt5a overexpression/ interference during the process of inducing ESCs to form PGC with BMP4. The expression of signaling molecules in 4d-induced cells was assessed by qRT-PCR, which indicated significant down-regulation of β-Catenin and TCF7L2 expression (P < 0.01) following Wnt5A interference, whereas overexpression of Wnt5A significantly up-regulated β-Catenin and TCF7L2 expression (P < 0.01) (Fig. 1B). In vitro experiments produced similar results (Fig. 1C). Co-immunoprecipitation (co-IP) of lysates of DF1 cells co-transfected with pcDNA3.1-Myc-TCF7L2 and pcDNA3.1-β-Catenin revealed that TCF7L2 and β-catenin interact (Fig. 1D). These data indicate that Wnt5A mediates Wnt5A/β-catenin/TCF7L2 signaling during the differentiation of ESCs into PGC. We previously confirmed that Wnt5A/β-catenin/TCF7L2 signaling promotes PGC formation25 (Fig. 1E). Collectively, our data indicate that Wnt5a-β-catenin-TCF7L2 positively regulates PGC formation.
Lin28A is a Specific Target of Wnt5a Signaling in PGC
To identify the specific target genes regulated by Wnt during PGC formation, we examined enrichment of target genes of TCF7L2 in humans (15727), rat (15652), and mouse (15374) in the GTRD (no bird database available) (Fig. 2A). Lin28A was the only common gene in GO and Venny analyses of 7473 target genes identified in these species (Fig. 2B; Supplementary Fig. 2), suggesting that Lin28A, a highly conserved gene targeted by Wnt, is involved in the generation of reproductive stem cells. These results are consistent with other reports26,27.
To confirm that Wnt-associated regulation of Lin28A also plays a role in the formation of chicken PGC, we compared the structure of Lin28A across species and found it is highly conserved (Supplementary Fig. 3). Chicken and mammalian Lin28A were exactly the same (Supplementary Fig. 3). Most importantly, the binding site for the transcription factor TCF7L2 was also present in the chicken Lin28A promoter (Supplementary Fig. 4A), which further suggests that Lin28A is a target of Wnt during the formation of chicken PGC.
To confirm that Wnt signaling regulates Lin28A, we examined the expression of Lin28A after activation/inhibition of Wnt signaling during PGC formation in vivo. Overexpression of Wnt5A significantly increased Lin28A expression (Fig. 2C), whereas inhibition of Wnt5A expression significantly inhibited Lin28A expression (Fig. 2D). Lin28A expression also decreased/increased significantly following inhibition/overexpression of β-catenin (Fig. 2C,D). Similar results were observed in induction experiments in vitro (Fig. 2C,D). Collectively, these results indicate that Lin28A responds to Wnt signaling. To examine this response further, we identified the core promoter of Lin28A (− 584 ~ + 100 bp) using the dual luciferase detection system (Fig. 2E; Supplementary Fig. 4A–C). Activation of Wnt signaling (overexpression of β-catenin) significantly increased the Lin28A promoter activity (P < 0.01) (Fig. 2F). However, mutation of the TCF7L2 binding site significantly reduced the activity of the Lin28A promoter (P < 0.01) (Fig. 2E,F; Supplementary Fig. 4D); Activation of Wnt signaling could not rescue Lin28A promoter activity following introduction of point mutations (Fig. 2E − G), indicating that Lin28A responds to Wnt signaling via the TCF7L2 binding site in the promoter.
Previous analyses indicated that there are two TCF7L2 binding sites in the Lin28A promoter. Therefore, we examined the binding of TCF7L2 to the Lin28A promoter using ChIP-qPCR and found enrichment of β-catenin/TCF7L2 complexes in the Lin28A promoter (Fig. 2H). Activation of Wnt signaling significantly increased binding of the β-catenin/TCF7L2 complex to Lin28A (P < 0.01), whereas inhibition of Wnt signaling significantly reduced this binding (P < 0.01) (Fig. 2H). These results indicate that Lin28A is a downstream target of Wnt signaling.
Lin28A Positively Regulates PGC formation in vitro and in vivo
Next, we investigated the function of Lin28A in PGC formation (Fig. 3A). Lin28A was inhibited/overexpressed during BMP4-induced differentiation of ESCs into PGC in vitro (Fig. 3A; Supplementary Fig. 5A). Morphologic observations on day 2 after BMP4 induction indicated that the cells had begun to grow larger. A few embryoid bodies (EBs) appeared on day 4, and the number of EBs increased on day 6; however, no EBs appeared between days 2 and 6 after Lin28A inhibition. In contrast, small EBs began to appear on day 2 after Lin28A overexpression, and on day 4 these EBs became larger and began to break. The number of EBs increased on day 6, the cell edges began to rupture, and a few cells were released from the EBs (Fig. 3B; Supplementary Fig. 5B). Lin28A overexpression significantly decreased expression of the pluripotency marker gene Nanog and increased Cvh, C-kit, and Blimp1 expression. Flow cytometry analyses demonstrated that Lin28A overexpression promoted PGC formation in the BMP4 model (Fig. 3C; Supplementary Fig. 5D,E). Similar results were observed in in vivo experiments (Fig. 3D,E; Supplementary Fig. 5E). Periodic acid–Schiff staining was used to monitor changes in the number of PGC formed in the genital ridge after Lin28A overexpression/interference. Compared with the number of PGC in the genital ridge during the normal in vivo hatching process (38 ± 1.53), the number of PGC in the genital ridge significantly increased following Lin28A overexpression (46 ± 2.10; P < 0.01) and significantly decreased (20 ± 1.64; P < 0.01) following Lin28A interference (Fig. 3D; Supplementary Fig. 5C). Collectively, these results indicate that Wnt/β-catenin signaling promotes PGC formation by activating Lin28A expression.
Lin28A is Regulated by H3K4me2
In a previous study examining H3K4me2 regulation of SSC formation, we performed RNA-seq analysis of LSD1-treated SSCs (LSD1: H3K4me2 demethylation modifying enzyme) and identified Lin28A as a DEG (Supplementary Fig. 6A). qRT-PCR analyses indicated significant up-regulation of Lin28A expression in SSCs following LSD1-mediated interference (Supplementary Fig. 6B). ChIP-qPCR analyses indicated that LSD1 regulates H3K4me2 enrichment in the Lin28A promoter (Supplementary Fig. 6C). To examine the effect of H3K4me2 on Lin28A regulation during PGC formation in the present study, we interfered with LSD1 and MLL2 expression in the in vitro BMP4 induction model. Lin28A expression was significantly higher than that induced by BMP4 after interfering with LSD1 expression, and the opposite trend was observed after interfering with MLL2 (Fig. 4A), indicated that H3K4me2 positively regulates Lin28A transcription in vitro. Lin28A expression in PGC on day 4.5 PGC was examined using qRT-PCR. Compared with the normal hatching process, H3K4me2 demethylation (interfering with MLL2) significantly decreased Lin28A expression, whereas H3K4me2 methylation (interfering with LSD1) significantly increased Lin28A expression (Fig. 4B). These results suggest that Lin28A is regulated by H3K4me2. To confirm that Lin28A is a target of H3K4me2, we examined the level of H3K4me2 enrichment in the Lin28A promoter in PGC using ChIP-qPCR. Compared with the control, H3K4me2 in the Lin28A promoter was significantly down-regulated following MLL2 interference and significantly up-regulated following LSD1 interference (Fig. 4C). Further results confirmed that changes in H3K4me2 regulate the Lin28A promoter (Fig. 5A). Collectively, these results indicate that in addition to Wnt signaling, H3K4me2 also regulates Lin28A expression during PGC formation.
Competition Between β-Catenin and LSD1 in Combination with TCF7L2 Regulates Lin28A Expression during PGC Formation
To further elucidate the molecular mechanism regulating Lin28A expression, we investigated interactions between Wnt and H3K4me2 using the dual luciferase system. Interference with Mll2 expression suppressed the response of Lin28A to Wnt signaling, whereas interference with LSD1 significantly enhanced the response (Fig. 5A). The position of H3K4me2 enrichment in the Lin28A promoter is near the TCF7L2 binding site. It is reasonable to speculate that β-catenin/TCF7L2 complexes affect the level of H3K4me2 enrichment to regulate Lin28A expression by altering the binding of LSD1 or MLL2 to the Lin28A promoter. Considering that the complex involving Mll2 is relatively fixed28,29, we used Co-IP to assess interactions between β-catenin, TCF7L2, and LSD1. Co-IP performed after co-transfection of DF1 cells with LSD1-Flag and β-catenin vectors indicated no interaction between LSD1 and β-catenin (Fig. 5B). However, in cells co-transfected with LSD1-Flag and TCF7L2-Myc, interaction between Flag and TCF7L2 was observed (Fig. 5C). Considering the correlation between TCF7L2 and β-catenin30, we hypothesize that in ESCs, TCF7L2 binding in the Lin28A promoter recruits LSD1, which reduces the level of H3K4me2 enrichment, inhibiting Lin28A transcription; during PGC formation, β-catenin enters the nucleus ectopically and competes with LSD1 for binding to TCF7L2, which increases the level of H3K4me2 enrichment and promotes Lin28A transcription. To test this hypothesis, DF1 cells were co-transfected with LSD1-Flag, TCF7L2-Myc, and β-catenin vectors. Co-IP indicated that LSD1 did not bind to TCF7L2, whereas β-catenin did bind to TCF7L2 (Fig. 5D). Collectively, these results indicate that β-catenin competes with LSD1 for binding to TCF7L2, which de-methylates H3K4me2 in the Lin28A promoter via LSD1 and activates Lin28A expression during PGC formation.
Lin28A Activates Blimp1 to Regulate PGC Formation by Inhibiting gga-let-7a-2-3p Maturation
Another study demonstrated that as an RNA-binding protein, Lin28A regulates the expression of related genes by inhibiting micRNA-let7 maturation26. However, the micRNA-let7 that interacts with Lin28A during chicken PGC formation remained to be identified. To determine the key micRNAlet7s targeted by Lin28A, 17 gga-let7s in the chicken micRNALet7 family were screened using miRDB (Fig. 6A,B). Expression of these mature microRNAs in chicken ESCs and PGC was evaluated by qRT-PCR after Lin28A overexpression/interference (Fig. 6A,B), which indicated that gga-let-7a-2-3p was significantly regulated by Lin28A in ESCs and PGC (Fig. 6A,B). gga-let-7a-2-3p was significantly up-regulated following Lin28A overexpression and significantly down-regulated following Lin28A interference (Fig. 6C,D), indicating that gga-let-7a-2-3p expression is regulated by Lin28A during chicken PGC formation.
Screening the miRDB identified 1143 genes targeted by gga-let-7a-2-3p. In particular, Blimp1 (PRDM1), which plays an important regulatory role in PGC formation, attracted our attention31,32. To determine whether gga-let-7a-2-3p targets Blimp1, we synthesized a gga-let-7a-2-3p mimic and inhibitor and transfected them into DF1 cells and PGC. qRT-PCR analysis indicated Blimp1 expression was significantly down-regulated in DF1 cells transfected with the mimic (Fig. 6E) and significantly up-regulated in cells transfected with the inhibitor, indicating that gga-let-7a- 2-3p negatively regulates Blimp1 (P < 0.01 for both) (Fig. 6E, left). As Blimp1 is a PGC marker, we performed the same experiment with PGC and observed similar results (Fig. 6E, right). To further confirm that gga-let-7a-2-3p targets Blimp1, we predicted the gga-let-7a-2-3p binding site in the Blimp1 3'UTR (UUGUACA). Wild-type and mutant (complete deletion of binding site) luciferase reporter vectors of the Blimp1 3'UTR were constructed separately. DF1 cells were then co-transfected with vectors for the gga-let-7a-2-3p mimic and inhibitor with Blimp1-3'UTR-WT and Blimp1-3'UTR-Mut. The gga-let-7a-2-3p inhibitor significantly increased Blimp1-3'UTR-WT luciferase activity in the double luciferase reporter assay (P < 0.01) but had no significant effect on Blimp1-3'UTR-Mut (P > 0.05) (Fig. 6F, left). The gga-let-7a-2-3p mimic significantly reduced Blimp1-3'UTR-WT luciferase activity (P < 0.01) but had no significant effect on Blimp1-3'UTR-Mut (P > 0.05) (Fig. 6F, right). These results indicate that Blimp1 is a direct target of gga-let-7a and that gga-let-7a binds to the 3'UTR of Blimp1 to inhibit its expression.
Blimp1 Interacts with LSD1 to Regulate the Expression of Related Genes in Wnt Signaling and Participates in PGC formation
As Blimp1 is known to affect the level of H3K4me233, we examined whether Blimp1 regulates PGC formation by altering the H3K4me2 level of key genes. Therefore, the correlation between H3K4me2 and Wnt signaling was examined during PGC formation. ChIP-qPCR analyses revealed two, four, and two H3K4me2 enrichment sites in the Wnt5A, β-Catenin, and TCF7L2 promoters, respectively. PGC exhibited significantly higher binding of H3K4me2 than ESCs (P < 0.01) (Fig. 7A). Enrichment of H3K4me2 levels in these sites in PGC was regulated by LSD1 and MLL2 (Supplementary Fig. 7), indicating that H3K4me2 regulates key Wnt signaling molecules. We then investigated in detail whether Blimp1 regulates H3K4me2 in the promoters of Wnt5A, β-Catenin, and TCF7L2. Notably, there is a Blimp1 binding site near the Wnt5A promoter H3K4me2 enrichment site (Fig. 7B). To confirm that Blimp1 binds to the Wnt5A promoter, a double luciferase reporter vector for the Wnt5A promoter was constructed and co-transfected into DF1 cells along with Blimp1 overexpression/interference vectors. The double luciferase reporter assay showed that Blimp1 overexpression significantly enhanced Wnt5A promoter activity, whereas interference with Blimp1 expression decreased promoter activity (Fig. 7C). However, Blimp1 overexpression/ interference had no effect on promoter activity after mutation of the Blimp1 binding site (Fig. 7D), indicating that Blimp1 binds to the Wnt5A promoter. Expression of Wnt5A was significantly up-regulated after Blimp1 over-expression in DF1 cells (Fig. 7E), as was the level of H3K4me2 in the Wnt5A promoter (Fig. 7F). Interestingly, the level of LSD1 binding in the Wnt5A promoter was significantly down-regulated (Fig. 7G). These results indicate that Blimp1 and LSD1 interact to regulate the expression of genes related to Wnt5A signaling.
Morphologic observation after interference with LSD1 and MLL2 expression in the in vitro BMP4 induction model revealed that LSD1 interference via shLSD1 promoted PGC formation, whereas interference with MLL2 expression inhibited PGC formation (Supplementary Fig. 8A). Expression of genes that activate Wnt signaling, such as Wnt5A, β-Catenin, FZD4, and TCF7L2, increased significantly after interference using shLSD1 (P < 0.01), whereas expression of genes that suppress Wnt signaling, such as AXIN1 and APC, decreased significantly (P < 0.01) (Supplementary Fig. 8B). Completely opposite results were obtained after interference with MLL2 expression (Supplementary Fig. 8B) and in vivo (Supplementary Fig. 8C). Collectively, these data indicate that H3K4me2 regulates PGC formation by activating Wnt5A/β-catenin/TCF7L2 signaling.
As our collective results indicated that Wnt-Lin28-Blimp-Wnt functions as a positive feedback loop during PGC formation, we sought to identify the factors that activate this feedback pathway. Previously, we confirmed that BMP4 plays an important role in PGC formation. Incubation of ESCs or PGC for 6 h in medium to which BMP4 was added led to significantly increased expression of signaling molecules such as Wnt5A, β-catenin, and TCF7L2 (P < 0.01) (Fig. 8A), whereas the AXIN1 and APC genes were significantly down-regulated (P < 0.01) (Fig. 8A). These preliminarily results indicate that BMP4 signaling activates Wnt5A signaling and that BMP4/Smads is upstream of WNT signaling. No significant change in BMP4 expression was observed 6 h after Wnt5A overexpression/interference in ESCs or PGC (Fig. 8B), indicating that Wnt signaling is downstream of BMP4 signaling. Considering the function of both BMP4 and Wnt in PGC formation suggests that BMP4 activates downstream WNT5A/β-catenin/TCF7L2 signaling to regulate PGC formation. Therefore, we preliminarily conclude that BMP4 signaling activates the Wnt-Lin28-Blimp-Wnt feedback system. To provide additional evidence, we changed the culture medium at 6 hours (after Wnt signaling activation) during induction with BMP4 (Fig. 8C,D). Flow cytometry analysis revealed that the absence of BMP4 had no effect on formation of normal PGC (Fig. 8E). Therefore, we conclude that BMP4 signaling mediates the normal development of PGC by activating Wnt signaling.