Ezh2 expression in the teratoma-forming cells and germ cells in Dnd1^ter/ter testes and in wild-type/Dnd1^ter/+ (WT) testes at E12.5 to E18.5. (A-E) The Oct4-ΔPE-GFP-positive germ cells (arrowheads in C,D,E) in WT testes showed higher Ezh2 signals than the 4C9-positive teratoma-forming cells at E17.5 (arrows in D) and E18.5 (arrows in E) as well as the GFP-positive germ cells at E16.5 (arrows in C) and E17.5 (D) in Dnd1^ter/ter testes. The GFP-positive germ cells in WT testes (arrowheads) showed slightly less Ezh2 signals than those in Dnd1^ter/ter testes (arrows) at E12.5 (A) and E14.5 (B). Results of the quantitative analysis of the Ezh2 signal intensity in WT or Dnd1^ter/ter germ cells (GC) and teratoma-forming cells (teratoma) relative to the surrounding somatic cells are shown at the bottom of the pictures. Comparisons of the somatic cells with germ cells and with teratoma-forming cells are shown in Fig. S4. The average signal intensity of 10 randomly selected somatic cells in each section was set as 1, and the signal intensity of each germ cell or teratoma-forming cell relative to the average value of the somatic cells in the same observed section was estimated. In total, three to five sections from three embryos of each genotype were observed. *P<0.05, ***P<0.001; n.s., not significantly different. Scale bars: 25 μm.

Dnd1 binds to the 3′-UTR of Ezh2 mRNA, and suppresses the inhibitory effect of miR-26a on Ezh2 mRNA. (A) A schematic representation of the effect of Dnd1 on the luciferase (luc) 3′-UTR of an Ezh2 reporter. In the absence of Dnd1, miR-26a associates with the 3′-UTR of Ezh2 mRNA and suppresses the expression of luciferase; in contrast, in the presence of Dnd1, the binding of miR-26a is inhibited, and luciferase activity is increased. (B) An expression vector containing mouse Dnd1, siRNA corresponding to a mature form of mouse miR26a, and the Renilla luc-3′-UTR of a mouse Ezh2 reporter vector were co-transfected into HEK293T cells. As a negative control for miR-26a and the Dnd1 expression vector, AllStras (AS) siRNA and an empty vector, respectively, were used. Renilla luciferase activity was normalized to the firefly luciferase activity that was expressed from the same vector, and the normalized luciferase activity of the cells transfected with the AS siRNA and the control empty vector (cont. plasmid) was set as 1. Data were obtained from four independent experiments. **P<0.01, ***P<0.001. (C) RIP assay results show the binding of Dnd1 to Ezh2 mRNA. An expression vector containing mouse Dnd1-HA (Dnd1) or an empty vector (cont) was transfected into HEK293T cells, and the cell extracts were immunoprecipitated using an anti-HA antibody. After purification of the RNA, the amount of precipitated mRNA was quantified by real-time qPCR. As a positive and a negative control, p27^kip1 and Gapdh, respectively, were used. The vertical axis shows the data relative to the input values. Shown are representative data from two independent experiments. The results of another independent experiment are shown in Fig. S8D.

Ccnd1 is a target of Ezh2. (A) Enrichment of H3K27me3 at the Ccnd1 locus in PGCs and ES cells. Representative images of the H3K27me3 ChIP-seq read density at the Ccnd1 promoter in E13.5 male PGCs (green) and ES cells (red). ChIP-seq data (SRX149169 and SRX186071) were visualized using the Integrative Genomics Viewer (http://software.broadinstitute.org/software/igv/). H3K27me3 was enriched at the transcription start site of Ccnd1 in E13.5 male PGCs, but not in ES cells. (B) Ezh2 KD and Dnd1 KD reduced the enrichment of H3K27me3 at Ccnd1 in ES cells. The histogram shows the ratios of the immunoprecipitated chromatin to the input chromatin determined by ChIP-qPCR analysis using the anti-H3K27me3 antibody. Oct4 and Hoxb1 were shown to be a negative locus and a positive locus, respectively. Shown are representative data from two independent experiments. The results of another independent experiment are shown in Fig. S8E. (C) Upregulation of Ccnd1 expression by Ezh2 KD in ES cells. The expression of Ezh2 and Ccnd1 was determined by real-time qPCR. (D) Downregulation of Ezh2 expression by Dnd1 KD in ES cells. The expression of Dnd1 and Ezh2 was determined by real-time qPCR. An empty vector was transfected as a control. Data were obtained from four independent experiments (C,D). *P<0.05, **P<0.01.

Ccnd1 expression in the teratoma-forming cells in Dnd1^ter/ter testes, and in germ cells in Dnd1^ter/ter testes and in wild-type/Dnd1^ter/+ (WT) testes at E14.5 to E18.5. (A,B) In Oct4-ΔPE-GFP-positive germ cells in WT (arrowheads) and Dnd1^ter/ter (arrows) testes at E14.5 (A) and E16.5 (B), the expression of Ccnd1 was undetectable. (C,D) 4C9-positive teratoma-forming cells in Dnd1^ter/ter testes (arrows) weakly expressed Ccnd1 at E17.5 (C), then more strongly expressed Ccnd1 at E18.5 (D); in contrast, Ccnd1 was undetectable in GFP-positive germ cells in WT testes (arrowheads) at both E17.5 and E18.5. Scale bars: 25 µm.

Expression of teratoma-related genes during PGC reprogramming in culture. (A-F) Purified E12.5 PGCs (E12.5) of Oct4-ΔPE-GFP transgenic embryos were cultured in the medium for PGC reprogramming without a feeder layer for 2, 4, or 6 days (EG d2, EG d4 or EG d6). The expressions of Ezh2 (A), Dnd1 (B), Ccnd1 (C), Sox2 (D), Nanog (E), and Mvh (F) were determined by RT-qPCR. (G) The expression of Ezh2 in germ cells at E12.5 to E16.5 and in teratoma-forming cells at E17.5 and E18.5 in Dnd1^ter/ter testes. The plots show the relative intensity of the fluorescent signals of Ezh2 in germ cells or teratoma-forming cells in comparison to those in the surrounding somatic cells based on the data shown in Figs 1 and 2. Data were obtained from three (d4 and d6) and four (d2 and d4) independent experiments (A-F). *P<0.05, **P<0.01, ***P<0.001.