An EED/PRC2-H19 Loop Regulates Cerebellar Development

Abstract

EED (embryonic ectoderm development) is a core subunit of the polycomb repressive complex 2 (PRC2), which senses the trimethylation of histone H3 lysine 27 (H3K27). However, its biological function in cerebellar development remains unknown. Here, we show that EED deletion from neural stem cells (NSCs) or cerebellar granule cell progenitors (GCPs) leads to reduced GCPs proliferation, cell death, cerebellar hypoplasia, and motor deficits in mice. Joint profiling of transcripts and ChIP-seq analysis in cerebellar granule cells reveals that EED regulates bunches of genes involved in cerebellar development. EED ablation exhibits overactivation of a developmental repressor long non-coding RNA H19. Importantly, an obvious H3K27ac enrichment is found at Ctcf, a trans-activator of H19, and H3K27me3 enrichment at the H19 imprinting control region (ICR), suggesting that EED regulates H19 in an H3K27me3-dependent manner. Intriguingly, H19 deletion reduces EED expression and the reprogramming of EED-mediated H3K27me3 profiles, resulting in increased proliferation, differentiation, and decreased apoptosis of GCPs. Finally, molecular and genetic evidence provides that increased H19 expression is responsible for cerebellar hypoplasia and motor defects in EED mutant mice. Thus, this study demonstrates that EED, H19 forms a negative feedback loop, which plays a crucial role in cerebellar morphogenesis and controls cerebellar development.

Keywords: EED, H19, Motor movement, PRC2; cerebellum.

Figure 1

Eed cKO mice develop motor impairments. A) Top, co‐immunostaining of EED (red) with Pax6 (green) in the cerebella at P21. Middle, co‐immunostaining of EED (red) with NeuN (green) in the IGL of cerebella at P21. Bottom, cerebellar sections from P21 mice were stained with antibodies against EED (red) and Calbindin(green). The co‐localization cells are defined by arrowheads. IGL, internal granule cell layer. Scale bars, 20 µm. B)Highly efficient depletion of EED protein from the nervous system in hGFAP‐Cre mediated conditional knockout mice as shown by qRT‐PCR using P14 cerebellar lysates. ^*** p < 0.001, n = 3, mean ± SEM. C)Western blot analysis of EED protein from the nervous system in hGFAP‐Cre mediated conditional knockout mice using P14 cerebellar lysates. D) Quantification of the density of the EED protein bands by normalization to the intensity of GAPDH bands in C. ^* p < 0.05, n = 4, mean ± SEM. E) Immunostaining of EED (green) in P21 cerebella of WT and EED^hGFAP mice. DNA was stained with DAPI. Scale bars, 50 µm. F) Violet staining of cerebellar sections from WT and EED^hGFAP mice shows dysmorphology in mutant mice during P7‐P21. Scale bars, 500 µm. G) Area of sagittal sections of the cerebellar vermis in WT and EED^hGFAP mice at the indicated postnatal developmental stages. ^** p < 0.01, ^*** p < 0.001, n = 3, mean ± SEM. (H) EED^hGFAP mice had reduced locomotivity to WT littermate mice in an open field test over a 5 min period. ^*** p < 0.001, n = 6, mean ± SEM. I) EED^hGFAP mice had reduced mean speed to WT littermate mice in an open field test over a 5 min period. ^*** p < 0.001, n = 6, mean ± SEM. J)The mean latency of mice to fall from the rotarod. Note the significant difference between the EED^hGFAP and WT mice. ^** p < 0.01, n = 7, mean ± SEM.

Figure 2

EED ablation exhibits defective GCP proliferation and leads to cell death in vivo. A) GCPs proliferation in EGL was labeled with anti‐BrdU in WT and EED^hGFAP mice at P7. Differentiated granule cells were co‐immunolabeled with anti‐NeuN antibodies. Arrowheads show BrdU‐positive cells. Scale bars, 20 µm. EGL, external granule cell layer. IGL, internal granule cell layer. ML, molecular layer. B) GCPs proliferation in EGL was evaluated after a 2 h chase following BrdU injection in WT and EED^hGFAP mice at P7. Quantification of BrdU‐positive cells within the EGL. ^*** p < 0.001, n = 3, mean ± SEM. C) Migrating GCPs in WT and EED^hGFAP mice were labeled with anti‐BrdU. Differentiated granule cells were co‐immunolabeled with anti‐NeuN antibodies. Arrowheads show a decreased number of BrdU‐positive and NeuN⁺ cells in the IGL, indicating impaired migration. Scale bar, 20 µm. D) Tracing of granule neuron migration was evaluated in WT and EED^hGFAP mice that were injected with BrdU at P7 and chased until P14. Quantification of BrdU‐positive and NeuN⁺ cells in the IGL. ^** p < 0.01, n = 3, mean ± SEM. E) BrdU (red), Ki67 (green), and DAPI (blue) immunofluorescent staining of WT and EED^hGFAP mice cerebellums at P10. White frames indicate folium at P10. Higher magnification of folium is shown in the adjacent right panels. Arrowheads indicate BrdU⁺Ki67⁻ cells. Left, Scale bars, 200 µm. Right, Scale bars, 10 µm. F) Apoptosis of newborn granule cells was evaluated in WT and EED^hGFAP mice that were injected with BrdU at P7 and chased 72 h after BrdU administration at P10. Proportion of BrdU⁺/Ki67⁻ cells in the EGL of WT and EED^hGFAP mice 72 h post–BrdU injection. ^*** p < 0.001, n = 3, mean ± SEM.

Figure 3

Up‐regulation of H19 in EED^hGFAP cerebellum. A) PCA cluster analysis of WT (green) and Eed cKO(red) mice cerebellum RNA‐seq samples. B) Heatmap of differentially expressed genes in WT and Eed cKO mice cerebellum samples. C) GO biological process analysis of upregulated genes (cKO vs WT). D) GO biological process analysis of downregulated genes (cKO versus WT). E) KEGG analysis of genes with down‐regulated expression in P14 Eed cKO mice cerebellums as compared with WT mice. F) qRT‐PCR analyses of genes that are up‐regulated expressed in Eed cKO cerebellar cells at P14. ^* p < 0.05, ^** p < 0.01, n ≥ 3, mean ± SEM. G) qRT‐PCR analyses of representative cerebellum development genes that are down‐regulated expressed in Eed cKO cerebellar cells at P14. ^* p < 0.05, ^*** p < 0.001, n ≥ 3, mean ± SEM.

Figure 4

EED controls the transcriptional program necessary for cerebellum development. A) Average genome‐wide occupancies of H3K27ac and H3K27me3 within ±5 kb of the gene body of all GENCODE. B) The levels of H3K27me3 and H3K27ac in isolated cerebellar tissues from P14 WT and EED^hGFAP mice were measured by immunoblotting. C) Quantified the density of the histone protein bands by normalization to the density of total histone H3. ^* p < 0.05, n = 3, mean ± SEM. D) ChIP‐seq binding profiles of H3K27me3 and H3K27ac at H19 ICR (loci −2–4 kb), Ctcf, Igf2 as indicated in WT and EED cKO cerebellum at P14. The regions marked in green show the gain or loss of H3K27me3 and H3K27ac. E) ChIP‐qPCR assessment of EED and H3K27me3 enrichment at ICR regions of H19. n = 3–4 wells per condition. ^* p < 0.05, mean ± SEM.

Figure 5

H19 deletion promotes the proliferation and migration of GCPs. A) Highly reduced EED protein from H19 KO mice as shown by qRT‐PCR using P14 cerebellar lysates. ^*** p < 0.001, n = 3, mean ± SEM. B) Western blot analysis of EED protein and H3K27me3 from H19 KO mice using P14 cerebellar lysates. C) Quantification of the density of the EED protein bands by normalization to the intensity of GAPDH bands in B. Quantification of the density of the H3K27me3 protein bands by normalization to the intensity of H3 bands in B. ^* p < 0.05, ^*** p < 0.001, n = 3, mean ± SEM. D) GCPs proliferation in EGL was labeled with anti‐BrdU in WT and H19 KO mice after a 2 h chase following BrdU injection at P7. Differentiated granule cells were co‐immunolabeled with anti‐NeuN antibodies. Arrowheads show BrdU‐positive cells. Scale bars, 20 µm. EGL, external granule layer. E) Quantification of BrdU‐positive cells within the EGL. ^* p < 0.05, n = 3, mean ± SEM. F) Migrating GCPs in WT and H19 KO mice were labeled with anti‐BrdU. Differentiated granule cells were co‐immunolabeled with anti‐NeuN antibodies. Tracing of granule neuron migration was evaluated in WT and H19 KO mice that were injected with BrdU at P7 and chased until P14. Arrowheads show BrdU‐positive and NeuN⁺ cells in the IGL. Scale bars, 20 µm. G) Quantification of BrdU⁺NeuN⁺ cells in the IGL. ^* p < 0.05, n = 3, mean ± SEM. IGL, internal granule cell layer.

Figure 6

H19 downregulation rescues the proliferation defect and cell survival in Eed cKO GCPs and cerebellar granule cells. A) RT‐qPCR analyses of H19 in GCP cells at 48 h after transfection with NC or siRNA‐H19 (n = 3 experiments). GCP cells were isolated from WT and Eed cKO mice at P6. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, mean ± SEM. B) Immunostaining for BrdU(red) in WT and Eed cKO GCPs transfection with NC or siRNA‐H19 for 48 h (n > 3 experiments). Scale bars, 50 µm. C) Quantification of BrdU⁺ cells (right) in WT and Eed cKO GCPs transfection with NC or siRNA‐H19 for 48 h (n > 3 experiments). ^* p < 0.05, ^** p < 0.01, mean ± SEM. D) Immunostaining for MAP2(green) and NeuN(red) in WT and Eed cKO GCPs transfection with NC or siRNA‐H19 for 48 h (n > 3 experiments). Scale bars, 20 µm. E) Quantification of NeuN⁺ cells (right) in WT and Eed cKO GCPs transfection with NC or siRNA‐H19 for 48 h (n > 3 experiments). ^* p < 0.05, ^** p < 0.01, mean ± SEM.

Figure 7

H19 downregulation partially rescues the motor defect in Eed cKO mice. A) Immunostaining for BrdU(green), NeuN(red), and DAPI in WT, EED^hGFAP, and EED^hGFAP‐H19 Het mouse cerebellums 2 h after BrdU injection at P7. White frames indicate folium at P7. Higher magnification of folium is shown in the adjacent right panels. Left: Scale bars, 100 µm. Right, Scale bars, 20 µm. B) Quantification of BrdU⁺ cells in EGL in WT, EED^hGFAP, and EED^hGFAP‐H19 Het mouse cerebellums. ^** p < 0.01, mean ± SEM. C) Migrating GCPs in WT, EED^hGFAP, and EED^hGFAP‐H19 Het mouse cerebellums were labeled with anti‐BrdU. Differentiated granule cells were co‐immunolabeled with anti‐NeuN antibodies. Tracing of granule neuron migration was evaluated in WT, EED^hGFAP, and EED^hGFAP‐H19 Het mouse cerebellums that were injected with BrdU at P7 and chased until P14. White frames indicate folium at P14. Higher magnification of folium is shown in the adjacent right panels. Left: Scale bars, 100 µm. Right, Scale bars, 20 µm. D) Quantification of BrdU⁺NeuN⁺ cells in the IGL. ^* p < 0.05, n = 3, mean ± SEM. IGL, internal granule cell layer. E) Representative motion trail from the open field test. F) EED^hGFAP‐H19 Het mice had significant improvement in locomotivity to EED^hGFAP littermate mice in an open field test over a 5 min period. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, n = 4–6, mean ± SEM. G) EED^hGFAP‐H19 Het mice had no significant mean speed to EED^hGFAP littermate mice in an open field test over a 5 min period. ^** p < 0.01, ^*** p < 0.001, n = 4–6, mean ± SEM. H)The mean latency of mice to fall from the rotarod. Note the significant difference between the EED^hGFAP and EED^hGFAP‐H19 Het mice. ^* p < 0.05, ^** p < 0.01, n = 6–9, mean ± SEM.