Role of Tet1/3 Genes and Chromatin Remodeling Genes in Cerebellar Circuit Formation

Abstract

Although mechanisms underlying early steps in cerebellar development are known, evidence is lacking on genetic and epigenetic changes during the establishment of the synaptic circuitry. Using metagene analysis, we report pivotal changes in multiple reactomes of epigenetic pathway genes in cerebellar granule cells (GCs) during circuit formation. During this stage, Tet genes are upregulated and vitamin C activation of Tet enzymes increases the levels of 5-hydroxymethylcytosine (5hmC) at exon start sites of upregulated genes, notably axon guidance genes and ion channel genes. Knockdown of Tet1 and Tet3 by RNAi in ex vivo cerebellar slice cultures inhibits dendritic arborization of developing GCs, a critical step in circuit formation. These findings demonstrate a role for Tet genes and chromatin remodeling genes in the formation of cerebellar circuitry.

Figure 1. Metagene Analysis of Cerebellar GC Gene Expression Using TRAP Methodology

(A) NeuroD1-EGFP expression in cerebellar GCs of Tg(NeuroD1-Egfp-L10a) TRAP mice at postnatal days 0, 7, 12, 18, 21 and 56 (P0, P7, P12, P18, P21 and P56). Scale bars indicate 80 μm (P56, 125 μm). At P0, most postmitotic, NeuroD1-EGFP+ cells are located in the EGL with only a few positive cells in the IGL. The number of NeuroD1-EGFP+ cells increases between P7 and P12 until P15, when all postmitotic NeuroD1-EGFP+ GCs have migrated into the IGL as shown in (B). (B) Progressive developmental stages of cerebellar GCs. By P12, although some GCs are still migrating, the vast majority are post-migratory and are forming synaptic connections. After GCs migrate by the PCs (P18-P21), they become multi-polar, after which they extend dendrites which form connections with ingrowing mossy fiber afferents (P21). (C) Transcriptional profiles of cerebellar GCs at P0, P7, P12, P18, P21 and P56, are decomposed into two matrices and projected into 7 time-dependent metagenes (C1_7-C7_7). “Metagenes” refer to aggregate patterns of gene expression (see text). (D) Expression profiles of epigenetic pathways, including DNA methyltransferases (Dnmts), histone acetyltransferases (HATs), Chromodomain helicase DNA binding proteins (Chds), histone methyltransferases (HMTs), Tet enzymes (Tets) and histone deacetylases (HDACs). The metagene that includes a given set of genes is listed on the right. (E) Expression profiles of gene categories with the most significant difference between immature (P0-P7) and mature cerebellar GCs, including transcription factors and genes highly expressed in adult GCs (Adult cerebellum), ion channels, synaptic genes and axon guidance genes, all of which fall in different metagenes. In the study, expression levels of metagenes, pathways and individual genes are shown as heatmaps which are row normalized ((X - mean(X))/standard_deviation(X)) and cut off at +/- 2. Color bar indicates color intensity for gene expression changes between +/- 2 fold change. See also Figure S1 and S3.

Figure 2. Sequential Timing of the Expression of Epigenetic Pathway Genes during Cerebellar GC Development

(A-D) Expression patterns of epigenetic genes encoding ATP-dependent chromatin remodeling complexes and centromere proteins (Smarcs and Cenps, C1_7, A), HATs (C1_7, B), histone demethylases (HDMs, C), and HMTs (C1_7 and C2_7, D). The heatmaps for different categories of epigenetic genes are organized from A-D according to their sequential timing during GC development. Schematic summary of temporal expression of epigenetic genes during cerebellar GC development (P0-P56) is shown in Figure S2D. Genes with known roles in cerebellar GC development (Smarca5 and Smarcb1) and encoding centromere proteins (Cenpa and Cenph), along with those selected for qRT-PCR analysis (Figure 2A-D), are marked with arrows. (E-H) Expression changes of selected genes were verified by qRT-PCR, including four histone methyltransferases involved in forming the methylation of histone H3 on lysine 4, 9 or 27 (K4, K9 or K27) (Suv39h2 for H3K9me3, Ezh2 for H3K27me3, Setd7 for H3K4me1 and Mll2 for H3K4me3). In qRT-PCR curve diagrams, Y-axis indicates relative mRNA level normalized to Beta-2-Microglobulin and HPRT1. Data represent the mean. Error bars indicate SD. Each data point represents nine replicates from three independent biological samples. See also Figure S2.

Figure 3. Increase in Tet Gene Expression and 5hmC Levels During the Period of Circuit Formation in Developing GCs

(A) Expression profiles of Tet genes (Tet1, Tet2 and Tet3), which generates 5hmC, and vitamin C transporter genes (Slc2a1, Slc2a3, Slc23a1 and Slc23a2), which regulate vitamin C levels, in developing cerebellar GCs (P0-P56). (B-D) qRT-PCR analysis of the expression of Tet genes (B-C), and vitamin C transporter genes (D). In qRT-PCR curve diagrams, Y-axis indicates relative mRNA level with the data normalized to Beta-2-Microglobulin and HPRT1. Data represent the mean. Error bars indicate SD. Each data point represents nine replicates from three independent biological samples. (E) Percentage of 5hmC and 5mC in genomic DNA purified from mouse cerebellar cortex at different developmental stages (P0, P7, P12, P18, P21 and P56), determined by liquid chromatography-mass spectrometry (LC-MS) analysis. 5mC and 5hmC contents are expressed as the percentage of 5hmC or 5mC in the total pool of cytosine. Data are the mean ± SD from triplicate analyses. See also Figure S4.

Figure 4. Vitamin C Treatment Enhances the Expression of Axon Guidance and Ion Channel Genes and 5hmC Levels

(A) Expression profiles of genes involved in signaling pathways (SHH, WNT, FGF, TGF-β and BMP), focal adhesion genes, genes highly expressed in adult GCs (Adult cerebellum), ion channels, synaptic genes and axon guidance genes are shown as a heatmap. A dramatic change in the expression of axon guidance genes (red text) is evident in ES cell-derived GCs treated with vitamin C (vit C, green text and delimited by yellow rectangles). (B-D) Expression profiles of axon guidance ligands (B), axon guidance receptors (C) and ion channels (D) in ES cell-derived GCs treated with vitamin C. Axon guidance genes that function in cerebellar development are marked with arrows. (E-F) Global 5hmC and 5mC levels in ES cell-derived GCs assayed by dot blot analysis (E), quantified in (F). Data represent the mean, n=3 independent experiments. Error bars indicate SD. ***p<0.001; t-test. See also Figure S5.

Figure 5. Increased Expression Levels of Axon Guidance and Ion Channel Genes Correlate with Higher Levels of 5hmC at Exon Start Sites

(A) The relative enrichment of 5hmC (dark color) and input (light color) at genomic elements in ES cell-derived GCs without treatment (Control, blue) and with vitamin C treatment (vit C, red), and cerebellar GCs at P7 (P7, green) and P56 (P56, purple). In the diagram, Y-axis indicates the percentage of 5hmC peaks distributed into each genomic element over all called peaks in each cell type (Figure S6A). The enrichment ratio (5hmC/Input) is highest for exons (red text) and is indicated at the top of the exon columns. (B) Highest 5hmC levels map to the exon start sites of the most highly expressed genes. Aligned profiles of 5hmC enrichment over exons are categorized by gene expression levels (High, Medium, Low and Input) (see Supplemental Experimental Procedures). “Exon start site” refers to all exons. The X axis is the distance from the exon start site, which is set as “0” in kilobases (kb). (C-E) Increased 5hmC levels in Robo1 and Slit2, a representative axon guidance receptor and ligand gene, respectively, and Scn8a, an ion channel important for cerebellar development, in control and vitamin C treated ES cell-derived GCs. The 5hmC line plots (C-E, marked in red) are indicated with normalized average 5hmC summits (Y axis). The X axis is the distance from the exon start site, which is set as “0” in kilobases (kb). The maximal 5hmC level is seen at exon start sites (green arrow). Input line plots are marked in black at the bottom. The increase in Robo1 and Slit2 expression in ES cell-derived GCs treated with vitamin C is seen in Figure 4B-C. (F-H) Intragenic 5hmC peaks in Robo1, Slit2 and Scn8a in ES cell-derived GCs without (control, blue) and with vitamin C treatment (vit C, red), are shown using Integrative Genomics Viewer (IGV). The last row in black represents the gene bodies located in this particular region of the genome. (The scale of the peaks is indicated in the upper right side of the control panel.) See also Figure S6 and S7.

Figure 6. Tet RNAi Knockdown Decreases 5hmC Levels and Down-regulates the Expression of Axon Guidance and Ion Channel Genes

(A) Schematic diagrams of Venus-shRNA constructs. The human U6 promoter (U6) drives RNA Polymerase III transcription for generation of shRNA transcripts, and the human phosphoglycerate kinase promoter (hPGK) drives expression of Venus. (B) Purified P7 cerebellar GCs transfected with Tet1/3 shRNA plasmids that co-express the fluorophore Venus and stained for Venus and Tuj1 (red) are viable after 3DIV. Scale bar: 25 μm. (C and E) The expression levels of selected genes in purified P7 cerebellar GCs when Tet1 and Tet3 were knocked down were determined by qRT-PCR. Purified GCs at P7 were infected with lentiviral particles of shRNA plasmids targeting Tet1 or Tet3 and co-expressing the fluorophore Venus. Scrambled shRNA was used as a control and the expression level of selected genes in the control cells was set as 1. In the qRT-PCR histogram, Y-axis indicates relative mRNA level normalized to Beta-2-Microglobulin and HPRT1. Data represent mean. Error bars indicate SD. Each data point represents nine technical replicates from three independent biological samples. *p<0.05; t-test. (D) RNAi knockdown of Tet1/3 in purified P7 cerebellar GCs. By dot blot analysis, 5hmC levels decrease and 5mC levels increase slightly. See also Figure S7.

Figure 7. Tet Knockdown Impairs Dendritic Arborization of Cerebellar GCs

(A) Coronal slices of P7 mouse cerebella electroporated with equal amounts of Tet1 and Tet3 shRNA plasmids or scrambled control shRNA plasmid that co-express the fluorophore Venus were cultured for 3 days and immunostained for Venus (green) to visualize the transfected GCs and for Calbindin (red) to visualize PCs. Major developmental layers are indicated. EGL, external granule layer; PCL, Purkinje cell layer; IGL, internal granular layer. Scale bar: 1 mm. (B) Higher magnification images showing dendritic growth of labeled GCs in cerebellar slices transfected with the scrambled or Tet shRNA constructs. Multipolar GCs extending dendrites are seen throughout the IGL deep to the PCs. (C-D) Tet knockdown impairs dendritic arborization of cerebellar GCs without affecting GC migration. Quantification of the effect of Tet1 and Tet3 gene knockdown on dendritic growth (C) and GC migration (D) observed at 3 DIV. Data depicts average percentage (Error bars, SD) from four different cultures (approximately 30-40 cells counted in each selected region) for each condition. (***p < 0.001; t-test). See Figure 1B for a schema of GC developmental stages. See also Figure S7.