Supplementary Figure 11: H3K27me3 is differentially enriched in stMg and cbMg and genetic deletion of Eed in adult microglia leads to progressive changes.
From: Epigenetic regulation of brain region-specific microglia clearance activity
(a) H3K27me3 enrichment at the transcriptional start site (TSS) of genes in adult cxMg is negatively correlated with cxMg gene expression. Heatmap shows (left) the abundance of H3K27me3 (MACS, n=1) ranked by log2 fold change (log2fc) of H3K27me3 ChIP over input in cxMg nuclei at the TSS ± 5 kb of individual genes; (right) log2fc (DESeq2, n=2) of mRNAs from cxMg-TRAP over its unbound fraction. p< 2.2e-16, r=-0.3241428; Pearson correlation. (b) A large number of the H3K27me3+ genes are shared between cbMg and stMg. Scatter plot shows log2fc of H3K27me3 ChIP in cbMg over its input (x-axis) versus log2fc of H3K27me3 ChIP in stMg over its input (y-axis). 7312 genes are H3K27me3+ both in cbMg and stMg (red). GO annotations (Enrichr) enriched for 7312 H3K27me3+ genes shared in stMg and cbMg. y-axis: -log10 (p-value). (c) H3K27me3 is differentially enriched in cbMg and stMg. Profile plots depict mean Reads Per million Mapped Reads of H3K27me3 in stMg and cbMg with coverage over the TSS ± 5 kb. Left: 6571 genes equally enriched in stMg and cbMg. Middle: 1382 genes with higher H3K27me3 in stMg. Right: 592 genes with higher H3K27me3 in cbMg. (d) Schematic showing the generation of EedloxP/loxP mouse and the Cre-dependent deletion of Eed. Brackets indicate the two homology arms mediating recombination in embryonic stem cells. loxP (red) and frt (yellow) sites are indicated by arrow heads. Representative genotyping gel image of Eed+/+, EedloxP/+, and EedloxP/loxP mice is shown from > 100 independent experiments. (e) Deletion of Eed results in progressive up-regulation of direct PRC2-target genes. Box-and-whisker plots show relative expression of H3K27me3+ genes up-regulated in Eed-deficient stMg, cbMg, and cxMg related to Fig. 4. stMg: control: min=-5.736, 25%=-2.92, median=-1.165, 75%=-0.506, max=1.94; 3mo: min=-3.79, 25%=-0.809, median=-0.308, 75%=0.29, max=2.83; 6mo: min=-6.51, 25%=-2.009, median=-0.33, 75%=0.086, max=1.598; and 9mo: min=-0.05, 25%=1.09, median=1.97, 75%=5.376, max=8.77; p<0.0001; F=135.8, KWS=208.1. cbMg: control: min=-6.455, 25%=-3.629, median=-0.91, 75%=-0.287, max=0.33; 3mo: min=-3.975, 25% -0.79, median=-0.28, 75%=0.376, max=2.14; 6mo: min=-5.08, 25%=-0.898, median=-0.32, 75%=0.1988, max=3.18; and 9mo: min=0.487, 25%=1.07, median=2.03, 75%=3.38, max=7.47; p<0.0001; F=86.18, KWS=139.7. cxMg: control: min=-2.945, 25%=-1.09, median=-0.737, 75%=-0.4977, max=0.0658; 3mo: min=-2.34, 25%=-0.98, median=-0.60, 75%=-0.30, max=0.839; 6mo: min=-1.98, 25%=-0.4999, median=0.117, 75%=0.519, max=1.379; and 9mo: min=0.42, 25%=0.86, median=1.59, 75%=2.96, max=5.056; p<0.0001; F=282.5, KWS=302.8. One-way ANOVA (Kruskal-Wallis test) with Dunn’s Multiple Comparison post-hoc analysis to compare all pairs of columns. Box-and-whisker plots (Tukey) show normalized expression values in z-scored log2 RPKM. **p ≤ 0.01, ***p ≤ 0.001. (f) Deletion of Eed does not lead to a loss of microglial identity or pro-inflammatory activation but leads to activation of transcription factors and phagocytic genes in stMg. Heatmaps show the variation in the expression levels (z-scored log2 RPKM) of pro-inflammatory / immediate early genes and, pan-microglia genes in control, 3 & 9mo Eed-deficient stMg / cbMg; and, phagocytosis/metabolism-related, and transcription factor-encoding genes in in control, 3 & 9mo Eed-deficient stMg (DESeq2, n=2/group). (g) Genome browser views (IGV) of selected genes in stMg TRAP from control and Cx3cr1CreErt2/+; Eef1a1LSLeGFPL10a/+;EedloxP/loxP mice. (h) Venn diagrams show overlap between 215 up-regulated (red) and 643 down-regulated (blue) genes in Eed-deficient stMg in 9mo mice with genes up- (Down: p=0.0416, OR=1.472, χ2=4.153; red; Up: p=0.007, OR=2.088, χ2=7.283); and down-(blue; Down: p < 0.0001, OR=5.365, χ2=181.9; Up: p=0.6152, OR=1.232, χ2=0.2526) regulated in Alzheimer’s Disease (AD)-associated microglia in mice12. p values and odds ratios (underneath each overlap) were calculated using the non-parametric χ2 test from a total 22,706 protein-coding genes analyzed in DESeq2. (i) Quantification of regional density and morphology of stMg Cx3cr1CreErt2/+; Eef1a1LSLeGFPL10a/+; EedloxP/loxP (mutant) and Cx3cr1CreErt2/+; Eef1a1LSLeGFPL10a/+; EedloxP/+ (control) mice. Cell body densities (control: mean=163.0, SEM=20.62; mutant: mean=127.4, SEM=10.00; p=0.1958, F=4.250, t4=1.551). Sholl intersections at 10 μm interval (control: mean=17.91, SEM=2.889; mutant: mean= 15.21, SEM=1.220; p=0.4373, F=5.605, t4=0.8619). Number of branch endings (control: mean=23.18, SEM=3.881; mutant: mean=22.90, SEM=2.362; p=0.9544, F=2.70, t4=0.060). 20-25 cells from n=3 mice/genotype. Bar graphs with individual data points show mean ± SEM. All t-tests were two-tailed unpaired.
