Neuronal activity modifies the chromatin accessibility landscape in the adult brain

Abstract

Neuronal activity-induced gene expression modulates the function and plasticity of the nervous system. It is unknown whether and to what extent neuronal activity may trigger changes in chromatin accessibility, a major mode of epigenetic regulation of gene expression. Here we compared chromatin accessibility landscapes of adult mouse dentate granule neurons in vivo before and after synchronous neuronal activation using an assay for transposase-accessible chromatin using sequencing (ATAC-seq). We found genome-wide changes 1 h after activation, with enrichment of gained-open sites at active enhancer regions and at binding sites for AP1-complex components, including c-Fos. Some changes remained stable for at least 24 h. Functional analysis further implicates a critical role of c-Fos in initiating, but not maintaining, neuronal activity-induced chromatin opening. Our results reveal dynamic changes of chromatin accessibility in adult mammalian brains and suggest an epigenetic mechanism by which transient neuronal activation leads to dynamic changes in gene expression via modifying chromatin accessibility.

Figure 1

Modification of the chromatin accessibility landscape in the adult mouse dentate gyrus by transient neuronal activation. (a) UCSC genome browser visualization of gained-open and gained-closed chromatin profiling coverage at the Arc and Gabrr1 loci before (E0; magenta) and 1 h after synchronous neuronal activation (E1; green). Data from each individual sample are shown. Significant gained-open regions (red bars) and a gained-closed region (blue bar) are indicated (P < 1e-5; fold changes > 2). (b) Summary of FAIRE-qPCR results for gained-open and gained-closed regions at the Arc and Gabbr1 loci. Data from three individual samples for each condition are shown (*P < 0.05; permutation test). (c) Comparison of open chromatin profiles of the dentate gyrus of adult mice between E1 and E0. Differential regions are shown in a volcano plot. Gained-open sites are shown in red and gained-closed sites are shown in blue (P < 1e-5; Fold changes > 2; T-test; n = 4 in each group). (d) GO analysis of genes associated with gained-open (red) and gained-closed (blue) regions at E1 compared with E0.

Figure 2

Characterization of gained-open and gained-closed regions at E1 compared to E0. (a) Genomic features of gained-open (n = 11,438) and gained-closed regions (n = 1,739) at E1. (b) Genomic regions with neuronal activation-induced changes in chromatin accessibility are enriched with H3K4me1 and H3K27Ac active histone marks, but not with histone modifications for repressive (H3K27me3) markers. Regions between start and end define boundaries of differential sites. ChIP-seq mapped reads of various histone modifications and CTCF (see Supplementary Table 2 data sources) were plotted in heatmap views. (c) Characterization of histone modification states of gained-open and gained-closed regions using ChromHMM. The Star symbol highlights the most enrichment of gained-open and gained-closed regions are in active enhancers (d) Expression levels of genes associated with gained-open and gained-closed regions between E0 and E1. Significantly upregulated (pink) and downregulated (light blue) genes between E1 and E0, and those overlapped with genes associated with gained-open peaks (red) or with genes associated with gained-closed peaks (blue) are coded in color (P < 0.05; Fold changes >2; Fisher’s exact test).

Figure 3

Enrichment of cFos-binding sites at neuronal activity-induced chromatin opening regions. (a) Common motifs within chromatin gained-open regions at E1 (top) predicted using MEME-ChIP. (b) Pie chart of gained-open peaks with the specific DNA logo. (c) Venn diagram of cFos ChIP binding sites and chromatin gained-open regions at E1. (d) Aggregate plots of ATAC-seq signals centered at cFos, FosB and JunB and CREB binding sites at E0 and E1. (e) Aggregate plots of cFos, FosB, JunB and CREB signals before and after KCl stimulation centered at gained-open (top panel) and gained-closed (bottom panel) sites at E1. The ChIP-seq data used for plots are from a previously published dataset.

Figure 4

Critical role of cFos in neuronal activity-induced chromatin opening of regions with cFos binding sites. (a) UCSC genome browser visualization of ATAC-seq profiles at the Hivep3 locus under different conditions (top panels). Data from merged biological replicates (n = 2–3 mice in each group) are shown. Red bar indicates the region where cFos knockdown blocked activity-induced chromatin opening. Also shown is a plot of expression levels of Hivep3 under different conditions (bottom panel; n = 3 mice in each groups; *P < 0.05; **P < 0.01; ^#P > 0.1; permutation test) (b) Principal component analysis of ATAC-seq of neurons expressing shRNA-Ctrl and shRNA-cFos at E0 and E1. (c) Aggregate plot of ATAC-seq signals centered on neuronal activity-induced gained-open regions with cFos-binding sites for neurons expressing shRNA-Ctrl and shRNA-cFos at E0 and E1. (d) Box-plot of mean expression levels of up-regulated genes in neurons expressing shRNA-Ctrl and shRNA-cFos in response to neuronal activation (*P < 0.01; Wilcoxon rank-sum test; P < 2.2e-16). Center line: median value; box limits: 25^th and 75^th percentile values; whiskers; maximum and minimum data points excluding the outliers; outliers: data points that are higher or lower than the 1.5 times the quartile (25^th and 75^th) values. (e) Principal components analysis of ATAC-seq of neurons expressing EYFP or cFos in the absence of neuronal stimulation. (f) Aggregate plot of ATAC-seq analysis of neurons at E0 and E1 and neurons overexpressing cFos (cFos-OE) and/or EYFP (EYFP-OE) at E0. Similar to (c). (g) Box-plot of mean expression levels of up-regulated genes induced by cFos expression at E0 or neuronal activation at E1 without exogenous cFos expression. Similar to (d) (^#P > 0.01; Wilcoxon rank-sum test; P = 0.81).

Figure 5

Characterization of neuronal activity-induced chromatin accessibility changes at different time points. (a) Principal component analysis of ATAC-seq datasets. Each dot represents data from one sample and is colored coded for before (E0), 1 h (E1), 4 h (E4), or 24 h (E24) after synchronous neuronal activation. (b, c) Venn diagrams of differential chromatin accessible regions (b) and their associated genes (c). (d) GO analysis of genes with sustained gained-open chromatin accessibility until 24 h.

Figure 6

Characterization of chromatin accessibility gained-open regions sustained for 4 h and 24 h. (a) Genomic regions with gained-opening sustained for 4 h and 24 h are enriched with H3K4me1 and H3K27Ac active histone marks, but not with histone modifications for repressive (H3K27me3) markers. Similar to Fig. 2b. (b) Characterization of histone modification states of gained-open regions sustained for 4 h and 24 h using ChromHMM. Similar to Fig. 2c. (c) Expression levels of genes associated with gained-open regions sustained for 4 h at E1 and E4. Significantly upregulated (pink) and downregulated (light blue) genes at both E1 and E4 compared to E0, and those overlapped with genes associated with gained-open peaks sustained for 4 h are coded in red (P < 0.05; Fold changes >2).

Figure 7

Gained-open sites can be maintained without cFos occupation. (a) Common motifs within chromatin gained-open regions sustained for 4 h (top panel) and 24 h (bottom panel) predicted using MEME-ChIP. (b) Pie chart of gained-open regions sustained for 4 h and 24 h with cFos-binding sites. (c) ChIP of cFos and IgG followed by qPCR analysis of Hivep3 and Gria2 loci in the adult dentate gyrus at E0, E4 and E24 (normalized percentage input). Data from individual samples are shown (^#P > 0.05; permutation test). (d) A working model for neuronal activity-induced changes in chromatin accessibility and dynamic changes in neuronal gene expression over time.