Developmental regulation and individual differences of neuronal H3K4me3 epigenomes in the prefrontal cortex

Abstract

Little is known about the regulation of neuronal and other cell-type specific epigenomes from the brain. Here, we map the genome-wide distribution of trimethylated histone H3K4 (H3K4me3), a mark associated with transcriptional regulation, in neuronal and nonneuronal nuclei collected from prefrontal cortex (PFC) of 11 individuals ranging in age from 0.5 to 69 years. Massively parallel sequencing identified 12,732-19,704 H3K4me3 enriched regions (peaks), the majority located proximal to (within 2 kb of) the transcription start site (TSS) of annotated genes. These included peaks shared by neurons in comparison with three control (lymphocyte) cell types, as well as peaks specific to individual subjects. We identified 6,213 genes that show highly enriched H3K4me3 in neurons versus control. At least 1,370 loci, including annotated genes and novel transcripts, were selectively tagged with H3K4me3 in neuronal but not in nonneuronal PFC chromatin. Our results reveal age-correlated neuronal epigenome reorganization, including decreased H3K4me3 at approximately 600 genes (many function in developmental processes) during the first year after birth. In comparison, the epigenome of aging (>60 years) PFC neurons showed less extensive changes, including increased H3K4me3 at 100 genes. These findings demonstrate that H3K4me3 in human PFC is highly regulated in a cell type- and subject-specific manner and highlight the importance of early childhood for developmentally regulated chromatin remodeling in prefrontal neurons.

Fig. 1.

Epigenome profiling in PFC neurons. (A) Representative examples of fluorescence-activated sorting of NeuN+ tagged neuronal nuclei. (Scale bar, 10 μm.) (B and C) H3K4me3 profiles for the 11 PFC NeuN+ neuronal samples (red), and for comparison, NeuN− (orange) and CD4⁺ lymphocytes (blue) in a (B) 20-kb region of Chr 17 and (C) 1.3-MB region of Chr 8.

Fig. 2.

The H3K4me3 epigenome of PFC neurons is different from nonneuronal cells. Heatmaps showing Pearson correlation coefficients for pair-wise comparison of raw tag counts within 2 kb of annotated TSS. (Lower) 11 neuronal (NeuN+) samples from subjects 1–11 (female pink, male blue), two nonneuronal (NeuN−) samples from subjects 6 and 11, and CD4⁺ T cells and two lymphocyte-derived cell lines K562 and GM12878. (Upper) The same comparison for the 11 neuronal samples as in Bottom, but with a different color scale to highlight differences between neuronal samples. Notice (Lower) very high levels of correlation between NeuN+ samples, but not between NeuN+ and NeuN− samples or lymphocytes. Notice further (Upper) that correlations are higher for neuronal samples closer in age.

Fig. 3.

Peaks common to all 11 NeuN+ samples are enriched in neuron development and synaptic transmission biological processes. Only the most significant GO categories are shown.

Fig. 4.

H3K4me3 map as guide to find novel transcripts expressed in prefrontal neurons. (A) 4-kb region of Chr 9 showing, proximal to ESTs, a representative example (subject no. 11) of a peak present in all NeuN+ samples but not present in lymphocytes or NeuN− samples. y axis, number of reads normalized by sequencing depth. Purple tag marks sequence used for in situ hybridization of sections from adult PFC specimen shown in (B Left to Right) Nissl, antisense, and sense probe. Notice intense labeling of layer III pyramidal neurons. (Scale bar, 100 μm.)

Fig. 5.

Numbers and distribution of H3K4me3 peaks in PFC neurons. (A) Distribution of 7,949 peaks common to all 11 subjects in relation to annotated TSSs (<2 kb, 2–10 kb, > 10 kb as indicated) (B) Same as in A but for subject specific peaks. Numbers are expressed as mean ± SEM with n = 11. Notice different distribution patterns of “common,” in A, and “subject specific” peaks, in B.