The genome-wide landscape of DNA methylation and hydroxymethylation in response to sleep deprivation impacts on synaptic plasticity genes

Abstract

Sleep is critical for normal brain function and mental health. However, the molecular mechanisms mediating the impact of sleep loss on both cognition and the sleep electroencephalogram remain mostly unknown. Acute sleep loss impacts brain gene expression broadly. These data contributed to current hypotheses regarding the role for sleep in metabolism, synaptic plasticity and neuroprotection. These changes in gene expression likely underlie increased sleep intensity following sleep deprivation (SD). Here we tested the hypothesis that epigenetic mechanisms coordinate the gene expression response driven by SD. We found that SD altered the cortical genome-wide distribution of two major epigenetic marks: DNA methylation and hydroxymethylation. DNA methylation differences were enriched in gene pathways involved in neuritogenesis and synaptic plasticity, whereas large changes (>4000 sites) in hydroxymethylation where observed in genes linked to cytoskeleton, signaling and neurotransmission, which closely matches SD-dependent changes in the transcriptome. Moreover, this epigenetic remodeling applied to elements previously linked to sleep need (for example, Arc and Egr1) and synaptic partners of Neuroligin-1 (Nlgn1; for example, Dlg4, Nrxn1 and Nlgn3), which we recently identified as a regulator of sleep intensity following SD. We show here that Nlgn1 mutant mice display an enhanced slow-wave slope during non-rapid eye movement sleep following SD but this mutation does not affect SD-dependent changes in gene expression, suggesting that the Nlgn pathway acts downstream to mechanisms triggering gene expression changes in SD. These data reveal that acute SD reprograms the epigenetic landscape, providing a unique molecular route by which sleep can impact brain function and health.

Figure 1

(a) Non-rapid eye movement (NREM) sleep spectral power during the first hour of recovery (REC) after sleep deprivation (SD) expressed relative to the 24-h baseline (BL) (n=9 Nlgn1^+/+, 13 Nlgn1^+/−, 12 Nlgn1^−/−). Differences between Nlgn1^−/− and Nlgn1^+/+ are highlighted by red (P≤0.05) and pink (P<0.1) symbols (also in c). (b) Slow wave (SW) density and properties averaged during light or dark periods of BL and REC. SW density was higher in Nlgn1^−/− than in Nlgn1^+/+ and Nlgn1^+/− for the 12-h light (genotype: F_2,24≥3.6, P<0.04), and was higher in REC than BL for the light period and lower in REC than BL for the dark period (condition: F_1,34≥50.7, P<0.0001). Amplitude was higher in Nlgn1^−/− than Nlgn1^+/+ and Nlgn1^+/− for both periods (genotype: F_2,34≥3.8, P<0.03); and higher in REC than BL for the 12-h light, whereas lower in REC than BL for the 12-h dark (condition: F_1,34≥5.3, P<0.03). Duration of positive phase and of negative phase was higher in REC than BL for both periods (condition: F_2,34≥46.2, P<0.0001). For the 12-h light, slope was higher in REC than in BL only in Nlgn1^−/− and Nlgn1^+/− (interaction: F_2,34≥3.2, P≤0.05). For the 12-h dark, slope was lower in REC compared with BL (condition: F_1,34=66.5, P<0.0001) and higher in Nlgn1^-/- than Nlgn1^+/+ or Nlgn1^+/- (genotype: F_2,34=5.4, P<0.01). (c) Time course of SW properties averaged across equal intervals during BL and REC. During light periods, amplitude was higher in Nlgn1^−/− than Nlgn1^+/+ in the first six intervals of REC sleep as was slope for specific intervals (Interactions: F_38,589≥2.3, P<0.03). During dark periods, Nlgn1^−/− showed higher slope than Nlgn1^+/+ and Nlgn1^+/− (genotype: F_2,31=4.6, P<0.02). All SW properties varied significantly with time, and SD is indicated by the black rectangle. (d) Decay of SW density, amplitude and slope between the first and the last intervals of REC light. The amplitude and slope decay was higher in Nlgn1^−/− than Nlgn1^+/− and Nlgn1^+/+ (F_2,31>6.6, P<0.01). Nlgn1: Neuroligin-1.

Figure 2

(a) Heat maps of the 184 transcripts displaying an interaction Genotype × sleep deprivation (SD) at P<0.05. Columns refer to individual microarray data (n=6 per group). Transcripts were ordered by hierarchical clustering (complete linkage) with Nlgn1^−/− fold change taken as a reference. (b) Scatter plot of the fold change in expression induced by SD in Nlgn1^+/+ versus Nlgn1^−/− mice. Only the 1298 probes significantly affected by SD with a false discovery rate (FDR) <0.05 (two-way analysis of variance) are shown. The 184 transcripts that showed an interaction with at P<0.05 before FDR correction are highlighted in red. (c) Microarray data and quantitative PCR (qPCR) validations of selected targets showing measurements in Nlgn1^+/+ versus Nlgn1^−/− mice under control (Ctrl) and SD conditions. *P<0.05, **P<0.01, ***P<0.001 compared with control (and between genotypes for Arid4b). (d) Correlation between fold change relative to Nlgn1^+/+ in control condition observed in microarray and qPCR validation data sets. Nlgn1: Neuroligin-1.

Figure 3

(a) Representative examples of changes in 5mC and mRNA expression after a 6-h sleep deprivation (SD) in the mouse cerebral cortex. (b) Correlation between DNA methylation differences measured by 5mC-immunoprecipitation (IP)-arrays and validated by 5mC-IP-quantitative PCRs. (c) Expanded views from the UCSC genome browser at the Dlg4 gene location. The first track shows average methylation probe fold differences (Log2) and the second shows regions significantly differentially methylated. The last track shows exons and introns taken from the mouse NCBI RNA reference sequence collection (RefSeq). ^#P<0.1, *P<0.05, **P<0.01, ***P<0.001 compared with control (Ctrl).

Figure 4

(a) Heat map of 4000 more significant probes showing an effect of sleep deprivation (SD) on 5hmC in the mouse cerebral cortex. Columns refer to three pools of DNA of three control (Ctrl) and three SD mice (total nine per group). Transcripts were ordered by hierarchical clustering (complete linkage). (b) Differentially 5hmC sites were enriched in exons (P<10E–300, Fisher's Exact Test); TTS/3′-untranslated region (P<10E–300) but negatively enriched in promoters (P<10E–300), introns (P<6.8E–11) and intergenic regions (P<1.6E–43). (c) Correlation between DNA hydroxymethylation differences measured by 5hmC-IP-arrays and 5hmC-IP-quantitative PCRs for 15 different genomic regions. (d) Changes in 5hmC after a 6-h SD for selected targets. (e) Selected changes in 5hmC and mRNA expression after a 6-h SD. ^#P<0.1, *P<0.05, **P<0.01, ***P<0.001 compared with control (Ctrl).