Disrupted sleep without sleep curtailment induces sleepiness and cognitive dysfunction via the tumor necrosis factor-α pathway

Abstract

Background: Sleepiness and cognitive dysfunction are recognized as prominent consequences of sleep deprivation. Experimentally induced short-term sleep fragmentation, even in the absence of any reductions in total sleep duration, will lead to the emergence of excessive daytime sleepiness and cognitive impairments in humans. Tumor necrosis factor (TNF)-α has important regulatory effects on sleep, and seems to play a role in the occurrence of excessive daytime sleepiness in children who have disrupted sleep as a result of obstructive sleep apnea, a condition associated with prominent sleep fragmentation. The aim of this study was to examine role of the TNF-α pathway after long-term sleep fragmentation in mice.

Methods: The effect of chronic sleep fragmentation during the sleep-predominant period on sleep architecture, sleep latency, cognitive function, behavior, and inflammatory markers was assessed in C57BL/6 J and in mice lacking the TNF-α receptor (double knockout mice). In addition, we also assessed the above parameters in C57BL/6 J mice after injection of a TNF-α neutralizing antibody.

Results: Mice subjected to chronic sleep fragmentation had preserved sleep duration, sleep state distribution, and cumulative delta frequency power, but also exhibited excessive sleepiness, altered cognitive abilities and mood correlates, reduced cyclic AMP response element-binding protein phosphorylation and transcriptional activity, and increased phosphodiesterase-4 expression, in the absence of AMP kinase-α phosphorylation and ATP changes. Selective increases in cortical expression of TNF-α primarily circumscribed to neurons emerged. Consequently, sleepiness and cognitive dysfunction were absent in TNF-α double receptor knockout mice subjected to sleep fragmentation, and similarly, treatment with a TNF-α neutralizing antibody abrogated sleep fragmentation-induced learning deficits and increases in sleep propensity.

Conclusions: Taken together, our findings show that recurrent arousals during sleep, as happens during sleep apnea, induce excessive sleepiness via activation of inflammatory mechanisms, and more specifically TNF-α-dependent pathways, despite preserved sleep duration.

Figure 1

Circadian sleep characteristics and sleep state distribution during baseline conditions (solid line) and after 15 days of sleep fragmentation (dotted line) in mice. (A–C) There was an absence of significant alterations in sleep architecture, despite (D) continuing successful induction of episodic awakenings. (E) No changes in delta frequency power of the EEG emerged; but there were (F) marked reductions in mean sleep latency. (BL) baseline vs. sleep fragmentation (SF); *P < 0.0001; n = 12/group). Shaded regions represent dark phase of the circadian cycle (07.00 to 19.00 hours).

Figure 2

Spatial learning performance. (A) (a, b) Mean path lengths (cm) and latencies (seconds) to locate the target platform during place training, (c) swim speed and (d) mean percentage time in the target quadrant during probe trials after completion of water maze testing, with (e) path lengths and (f) latencies during assessment of spatial task retention in the water maze in mice exposed to 15 days of sleep fragmentation and those maintained in control sleep conditions (n = 18 per group; *P<0.0001). (B) Mice exposed to sleep fragmentation (a, b) showed less immobility during the forced swim test, (c) performed significantly more entries into the closed arms, and (d) spent significantly less time in the open arms of the elevated plus maze compared with mice having control sleep conditions (n = 18/group; * P<0.002).

Figure 3

Effects on cyclic AMP response element-binding protein (CREB). (A) Phosphorylated CREB expression (red fluorescence) and NeuN(green fluorescence) in hippocampus of a representative animal exposed to sleep fragmentation (SF) for 15 days and a control (n = 3). Right panel shows merged images. (B) Transcriptional CREB activity in untrained mice and in mice trained in the spatial task water maze (both exposed to SF for 15 days) compared with trained and untrained non-SF controls (P <0.01; n = 5/group). (C) Time course of phosphodiesterase (PDE) 4 gene expression in cortex of mice exposed to SF (P <0.01 for all time points; n = 6).

Figure 4

Effect of 15 days of sleep fragmentation (SF) on energy metabolism in the cortex. (A) ATP levels in cortical tissues from mice with or without SF; there was no significant difference between the groups. (n = 4/group) (B) Representative western blots showing the lack of -activated protein kinase (AMPK)α phosphorylation in the same tissues. E10 epithelial cells exposed to 0.2% O2 for 24 hours were used as a positive control (+) for phospho- (p)AMPK. (C) Homer1a expression remains unaltered over the course of SF (n = 6/time point).

Figure 5

(A) Time course of tumor necrosis factor (TNF)-α, TNF receptors, interleukin (IL)-1, and IL-6 gene expression in cortex of miceexposed to SF (*P <0.01; n = 6). (B) TNF-α protein concentrations in cortex of mice exposed to sleep fragmentation (SF) and controls(*P <0.01). (C) Immunofluorescence photomicrographs in frontal cortex of two representative mice exposed to SF for 15 days and controls(n = 4), showing TNF-α immunoreactivity (red) and NeuN (green). Thee was an intense increase in TNF-α expression in neurons, although thesource of such immunoreactivity might also be derived from other cellular sources (for example, microglia) or from the circulation viablood–brain barrier transport (see text).

Figure 6

Changes in sleep patterns in mice after sleep fragmentation (SF). (A) Changes in mean sleep latency after 15 days of sleep fragmentation in 57BL6/J and tumor necrosis factor receptor (TNFR) knockout mice when compared with corresponding controls (n = 6;P <0.001). (B) Changes in mean slow wave sleep latency during the circadian cycle in mice exposed to SF for 15 days (open circles) and treated with vehicle (grey line) or TNF-α neutralizing antibodies (dotted line) and mice under control sleep conditions. (*P <0.01; n = 7/group). (C) Increases in theta frequency during quiet waking in C57BL/6 J mice, indicating = increased sleepiness after SF. (D) Mean latencies (seconds)and path lengths (cm) to locate the target platform during (a, b) place training, (c, d) reference memory after training, and (e, f) immobility in the forced swim test in C57BL6/J and TNFR knockout mice exposed to SF or control sleep conditions (*P <0.01; n = 12/group). (E) Mean latencies(seconds) and path lengths (cm) to locate the target platform during place training in mice exposed to SF or control sleep conditions, and treated with vehicle or TNF-α neutralizing antibodies. (*P <0.01; n = 12/group).

Figure 7

Putative schematic diagram linking sleep fragmentation (SF) to increased activation of pathways medicated by tumor necrosis factor (TNF-α) and those mediated by NADPH oxidase (see main text and Nairet al[48]for more details).