Sleep deprivation leads to non-adaptive alterations in sleep microarchitecture and amyloid-β accumulation in a murine Alzheimer model

Abstract

Impaired sleep is a common aspect of aging and often precedes the onset of Alzheimer's disease. Here, we compare the effects of sleep deprivation in young wild-type mice and their APP/PS1 littermates, a murine model of Alzheimer's disease. After 7 h of sleep deprivation, both genotypes exhibit an increase in EEG slow-wave activity. However, only the wild-type mice demonstrate an increase in the power of infraslow norepinephrine oscillations, which are characteristic of healthy non-rapid eye movement sleep. Notably, the APP/PS1 mice fail to enhance norepinephrine oscillations 24 h after sleep deprivation, coinciding with an accumulation of cerebral amyloid-β protein. Proteome analysis of cerebrospinal fluid and extracellular fluid further supports these findings by showing altered protein clearance in APP/PS1 mice. We propose that the suppression of infraslow norepinephrine oscillations following sleep deprivation contributes to increased vulnerability to sleep loss and heightens the risk of developing amyloid pathology in early stages of Alzheimer's disease.

Keywords: CP: Neuroscience; EEG; biomarkers; cerebrospinal fluid; extracellular fluid; glymphatic system; microdialysis; neurodegeneration; proteomics; sleep deprivation; ubiquitin pathway.

Figure 1.. Sleep is shorter in young APP/PS1 mice

(A) Scheme illustrating the mouse activity experimental setup. (B and C) Activity tracking of WT and APP/PS1 mice during the active phase (ZT 12–24 and ZT 36–48) and inactive phase (ZT 0–12 and ZT 24–36), indicated in arbitrary units (A.U.) at (B) 3–4 months and (C) 5–6 months of age. Traces indicate mean ± SEM. Mean activity within each phase is shown at the bottom; *p = 0.0158. Orange bar indicates 1-h time window at the end of active phase; **p = 0.0099, *p = 0.0180, n = 6, unpaired t test. (D) Schematic of EEG recording experiments. (E) Representative EEG/EMG trace for one APP/PS1 mouse. (F) Pie chart showing total percentage of time spent awake, in NREM sleep, and in REM sleep across a 48-h duration in the 3–4 months group, n = 4. (G–I) Total percentage of time, duration, and bout assessment of (G) wakefulness, *p = 0.031, **p = 0.0081, and **p = 0.0048; (H) NREM sleep, *p = 0.027, **p = 0.0070, and ns = 0.039; and (I) REM sleep, ns = 0.145, *p = 0.00182, and ns = 0.135, unpaired t test, n = 4. All bar graphs show ±SD.

Figure 2.. Suppression of NE oscillations after sleep deprivation in young APP/PS1 mice

(A) Scheme for fiber photometry experiment with NE biosensor in the hippocampus. (B) Representative trace of fiber photometry NE fluorescent signal during natural sleep and 24 h after sleep deprivation. (C) Power spectral density (PSD) analysis during natural NREM sleep, ns = 0.277, ns = 0.494. Unpaired t test, n = 5. (D) NE peak oscillation analysis during natural NREM sleep (3-h interval), ns = 0.482, **p = 0.006. Unpaired t test, n = 5. (E) PSD analysis at 0 h interval after sleep deprivation in NREM sleep (3-h interval), *p = 0.02, ns = 0.60. Unpaired t test, n = 5. (F) NE peak oscillation analysis at 0 h after sleep deprivation in NREM sleep (3-h interval), ns = 0.19, ns = 0.09. Unpaired t test, n = 5. (G) PSD analysis 24-h after sleep deprivation (3-h interval), NREM, **p = 0.009, ns = 0.583. Unpaired t test, n = 5. (H) NE peak oscillation analysis 24 h after sleep deprivation in NREM (3-h interval), *p = 0.0383, ****p < 0.0001. Unpaired t test, n = 5. (I) Schematic of low-flow microdialysis experiment. (J) Total extracellular NE levels during different brain states, all ns, unpaired t test, n = 4–5. PSD traces are presented as the mean ± SEM. Bar graphs show mean ± SD. A.U., arbitrary units.

Figure 3.. Elevation of Aβ₄₂ isoform 24 h after sleep deprivation in APP/PS1 mice

(A) Experimental setup for brain tissue collection. (B) Preparation of tissue samples. Brain homogenates were either centrifuged for isolation of the soluble fraction (supernatant) or extracted with Triton X, termed as the Triton-soluble fraction. (C) Concentration of Aβ levels in the soluble fraction measured with ELISA using the 4G8 antibody for WT (top) and APP/PS1 mice (bottom). One-way ANOVA with Tukey’s multiple comparisons test, all ns, n = 6–7. (D) Concentration of total Aβ, Aβ₄₀, and Aβ₄₂ isoforms in the Triton-soluble fraction, quantified with ELISA. One-way ANOVA with Tukey’s multiple comparisons test, n = 5–6, *p = 0.046, **p = 0.007. Bar graphs indicate mean ± SD.

Figure 4.. The brain extracellular proteome of WT and APP/PS1 mice is changed 24 h after sleep deprivation

(A) Proteomic workflow: microdialysis sampling was performed to characterize proteome overlaps between the extracellular fluid and the CSF. (B) Coefficient of experimental group-wise extracellular fluid sample variation in the original dataset, n = 5–7. (C) PCA portraying brain state separation, but no genotypic difference, n = 5–7. (D and E) Volcano plots indicating differential protein changes in the (D) WT and (E) APP/PS1 groups between 24 h after sleep deprivation and natural sleep. Differential changes are indicated by red (upregulated) and blue (downregulated) dots with p < 0.05 (Student’s t test with permutation correction) and 2-fold change in log2. Gray dots are not significant, n = 5–7 mice per genotype. (F and G) Total protein levels of (F) Chi3l1 and (G) Ptgds, expressed as label-free quantification (LFQ) values. One-way ANOVA with Tukey’s multiple comparisons test; Chi3l1 (sleep WT vs. 24 h after sleep deprivation WT, *p = 0.0153; sleep WT vs. 24 h after sleep deprivation APP/PS1, *p = 0.0260; sleep APP/PS1 vs. 24 h after sleep deprivation WT, **p = 0.0033; sleep APP/PS1 vs. 24 h after sleep deprivation APP/PS1, **p = 0.0054) and Ptgds (sleep WT vs. sleep APP/PS1, *p = 0.0109; sleep APP/PS1 vs. 24 h after sleep deprivation APP/PS1, *p = 0.0169), n = 5–7 mice. Violin plots show mean ± SD.

Figure 5.. Marked changes in the CSF proteome 24 h after sleep deprivation in young APP/PS1 mice

(A) Proteomic workflow for CSF samples. (B) Coefficient of experimental group-wise variation in the original CSF dataset, showing the median/mean and interquartile range, n = 5–7. (C) Group-wise comparison of all detected proteins, shown as percentage data completeness, n = 5–7. (D) PCA plot reflecting sample stratification across experimental groups. Ellipses around groups are drawn for illustrative purpose only, n = 5–7. (E–H) Volcano plots representing protein changes in (E) the WT group, (F) the APP/PS1 group, (G) during natural sleep, and (H) 24 h after sleep deprivation. Differential changes are indicated by red (upregulated) and blue (downregulated) dots, with p < 0.05 (Student’s t test with permutation correction) and 2-fold changes in log2. Gray dots are not significant, n = 5–7 mice per genotype. Proteins in boxes inside of volcano plot (F and H) overlap with the Higginbotham et al. proteomic dataset. (I) Total protein levels of App, Mapt, and Gfap proteins between genotypes, expressed as label-free quantification (LFQ) values. One-way ANOVA with Tukey’s multiple comparisons test; App (sleep APP/PS1 vs. 24 h after sleep deprivation WT, *p = 0.0304; 24 h after sleep deprivation WT vs. 24 h after sleep deprivation APP/PS1, *p = 0.0122), Mapt (****p < 0.0001), and Gfap (sleep APP/PS1 vs. sleep WT, *p = 0.0363; sleep WT vs. 24 h after sleep deprivation WT, *p = 0.0103; sleep WT vs. 24 h after sleep deprivation APP/PS1, **p = 0.0022); n = 5–7 mice. Violin plots show mean ± SD. (J) Heatmap for LFQ intensities of ANOVA significant proteins, color coded by Z-scored values. Blue tones indicate reduced values, whereas red tones are elevated values.

Figure 6.. Abnormal ubiquitin proteolysis 24 h after sleep deprivation

(A–D) Enrichment analysis of upregulated (red tones) and downregulated (blue tones) protein pathways in (A) WT group between 24 h after sleep deprivation and natural sleep groups, (B) APP/PS1 group 24 h after sleep deprivation and natural sleep, (C) APP/PS1 and WT groups during natural sleep, and (D) APP/PS1 and WT 24 h after sleep deprivation, n = 5–7. Numbers of proteins in each pathway are presented in parentheses. (E) Protein interaction network of the top GO terms (33 proteins), based on the STRING database. Dark red tones indicate highly connected nodes and stronger connections between nodes in the network. The KEGG pathways of interest and their members are color coded. BP, biological process; MF, molecular function; CC, cellular compartment.