Voluntary wheel running exercise improves sleep disorder, circadian rhythm disturbance, and neuropathology in an animal model of Alzheimer's disease

APP^SWE/PS1^dE9 (TG) mice were obtained from Jackson Laboratory (Bar Harbor, United States) [B6; C3‐Tg (APPswe, PSEN1dE9) 85Dbo/Mmjax, No:004462]. The TG mice and their littermate wild‐type (WT) mice (6‐month‐old) were used in this study. The male mice were used in this study and randomly divided into four groups based on their genotypes: WT mice (WT‐Ctrl), WT mice with VWR exercise (WT‐Ex), TG mice (TG‐Ctrl), and TG mice with VWR exercise (TG‐Ex). The mice from four groups were randomly sacrificed at Zeitgeber Time (ZT)1 (9:00) and ZT13 (21:00). The number of animals used in this study is given in each method section.

Male TG mice were hybridized with female WT mice to generate offspring. At 21 days, the genotype of mice was identified by polymerase chain reaction (PCR) assay through tail biopsies. The sequence of the transgene forward primer is 5′AGGACTGACCACTCGACCAG 3′ and the transgene reverse primer is 5′CGGGGGTCTAGTTCTGCAT 3′.

All animals were housed under specific pathogen‐free conditions (temperature, 22 ± 2°C; humidity, 50% ± 10%; air exchange per 20 min; 12 h/12 h light/dark cycle with lights at 8 a.m.). All animal experiments were performed according to the guidelines approved by the Institutional Animal Care Committee of Dalian Medical University.

Animals in WT‐Ex and TG‐Ex groups were housed in a cage with a running wheel (activity wheels, Shanghai Xinruan Information Technology Co, Ltd, China). Each mouse was free to access the running wheel, food, and water. To minimize the stress and depression‐like symptoms by the influence of single housing,15 two mice were housed in one cage. Mice in the WT‐Ctrl and TG‐Ctrl groups were housed in the same cages and conditions without the running wheel. After 2 months, the mice were housed in a single cage to record the circadian behavior and running distance for 3 days.

After 2 months of VWR exercise, all four groups of mice (WT‐Ctrl, WT‐Ex, TG‐Ctrl, and TG‐Ex) were examined for cognitive function by the Morris water maze test and Y‐maze test. The time for conducting behavioral tests was consistent, at 2 p.m. in the afternoon. We analyzed the escape latency, the number of crossing target quadrants, and the percentage of total time spent in the target quadrant in the Morris water maze. We also analyzed the alternation rate in the Y‐maze based on our previous studies, as described.16 There were 7–11 mice in each group.

All four groups of mice (WT‐Ctrl, WT‐Ex, TG‐Ctrl, and TG‐Ex) were arranged for EEG/EMG analysis after the electrodes were implanted. Based on our previous study,17 we analyzed the EEG signals including the time of wake stages, rapid eye movement (REM) sleep, non‐REM (NREM) sleep,18 and the sleep or wake phase shifts in a 24‐h light and dark cycle. There were 3–4 mice in each group.

After 2 months of VWR exercise, we conducted EEG/EMG examination and analysis. Then, all four groups of mice were anesthetized at 8:00–9:00 a.m. with isoflurane and perfused with pre‐cooled phosphate buffered saline (PBS, 0.1 M, pH 7.2). The brains were fixed in 4% paraformaldehyde for 24 h, followed by a dehydration process. After dehydration, the tissues were coated by Optimal Cutting Temperature Compound (Tissue‐Tek, 4583, SAKURA, Torrance, CA, USA) and sliced with the cryostat (CM‐1950S, Leica, Germany). A series of 40‐µm slices was cut coronally from the olfactory bulb to the brainstem. We used a coordination map to localize SCN, hippocampus, and cortex. The sections of SCN, hippocampus, and cortex were washed three times with PBS and incubated with blocking buffer for 1 h at room temperature, and then incubated with the primary antibodies overnight at 4°C, washed with PBS, and incubated with secondary antibodies. Finally, the sections were attached to glass slides for imaging.

We further assessed the immunofluorescence (IF) staining to detect the levels of several core ciradian proteins such as BMAL1, CLOCK, REV‐ERBα, and RORα in the SCN. Phospho‐tau protein was stained with the p‐tau231 antibody, and GABAergic neurons were stained with the vesicular GABA transporter (VGAT) antibody. The neurofilament protein heavy chain (NF‐H) was detected with the NF‐H antibody. To assess neuroinflammation in the hippocampus and cortex, microglia were stained with ionized calcium‐binding adapter molecule 1 (Iba1) antibody, and astrocytes were stained with glial fibrillary acidic protein (GFAP) antibody. IF was used to stain the Aβ plaques by 6E10 antibody. The primary antibodies are listed in Table S1. Finally, the sections were imaged using a laser scanning confocal microscope (A1confocal, Nikon, Japan) under 60× objective lenses. Data were collected from 7–11 sections per animal and 3 mice per group. The analysis of CLOCK, BMAL1, REV‐ERBα, RORα, and p‐tau231 was used to calculate the integrated density (normalized to WT‐Ctrl) by Image J software (Rockville, United States). The analysis of VGAT, NF‐H, Iba1, GFAP and 6E10 was used to calculate the area fraction (normalized to WT‐Ctrl) by Image J software. The details of IF used for this study were referred to in our previous study.19

We then extracted the hypothalamus from four groups of mice, measured the messenger RNA (mRNA) levels of Clock, Bmal1, Revα, Revβ, Rorα, Rorβ, Cry1, and Per1 by RT‐qPCR at ZT1 (9:00) and ZT13 (21:00). The tissue samples were mixed with Trizol and chloroform separately. The mixtures were grinded and centrifuged for 15 minutes. Then the supernatant was removed and mixed with isoproanol for washing. Centerfuged again, the supernatant was discarded and the remain was washed by 75% ethanol twice. After dried, the total RNA was extracted. Then we synthesized total RNA into complementary DNA (cDNA) and performed qPCR following the instructions of Hifair III 1st Strand cDNA Synthesis SuperMix (YEASEN, 11141es60, China) and Taq SYBR Green qPCR Premix (YUGONG, EG20117M, China). The sequence of primers is listed in Table S2. The relative expression levels were calculated using the 2‐ΔΔCt method. There were 3–4 mice in each group.

We performed western blotting to determine the protein levels of CLOCK and BMAL1, APP, p‐tau231, BACE1 and GSK3β, in the hypothalamus extracts obtained from four groups of mice (WT‐Ctrl, WT‐Ex, TG‐Ctrl, and TG‐Ex). The primary antibodies are listed in Table S1. The results were obtained from five mice for each group, and the experiments were repeated three times. The details of western blotting methods used for this study were referred on our previous study.19

The hypothalamus tissue from four groups of mice was dissected rapidly on ice at ZT1 (9:00) and sonicated in lysis buffer (containing 50 mM Tris pH 7.4, 150 mM sodium chloride, 1% triton X‐100, and 1% sodium deoxycholate), after grinding and centrifugation, we collected the supernatant to determine human Aβ42 and Aβ40 levels by Human Aβ42 ultrasensitive ELISA Kit (KHB3544, Thermo Fisher Scientific, Waltham, MA, USA) and Human Aβ40 ELISA Kit (KHB3481, Thermo Fisher Scientific, Waltham, MA, USA). There were four mice in each group.

All data were assessed using GraphPad Prism 9.0 software (GraphPad Software Inc, USA), and statistical data were expressed as means ± standard error of the mean (SEM). The data from behavioral analyses, EEG, and biochemical results were analyzed using two‐way analysis of variance (ANOVA) followed by Sidak's multiple comparisons tests. p < 0.05 was considered statistically significant. All experiments were repeated at least three times.

TG‐Ctrl mice showed higher activities than the WT‐Ctrl mice during the 12‐h light stage (Figure 1B); the running distance was significantly increased by 123% in TG‐Ctrl mice (p < 0.05; Figure 1C). After 2 months of VWR exercise, the running distance was markedly decreased by 64.9% in the 12‐h light stage in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.05; Figure 1C). There was no significant difference between the four groups in the running distance during the 12‐h dark stage and the mean distance per day. As mice primarily have nocturnal activity and sleep during the daytime, the VWR exercise reduced diurnal activity and improved the behavioral circadian rhythm disorder in the TG‐Ex mice.

To further investigate sleep structure change we recorded day–night EEG in all four groups of mice. We found that the percentage of REM sleep in the TG‐Ctrl mice was significantly decreased by 50.6%; the NREM sleep was decreased by 49.2%; and wakefulness was increased by 43.2% during the 12‐h light stage versus the WT‐Ctrl mice (both p < 0.01; Figure 2B). After VWR exercise, the percentage of REM sleep was significantly increased by 89% (p < 0.05), the NREM sleep was increased by 119%, and the wake stages were decreased by 36.8% during the 12‐hour light stage in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.01; Figure 2B).

Furthermore, we documented that the number of sleep phase shifts (NREM to REM, REM to NREM) was increased by 109%, and wake phase shifts (NREM to wake, REM to wake) was increased by 107% during the 12‐h light stage in the TG‐Ctrl mice versus the WT‐Ctrl mice (both p < 0.05; Figure 2C). After the VWR exercise, the sleep phase shifts were decreased by 47.7% (p < 0.05; Figure 2C), and the wake phase shifts were decreased by 47.1% (p < 0.01; Figure 2C) during the 12‐h light stage in the TG‐Ex mice versus the TG‐Ctrl mice. These findings suggest that VWR exercise improved the disturbance in sleep structure and decreased sleep fragmentation, especially during the light stage.

To evaluate the effect of VWR exercise on the cognitive function in the TG mice, we conducted the Morris water maze and Y‐maze tests. They showed that TG‐Ex mice displayed a significant decrease in escape latency on day 4 (by 44%, p < 0.05; Figure 3A), a significant increase in the number of platform crossings (by 147%, p < 0.01; Figure 3A‐B), and total time in the target quadrant (by 59%, p < 0.05; Figure 3A‐B) compared with TG‐Ctrl mice. Consistently, the Y‐maze alternation percentage was increased markedly in the TG‐Ex mice by 46.5% versus the TG‐Ctrl mice (p < 0.05; Figure 3B). There was no significant difference between the WT‐Ctrl and the WT‐Ex mice in the Morris water maze and Y‐maze tests. Together, our results reveal that VWR exercise can improve the learning and memory abilities in the TG mice.

SCN in the hypothalamus is the master circadian pacemaker that maintains circadian outputs. We focused on the core circadian clock genes in the SCN and observed an increased expression of BMAL1 by 46.9% (p < 0.01; Figure 4A‐B) in the TG‐Ctrl mice versus the WT‐Ctrl mice. After the VWR exercise, the expression of BMAL1 was decreased by 45.7% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.001; Figure 4A‐B). Furthermore, we measured the expression of CLOCK in the SCN of four groups and found no significant difference.

In addition, we measured the expressions of REV‐ERBα and RORα, which regulate the BMAL1. We found a 54.7% reduction in the expression of REV‐ERBα in the TG‐Ctrl mice versus the WT mice in the SCN (p < 0.001; Figure 4C–D). After VWR exercise, the expression of REV‐ERBα was increased by 119% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.001; Figure 4C‐D). Meanwhile, the level of RORα was significantly increased by 116% in the TG‐Ctrl mice versus the WT‐Ctrl mice. After VWR exercise, it was decreased by 36.4% in the TG‐Ex mice versus the WT‐Ex mice (both p < 0.001; Figure 4C‐D). These results indicate that VWR exercise can change the expression levels of the core circadian clock genes.

Hyperphosphorylated tau protein aggregated in the cytoplasm is a typical hallmark of AD. Because most neurons in the SCN are GABAergic,20 we further elucidate the pathological changes in the SCN. A significant increase in the integrated density of intracellular p‐tau231 was found in the SCN of TG‐Ctrl mice versus WT‐Ctrl mice (by 129%, p < 0.001; Figure 5A,B). Meanwhile, the VGAT, a marker of GABAergic neurons, was decreased by 94.5% in the TG‐Ctrl mice versus the WT‐Ctrl mice (p < 0.001; Figure 4A,B). Double‐staining showed that some p‐tau231 in the cytoplasm was co‐localized with GABAergic neurons in the TG mice (Figure 5A). After VWR exercise, the integrated density of p‐tau231 was markedly reduced by 35% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.05; Figure 5A,B). It is important to note that the VGAT was increased by 38.7% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.05; Figure 5A,B).

NF‐H is one of the important components of cytoskeletal proteins in the large myelinated axons, which is often released into the extracellular fluid due to axonal damage in AD.21 We found that the integrated density of NF‐H was reduced by 140% in the SCN of TG‐Ctrl mice versus WT‐Ctrl (p < 0.001; Figure 5C,D). After VWR exercise, the integrated density was elevated by 74.5% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.05; Figure 5C,D). Furthermore, NF‐H was co‐localized with VGAT in the SCN, indicating that VWR exercise reduces the degeneration of GABAergic neurons in the TG mice.

Because the expression of core clock genes changed in SCN, we further examined the mRNA level of clock genes in the hypothalamus at two time points: ZT1 (9:00) and ZT13 (21:00). We found that the abnormal expression of clock genes mainly occurred during the light stage in the TG mice, and VWR exercise significantly alleviated such changes (Figure 6A).

At ZT1, we found an increased expression of Bmal1 and Rorα (by 180% and 169%, both p < 0.05; Figure 6A) in the TG‐Ctrl mice versus the WT‐Ctrl mice. After the VWR exercise, the expression decreased by 57% and 68% in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.05; Figure 6A). In addition, the expressions of Rev‐erbα were reduced by 77% (p < 0.05) in the TG‐Ctrl mice versus the WT mice, and elevated by 79% after VWR exercise in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.01; Figure 6A). There were no significant changes in the expression of other clock genes (Clock, Rev‐erbβ, Rorβ, Cry1, Per1). This tendency in the hypothalamus was similar to that in the SCN. At ZT13, only the mRNA level of Cry1 declined by 65% in the TG‐Ctrl mice versus the WT‐Ctrl mice (p < 0.05; Figure 6A), and there was no significant change after VWR exercise (Figure 6A).

Furthermore, we found that the expression of p‐tau231 in the hypothalamus was reduced by 25.7% (p < 0.05; Figure 6B‐C), and the level of APP protein was reduced by 32.3% (p < 0.01; Figure 6B‐C) in the TG‐Ex mice versus the TG‐Ctrl mice. In addition, we measured the level of Aβ40 and Aβ42 in the hypothalamus by ELISA. We found a significant reduction of Aβ40 (by46.8%), Aβ42 (by 62.5%), and Aβ42/Aβ40 ratio (by 28.8%) in the TG‐Ex mice versus the TG‐Ctrl mice (all of them p < 0.05; Figure 6D).

BACE1 is a rate‐limited enzyme for Aβ production and plays a critical role in AD pathology.22 Meanwhile, we found that the expression of BACE1 was decreased by 26.7% in the TG‐Ex mice versus the TG‐Ctrl mice (p < 0.05; Figure 6B‐C). The GSK3β promotes tau hyperphosphorylation and induces Aβ formation.23 We also found that the expression of GSK3β was lower by 26.5% in the TG‐Ex mice compared to the TG‐Ctrl mice (p < 0.01; Figure 6B‐C), suggesting that VWR exercise may reduce the pathological accumulation via suppression of BACE1 and GSK3β in the hypothalamus of TG mice.

Next, we measured the p‐tau231 protein by IF staining and found that the integrated density of p‐tau231 was increased significantly in the hippocampus (by 259%, p < 0.0001) and cortex (by 419%, p < 0.0001) in the TG‐Ctrl mice versus WT‐Ctrl mice (Figure 7A‐B). In addition, we documented a marked decrease of p‐tau231 IF staining after VWR exercise in the hippocampus (by 57.1%) and cortex (by 51.8%) in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.001; Figure 7A‐B). These results were consistent with the change of p‐tau231 in the SCN.

In our study, we documented that VWR exercise improved cognitive impairment in the TG mice. Then we further assessed the neuropathological changes of the hippocampus and cortex of TG mice. TG‐Ctrl mice showed abundant Aβ plaque in the hippocampus and cortex (Figure 8A‐B). After VWR exercise, it was decreased significantly in the hippocampus (by 60.4%) and cortex (by 57%) in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.001, Figure 8C). It is well known that neuroinflammation plays an important role in the pathogenesis of AD.24 We then examined the levels of microglia with the Iba1 immunofluorescent staining and astrocytes with the GFAP immunofluorescent staining. We found that microglia and astrocytes were concentrated around the Aβ plaques. The area fraction of Iba1 was higher in the hippocampus (by 230%, p < 0.0001) and cortex (by 180%, p < 0.0001) of TG‐Ctrl mice versus WT‐Ctrl mice (Figure 8A‐C). Similarly, the area fraction of GFAP was elevated in the hippocampus (by 318%) and cortex (by 294%) in the TG‐Ctrl mice versus WT‐Ctrl mice (both p < 0.0001; Figure 8B‐D).

After the VWR exercise, we found that the area fraction (%) of Iba1 was markedly downregulated in the hippocampus (by 27.9%) and cortex (by 28.6%) in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.001, Figure 8A‐C). In addition, the area fraction (%) of GFAP was reduced in the hippocampus (by 47.2%) and cortex (by 66.7%) in the TG‐Ex mice versus the TG‐Ctrl mice (both p < 0.001, Figure 8B‐D).