Chronic sleep fragmentation shares similar pathogenesis with neurodegenerative diseases: Endosome‐autophagosome‐lysosome pathway dysfunction and microglia‐mediated neuroinflammation

Abstract Aims Insufficient sleep has been found to result in varying degrees of cognitive impairment and emotional changes. Sleep was reported probably responsible for cleaning metabolic wastes in brain by increasing extracellular bulk flow. Herein, we propose that chronic sleep insufficiency in young adult wild‐type mice is also linked with dysfunction of intracellular protein degradation pathways and microglia‐mediated neuroinflammation, which are potentially important mechanisms in the initiation of neurodegeneration. Methods We applied the chronic sleep fragmentation (CSF) model to induce chronic sleep insufficiency in wild‐type mice. After 2 months of CSF, cognitive function, amyloid‐β accumulation, dysfunction of endosome‐autophagosome‐lysosome pathway, and microglia activation were evaluated. Results Following CSF, impairment of spatial learning and memory, and aggravated anxiety‐like behavior in mice were identified by behavioral experiments. Increased intracellular amyloid‐β accumulation was observed in cortex and hippocampus. Mechanistically, CSF could significantly enhance the expression of Rab5 (early endosome marker), Rab7 (late endosome marker), as well as LC3B (autophagosome marker), and autophagy‐positive regulatory factors in brain detected by immunofluorescent staining and Western blot. In addition, activation of microglia was evident by enhanced CD68, CD16/32, and CD206 levels after CSF treatment. Conclusions Chronic sleep fragmentation could initiate pathogenetic processes similar to the early stage of neurodegeneration, including dysfunction of endosome‐autophagosome‐lysosome pathway and microglia‐mediated neuroinflammation. Our findings further strengthen the link between chronic sleep insufficiency and the initiation of neurodegeneration even if lack of genetic predisposition.


| INTRODUC TI ON
Lack of sleep has already become a pervasive trend in many populations in our society, mostly due to electronic devices use, excessive workload, and life stress. People, especially the younger population, are usually ignorant of the importance of sleep and holding inappropriate sleep habits for long. Adequate sleep has been believed partially participate in waste clearance in the brain and memory consolidation. 1,2 Given substantial evidence, chronic sleep insufficiency is associated with cognitive decline and psychological problems ranging from mood changes to psychotic symptoms. 3,4 According to the most recent meta-analysis, individuals with sleep disturbance were 1.55 (95% confidence intervals: 1.25-1.93) and 3.78 (95% confidence intervals: 2.27-6.30) times as likely to have Alzheimer's disease (AD) and preclinical AD than individuals with normal sleep. 5 Clinical observations suggested that neurodegenerative diseases such as AD and Parkinson disease, among others, usually exhibited sleep disturbance in the preclinical and early stages of the disease process. 6 Sleep disruption that was untreated could in turn accelerate neurodegenerative disease progression. 7 Briefly, sleep insufficiency could potentially aggravate neurodegeneration and vice versa.
Alzheimer's disease, the most common type of neurodegenerative diseases, is generally characterized by memory loss, spatial learning disorder, and behavioral changes. 8 Only a small proportion of AD cases are familial, which is caused by gene mutations in the amyloid precursor protein (APP), presenilin 1 (PS1), and presenilin 2 (PS2) leading to increased production of Aβ. 9,10 However, in the far more common sporadic cases, Aβ generation is normal, but its clearance is impaired. 11 It was recently reported that sleep was an important physiological process, during which extracellular metabolic wastes such as β-amyloid protein were cleared via paravascular pathway. 1,12,13 Meanwhile, this paravascular cleaning processes were found to be significantly suppressed by sleep deprivation treatment and in AD mouse models. [14][15][16] Clinical studies have confirmed that neurodegenerative pathogenesis was initiated more than 20 years prior to positive detection of extracellular mis-folded protein deposits and symptoms of clinically evident cognitive decline. 17 The preclinical stage could thus be much more important than late stages for the development of effective interventions for neurodegenerative diseases in clinical practice.
Intracellular and extracellular amyloid-β (Aβ) protein accumulation is a typical pathological change that plays a pivotal role in the evolution of AD. 18,19 In normal conditions, APP, the precursor of Aβ, is internalized into cells by endocytosis and then fused with either the lysosome or autophagosome for lysosomal degradation. 19 The intracellular integrated protein degradation pathway involving endosomes, autophagosomes, and lysosomes is termed the endosome-autophagosome-lysosome (EAL) pathway. 20 EAL pathway dysfunction is characterized by progressive accumulation of autophagic vacuoles (AVs) and enlargement of endosomes, which can also be seen early in AD neuropathogenesis. 20,21 Our recent data reported altered expression of endosome, autophagosome, and lysosome markers in AD transgenic mice, 20 which supported the hypothesis that EAL pathway dysregulation could be a causative mechanism for AD initiation. Furthermore, microglia activation due to accumulation of Aβ has also been believed to participate in the pathogenesis of AD. [22][23][24] Microglia activation could induce neuronal damage by producing proinflammatory cytokines, chemokines, complement proteins, and upon strong activation may release toxicfree radicals. 25,26 Neuroinflammation was proposed closely linked to and may even precede the development of other neuropathological characteristics of AD. 27 Previous study indicated that sleep deprivation treatment accelerates neurodegeneration in APP/PS-mutated mice. 28 However, in the current study, we provided evidence that chronic sleep fragmentation (CSF) could induce similar pathogenesis of neurodegeneration in young adult wild-type mice, involving dysfunction of intracellular protein degradation as well as microglia activation. Our data would highlight the potential risk of initiating neurodegeneration by chronic sleep insufficiency, even if in young adults without genetic predisposition of neurodegenerative diseases.  30 Briefly, we used an orbital rotor (Shiping) with a speed of 55 Hz and a repeated cycle of 10-second on, 110-second off, during light-ON phase (8:00 am-8:00 pm) continuously for 2 months. Prior to the study, we ensured that all mice were able to groom, eat, and drink during orbital rotor movement of the platform, yet would arouse from sleep upon orbital rotor movement. The experimental design procedure is illustrated in order of time ( Figure 1A). No difference was observed in the weight increase in mice between two groups during the first month of CSF ( Figure 1B).

K E Y W O R D S
Alzheimer's disease, amyloid-β, autophagy, lysosome, microglia, sleep fragmentation

| Morris water maze (MWM) test
The Morris water maze test is widely used to assess spatial learning and memory performance in rodents. 31 The apparatus consisted of a circular tank (150 cm in diameter and 40 cm deep) filled with warm water (20-23°C). Four different pictures were hung on the curtain surrounding the tank in four quadrant directions to function as permanent distal cues. The water was made to appear opaque by the addition of powdered milk. A platform (10 cm in diameter) was located in the middle of the southwest quadrant. Mice were subjected to four consecutive trials between 8:00 and 12:00 am each day over a 5-days training period. Each mouse was released from four different positions around the perimeter of the tank (north, northwest, east, and southeast) in four trials in the same order. 32 In each trial, every mouse was allowed to swim until it found the platform (for a maximum of 60 seconds). If the platform was not found in 60 seconds, the mouse was guided to the platform and remained there for 15 seconds. The escape latency to find the hidden platform was automatically recorded using a video tracking system (XR). On the sixth day after 5-training days, a probe test was conducted. The platform was removed while each mouse was released from the northeast quadrant and was allowed to swim for 60 seconds. Memory retention was measured by quantifying the time spent in the target quadrant (southwest) and the number of times the mouse crossed the previous platform location.

| Novel object recognition (NOR) test
Mice were subjected to the novel object recognition test to assess their object recognition and short-time working memory. 33 During the familiar phase, mice were placed in the box (length 30 cm, width 28 cm, height 35 cm) in sequence which contained two copies of objects (A1 and A2 both in red, cylindrical wood), and allowed to explore freely (10 minutes per trial). After a 1-hour delay, the test trial was conducted. The mice were returned to the box in which one of the original objects was replaced by a novel object in green, cubic plastic with similar size. ("novel"), while the other object remained unchanged ("familiar") (5 minutes per trial).
Exploration of the object was defined by sniffing, licking, chewing, or moving vibrissae while directing the nose toward and less than 1 cm from the object. The number of episodes and time spent in exploration of each object by each animal were recorded using a video camera (XR). The discrimination index between the two objects was calculated as DI = (TN − TF)/(TN + TF) (TN = time spent exploring the "novel" object, TF = time spent exploring the "familiar" object). In the familiar phase, the total exploration time (Te) and the exploration time for A1 and A2 objects were also recorded.

| Open field test (OFT)
The open field test was used to assess the locomotor activity and anxiety-like behavior 34 During the tests, each animal was placed into the center of the tank and allowed to explore freely for 5 minutes. The tank was cleaned with 75% ethanol after every test to avoid the leftover effects by previous mice. The total distances and average speed during the 5-minute period were recorded automatically as the index of locomotor activity. The distance to the center zone and inner toroid time was used as exploratory behavior to assess the level of anxiety.

| Forced swimming test (FST)
The forced swimming test was performed to measure the depression-like behavior of the control and CSF groups. 37 Mice were individually forced to swim in an open cylindrical vessel, which contained water 15 cm deep at a temperature of 20-23°C, so that they could not escape or touch the bottom. Each mouse was gently placed in the cylinder, and the total duration of floating was recorded during a 6-minute period. The immobility time was measured during the last 4 minutes of the test. Each mouse was judged to be immobile when it stopped struggling and maintained motionless floating The schematic figure of experimental design procedure. A, The experimental design procedure that indicates the timing of CSF model, behavioral tests, immunofluorescence, electron microscopy, and Western blot. B, Body weight curves of the CSF and control mice during the first month after the CSF model were established in the water, making only those movements necessary to keep its head above water.

| Tissue preparation and immunofluorescent staining
Mice were anesthetized and then perfused with 20 mL of 4% paraformaldehyde. After perfusion, brains were removed, postfixed in

| Electron microscopy
Mice were anesthetized and perfused with 4% paraformaldehyde/2.5% glutaraldehyde buffered with PBS. The prefrontal cortex was removed, cut into 1-mm-thick sections, and then immersed in 2.5% glutaraldehyde overnight at 4°C. The samples were postfixed in 1% OsO4 for 2 hours at 4°C in the dark, block-stained in 1% uranyl acetate for 1 hour, dehydrated in a graded series of aqueous alcohol solutions, and embedded in epoxy resin. Slices with thickness of 70 nm were prepared and mounted using an ultramicrotome (UC7, Leica). Slices were observed using a CM-120 transmission electron microscope (FEI UK Ltd.). Fifteen fields per sample were randomly captured. The number and size of lysosomes between the CSF and control groups were compared.

| Western blot
For Western blotting, the cortex and hippocampus of the mice were quickly removed and homogenized in RIPA lysis buffer with protease inhibitor cocktail (Beyotime). After centrifugation at 10 600 g at 4°C for 15 minutes, the supernatants were collected, and the protein concentration was determined using a BCA Kit (AF2535, R&D system), rabbit anti-iNOS (A0312, Abclonal), and rabbit anti-β-actin (GB11001, Servicebio). Thereafter, the membranes were incubated with the appropriate HRP-conjugated secondary antibodies (Jackson ImmunoResearch) for 1 hour at room temperature. Finally, the blots were visualized using a Bio-Rad ChemiDoc XRS+ imaging system with enhanced chemiluminescence kits (Advansta ECL). The bands were analyzed blindly using ImageJ software to obtain the optical densities (OD) of the signals.
The value was expressed as the ratio of OD of the tested proteins to OD of β-actin.

| Statistical analysis
All measurements were performed by researchers blinded to each group and condition. Data were expressed as the means ± SEM.
Differences between the two groups of mice in the escape latency

| CSF caused cognitive decline and aggravated anxiety
To assess the effects of CSF on learning and memory, we performed behavioral tests of MWM and NOR. The MWM test showed that CSF group had longer escape latency to find the platform through five training days vs the control group. Moreover, in the probe test, CSF mice spent significantly less time in the targeted quadrant and exhibited a fewer number of crossings in the area where the platform was previously located (Figure 2A). These results suggested significant deficits in spatial learning and memory after CSF. In the NOR test, no difference was found in the exploration time for the object F I G U R E 3 Intracellular endosomes and lysosomes in the cortex were dysregulated after CSF. A, Representative confocal images of coronal sections labeled by Rab5, Rab7, and Lamp1 staining in the cortex of CSF and control mice. Scale bar = 40 μm. Local enlarged images were presented in the white squares. Scale bar = 10 μm. B, Representative Western blots showing the changes in the expression of Rab5, Rab7, and Lamp1 in the cortex of CSF and control mice. C, Histograms showed the quantitative density of cells immunoreactive for Rab5, Rab7, and Lamp1. n = 5 per group, *P < .05, **P < .01. D-F, Quantitative analysis of the Rab5, Rab7, and Lamp1 protein levels was performed. Protein expression levels were normalized to the level of β-actin. n = 5 per group, *P < .05, **P < .01, ***P < .001 A1 and the object A2 in the familiar phase between the two groups.
However, in the test phase, the Discrimination Index (DI) of the CSF group was significantly decreased compared with that of the control group ( Figure 2B), which indicates deficits in object recognition.
To further explore the influence of CSF on behavior changes, the OFT and FST were conducted. Interestingly, in the OFT, we found that the CSF group displayed longer total distance moved in the tank, indicating an increase in spontaneous activities. Additionally, CSF mice spent less time in the central zone during the observed 5 minutes (Figure 2C), which demonstrated that sleep disturbance could aggravate anxiety-like behavior. Nevertheless, no evident depression-like behavior was observed in this model, considering nonsignificant difference in the immobility time between two groups subjected to the FST ( Figure 2D).

| CSF significantly increased intracellular βamyloid(Aβ) accumulation
Using immunofluorescence staining, we observed elevated number of Aβ + cells in the cortex and hippocampus of CSF mice compared with control mice ( Figure 2E,F). The detected Aβ was primarily distributed within the cells, while extracellular Aβ deposits were not detected after 2-month CSF.

| CSF caused accumulation of endosomes and lysosomes
It has been found that APP is first internalized by endocytosis and sorted in early endosomes, then delivered to late endosomes, both of which are enriched in APP secretases. 38 The late endosomes fuse with either lysosomes or autophagosomes for Aβ degradation. 19,20 Herein, we measured the expression of the early endosome marker Rab5 and the late endosome marker Rab7, which were both involved in the trafficking, catabolism, and elimination of APP to generate Aβ. [39][40][41] The CSF-induced increase in the numbers of Rab5 + and Rab7 + cells in the cortex was evident by immunofluorescence staining (Figure 3A,C).

Consistently, Western blot results showed similar alterations in Rab5
and Rab7 expression at protein level in this CSF model ( Figure 3B,D,E).
We next evaluated the lysosome marker Lamp1. The density of Lamp1 + cells and its protein expression in the cortex were both enhanced by CSF, which indicates the accumulation and dysregulation of cellular lysosomes ( Figure 3A-C,F). Based on the above observations, we further used the electron microscopy to determine the ultrastructural changes in lysosomes in CSF mice. As shown in these electron microscopy images, we found more abundant and enlarged lysosomes in the CSF cortex vs the control cortex (Figure 4).

| CSF induced disorder of the autophagy process
Autophagy is responsible for degrading aggregated proteins and damaged organelles within autophagosomes or by fusion with the lysosomes. 42 Overactive autophagy and accumulated AVs are characteristics of the early stage of neurodegeneration. 20,21 Our data showed a significant increase in the number of LC3B + (autophagosome marker) cells in the cortex of CSF mice ( Figure 5A,C). In agreement with the immunofluorescence results, the LC3B-II protein level was also elevated after CSF ( Figure 5B,D).
To further explore the changes in autophagy, we examined the expression of two autophagy-positive regulatory factors (Beclin1 and UVRAG) in the cortex by immunofluorescence staining and Western blot. As shown in the figures, the expression of Beclin 1 and UVRAG both increased significantly in the CSF group compared with the control group ( Figure 5A-C,E,F).

| CSF induced microglia activation
Microglia activation with enhanced release of inflammatory mediators is known to exist in AD. 43 Here, we investigated whether microglia activation also participated in the mechanism of CSF-induced cognitive disorders. We doubled-labeled microglia in brain slices by immunostaining for the general microglia marker Iba1 and the phagocytosis marker CD68 which was expressed in activated microglia ( Figure 6A).
Data showed that CSF could significantly increase the number of Iba1 + cells and CD68 + cells in the hippocampus region compared with the control (Figure 6B,C). While in the cortex, Iba1 + cells and CD68 + cells also tended to be higher in CSF mice, but the differences between the two groups were not statistically significant ( Figure S1A-C). we performed double staining for Iba1 with either CD16/32 or CD206 ( Figure 7A). The immunostaining results showed that the numbers of CD16/32 + Iba1 + cells and CD206 + Iba1 + cells were both increased in the hippocampus of CSF mice compared with control mice ( Figure 7B,C). The Western blot for CD16/32, iNOS, and CD206 further confirmed that CSF could activate the microglia with increases in both M1-and M2-type markers in the hippocampus ( Figure 7D,E). In cortex, there also existed elevated CD16/32 expression in CSF mice, but the CD206 level appeared consistent in two groups ( Figure S1D,E).
F I G U R E 5 Intracellular autophagy process in the cortex was disordered after CSF. A, Representative confocal images of coronal sections labeled by LC3B, Beclin 1, and UVRAG staining in the cortex of CSF and control mice. Scale bar = 40 μm. Local enlarged images were presented in the white squares. Scale bar = 10 μm. B, Representative Western blots showing the changes in the expression of LC3B-II, Beclin 1, and UVRAG in the cortex of CSF and control mice. C, Histograms showed the quantitative density of cells immunoreactive for LC3B, Beclin1, and UVRAG. n = 5 per group,*P < .05, **P < .01. D-F, Quantitative analysis of the LC3B-II, Beclin1, and UVRAG protein levels was performed. The protein expression levels were normalized to the level of β-actin. n = 5 per group, *P < .05, **P < .01

| D ISCUSS I ON
In the present study, we found that CSF could induce pathogenic processes similar to those of early-stage AD. In addition to impaired cognitive function, which was evident by behavior tests, we found intracellular Aβ accumulation after 2-month CSF. Aβ, generated by cleavages of its precursor protein APP by β-and γ-secretases, could be distributed in both intracellular and extracellular compartments in normal conditions. 45,46 In pathological conditions, the imbalance in the production, degradation, and clearance of Aβ was the key in the initiation of fibril and plaque formation and neurodegeneration.
It was proposed that intracellular Aβ could be found in early-stage but not late-stage AD seen in human postmorterm autopsies. 46 In another study, 5xFAD transgenic mice begin to develop intracellular Aβ accumulation as early as 2 months after birth, then gradually develop substantial extracellular plaques in the following months. 47 Moreover, intracellular Aβ could be found in neurons, in nonneuronal cells, as well as could be located on the cell membrane. 48 Therefore, the intracellular accumulation of Aβ evident in our study indicated that chronic sleep insufficiency could induce similar pathological changes with early stage or presymptomatic stage of AD. A few studies have shown that enhanced intracellular p-Tau and Aβ accumulation were present after sleep deprivation in AD transgenic mice. 28 However, our study is the first to report that intracellular Aβ can be found in wild-type mice subjected to CSF. This implies a link between long-term sleep insufficiency and cognitive impairment, which is probably related to AD pathogenesis, even without the genetic susceptibility.
In recent years, researchers in the AD field started to pay attention to the processes of trafficking, catabolism, and degradation of APP, which were together referred to as the EAL pathway in AD pathogenesis. In the EAL pathway, APP is processed and metabolized through the endocytosis of endosomes, the turnover of autophagosomes, and then degradation in lysosomes. 19 Dysregulation of the EAL pathway was found in the different pathologic stages in AD transgenic mice, which was characterized by the accumulation of enlarged vacuoles containing APP and its metabolite Aβ. 20 We previously published the evidence for the increased expression levels of endosome, autophagosome, and lysosome markers in APP/ PS1 mice. 20 In the current study, we found a consistent trend in the can also rapidly activate microglia and astrocytes, but nevertheless, sustained microglial activation and unresolved inflammation in the brain are harmful to neurons and synapses and even cause subsequent deterioration of brain structure and function.  58,59 Aβ turnover in the brain follows circadian rhythms and is maintained in a dynamic status. 60 F I G U R E 7 M1 and M2 polarized microglia in the hippocampus were activated after CSF. A, The M1 phenotype of microglia was detected by double staining for CD16/32 (green) and Iba1 (red), while the M2 phenotype was detected by double staining for CD206 (green) and Iba1 (red) in the hippocampus of CSF and control mice. Scale bar = 40 μm. Local enlarged images were presented in the white squares. Scale bar = 10 μm. B-C, Quantitative analysis of Iba1 + CD16/32 + double-positive cells and Iba1 + CD206 + double-positive cells was shown in the histogram. n = 5 per group, **P < .01. D, Representative Western blots of CD16/32, iNOS, and CD206 in the hippocampus of CSF and control mice. E, Statistical analysis of Western blots of CD16/32, iNOS, and CD206 in the CSF and control groups. Protein expression levels were normalized to the level of β-actin. n = 5 per group, *P < .05, **P < .01, ***P < .001 F I G U R E 8 Schematic model of how CSF induced accumulation of endosomes, autophagosomes, and lysosomes, which led to increased intracellular Aβ accumulation. In normal conditions, the trafficking, catabolism, and elimination of APP are via the EAL pathway. First, APP is internalized into cells by endocytosis, then sorted in the early endosomes, and delivered to the late endosomes. APP can be processed by APP secretases to form Aβ in this stage. Then, the late endosomes fuse with either lysosomes or autophagosomes for lysosomal degradation. The balance of formation and clearance of Aβ is maintained by normal EAL pathway. After 2-mo CSF, the EAL pathway is dysregulated and autophagic flux flow is impaired, leading to accumulation of abnormal vesicles in cells. It is evident by increased expression of Rab5 (early endosome marker), Rab7 (late endosome marker), Lamp1 (lysosome marker), as well as LC3B (autophagosome marker), and autophagy-positive regulatory factors Beclin 1 and UVRAG. The accelerated APP processing in EAL pathway results in imbalance of formation and clearance of intracellular Aβ. In pathogenic process of Alzheimer's disease, overloaded intracellular Aβ can be released to extracellular space, potentially forming Aβ plaques eventually via oligomer polyermerization

ACK N OWLED G M ENTS
This work was supported by the National Natural Science Foundation of China (61327902-6 to W. Wang and 81801318 to FF Ding).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.