Background
McGeer and McGeer [
1] suggested the pathological importance of neuroinflammation mediated by microglia rather than amyloid β (Aβ) in the pathogenesis of Alzheimer’s disease (AD), as anti-inflammatory agents have a substantial sparing effect on AD [
2]. Although the neuroinflammation regularly observed in AD has long been incorporated into the amyloid cascade hypothesis of AD, two recent reports have suggested an inflammatory hypothesis for AD. Treatment with the colony-stimulating factor 1 receptor (CSF1R) inhibitors, that block microglial proliferation and rescue inflammatory alterations, improved the performance in memory and behavioral tasks without affecting the accumulation of Aβ in mouse models of AD [
3,
4]. Furthermore, previous studies have provided evidence that infection of the brain with microbes is linked to AD [
5,
6]. Some microbes can remain latent in the nervous system with the potential for reactivation, and neuronal damage caused by direct microbial action and microbe-induced inflammation might occur years after the initial infection. In line with this infection hypothesis of AD, we previously reported that virulence factors of
Porphylomonas gingivalis, a major pathogen in periodontal disease, including lipopolysaccharide (LPS) and gingipains, were able to activate microglia to induce neuroinflammation through the activation of Toll-like receptor 2 and protease-activated receptor-2, respectively, [
7,
8]. These observations suggest that excessive neuroinflammation mediated by microglia is a key driver of AD rather than a result of the disease.
Circadian rhythm disturbances have long been considered a consequence of neurodegeneration associated with AD. AD patients exhibit profound disruptions in their circadian rhythm concerning sleep-wakefulness and other processes. In the post-mortem brain tissue of AD patients, altered synchronization in the rhythms of clock gene expression, including period 1 (
PER1), period 2 (
PER2), and brain and muscle Arnt-like protein-1 (
BMAL1), was noted in different regions of AD patients compared with control subjects [
9]. The clock gene system is known to be capable of influencing the inflammatory process in a number of ways. BMAL1 is the central mediator of the circadian control of the immune system and promotes an anti-inflammatory state [
10]. A recent clinical study showed that the significant deregulation of
BMAL1 in the brain was associated with early AD [
11]. An intrinsic molecular clock also exists in microglia that control diurnal morphological changes in their processes, and these cells regulate the sleep-wake cycle-dependent changes in synaptic strengths [
12‐
14]. Furthermore, infection of
P. gingivalis induced process extension of cortical microglia, and this response was significantly greater during the day than night [
15], suggesting that the intrinsic microglial clock limits the over-reaction and inflammatory responses of microglia during the active phase. However, little is known about the possible contribution of microglial intrinsic molecular clock to the neuroinflammatory response in AD pathology.
Although a number of AD mouse models have been developed based on amyloid precursor protein (APP) overexpression, the overexpression paradigm may cause additional phenotypes unrelated to AD, including the overproduction of soluble N-terminal fragments, C-terminal fragment-α, C-terminal fragment-β, and APP intracellular fragments [
16]. To overcome these drawbacks, two novel AD mouse models (single humanized APP knock-in (KI) mice carrying Swedish (NL), Beyreuther/Iberian (F), or Arctic (G) mutations in different combinations) were generated by knock-in (KI) of a humanized Aβ sequence bearing AD-associated mutations into the mouse APP locus [
17]. These models exhibit unique pathophysiologic properties in the brain. For example, APP-KI
NL-G-F/NL-G-F mice, which bear all three mutations, show aggressive Aβ pathology starting at 2 months of age [
18].
We showed for the first time in the present study that a reduced expression level of BMAL1 was responsible for the increase in the inflammatory phenotype of microglia in APP-KI mice through a reduction in RORα, which in turn reduced IκBα and enhanced NF-κB activation. To determine the pathological roles of REV-ERB in the inflammatory response and learning ability of APP-KI mice, we evaluated the effects of SR9009, a synthetic agonist for REV-ERB, because REV-ERBα nuclear receptor plays a pivotal role in the negative feedback loop regulating the BMAL1 and CLOCK expression. These results suggest that an impaired intrinsic microglial clock system contributes to neuroinflammatory responses and the resultant cognitive impairment in the early stage of AD.
Methods
Animals
Wild-type (WT) and APP-KI mice on a C57BL/6 background were kept in a specific pathogen-free environment at Kyushu University Faculty of Dental Science. Under light-dark conditions, the Zeitgeber time 0 (ZT0) was designated as lights on and ZT12 as lights off. The line of APP-KI mice carried the Arctic mutation, Swedish, and Beyreuther/Iberian mutations. The selection of APP-KI homozygous mice from their littermates obtained by heterozygous coupling was performed by examining the template genomic DNA isolated from tail biopsies, using primers 5′-ATCTCGGAAGTGAAGATG-3′, 5′-ATCTCGGAAGTGAATCTA-3′, 5′-TGTAGA TGAGAACTTAAC-3′, and 5′-CGTATAATGTATGCTATACGAAG-3′. Male mice were used in the whole study. Two-month-old WT and APP-KI mice were administered SR9009 (Millipore) 100 mg/kg (intraperitoneally) for 14 days. All animal experiments were conducted in accordance with the guidelines contained in the Act on Welfare and Management of Animals (Ministry of Environment of Japan) and Regulation of Laboratory Animals (Kyushu University) and under the protocols approved by the Institutional Animal Care and Use Committee review panels at Kyushu University.
Immunoblotting analyses
The brain cortical tissues were collected from APP-KI mice with or without SR9009 injection. The brain lysates were prepared as described previously [
23]. The following primary antibodies were used: rabbit anti-Iba1 (1:1000; WAKO), mouse anti-NOS2 (1:1000; Abcam), mouse anti-IL-1β (1:1000, Santa Cruz Biotechnology), mouse anti-actin (1:5000; Abcam), mouse anti-phospho-IκBα (1:1000, Santa Cruz Biotechnology), rabbit anti-IκBα (1:1000, Santa Cruz Biotechnology), and mouse anti-Aβ (6E10, 1:1000, Covance). The following were used as secondary antibodies: horseradish peroxidase (HRP)-labeled anti-rabbit (1:2000; GE Healthcare) and anti-mouse (1:2000; R&D Systems). The HRP-labeled antibodies were detected using an enhanced chemiluminescence detection system (ECL Kit; GE Healthcare) with an image analyzer (LAS-1000; Fuji Photo Film).
Real-time polymerase chain reaction (PCR)
The mRNA isolated from the isolated microglia of each group at different time points were subjected to a real-time quantitative RT-PCR. The total RNA was extracted with the RNAiso Plus (Takarada, Japan) according to the manufacturer’s instructions. A total of 1000 ng of extracted RNA was reverse transcribed to cDNA using the QuantiTect Reverse Transcription Kit (Qiagen, Japan). After an initial denaturation step at 95 °C for 5 min, temperature cycling was initiated. Each cycle consisted of denaturation at 95 °C for 5 s, annealing at 60 °C for 10 s, and elongation for 30 s. In total, 40 cycles were performed. The cDNA was amplified in duplicate using a Rotor-Gene SYBR Green RT-PCR Kit (Qiagen, Japan) with a Corbett Rotor-Gene RG-3000A Real-Time PCR System. The data were evaluated using the RG-3000A software program (version Rotor-Gene 6.1.93, Corbett). The sequences of primer pairs were as follows:
BMAL1: 5′-CTATCTTCCTCGGACACTGC-3′ and 5′-CTTCTTGCCTCCTGGAGAAG-3′; PER1: 5′-CCAGATTGGTGGAGGTTACTGAGT-3′ and 5′-GCGAGAGTCTTCTTGGAGCAGTAG-3′; PER2: 5′-TTCCACTATGTGACAGCGGAGG-3′ and 5′-CGTATCCAT TCATGTCGGGCTC-3′; REV-ERBα: 5′-CCCTGGACTCCAATAACAACACA-3′ and 5′-GCCATTGGAGCTGTCACTGTAG-3′; TNF-α: 5′-CTGTAGCCCACGTCGTAGC-3′ and 5′-TTGAGATCCATGCCGTTG-3′; IL-1β: 5′-CAACCAACAAGTGATATTCTCCATG-3′ and 5′-GATCCACACTCTCCAGCTGCA -3′; NOS2: 5′-GCCACCAACAATGGCAAC-3′ and 5′-CGTACCGGATGAGCTGTGAATT-3′; IL-6: 5′-TCAATTCCAGAAACCGCT ATGA-3′ and 5′-CACCAGCATCAGTCCCAAGA-3′; RORα: 5′-TTCTAAAAGCAGGCT CGCTAGAG-3′ and 5′-AAGTACACGGTGTTGTTCTGAGAGTC-3′; IκBα: 5′- GAAGCCGCTGACCATGGAA-3′ and 5′-GATCACAGCCAAGTGGAGTGGA-3′; Actin: 5′-AGAGGGAAATCGTGCGTGAC-3′ and 5′-CAATAGTGATGACCTGGCCGT-3′.
For data normalization, an endogenous control (actin) was assessed to control for the cDNA input, and the relative units were calculated by a comparative Ct method. All of the real-time RT-PCR experiments were repeated three times, and the results are presented as the means of the ratios ± standard error of the mean (SEM).
Cell isolation
CD11b
+ microglial cells were isolated from the mouse brain by the magnetic cell sorting (MACS) method as described previously [
19]. Mice in each group were anesthetized and transcardially perfused with phosphate-buffered saline (PBS). The brains were separated as contralateral and ipsilateral samples and cut into small pieces. After enzymatic digestion using a Neural Tissue Dissociation Kit (Papain), the cell suspensions were further mechanically dissociated using a gentle MACS Dissociator (Milteny Biotec, Bergisch Gladbach, Germany). The single-cell suspensions were obtained after application to a 30-mm cell strainer. After magnetic labeling with CD11b Microbeads, the cell suspension was loaded onto a magnetic column placed in the magnetic separator (Milteny Biotec). The MACS column was then rinsed with PBS, and the CD11b-positive fraction was collected.
Cell culture
The mouse microglial cell line MG6 (Riken Cell Bank, RCB2403) was maintained in DMEM containing 10% fetal bovine serum (Gibco) supplemented with 450 mg/ml glucose (Gibco), penicillin-streptomycin (Gibco), 10 μg/ml insulin, and 100 μM β-mercaptoethanol in accordance with the previously described methods [
20,
21]. They were synchronized by treatment with 100 nM dexamethasone (Wako) for 2 h and then stimulated with 10 μM SR9009 and 1 μM oligomeric Aβ (OAβ) at the indicated time points. Non-synchronizing cells were treated with 50 ng/ml LPS (Sigma) or combination with 10 μM SR9009 for 8 h.
CAGE RNA sequencing of microglia from WT and APP-KI mice
ZT2 and ZT14 were 2 h after light turning on and turning off which can typically represent the day time and night time. Therefore, CD11b
+ microglial cells were acutely isolated from WT and APP-KI mouse brain at ZT2 and ZT14 by MACS methods, and then, 5 μg RNA of each group was prepared for sequencing. Cap analysis gene expression (CAGE) library preparation, sequencing, mapping, and gene expression were performed by DNAFORM (Yokohama, Kanagawa, Japan). In brief, the RNA quality was assessed by a Bioanalyzer (Agilent) to ensure that the RNA integrity number (RIN) was over 7.0 and that the A260/A280 and 260/230 ratios were over 1.7. First-strand cDNAs were transcribed to the 5′ end of the capped RNAs and attached to CAGE “bar code” tags, and then, the sequenced CAGE was mapped to the mouse mm9 genomes using the BWA software program (v0.5.9) after discarding ribosomal or non-A/C/G/T base-containing RNAs. For tag clustering, the CAGE-tag 5′ coordinates were input for Reclu clustering, with a maximum irreproducible discovery rate (IDR) of 0.1 and minimum count per million (CPM) value of 0.1 [
22].
Locomotor activity
Fourteen days after SR9009 treatment, locomotor activity was examined (day 15). Mice were removed from their home cages and placed in a novel home cage (clean and without bedding), which provided a floor area of 28 × 18 cm, and then, the locomotor activity of mice of each genotype and each different age group was scored for 3 min. The novel home cage was divided into six identical rectangles and a trained observer determined the incidence of line crossing.
Novel object recognition test
Fourteen days after SR9009 treatment, novel object recognition tests were examined (day 15–day 18). Mice were individually habituated to an open-field box (58 × 42 × 35 cm) by being given 10 min of exploration time in an empty arena for 3 days (habituation session). During the acquisition phase, two objects of the same material were placed in symmetrical positions at the center of the box for 10 min. One hour after the acquisition phase training, one of the objects was replaced with a novel object, and the exploratory behavior was again analyzed for 3 min. After each session, the objects were thoroughly cleaned with 75% ethanol to prevent odor recognition. Exploration of an object was defined as rearing on the object or sniffing it at a distance of < 1 cm, touching it with the nose, or both. Successful recognition of a previously explored object was reflected by preferential exploration of the novel object. Discrimination of spatial novelty was assessed by comparing the difference between the time spent exploration of the novel and familiar objects and the total time spent exploring both objects, which made it possible to adjust for differences in total exploration time.
Y-maze test
Y-maze test was examined prior to Novel object recognition test on the same day (day 15). Testing occurred in a Y-shaped maze with three identical black Plexiglas arms at 120° angles from each other (40 × 10 × 20 cm; Shinfactory, Fukuoka, Japan). After being introduced to the end of one arm, the mice were allowed to freely explore the three arms in a 5-min session. The number of arm entries and the number of triads were recorded in order to calculate the percentage of alternation. The three consecutive choices were defined as an alternation. The percentage of alternations was calculated as (actual alternations/maximum alternations) × 100.
Immunofluorescent staining
Mouse brain samples for immunofluorescent staining were prepared as previously reported [
20,
21]. Sections for staining were incubated with the following primary antibodies: rabbit anti Iba1 (1:2000, Wako) and mouse anti-Aβ (6E10, 1:1000, Covance) at 4 °C overnight. After being washing with PBS, the sections were incubated with donkey anti-rabbit Alexa 488 (1:500; Jackson ImmunoResearch) and donkey anti-mouse Cy3 (1:500; Jackson ImmunoResearch) at 4 °C for 2 h. After further washing, the sections were mounted in Vectashield anti-fading medium (Vector Laboratories). The fluorescence images were observed using a confocal laser scanning microscope (CLSM; C2si, Nikon).
Quantitative morphological analyses of microglia
Confocal Z stack images were captured from the brain of APP-KI mice with or without SR9009 injection. Somata of microglia were quantified after outlining using the ImageJ software program as reported previously [
23]. The morphological analyses of microglia were performed using Z-projections of confocal images. Microglial processes were traced and reconstructed as a single microglia image using the Simple Neurite Tracer program, and the total process length was semi-automatically traced using three-dimensional image data. Single microglia with topological skeletonized images were converted using the skeletonize program.
Statistical analysis
The data are represented as the means ± SEM. The statistical analyses were performed by Student’s
t test or a one- or two-way analysis of variance (ANOVA) with a post hoc Tukey’s test using the GraphPad Prism software package (GraphPad Software). A value of
p < 0.05 was considered to indicate statistical significance. The significance, phase, and amplitude of 24 h rhythms in clock gene expression were evaluated statistically using the non-parametric JTK_Cycle test implemented in R [
24].
Discussion
There is increasing evidence that immune parameters change with the time of day and that disruption of the circadian rhythms is linked to inflammatory pathologies. Previous studies have noted the importance of the circadian clock in age-related neuroinflammatory sensitization [
27,
28]. BMAL1 is the central mediator of the circadian control of the immune system and anti-inflammatory system, as it drives the expression of
RORα, which can increase the levels of IκBα [
10,
29], a major negative regulator of NF-κB [
30]. Furthermore, REV-ERBα is also a transcriptional repressor that inhibits the
BMAL1 expression [
26]. LPS-induced sepsis is less severe when injection at ZT10 than at ZT12, coinciding with a decrease in pro-inflammatory cytokines TNF-α, IL-6, and CXCL1 at ZT12, and increase in the anti-inflammatory cytokine IL-10. This effect is blunted a myeloid-specific
BMAL1−/− mouse, indicating the anti-inflammatory role of BMAL1 in myeloid-lineage cells [
31]. A recent report has demonstrated that activation of the NF-κB pathway in both primary microglia experiments and
REV-ERBα−/− brain, demonstrating that REV-ERBα negatively controls microglial activation and neuroinflammation [
32]. These observations are consistent with the previous observations that the NF-κB-mediated transcriptional repression of the clock feedback limb could cause circadian disruption in response to inflammation [
33,
34]. These results suggest that REV-ERBα/BMAL1 is a regulator of microglial activation and neuroinflammation. However, little is known about the impairment of the intrinsic microglial clock in the early phase of AD. Furthermore, the involvement of an altered circadian clock gene expression in the increased presence of microglia with an inflammatory phenotype in AD remains poorly understood. To address these questions, we explored mRNA expression profiles of circadian clock genes and pro-inflammatory genes in cortical microglia isolated from 2-month-old APP-KI and WT mice using a combination of CAGE sequencing and q-PCR analyses. We first found that the mean mRNA expression of
BMAL1 and
REV-ERBα was significantly lower in APP-KI microglia than in WT microglia. In contrast, the mean mRNA expression of pro-inflammatory genes, including
TNF-α,
IL-1β, and
IL-6, at ZT14 was significantly higher in APP-KI microglia than in WT microglia. It was also noted that the diurnal mRNA expression of these pro-inflammatory genes was negatively associated with that of
BMAL1. In contrast, the diurnal mRNA expression of
IκBα was positively associated with that of
BMAL1. Previous study reported that RORα inhibits NF-κB signaling by inducing IκBα gene expression [
29]. We reported that chronic activation of NF-κB promoted increased expression of pro-inflammatory mediators including IL-1β, TNFα, and iNOS [
21]. On the basis of these previous observations, it is likely to speculate that reduced expression of BMAL1 may promote the NF-κB activation and the subsequent increased expression of pro-inflammatory genes in APP-KI microglia.
A cross-sectional study had been conducted to examine the association between circadian function, aging, and preclinical AD pathology which included 189 cognitively normal participants. After correction for age and sex, the presence of preclinical amyloid plaque pathology, assessed by positive PiB imaging or increasing cerebrospinal fluid was associated with increased intradaily variability, suggesting preclinical AD is associated with rest-activity rhythm fragmentation and circadian dysfunction could contribute to the early pathogenesis of neurodegenerative disease [
35]. In the present study, we used single humanized 2-month-old APP-KI mice. Compared to 6-month- old APP-KI mice, the 2-month-old mice did not exhibit Aβ deposition and activated microglia at the site of Aβ deposition in the hippocampus and cerebral cortex. Therefore, 2-month-old APP-KI mice were reasonable to be used to study the pathogenesis in the early stage of AD.
SR9009, which was developed as a synthetic REV-ERBα agonist, altered the circadian behavior and the circadian pattern of the core clock gene expression in the hypothalamus of mice [
36]. Furthermore, LPS-induced IL-6 expression was attenuated in microglia from BMAL1-deficient mice. This phenotype was recapitulated by pharmacological disruption of oscillatory rhythmicity using SR9009 [
37]. In a previous study, the REV-ERBα agonists, GSK4112, and SR9001 appeared to prevent neuroinflammation and cell death when infused into the brain [
38]. These observations prompted us to examine whether or not the pharmacological activation of REV-ERBα could ameliorate neuroinflammatory response culminating in the disruption of learning and memory. The WT mice showed a response to the novel object and were able to discern a change in the object even after treatment with SR9009. Rather surprisingly, APP-KI mice did not show a response and could not discern a change in the object after treatment with SR9009, suggesting that pharmacological activation of REV-ERBα induced memory disruption in APP-KI mice but not in WT mice. On the other hand, SR9009 has been reported to have REV-ERB-independent proliferation and metabolism [
39]. Therefore, additional experiments using REV-ERB deficient microglia are necessary to examine whether a memory impairment effect of SR9009 is due to an activation of REV-ERB.
It is interesting to speculate that OAβ enhances the effects of SR9009, as OAβ is generated even in the brain of 2-month old APP-KI mice. Furthermore, the APP expression is required for microglial activation, particularly in response to OAβ [
40]. Our observations in the present study showed that treatment with SR9009 alone was able to increase the expression of
IL-1β and
IL-6 but not that of
TNF-α in microglial cells. In contrast, OAβ failed to increase the expression of these pro-inflammatory genes. The combined treatment with SR9009 and OAβ significantly increased the
TNF-α expression and further enhanced the expression of
IL-1β and
IL-6 in microglial cells. The present finding that OAβ significantly suppressed
BMAL1 expression in microglia might suggest the possible involvement of decreased the
BMAL1 expression and the subsequent activation of NF-κB in OAβ-induced neuroinflammation. Additional experiments will be needed to confirm this possibility. In the present study, systemic treatment with SR9009 induced activation of hippocampal microglia, increased expression of inflammatory mediators in the hippocampus, and memory disruption in APP-KI mice. SR9009 alone also increased the expression of IL-1β and IL-6 in MG6 microglia. The combined treatment with OAβ and SR9009 further increased expression of these cytokines. In contrast, Griffin et al. have reported that SR9009 exhibited dose-dependent suppression of LPS + ATP-induced IL-1β protein secretion in BV-2 cells and primary cultured microglia [
32]. Moreover, an application of REV-ERBα agonist GSK4112 suppressed LPS-induced microglia activation through NF-κB pathway resulting in suppressing the expression and secretion of the pro-inflammatory cytokines [
38]. In the present study, however, we could not see anti-inflammatory effects of SR9009. On the contrary, SR9009 induced pro-inflammatory mediators in both cultured microglia and the cerebral cortex of APP-KI mice. This discrepancy may stem from differences in experimental procedures. In our in vitro studies, MG6 microglia were synchronized by treatment with dexamethasone before treatment with SR9009. Dexamethasone was used in the concentration that does not induce anti-inflammatory effects. The inflammatory effect of SR9009 on cultured microglia corresponded well with our in vivo findings.
It has been reported that RORα can exert bi-directional regulation of IL-6 expression dependent on the activation state of astrocytes [
41]. RORα may negatively regulate IL-6 expression through the NF-κB pathway in reactive astrocytes. On the other hand, RORα may trans-activate IL-6 expression by interacting with a RORE in the promoter of non-reactive astrocytes. For these two pathways, RORα competes with REV-ERBα that binds the same response elements with a repressor activity. These observations prompted further examination of an anti-inflammatory effect of SR9009 on the strongly activated microglia, because Griffin et al. used a high concentration of LPS (i.e., 50 ng/mL) to activate microglia [
32]. As shown in (Additional file
2: Figure S1), SR9009 exerted an anti-inflammatory effect on non-synchronizing and strongly activated MG6 microglia treated with LPS (50 ng/mL). Therefore, it is considered that SR9009 activates NF-kB pathway in synchronizing and mildly activated microglia through activation of REV-ERBα, while SR9009 may repress NF-κB in non-synchronizing and strongly activated microglia by interacting with RORE through activation of REV-ERBα. However, the exact action mechanism of SR9009 on microglia is to be elucidated in future studies.
Several recent reports suggest that circadian dysfunction has a causal role in AD-related neurodegeneration [
42‐
44]. However, neuronal circadian dysfunction alone cannot fully explain the mechanistic relationship between circadian dysfunction and neurodegeneration, because neuroinflammation may play a crucial role in AD-related neurodegeneration [
3,
4]. In the present study, we have found that microglia isolated from APP-KI mice exhibited circadian alterations that may be associated with excessive neuroinflammation. Therefore, it may be concluded that circadian disturbances of BMAL1/RORα/NF-κB crosslinking in microglia may contribute to the early stage of AD pathologies through induction of excessive neuroinflammation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.