Downregulation of TREM2 Expression Exacerbates Neuroin ammatory Responses Through TLR4- Mediated MAPK Signaling Pathway in a Transgenic Mouse Model of Alzheimer’s Disease


 Activation of glial cells and neuroinflammation play an important role in the onset and development of Alzheimer’s disease (AD). Triggering receptor expressed on myeloid cells 2 (TREM2) is a microglia-specific receptor in the brain that is involved in regulating neuroinflammation. However, the precise effects of TREM2 on neuroinflammatory responses and its underlying molecular mechanisms in AD have not been studied in detail. Here, we employed a lentiviral-mediated strategy to downregulation of TREM2 expression on microglia in the brain of APPswe/PS1dE9 (APP/PS1) transgenic mice and BV2 cells. Our results showed that TREM2 downregulation significantly aggravated AD-related neuropathology including Aβ accumulation, peri-plaque microgliosis and astrocytosis, as well as neuronal and synapse-associated proteins loss, which was accompanied by a decline in cognitive ability. The further mechanistic study revealed that downregulation of TREM2 expression initiated neuroinflammatory responses through toll-like receptor 4 (TLR4)-mediated mitogen-activated protein kinase (MAPK) signaling pathway and subsequent stimulating the production of pro-inflammatory cytokines in vivo and in vitro. Moreover, blockade of p38, JNK, and ERK1/2 inhibited the release of tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and interleukin-6 (IL-6) induced by Aβ1−42 in TREM2-knocked down BV2 cells. Taken together, these findings indicated that TREM2 might be a potential therapeutic target for AD and other neuroinflammation related diseases.


Introduction
Alzheimer's disease (AD) is the most common age-dependent neurodegenerative disorder characterized by severe memory loss, unusual behavior, and a progressive decline in cognitive function. The classical neuropathological hallmarks of AD include extracellular β-amyloid (Aβ)-associated plaques, intraneuronal tau-associated neuro brillary tangles (NFTs), and loss of neurons and synapses. Increasing evidence demonstrated that neuroin ammation plays a signi cant role in the development and pathology of AD [1][2][3]. The neuroin ammatory cascade is principally initiated through microglia and astrocytes. Microglia, the brain resident innate immune cells, are the critical cell type that contributes to neuroin ammatoryrelated pathology during AD [4]. Microglia activation was found to be related to the interaction between Aβ with several receptors including toll-like receptor 4 (TLR4) in AD patients and animal models [5]. Aβ binding to TLR4 activates the mitogen-activated protein kinase (MAPK) signaling pathway by its downstream molecule myeloid differential protein 88 (MyD88) and TNF-receptor-associated factor 6 (TRAF6) [6,7]. When activated, microglia aggregate around Aβ plaques and produce various proin ammatory cytokines and in ammatory mediators, such as tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), interleukin-6 (IL-6), reactive oxygen species, and nitric oxide [3,8]. Pro-in ammatory cytokines promote amyloid precursor protein (APP) production and the process of APP proteolytic cleavage to increase the production of Aβ, resulting in a vicious cycle in AD [9][10][11].
Triggering receptor expressed on myeloid cells 2 (TREM2) is a type I transmembrane receptor of the immunoglobulin superfamily, which is selectively and highly expressed on microglia in the brain [12,13]. In AD, increased expression of TREM2 has been con rmed in patients and in mouse models of amyloid

Intracerebral injection of lentiviral particles
The lentiviruses containing short hairpin sequences (TREM2 shRNA) (mouse TREM2 gene, NCBI ID: NM_031254.3) and empty vector (as controls) were provided by Genechem Co., Ltd. (Shanghai). TREM2 was expressed under the promoter of fractalkine receptor (CX3CR1) in vivo and in vitro. The titer for lentiviral particles containing TREM2 shRNA (LV-shTREM2) was 8×10 8 transduction units (TU)/mL, and the titer for control lentiviral particles (LV-Con) was 2.5×10 9 TU/mL. 6-month-old male APP/PS1 mice and their age and sex-matched WT mice were randomized into the WT group (n = 10), APP/PS1 group (n = 10), APP/PS1 + LV-Con group (n = 10), and APP/PS1 + LV-shTREM2 group (n = 10). Intracerebral injection of lentiviral particles was performed as previously described [6,24]. Brie y, mice were anesthetized by intraperitoneal injection of sodium pentobarbital and placed on a stereotaxic apparatus (Shenzhen RWD Life Science Co., Ltd. China). The lentiviral particles were delivered into the cerebral cortex (two deposits) and hippocampus (one deposit) of each hemisphere at the speed of 0.5 µl/min using a 28-gauge stainless steel needle. The coordinates from bregma for the cortex: AP − 0.3 mm, ML ± 2 mm, and DV − 1.5 mm; as well as AP − 2 mm, ML ± 1.2 mm, and DV − 1.2 mm. The coordinates from bregma for the hippocampus: AP − 2 mm, ML ± 1.2 mm, and DV − 2 mm. To allow the solution to diffuse into the brain and to prevent its re ux, the needle was left for an additional 5 min after the injection. After the scalp was sutured, mice were kept on a warm pad and returned to their home cages until they awoke. Behavioral tests were performed 2 months after the lentiviral injection (Fig. 1a).

Cell culture
The immortalized murine microglial cell line BV2 retains many of the morphological and functional properties of primary microglia [34]. BV2 cells were grown in DMEM (high glucose) supplemented with 10% FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin. The cells were cultured under a humidi ed atmosphere of 95% air and 5% CO 2 at 37°C.

Preparation of Aβ 1−42 oligomers
The toxic oligomers of Aβ 1−42 were prepared as previously described [6,35]. Brie y, lyophilized Aβ 1−42 peptide was dissolved in HFIP at a concentration of 1 mg/mL, separated into aliquots in sterile microcentrifuge tubes, dried under vacuum, and stored at − 80°C. Immediately before using, the peptide lm was rst dissolved in DMSO at 2 mM and diluted with DMEM and then incubated at 4°C for 24 h to form diffusible oligomers.

Cell transfection and drug treatment
Cell transfection was conducted as previously described [6,24]. Brie y, 1 × 10 4 BV2 cells were seeded on 6-well plates and incubated in a growth medium until 20-30% con uence. Lentiviral particles and 25×HitransG (Genechem Co., Ltd.) were mixed gently and added to the cultures according to the manufacturer's instructions. After 8 h incubation in transfection medium, the supernatant was discarded immediately and the cells were incubated with a fresh medium containing 10% FBS. Each transfection was performed in triplicate and repeated 2 to 3 times.

Novel object recognition (NOR) test
To evaluate the object recognition memory, the NOR test was performed in a square-shaped box (40 cm × 40 cm × 40 cm) for 2 days as described previously [36]. In the acquisition (training) phase, the mice were exposed to two identical objects (A1 and A2) for 5 min, and the time spent exploring each object was recorded. Twenty-four hours later, in the testing (consolidation) phase, one object (A1) was replaced by a novel object (B) of a different material, shape, and color, and the mice were allowed to freely explore the objects for 5 min. Time spent exploring each object was de ned as directing the nose to the object at a distance of maximum 2 cm and exploring it. The recognition index (RI) = (exploring time for novel object B/total exploring time for object B and object A2) × 100%. The box and objects after each trial were thoroughly cleaned with 70% alcohol to minimize odor cues.

Y-maze test
Spontaneous alternation in the Y-maze test, which was used to assess short-term working spatial memory and exploratory activity of mice, was performed as described before [7,35]. The Y-maze apparatus consists of three arms at 120° angles (width 6 cm × length 30 cm × height 15 cm) to each other. Each mouse was initially placed in one arm and allowed to explore the maze freely for 8 min, and the sequence (i.e., ACABC, etc.) and number of arm entries were counted. A successful alternation was de ned as entries into all three arms on consecutive choices (i.e., ABC, BCA, or CAB but not ABA). The spontaneous alternation (%) = [(Number of alternations)/(Total arm entries − 2)] × 100%. The Y-maze arms were cleaned with 70% ethanol between trials to minimize odor cues.

Morris water maze (MWM) test
Spatial learning and reference memory of mice was evaluated by the MWM test as previously described [7,35]. The apparatus consisted of a 120 cm diameter circular plastic pool, lled with opaque water at the depth of 35 cm and controlled temperature of 25 ± 2°C. A white 9 cm diameter transparent platform was located 1 cm below the water surface in the center of one of the four quadrants of the maze. Mice of all groups were given four trials per day for 5 consecutive days in the hidden platform acquisition training. On each training trial, the mice were allowed to swim freely for 90 s to nd the platform. If they failed to locate the platform within 90 s, they were gently guided to the platform and allowed to stay there for 10 s. On the sixth day, mice were subjected to a probe test to evaluate memory retention. During this test, the platform was removed and the mice were allowed to search the maze for 90 s. The escape latency, the number of crossing the platform area, and the time spent in each quadrant were monitored by a video tracking system (Chengdu TME Technology Company, Chengdu, China).

Preparation of brain tissue
After all the behavioral tests, 5 mice per group were randomly selected and deeply anesthetized before transcardiac perfusion with saline followed by a cold 4% paraformaldehyde (PFA). The brains were immediately removed and post xed with 4% PFA for 24 h, cryoprotected with 30% sucrose for 48 h, and cut into coronal sections at a thickness of 30 µm and stored at − 20°C for neuropathological analysis. The cortex and hippocampus of the remaining mice were immediately isolated on an ice glass plate and then stored at − 80°C for subsequent biochemical analysis.

Th-S staining
To detect dense-core amyloid plaques, Th-S staining was performed as previously described [35,36]. Brie y, the free-oating sections were washed with phosphate buffered saline (PBS, pH 7.4) three times, and then were brie y rinsed in distilled water. Next, the sections were incubated in freshly prepared 1% Th-S aqueous solution at room temperature in the dark for 5 min followed by differentiation using 70% ethanol for 1 min. Finally, the sections were covered with aqueous mounting media and examined by the uorescence microscope. Aβ plaques number and Th-S staining area (plaque area/total area selected × 100%) were quanti ed in ve elds of each section (6 sections per mouse) using Image-Pro Plus 6.0 software.

Immunohistochemistry and immuno uorescence
Immunohistochemistry staining was carried out as previously described [35,36]. Brie y, free-oating sections were washed with PBS, pre-treated in 3% hydrogen peroxide, and maintained in blocking solution (1% BSA and 0.3% Triton X-100) for 1 h at room temperature. Subsequently, sections were incubated overnight at 4°C with primary antibodies diluted in blocking solution (anti-Iba1: 1/900; anti-NeuN:1/10000). After washing with PBS, the sections were reacted with secondary antibody at 37°C for 2 h, and visualized using a 3, 3'-diaminobenzidine (DAB) substrate kit. Finally, the sections were mounted with neutral balsam and examined by the light microscopy. For immuno uorescence staining, the sections were incubated overnight at 4°C with primary antibodies diluted in blocking solution (anti-GFAP: 1/5000; anti-Iba1: 1/900; anti-MOAB2:1/2000), followed by incubation with Cy3 or FITC-conjugated secondary antibody at 37°C for 2 h. After washing with PBS, the sections were counterstained with DAPI and visualized by the uorescent microscope. Six random visual elds of the cortex and hippocampus were photographed in each section, and the number of staining cells in each eld was counted by a researcher who was blind to the condition with Image-Pro Plus 6.0 software.

Quantitative real-time reverse transcription PCR (qRT-PCR)
Total RNA was extracted from the cortex, hippocampus, and BV2 cells with RNAiso Plus reagent. An equal amount of total RNA (500 ng) was reverse transcribed into cDNA in a 10 µL reaction volume according to the manufacturer's instructions. qRT-PCR was carried out in 96-well plates on a uorescence thermocycler iQ5 (Bio-Rad) with initial denaturation at 95°C for 30 s followed by 40 cycles of ampli cation (95°C for 10 s, 60°C for 30 s, and 72°C for 30 s). The primer sequences used in the current experiment are as follows: 5'-CCT GAA GAA GCG GAA TGG-3' and 5'-GGA GAC TCT TGA CAC TGG TA-3' for TREM2; 5'-GTC TAC TGA ACT TCG GGG TGA T-3' and 5'-ATG ATC TGA GTG TGA GGG TCT G-3' for TNF-α; 5'-GAA GAG CCC ATC CTC TGT GA-3' and 5'-ATG ATC TGA GTG TGA GGG TCT G -3' for IL-1β; 5'-ACA AAG CCA GAG TCC TTC AGA G-3' and 5'-CAT TGG AAA TTG GGG TAG GA-3' for IL-6; and 5'-ACC ACA CCT TCT ACA ATG AG-3' and 5'-GGT TGG TGA AGT TGG TAG G-3' for β-actin. β-actin was used as an internal control for normalization. Relative gene expression was quanti ed using the 2 −ΔΔCt method. To detect the e ciency of lentiviral-mediated downregulation of TREM2 expression, the PCR products were separated by electrophoresis in 1.5% agarose gels and stained with ethidium bromide.
In brief, cortical and hippocampal tissues of each mouse were homogenized in cold homogenization buffer containing protease inhibitor mixtures. Homogenates were centrifuged at 15,000 rpm for 1h at 4°C, and the supernatants (soluble fraction) were collected. The acquired pellet was incubated with guanidine buffer containing protease inhibitor mixtures for 2 h at room temperature and further centrifuged as above [37]. The supernatants were collected and used as an insoluble fraction. BV2 cells were seeded in 6-well culture plates. After treatment, the supernatants were collected and centrifuged at 8,000 rpm for 30 min at 4°C to remove cell debris. The supernatants were assayed by ELISA for soluble/insoluble Aβ 1−40 and Aβ 1−42 , TNF-α, IL-1β, and IL-6 according to the manufacturer's instructions. Absorbance measurements were read at 450 nm with a microplate reader.

Western blotting
Western blotting was performed as described previously [38]. Brie y, the brain tissues and BV2 cells were homogenized in ice-cold lysis buffer containing protease inhibitor mixtures. After centrifugation, the levels of protein in the supernatant were measured using the BCA protein assay kit. The lysates (20 µg) were boiled and separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred onto NC membranes. Membranes were blocked and incubated overnight at 4°C with primary antibodies including TREM2, Iba1, GFAP, TLR4, MyD88, TRAF6, phospho-p38, total-p38, phospho-JNK, total-JNK, phospho-ERK1/2, total-ERK1/2, Syn, PSD-95, and β-actin. After washing, the membranes were incubated with the appropriate secondary antibodies and visualized with the ECL detection reagents. Western blotting results were analyzed using the Image J software.

Statistical analysis
Data were expressed as mean ± standard deviation (SD). The individual experiment was carried out at least three times. All statistical analyses were performed using the SPSS 13.0 software. Group differences in the escape latency in the MWM test were analyzed using two-way analysis of variance (ANOVA) with repeated measures. The other data were analyzed using one-way ANOVA followed by Tukey's post-hoc analysis. Statistical signi cance was considered at p < 0.05.

Results
TREM2 is upregulated in APP/PS1 mice and is localized mainly on microglia First, double immuno uorescence staining was performed to determine the cellular localization of TREM2 in the brain of 5-month-old APP/PS1 mice. The results showed a good co-localization of TREM2 and microglia marker Iba1 in the brain as we recently reported [24] (Fig. 1b). Afterward, qRT-PCR and western blotting were performed to examine the dynamic changes of TREM2 expression in APP/PS1 mice at 1, 3, 5, and 7 months of age. As revealed in Fig. 1c, TREM2 mRNA levels in the cortex and hippocampus of APP/PS1 mice were signi cantly increased at 3 months, and gradually increased from 3 to 7 months. Simultaneously, TREM2 protein levels in the cortex and hippocampus of APP/PS1 mice showed a similar trend during disease progression (Fig. 1d).
Downregulation of TREM2 expression exacerbated Aβ deposition in APP/PS1 mice To further explore the effects of TREM2 on learning and memory ability and AD-related pathophysiologic changes, a lentiviral-mediated strategy was performed to downregulate TREM2 expression in the brain of APP/PS1 mice. TREM2 mRNA levels in the cortex and hippocampus were successfully reduced by 53.06% and 40.23% respectively, at 2 months after injection of lentiviral particles (Fig. S1a). Similarly, TREM2 protein levels in the cortex and hippocampus were effectively decreased by 46.15% and 35.71% respectively (Fig. S1b).
Amyloid plaques derived from Aβ deposition are one of the most signi cant characteristics in the development of AD. Th-S staining was used to determine the amyloid plaques in the brain. The deposition of Aβ in the brain of APP/PS1 mice was much more severe after downregulation of TREM2 expression (Fig. 2a). The percentages of the number of Aβ plaques (number per mm 2 ) were markedly increased by 50.41% and 55.8% respectively, compared to the APP/PS1 group (Fig. 2b). Meanwhile, the Th-S staining area in the cortex and hippocampus showed a similar trend (Fig. 2c). Aβ levels in the brain tissue which were paralleled with Aβ deposition were further detected by ELISA. As expected, higher levels of soluble Aβ 1−40 and Aβ 1−42 in the cortex and hippocampus were observed in APP/PS1 mice that received TREM2 lentivirus (Fig. 2d, e). Simultaneously, insoluble Aβ 1−40 and Aβ 1−42 in the cortex and hippocampus were also dramatically increased (Fig. 2f, g). Overall, these data indicated that Aβ plaque burden was signi cantly elevated by the downregulation of TREM2 expression in APP/PS1 mice.
Downregulation of TREM2 expression induced microgliosis and astrogliosis in APP/PS1 mice Reactive microglia and astrocytes surrounding Aβ plaques is an early feature of AD pathogenesis, which may trigger neuroin ammation and further amplifying neuronal damage. To explore whether downregulation of TREM2 expression could induce widespread and striking microgliosis and astrogliosis, immunohistochemistry, immuno uorescence, and western blotting were performed. The number of Iba1positive microglia and GFAP-positive astrocytes were signi cantly increased in the brain of APP/PS1 mice that received TREM2 lentivirus (Fig. 3a, b). Furthermore, Iba1 and GFAP protein levels were also markedly elevated in the cortex (Fig. 3c) and hippocampus (Fig. 3d) in these mice. As expected, large numbers of Iba1-positive reactive microglia and GFAP-positive reactive astrocytes were distributed around the more Aβ plaques in both the cortex and hippocampus of APP/PS1 mice after downregulation of TREM2 expression as assessed by double staining of Iba1/Aβ (MOAB2) plaques (Fig. 4) and GFAP/Aβ (MOAB2) plaques (Fig. 5). These data support the notion that TREM2 downregulation activated microglia and astrocytes, which subsequently induced widespread and striking microgliosis and astrogliosis in the brain of APP/PS1 mice.
Downregulation of TREM2 expression promoted proin ammatory cytokines production in APP/PS1 mice Activated microglia and astrocytes stimulated by Aβ may be involved in the acceleration of the AD pathological process by releasing pro-in ammatory cytokines and other toxic mediators, which leads to an increase in Aβ production and accumulation, neuronal damage, as well as cognitive impairment [39][40][41]. To examine the potential role of TREM2 on the production of pro-in ammatory cytokines, we rst investigated the mRNA levels of TNF-α, IL-1β, and IL-6 in the cortex and hippocampus. As shown in Fig. 6a, the mRNA levels of TNF-α, IL-1β, and IL-6 were signi cantly increased in the brain of APP/PS1 mice that received TREM2 lentivirus. We next evaluated the protein levels of TNF-α, IL-1β, and IL-6 by ELISA to further con rm the above results. Similarly, corresponding increments in protein levels of TNF-α, IL-1β, and IL-6 in the cortex and hippocampus were also observed (Fig. 6b). Collectively, these ndings strongly suggested that TREM2 downregulation enhanced the secretion and production of proin ammatory cytokines released by activated glial cells in APP/PS1 mice.
Downregulation of TREM2 expression enhanced neuroin ammatory responses via TLR4-mediated MAPK signaling pathway in APP/PS1 mice After binding to Aβ, TLR4 recruits the pivotal downstream signaling adaptor molecules MyD88 and TRAF6, activating multiple downstream signaling molecules, including the MAPK family. It is well established that MAPK signaling pathway activation is critical for the production of various proin ammatory cytokines and mediators [6,42]. To demonstrate that downregulation of TREM2 expression can aggravate neuroin ammation via TLR4-mediated MAPK signaling pathway. We rst measured the expression of TLR4/MyD88 signaling cascade by western blotting. As revealed by Fig. 7a, b, TLR4, MyD88, and TRAF6 were dramatically increased in the cortex and hippocampus of APP/PS1 mice that received TREM2 lentivirus. Furthermore, we examined the activation of MAPK signaling proteins. Downregulation of TREM2 expression effectively enhanced phospho-p38, JNK, and ERK1/2 in the cortex ( Fig. 7c) and hippocampus (Fig. 7d) of APP/PS1 mice. Therefore, these ndings suggested that TREM2 downregulation regulates TLR4-mediated activation of MAPK signaling pathway.
Downregulation of TREM2 expression promoted neuronal and synaptic loss in APP/PS1 mice The synaptic dysfunction and neuronal loss in the hippocampus and association cortices, is widely considered as the main contributors to cognitive de cits in AD [43][44][45]. Therefore, we explored whether downregulation of TREM2 expression can exacerbate neurodegenerative changes in APP/PS1 mice. First, the number of NeuN-positive neurons was explored by immunohistochemical staining. As revealed by Fig. 8a, the number of NeuN-positive neurons was signi cantly decreased in the cortex and hippocampal cornu ammonis 1 (CA1) region of APP/PS1 mice that received TREM2 lentivirus. Next, the synapticassociated proteins Syn and PSD-95 were examined by western blotting. As anticipated, downregulation of TREM2 expression markedly decreased the levels of pre-synaptic protein Syn and post-synaptic protein PSD-95 in the cortex (Fig. 8b) and hippocampus (Fig. 8c). Collectively, these ndings indicated that TREM2 downregulation increases neuronal and synaptic loss in APP/PS1 mice. Downregulation of TREM2 expression exacerbated learning and memory de cits in APP/PS1 mice To demonstrate downregulation of TREM2 expression could aggravate behavioral de cits in APP/PS1 mice, NOR test, Y-maze test, and MWM test were conducted. The NOR task is gaining popularity for its ability to evaluate the non-spatial memory and the preference for novelty in rodents (Fig. 9a). As shown in Fig. 9b, no signi cant differences were observed in the time exploring two identical objects during the acquisition phase, which suggested that the experimental groups were equally motivated to explore the objects. After 24 h, in the testing phase with two different objects, RI was dramatically decreased in APP/PS1 mice compared to WT mice. Importantly, a more obvious de cit in exploring the novel object was found in APP/PS1 mice that received TREM2 lentivirus (Fig. 9c). After that, we investigated alterations of hippocampal-dependent spatial memory by measuring Y-maze spontaneous alternation. In contrast to WT mice, APP/PS1 mice displayed a reduced percentage of spontaneous alternation.
Interestingly, this decreased spontaneous alternation behavior was effectively exacerbated by downregulation of TREM2 expression (Fig. 9d). Finally, the MWM test was performed to evaluate the ability of spatial learning and long-term memory of mice. In the hidden platform test, all mice gradually learned the location of the platform as training progressed, and the escape latency fell to the shortest on the fth day. As anticipated, a longer escape latency was observed in APP/PS1 mice that received TREM2 lentivirus (Fig. 9e, f). Moreover, APP/PS1 mice that received TREM2 lentivirus showed a fewer number of platform traverses and a lesser percentage of time spent in the target quadrant, implying an obvious spatial memory impairment (Fig. 9g, h).

Knockdown of TREM2 expression increased proin ammatory cytokines production in Aβ 1−42 -induced BV2 cells
To further assess the potential regulatory effects of TREM2 on neuroin ammation induced by oligomeric Aβ 1−42 in microglia, a lentiviral-mediated strategy was used to knock down TREM2 expression in BV2 cells. The expression levels of TREM2 were validated using qRT-PCR at 48 h and western blotting at 72 h after transfection, respectively. As revealed by Fig. S2a, the mRNA levels of TREM2 signi cantly decreased by nearly 54.35% compared with control. Similarly, the protein levels of TREM2 markedly decreased by nearly 27.94% (Fig. S1b).
After transfection for 48 h, BV2 cells were cultured with 5 µM oligomeric Aβ 1−42 for another 24 h, and the expression levels of pro-in ammatory cytokines were examined. As indicated by Fig. 10a, the mRNA levels of TNF-α, IL-1β, and IL-6 were dramatically increased upon Aβ 1−42 stimulation. Interestingly, knockdown of TREM2 expression enhanced the Aβ 1−42 -induced in ammatory responses, as the mRNA levels of TNF-α, IL-1β, and IL-6 were further increased. Similarly, the protein levels of TNF-α, IL-1β, and IL-6 in the media were consistent with the mRNA levels (Fig. 10b).

Blockade of MAPK inhibited Aβ 1−42 -induced proin ammatory cytokines production in TREM2-knocked down BV2 cells
To investigate whether knockdown of TREM2 expression plays an important role in regulating Aβ 1−42induced pro-in ammatory cytokines production via the MAPK signaling pathway, BV2 cells were transfected with lentiviral particles for 48 h, then pretreated with 10 µM SB202190 (p38 speci c inhibitor) ,10 µM SP600125 (JNK speci c inhibitor), or 10 µM U0126 (ERK1/2 speci c inhibitor) for 30 min, followed by 5 µM oligomeric Aβ 1−42 for another 24 h. As shown in Fig. 12a, the mRNA levels of TNF-α, IL-1β, and IL-6 were increased upon knockdown of TREM2 expression. However, the presence of SB202190, SP600125, and U0126 eliminated the increased mRNA levels of TNF-α, IL-1β, and IL-6 in BV2 cells resulted from TREM2 down-regulation. Meanwhile, the protein levels of TNF-α, IL-1β, and IL-6 in the media were also increased upon TREM2 knockdown in BV2 cells. Likewise, these inhibitors ameliorated the higher protein levels of TNF-α, IL-1β, and IL-6 in the media (Fig. 12b). From the above results, we concluded that TREM2 regulates pro-in ammatory cytokines production by the MAPK signaling pathway.

Discussion
In the current study, to explore the effects and the molecular mechanism of TREM2 on Aβ pathology, neuroin ammatory responses, neurodegenerative changes, and behavioral de cits, we used a lentiviralmediated strategy to downregulation of TREM2 expression in the brain of APP/PS1 mice and BV2 cells.
Our results demonstrated that TREM2 was uniquely expressed by microglia, and the expression of TREM2 was signi cantly increased with aging in the cortex and hippocampus of APP/PS1 mice. Furthermore, downregulation of TREM2 expression exacerbated neuropathologies including Aβ deposition, microgliosis, astrogliosis, as well as neuronal and synaptic loss, which was accompanied by a decline in cognitive function in APP/PS1 mice. Moreover, the mechanistic study revealed that downregulation of TREM2 expression increases pro-in ammatory cytokines production via TLR4mediated MAPK signaling pathway in the brain of APP/PS1 mice and BV2 cells. These data coupled with the previous ndings strongly suggested that TREM2 is involved in AD pathogenesis, and it may represent a potential therapeutic target against AD [20,24,29].
Neuroin ammation is a critical pathological feature and considered a major contributor to AD pathogenesis [1]. Activated microglia and astrocytes are both involved in the initialization and progression of neuroin ammation in AD. There is increasing evidence to suggest that neuroin ammatory responses were closely associated with the several AD-related pathological processes, including Aβ accumulation, neuronal and synaptic loss, tau pathology, as well as cognitive de cits [39][40][41]. In the present study, after downregulation of TREM2 expression, the number of Iba1-positive microglia and GFAP-positive astrocytes were markedly increased, and they clustered around Aβ plaques. Furthermore, the protein levels of Iba1 and GFAP were effectively elevated in the cortex and hippocampus of APP/PS1 mice. The current study indicated that TREM2 downregulation leads to more obvious microgliosis and astrogliosis in the brain of APP/PS1 mice, which are closely associated with neuroin ammation.
However, the molecular mechanisms by which TREM2 downregulation aggravated neuroin ammatory responses remains unclear. TLR4, a pattern recognition receptor for Aβ, is highly expressed on microglia and astrocytes surfaces. TLR4 ligation by Aβ can activate multiple downstream signaling pathways, including MAPK. In line with this notion, the expression of TLR4/MAPK signaling pathway molecules was determined. The present study revealed that TLR4, MyD88, and its adapter protein TRAF6 were markedly upregulated in the brain of APP/PS1 mice after downregulation of TREM2 expression. Furthermore, the phosphorylation of p38, JNK, and ERK1/2 was signi cantly increased in the cortex and hippocampus of APP/PS1 mice. In Aβ 1−42 -induced BV2 cells, we found that TREM2 downregulation signi cantly elevated the levels of TLR4, MyD88, and TRAF6. Meanwhile, a signi cant increase of phospho-p38, JNK, and ERK1/2 were also observed after knockdown of TREM2 expression. Taken together with previous ndings, our results suggested that TREM2 downregulation aggravated neuroin ammatory responses through TLR4/MAPK signaling pathway in AD [46].
Mounting evidence suggests that Aβ induces TLR4/MAPK signaling pathway activation, which subsequently stimulates the activated microglia and astrocytes to produce and release pro-in ammatory cytokines and other in ammatory mediators.
TNF-α can stimulate β-and γ-secretase enzyme activity, which results in increased synthesis of Aβ peptides and a further increase in TNF-α release [10,47]. This auto-ampli ed loop in the AD brain can contribute to the maintenance of excessive levels of TNF-α, which then stimulate Aβ synthesis and neuronal loss, also suppressing microglia phagocytosis of Aβ [47,48]. Besides, TNF-α could contribute to promoting insulin resistance and nally lead to cognitive decline in AD [49,50]. IL-1β activates p38 pathway, which could lead to tau hyperphosphorylation and further exacerbate synaptic and neuronal cell dysfunction [9,51]. Similarly to TNF-α, IL-1β can promote Aβ production by modulating γ-secretase enzyme activity in neurons [10]. IL-6 stimulates and promotes the recruitment of microglia and astrocytes to release pro-in ammatory cytokines, and it also promotes β-secretase enzyme activity and tau phosphorylation in AD [11,52,53]. Our results showed that pro-in ammatory cytokines TNF-α, IL-1β, and IL-6 were increased in the cortex and hippocampus of APP/PS1 mice after downregulation of TREM2 expression. Furthermore, knockdown of TREM2 expression also elevated the production of TNF-α, IL-1β, and IL-6 in BV2 cells, similar to the results of in vivo studies. The current ndings were coincident with previous reports, which identi ed that downregulation of TREM2 expression may contribute to the development of neuroin ammation in AD [27,28]. Afterward, we attempted to elucidate whether knockdown of TREM2 expression could regulate Aβ 1−42 -induced neuroin ammatory responses via the MAPK signaling pathway. We treated TREM2-knocked down BV2 cells with the p38 inhibitor SB202190, JNK inhibitor SP600125, and ERK1/2 inhibitor U0126. The current results showed that the higher levels of pro-in ammatory cytokines TNF-α, IL-1β, and IL-6 seen in TREM2 knockdown BV2 cells were signi cantly suppressed by SB202190, SP600125, and U0126. Collectively, these ndings demonstrated that TREM2 regulates pro-in ammatory cytokines production by the MAPK signaling pathway.
Previous studies reported that neuroin ammation is related to the aggregation of Aβ in the brain of AD [54,55]. Abnormal Aβ deposition can stimulate nuclear factor-κB (NF-κB) and MAPK signaling pathways, which are associated with the transcription of pro-in ammatory cytokines and mediators. In turn, proin ammatory cytokines and mediators cause the initiation of APP cleavage through the MAPK and NF-κB signaling pathways [10,56]. Therefore, Aβ can be viewed both as a cause and consequence of neuroin ammation in AD. TREM2 facilitates microglia clustering around Aβ plaques and promotes Aβ phagocytosis and degradation. However, TREM2 deletion decreased the microglial phagocytic clearance of Aβ, resulting in exacerbated brain Aβ deposition and cognitive de cits in the AD animal model [22,57]. In this study, after downregulation of TREM2 expression, more Aβ plaques in the cortex and hippocampus of APP/PS1 mice were observed. Likewise, the protein levels of both soluble/insoluble Aβ 1−40 and Aβ 1−42 were higher in the brain of APP/PS1 mice after TREM2 downregulation. Thus, combined with previous reports, downregulation of TREM2 expression may lead to Aβ aggregation and accumulation in the brain, and further exacerbating spatial cognitive impairment [27, 28, 58].
Both activated microglia and astrocytes may gradually release pro-in ammatory cytokines TNF-α, IL-1β, and IL-6. These cytokines are destructive to neurons by altering synaptic proteins and nally lead to cognitive de cits [59,60]. The synapse-associated proteins, especially pre-synaptic Syn and post-synaptic PSD-95, play important roles in synaptic plasticity and memory formation [61,62]. It has been shown that de cits in Syn and PSD-95 correlate with cognitive impairment in AD and aging [63,64]. Previous studies reported that there is a direct interaction between neuroin ammation and Syn as well as PSD-95 in newborn neurons [65,66]. In this study, TREM2 downregulation remarkably increased neuronal loss in the cortex and hippocampal CA1 region of APP/PS1 mice. Additionally, the levels of Syn and PSD-95 markedly decreased in the cortex and hippocampus of APP/PS1 mice after downregulation of TREM2 expression. These data showed that TREM2 downregulation led to the exacerbation of neurodegenerative changes, as the more severe neuronal and synaptic loss was observed. Cognitive dysfunction is associated with neuronal and synaptic loss in the cortex and hippocampus in AD [27,67,68]. Thus, we assessed the learning and memory capacity of APP/PS1 mice using established behavioral tests. In the NOR test, APP/PS1 mice exhibited a signi cantly lower preference for exploring the novel object in the testing phase following TREM2 downregulation, indicating the recognition memory impairment. Furthermore, short-term working memory impairment was evaluated by the Y-maze test. APP/PS1 mice showed a markedly reduced percentage of spontaneous alteration behavior after downregulation of TREM2 expression. At last, in the hidden platform test, the escape latency was increased in APP/PS1 mice that received TREM2 lentivirus. When removing the hidden platform, APP/PS1 mice had a signi cant decrease in the number of target crossing and the time spent in the target quadrant, suggesting an obvious spatial memory disorder. Our study coupled with the work of others strongly revealed that exacerbation of neuronal and synaptic loss may contribute to worsening cognitive de cits after downregulation of TREM2 expression [22,27,57].
In conclusion, this study revealed that downregulation of TREM2 expression in the brain of APP/PS1 mice markedly exacerbated AD-related neuropathology including Aβ deposition, gliosis, neuroin ammation, as well as neuronal and synaptic loss, which was accompanied by a decline in cognitive function in APP/PS1 mice. Furthermore, in vivo and in vitro evidence demonstrated that TREM2 downregulation increases the production of pro-in ammatory cytokines through TLR4-mediated MAPK signaling pathway activation. Taken together, our results indicated that TREM2 could be a promising treatment to improve cognitive de cits and neuroin ammation in AD and other central nervous system diseases.   blotting and densitometry analysis of Iba1and GFAP in the cortex. d Western blotting and densitometry analysis of Iba1and GFAP in the hippocampus. n = 5. Data were presented as mean ± SD and analyzed by one-way ANOVA followed by Tukey's post-hoc tests. #p < 0.05 or ##p < 0.01 vs. WT group, and *p < 0.05 or **p < 0.01 vs. APP/PS1 group.  Reactive astrocytes surround Aβ plaques in APP/PS1 mice. a Immuno uorescence staining of MOAB2positive Aβ (red), GFAP (green), and DAPI (blue) in the cortex. b Immuno uorescence staining of MOAB2positive Aβ (red), GFAP (green), and DAPI (blue) in the hippocampus. Scale bar = 200 μm. n = 5.

Figure 6
Downregulation of TREM2 expression promoted pro-in ammatory cytokines production in APP/PS1 mice. a The mRNA levels of TNF-α, IL-1β, and IL-6 in the cortex and hippocampus were detected by qRT-PCR. b The protein levels of TNF-α, IL-1β, and IL-6 in the cortex and hippocampus were measured by ELISA. n = 5. Data were presented as mean ± SD and analyzed by one-way ANOVA followed by Tukey's post-hoc tests. ##p < 0.01 vs. WT group, and *p < 0.05 or **p < 0.01 vs. APP/PS1 group.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryFigurecaptions.docx