The deficient CLEC5A ameliorates the behavioral and pathological deficits via the microglial Aβ clearance in Alzheimer’s disease mouse model

The APP transgenic mouse model used in this study develops spatial memory impairment after 4 months of age [37]. The Morris water maze test was performed to examine spatial learning and memory in 6-month-old mice. In a 5-day hidden platform training, the APP mice had significantly longer escape latency than WT mice, indicating a deficiency in memory acquisition. The APP5A^−/− mice exhibited shorter escape latency than APP mice, suggesting that the Clec5a knockout improved the memory acquisition deficit in APP mice. The 5A^−/− mice showed no significant alteration in escape latency compared to WT mice (Fig. 2A).

The probe trial assessed the memory retention of the mice by monitoring their swimming behavior in the pool without the platform for 60 s. The APP mice spent significantly less time in the target quadrant than WT mice, indicating a deficit of memory retention in APP mice. The APP5A^−/− mice spent a significantly longer time in the target quadrant than APP mice, but 5A^−/− mice performed as well as WT mice (Fig. 2B, C). These results suggested that Clec5a knockout in the AD mouse model improves their memory retention deficit at six months of age. There was no difference in swimming speed between the groups, suggesting that Clec5a knockout does not affect motor function (Fig. 2D).

We previously reported that APP transgenic mice are less sensitive to nociceptive stimulation [38]. In addition, aberrant microglial activation has been associated with changes in nociception sensitivity [39–41]. To investigate the role of Clec5a on nociception, we examined the thermal and mechanical sensations of these 4 groups of mice by von Frey filaments and hot plate tests. In the von Frey test, APP mice required higher force to induce paw withdrawal than WT mice (Fig. 3A), which suggested that APP mice had higher mechanical sensation thresholds. Meanwhile, the APP5A^−/− mice had a lower threshold than the APP mice in response to mechanical stimulation (Fig. 3A), indicating that Clec5a knockout attenuates the mechanical nociception in the AD mouse model. In the hot plate test, APP mice required a higher temperature to induce paw withdrawal than WT mice, suggesting an impairment in thermal nociception. However, Clec5a knockout did not significantly alter the thermal-induced paw withdrawal threshold in the WT mice or the APP mice (Fig. 3B). These findings suggested that the absence of Clec5a rescues the deficits of mechanical nociception but not thermal nociception in the AD mouse model.

In addition, we examined whether the Clec5a knockout affects locomotor and anxiety-like behaviors by the open field and elevated plus maze tests. This APP mouse model showed hyperactivity in the open field test and lower anxiety in the elevated plus maze [37]. In the open field test, APP mice traveled longer distances than WT mice, suggesting that APP mice have higher locomotor activities. However, the Clec5a knockout did not alter the locomotor activity in APP mice (Fig. 4A). The APP and WT mice did not show a significant difference in the time spent in the center area, but APP5A^−/− mice spent significantly more time than APP mice in the center area (Fig. 4B). In the elevated plus maze test, the APP mice spent significantly more time traveling in the open arms than WT mice (Fig. 4C), indicating that APP mice have lower anxiety-related behavior. However, APP5A^−/− mice did not significantly alter the open arm time compared with APP mice. In both tests, the performance of 5A^−/− mice was no different from WT mice. Therefore, deleting Clec5a did not alter locomotor activity, but its impact on anxiety-related behaviors requires further investigation.

Aβ deposition is one of the pathological markers of AD. These mice were sacrificed after behavioral tests to examine whether Clec5a knockout alters Aβ levels. ELISA was used to examine the guanidine-soluble Aβ levels in the hippocampus of APP and APP5A^−/− mice. Both total Aβ and Aβ₁₋₄₂ levels in the hippocampus were significantly lower in APP5A^−/− mice than in APP mice (Fig. 5A, B).

To investigate amyloid plaque distribution in the hippocampus, brain slices from APP and APP5A^−/− mice were stained with Thioflavin-S (Thio-S) to detect amyloid plaque with β-sheet conformation (Fig. 5C) and with anti-Aβ antibody to detect total Aβ deposition (Fig. 5D). In the Thio-S staining, the number of amyloid plaques was significantly decreased in the hippocampus (Fig. 5E). Furthermore, the area of total Aβ deposition was measured by anti-Aβ antibody staining (Fig. 5D). The area covered by Aβ in the hippocampus was significantly lower in APP5A^−/− mice than in APP mice (Fig. 5F). In addition, the plaque size showed a downtrend in APP5A^−/− mice (Fig. 5G). These results revealed that the Clec5a knockout reduces the Aβ deposition in the AD mouse model.

Microglia are recruited to amyloid plaques in AD patients and mouse models. These plaque-associated microglia are hyperreactive for immune response and phagocytosis [42]. The plaque-associated microglia, which is defined as microglia within 10 μm from the edge of plaque, were analyzed in the hippocampus of APP and APP5A^−/− mice (Fig. 6A). The ratio of microglia coverage around the plaque was significantly increased in APP5A^−/− mice (Fig. 6B). We analyzed the correlation between the size of the plaque and the microglia coverage around the plaque. The plaque size was negatively correlated with the microglia coverage in the hippocampus (Fig. 6C). The increase in plaque-associated microglia is not due to the increase in the total number of microglia. We found that the density of microglia in the hippocampus had no significant difference among all four genotypes (Fig. S2A, B). Upon Aβ stimulation, microglia usually change their morphology by shortening their processes and enlarging their cell body [43]. The coverage of microglial cells in the hippocampus was significantly higher in APP mice than WT mice, but it was lower in APP5A^−/− mice than in APP mice (Fig. S2C). Moreover, the average size of microglia is significantly smaller in the APP5A^−/− mice than in APP mice (Fig. S2D), suggesting the reduction of activating microglia in Clec5a knockout.

Plaque-associated microglia have direct access to amyloid plaque, which they remove through phagocytosis. Thus, we examined the level of CD68, a marker of phagocytosis, in APP and APP5A^−/− mice. The level of CD68 in the microglia was significantly higher in APP5A^−/− mice than in APP mice (Fig. 6D, E). These results indicated that the Clec5a knockout in APP mice increased the recruitment of microglia to amyloid plaques and implied the increasing phagocytosis of Aβ in microglia.

Multiple inflammatory response pathways might be modulated by CLEC5A. CLEC5A can activate the Syk signaling, which has been implicated in the regulation of Aβ production and neuroinflammation [44, 45]. Using Western blot analysis, the ratio of phosphorylated Syk is significantly lower in APP5A^−/− mice than in APP mice (Fig. 7A, B). Previous studies indicated that CLEC5A regulates the NLRP3 inflammasome activation [35, 46]. Our Western blot analysis indicated that the level of NLRP3 is lower in 5A^−/− mice than in WT mice (Fig. 7C). The expression of NLRP3 is regulated by NF-κB, which could be activated by Syk signaling in macrophages [47, 48]. In the hippocampus, knockout Clec5a significantly reduced the total protein level of NF-κB (Fig. 7E), but no difference in phosphorylated NF-κB level (Fig. 7F). However, we did not detect significant alteration in other potential targets, such as AKT (Fig. S3A), and the cytokines, such as IL-1β, IL-18, IL-6, and IL-10 (Fig. S3B-E) in these mice.

To knockdown Clec5a in vitro, we generated two stable BV2 microglial lines with two different shRNAs (Sh1 and Sh2, Fig. 8A). The reduction of Aβ deposition in APP5A^−/− mice may be due to the enhancement of microglial phagocytosis. To monitor the phagocytosis efficiency, the oligomeric Aβ-stimulated BV2 cells were incubated with the fluorescence-conjugated Aβ for 24 h. The percentage of Aβ-colocalized cells was higher in both Clec5a knockdown cells (Fig. 8B, C). These results suggested that Aβ-induced phagocytosis is increased upon Clec5a knockdown. Similarly, the primary microglia generated from 5A^−/− mice also had higher Aβ phagocytosis ability than from WT mice (Fig. 8D, E). This finding aligns with the decrease of Aβ deposition observed in vivo (Fig. 5C-F).

The blockade of CLEC5A has been reported to decrease Nlrp3 expression after virus infection [35]. To compare with the in vivo finding, we investigated the expression of Nlrp3 upon Clec5a knockdown after 24 h of Aβ treatment. In WT cells, the level of Nlrp3 mRNA was higher in Aβ treated cells than in untreated cells. In both Clec5a knockdown cell lines, the levels of Aβ-induced Nlrp3 mRNA were significantly reduced (Fig. S4A). Next, we examined if Clec5a modulates the Aβ-induced pro-inflammatory response in microglia. The level of pro-inflammatory cytokines TNF-⍺ in the Aβ conditioned media and Il-6 mRNA level was lower in Clec5a knockdown cells than in the WT cells (Fig. S4B, C).

The APP transgenic mouse line J20 (B6.Cg-Zbtb20^{Tg(PDGFB−APPSwInd)20Lms}/2Mmjax) was used in this study as an AD mouse model. This transgenic mouse line expresses a human APP minigene with the Swedish (K670N/M671L) and Indiana (V717F) familial AD mutations. The hemizygous hAPP transgenic mice express high levels of Aβ in the mice brain [74]. The Clec5a knockout mouse, in which exons 3 to 5 of Clec5a were excised, was generated in a previous study [30]. The genotype of these mice was verified by PCR.

The 6-month-old and 10-month-old mice were sacrificed to isolate their hippocampi. The hippocampal tissues were homogenized on the Single cell suspension dissociator (DSC-400; RWD) by enzymatic digestion with cell debris removal kit (DHABE-5003; RWD). The microglia were purified by incubating with CD11b MicroBeads (130-093-634; Miltenyi Biotec) and separated by MACS separator (MiniMACS™ Separator; Miltenyi Biotec).

The Morris water maze was performed to assess the learning and memory in mice. The mice were placed in a rounded pool (122 cm in diameter) with a platform (14 cm in diameter) submerged 1 cm below the water surface. The 5-day hidden platform test consisted of 10 sessions (2 per day). In each session, the mice were allowed to explore the pool in three 60-second trials with 15-minute inter-trial intervals. The escape latency serves as the index for memory acquisition. For each day, the entry points were changed semi-randomly. The probe trial was conducted one day after the last session of hidden platform training by removing the platform and allowing the mice to explore the maze for 1 min. The time spent in the target quadrant, where the platform had previously been located, was used to measure memory retention. The time spent in the target quadrant and the swim speed were recorded and analyzed with the EthoVision video tracking system (Version 3.1, Noldus).

The elevated plus maze consisted of an elevated plus-shape apparatus with two open arms and two closed arms. The mice were placed individually at the center of the maze and allowed to explore the maze freely for 10 min. The time spent on each arm was recorded and calculated using the EthoVision video tracking system.

The open field test was performed using an automated Flex-Field/Open field Photo-beam Activity System (Version 2.0, TRU Scan Photobeam LINC, Coulbourn Instruments), which is a clear plastic open chamber (41 × 41 × 38 cm) with 16 × 16 infrared photo-beam arrays placed 1.5 cm above the bottom of the chamber. Mice were placed in this chamber for 15 min, and the arena’s beam breaks were counted.

Mice were placed on an elevated rack with meshes for 30 min for habituation. Several von Frey filaments (VFF) with different forces (0.07, 0.16, 0.4, 0.6, 1.4, 4, and 6 g; NC12775; North Coast Medical) were applied to the hind paws of these mice using the up-and-down method. The withdrawal responses were determined by the lift of their stimulated paw.

Mice were placed on a thermal apparatus with an analgesia meter (Model PE34; IITC Life Science). The surface of the plate was heated from 30℃ with an increase of 6℃ per minute until the temperature reached 55℃. The temperature was recorded when the withdrawal responses were observed. The withdrawal responses were determined by the lift or lick of their hind paws.

For measuring Aβ levels, the mice hippocampal tissues were homogenized in 5 M guanidine/5 mM Tris buffer (pH 8.0). The samples were diluted in 0.25% casein-blocking buffer containing 0.5 M guanidine and protease inhibitor (04693116001; Roche) before applying to total Aβ and Aβ₁₋₄₂ ELISA kits (27729, 27711; IBL). For measuring the murine IL-1β, the mice hippocampal tissues were homogenized in RIPA buffer (RB4477; Bio Basic) containing protease and phosphatase inhibitors (04693132001, 04906837001; Roche). The homogenates were applied to the IL-1β ELISA kit (432604; BioLegend). The concentration was determined from the standard curve and normalized to tissue weight according to the manufacturer’s protocol.

Frozen mice cortical tissues were homogenized with lysis buffer (150 mM NaCl, 7.6 mM NaH₂PO₄, 32.4 mM Na₂HPO₄, 1% Triton-X) containing protease and phosphatase inhibitors. The levels of murine IL-6, IL-10, and IL-18 of cortical lysates were measured by Multi-Plex immunoassay service in the Inflammation Core Facility, Academia Sinica, Taiwan.

The mice hippocampal tissues were isolated and homogenized with RIPA buffer (RB4477; Bio Basic) containing protease and phosphatase inhibitors. The total protein concentration in tissue homogenates was quantified by BCA assay (23225; Thermo Scientific), and 30 µg of total protein was used in the western blot. The protein samples were mixed with SDS sample buffer and 2-Mercaptoethanol (M6250; Sigma-Aldrich) and boiled to 98℃ for 15 min. The boiled protein samples were then separated via 10% Tris-glycine SDS-polyacrylamide gel electrophoresis and transferred to PVDF membranes (IPVH00010; Merck Millipore). The membranes were probed with rabbit anti-Syk (#13198; Cell Signaling Technology), rabbit anti-p-Syk (AF8404; Affinity Biosciences), mouse anti-NLRP3 (AG-20B-0014-C100; AdipoGen Life Sciences), rabbit anti-NF-κB (#8242; Cell Signaling Technology), rabbit anti-p-NF-κB (#3033; Cell Signaling Technology), rabbit anti-AKT (#9272; Cell Signaling Technology), rabbit anti-p-AKT (#9271; Cell Signaling Technology), and mouse anti-actin (66009-1-Ig; Proteintech Group) antibodies. The membranes were washed and probed with the HRP-conjugated goat anti-mouse IgG and goat anti-rabbit IgG antibodies (115-035-003, 111-035-003; Jackson ImmunoResearch). Protein signals were developed using a chemiluminescent substrate ECL detection system (CCH345; BIO-HELIX) and quantified by using a luminescence imaging system (LAS-4000; Fujifilm).

The brain was fixed by 4% paraformaldehyde (PFA) for 24 h and dehydrated in gradient sucrose buffer (10%, 20%, and 30% in sterile water; 7, 4, and 12 h for each concentration, respectively). The brain is embedded in optimal cutting temperature compounds (Tissue-Tek^® O.C.T. Compound; Sakura) and frozen by liquid nitrogen. The tissues were sectioned coronally in 30 μm slices using a sliding microtome (CM1900; Leica). The brain slices were stored in cryoprotectant medium (30% glycerol, 30% ethylene glycol in PBS) at -20 °C. For IHC, the brain slices were blocked by phosphate-buffered saline (PBS) containing 10% FBS (fetal bovine serum, 35-010-CV; Corning) and 0.5% Triton X-100 (TB0198; Bio Basic). The slices were stained with primary antibodies including anti-Aβ antibody (D54D2, #8243; Cell Signaling Technology or 6E10; 803014; BioLegend), anti-Iba-1 antibody (019-19741; Wako), anti-CLEC5A antibody (FAB1639R; R&D Systems), and anti-CD68 antibody (ab31630; Abcam) for 24 h in 4℃. The CF^® Dye labeled secondary antibodies conjugated with different fluorescents (20303, 20019, 20023; Biotium) were applied to sections for 1 h at room temperature. For the Thioflavin-S staining, the slices were incubated with 0.015% Thioflavin-S (T1892; Sigma-Aldrich) for 20 min at room temperature. After mounting, the brain slices were imaged using a Zeiss fluorescence microscope (Axio Observer A1; Zeiss).

The number of Thioflavin-S⁺ plaque and microglia density in the hippocampal area were counted manually. The coverage of Aβ was defined as the percentage of Aβ area in the hippocampal area. The microglia coverage was defined as the percentage of Iba-1 positive area normalized to the hippocampal area. The microglial CD68 expression was defined as the percentage of CD68 positive area in the hippocampal Iba-1 positive area. The Aβ, Iba-1, CD68, and hippocampal area were measured by ImageJ software (ImageJ; NIH). To analyze the plaque-associated microglia, a circular region encompassing the entire plaque was established. Subsequently, the circle was expanded by a radius of 10 μm. Microglia within this enlarged circle were identified as plaque-associated, and their area was quantified using ImageJ software (ImageJ; NIH).

Mouse microglial BV2 cell line was maintained in RPMI 1640 medium (50-020-PB; Corning) supplemented with 10% fetal bovine serum (FBS; 35-010-CV; Corning) and incubated at 37℃ in 5% CO₂.

The plasmids containing Clec5a shRNAs (Table 1) were purchased from the National RNAi Core Facility, Academia Sinica, Taipei, Taiwan. The plasmid containing Clec5a shRNA was transfected into BV2 cells by Lipofectamine 2000 (11668500; Invitrogen) for 4 h and then recovered in RPMI 1640 medium (50-020-PB; Corning) supplemented with 10% fetal bovine serum (FBS; 35-010-CV; Corning) for 24 h. To select transfected cells, 5 µg/ml puromycin (P8833; Sigma-Aldrich) was applied to the medium for 24 h. Single colonies were picked and diluted in a 96-well plate with RPMI + FBS. The levels of Clec5a in each clone were measured by qPCR. Two stable Clec5a-knockdown cell lines (Sh1, Sh2) were selected for in vitro study.

Cortices from the postnatal day 0–5 mice were dissected in ice-cold Hank’s Balanced Salt Solution media (HBSS; 20-021-CVR; Corning). Tissues were dissociated by Single Cell Suspension Dissociator (DSC-410; RWD Life Science) with High Activity Neonatal Brain Enzymatic Digestion Kit (DHNBE-5002; RWD Life Science), and then the dissociated cells were collected by a cell strainer (130-110-916; Miltenyi Biotec). The cells were centrifuged at 2400×g at room temperature and re-suspended by Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 media (DMEM/F12; 12400024; Gibco) containing 10% FBS and 1% Penicillin-Streptomycin solution (CC502-0100; GeneDireX). The cells were seeded on the 25 cm² Cell Culture Flask (PC272-0200; GeneDireX) and renewed the culture medium every 3 days. After the cells reached the confluency, the cells were treated with 0.25% trypsin solution (CC511-0100; GeneDireX) in DMEM/F12 medium without serum at 37℃ for 10 min to obtain the microglia-enriched cells. Next, the microglia-enriched cells were detached with 1.25% trypsin solution (CC511-0100; GeneDireX) in PBS at 37℃ for 1 h. The microglia cells were centrifuged at 1800 ×g for 10 min and seeded at the number of 2 × 10⁴ cells on poly-D-lysine (P7280; Sigma-Aldrich) pre-coated coverslip.

For preparing the oAβ, HFIP-treated Aβ₁₋₄₂ peptides (A-1163-2; rPeptide Inc.) were dissolved in 10% Dimethyl sulfoxide (DMSO, 97063-136; VWR International) at 100 µM and stored at -80℃. The oAβ was generated by incubating the Aβ stock at 4℃ for 24 h, snap frozen by liquid nitrogen, and storing at -80℃.

For measuring the phagocytosis, the primary microglia cells and BV2 cells were activated by 5 µM oAβ for 24 h and then followed by incubation with 0.2 µM fluorescent-conjugated Aβ_{1− 42} peptide (AS-60479-01; AnaSpec). After 24 h, cells were fixed by 4% PFA. The primary microglia cells were stained with anti-Iba-1 antibody (019-19741; Wako) followed by the CF^® Dye labeled secondary antibody (20019; Biotium). The BV2 cells were stained with fluorescent-conjugated phalloidin (#00045; Biotium). The cells were imaged by the Zeiss fluorescence microscope (Axio Observer A1; Zeiss), and the percentage of phagocytic microglial cells was analyzed by ImageJ software (ImageJ; NIH).

The WT and Clec5a-knockdown BV2 microglial cells were treated with 5 µM oAβ for 24 h. The cultured media were diluted and applied to the TNF-⍺ ELISA kit (430916; BioLegend) according to manufactural instructions. TNF-⍺ concentration was determined from the standard curve and normalized to the total cell number.

The RNA was extracted from the isolated microglia and BV2 cells using GENEzol™ Reagent (GZR100; Geneaid). The purified RNA was reverse-transcribed into cDNA using the High-Capacity RNA-to-cDNA™ Kit (4387406; Applied Biosystems). The cDNA was mixed with qPCRBIO SyGreen Blue Mix Hi-ROX (PB20.16; PCR Biosystems Inc.) and analyzed by a StepOnePlus Real-Time PCR System (Applied Biosystems). Iba-1 was used as the internal control for measuring the Clec5a level in adult microglia, whereas the Gapdh was used as the internal control for measurements of Clec5a, IL-6, and Nlrp3 in BV2 cells. The sequences of primers are listed in Table 2. The data was analyzed using StepOne software version 2.0.

Data are presented as the mean ± s.e.m. Statistical analyses were performed with GraphPad Prism (Version 8.0; GraphPad). Differences among multiple means were assessed by one-way or two-way ANOVA, followed by Tukey’s post-hoc test. Unpaired t-tests analyzed differences between two means and the inter-group comparison of behavior tests. Pearson correlation analysis was used to analyze the correlation between Aβ plaques and microglia coverage, and linear regression was applied to determine the confidence interval of the best-fit line. The threshold for significance was defined as p < 0.05.

The animal study was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of the National Yang Ming Chiao Tung University (Taipei, Taiwan). The behavioral studies were conducted at the Animal Behavioral Core at Brain Research Center, National Yang Ming Chiao Tung University.

Not applicable.

The authors declare no competing interests.

No datasets were generated or analysed during the current study.