Tau induces inflammasome activation and microgliosis through acetylating NLRP3

The primary antibodies used in this study are all listed in Table. The mouse immunoglobulin G (IgG) and rabbit IgG were purchased from Sigma. The fluorescein and horseradish peroxidase (HRP) labelled secondary antibodies were obtained from Li‐Cor Biosciences and Proteintech Group, respectively. The Alexa Fluor 647‐, 594‐ and 488‐conjugated secondary antibodies were from Thermo Fisher Scientific. The enhanced chemiluminescent substrate was purchased from Biosharp. Neofect DNA transfection reagent was from Neofect Biological Technology Co. Ltd. TPOP146, lipopolysaccharides (LPS) and nigericin were purchased from MedChemExpress. L‐Moses hydrochloride (L‐45) was from GlpBio. The protein marker, dulbecco's modified eagle medium (DMEM), high glucose, DMEM/F12, foetal bovine serum (FBS) and Lipofectamine 3000 were from Invitrogen and Gibco (Thermo Fisher Scientific). Protein A/G agarose was from Smart Lifesciences. All other chemical reagents were standard commercial products with analytical reagent grade. S1

Human brain samples were procured from the China Brain Bank (Zhejiang University School of Medicine). More detailed information is provided in Figure S1C. C57BL/6 male mice (WT) were obtained from the Experimental Animal Center of Tongji Medical College, Huazhong University of Science and Technology. Fifty‐six male C57BL/6 mice aged 12 weeks were randomly divided into four groups for different Tau protein overexpression using adeno‐associated virus (AAV) injection: vehicle (empty vector AAV, n = 14), Tau441 (n = 15), TauS262E (n = 15) and TauK18– (n = 12). Three months after virus injection, the behaviours of the mice were tested, which lasted for 2 weeks, followed by positron emission tomography/computed tomography (PET/CT) imaging. Finally, the mice were sacrificed for subsequent molecular biochemical assays and RNA‐Seq of the brain tissues (Figure 4A). In addition, 16 male C57BL/6 mice aged 12 weeks were randomly divided into two groups for subsequent TNB peptide intervention experiment: Tau441 (AAV‐Tau441, n = 8) and Tau441 + TNB (n = 8). The detailed TNB peptide injection method and other tests can be found in Section 2.13 and Figure 8A. Tau P301S (PS19, strain #: 008169) transgenic mice with age of 3 or 6 months were from Jackson Lab. These mice harbour the T34 isoform and four microtubule‐binding repeats (1N4R) of the Tau protein with P301S mutation under the regulatory control of the murine prion promoter. 3xTg‐AD mice (9‐month old) carrying human mutated APP, PS1 and Tau genes were also from Jackson Lab. During the experiments, all animals were maintained under appropriate temperature (22 ± 1°C), humidity (55 ± 15%) and a 12 h light/12 h dark cycle. All animal experiments were approved by the Animal Care and Use Committee of Huazhong University of Science and Technology and performed in compliance with the NIH Guide for the Care and Use of Laboratory Animals.

The plasmid pFLAG‐Tau‐2N4R (Tau441), encoding human Tau, was from our laboratory. The plasmid TauK18– (or TauK18△) lacking microtubule‐binding repeats domain (243–372) was generated by PCR and cloned in a myc‐tagged pcDNA3.0 vector/EGFP N1 vector in EcoRI and BamHI restriction sites. The plasmids GV362‐EGFP‐FLAG‐mouse NLRP3, ‐mouse ASC and ‐mouse Casp1 (fusion with Flag, CMV‐MCS‐3FLAG‐IRES‐EGFP‐SV40‐Neomycin) were provided by Genechem. We also constructed the plasmid pcDNA3.0‐EGFP‐mouse NLRP3 (fusion with EGFP). Mutations of NLRP3 lysine‐21, lysine‐22, lysine‐24 into arginine (K21R, K22R and K24R) and Tau threonine‐181, serine‐199, threonine‐217 and serine‐262 into glutamic acid (T181E, S199E, T217E and S262E) were carried out using the Mut Express II Fast Mutagenesis Kit by following the manufacturer's instructions (Vazyme Biotech). The plasmid SIRT2 was obtained from Addgene (Plasmid #13813).

HEK‐293T, N2a, SH‐SY5Y and BV‐2 microglial cells were cultured in DMEM high glucose medium or DMEM/F12 medium. All cell lines were supplemented with 10% FBS and 1% penicillin/streptomycin, and maintained at 37°C in a humidified atmosphere with 5% CO₂ and 95% air. The cells were seeded into six‐well plates or 96‐well plates. When the confluence reached to 70%–80%, we added the fresh medium into the cells and then transfected them with relevant plasmids for 24–48 h by Neofect DNA transfection reagent/Lipofectamine 3000 according to the manufacturer's protocols. In addition, primary astrocytes and primary neurons were successfully obtained and cultured according to the previous studies.47, 48

Briefly, HEK‐293T cells were seeded in a 96‐well plate and cultured overnight (80% confluence, 100 µL). After different Tau plasmids transfection (Tau441, TauT181E, TauS199E, TauT217E, TauS262E and TauK18–) for 48 h, the cell viability was assessed by the cell counting kit‐8 (CCK8) assay respectively. CCK8 assay was also used to evaluate the effect of the TNB on the cells including BV‐2, N2a, SH‐SY5Y, primary astrocytes and primary neurons. The peptide TNB/FITC‐TNB with purity of 95.93% (2574.89 Da, LEDLEDVDLKKFKMHLEDYPP) was chemically synthesised by Bioyeargene Biosciences using conventional Fmoc solid‐phase synthetic strategy. The cells were plated and grown to 80% confluence prior to treatment with the peptide with different dosages (0, 25, 50, 100, 200 and 400 µM). After 48 h, the cells were exposed to 10 µL CCK8 reagent for another 4 h. The absorbance at 450 nm was determined using a microplate reader (BioTek SynergyH1). The data from three parallel wells were calculated.

IL‐1β levels in supernatant of BV‐2 cells after corresponding treatments were quantitatively measured through an enzyme‐linked immunosorbent assay (ELISA) using the mouse IL‐1β ELISA kit (Beijing 4A Biotech Co., Ltd.). All detection steps and calculation methods were carried out according to the instructions of the kit.

Based on the previous relevant methods,49, 50 we reconstituted the mouse inflammasome system in HEK‐293T cells without endogenous NLRP3, ASC and Caspase‐1, aiming to detect Tau‐induced NLRP3 inflammasome activation. This biochemical assay separated ‘priming’ step (S1) from NLRP3 ‘activation’ step (S2) (Figure 2A). In brief, we set up HEK‐293T cells transiently expressing either NLRP3 (293T NLRP3, ‘activator’ or ‘donor’) or ASC and Caspase‐1 (293T ASC‐Casp1, ‘recipient’). It was worth mentioning that excess ASC promotes self‐aggregation and induces NLRP3‐independent Caspase‐1 activation. Thus, the amount and proportion of the plasmids, especially ASC, should be strictly controlled as the following: NLRP3 (2000 ng/well of six‐well plate), ASC (500 ng/well) and Casp1 (2000 ng/well). After NLRP3‐overexpressing cells were treated with the stimulus (S1) such as nigericin (Nig) or overexpressed the Tau proteins (Tau441, TauT181E, TauS199E, TauT217E, TauS262E and TauK18–), the cells were collected in lysis buffer consisting of 10 mM KCl, 1.5 mM MgCl₂, 10 mM Tris–HCl (pH 7.5), .2% NP‐40 and protease inhibitor cocktail and centrifuged at 1200 g for 5 min. The cell extracts (20 µg) were then mixed with 10⁶ recipient cells (S2) semi‐permeabilised with 60 ng perfringolysin O (PFO, Cusabio), a bacterial toxin that forms pores in the plasma membrane. Additional lysis buffer was added to reach a final volume of 20 µL. After incubation at 30°C for 80 min, the reaction mixture was incubated with .5% NP‐40 on ice for 15 min before centrifugation at 16 000 g for 15 min. We took part of cell supernatant from S1 or S2 for immunoprecipitation. The remaining supernatant was boiled in sodium dodecyl sulfate (SDS) loading buffer and used for immunoblotting.

The cells and brain tissues were lysed and homogenised with radio immunoprecipitation assay (RIPA) lysis buffer (Beyotime Biotechnology) containing phenylmethanesulfonyl fluoride (PMSF), 1:100 and proteinase inhibitor cocktail (1:100, Yeasen Biotech) before centrifugation for 10 min at 16 000 g at 4°C. The obtained supernatant was added to 4× loading buffer (50 mM Tris–HCl, pH 7.6, 2% SDS, 10% glycerol, 10 mM dithiothreitol [DTT]) at the ratio of 3:1, then boiled at 100°C for 10 min followed by sonication and determination of protein concentration by BCA kit (Thermo Fisher Scientific). In order to visualise the protein samples during loading, we added .2% bromphenol blue into them. After electrophoresis in 10% SDS–polyacrylamide gel electrophoresis (PAGE), the separated proteins were transferred onto nitrocellulose membranes (Amersham Biosciences). The membranes were then blocked with 5% nonfat milk dissolved in tris buffered saline (TBS), 50 mM Tris–HCl, pH 7.6, 150 mM NaCl) for 1 h and incubated with primary antibody at 4°C overnight. The membranes were washed by TBST buffer (TBS + .1% Tween‐20) three times for 5 min each and incubated with 1:10 000 anti‐mouse or anti‐rabbit secondary antibody for 1 h at room temperature. The enhanced chemiluminescence kit was then employed to detect signals. The visualised images were collected with Odyssey Infrared Imaging System (Li‐Cor Biosciences) or Chemiscope 6000 (Clinx). For quantification of protein levels, appropriate film exposures were scanned and the density of bands was determined by ImageJ software and normalised to band intensity of β‐actin or GAPDH.

To determine NLRP3 acetylation level and analyse protein–protein interactions, we performed IP/Co‐IP using cell lysates with a diluted concentration of 1 µg/µL. The lysates were incubated with specified primary antibodies and protein A/G agarose beads at 4°C overnight. The beads were collected and washed three times with RIPA buffer/phosphate‐buffered saline (PBS), then resuspended with 1× loading buffer and boiled for 10 min. Finally, the supernatant was collected and analysed by Western blotting.

To confirm whether Tau protein can acetylate NLRP3, purified human WT full‐length Tau protein (Tau441) and the NLRP3‐protein A/G agarose beads complexes were incubated in test tube (Figure 3D). Purification of Tau441 was accomplished by HUABIO. Briefly, the cDNA of Tau441 was cloned in a His‐tagged PET‐28a (+) vector and amplified in ROSETTA 2(DE3)PLYSS competent cells (1 mM isopropyl β‐D‐thiogalactopyranoside for 4 h induction) before affinity purification by using Ni‐NTA resin. The molecular weight and purity of recombinant Tau are 75 kDa and >85% as estimated by SDS–PAGE gel, respectively. To obtain NLRP3 protein, we transfected the plasmid GV362‐EGFP‐FLAG‐NLRP3 (or the mutant NLRP3) into HEK‐293T cells in a 10 cm diameter dish for 48 h. The NLRP3 protein in cells was completely recruited to agarose beads by IP with anti‐Flag primary antibody. For the in test‐tube acetylation assay, 50 nM each of the purified Tau protein, one‐third of the NLRP3‐protein A/G agarose beads complexes and 1 mM acetyl coenzyme A (acetyl CoA, Sigma) were incubated at 37°C for 2 h with constant shaking in the reaction buffer (50 mM HEPES [pH 8.0], 10 mM sodium butyrate, 1 mM DTT and 10% glycerol).46 The acetylation level of NLRP3 was analysed by Western blotting using anti‐acetylated lysine antibody. Using this system, we also detected the in vitro binding between purified Tau and NLRP3 in this assay.

The mice under isoflurane anesthesia were perfused intracardially with normal saline and 4% paraformaldehyde (PFA) to pre‐fix the brain tissues. Immediately, the mice were euthanised and the brains were quickly removed into fresh 4% PFA for a further 24 h fixation at 4°C. After washing with PBS, the brains were immersed into 30% sucrose–PBS for thorough dehydration. Then, the brains were embedded in opti‐mum cutting temperature (OCT compound), frozen and coronally cut into 30 µm sections using a cryotome (CM1950, Leica). Slices were permeabilised with .5% Triton–PBS and blocked with 3% bovine serum albumin (BSA) for 30 min, respectively. After staining with indicated antibodies (overnight for primary antibodies staining and 1 h for Alexa Fluor 647‐, 594‐ or 488‐conjugated secondary antibodies staining), they were incubated with 4',6‐diamidino‐2‐phenylindole (DAPI) for 8 min at room temperature. For cultured cells, the cells were washed three times with PBS for 5 min after indicated treatment, and then fixed with 4% PFA for 30 min at 4°C. The remaining steps were the same as described in the brain slice samples above. Eventually, the samples were transferred to microscope slides for observation and images acquisition under a LSM800 fluorescence microscope (Zeiss) or a VS120 virtual slide microscope (Olympus). All quantitative and morphological analyses were fulfilled by using ImageJ software.51

After Tau‐induced acetylation reaction (Figure 3D), the NLRP3–protein A/G agarose beads complexes were washed with RIPA buffer/PBS three times. Then, we added 40 µL 1× loading buffer into the complexes, boiled the sample for 10 min and performed SDS–PAGE and Coomassie blue staining. Protein samples in gel (about 1 cm × 1 cm in size) were collected and sent to SpecAlly Life Technology Co., Ltd. for liquid chromatography–mass spectrometry (MS) analysis.

Molecular docking was performed in the ClusPro server (https://cluspro.org↗)52 using existing NLRP3 PYD monomer structure (PDB: 2NAQ) and Tau filament structure (PDB: 7NRS). The docking results were modified in Coot before going through the PDBePISA server. The potential docking solutions with low energy were chosen since they all displayed Tau–NLRP3 interaction and relevant sites. Structure images were generated using the PyMOL Molecular Graphics System.

C57BL/6 mice under deep anesthesia were injected with the AAV viruses (AAV9 serotype) by using a stereotaxic instrument (Kopf). To achieve transfection of neurons in vivo, we selected the neuron‐specific promoter hSyn. The AAV viruses including pAAV9‐hSyn‐Tau441‐mCherry‐3FLAG (Tau441), AAV9‐hSyn‐TauS262E‐mCherry‐3FLAG (TauS262E), AAV9‐hSyn‐TauK18△‐mCherry‐3FLAG (TauK18–) and control vector (vehicle) were constructed and packaged by Genechem. Left hippocampal CA1 region stereotaxic injections (−1.9 mm anterior/posterior [A/P], −.7 mm medial/lateral to bregma [M/L] and −1.5 mm dorsal/ventral [D/V] to the dura surface) of 2 µL high‐titre AAV (∼10¹³ v.g./mL) were performed with a Hamilton syringe at a rate of .2 µL/min. Three months later, the mice were tested for the behaviours and PET imaging, and then sacrificed for further detections (Figure 4A).

Two weeks after pAAV9‐hSyn‐Tau441‐mCherry‐3FLAG injection into the left hippocampal CA1, C57BL/6 mice were repeatedly administered with the TNB peptide of 5 mM via lateral ventricle (−.2 mm A/P, −.9 mm M/L, −2.3 mm D/V)‐implanted guiding cannulas (RWD) once every 3 days for 6 weeks. During dosing, mice were restricted by a stereotaxic apparatus and kept awake. One month after the last peptide administration, the mice were tested for the behaviours and sacrificed to further detections (Figure 8A). To confirm the effective delivery of the FITC–TNB peptide, we also injected the peptide into lateral ventricle of C57BL/6 mice for 1 h and euthanised them immediately to observe the location of TNB in brain through confocal microscopy imaging.53

In open field test (OFT), the mice were put into a square‐shaped 50 cm × 50 cm plastic apparatus surrounded by a 40 cm wall, and allowed to freely explore for 5 min. A 25 × 25 (or 30 × 30) cm² square area at the centre of the apparatus was defined as the centre zone. The mice were first placed into the centre zone and their behaviour record was started simultaneously. Data were gathered and analysed with a video‐tracking system (Chengdu Techman Software Co., Ltd.). After each trial, the open field was cleaned with a paper towel dampened with 75% ethanol to remove odour cues. For analysis, the ratio of time spent in the centre zone to the total time was quantified.

In novel object recognition (NOR) test, for adapting the environment, the mice were put in the same plastic apparatus as in OFT for 5 min without any other object 24 h before the experiment. On the next day, the mice re‐entered the apparatus from the same starting point and were allowed to freely explore the symmetrically placed objects A and B (different shape and colour) for 5 min. To measure long‐term memory, after 24 h, the object B was replaced by object C with different shape and colour from A and B, and the mice were reintroduced into the apparatus to explore for 5 min (test phase). Between mice trials, the equipment was wiped with ethanol to remove odour cues. The exploration behaviour of mice on the two objects (A/B and A/C) was recorded using a video‐tracking system (Chengdu Techman Software Co., Ltd.). The novel object discrimination index was calculated as the time spent in exploring the novel object divided by the time spent on both objects, which was quantified in both the training phase and test phase.

In fear conditioning test (FCT), the mice were placed into the conditioning chamber (33 cm × 33 cm × 33 cm) equipped with white board walls, a transparent front door, a grid floor and a speaker (Chengdu Techman Software Co., Ltd.). On day 1, the mice were allowed to conduct a free exploration for 3 min before delivery of a tone conditioned stimulus (CS, 20 s, 80 dB, 300 Hz) through the speaker paired with a foot shock unconditioned stimulus (US, 3 s, .8 mA) at the end of the tone through the grid floor. A total of three CS–US pairs with a 60 s intertrial interval (ITI) were exerted to each animal in the training stage. The mouse was removed from the chamber 1 min after the last foot shock and placed back into the home cage. The contextual fear conditioning stage started 24 h after the training phase. During the test, the mouse was put back into the same conditioning chamber without sound stimulation for 5 min, and the time of its freezing responding to the same context was recorded. On the third day, the mouse was placed back into the same chamber with different contextual cues including coloured walls and a smooth plastic floor for 5 min. After 2 min of free exploration, the mouse was exposed to the exact same 3‐CS tones with 20 s ITI as in the training stage without the foot shock, and its freezing response to the tone in the altered context was recorded.

The Morris water maze (MWM) was performed as previously described to detect spatial learning and memory.54, 55 The apparatus used in MWM was a circular tank (120 cm in diameter, 60 cm in height) filled with nontoxic white dye‐tinted water (25 ± 2°C). Different posters are plastered on the walls of the tank. The maze was theoretically divided into four equal quadrants. An escape platform with a diameter of 10 cm was placed in the middle of one of the quadrants (target quadrant) and submerged 1 cm below the water surface and kept in the same position during the trial. A video‐tracking camera above the centre of the pool surface monitored the trajectory of the mice, and the video signal was transmitted to a computer in an adjacent room. Each mouse underwent four training trials a day for consecutive 5 days from semi‐random start positions to find the hidden platform within 60 s. If the mouse could find the platform, it would stay on the platform for 20 s. Otherwise, the mouse would be gently directed to the platform and kept there for 20 s. Latency (s) to find the hidden platform was recorded after each trial of each learning session. The hidden platform was removed 48 h after the end of training. The mice were placed into the pool for the probe trial with a duration of 60 s. The target quadrant crossing times was recorded.

As previous procedure,56 we labelled Tosylate‐DPA‐714 (CAS: 958233‐17‐5, Huayi Isotopes) with 18F produced by a cyclotron (MINItrace, GE Healthcare) to obtain the radiotracer [18F]‐DPA‐714 (radiochemical purity [RCP] = 99%) using a TRACERlab FX FN synthesis module (GE Healthcare). The RCP was determined by means of HPLC, using an U3000 HPLC system (Thermo Scientific) with a Flow‐Ram radio‐HPLC detector (LabLogic, Sheffield). Simultaneous PET and CT scanning in mouse brain was performed by using a small animal microPET/SPECT/CT multimodal imaging system (InliView‐3000B, Novel Medical, Yongxin Medical Equipment Co., Ltd.). The mice were fasted for 12 h and weighted before the radiotracer injection. Each mouse was intravenously injected with 200 µCi ± 10 µCi of [18F]‐DPA‐714 via tail vein. After an uptake period of 50 min, the mice were quickly anaesthetised with isoflurane inhalation and prone positioned in the PET/CT scanning bed board with a mask covering its mouth to ensure continuous anesthetic gas inhalation throughout the scanning. CT localisation was first conducted to select the head as the scanning range of CT and PET. PET scan was performed for 10 min. Image analysis was performed using PMOD v4.104 (PMOD Technologies). A predefined mouse brain atlas template was used to analyse different brain regions including striatum, cortex, hippocampus, thalamus and cerebellum. Standard uptake value (SUV) of [18F]‐DPA‐714 in the region of interest (ROI) was calculated as: SUV = the radioactivity concentration of ROI (kBq/mL)/injected dose (MBq)/body weight (kg), 1 µCi = .037 MBq.

All data were expressed as means ± SEMs of at least three independent assays unless otherwise stated. Statistical analyses were carried out using GraphPad Prism 7.0 (GraphPad Software Inc.). Differences between groups were evaluated by unpaired two‐tailed Student's t test or one‐way analysis of variance. Correlations between groups were determined by Pearson's correlation test. p < .05 was considered statistically significant: ^****p < .0001, ^***p < .001, ^**p < .01 and ^*p < .05.

To identify the potential role of NLRP3 acetylation in inflammasome activation in transgenic mice and patients with Tauopathy, we analysed hippocampal samples from 3xTg‐AD transgenic mice (9‐month old). The elevated Caspase‐1 cleavage and NLRP3 and ASC levels indicated NLRP3 inflammasome activation, and meanwhile NLRP3 acetylation increased in AD mice (Figure S1A,B). Although we observed a great increase of Tau/pTau (AT8 and Phospho‐Tau (Ser262)), it is noteworthy that extracellular Aβ deposits commonly arise in 3xTg‐AD transgenic mice at 6 months or older.57 Thus, to further identify the potential relationship between Tau and NLRP3 acetylation, we tested the hippocampal samples from PS19 transgenic mice with age of 3 or 6 months. PS19 mouse line overexpresses MAPT with the P301S mutation and is a representative Tauopathy model. As shown in Figure 1A,B, the levels of NLRP3, ASC, cleaved Caspase‐1 as well as Ace‐NLRP3 (Ace‐lys/NLRP3) in hippocampal samples from 3‐ or 6‐month‐old PS19 mice were significantly increased compared with 6‐month‐old WT mice. More importantly, NLRP3 acetylation and NLRP3 inflammasome activation were also observed in the hippocampal tissues and temporal cortices of AD patients (Figure 1D,E). Pearson correlation analysis based on the above results revealed the strong correlations among Tau/pTau level, NLRP3 acetylation and NLRP3 inflammasome activation: Tau/pTau level was positively correlated with NLRP3 acetylation, and NLRP3 acetylation level was positively correlated with inflammasome activation (Figures 1C and S1D). These data suggest that NLRP3 acetylation is increased with inflammasome activation in Tauopathy transgenic mice and patients with AD, and Tau/pTau level is highly correlated with NLRP3 acetylation and inflammasome activation.

In order to determine the role of Tau proteins in NLRP3 acetylation and inflammasome activation in microglia, according to disease progression and Tau‐associated early diagnostic markers,31 we chose to explore the effects of four representative pTau with relatively high abundances in the early stage of disease: pTau‐T181, pTau‐S199, pTau‐T217 and pTau‐S262 through overexpressing their mimics TauT181E, TauS199E, TauT217E and TauS262E in cells. Additionally, as controls, full‐length WT Tau441 and microtubule‐binding repeats domain lacked TauK18‐overexpression plasmids were used. As shown in Figure S2A, after 4 h pretreatment with 1 µg/mL LPS, the BV‐2 microglia overexpressed Tau protein for 48 h (the transfection efficiency of GFP‐Tau is relatively low, at 18.4%), or was incubated with the supernatant/lysate from HEK‐293T cells with Tau protein overexpression for 18 h or nigericin (Nig, inflammasome activation positive control, 20 µM) for 3 h. Tau overexpression in HEK‐293T cells did not reduce the viability of cells, and resulted in Tau release into the culture medium (Figure S2B,D). All the three treatments on BV‐2 cells, including incubation with the supernatant (S) or lysate (P) from Tau‐overexpressing HEK‐293T cells, or direct overexpression (T) of WT Tau441 or pTau mimics, induced NLRP3 inflammasome activation, manifested by significantly increased NLRP3, ASC and cleaved Caspase‐1 levels as well as IL‐1β release (Figure S2E–K). By comparison, the supernatant and cell lysate containing TauS262E protein induced the strongest NLRP3 inflammasome activation in BV‐2 cells, followed by TauT181E, TauS199E, TauT217E and then Tau441, while TauK18– had no effect on promoting NLRP3 inflammasome activation compared with LPS treatment alone group (Figure S2E–H). Tau also induced NLRP3 acetylation in BV‐2 cells, and TauS262E again, showed the strongest effect, whereas TauK18– had no impact on NLRP3 acetylation (Figure S2I,J). Similarly, Pearson correlation analysis at the cellular level indicated that NLRP3 acetylation may contribute to inflammasome activation (Figure S2L). To verify the acetylation of NLRP3 and activation of inflammasome induced by Tau and pTau, we next adopted a more accurate method to evaluate the activity of NLRP3 inflammasome based on HEK‐293T cell culture system (Figure 2A). Tau/NLRP3 and ASC/Caspase‐1 were separately expressed in donor and recipient cells. The latter, after PFO permeabilisation, were incubated with extracts from Tau/NLRP3‐overexpressing donor cells (activator), then the cleavage of Caspase‐1 was detected to access the NLRP3 inflammasome activation. As shown in Figure 2B,C, in donor cells, full‐length Tau and pTau‐induced NLRP3 acetylation, with TauS262E showing the most pronounced effect; when recipient cells with ASC/Caspase‐1 expression were incubated with the Tau‐stimulated NLRP3 from donor cells, Caspase‐1 was cleaved, indicating effective NLRP3 inflammasome assembly and activation. These results demonstrated that Tau/pTau‐induced NLRP3 acetylation can potently increase Caspase‐1 cleavage, whereas TauK18– had no effect on NLRP3 acetylation or Caspase‐1 cleavage. By comparing the NLRP3 acetylation level in Tau treated groups in S1 and S2, it could be found that NLRP3 acetylation levels were further increased during the 80‐min incubation, implying that Tau protein continue exerting its effect of acetylating NLRP3 in S2. By contrast, NLRP3 did not further increase its acetylation level in S2 in positive control group since in this incubation system the acetylation inducer and NLRP3 activator nigericin42 was not added. Next, acetylation of NLRP3 induced by Tau was further identified by incubation of BV‐2 cells with purified Tau protein. After 4 h pretreatment with 1 µg/mL LPS, BV‐2 cells were incubated with purified Tau441 protein (5 µM, by referring to the previous publications58, 59) with or without Sirt2 overexpression/acetyltransferase inhibitor TPOP146 or L‐45 treatment for 18 h. As shown, exogenous recombinant Tau441 protein was found in BV‐2 cells (Figure 2D), and induced NLRP3 acetylation and inflammasome activation, which could be prevented by deacetylase Sirt2 and acetyltransferase inhibitor TPOP146/L‐45 (Figure 2E,F). Together, these data suggest that Tau and pTau proteins which increase in early stage of Tauopathy promote NLRP3 acetylation and activate NLRP3 inflammasome.

It has been reported that NLRP3 K21 and K22 sites are modified by acetylation in macrophages and are targeted by an NAD+‐dependent deacetylase Sirt2 for deacetylation, and NLRP3 acetylation facilitates the NLRP3 inflammasome activation.41 In addition, K24 acetylation of NLRP3 regulated by KAT5, a histone acetyltransferase belonging to MYST family, is critical for its full activation.42 Coincidentally, Tau was identified as a bona fide acetyltransferase that shares sequence and functional similarities to the MYST family of HATs, and is capable of self‐acetylation.43 Therefore, we tested whether these sites were involved in Tau‐induced NLRP3 acetylation by overexpressing mutant plasmids of NLRP3 (K21R, K22R and K24R) and Sirt2 in TauS262E treated HEK‐293T cell model (Figures 2A and 3A). The results showed that the levels of TauS262E‐induced NLRP3 acetylation, as well as NLRP3 inflammasome activation, were significantly decreased in K21R, K22R and K24R mutant‐overexpressing cells (Figure 3B,C). Likewise, Sirt2 overexpression also significantly suppressed Tau‐induced NLRP3 acetylation and inflammasome activation (Figure 3B,C). Altogether, these results indicate that the highly conserved residues K21, K22 and K24 in the NLRP3‐PYD domain are potential sites of Tau‐induced NLRP3 acetylation, which is crucial for the activation of NLRP3 inflammasome.

Although the previous results demonstrated that different Tau proteins exhibited varying effects on promoting NLRP3 acetylation and inflammasome activation, there was no evidence supporting a direct interaction between Tau and NLRP3. To confirm whether Tau directly acts on NLRP3, we first transfected EGFP–NLRP3 and Flag–Tau441 plasmids into HEK‐293T cells for 48 h, and then performed IF staining or Co‐IP (Figure S3A). As shown in Figure S3B,C, Tau visibly co‐localized with and was bound to NLRP3. Next, we incubated purified WT Tau441 protein with agarose beads bound to NLRP3 or mutant NLRP3 proteins (K21R, K22R and K24R NLRP3) in the test tube, and then measured the levels of NLRP3 acetylation and Tau–NLRP3 binding (Figure 3D). We observed the binding of Tau protein to NLRP3 again and confirmed the acetyltransferase activity of Tau441 protein in NLRP3 acetylation, while the mutations of K21R, K22R and K24R markedly impeded the binding and subsequent acetylation (Figure 3E–H), which convincingly illustrates that Tau can directly bind to and acetylate NLRP3, and K21, K22 and K24 at NLRP3–PYD domain are key sites for Tau‐induced NLRP3 acetylation. Coomassie staining combined with MS detection revealed the presence of Tau protein on the agarose beads (Figure S3D,E), which further suggested the direct binding of Tau and NLRP3. Since NLRP3 K21, K22 and K24 mutations dampened the NLRP3 binding to Tau and NLRP3 acetylation induced by Tau, we speculated that the region containing these sites might mediate the interaction with Tau. Therefore, we conducted molecular docking to confirm the interaction region between Tau and NLRP3. As shown in Figure S3F, three potential docking models displayed that K21 and K22 sites of NLRP3 were in close proximity (<3.0 Å) to the Tau S324 and D348 sites, which met our expectation. In addition, MS detection of the Tau‐incubated NLRP3 identified abundant conserved NLRP3 unique peptide LAQYLEDLEDVDLKK containing K21 and K22 sites, reinforcing the possibility of their acetylation modification by Tau (Figure S3E).

To further confirm the role of Tau‐induced NLRP3 acetylation in the development of neurodegeneration in Tauopathy, we injected AAV carrying Tau441 (AAV‐Tau441), TauS262E (AAV‐TauS262E), TauK18– (AAV‐TauK18–, loss of acetyltransferase activity) or the vector control with 3×Flag‐mCherry into hippocampal CA1 region of 3‐month‐old C57BL/6 mice (Figure 4A). Three months after AAV injection, the successful overexpression of Tau proteins in hippocampus was confirmed (Figures 4A and 5A,B). Then, the cognitive function was evaluated by using NOR, MWM and FCT, respectively. In NOR, the significant memory impairment appeared in the mice of Tau441 and TauS262E‐overexpressing group, while TauK18– overexpression did not impair the visual episodic memory (Figure 4B,C). In addition, the results of FCT showed that the mice overexpressing Tau441 and TauS262E had decreased total freezing time (s) in the context and tone conditioning paradigm compared with the mice in vehicle group, indicating the fear memory impairment. Memory impairment was more pronounced in the TauS262E group than in the Tau441 group, and was unaffected in the TauK18– group (Figure 4D,E). The spatial learning and memory ability in the mice of four groups was determined by MWM. The mice in Tau441 and TauS262E group showed impaired learning ability with significantly longer latency to find the hidden platform on the fifth day of the training period. In the memory test on day 7, the number of times mice crossed the location of the platform was significantly lower in the Tau441 and TauS262E groups than in the vehicle group. However, no significant differences were observed between the TauK18– group and vehicle group during both the training and testing stages (Figure 4F–H). In addition, in OFT, there was no significant difference in the time and distance of moving among the four groups, indicating that Tau441, TauS262E and TauK18– overexpression for 3 months had no effect on the motor ability of mice (Figure S4). Taken together, these results suggest that Tau441 and TauS262E overexpression in the hippocampus for a duration of 3 months induces cognitive impairment in C57BL/6 mice.

As previously shown, TauK18– with deficiency of acetyltransferase activity cannot cause the activation of NLRP3 inflammasome in microglia (Figure S2I–K). Combined with the behavioural results (Figure 4), we speculate that the cognitive impairment caused by Tau441 and TauS262E overexpression may be linked to their effects on the activation of NLRP3 inflammasome in microglia and promotion of microglia activation. To verify this, we performed Western blotting to identify the changes of NLRP3 inflammasome activation related protein molecules in hippocampus. We observed elevated Caspase‐1 cleavage，and NLRP3 and ASC levels indicating NLRP3 inflammasome activation in Tau441 and TauS262E overexpression group, as well as increased NLRP3 acetylation (Figure 5A,B). IF staining of Iba‐1 and sholl analysis further confirmed that overexpression of Tau441 and TauS262E caused reactive microgliosis which was manifested by the typical activation morphology (decrease in branch length and number) and increased number of microglia in AAV‐infected brain regions (Figure 5C–F).

The widespread propagation of Tau pathology is a key contributing factor for the continuous deterioration of Tauopathies.60 It has been demonstrated that microglia, especially in an inflammatory or reactive state, can drive Tau transmission and toxicity via various pathways.61, 62 Based on the observation that Tau acetylated NLRP3 and activated microglia, we speculated that the spatial distribution of Tau pathology would share the similar pattern with that of microglia activation. Activation of microglia leads to upregulation of the mitochondrial 18 kDa translocator protein (TSPO), which has been used as an in vivo PET imaging biomarker for neuroinflammation.63 To trace microglial activation in the whole brain, we performed PET/CT imaging using the radiotracer [18F]‐DPA‐714, which has been identified as a selective TSPO ligand with demonstrated good binding potential and bioavailability.64 As shown in Figure 6A–C, wide range of DPA‐714 high uptake signals were monitored in various regions throughout the whole brain of the mice in TauS262E‐overexpressing group, which represented the occurrence of neuroinflammation. Neuroinflammation was relatively milder in the Tau441 group than in the TauS262E group, and was almost absent in the TauK18– group, which aligns with the inability of TauK18– to induce NLRP3 acetylation. These results further confirmed the importance of NLRP3 acetylation induced by Tau in promoting microgliosis and neuroinflammation in Tauopathies. We also performed Iba‐1 IF staining on coronal sections of the entire mouse brain in the vehicle and TauS262E groups to confirm microglial activation for double validation of inflammatory response characteristics in whole brain (Figure S5A). As expected, striking microglia activation and aggregation appeared at coronal brain sections adjacent to the initiation site of AAV‐TauS262E injection (Figure 6D), while no obvious microgliosis was observed in vehicle group (Figure S5B). Furthermore, as depicted in Figures 6D and S5B, the mCherry‐labelled TauS262E proteins (red) showed a three‐dimensional, multi‐directional propagation that partly overlapped with the diffusion pattern of the AAV itself. However, through meticulous comparison, we found that the diffusion of AAV–vehicle was less pronounced particularly towards the anterior/posterior regions of the brain compared to AAV–TauS262E. Therefore, by discounting the diffusion effect of AAV, our data demonstrated that TauS262E proteins actually spread more rapidly and extensively throughout the brain compared to the virus itself. Since TauS262E potently activated microglia, this partially suggests that reactive microgliosis may reciprocally promote TauS262E propagation, which is in line with previous observations.61, 62

To explore whether blocking NLRP3 acetylation and NLRP3 inflammasome activation induced by Tau can prevent microgliosis in Tauopathies, we designed a peptide TNB (LEDLEDVDLKKFKMHLEDYPP) based on region of NLRP3 binding with Tau to competitively inhibit Tau–NLRP3 interaction. First, we assessed the cytotoxicity of various concentrations (0, 25, 50, 100, 200 and 400 µM) of TNB on BV‐2 cells, N2a cells, SH‐SY5Y cells, primary astrocytes and primary neurons by the CCK8 assay. As shown in Figure 7A, application of TNB at concentrations lower than 200 µM did not cause any significant change in cell viability, whereas TNB at 400 µM induced significant decrease in cell viability. Therefore, we added the purified Tau441 protein and peptide TNB within a safe dose range into the BV‐2 cells pretreated with LPS (Figure 7B) in the following cell experiments. TNB could markedly inhibit IL‐1β production and Caspase‐1 cleavage induced by LPS and Tau441 protein. But the levels of NLRP3 and ASC showed no change after TNB treatment (Figure 7C–E). As expected, TNB disrupted the interaction between Tau and NLRP3 through competitive binding with Tau, consequently impeding the NLRP3 acetylation (Figure 7C–F). In addition, TNB did not interfere with Tau uptake by BV‐2 cells (Figure 7F), as alterations in this process can potentially impact NLRP3 inflammasome activation. Since we previously observed that nigericin could significantly promote NLRP3 acetylation and inflammasome activation in BV‐2 cells pretreated with LPS (Figure 2E,F), possibly in a KAT5‐dependent manner,42 we further verified the specificity of TNB by treating BV‐2 cells with 200 µM TNB peptide following LPS + nigericin pre‐treatment. We found that TNB could not inhibit NLRP3 acetylation and inflammasome activation induced by LPS + nigericin (Figure S6). Thus, TNB specifically and competitively binds to Tau and inhibits Tau‐mediated NLRP3 acetylation, thereby preventing inflammasome activation in BV‐2 microglia (Figure 7G).

To further confirm the effect of TNB peptide on Tau‐induced NLRP3 acetylation and inflammasome activation in vivo, we repeatedly administered the model mice (Tau441 overexpression in hippocampus for 3 months) with the TNB peptide (5 mM) via lateral ventricle‐implanted guiding cannulas once every 3 days for 6 weeks (Figure 8A). All mice were alive at 3 months after AAV injection (Figure 8B). We injected FITC–TNB into the lateral ventricle of C57BL/6 mice and euthanised the animals 1 h later to observe the localisation of TNB in the brain by confocal microscopy. As shown in Figure 8C, TNB was extensively distributed along the edges of several ventricles and in the hippocampus, especially in the dentate gyrus (DG). TNB was also scattered throughout the cortex and other brain regions. In the OFT, we found that Tau441 overexpression and/or TNB peptide administration did not impair motor ability of the mice (Figure 8D). TNB significantly ameliorated cognitive impairment in model mice, as indicated by increased recognition index in NOR, increased total freezing time in FCT, and shorter latency to find the hidden platform as well as increased number of platform crossings in MWM (Figure 8E–H). Consistent with the in vitro experimental results (Figure 7D,E), it was found that TNB peptide could inhibit NLRP3 acetylation and inflammasome activation in vivo without changing the levels of NLRP3 and ASC through Western blotting analysis of hippocampal samples (Figure 9A,B). Moreover, TNB peptide significantly ameliorated the activation phenotype of microglia induced by Tau (Figure 9C–G). Collectively, blocking Tau‐induced NLRP3 acetylation to inhibit inflammasome activation by TNB peptide rescues cognitive impairment in Tau‐overexpressing mice.

Data will be made available on request.