Abstract Background Reactive astrocytes are associated with the progression of neurodegenerative diseases such as Tauopathies represented by Alzheimer’s disease (AD), particularly in the human APOE4 background. Previous studies using P301S tauopathy model mice have shown the potential contribution of activated astrocytes to tau-associated neurodegeneration and accelerated mortality. However, the impact of Aβ pathology on modulating mortality, astrocyte activation and neurodegeneration in this model remains unclear. Methods We established a novel AD mouse model (E4-3Tg) by crossing P301S-Tau mice with APPswe/PSEN1dE9 mice on human APOE4 knock-in background and compared its phenotypes with the human APOE4 knock-in P301S-Tau (TE4) mouse model. The impact of Aβ in P301S tauopathy mice was evaluated by analyzing survival rate, astrocyte activation, tau-associated neurodegeneration and behavior performance. We also assessed degeneration in mature and immature hippocampal neurons using scanning electron microscopy and RNA-sequencing analysis. Moreover, we suppressed A1 astrocyte activation to evaluate its contribution in the pathology observed in E4-3Tg and TE4 mice. Results E4-3Tg mice exhibited significantly accelerated mortality, with a median survival of approximately 6.5 months, compared to around 9 months in TE4 mice. At 6.5 months of age, compared to TE4 mice, E4-3Tg mice had already developed prominent A1 astrocyte activation, accompanied by exacerbated tau pathology, neuronal loss, and cognitive impairment. Notably, E4-3Tg mice also showed a marked reduction in hippocampal immature neurons, associated with tau-related mitochondrial dysfunction and cellular senescence, a phenotype absent in TE4 mice even at 9 months of age. Administration of anserine (beta-alanyl-3-methyl-L-histidine), which was demonstrated to be a dual IRAK1/TAK1 inhibitor targeting astrocytes, effectively suppressed A1 astrocyte activation and reduced the accelerated mortality observed in both E4‑3Tg and TE4 mice. Moreover, this protective effect of anserine was accompanied by attenuated tau-associated mitochondrial dysfunction and neuronal loss in both mature and immature hippocampal neurons. Conclusion Collectively, our findings demonstrated that Aβ pathology accelerates mortality in tauopathy model by promoting A1 astrocyte activation and aggravating tau-associated neurodegeneration. Therapeutic inhibition of astrocytic IRAK1/TAK1 signaling reduced mortality and alleviated tau-associated neurodegeneration, highlighting A1 astrocyte activation as a promising target for reducing accelerated mortality in Alzheimer’s disease. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-025-03564-7. Keywords: Astrocyte activation, Mortality, Tau, Neurodegeneration, Anserine Introduction In recent years, various studies have provided evidence that astrocytic neuroinflammation contributes to the progression of neurodegenerative diseases [[48]1, [49]2]. Astrocytes, the most abundant glial cell type in the brain, normally play an important role in supporting neurons through astrocyte–neuron communication, including energetic and metabolic support [[50]3]. However, in neurodegenerative diseases, astrocytes lose the neuron-supporting functions and become over-activated through the activation of the NF-κB pathway, ultimately acquiring a neurotoxic phenotype known as A1 astrocytes [[51]2, [52]4, [53]5]. Activated A1 astrocytes are closely associated with increased mortality, potentially through the promotion of neurodegenerative processes. Toxin-induced severe astrocyte activation has been found to lead to neurodegeneration and mortality in an astrocyte-specific toxin receptor model (GiD) [[54]6]. Furthermore, inhibition of astrocyte reactivity through knocking out Tnf, Il1a and C1qa has been reported to reduce mortality in an amyotrophic lateral sclerosis (ALS) mouse model [[55]7]. These studies suggest the contribution of activated astrocytes to accelerated mortality in neurodegenerative diseases. Alzheimer’s disease (AD), the most representative neurodegenerative disease characterized by progressive neuronal loss and cognitive decline, is also one of the leading causes of mortality worldwide [[56]8, [57]9]. Among its hallmark pathologies, tauopathy, characterized by the accumulation of hyperphosphorylated tau (p-Tau) in neurons, is considered a key driver of neurodegeneration [[58]10]. Clinical studies have shown the association between aggravated tau pathology and high mortality risk in AD patients [[59]11, [60]12]. This pathological association has been recapitulated in P301S-Tau transgenic mice, a tauopathy model with the human MAPT P301S mutation linked to frontotemporal dementia with parkinsonism on chromosome 17 (FTDP-17) [[61]13]. Although originally developed as a tauopathy model, P301S-Tau mice also exhibit several AD phenotypes including tau-associated synaptic and neuronal loss, and they are widely used to study tau pathology in AD in recent year. Importantly, recent studies in this model further revealed that tau pathology and neurodegeneration are accelerated by activated A1 astrocytes through over-production of Complement component 3 (C3), the most representative marker of A1 astrocytes [[62]14–[63]16]. Moreover, A1 astrocyte activation [[64]17] and C3 over-production are also associated with the impairment of adult hippocampal neurogenesis (AHN) [[65]18], a process that is known to be severely impaired during AD progression, particularly in the presence of tau pathology [[66]19–[67]21]. Notably, astrocyte activation and tau-associated neurodegeneration are markedly exacerbated in human apolipoprotein E4 (APOE4) background [[68]22–[69]24]. APOE4, one of the strongest genetic risk factors for AD [[70]25], has also been reported to be associated with increased mortality in older adults [[71]26] and AD patients [[72]27]. Together, these findings suggest a potential link between A1 astrocyte activation and accelerated mortality in AD through exacerbation of tau-associated neurodegeneration. In line with this, inhibition of immune responses attenuated tau-associated neurodegeneration and reduced mortality in tauopathy mice [[73]13], indicating that targeting astrocyte activation may represent a therapeutic approach to reduce mortality. Meanwhile, amyloid beta (Aβ), the earliest pathological feature of AD, known to activate astrocytes via Toll-like receptor (TLR) signaling pathways [[74]28, [75]29] and induce NF-κB-mediated C3 over-production [[76]30], is closely linked to exacerbated tau pathology through astrocytic reactivity [[77]31, [78]32]. Therefore, the effects of Aβ on astrocyte activation, tau pathology and related mortality should not be overlooked. However, due to the limitation of AD mouse models, this potential impact remains unclear [[79]33]. To address this limitation, in this study, we crossed APPswe/PSEN1dE9 mice with P301S-Tau mice. To further exacerbate A1 astrocyte activation and tau-associated neurodegeneration, we additionally knocked-in (KI) human APOE4, and established a unique human APOE4 KI APPswe/PSEN1dE9/Tau-P301S triple-transgenic (E4-3Tg) mouse model. Compared with APPswe/PSEN1dE9 mice and P301S Tau mice, E4-3Tg mice are able to recapitulate both Aβ and tau pathologies. Furthermore, unlike the commonly used 3xTg-AD model [[80]33], E4-3Tg mice express human ApoE4 protein instead of mouse ApoE, providing a better environment to mimic AD pathology. It is expected that E4-3Tg mice would exhibit comprehensive AD-related pathologies at an earlier age including mortality. In parallel, we also used human APOE4 KI P301S-Tau (TE4) mice for comparison of phenotypes. To investigate the contribution of activated astrocytes, we inhibited astrocyte activation with anserine (β-alanyl-3-methyl-L-histidine), a natural imidazole dipeptide abundant in vertebrate muscle tissues and known for its anti-inflammatory activity [[81]34, [82]35]. Anserine can reach the brain parenchyma, including the hippocampus after oral administration [[83]36]. Anserine is also detected in high levels in human cerebrospinal fluid (CSF) compared with other dipeptides, with a relatively high plasma-to-CSF transportation [[84]37], suggesting efficient blood–brain barrier (BBB) penetration. Once in the brain, anserine can be taken up almost exclusively by astrocytes via the high-affinity transporter PEPT2 (encoded by Slc15a2), which is predominantly expressed in astrocytes [[85]38–[86]42]. Our previous studies have further demonstrated the ability of anserine to modulate astrocyte-mediated neuroinflammation in APPswe/PSEN1dE9 model mice [[87]43]. In the present study, our findings revealed that Aβ pathology accelerated mortality in the tauopathy model through promoting A1 astrocyte-mediated tau-associated neurodegeneration. Moreover, inhibition of A1 astrocyte activation with anserine, which was demonstrated as a dual IRAK/TAK1 inhibitor, mitigated this pathological effect. These findings highlight A1 astrocyte activation as a promising therapeutic target for reducing mortality in AD. Materials and methods Animals APPswe/PSEN1dE9 (B6/C3) and P301S-tau (B6/C3) transgenic model mice were purchased from Jackson Laboratories (Bar Harbor, Maine, USA). APPswe/PSEN1dE9 model mice express the Swedish mutation, carrying a chimeric human APP transgene (Mo/HuApp695swe) and a human PS1 transgene (missing exon 9). P301S-tau transgenic model mice express human P301S 1N4R tau driven by PrP promoter. Human APOE4 Knock-in (C57BL/6) model mice were purchased from Riken BRC (Koyadai, Tsukuba, Japan). APOE4 Knock-in mice were crossed to APPswe/PSEN1dE9 model mice or P301S-tau model mice to generate APOE4-P301S-Tau or APOE4-APPswe/PSEN1dE9 mice. To generate APOE4-APPswe/PSEN1dE9/P301S-Tau model mice, we crossed female APOE4-APPswe/PSEN1dE9 to male APOE4-P301S-Tau mice because female APOE4-APPswe/PSEN1dE9/P301S-Tau mice were not suitable for breeding due to increased risk of postpartum mortality. APOE4-P301S-Tau mice can be generated by either mating APOE4-P301S-Tau mice with APOE4-Control mice or as part of the breeding process for generating APOE4-APPswe/PSEN1dE9/P301S-Tau mice. For this reason, APOE4-P301S-Tau mice have larger litter size than APOE4-APPswe/PSEN1dE9/P301S-Tau mice, and we used APOE4-P301S-Tau mice for several experiments requiring large cohorts, including ELISA, RNA-Sequencing and AAV-mediated PEPT2 KD to further confirm the effects of anserine. Only APOE4 homozygote mice were used for the experiments. APOE4 Knock-in mice were used as controls. We used abbreviations to describe APOE4-P301S-Tau mice and APOE4-APPswe/PSEN1dE9/P301S-Tau mice. APOE4-P301S-Tau mice were called TE4 mice and APOE4-APPswe/PSEN1dE9/P301S-Tau mice were called E4-3Tg mice, respectively. E4-3Tg mice exhibited no sex-based difference and the male TE4 mice displayed a relatively higher mortality compared to female TE4 mice, these results are shown in Fig. S1. Because anserine reduced mortality equally in both sexes as shown in Fig. S13, and to minimize potential sex bias, all subsequent experiments were conducted with age-matched, sex-balanced cohorts of male and female unmated mice. All mice were sorted by age and genotype, maintained under a 12-hour light/dark cycle at ~ 22 °C, and had free access to food (solid feed MF; Oriental Yeast Co., Ltd., Tokyo, Japan) and water. All animal procedures and experiments in this study were approved by the ethical committee of the University of Tokyo (C-20-2) and were conducted according to the guidelines for animal experimentation required by the University of Tokyo. Anserine treatment Anserine used in the in vivo study was a high-purity product (> 93%) purified from salmon muscle and provided by Tokai Bussan (Tokyo, Japan). Mice were randomly assigned to anserine-treated or untreated groups. Anserine-treated mice were maintained on a steady dosage of anserine diluted in autoclaved drinking water, at a concentration of 2.0 g/L (~ 10 mg per mouse per day) for 8 weeks. For TE4 mice, anserine treatment was performed from about 7 months of age to 9 months of age. For E4-3Tg mice, anserine treatment was performed from about 4.5 months of age to 6.5 months of age. Behavioral tests were performed following the completion of the treatment period. Anserine used in the in vitro study was obtained from Toronto Research Chemicals (Canada, TRC-A678253). This same anserine compound was also used in TE4 mice to confirm its effect on reducing mortality. BrdU injection BrdU (5-Bromo-2’-deoxyuridine) injection was conducted to evaluate the number of matured newborn neurons as we previously described [[88]44]. Briefly to say, BrdU (Wako) (10 mg/ml) was dissolved in physiological saline and mice were received intraperitoneal injection with the dose of 100 mg/kg for 3 days. After 4 weeks mice were sacrificed and BrdU and NeuN co-staining was conducted for counting the number of matured newborn neurons. Y-maze Y-maze was conducted to assess short-term spatial memory of mice, which reflects spontaneous alternation behavior. During the 10-min test, mice were placed on the starting arm and allowed to explore the maze freely. The sequence and total number of arm entries were recorded using the SMART video tracking system (Panlab, Barcelona, Spain). When mice entered 3 different arms in sequence, it was considered an alternation. The alternation rate was calculated as: (number of actual alternations/[total number of arm entries – 2]) × 100. After each session, the maze was cleaned with a 70% ethanol solution. Radial arm maze The radial arm maze (RAM) test was conducted to assess spatial reference memory in mice. Mice were given limited food (2 g per day per mouse) starting at least three days before the test and continuing throughout the testing period. To allow mice to adapt to the maze, they were allowed to freely explore and feed in the maze for two days (habituation). During habituation, a 20 mg food reward was placed at the end of each arm. This was followed by a training trial. The training trials were performed for 5 days and two sessions daily with 1-hour intervals. Mice were placed on the start arm and allowed to search for rewards, which were placed in only three of the arms. The time taken to find all three rewards was recorded using the SMART video tracking system (Panlab, Barcelona, Spain), or until 4 min had elapsed. After each test session, the maze was cleaned with 70% ethanol solution. Contextual fear conditioning test Contextual fear conditioning (CFC) test was performed using P.O.BOX 319 (Med Associates Inc., Albans, UK) for 2 consecutive days. The design of the experiment was described previously [[89]45]. The conditioning session was conducted on the first day. Mice were placed into the box wiped with 70% isopropanol (Context A) for 370 s. Auditory tones (85 dB, 5000 Hz, 10 s) were delivered at 128, 212, and 296 s, each co-terminated with a foot shock (0.75 mA, 1 s). 24 h after the Conditioning test, the contextual test was performed. Mice were placed into the same box (Context A) for 512 s and freezing time was recorded. The freezing behavior was recorded using Video Freeze Software (Med Associates, Inc., St. Albans, UK). Grip strength test The grip strength test was conducted to evaluate the physiological function of each mouse. Before the test, the grip strength meter (MK-380 M; Muromachi Kikai) was cleaned with 70% ethanol. Mice were held by the tail and placed on the grip strength meter, allowing their two forelimbs to grasp the metal grid firmly. Then, the operator gently pulled each mouse away horizontally from the grid until its forelimbs detached, and the peak pulling force was recorded. Each mouse underwent three consecutive trials, with a 1-min interval between trials. The mean of the three measurements was recorded as the grip strength of each mouse (forelimbs). The body weight of each mouse was also recorded simultaneously to exclude any confounding effect of body weight on grip strength. Immunohistochemical staining Mice were perfused with TBS to remove blood, followed by perfusion with 4% paraformaldehyde (PFA) to fix the brain tissue. Brain samples were taken and postfixed in 4% PFA for 24 h, then incubated in 30% sucrose/TBS solution for 48 h. Tissue-Tek OCT compound (Sakura Finetek, Japan) was used to embed brain samples. All the brain samples were stored at −80 °C before use. Brain samples were sliced into 50 μm-thick coronal sections with a Cryostat (Microm, Germany) and kept in cryoprotectant solution at −30 °C. For immunofluorescence staining, sections were washed twice with 0.3% TBS-X (Triton) for 10 min and blocked with 3% normal donkey serum (NDS) diluted in 0.3% TBS-X for 60 min at room temperature and then incubated with primary antibodies overnight at 4 °C with shaking. The primary antibodies used in this study were anti-GFAP (Mouse, 1:1000, Sigma), anti-C3 (Rat, 1:200, Hycult), anti-DCX (Rabbit, 1:500, Cell Signaling), anti-p21 (Rabbit, 1:100, Abcam), anti-DCX (Sheep, 1:100, R&D Systems), anti-Tom20 (Rabbit, 1:500, Abcam), anti-p-TAK1 (Rabbit, 1:200, Cell Signaling), anti-pT181 (Rat, 1:500, Wako), anti-AT8 (Mouse, 1:500, Thermo Fisher), anti-Aβ (Mouse, 1:1000, Wako), anti-NeuN (Mouse, 1:500, Millipore), anti-BrdU (Rat, 1:100, Abcam), anti-Iba1 (Goat, 1:500, Wako) and anti-GFP (Rabbit, 1:500, MBL life science). Sections were then washed 3 times with TBS for 10 min and incubated with secondary antibodies for 2 h at room temperature. The secondary antibodies used in this study were donkey anti-mouse IgG Alexa 488 (1:1000, Invitrogen), goat anti-rabbit IgG Alexa 488 (1:1000, Invitrogen), donkey anti-rabbit IgG Alexa 647 (1:1000, Invitrogen), goat anti-Rat IgG Alexa 594 (1:1000, Invitrogen), donkey anti-goat IgG Cy3 (1:200, Jackson ImmunoResearch) and donkey anti-sheep IgG NL557 (1:200, R&D Systems). After washing with TBS for 10 min, sections were incubated with DAPI (1:10000, Sigma) in TBS for 5 min and then washed twice with TBS for 15 min. For NeuN staining, sections were not incubated with DAPI. For AT8 and Thioflavin S co-staining, sections were incubated with secondary antibody and then stained with 0.1% Thioflavin S in 50% ethanol for 8 min. Sections were finally mounted on microscope slides and visualized using a confocal microscope (FV3000-L4EN-TN21; Olympus, Japan). Specific information on antibodies used in this study can be seen in Supplementary Table 1. Image analysis of immunohistochemical staining Microscopy images were analyzed using ImageJ software. For area-based staining analysis (e.g., Aβ, AT8 and pT181 staining), immunoreactivity was quantified as the percentage (%) of the area covered by the positively stained regions. The data were expressed as positive staining area (%) ± SEM per group. For cell-counting analyses (GFAP staining), data were presented as the number of positive cells ± SEM per group. Additionally, for p-TAK1 and p21 staining, data were reported as the percentage (%) of positive cells ± SEM per group. To quantify staining intensity or density within specific cells (e.g., C3), staining area was measured individually in selected cells using ImageJ, and data were expressed as the intracellular stained area (%) ± SEM per group. More than 12 random cells were selected from four randomly acquired fields per mouse for analysis. For AT8 and pT181 staining, positively stained cells were counted as p-Tau-positive cells. The data were presented as AT8- or pT181-positive cell number per section ± SEM per group. For counting DCX-positive and BrdU-positive cells, the entire dentate gyrus (DG) region was analyzed through the complete z-axis (1.0 μm interval per image), and all images were collected. These images were processed in ImageJ through Image > Stacks > Images to Stack and Image > Stacks > Z-Projection > Max Intensity to obtain a single image representing the entire z-stack. Cell counts were converted into the total number of cells in the DG region based on standardized DG lengths. For Tom20 and DCX co-staining, more than 10 randomly selected DCX-positive cells per mouse were analyzed, and data were presented as Tom20-positive area size within DCX-positive cells ± SEM per group. For all immunohistochemical analyses, at least three randomly acquired fields per brain section were considered, and at least two brain sections between Bregma − 1.70 and − 2.70 were analyzed per mouse. Quantification of neuron density Neuron density in the hippocampal CA1 region was quantified using three brain sections per mouse (at Bregma − 1.4, − 1.7, and − 2.0 mm). For the dentate gyrus (DG) region, one section per mouse (at Bregma − 2.0 mm) was used. This approach minimized variations caused by sampling from different hippocampal positions. NeuN immunostaining was performed to identify mature neurons. The number of NeuN-positive cells was counted, and neuron density was presented as the number of NeuN-positive cells per mm² ± SEM per group. Sample preparation for scanning electron microscopy For scanning electron microscopy (SEM) observation, mice were perfused with TBS and fixed with 2% PFA and 2% glutaraldehyde buffered with 0.1 M phosphate buffer (PB), (pH 7.2). Fixed brains were cut into 0.5–1 mm sections and hippocampal regions were obtained. Sections were then treated with 2% OsO4 in PB for 2 h, dehydrated with a graded series of ethanol and embedded in epoxy resin (EPOK812, Okenshoji, Tokyo, Japan). Samples were sectioned into 90 nm-thick slices using an ultramicrotome (Leica UC7) and then stained with uranyl acetate and lead citrate. After coating, the sections were observed by an electron microscope (Regulus 8240, Hitachi High-Tech Corporation, Tokyo, Japan). For the cell samples, fixed cells were treated with 1% OsO4 in PB for 2 h and 0.5% uranyl acetate for 30 min. Then the samples were dehydrated with a graded series of ethanol and embedded in epoxy resin, cut into 300 nm-thick sections with the ultramicrotome and stained for SEM observation as describe above. Preparation for in-resin correlative light and electron microscopy In-resin correlative light and electron microscopy (CLEM) was performed as previously described [[90]46, [91]47]. Briefly, mice were perfused with TBS and fixed with 4% PFA and 0.25% glutaraldehyde buffered with 0.1 M PB. After post-fixation overnight with the same solution, brains were stored in 0.1 M PB at 4 °C before use. The brains were then sliced into 50 μm-thick sections by semi-automatic Leica VT1200 vibrating blade microtome and stored in 0.05% sodium azide in 0.1 M PB at 4 °C. Brain sections were stained with primary and secondary antibodies, then post-fixation with 2% PFA and 2% glutaraldehyde in 0.1 M PB for at least 1 h. Then brain sections were treated with 1% OsO4 in PB for 15 min, dehydrated with a graded series of ethanol and embedded in epoxy resin (EPOK812, Okenshoji, Tokyo, Japan). Samples were cut into 300 nm-thick sections with the ultramicrotome (Leica UC7) and observed by Nikon A1RHD25 confocal laser-scanning microscope to identify the target cells. Finally, the same sections were stained with uranyl acetate and lead citrate and observed using an electron microscope (Regulus 8240, Hitachi High-Tech Corporation, Tokyo, Japan). Isolation of astrocytes and newborn neurons Magnetic-activated cell sorting (MACS) was used to isolate astrocytes and newborn neurons from the mouse brain. Mice were sacrificed under deep anesthesia, and brains were removed carefully and washed in cold D-PBS to eliminate residual blood. The cortex and hippocampus were dissected and enzymatically dissociated into a single-cell suspension using the Adult Brain Dissociation Kit (Miltenyi Biotec), following the manufacturer’s instructions. For newborn neuron isolation, only the hippocampus was used. Astrocytes and newborn neurons were isolated using the Anti-ACSA-2 MicroBead Kit and Anti-PSA-NCAM MicroBead Kit (Miltenyi Biotec), respectively. Briefly, the single-cell suspension was first incubated with a mouse Fcγ receptor blocking reagent at 4 °C for 10 min, followed by incubation with Anti-ACSA-2 or Anti-PSA-NCAM MicroBeads at 4 °C for 15 min. Target cells were then collected using MACS with MS columns (Miltenyi Biotec). Adeno-associated virus preparation AAV-shPEPT2 and AAV-scramble RNA were packaged in HEK293T cells using the TransIT-VirusGEN^® Transfection Reagent (Mirus) and purified with the AAVpro^® Purification Kit (Takara Bio, Japan), according to the manufacturer’s instructions. The pAAV-U6-shPEPT2-EGFP vector (expressing shRNA targeting Slc15a2) and the pAAV-U6-scramble RNA vector were obtained from VectorBuilder. The pHelper vector was a generous gift from Dr. Haruo Okado (Tokyo Metropolitan Institute of Medical Science), and the pUCmini-iCAP-PHP.eB vector was obtained from Addgene (USA). Intracerebroventricular injection Intracerebroventricular (ICV) injection was performed at approximately 6.5 months of age. TE4 mice were anesthetized with a mixed narcotic solution containing medetomidine (0.3 mg/kg), midazolam (4 mg/kg), and butorphanol tartrate (5 mg/kg), administered intraperitoneally (i.p.). Each mouse received a 4 µl injection into both lateral ventricles of either AAV-PHP.eB-U6-shPEPT2-EGFP or AAV-PHP.eB-U6-scramble_RNA-EGFP. The stereotaxic coordinates used were: anterior-posterior (AP) = − 0.4 mm, medial-lateral (ML) = ± 1.0 mm, and dorsal-ventral (DV) = − 2.0 mm. Injections were performed using an automated nanoliter injector (Nanoject II, Drummond Scientific, USA), and the needle was kept in place for at least 5 min after injection to prevent backflow of the AAV solution. Mice were allowed to recover for 2 weeks, and anserine treatment began at 7 months of age as described above. Mouse striatal precursor-1 (MSP-1) cell culture The MSP-1 cell line, derived from neural stem cells of the ventral telencephalon of p53 knockout mice, was obtained as previously described [[92]48]. MSP-1 cells were cultured on poly-L-lysine/fibronectin-coated dishes (Sigma) in DMEM/F12 medium (Gibco) supplemented with 10% fetal bovine serum (FBS; Gibco) and 10 ng/ml basic fibroblast growth factor (bFGF; Wako). This medium is referred to as MSP-1 proliferation medium. The culture medium was refreshed every 2–3 days. To obtain differentiated astrocytes, MSP-1 cells were first seeded in proliferation medium overnight, then switched to astrocyte differentiation medium consisting of serum-free DMEM/F12 supplemented with N-2 supplement (R&D Systems), 10 ng/ml leukemia inhibitory factor (LIF; Wako), and 10 ng/ml ciliary neurotrophic factor (CNTF; Wako). The medium was changed every 2–3 days. On day 5, cells were stimulated with either 100 ng/ml tumor necrosis factor-α (TNF-α; Wako) or 100 ng/ml interleukin-1β (IL-1β; MedChemExpress) for 24 h. For the anserine treatment group, cells were pre-treated with L-anserine (Toronto Research Chemicals) before exposure to TNF-α or IL-1β. To obtain differentiated neurons, MSP-1 cells were cultured in neuron differentiation medium consisting of serum-free DMEM/F12 supplemented with N-2 supplement. The medium was replaced every 2 days. On day 5, cells were treated for 24 h with either the conditioned medium from astrocyte-differentiated MSP-1 cells or recombinant mouse C3/C3a protein (MedChemExpress). For blocking experiments, necrostatin-1 (Nec-1) and the C3a receptor antagonist SB290157 (Selleck) were used. Reagent details are listed in Supplementary Table 2. Kinase activity assay The potential targets of anserine within the NF-κB signaling pathway, including IKK-α, IKK-β, IRAK1, IRAK4, and TAK1-TAB1 were identified using an activity-based biochemical screening/profiling assay (CARNA BIOSCIENCES, Kobe, Japan). Briefly, anserine was mixed with a solution containing substrate, ATP, metal ions, and additives, followed by the addition of kinase solution. The mixture was incubated in a polypropylene 384-well microplate at room temperature for 1 h. After the reaction was terminated by adding termination buffer, the samples were analyzed using the BioPhase™ 8800 system (AB Sciex), which separates and quantifies product and substrate peptide peaks. The extent of kinase inhibition was evaluated based on the product ratio calculated from the peak heights. Initially, three concentrations of anserine (2.5 mM, 5 mM, and 10 mM) were used in a simple inhibition assay to screen for potential targets (Supplementary Table 4). Based on the results, TAK1–TAB1 and IRAK1 were identified as the primary candidates and selected for further analysis. Their IC₅₀ values were determined using a 10-point dose–response assay (Fig. [93]2B and C). Fig. 2. [94]Fig. 2 [95]Open in a new tab Anserine inhibited TAK1- and IRAK1-mediated NF-κB activation in cultured astrocytes. A Schematic overview and representative images of the experimental design used to evaluate the effect of the natural anti-inflammatory dipeptide anserine on astrocyte activation. MSP-1 astrocytes were differentiated from neural stem-like MSP-1 cells by stimulation with CNTF and LIF for 5 days in vitro (scale bar, 50 μm). Astrocyte activation was induced by TNF-α or IL-1β (100 ng/ml, 24 h), confirmed by increased nuclear p-p65 immunoreactivity (scale bar, 10 μm). To study the effects of anserine on NF-κB activation, TNF-α or IL-1β stimulation was performed for 30 min. Anserine was administered before the stimulation. Except for C3 ELISA (D and E), anserine was used at 10 mM for all experiments. B and C In vitro kinase activity curves of TAK1-Table 1 (B) and IRAK1 (C) treated with L-anserine. D Quantification of C3 concentration in culture medium of Sham, IL-1β and IL-1β + Ans (0.8, 2, 10, and 50 mM) groups (n = 6–8), with comparisons made versus the IL-1β group. E Quantification of C3 concentration in culture medium of Sham, TNF-α and TNF-α + Ans (0.8, 2, 10, and 50 mM) groups (n = 6–8), with comparisons made versus the TNF-α group. F Quantification of nuclear p-p65 intensity of Sham, IL-1β and IL-1β + Ans (10 mM) groups (n = 116–140), with comparisons made versus the IL-1β group. G Quantification of nuclear p-p65 intensity of Sham, TNF-α and TNF-α + Ans (10 mM) groups (n = 103–109), with comparisons made versus the TNF-α group. H Quantification of p-IRAK1 fluorescence intensity in Sham, IL-1β and IL-1β + Ans (10 mM) groups (n = 42–50), with comparisons made versus the IL-1β group. I Quantification of p-IRAK1 fluorescence intensity Sham, TNF-α and TNF-α + Ans (10 mM) groups (n = 39–52), with comparisons made versus the TNF-α group. J Quantification of p-TAK1 fluorescence intensity in Sham, IL-1β and IL-1β + Ans (10 mM) groups (n = 34–44), with comparisons made versus the IL-1β group. K Quantification of p-TAK1 fluorescence intensity in Sham, TNF-α and TNF-α + Ans (10 mM) groups (n = 52–54), with comparisons made versus the TNF-α group. L GSEA of the Hallmark “TNF-α signaling via NF-κB” gene set. Ans, Anserine. M Volcano plots of differentially expressed genes (DEGs) (|log₂ fold change| >0.5, p < 0.05): left, TNF-α vs. Sham; right, TNF-α + Ans vs. TNF-α. Up-regulated genes are shown in red and down-regulated genes are shown in blue. Data are presented as mean ± SEM (D and E) or presented in violin plots (F–K) (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) and were analyzed by one-way ANOVA with Dunnett’s multiple comparisons test (D and E) or Kruskal-Wallis test with Dunn’s multiple comparisons test (F–K). All comparisons were made versus the TNF-α group in TNF-α–stimulated MSP-1 astrocytes or the IL-1β group in IL-1β–stimulated MSP-1 astrocytes. **p < 0.01, ****p < 0.0001. Some elements in panel A were created with BioRender.com Protein isolation For isolation of brain proteins, mice were perfused with TBS and hippocampal tissues were dissected out. Tissues were lysed on ice for 1–2 h in RIPA buffer containing 1 mM PMSF (Cell Signaling), Protease Inhibitor Cocktail (Wako), and Phosphatase Inhibitor Cocktail (Wako). For cell samples, harvested cells were lysed in the same RIPA buffer on ice for 30 min. Then, the lysates were centrifuged at 15,000 g for 30 min at 4 °C and the supernatants were collected (RIPA-soluble fraction). Protein samples were stored at −80℃ before use. Western blot Equal amounts of isolated proteins (20–40 µg) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad). After transfer, the blotting membranes were blocked with EveryBlot blocking buffer (BioRad) with shaking for 5 min at room temperature. Subsequently, blocked membranes were incubated with diluted primary antibodies at 4 °C overnight. The primary antibodies used in this study were anti-α-Tubulin (Mouse, 1:2000, Proteintech), anti-GAPDH (Rabbit, 1:1000, Cell Signaling), anti-p-IRAK1 (Rabbit, 1:1000, Assay BioTech), anti-IRAK1 (Rabbit, 1:1000, Cell Signaling), anti-p65 (Rabbit, 1:1000, Cell Signaling), anti-p-p65 (Rabbit, 1:1000, Cell Signaling), anti-TAK1 (Rabbit, 1:500, Cell Signaling), anti-p-TAK1 (Rabbit, 1:500, Cell Signaling), anti-IKK (Rabbit, 1:1000, Proteintech) and anti-p-IKK (Rabbit, 1:500, Cell Signaling). All the membranes were washed with 0.05% TBS-T for 5 cycles for 5 min each, and then incubated with diluted secondary antibodies for 2 h at RT. The secondary antibodies used in this study were HRP-conjugated Goat Anti-Mouse IgG (1:10000, Proteintech) and HRP-conjugated Goat Anti-Rabbit IgG (1:10000, Proteintech). Afterwards, the membranes processed the same washing procedure as before. Finally, Clarity Max Western ECL Substrate (Bio-Rad) and Image Gauge version 3.41 (Fujifilm, Japan) were used to detect the signals. Relative expression of target protein was analyzed by ImageJ using α-Tubulin or GAPDH as loading controls. All the band images were quantified under identical background conditions. The information on antibodies used in this study can be seen in Supplementary Table 1. Immunocytochemical staining Harvested MSP-1 cells were cultured in poly-L-lysine/fibronectin-coated chambered glass slides (4-well or 8-well; Watson) using MSP-1 differentiation medium to induce astrocyte or neuronal differentiation. After treatment, cells were fixed with 4% PFA for 10 min, followed by washing with TBS for 10 min. Cells were then blocked with 5% BSA diluted in 0.05% TBS-T for 30 min at room temperature and incubated overnight at 4 °C with primary antibodies diluted in blocking solution. The primary antibodies used in this study were anti-AT8 (Mouse, 1:1000, Thermo Fisher), anti-PSA-NCAM (Mouse, 1:1000, gift from Dr. Seki), anti-MAP2 (Rabbit, 1:1000, Proteintech), anti-p-p65 (Rabbit, 1:500, Cell Signaling), anti-p-IKK (Rabbit, 1:1000, Cell Signaling), anti-p-TAK1 (Rabbit, 1:1000, Cusabio), anti-p-IRAK1 (Rabbit, 1:1000, Assay BioTech) and anti-p-MLKL (Rabbit, 1:2000, Abcam). After incubation, cells were washed three times in TBS for 5 min each, followed by incubation with appropriate secondary antibodies for 2 h at room temperature. The secondary antibodies used in this study were donkey anti-mouse IgG Alexa 488 (1:1000, Molecular Probes), goat anti-mouse IgM Alexa 488 (1:1000, Abcam) and donkey anti-rabbit IgG Alexa 568 (1:1000, Molecular Probes). After washing with TBS for 5 min, cells were incubated with DAPI (1:10000, Sigma) in TBS for 5 min, followed by two additional washes with TBS for 5 min each. Cells were visualized using a confocal microscope (FV3000-L4EN-TN21; Olympus, Japan). Specific information on antibodies used in this study can be seen in Supplementary Table 1. Image analysis of immunocytochemical staining For AT8 and p-MLKL staining, signal intensity was quantified across the whole image field and presented as relative intensity (ratio to Sham) ± SEM per group. For p-TAK1, p-IRAK1, and p-IKK staining, signal intensity was measured on a per-cell basis and presented as relative intensity (ratio to Sham) ± SEM per group. For p-p65 staining, only nuclear signal intensity was quantified, and results were presented as relative intensity (ratio to Sham) ± SEM per group. Mitochondrial function assay Mitochondrial membrane potential, morphology, and oxidative stress in MSP-1-derived neurons were assessed using the MT-1 MitoMP Detection Kit (Dojindo), MitoBright IM Red (Dojindo), and the ROS Assay Kit–Photo-oxidation Resistant DCFH-DA– (Dojindo), according to the manufacturer’s instructions. Briefly, after conditioned medium (CM) treatment, cells were incubated with the respective reagents for 30 min and then washed with HBSS. Cells were subsequently fixed with 4% PFA and visualized using a confocal microscope (FV3000-L4EN-TN21; Olympus, Japan). Acquired images were analyzed using ImageJ. RNA sequencing Total RNA was extracted using the RNeasy Mini Kit (Qiagen) from three sources: isolated astrocytes, isolated newborn neurons, and MSP-1-derived differentiation-induced astrocytes. 1 ng of RNA (from isolated astrocytes and newborn neurons) or 10 ng of RNA (from cultured astrocytes) was processed using the SMART-seq Stranded Kit (Takara Bio, Japan). Sequencing was performed on an Illumina NovaSeq 6000 platform (Illumina, USA). Transcript abundance was quantified as transcripts per million (TPM). Differentially expressed genes (DEGs) were identified using the DESeq2 package in the R environment. DEGs were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis using DAVID ([96]https://david.ncifcrf.gov/). Gene Set Enrichment Analysis (GSEA) was conducted based on TPM values using GSEA software (UC San Diego and Broad Institute). All visualizations were generated using the ggplot2 package in R. Reverse transcription quantitative PCR (RT-qPCR) Total RNA extracted from isolated astrocytes and newborn neurons (0.5–1 ng, low input due to limited cell numbers) was reverse transcribed into cDNA using SuperScript™ III Reverse Transcriptase (Thermo Fisher), according to the manufacturer’s instructions. Quantitative PCR (qPCR) was performed on a TP-850 system (Takara Bio, Japan) using TB Green^® Premix Ex Taq (Takara Bio, Japan). To accommodate the low RNA input, the amplification protocol was modified to include an initial denaturation at 95 °C for 30 s, followed by 50 cycles of 95 °C for 5 s and 60 °C for 30 s. All samples were run in duplicate, and β-actin was used as the reference gene. Relative gene expression was calculated using the 2^−ΔΔCt method and expressed as a fold change. Primer sequences were designed using Primer3 software and are listed in Supplementary Table 3. Statistical analyses All data are presented as mean ± SEM, as violin plots (median: bold dashed line; quartiles: dashed lines) or as box plots (line at median), and were analyzed using GraphPad Prism (version 8; GraphPad Software). Normality was assessed with the Shapiro–Wilk test in combination with Q–Q plots. For normally distributed data, homogeneity of variance was examined using the F-test (two groups) or Brown–Forsythe test (more than two groups). When variances were equal, comparisons between two groups were made using a two-tailed Student’s t-test, while comparisons among more than two groups were made using one-way ANOVA followed by Dunnett’s multiple comparisons test (in vitro studies, comparison with the pathological group) or Tukey’s post hoc test (in vivo studies). For datasets across multiple time points or conditions, two-way ANOVA with Tukey’s post hoc test was used. When data were not normally distributed, the Mann–Whitney test (two groups) or Kruskal–Wallis test with Dunn’s multiple comparisons test (more than two groups) was applied. When variances were unequal, Welch’s ANOVA followed by Dunnett’s T3 multiple comparisons test was used. A p-value less than 0.05 was considered statistically significant. Results E4-3Tg mice exhibited accelerated mortality, promoted astrocyte activation, and exacerbated Tau pathology Firstly, we compared the survival rate between E4-3Tg mice and TE4 mice, hypothesizing that Aβ pathology would accelerate mortality at an earlier age. Both E4-3Tg and TE4 mice exhibited increased mortality compared to Control mice (Fig. [97]1A). Notably, E4-3Tg mice showed significantly accelerated mortality compared to TE4 mice, and this difference remained significant up to 28 weeks of age (Fig. [98]1A; Fig. S1A). The median survival dropped from 40 weeks (approximately 9 months) in TE4 mice to 28.5 weeks in E4-3Tg mice (approximately 6.5 months; p < 0.0001; Fig. [99]1A). No sex differences in survival were observed in E4-3Tg mice (Fig. S1C). Regarding physical performance, a subset of E4-3Tg mice developed typical tauopathy-related paralysis, which in TE4 mice was usually observed at 9 months of age (Fig. [100]1B; Supplementary Video 1). Furthermore, a downward trend in body weight and motor function was observed at 6.5 months of age in E4-3Tg mice (Fig. S2A), while TE4 mice exhibited severe motor impairment and weight loss at 9 months of age (Fig. S2B), consistent with previous reports [[101]13]. Fig. 1. [102]Fig. 1 [103]Open in a new tab Accelerated mortality accompanied by exacerbated tau pathology and astrocyte activation in human APOE4 KI APPswe/PSEN1dE9/P301S-Tau model mice. A Kaplan-Meier survival curves of Control, TE4 and E4-3Tg mice (Control, n = 32; TE4, n = 38; E4-3Tg, n = 34). Kaplan–Meier survival analysis was performed using the log-rank (Mantel–Cox) test. Significant differences were observed among the three groups (p < 0.0001). Pairwise comparisons: Control and TE4, p < 0.0001; Control and E4-3Tg, p < 0.0001, TE4 and E4-3Tg, p < 0.0001. B Representative images of Control, E4-3Tg and TE4 mice at either 6.5 or 9 months of age (scale bar, 1 cm). C Results of the Y-maze test of Control, TE4 and E4-3Tg mice (n = 7). The left panel is schematic of the Y-maze test protocol. Mice were introduced from arm A. The results are shown as alternation rates (%). D Result of the CFC test of Control, TE4 and E4-3Tg mice (n = 6). The left graph is schematic of the CFC test. The results are shown as freezing time during the contextual test. E Representative images of AT8 staining in the CA1 region of TE4 and E4-3Tg mice at 6.5 months of age (scale bar, 25 μm). F Representative images of NeuN staining in the CA1 region of Control, TE4 and E4-3Tg mice at 6.5 months of age (scale bar, 25 μm). G Quantification of AT8-positive area in the CA1 region in each group (n = 4). H Quantification of neuronal density in the CA1 region in each group (n = 4). I Representative images of GFAP and p-TAK1 co-staining in the hippocampus region of Control, TE4 and E4-3Tg mice at 6.5 months of age (scale bar, 10 μm). J Quantification of the number of GFAP-positive cells in each group (n = 4). K Quantification of the proportion of p-TAK1^+/GFAP^+ cells in each group (n = 3–4). Data are presented as mean ± SEM and were analyzed by one-way ANOVA with Tukey’s multiple comparisons test (C, D, H, J and K) or Student’s t-test (G). ns, not significant, *p < 0.05, **p < 0.01 To evaluate the cognitive function of E4-3Tg mice at 6.5 months of age, we conducted the Y-maze and contextual fear conditioning (CFC) tests. As a result, E4-3Tg mice showed a significantly reduced alternation rate in the Y-maze test and decreased freezing time in the CFC test compared with Control mice, while TE4 mice did not exhibit such cognitive deficits at this age (Fig. [104]1C, D; n = 6–7 per group). We also observed that E4-3Tg mice exhibited a trend toward more severe cognitive impairment in CFC compared with TE4 mice (Fig. [105]1D). Immunostaining showed increased AT8 immunoreactivity (p-Tau at Ser202 and Thr205) and increased neuronal loss in the hippocampal CA1 region compared with TE4 mice (Fig. [106]1E–H; n = 4 per group). Importantly, we observed the co-localization of Thioflavin S and AT8-positive tau (Fig. S3). These results suggested earlier pathological tau accumulation and neurodegeneration in E4-3Tg mice. Additionally, E4-3Tg mice exhibited a severe impairment in adult hippocampal neurogenesis (AHN), with a sharp reduction in the number of DCX-positive immature neurons (Fig. S4A, B). This phenotype was not observed in TE4 mice, even at 9 months of age (Fig. S4C, D). We further assessed astrocyte activation by GFAP and p-TAK1 (MAP3K7) co-staining. TAK1 is a key regulator of NF-κB signaling downstream of both TNFR and IL-1R/TLR pathways [[107]49], which can induce A1 astrocyte activation in response to TNF-α or IL-1β [[108]2]. At 6.5 months of age, E4-3Tg mice exhibited a significantly increased number of GFAP-positive cells compared with Control mice and an increased trend compared with TE4 mice (Fig. [109]1I, J; n = 4 per group). Importantly, the proportion of p-TAK1⁺/GFAP⁺ cells was significantly higher in E4-3Tg mice than in both Control and TE4 mice (Fig. [110]1K; n = 3–4 per group). These results indicated accelerated A1 astrocytic activation in E4-3Tg mice. By contrast, TAK1 activation was not evident in microglia at this age (Fig. S5). Collectively, our findings through the E4-3Tg mouse model revealed that Aβ pathology accelerated mortality in the P301S tauopathy model, accompanied by enhanced A1 astrocyte activation and exacerbated tau-associated neurodegeneration. Dual inhibitory effects of Anserine on TAK1- and IRAK1-mediated A1 astrocyte activation in vitro Previous studies showed the beneficial effects of inhibition of TAK1 signaling in ameliorating the innate immune response and NF-κB activation [[111]50], and this was further confirmed in cultured astrocytes [[112]51]. As E4-3Tg mice exhibited a significantly increased astrocytic TAK1 activation, we considered TAK1 to be a potential target to effectively block astrocyte activation. Previously, we reported that anserine, an imidazole dipeptide that is actively taken up by astrocytes [[113]39], can inhibit astrocyte activation [[114]43]. We therefore tested the role of anserine in TAK1-mediated astrocyte activation. We established an in vitro cultured astrocyte model using an immortalized neural stem cell line, MSP-1, as we previously described [[115]48, [116]52–[117]55]. Consistent with a previous study [[118]56], MSP-1 astrocytes could be strongly activated by TNF-α or IL-1β, evidenced by p-p65 staining (Fig. [119]2A). Next, we performed kinase assays to investigate the potential inhibitory effects of anserine on kinases involved in inflammation-stimulated NF-κB activation. Although anserine showed a dose-dependent inhibitory effect on TAK1, the IC[50] was relatively high (Fig. [120]2B). In contrast, anserine exhibited a stronger inhibitory effect on IRAK1 (Interleukin-1 receptor-associated kinase 1) (Fig. [121]2C), a key kinase that functions upstream of TAK1 in the IL-1R–NF-κB pathway. These results indicated that anserine exerted inhibitory effects on both IRAK1 and TAK1, suggesting its dual inhibitory role in inhibiting A1 astrocyte activation via the TAK1/IRAK1 signaling axis. We then confirmed this dual inhibitory effect of anserine in both IL-1β–stimulated and TNF-α–stimulated MSP-1 astrocytes. We first assessed the effect of anserine on the production of A1 astrocyte marker C3 over a concentration range of 0.4 to 50 mM, within which we confirmed the low cytotoxicity of anserine (Fig. S6). MSP-1 astrocytes pre-treated with anserine showed a significantly reduced C3 production beginning at 2 mM in IL-1β–stimulated MSP-1 astrocytes (Fig. [122]2D; n = 6–8 per group), and reduced C3 production beginning at 10 mM in TNF-α–stimulated MSP-1 astrocytes (Fig. [123]2E; n = 6–8 per group). The relatively stronger inhibitory effect in IL-1β–stimulated MSP-1 astrocytes was consistent with our findings in kinase assays. We selected the concentration of 10 mM for subsequent experiments because 10 mM was sufficient to reduce C3 production (Fig. [124]2D, E). Immunostaining revealed that anserine significantly reduced nuclear p-p65 intensity in both IL-1β– and TNF-α–stimulated MSP-1 astrocytes (Fig. [125]2F, G), indicating the effective suppression of NF-κB activation. This inhibitory effect resulted from the suppressed IRAK1–TAK1–IKK signaling in IL-1β–stimulated MSP-1 astrocytes and suppressed TAK1–IKK signaling in TNF-α–stimulated MSP-1 astrocytes, evidenced by both immunostaining and western blot (Fig. [126]2H–K; Fig. S7; Fig. S8). To further confirm this, we performed RNA sequencing (RNA-Seq) analysis using samples of TNF-α–stimulated group (n = 4 per group) because MSP-1 astrocytes exhibited a stronger response to TNF-α than IL-1β. Through gene set enrichment analysis (GSEA), we confirmed suppressed NF-κB activation in anserine-pretreated TNF-α–stimulated MSP-1 astrocytes (Fig. [127]2L). We also observed decreased expression of pro-inflammatory factors such as C3, Ccl2 and Ccl7 (Fig. [128]2M). Taken together, our results demonstrated the dual inhibitory effect of anserine on astrocytic TAK1- and IRAK1-mediated NF-κB activation in MSP-1 astrocytes. Consequently, inhibited NF-κB activation led to reduced production of pro-inflammatory factors including C3 (Fig. S7D; Fig. S8C). Inhibition of astrocyte activation and C3 over-production alleviated tau-associated neuronal mitochondrial dysfunction in differentiating neurons Then, we evaluated the effects of MSP-1 astrocyte activation on neurons. We induced MSP-1 cells to differentiate into neurons (Fig. S9) and treated them with cultured medium (CM) derived from TNF-α–stimulated MSP-1 astrocytes, which contained relatively elevated C3 levels (Fig. [129]3A). MSP-1 neurons treated with CM from the TNF-α group (TNF CM) exhibited significantly increased AT8 immunoreactivity (Fig.[130]3B, C; n = 3–5). Given that pathological p-Tau has been reported to induce mitochondrial elongation along with mitochondrial dysfunction in both Drosophila neurons [[131]57] and mouse neurons [[132]58], we next evaluated mitochondrial morphology and function in MSP-1 neurons. MitoBright staining and scanning-electron microscopy (SEM) revealed severe mitochondrial swelling in MSP-1 neurons exposed to TNF CM (Fig. [133]3D, E; Fig. S10). Consistently, measurement of MT-1 and ROS indicated reduced mitochondrial membrane potential and elevated oxidative stress in TNF CM-treated MSP-1 neurons (Fig. [134]3F–I; n = 4–6). All of these were indicative of tau-associated mitochondrial dysfunction. These tau-associated neuronal mitochondrial defects were significantly attenuated when MSP-1 neurons were exposed to CM from the TNF-α + Ans group (TNF + Ans CM) or when C3a signaling was blocked with a selective C3a receptor antagonist (TNF CM-C3aRA) (Fig. [135]3B–I; Fig. S10). Mitochondrial dysfunction is linked to activation of the cell-death pathway [[136]59]. Correspondingly, we observed an increased intensity of p-MLKL (necroptosis marker) in MSP-1 neurons treated with TNF CM (Fig. S11). In contrast, MSP-1 neurons treated with TNF + Ans CM or TNF CM-C3aRA exhibited significantly reduced p-MLKL intensity, comparable to MSP-1 neurons treated with Necrostatin-1 (Nec-1), a necroptosis inhibitor (Fig. S11). Fig. 3. [137]Fig. 3 [138]Open in a new tab Inhibition of A1 astrocyte activation alleviated mitochondrial dysfunction in differentiating neurons. A Schematic overview of the in vitro study using differentiating MSP-1 newborn neurons. MSP-1 neurons were treated with cultured medium (CM) collected from MSP-1 astrocytes for 24 h. B Representative images of AT8 staining of MSP-1 neurons treated with TNF CM and TNF+Ans CM (scale bar, 10 μm). C Quantification of AT8 fluorescence intensity in each group, (n = 3–5). D Representative electron microscopy images of MSP-1 neurons treated with Sham, TNF CM, TNF+Ans CM and TNF CM-C3aRA (scale bar, 2 μm (up), 1 μm (down)). N, nucleus, MT, normal mitochondria (blue), sMT, swollen mitochondria (yellow). E Quantification of mitochondrial length and the ratio of swollen mitochondria in each group (neuron number, 15–18; mitochondria number, 45–49). F Representative images of MT-1 staining of MSP-1 neurons treated with Sham, TNF CM, TNF+Ans CM and TNF CM-C3aRA (10 μM) (scale bar, 25 μm). G Quantification of MT-1 fluorescence intensity in each group (n = 4–6). H Representative images of ROS staining of MSP-1 neurons treated with Sham, TNF CM, TNF+Ans CM and TNF CM-C3aRA (10 μM) (scale bar, 25 μm). I Quantification of ROS fluorescence intensity in each group (n = 5–6). Data are presented as mean ± SEM (C, G and I) or presented as violin plots (E) (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) and were analyzed by one-way ANOVA with Dunnett's multiple comparisons test (C, G and I) or Kruskal-Wallis test with Dunn's multiple comparisons test (E). All comparisons made were versus the TNF CM group. *p < 0.05, **p < 0.01, ***p < 0.001. Some elements in panel A were created with BioRender.com To further confirm C3 as the upstream trigger in this pathology, we directly applied recombinant C3 protein to MSP-1 neurons (Fig. S12A). C3 treatment alone also induced tau-associated mitochondrial dysfunction in MSP-1 neurons, resulting in increased p-MLKL expression, which was significantly alleviated by C3aRA treatment (Fig. S12B–M). Collectively, our results indicated that suppression of C3 production derived from activated A1 astrocytes alleviated tau-associated neuronal mitochondrial dysfunction, and subsequently prevented the activation of the necroptosis pathway in MSP-1 neurons. Inhibition of astrocyte activation reduced mortality and cognitive impairment in E4-3Tg mice Our in vitro study demonstrated the neurotoxic effects of activated A1 astrocytes, and showed that anserine acts as a neuroprotective compound through dual inhibition of IRAK1/TAK1 signaling in astrocytes. Since the accelerated mortality observed in E4-3Tg mice was accompanied by astrocytic TAK1 activation and exacerbated tau-associated neurodegeneration, we hypothesized that anserine administration could inhibit astrocyte activation and mitigate this phenotype. Following our previous protocol [[139]43], we treated E4-3Tg mice with anserine (2 g/L in drinking water), from 4.5 to 6.5 months of age (8 weeks), generating the anserine-treated E4-3Tg mice, simply called E4-3Tg (A) mice (Fig. [140]4A). We observed that E4-3Tg (A) mice rarely died before 28 weeks (1/20 in E4-3Tg (A) mice vs. 10/24 in E4-3Tg, p < 0.01), and exhibited a survival rate comparable to that of Control mice (Fig. [141]4B). This survival benefit was observed in both sexes (Fig. S13A, B). Though E4-3Tg mice did not exhibit significant weight loss and motor impairment compared to Control mice (with a downward trend, p < 0.1), anserine treatment restored both body weight (Fig. [142]4C; n = 11 per group) and motor function (Fig. [143]4D; n = 6–7 per group) to levels comparable to those of Control mice. Additionally, E4-3Tg (A) mice showed improved alternation behavior in the Y-maze test (Fig. [144]4E, n = 6–8 per group) and improved contextual memory in the CFC test (Fig. [145]4F, n = 10 per group) compared with untreated E4-3Tg mice. Fig. 4. [146]Fig. 4 [147]Open in a new tab Inhibition of A1 astrocyte activation reduced mortality and improved cognitive function in E4-3Tg mice. A Schematic overview of the in vivo study using E4-3Tg model mice. An 8-week oral anserine treatment (2 g/L) was administered starting at 4.5 months of age. B Kaplan–Meier survival curves of Control, E4-3Tg and E4-3Tg (A) mice (Control, n = 16; E4-3Tg; n = 24, E4-3Tg (A), n = 20). Kaplan-Meier survival analysis was performed by log-rank (Mantel-Cox) test. Among the three groups (p = 0.0004). Pairwise comparisons: Control and E4-3Tg, p = 0.0041; E4-3Tg and E4-3Tg (A), p = 0.0051. C Quantification of body weight of 3 groups of mice at the end of behavioral testing (n = 11). D Quantification of grip strength of 3 groups of mice at the end of behavioral testing (n = 6–7). E Quantification of alternation rates (%) in each group of mice during 10 min in the Y-maze test (n = 6–8). F Quantification of freezing time during the contextual test of CFC test (n = 10). G Representative images of GFAP, p-TAK1 and C3 co-staining in the hippocampus of Control, E4-3Tg and E4-3Tg (A) mice (scale bar, 10 μm). H Quantification of the proportion of p-TAK1^+/GFAP^+ cells in each group (n = 4). White Triangles indicate p-TAK1^+/GFAP^+ cells. I Quantification of C3-positive area in GFAP-positive cells in each group (n = 4). J Quantification of C3-positive area in p-TAK1^−/GFAP^+ cells and p-TAK1^+/GFAP^+ cells (n = 19–22). Data are presented in violin plots (J) (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) or presented as mean ± SEM (C, D, E, F, H and I) and were analyzed by Student’s t-test (J), one-way ANOVA with Tukey’s multiple comparisons test (C, D, E, H and I) or Kruskal-Wallis test with Dunn’s multiple comparisons test (F). *p < 0.05, **p < 0.01, ***p < 0.001 Then we investigated the changes in astrocyte activation in E4-3Tg mice. Due to antibody limitations, reliable p-IRAK1 signals could not be detected, making it infeasible to directly assess IRAK1 activation in astrocytes. Because IRAK1 activation is an upstream event that leads to TAK1 activation, we instead evaluated p-TAK1 signals in GFAP-positive astrocytes after anserine treatment. Anserine treatment significantly reduced the proportion of p-TAK1^+/GFAP^+ cells (Fig. [148]4G, H; n = 4 per group), accompanied by a markedly reduced immunoreactivity of the A1 astrocyte marker C3 (Fig. [149]4I; n = 4 per group). No change was observed in the proportion of p-TAK1⁺/Iba1⁺ cells (Fig. S14A, B). These results indicated the inhibition of A1 astrocyte activation after anserine treatment. Furthermore, quantitative colocalization showed that p-TAK1⁺ astrocytes exhibited higher C3 immunoreactivity (Fig. [150]4J), supporting the notion that reduced astrocytic C3 production resulted from anserine-mediated suppression of TAK1 activation. Considering that both IL-1R–IRAK1–TAK1 and TNFR–TAK1 axes are activated in P301S tau model mice for the increased production of TNF-α and IL-1β [[151]22], these findings provide indirect evidence that anserine effectively inhibited IRAK1/TAK1 signaling in astrocytes. Collectively, our results showed that inhibition of A1 astrocyte activation by anserine treatment reduced mortality and cognitive decline in E4-3Tg mice. Inhibition of astrocyte activation alleviated tau pathology and reduced neuronal loss in E4-3Tg mice To explore the underlying mechanisms of reduced mortality in E4-3Tg (A) mice, we next evaluated the effects of A1 astrocyte inhibition on neurodegeneration. We observed that E4-3Tg (A) mice exhibited significantly attenuated AT8 immunoreactivity and a reduced number of AT8-positive cells in the CA1 region (Fig. [152]5A, B; n = 4–5 per group). We did not find a significant change in the expression of Aβ (Fig. [153]5C, D; n = 5 per group), which was consistent with our previous study [[154]43]. Correspondingly, we observed increased neuronal density in the CA1 region of E4-3Tg (A) mice (Fig. [155]5E, F; n = 5–6 per group). Moreover, consistent with a previous study [[156]22], we also observed robust brain atrophy in E4-3Tg mice, which was attenuated in E4-3Tg (A) mice (Fig. S15A). Fig. 5. [157]Fig. 5 [158]Open in a new tab Inhibition of astrocyte activation attenuated tau pathology and reduced neuronal loss in E4-3Tg mice. A Representative images of AT8 staining of E4-3Tg and E4-3Tg (A) mice in the CA1 region (scale bar, 25 μm). B Quantification of AT8-positive area and AT8-positive cell number in the CA1 region in each group (n = 4–5). C Representative images of Aβ staining in the hippocampal CA1 and DG regions of E4-3Tg and E4-3Tg (A) mice (scale bar, 50 μm). D Quantification of Aβ-positive area in CA1 and DG regions of each group (n = 5). E Representative images of NeuN staining in the CA1 and DG regions of Control, E4-3Tg and E4-3Tg (A) mice (scale bar, 25 μm (upper), 50 μm (lower)). F Quantification of neuronal density in the CA1 (left) and DG regions (right) in each group (n = 5–6). Data are presented as mean ± SEM and were analyzed by Student’s t-test (B and D) or one-way ANOVA with Tukey’s multiple comparisons test (F). ns, not significant, *p < 0.05 We also observed p-Tau expression in newborn neurons of E4-3Tg mice, which was reduced in E4-3Tg (A) mice (Fig. [159]6A, B; cell number = 21–24 per group). In parallel, we found an increased number of DCX-positive cells as well as BrdU^+/NeuN^+ cells in E4-3Tg (A) mice (Fig. [160]6C–F; n = 3–6 per group). Notably, we also observed a strong negative correlation between the number of GFAP-positive astrocytes and the number of DCX-positive neurons in E4-3Tg mice (Fig. [161]6G), suggesting that improved AHN levels was likely a consequence of inhibited A1 astrocyte activation and C3 production. Fig. 6. [162]Fig. 6 [163]Open in a new tab Inhibition of astrocyte activation alleviated AHN dysfunction in E4-3Tg mice. A Representative images of DCX and AT8 co-staining in the DG area of E4-3Tg mice and E4-3Tg (A) mice (scale bar, 10 μm). E4-3Tg (A) mice showed minimal AT8 signal in DCX-positive cells. White Triangle, AT8-positive newborn neurons; Transparent triangle, AT8-negative newborn neurons; white arrow, AT8-positive mature neurons. B Quantification of AT8-positive area (%) in DCX-positive cells in each group (n = 21–24). C Representative images of DCX staining in the DG region of Control, E4-3Tg and E4-3Tg (A) mice (scale bar, 100 μm). D Representative images of BrdU and NeuN co-staining in the DG region (scale bar, 50 μm). E Quantification of the number of DCX-positive cells in the DG region in each group (n = 4–6). F Quantification of the number of BrdU^+/NeuN^+ cells in the DG region of each group (n = 3–4). G Simple linear regression curves showing the correlation between the number of GFAP-positive cells in the DG region and the number of DCX-positive cells in E4-3Tg mice (n = 15). H Enlarged SEM images showing mitochondrial swelling in DCX-positive newborn neurons of E4-3Tg mice (scale bar, 1 μm). MT, normal mitochondria. sMT, swollen mitochondria (defined as > 1 μm in long diameter with increased volume; highlighted in orange); N, Nucleus; SEM, scanning electron microscopy. I and J Quantification of mitochondrial length (long diameter) (I) and mitochondrial area size (J) of newborn neurons in Control (8 neurons, 39 mitochondria) and E4-3Tg mice (10 neurons, 41 mitochondria). Data are presented as mean ± SEM (E and F) or presented as violin plots (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) (B, I and J) and were analyzed by Mann-Whitney test (B, I and J) or one-way ANOVA with Tukey’s multiple comparisons test (E and F). *p < 0.05, **p < 0.01, ***p < 0.001 Collectively, these results showed that inhibition of A1 astrocytes attenuated tau-associated neurodegeneration in both hippocampal mature neurons and newborn neurons in E4-3Tg mice. This neuroprotective effect was accompanied by reduced mortality and improved cognitive function observed in E4-3Tg (A) mice, suggesting a link between attenuated tau pathology and reduced mortality. Inhibition of astrocyte activation prevented mitochondrial dysfunction and senescence in newborn neurons of E4-3Tg model mice Severe AHN dysfunction was observed in E4-3Tg mice, a phenotype not evident in TE4 mice. We hypothesized that newborn neurons are particularly vulnerable to Aβ-enhanced astrocytic neurotoxicity, as they are metabolically immature and highly dependent on astrocytic support [[164]60]. Importantly, the overall loss of mature neurons was not sharp and the proportion of AT8-positive neurons remained relatively low at 6.5 months of age. Therefore, focusing on the changes in newborn neurons may provide a more sensitive measure of the neurotoxic effects of A1 astrocytes in E4-3Tg mice. Based on this hypothesis, we further investigated the transcriptomic and structural changes in newborn neurons of E4-3Tg mice. Consistent with our findings in MSP-1 neurons, we observed mitochondrial dysfunction in newborn neurons of E4-3Tg mice. We first observed the DCX-positive newborn neurons at the ultrastructural level using correlative light and electron microscopy (CLEM) (Fig. S16). Compared to control mice, newborn neurons of E4-3Tg mice showed an increased number of swollen mitochondria with increased area and length (Fig. [165]6H–J), indicating the potential changes in mitochondrial function. Then we isolated newborn neurons using MACS and performed RNA-Seq analysis to assess transcriptomic differences in newborn neurons (Fig. [166]7A; n = 3–4 per group for Sequencing). GSEA results suggested down-regulated expression of genes related to the electron transport chain and oxidative phosphorylation, together with an up-regulated trend in oxidative damage response–related genes in newborn neurons of E4-3Tg mice (Fig. [167]7B). By contrast, newborn neurons in E4-3Tg (A) mice exhibited a restored trend in these pathways (Fig. [168]7B; Fig. S17). Analysis of DEGs further demonstrated that anserine treatment significantly restored the expression of electron transport chain–related genes (Fig. [169]7C). Moreover, Tom20 and DCX co-staining confirmed that mitochondrial swelling in newborn neurons was alleviated by anserine treatment (Fig. [170]7D, E). Taken together, in line with the increased p-Tau (AT8) expression in newborn neurons of E4-3Tg mice, our results suggest that accelerated tau pathology contributes to mitochondrial dysfunction, which can be prevented by inhibition of A1 astrocytes. Fig. 7. [171]Fig. 7 [172]Open in a new tab Inhibition of A1 astrocyte activation prevented mitochondrial dysfunction and senescence in newborn neurons. A Schematic overview of the isolation of newborn neurons from the hippocampus using PSA-NCAM Microbeads for RNA-Seq analysis. B Representative GSEA results using the Wiki Pathways database. C Heatmap showing the expression of mitochondrial function–related genes across the three groups. All genes were down-regulated in newborn neurons from E4-3Tg mice and up-regulated in those from E4-3Tg (A) mice. D Representative images of DCX and Tom20 co-staining in the DG region of Control, E4-3Tg and E4-3Tg (A) mice (scale bar, 10 μm (upper), 2 μm (lower)). E Quantification of Tom20-positive area in DCX-positive cells of each group (n = 53–57). E4-3Tg mice showed larger Tom20-positive area in DCX-positive newborn neurons, which was prevented by anserine treatment. F Heatmap showing expression levels of senescence-associated secretory phenotype (SASP) genes across the three groups. G Quantification of gene set expression score of SASP in each group (n = 3–4). H RT-qPCR analysis of the expression of Cdkn2a (p16) and Cdkn1a (p21) in each group (n = 4). I Representative images of p21 and DCX co-staining in the DG region of Control, E4-3Tg mice and E4-3Tg (A) mice (scale bar, 10 μm). J Quantification of the proportion of DCX^+/p21^+ cells in each group (n = 3–5). Data are presented in violin plots (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) (E), as box plots (line at median) (G) or as mean ± SEM (H, J) and were analyzed by one-way ANOVA with Tukey’s multiple comparisons test (H, J) or Kruskal-Wallis test with Dunn’s multiple comparisons test (E and G). *p < 0.05, **p < 0.01, ***p < 0.001. Some elements in panel A were created with BioRender.com Additionally, alleviated mitochondrial dysfunction in newborn neurons coincided with reduced cellular senescence. Studies have reported the well-established link between mitochondrial dysfunction and neuronal senescence, which is known to occur in neurons with tauopathy [[173]43–[174]45]. In newborn neurons of E4-3Tg mice, we observed increased expression of Senescence-Associated Secretory Phenotype (SASP) genes, with mean expression scores shifting from negative to positive (Fig. [175]7F, G), although this difference did not reach statistical significance (p = 0.11). Compared with E4-3Tg mice, newborn neurons in E4-3Tg (A) mice showed reduced SASP expression (Fig. [176]7F, G). RT-qPCR analysis further revealed a significant reduction in Cdkn1a (encoding p21) expression and a decreasing trend of Cdkn2a (encoding p16) in newborn neurons of E4-3Tg (A) mice (Fig. [177]7H; n = 4 per group). Immunostaining confirmed a reduced proportion of p21-positive newborn neurons in E4-3Tg (A) mice compared with E4-3Tg mice (Fig. [178]7I, J; n = 3–5 per group). In line with these changes, expression of genes related to synaptic plasticity and neuronal maturation—including Shank2, Syn, Nrn1, Olfm1 and Map1a were significantly decreased in newborn neurons of E4-3Tg mice, which was restored in E4-3Tg (A) mice (Fig. S18). By contrast, we did not observe a difference in the proportion of p21^+/GFAP^+ astrocytes among three groups (Fig. S19), suggesting that the change in astrocytic senescence was not prominent in E4-3Tg mice. Together, these findings suggested a cascade in which mitochondrial dysfunction led to neuronal senescence, leading to impaired maturation and loss of newborn neurons in E4-3Tg mice, a process that was prevented by inhibition of A1 astrocyte activation. Inhibition of A1 astrocyte activation reduced mortality and improved both cognitive and motor function in TE4 tauopathy mice In E4-3Tg mice, our findings showed that inhibition of A1 astrocyte activation with anserine reduced mortality, attenuated neurodegeneration, and improved cognition—effects that were likely mediated through the attenuation of tau pathology. To further investigate whether anserine exerts similar neuroprotective effects in other models with tauopathy background, we examined its impact on mortality, A1 astrocyte activation, and tau-associated neurodegeneration in TE4 tauopathy mice. Since this model was reported to exhibit prominent neurodegenerative phenotypes [[179]13, [180]28], we additionally performed transcriptomic analysis to deepen our understanding of the dual inhibitory effects of anserine on A1 astrocyte activation and used SEM to further validate the impact of p-Tau on neurodegeneration. Moreover, we knocked down anserine transporter PEPT2 to confirm that the neuroprotective effects of anserine were mediated by A1 astrocyte inhibition. We administered anserine to TE4 mice starting at 7 months of age for 8 weeks (Fig. [181]8A). In line with our findings in E4-3Tg mice, compared with untreated TE4 mice, which typically exhibit tau pathology-related paralysis (Fig. [182]1A), anserine-treated TE4 mice, TE4 (A) mice rarely developed paralysis and could easily live over 40 weeks (1/21 in TE4 (A) mice vs. 11/23 in TE4, p < 0.01; Fig. [183]8B). Although male TE4 mice exhibited a relatively higher mortality compared to female TE4 mice (Fig. S1C), consistent with a previous study [[184]61], anserine treatment exerted comparable protective effects in both male and female TE4 mice (Fig. S1 3 C, D). Importantly, similar effects were observed using an alternative commercially available anserine preparation, the same one used in our in vitro study, further supporting the reproducibility of our findings (Fig. S13E, F). Notably, TE4 (A) also exhibited restored body weight (Fig. [185]8C; n = 9–11 per group) and improved motor function (Fig. [186]8D; n = 6–7 per group). Regarding cognitive function, TE4 mice exhibited reduced alternation behavior in the Y-maze test (Fig. [187]8E). Since TE4 mice did not show significant impairment in the CFC test, we performed another behavior test, radial arm maze (RAM) test, in which TE4 mice took more time to find all the feed rewards (Fig. [188]8F). In contrast, these memory deficits were significantly alleviated in TE4 (A) mice (Fig. [189]8E, F; n = 6–8 per group). Fig. 8. [190]Fig. 8 [191]Open in a new tab Inhibition of A1 astrocyte activation reduced mortality and improved both cognitive and motor function in TE4 mice. A Schematic overview of the in vivo study using TE4 tau model mice. An 8-week oral anserine treatment (2 g/L) was administered starting at 7 months of age. B Kaplan-Meier survival curves of Control, TE4 and TE4 (A) mice (Control, n = 16; TE4, n = 23; TE4 (A), n = 21). Kaplan-Meier survival analysis was performed by log-rank (Mantel-Cox) test. Among the three groups (p < 0.0001). Pairwise comparisons: Control and TE4, p = 0.0017; TE4 and TE4 (A), p = 0.0017. In TE4 (A) group, seven mice were treated with an alternative commercially available anserine preparation. C Quantification of body weight of three groups of mice at the end of behavior tests (n = 9–11). D Quantification of grip strength of three groups of mice at the end of behavior tests (n = 6–7). E Results of the Y-maze. The results were presented as alternation rates (%) in the Y-maze of three groups of mice during 10 min (n = 8). F Results of the RAM. The time taken to find all the food rewards was recorded in a 5-days RAM test (n = 6–7). G Representative images of GFAP, p-TAK1 and C3 co-staining in the hippocampus region of Control, TE4 and TE4 (A) mice (Scale bar, 10 μm). White Triangle, p-TAK1^+/GFAP^+ cells. H Quantification of the proportion of p-TAK1^+/GFAP^+ cells in each group (n = 4–5). I Quantification of C3-positive area in GFAP-positive cells in each group (n = 4–6). J Quantification of C3-positive area in p-TAK1^−/GFAP^+ cells and p-TAK1^+/GFAP^+ cells (n = 21). Data are presented as violin plots (median: bold dashed line; quartiles (1/4 and 3/4: dashed lines)) (J) or presented as mean ± SEM (C, D, E, F, H and I) and were analyzed by Student’s t-test (J), one way ANOVA with Tukey’s multiple comparisons test (C, D, E, H and I) and two-way repeated ANOVA with Tukey’s multiple comparisons test (F). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Anserine treatment also inhibited A1 astrocyte activation in TE4 mice. Immunostaining revealed significantly inhibited astrocytic TAK1 activation in TE4 (A) mice (Fig. [192]8G, H; n = 4–5 per group), accompanied by reduced C3 immunoreactivity (Fig. [193]8I, J). Consistent with our findings in E4-3Tg mice, we did not observe a change of microglial TAK1 activation in TE4 (A) mice (Fig. S14C, D). RNA-Seq analysis of MACS-isolated astrocytes further revealed significantly down-regulated expression of A1-specific astrocyte markers, represented by C3, along with suppression of inflammatory responses, complement activation and NF-κB signaling in astrocytes of TE4 (A) mice (Fig. S20). We further found reduced production of C3, TNF-α and IL-1β in the hippocampus, as well as reduced expression of pro-inflammatory factors such as Tnf, Il1b and C1qa after anserine treatment (Fig. S20B, E; Fig. S21). Notably, these factors are known inducers of A1 astrocyte activation and genetic deletion of Tnf, Il1a and C1qa reduced mortality in ALS model mice [[194]7]. This suggested that reduced mortality was the consequence of inhibited astrocytic reactivity. Taken together, these results indicated that inhibition of A1 astrocyte activation reduced mortality and improved both cognitive and motor function in TE4 mice. Inhibition of astrocyte activation alleviated tau-associated neurodegeneration in TE4 mice We also confirmed that inhibition of A1 astrocyte activation alleviated tau-associated neurodegeneration in TE4 mice. We first confirmed the effect of anserine treatment on attenuating brain atrophy in TE4 mice (Fig. S15B), similar to our observations in E4-3Tg mice. Then we visualized ultrastructural changes in AT8-positive neurons by CLEM (Fig. [195]9A). AT8-positive neurons exhibited a significantly increased number of swollen mitochondria with enlarged size and elongated length, with some of the swollen mitochondria exhibiting reduced cristae density (Fig. S22). These results indicated the presence of tau-associated mitochondrial dysfunction in TE4 mice, consistent with our findings in MSP-1 neurons and E4-3Tg mice. In association with this, degenerating hippocampal neurons containing swollen mitochondria and aggregated tau were abundant in TE4 mice (Fig. [196]9B). Notably, we observed that some neurons displayed a “necroptosis-like” phenotype, with vacuolized cytoplasm and mitochondrial swelling, along with tau aggregation (Fig. [197]9C), indicating the occurrence of neuronal death driven by tau-associated mitochondrial dysfunction. Compared to TE4 mice, TE4 (A) mice exhibited significantly alleviated mitochondrial swelling and restored mitochondrial cristae density in hippocampal neurons (Fig. [198]9D and Fig. S23). Degenerating neurons were also rarely observed in TE4 (A) mice, consistent with increased neuronal density revealed by NeuN staining (Fig. [199]10A–C; n = 4–6 per group). Furthermore, in agreement with our observations in newborn neurons of E4-3Tg mice, alleviated neuronal mitochondrial dysfunction in TE4 (A) mice was associated with a reduced number of p21-positive neurons compared to TE4 mice (Fig. [200]10D, E; n = 3–4 per group). We also did not observe altered expression of senescence markers in astrocytes after anserine treatment (Fig. S24). In line with these changes, AT8 staining and pT181 staining (recognizing p-Tau at Thr181) demonstrated the attenuated tau pathology in TE4 (A) mice (Fig. S25A–D), resulting from reduced astrocytic C3 expression (Fig. S25E, F). Fig. 9. [201]Fig. 9 [202]Open in a new tab Inhibition of astrocyte activation alleviated tau-associated mitochondrial abnormalities in TE4 mice. A Representative CLEM image of phosphorylated tau-positive neurons (AT8-positive) in the hippocampus of TE4 mice (scale bar, 1 μm). Dapi staining (blue) shows nucleus (N), and AT8 antibody staining (magenta) shows the area of p-Tau expression. Lower panels show enlarged images of upper panel (depicted by white rectangle). The red arrows in the lower panel show aggregated-tau fibers in the neuron. Swollen mitochondria (sMT; defined as > 1 μm in long diameter, with increased volume; overlaid in yellow); FM, fluorescent microscopy; SEM scanning electron microscopy. B Representative SEM images of hippocampal neurons in TE4 mice (Scale bar, 1 μm). Nuclei of normal neurons appear bright, but nucleus of degenerated neuron becomes dark. N, nucleus; MT, normal size mitochondria (blue); sMT, swelling mitochondria (yellow). The red arrows indicate tau fibers. C Representative SEM image of hippocampal neurons in TE4 mice exhibiting necroptosis-like phenotypes, including dark nuclei and prominent cellular vacuolization, (Scale bar, 1 μm). The red arrows indicate aggregated-tau fibers, and blue arrow indicates the cell vacuolization. D Representative SEM images of hippocampal neurons in Control, TE4 and TE4 (A) mice (scale bar, 1 μm). N, nucleus; MT, normal mitochondria (blue); sMT, swollen mitochondria (yellow). The black arrow indicates mitochondria with reduced cristae density Fig. 10. [203]Fig. 10 [204]Open in a new tab Alleviated neuronal mitochondrial abnormalities reduced neuronal p21 expression and neuronal loss in TE4 mice. A Representative images of NeuN staining in the CA1 and DG regions of Control, TE4 and TE4 (A) mice (scale bar, 25 μm (upper), 50 μm (lower)). B and C Quantification of neuronal density in the CA1 (B) and DG regions (C) in each group (n = 4–6). D Representative images of NeuN and p21 co-staining of Control, TE4 and TE4 (A) mice (scale bar, 50 μm (upper), 10 μm (lower)). White arrows indicate the p21-positive neurons. E Quantification of the proportion of p21-positive neurons in the CA1 and DG regions in each group (n = 3–4). Data are presented as mean ± SEM and were analyzed by one-way ANOVA with Tukey’s multiple comparisons test. *p < 0.05, **p < 0.01 Collectively, these results indicated that the inhibition of A1 astrocyte activation and C3 over-production alleviated tau-associated neuronal mitochondrial abnormalities and senescence, thereby preventing neuronal loss in TE4 mice. This neuroprotective effect was associated with the reduced mortality observed in TE4 (A) mice. Blocking astrocytic uptake of anserine reversed its neuroprotective effects in TE4 mice Finally, we confirmed that the neuroprotective effects of anserine observed in TE4 mice were a consequence of the inhibition of TAK1-mediated A1 astrocyte activation. To block the effects of anserine on astrocytes, we used AAV-shPEPT2-PHP.eB (expressing shRNA targeting Slc15a2) to knock down (KD) PEPT2, an anserine transporter expressed predominantly in astrocytes [[205]38], delivered by intraventricular injection (Fig. [206]11A–C). Compared to TE4 (A) mice injected with a control AAV vector (TE4 (A)-shCon), PEPT2 KD (TE4 (A)-shPEPT2) led to increased TAK1 activation and C3 immunoreactivity in GFAP-positive astrocytes (Fig. [207]11D–F; n = 5–6 per group). Consequently, TE4 (A)-shPEPT2 mice exhibited elevated p-Tau expression and a higher number of p21-positive neurons (Fig. [208]11G–J; n = 3–4 per group), along with reduced neuronal density in the hippocampal CA1 and DG regions (Fig. [209]11K; n = 4–5 per group), indicating exacerbated tau-associated neurodegeneration following PEPT2 KD. In addition, PEPT2 KD reversed the anserine-mediated improvements in body weight and motor performance in TE4 mice (Fig. [210]11L, M; n = 6–7 per group). Furthermore, TE4 (A)-shPEPT2 mice exhibited impaired alternation behavior in the Y-maze test, comparable to TE4 mice, whereas TE4 (A)-shCon mice maintained significantly improved alternation performance (Fig. [211]11N; n = 7 per group). Together, these results demonstrated that inhibited A1 astrocyte activation mediated by anserine was essential for its neuroprotective effects, further highlighting the pathological role of A1 astrocytes in driving tau pathology and neurodegeneration. Fig. 11. [212]Fig. 11 [213]Open in a new tab Loss of the neuroprotective effects of anserine by knockdown of its astrocytic transporter in TE4 mice. A and B Schematic of the in vivo AAV experiment designed to block PEPT2-mediated intracellular uptake of anserine in astrocytes. An AAV vector expressing shRNA targeting Slc15a2 (AAV-shPEPT2) was injected into both lateral ventricles of TE4 mice at 7 months of age. C Representative images showing astrocytic expression of AAV-shPEPT2. Vector-driven GFP was observed in GFAP-positive astrocytes in the hippocampus of AAV-injected TE4 mice (scale bar, 50 μm). White arrows indicate GFP signals, suggesting expression of the AAV vector in GFAP-positive cells. D Representative images of GFAP and p-TAK1 (upper, scale bar, 25 μm) or GFAP and C3 (lower, scale bar, 50 μm) co-staining in hippocampus region of TE4, TE4 (A)-shCon and TE4 (A)-shPEPT2 mice. E Quantification of the proportion of p-TAK1^+/GFAP^+ cells in each group (n = 4). F Quantification of C3-positive area in GFAP-positive cells in each group (n = 5–6). G Representative images of pT181 staining in the hippocampal CA1 and DG regions in TE4, TE4 (A)-shCon and TE4 (A)-shPEPT2 mice (scale bar, 25 μm). H Quantification of pT181-positive staining area in each group (n = 3–4). I Representative images of p21 and NeuN co-staining in hippocampus area in TE4, TE4 (A)-shCon and TE4 (A)-shPEPT2 mice (scale bar, 20 μm). J Quantification of the proportion of p21^+/NeuN^+ cells in each group (n = 3–4). K Quantification of neuronal density in the CA1 (left) and DG regions (right) in each group (n = 4–5). L Quantification of body weight of TE4 (A)-shCon and TE4 (A)-shPEPT2 mice (n = 7). M Quantification of grip strength of TE4 (A)-shCon and TE4 (A)-shPEPT2 mice (n = 6). N Result of the Y-maze in TE4, TE4 (A)-shCon and TE4 (A)-shPEPT2 mice (n = 7). Data are presented as mean ± SEM and were analyzed by Student’s t-test (L and M) or one-way ANOVA with Tukey’s multiple comparisons test (E, F, H, J, K and N). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Some elements in panel B were created with BioRender.com Discussion In the present study, by utilizing a novel E4-3Tg mouse model, we revealed the contribution of Aβ pathology to accelerated mortality in the tauopathy mouse model and demonstrated the neuroprotective role of anserine in preventing this pathology by inhibiting IRAK1/TAK1 signaling in astrocytes (Fig. [214]12). Compared to TE4 mice, E4-3Tg mice exhibited accelerated mortality, with a median survival of approximately 6.5 months. At this age, E4-3Tg mice displayed enhanced A1 astrocyte activation, accelerated tau-associated neurodegeneration, and cognitive impairment. These pathological features only became apparent in TE4 mice at 9 months of age. Importantly, this pathology was prevented by a dual TAK1/IRAK1 inhibitor, anserine, for its role in suppressing A1 astrocyte activation. Therefore, our findings revealed the importance of targeting astrocyte activation to prevent accelerated mortality in Alzheimer’s disease (AD). Fig. 12. [215]Fig. 12 [216]Open in a new tab Graphical summary of this study. A Schematic illustration of the impact of Aβ pathology on the P301S tauopathy model. Aβ enhances the activation of A1 astrocytes, which in turn accelerates tau pathology and exacerbates tau-associated neurodegeneration in both immature and mature hippocampal neurons. Together, these pathological processes contribute to the accelerated mortality observed around 6.5 months of age. B Schematic illustration of the proposed mechanism by which anserine inhibits A1 astrocyte activation to prevent accelerated mortality in E4-3Tg and TE4 mice. In the AD environment, astrocytes are activated by Aβ or pro-inflammatory cytokines via the corresponding receptors. Anserine is actively taken up by astrocytes via PEPT2 and inhibits NF-κB activation mediated by IRAK1 and TAK1 signaling. This inhibition reduces the production of pro-inflammatory cytokines such as C3, TNF-α, IL-1β, and C1q. Inhibited astrocytic C3 production alleviates tau-associated neurodegeneration. Moreover, decreased secretion of pro-inflammatory cytokines limits further A1 astrocyte activation. Together, these effects mitigate neurodegeneration, protect against cognitive decline, and reduce mortality. Elements in panel A and B were created with BioRender.com Here, our results revealed the contribution of Aβ to accelerated mortality in AD. Consistent with other neurodegenerative diseases, AD is characterized not only by cognitive decline but also by reduced lifespan compared with healthy individuals [[217]9]. Aggravated tau pathology is associated with high mortality risk in AD patients [[218]11, [219]12], and this link was confirmed in P301S tauopathy model [[220]13]. As the earliest pathological feature of AD, Aβ is able to promote AD-related pathology through different mechanisms including tau pathology [[221]62], suggesting its potential contribution to mortality in AD. However, due to the limitation of AD models [[222]33], the impact of Aβ on mortality is still unclear. In contrast, E4-3Tg mice utilized in this study not only exhibit both Aβ and tau pathology, but also a clear mortality endpoint, earlier than that observed in TE4 mice. Importantly, the time point of the median survival was accompanied by aggravated tau pathology in E4-3Tg mice. Moreover, attenuating tau pathology with anserine reduced mortality without altering Aβ pathology. Therefore, we excluded the possibility that reduced Aβ pathology contributed to the observed survival benefit. Collectively, our findings provide evidence that Aβ contributes to accelerated mortality through the aggravation of tau pathology. Furthermore, compared with other commonly used AD mouse models [[223]33], our findings demonstrated that E4-3Tg mice recapitulate both Aβ and tau pathologies along with accelerated mortality. These results highlight the promise of E4-3Tg mice as a novel and valuable AD model for evaluating not only the effects of therapeutic strategies on AD-related pathology, but also on AD-associated mortality. We also showed the important role of A1 astrocyte activation in accelerating mortality in AD. Increased neuroinflammatory response in the central nervous system is closely associated with tau pathology [[224]63]. Although microglia are usually more recognized for their role in neuroinflammation, the important role of astrocytes cannot be ignored [[225]64]. Recent studies have revealed the crucial neurotoxic effects of activated A1 astrocytes in AD [[226]2, [227]65], and inhibition of astrocytic reactivity showed positive effects in protecting against cognitive decline in our previous studies [[228]66, [229]67]. Reactive A1 astrocytes contribute to tau-associated neurodegeneration through C3 over-production [[230]14], which has also been recently reported to be associated with accelerated mortality in neurodegeneration mouse models [[231]6, [232]13]. In this study, we found that accelerated mortality and aggravated tau pathology were accompanied by increased A1 astrocyte activation in both E4-3Tg and TE4 mice. Notably, inhibition of A1 astrocyte activation and C3 over-production by TAK1/IRAK1 inhibitor anserine significantly prevented tau-associated neurodegeneration and reduced mortality in both models. Importantly, anserine treatment did not alter p-TAK1 expression in microglia in either E4-3Tg or TE4 mice. By contrast, PEPT2 knockdown abolished its inhibitory effect on astrocytic p-TAK1 in TE4 mice (Fig. [233]11). These results further suggested that the effect of anserine was mainly on astrocytes. Moreover, anserine treatment also resulted in reduced production of TNF-α and IL-1β, which are associated with accelerated mortality in ALS model mice [[234]7], further verifying the link between inhibited astrocyte activation and reduced mortality observed in E4-3Tg mice and TE4 mice. As described above, the P301S mutation in TE4 mice was identified in individuals with the FTDP-17. Therefore, our findings suggest that targeting A1 astrocyte activation is a promising strategy not only to prevent accelerated mortality in AD, but also in other tauopathies. We suggested that exacerbated tau-associated neurodegeneration mediated by activated A1 astrocytes was closely associated with mortality observed in E4-3Tg mice and TE4 mice. The progression of tau pathology caused severe neuronal loss and brain atrophy in TE4 mice [[235]22], leading to neuronal dysfunction. In this study, we found attenuated neuronal loss and brain atrophy in both E4-3Tg (A) and TE4 (A) mice. We also observed tau-associated mitochondrial dysfunction in both hippocampal mature and immature neurons, linked to neuronal senescence and loss. Notably, cellular senescence has also been reported in glial cells in P301S tauopathy mice [[236]68, [237]69], although we did not observe these changes in astrocytes in this study. Previous studies have shown that APOE4 exacerbates neuronal senescence [[238]70] and enhances astrocytic activation along with pro-inflammatory responses [[239]22, [240]24, [241]71]. Thus, the presence of APOE4 may accelerate A1 astrocyte activation and promote neuronal senescence in E4-3Tg and TE4 mice. Importantly, inhibition of astrocytic IRAK1/TAK1 signaling by anserine did not alter the expression of senescence markers in astrocytes. Therefore, we suggest that suppression of A1 astrocyte activation is the main contributor to the attenuation of tau-associated neurodegeneration. The attenuated tau-associated neurodegeneration may reduce mortality for several possible reasons. Firstly, tauopathy-related paralysis is considered as one of the main reasons of mortality in P301S tauopathy mice [[242]13], and this phenotype is reported to be associated with tau-associated degeneration of motor neurons in spinal cord [[243]72]. In our study, both TE4 (A) and E4-3Tg (A) mice rarely exhibited paralysis when A1 astrocyte activation was inhibited, suggesting the possible changes in tau-associated neurodegeneration in motor neurons, thereby reducing paralysis-associated mortality. Regarding this possibility, we plan to further investigate the relationship between spinal motor neuron degeneration, A1 astrocyte activation, and mortality in future studies. However, not all mice, especially E4-3Tg mice, died with a typical paralysis phenotype. Some appeared to die suddenly, which may explain the absence of significant weight loss and motor decline at 6.5 months of age. E4-3Tg mice showed severe AHN dysfunction because of Aβ-enhanced tau pathology in newborn neurons. Newborn neurons play an important role in maintaining neuronal networks and preventing neuronal hyperexcitability [[244]73]. Therefore, it is possible that AHN dysfunction in E4-3Tg mice results in neuronal hyperexcitability, which induces epilepsy as previously report in epilepsy patients [[245]74, [246]75]. In addition, tau accumulation in the locus coeruleus (LC) can cause the degeneration of LC neurons [[247]76], providing another possible reason, such as sudden unexpected death in epilepsy (SUDEP) because of the impaired LC-heart noradrenergic pathway [[248]77, [249]78]. Notably, according to a recent study, inhibition of neuronal death is not enough to reduce mortality in tauopathy model mice [[250]79], while immunosuppressive agent tacrolimus (FK506), which can suppress astrocyte activation via NF-κB pathway, significantly improved survival [[251]13, [252]80, [253]81]. These findings further support our findings in this study. In this study, we established E4-3Tg and TE4 mouse models, both carrying human APOE4 background, to accelerate A1 astrocyte activation and tau-mediated neurodegeneration. Similar pathological processes of astrocyte activation may also occur with other APOE isoforms (APOE2 and APOE3). Previous studies have reported that APOE4 is associated with aggravated tau pathology, enhanced astrocyte-mediated neurotoxicity [[254]22], and increased mortality risk compared with APOE3 and APOE2 [[255]27], whereas APOE2 has been suggested to exert protective effects against tau pathology [[256]22]. These findings indicate that A1 astrocyte activation may contribute to mortality in a genotype-dependent manner. Therefore, future studies comparing the potential relationship between A1 astrocyte activation, tau-associated neurodegeneration, and mortality across different APOE isoforms will be important for providing broader mechanistic insights into accelerated mortality in AD. Furthermore, we identified anserine as a viable astrocyte-selective inhibitor that can suppress NF-κB-mediated A1 astrocyte activation, acting as a dual inhibitor of IRAK1 and TAK1. Anserine, as one of the imidazole dipeptides with anti-inflammatory properties [[257]35], has been found to suppress astrocyte activation and protect against cognitive decline in both aged APPswe/PSEN1dE9 model mice and human individuals with mild cognitive impairment (MCI) [[258]34, [259]43, [260]82]. Recently, our study in frail older adults provided additional evidence supporting the cognitive benefits of anserine, even at lower intake levels [[261]83]. In this study, we were the first to propose a novel mechanism by which anserine suppresses A1 astrocyte activation and subsequent neurodegeneration and mortality through the inhibition of IRAK1- and TAK1-mediated NF-κB activation. Since A1 subtype astrocytes are also the key contributors to other neurodegenerative diseases such as Parkinson’s disease (PD), Huntington’s disease (HD), amyotrophic lateral sclerosis (ALS), and multiple sclerosis (MS) [[262]2, [263]3], our findings suggest that anserine may have broader therapeutic potential for various neuroinflammation-associated neurodegenerative conditions. Considering clinical transformation, the dual inhibitory effects of anserine on TAK1 and IRAK1 also offer several advantages compared with other known compounds that can suppress NF-κB activation, such as metformin or (5Z)−7-Oxozeaenol [[264]84–[265]86]. Metformin has not shown consistent efficacy in clinical trials, and long-term use may even exacerbate cognitive impairment [[266]87, [267]88]. (5Z)−7-Oxozeaenol is a strong TAK1 inhibitor. However, it has primarily been investigated for cancer treatment, with studies reporting cytotoxic effects on cancer cells [[268]86, [269]89]. TAK1 inhibition by (5Z)−7-Oxozeaenol even promoted neuronal necroptosis in vitro [[270]90], raising concerns about its potential side effects in clinical applications. In contrast, we observed no detectable cytotoxicity of anserine in MSP-1 astrocytes, and its safety profile has also been confirmed in elderly populations in our previous clinical studies [[271]82, [272]91]. Although the IC₅₀ values are in the low-mM range, the presence of PEPT2-mediated active transport may allow the intracellular accumulation of anserine at higher concentrations in astrocytes, similar to the transport dynamics observed with metformin [[273]92]. Our in vitro results also showed that anserine inhibited C3 secretion in TNF-α–stimulated MSP-1 astrocytes at concentrations below the IC[50] values for TAK1 (Fig. [274]2B), consistent with possible intracellular accumulation. Importantly, one study demonstrated a stronger effect of pre-treatment with anserine and carnosine in suppressing inflammatory responses [[275]93]. Thus, the inhibitory effects of anserine on IRAK1/TAK1 may be stronger under pre-treatment conditions. To further elucidate the mechanism by which anserine inhibits IRAK1 and TAK1, molecular modeling (e.g., docking and dynamics simulations) and cryo-electron microscopy may need to be performed in future studies. Taken together, our findings highlight anserine as a viable and promising therapeutic strategy for clinical translation. Nevertheless, transcriptomic analysis of the human brain shows the expression of SLC15A2 both in astrocytes and microglia [[276]41, [277]94], although its expression in microglia is relatively low. This raises the possibility that the effects of anserine in humans may not be confined to astrocytes, which should be addressed in future studies. This study has several limitations. First, the E4-3Tg mouse model was generated by crossing separate APPswe/PSEN1dE9 mice and P301S-Tau transgenic mice, which resulted in relatively low offspring yields as described in the Methods section. This limited our ability to perform several experiments requiring large cohorts of mice, such as RNA-sequencing and AAV-mediated PEPT2 knockdown, in the E4-3Tg model. For the same reason, some experiments were conducted with relatively small N sample size. Second, we did not perform age-matched treatment because E4-3Tg and TE4 mice develop accelerated mortality and A1 astrocyte activation at different ages. Although this design ensured that both models were examined at comparable disease stages, further studies in which anserine is administered to TE4 mice from 4.5 to 6.5 months of age would be helpful to exclude potential confounding effects of age and treatment. Third, we did not perform dose–response analyses of the inhibitory effects of anserine on IRAK1 and TAK1 under pre-treatment conditions, nor did we determine the extent of intracellular accumulation of anserine. These need to be addressed in future studies. Lastly, the dual inhibitory effects of anserine on TAK1 and IRAK1 were demonstrated only in cultured mouse astrocytes. Further studies need to establish reliable methods for confirming IRAK1 inhibition in vivo, such as optimized antibodies and protein analysis using isolated astrocytes. Moreover, TAK1 has been reported to be associated with inflammatory reactive astrocyte signature in human iPSC-derived astrocytes [[278]95]. As increased astrocytic C3 expression [[279]2] and NF-κB activation [[280]4] were observed in AD patients, it would be important to further evaluate the effects of anserine on human astrocytes, for example by using iPSC-derived astrocyte cultures, to better evaluate the translational potential of our findings. Conclusion In summary, our findings demonstrated that Aβ pathology contributes to accelerated mortality in tauopathy mice, in association with enhanced A1 astrocyte activation and tau-associated neurodegeneration. We also demonstrated that inhibition of IRAK1/TAK1 signaling in astrocyte by anserine significantly attenuated tau-associated neurodegeneration and reduced mortality. Based on these results, we propose that targeting A1 astrocytes is promising for reducing mortality in AD. Supplementary Information [281]Supplementary Material 1^ (63.4MB, zip) Acknowledgements