Microglia TFEB activation attenuates Alzheimer’s disease pathology by enhancing autophagy-lysosomal function
Yeji Kim, Tae-Young Ha, Oksana Kondaurova, Myung-Shik Lee, Keun-A Chang
TL;DR
Activating TFEB in microglia improves Alzheimer’s disease by boosting waste clearance and reducing brain inflammation.
Contribution
This study shows that microglial TFEB activation specifically enhances Aβ clearance and reduces neuroinflammation in Alzheimer’s.
Findings
Microglial TFEB overexpression reduces amyloid burden in key brain regions of 5xFAD mice.
TFEB activation rescues memory deficits in both male and female 5xTFEB mice.
Transcriptomic analysis reveals upregulated autophagy and reduced inflammatory signaling.
Abstract
Alzheimer’s disease (AD) is characterized by amyloid-β (Aβ) accumulation, neuroinflammation, synaptic dysfunction, and cognitive decline. Impairment of microglial autophagy-lysosomal pathway (ALP) is increasingly recognized as a key driver of the disease progression. Transcription factor EB (TFEB), a master regulator of ALP, has emerged as a promising therapeutic target; however, its specific role in microglia remains unclear. Here, we aimed to determine the therapeutic effects of microglial TFEB expression in AD pathogenesis. We established a tamoxifen-inducible, microglia-specific TFEB-overexpressing 5xFAD mouse line (5xTFEB) and conducted behavioural testing, histopathology and biochemical analyses, live-cell imaging of Aβ phagocytosis, and bulk RNA sequencing. Differential gene expressions were analysed, and inflammasome activation was evaluated. Microglial TFEB overexpression…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10- —Gachon University Research Fund of 2024
- —https://doi.org/10.13039/501100003725National Research Foundation of Korea
- —Korea Dementia Research Project through the Korea Dementia Research Center
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAlzheimer's disease research and treatments · Neuroinflammation and Neurodegeneration Mechanisms · Autophagy in Disease and Therapy
Background
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by amyloid-β (Aβ) plaque deposition, neuroinflammation, synaptic dysfunction, and cognitive decline [1, 2]. Beyond neuronal pathology, accumulating evidence implicates microglia, the resident immune cells of the central nervous system, as critical contributors of AD progression. Under pathological conditions, microglia become activated and engage in both Aβ clearance and neuroinflammatory responses through cytokine release and inflammasome signalling [3]. However, during chronic exposure to pathological stimuli such as sustained Aβ accumulation, microglia often adopt a reactive phenotype characterized by impaired phagocytic capacity and excessive production of pro-inflammatory cytokines and inflammasome-related mediators [4].
The autophagy-lysosomal pathway (ALP) has emerged as a critical mechanism in AD pathogenesis [5]. ALP maintains cellular homeostasis by degrading misfolded proteins and damaged organelles through autophagosome formation and lysosomal proteolysis [6]. In AD models, genetic alterations such as presenilin-1 (PSEN1) mutations impair autolysosomal acidification and proteolytic activity, thereby disrupting Aβ clearance and exacerbating plaque accumulation [7, 8]. Consistent with this, several recent studies have demonstrated that pharmacological activation of autophagy enhances Aβ degradation and ameliorates AD-related pathology. For instance, crocetin promotes Aβ clearance via STK11/LKB1-mediated AMPK activation [9], while naringenin enhances AMPK-driven autophagy and protects neuronal cells from Aβ-induced toxicity [10]. These findings support ALP dysfunction as a key driver of AD progression.
Transcription factor EB (TFEB) is a master regulator of ALP that governs lysosomal biogenesis and autophagy-related gene expression [11]. In response to cellular stress, TFEB is dephosphorylated and translocated into the nucleus, where it binds to coordinated lysosomal expression and regulation (CLEAR) elements to initiate a transcriptional program promoting lysosome biogenesis, autophagosome formation, and vesicular trafficking [12]. Dysregulation of TFEB activity has been implicated in multiple neurodegenerative disorders, including AD [11], in which impaired TFEB function compromises autophagic clearance, accelerates the accumulation of pathological protein aggregates, and contributes to disease progression [13, 14].
Despite growing recognition of TFEB as a central regulator of cellular proteostasis, its cell type-specific role in microglia during AD progression remains incompletely understood. In this study, we investigated the functional significance of microglial TFEB in AD using the 5xFAD mouse model. We generated 5xFAD mice with microglia-specific TFEB overexpression (5xTFEB) and systematically evaluated cognitive performance, ALP-related molecular and cellular changes, and Aβ pathology. Our findings demonstrate that microglial TFEB activation is associated with restoration of autophagy-lysosomal function, enhanced Aβ clearance, and attenuation of neuroinflammatory responses, highlighting microglial TFEB as a potential therapeutic target for AD.
Results
ALP dysfunction and reduced microglial TFEB activity in 5xFAD mice
TFEB is known as a master regulator of the autophagy-lysosomal pathway (ALP), whose nuclear translocation is required for CLEAR-program activation under stress [12, 15]. To examine the status of TFEB in AD pathology, we performed subcellular fraction and immunoblotting of the cortex (CX) from 6-month-old WT and 5xFAD mice. Total TFEB abundance was comparable between genotypes (fold change, 1·08 ± 0·11; p = 0·55). In contrast, cytosolic TFEB (C-TFEB) was increased, whereas nuclear TFEB (N-TFEB) was reduced in the 5xFAD CX relative to the WT (C-TFEB: fold change, 3·26 ± 0·22, p < 0·0001; N-TFEB: fold change, 0·55 ± 0·07, p = 0·03) (Fig. 1a), indicating impaired nuclear import of TFEB. Fraction purity was confirmed using GAPDH (cytosol) and Lamin-B1 (nucleus) as loading controls.
Fig. 1ALP dysfunction and reduced microglial TFEB expression in 5xFAD mice. a Representative western blots of total, cytosolic (C-TFEB) and nuclear TFEB (N-TFEB) from cortical lysates of 6-month-old WT and 5xFAD mice, with quantification. (n = 7 per group). b Representative western blots of p62 and LC3-II in cortical lysates with quantification. (n = 7 per group). c Schematic illustration of microglial isolation by magnetic-activated cell sorting (MACS). RT-qPCR analysis of isolated microglia showing Tfeb mRNA levels in WT and 5xFAD mice. (WT: n = 5, 5xFAD: n = 6) (d) mRNA expression levels of TTFEB, SQSTM1, MAP1LC3B,* and LAMP1* in human microglia based on Single-cell RNA-seq data from Alzheimer’s disease patients and control subjects. Data are presented as mean ± SEM. Statistical significance was determined using Student’s t-test (a-d). * p < 0.05, *** p < 0.001, **** p < 0.0001
We next examined autophagy markers to evaluate ALP activity. Compared with WT, 5xFAD CX displayed elevated p62/SQSTM1 and LC3 II levels (p62: fold change, 2·42 ± 0·19, p < 0·0001; LC3II/LC3I ratio : fold change, 1·97 ± 0·12, p < 0·0001; LC3II/GAPDH fold change, 1·96 ± 0·67, p = 0·003) (Fig. 1b), consistent with autophagy insufficiency and accumulation of autophagosomes due to impaired clearance. LC3-II was normalized to LC3-I and p62 to GAPDH.
To determine whether this dysregulation extends to microglia, we isolated adult microglia using MACS and quantified Tfeb expression. RT-qPCR revealed significantly lower Tfeb mRNA levels in 5xFAD microglia compared with WT (ΔΔCt-derived fold change 0·13 ± 0·04, p < 0·001) (Fig. 1c). Together with reduced nuclear TFEB in whole CX, these findings suggest attenuated TFEB availability and transcriptional capacity in microglia under AD-like conditions.
We next analysed a publicly available SEA-AD single-nucleus transcriptomic dataset to assess whether similar alterations occur in human AD. Microglia from individuals with advanced AD pathology (Braak 5–6) exhibited significantly reduced TFEB expression relative to controls (Braak 0–2), whereas LAMP1, MAP1LC3B, and SQSTM1 were increased (Fig. 1d).
Together, these data indicate reduced microglial TFEB expression and nuclear localization in AD context, accompanied by increased expression of autophagy/lysosomal markers in 5xFAD mouse cortex and human AD microglia.
TFEB activation enhances autophagy–lysosomal function and Aβ clearance in adult microglia
To investigate the functional role of TFEB in microglia, we generated tamoxifen-inducible mice overexpressing human TFEB (hTFEB) specifically in microglia (TFEB-CA) (Supplementary Fig. 1a). PCR genotyping and immunofluorescence analysis confirmed microglia-specific hTFEB expression (Supplementary Fig. 1b, c). Primary microglia were isolated from 8-week-old WT and TFEB-CA mice using MACS and cultured in vitro for 7 days, during which cells acquired a ramified morphology characteristic of mature microglia (Supplementary Fig. 1d, e).
To assess TFEB-dependent lysosomal function, we employed pHrodo-labelled Aβ oligomers (pHrodo-oAβ), which fluoresce upon delivery into acidic lysosomal compartments, providing a dynamic readout of phagolysosomal trafficking. WT and TFEB-CA microglia were treated with 1 µM pHrodo-oAβ and monitored by live-cell imaging for 24 h (Fig. 2a). Red fluorescence progressively increased over time in both groups (Fig. 2b); however, TFEB-CA microglia exhibited significantly higher fluorescence intensity than WT cells from 2 h onward (p < 0·001; Fig. 2c), indicating enhanced trafficking of oAβ into acidic compartments.
Fig. 2TFEB overexpression enhances lysosomal function and Aβ clearance in adult microglia. a Experimental workflow for adult brain microglia culture from WT and TFEB-CA mice. b Scheme for assessing phagocytic activity using pHrodo-labelled oligomeric Aβ (oAβ-pHrodo+) and representative images of TFEB-CA microglia incubated with 1 μm oAβ-pHrodo at the indicated times (red, pHrodo; scale bar, 200 μm). c quantification of pHrodo intensity from live-cell imaging over 24 h (right). d Representative immunofluorescence images of LAMP1 (green) in WT and TFEB-CA microglia following 1 µM oAβ treatment for 2 h and 24 h (left; scale bar, 10 μm), with quantification of LAMP1 intensity (n = 4). e Representative immunostaining of hTFEB (green) and 6E10 (red) in TFEB-CA microglia following 24 h of 1 µM oAβ treatment (left; scale bar, 10 μm), with quantification of the 6E10-positive area using Image J. The region indicated by the red dotted square is shown at higher magnification (n = 5). f MTT assay showing cell viability of WT and TFEB-CA microglia following 24 h exposure to 0.5, 1 and 5 µM oAβ. Nuclei were counterstained with DAPI (blue). Data are presented as mean ± SEM. Statistical significance was assessed using Student’s t-test (c, e and f) and performed using one-way ANOVA followed by Tukey’s multiple comparisons test (d). * p < 0.05, ** p < 0.01, *** p < 0.001
Consistent with this finding, immunocytochemical analysis of LAMP1, a lysosomal membrane protein transcriptionally regulated by TFEB, revealed a significantly higher signal intensity in TFEB-CA microglia compared with WT controls at 2 h after oAβ exposure (p = 0·001; Fig. 2d). By 24 h, LAMP1 intensity in both groups had decreased to levels comparable to baseline (0 h), and no significant difference was observed between groups (Fig. 2d), suggesting a transient enhancement of lysosomal engagement during early oAβ processing.
To directly evaluate Aβ clearance, intracellular Aβ levels were quantified by 6E10 immunostaining 24 h after oAβ exposure. TFEB-CA microglia exhibited a significantly reduced intracellular Aβ burden compared with WT microglia (p < 0·001; Fig. 2e). In addition, MTT assays demonstrated that TFEB overexpression significantly preserved microglial viability following treatment with 0·5, 1, and 5 µM oAβ, relative to WT cultures (p < 0·0001; Fig. 2f).
Together, these results demonstrate that microglial TFEB activation enhances lysosomal engagement and phagolysosomal processing of oAβ, promotes more efficient Aβ clearance, and protects microglia from Aβ-induced cytotoxicity.
Microglia-specific overexpression of TFEB ameliorated cognitive deficits in 5xFAD mice
To determine the impact of TFEB activation in microglia on AD-related cognitive dysfunction, we generated a microglia-specific TFEB overexpression AD model (5xTFEB) by crossing 5xFAD mice with TFEB-CA mice (Fig. 3a). Successful genotyping of 4-week-old pups was confirmed by PCR (Supplementary Fig. 2). At 6 months of age, WT, 5xFAD, and 5xTFEB mice were subjected to microglial or non-microglia isolation using magnetic-activated cell sorting (MACS), followed by RT-qPCR and western blot analyses (Fig. 3b). RT-qPCR analysis of purified microglia revealed no significant difference in endogenous mouse Tfeb (mTfeb) expression between 5xFAD and 5xTFEB groups, whereas human TFEB (hTFEB) mRNA was robustly upregulated in 5xTFEB microglia (mTfeb: ΔΔCt-derived fold change, 0·54 ± 0·11, p = 0·01; vs. WT; hTFEB: ΔΔCt-derived fold change, 24·96 ± 4·61, p = 0·0004), confirming successful transgene expression without altering native TFEB levels (Fig. 3c). Western blot analysis of MACS-isolated cells demonstrated strong enrichment of the microglial marker Iba1, with minimal detection of the neuronal marker NeuN, indicating high microglial purity (Fig. 3d). Notably, hTFEB protein was detected exclusively in microglial isolated from 5xTFEB mice. Consistently, immunofluorescence analysis showed that hTFEB signals co-localized with Iba1-positive microglia, but not with GFAP-positive astrocytes (Fig. 3e) or NeuN-positive neurons (Fig. 3f), in both the cortex and hippocampus, confirming the microglia-specific expression of the transgene in vivo. Corresponding low-magnification images, together with higher-magnification views of the indicated regions, are provided in Supplementary Fig. 3, further illustrating the regional distribution and cellular specificity of hTFEB expression.
Fig. 3. Generation and validation of 5xTFEB mice with microglia-specific hTFEB overexpression. a Breeding strategy for generating 5xTFEB mice by crossing 5xFAD mice with TFEB-CA transgenic lines. b Experimental workflow for isolation of microglial and non-microglia cell population from 6-month-old WT, 5xFAD, and 5xTFEB mice using MACS, followed by RT-qPCR and western blot analyses. c RT-qPCR analysis showing that endogenous mouse Tfeb mRNA levels were not significantly altered, whereas human TFEB (hTFEB) transcripts were robustly induced specifically in microglia from 5xTFEB mice (WT: n = 3, 5xFAD: n = 4, 5xTFEB: n = 4). d Western blot analysis of MACS-isolated microglia and non-microglia from WT, 5xFAD, and 5xTFEB mice using the indicated antibodies. e Immunofluorescence analysis of cortical and hippocampal sections from 6-month-old 5xFAD and 5xTFEB mice stained for hTFEB (green), Iba1-positive microglia (red), and GFAP-positive astrocytes (magenta). f Immunofluorescence analysis of cortical and hippocampal sections from 6-month-old 5xFAD and 5xTFEB mice stained for hTFEB (green), Iba1-positive microglia (red), and NeuN-positive neurons (magenta). Nuclei were counterstained with DAPI (blue). Scale bar, 10 μm. Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparisons test (c). * p < 0.05, ** p < 0.01, *** p < 0.001
Given that amyloid plaque deposition in 5xFAD mice begins at approximately 2 months of age and becomes prominent in the hippocampus (HP) and CX by 6 months, coinciding with progressive cognitive decline [16, 17], behavioural experiments were conducted at 6.5 months of age (Fig. 4a). In the Y-maze test, which assesses spatial working memory and cognitive flexibility, 5xFAD mice exhibited significantly reduced spontaneous alternation compared with WT controls (male: WT, 65·27 ± 2·75%; 5xFAD, 45·01 ± 1·11%; p < 0·0001 vs. WT; female: WT, 60·15 ± 1·94%; 5xFAD, 42·81 ± 1·64%; p < 0·0001). In contrast, 5xTFEB mice showed significantly improved alternation performance relative to 5xFAD mice (male: 5xTFEB, 59·48 ± 1·42%, p < 0·0001 vs. 5xFAD; female: 5xTFEB, 57·92 ± 3·59%, p < 0·0001), with no differences in total arm entries among groups, indicating preserved locomotor activity (Fig. 4b). Comparable effects were observed in both sexes.
Fig. 4. Microglia-specific TFEB overexpression ameliorates cognitive deficits in 5xFAD mice. a Experimental timeline. Tamoxifen (80 mg/kg, i.p.) was administered once daily for five consecutive days at 1 month of age to induce microglial hTFEB expression. Behavioural assessments were conducted at 6.5–7 months, followed by brain tissue collection. b Y-maze test showing the spontaneous alternations (%) and the total arm entries in male (left) and female (right) mice. Both sexes of 5xTFEB mice exhibited significantly higher alternation rates compared with 5xFAD mice (male; n = 10 per group; female; WT: n = 11, 5xFAD: n = 10, 5xTFEB: n = 9). c Passive avoidance test (PAT) showing step-through latency measure 24 h after training in male (left) and female (right) mice. Latency was significantly increased in 5xTFEB mice relative to 5xFAD controls (male; n = 10 per group; female; WT: n = 11, 5xFAD: n = 10, 5xTFEB: n = 9). d Morris water maze (MWM) test showing escape latency during training (line graphs) and probe trial outcomes, including representative swim trajectories, time spent in the target quadrant (bar graphs), swim velocity, and distance. 5xTFEB mice demonstrated shorter escape latencies and increased target quadrant preference compared with 5xFAD mice, with no differences in swim speed or distance across groups (male; WT: n = 6, 5xFAD: n = 8, 5xTFEB: n = 8; female; WT: n = 10, 5xFAD: n = 12, 5xTFEB: n = 8). Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparisons test (b, d). ** p < 0.01, *** p < 0.001, **** p < 0.0001
In the passive avoidance test (PAT), which evaluates associative learning and memory retention, 5xFAD mice displayed markedly reduced step-through latencies compared with WT mice (male: WT, 248.7 ± 23.13; 5xFAD, 34·81 ± 6·89; p < 0·0001; female: WT, 237.6 ± 20.35; 5xFAD, 48·24 ± 6·65; p < 0·0001). In contrast, 5xTFEB mice exhibited significantly prolonged latencies relative to 5xFAD mice (male: 5xTFEB, 219·5 ± 25·04; p < 0·0001; female: 5xTFEB, 240·3 ± 33·19; p < 0·0001), indicating improved associative memory (Fig. 4c). This effect was consistent across sexes.
In the Morris water maze (MWM), which assesses spatial learning and long-term memory, 5xFAD mice showed impaired acquisition during training trials compared with WT controls. Both male and female 5xTFEB mice exhibited accelerated learning, as reflected by significantly shorter escape latencies relative to 5xFAD littermates. In the probe trial, 5xFAD mice spent significantly less time in the target quadrant compared with WT mice (male: WT, 25·2 ± 1·7; 5xFAD, 14·42 ± 1·33; p = 0·002; female: WT, 21·59 ± 1·21; 5xFAD, 13·20 ± 0·81; p = 0·003). In contrast, 5xTFEB mice spent significantly more time in the target quadrant than 5xFAD mice (male: 5xTFEB, 23·4 ± 2·26; p = 0·005; female: 5xTFEB, 21·5 ± 2·2; p = 0·006), indicating improved spatial memory retention (Fig. 4d).
Together, these results demonstrate that microglia-specific TFEB overexpression significantly ameliorates cognitive impairments in 5xFAD mice, improving working memory, associative learning, and spatial memory in both male and female mice.
Microglia-specific TFEB overexpression reduces the amyloid plaque burden in 5xFAD mice
To determine whether microglia hTFEB overexpression modulates Aβ pathology in vivo, we quantified amyloid plaque burden in 7-month-old WT, 5xFAD, and 5xTFEB mice. In male 5xFAD mice, extensive Aβ deposition was observed through the CX, HP, and entorhinal cortex (EC). In contrast, both 6E10 immunostaining for total Aβ and Thioflavin-S (Thio-S) staining for fibrillar amyloid plaques revealed significantly reduced diffuse and dense-core plaque deposition in 5xTFEB mice compared with 5xFAD littermates (Thio-S: CX fold change, 0·72 ± 0·08, p = 0·03; HP fold change, 0·57 ± 0·12, p = 0·02; EC fold change, 0·49 ± 0·08, p = 0·02; 6E10: CX fold change, 0·70 ± 0·11, p = 0·04; HP fold change, 0·56 ± 0·13, p = 0·03; EC fold change, 0·50 ± 0·09, p = 0·04) (Fig. 5a). A similar trend toward reduced amyloid burden were also observed in female 5xTFEB mice compared with age-matched 5xFAD females (Supplementary Fig. 4).
Fig. 5. Microglia-specific TFEB overexpression reduces amyloid plaque burden in 5xFAD mice. a Representative images of amyloid deposition in the cortex (CX), hippocampus (HP), and entorhinal cortex (EC) of WT, 5xFAD, and 5xTFEB male mice. Fibrillar amyloid plaques were visualized by Thioflavin-S (Thio-S, green) staining, and total Aβ plaques were detected by 6E10 immunostaining (red). No plaques were observed in WT brains. Quantification revealed a significant reduction in both fibrillar and total plaque burden in 5xTFEB compared with 5xFAD mice (n = 6 per group, scale bar, 25 μm). b Western blot analysis of cortical lysates probed with 6E10 antibody showing reduced Aβ monomer levels in 5xTFEB mice relative to 5xFAD mice (n = 4 per group). c ELISA quantification of cortical Aβ_1−42_ confirming significantly lower levels in 5xTFEB compared to 5xFAD mice (n = 6 per group). Data are presented as mean ± SEM. Statistical analyses were performed using Student’s t-test (a-b). * p < 0.05, ** p < 0.01
Biochemical analyses corroborated the histological findings. Western blotting of cortical lysates significantly reduced levels of monomeric Aβ in 5xTFEB mice relative to 5xFAD (fold change, 0·35 ± 0·02, p = 0·001). Normalization of Aβ monomer levels to APP demonstrated a marked decrease in the Aβ monomer/APP ratio in 5xTFEB mice relative to 5xFAD mice (fold change, 0·48 ± 0·13, p = 0·009) (Fig. 5b). Consistently, ELISA quantification demonstrated significantly lower cortical \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{A}{\upbeta\:}}_{1-42}$$\end{document} concentrations in 5xTFEB mice compared with 5xFAD littermates (5xFAD, 176·7 ± 27·26; 5xTFEB, 96·50 ± 19·76; p = 0·03) (Fig. 5c).
Together, these results indicate that microglia-specific hTFEB overexpression is associated with a substantial reduction in both diffuse and fibrillar amyloid deposition in 5xFAD brain. To assess whether amyloid burden was related to cognitive performance, we performed correlation analyses between cortical \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{A}{\upbeta\:}}_{1-42}$$\end{document} levels and individual behavioral outcomes, including the Y-maze, passive avoidance test (PAT), and Morris water maze (MWM) (Supplementary Fig. 5). Lower cortical \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{A}{\upbeta\:}}_{1-42}\:$$\end{document} concentrations were significantly associated with improved performance across these behavioural tasks, indicating a close relationship between reduced amyloid pathology and cognitive function.
Transcriptomic changes in 5xTFEB mice reveal enhanced ALP and reduced inflammatory signalling
To assess the transcriptomic impact of microglia hTFEB overexpression, we performed bulk RNA sequencing of HP and CX tissues from 7-month-old WT, 5xFAD, and 5xTFEB mice. Differential expression analysis revealed pronounced transcriptional differences between the 5xFAD and 5xTFEB groups. In the HP, 299 genes were significantly upregulated and 334 genes were downregulated, whereas in the CX, 144 genes were upregulated and 276 were downregulated in 5xTFEB mice relative to 5xFAD controls (Fig. 6; Supplementary Fig. 6). Volcano plots and hierarchical clustering of the top differentially expressed genes (DEGs) demonstrated clear segregation among WT, 5xFAD, and 5xTFEB samples in both regions (Fig. 6a), indicating genotype-dependent transcriptional signatures.
Fig. 6. Transcriptomics reveal ALP and inflammation gene changes in 5xTFEB mice. a Volcano plot and heatmap of differentially expressed genes (DEGs) in the hippocampus (HP) of 7-month-old WT, 5xFAD, and 5xTFEB mice obtained from bulk RNA-seq analysis. A total of 299 genes were upregulated and 334 downregulated in 5xTFEB compared to 5xFAD. b-d Gene Ontology (GO) enrichment analysis of DEGs categorized into (b) Biological Process (BP), (c) Cellular Component (CC), and (d) Molecular Function (MF). Upregulated terms included autophagy, lysosomal organization, endosome–lysosome transport, and ubiquitin-related functions, while downregulated terms included apoptotic process, inflammatory response, cytokine production, and immune receptor activity. e. Heatmap showing expression profiles of autophagy- and inflammation-related genes. f RT–qPCR analysis confirmed these transcriptomic findings, demonstrating increased expression of Gabarapl2 and Atg5 and decreased expression of Nlrp3 and Il1b in 5xTFEB mice. Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparisons test a-b. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. g STRING network analysis showing TFEB-associated interactions with autophagy- and lysosome-related proteins, including GABARAPL2, ATG5, Beclin-1
Gene Ontology (GO) enrichment analysis of DEGs revealed robust enrichment of autophagy- and lysosome-related biological process (BP) terms among genes upregulated in 5xTFEB mice. These included autophagy, endocytosis, autophagosome assembly, lysosomal organization, endosome to lysosome transport, and autophagosome maturation (Fig. 6b). In contrast, genes downregulated in 5xTFEB mice relative to 5xFAD were significantly enriched for BP terms associated with apoptosis and inflammatory signalling, including inflammatory response, positive regulation of inflammatory response, and regulation of cytokine production (Fig. 6b), indicating coordinated suppression of pro-inflammatory transcriptional programs.
At the cellular component (CC) level, upregulated DEGs were predominantly associated with lysosomes, autophagosomes, and late endosomes (Fig. 6c), consistent with enhanced intracellular degradative and trafficking capacity. Conversely, downregulated CC terms were enriched for the external plasma membrane and immunological synapse, suggesting reduced microglial immune activation. Molecular function (MF) analysis further showed enrichment of upregulated DEGs in categories such as ubiquitin protein ligase binding, ATP binding, and ubiquitin-protein transferase activity, whereas downregulated MF terms included receptor tyrosine kinase binding and cytokine receptor binding, again consistent with attenuation of inflammatory signalling (Fig. 6d).
Consistent with these pathway-level changes, key genes involved in autophagy and lysosomal function were dysregulated in 5xFAD mice and showed recovery in 5xTFEB mice. Notably, the autophagy-related Gabarapl2 and Atg5 were significantly upregulated in 5xTFEB mice compared with 5xFAD mice, whereas pro-inflammatory mediators, including Nlrp3 and, Il1b, was reduced (Fig. 6e).
These transcriptomic changes were validated by RT-qPCR. Gabarapl2 expression was significantly reduced in 5xFAD mice (fold change, 0.49 ± 0.12, p = 0.04 vs. WT) but was restored to near-WT levels in 5xTFEB mice (0·96 ± 0·25; p = 0.04 vs. 5xFAD). Atg5 showed a similar pattern, with reduced expression in 5xFAD mice (fold change, 0·51 ± 0·07, p = 0.005) and recovery toward WT levels in 5xTFEB mice (0·81 ± 0·10; p = 0.05). In contrast, Nlrp3 expression was elevated in 5xFAD mice relative to WT controls (fold change, 2·59 ± 0·41; p = 0.003) but was significantly reduced in 5xTFEB mice (1·46 ± 0·23; p = 0.03). A similar trend was observed for Il1b, which was increased in 5xFAD mice (2·83 ± 0·43; p = 0.001) and attenuated in 5xTFEB mice (1·78 ± 0·20; p = 0.05) (Fig. 6f).
Finally, STRING network analysis identified interactions between TFEB and multiple autophagy- and lysosome-related proteins, including GABARAPL2, ATG5, Beclin-1 (Fig. 6g), supporting a transcriptional network linking TFEB activation and enhanced ALP-related pathway.
Together, these transcriptomic data indicate that microglial hTFEB overexpression in 5xFAD mice is associated with a coordinated shift in gene expression toward lysosomal and autophagic programs, accompanied by suppression of inflammatory signalling, providing a molecular framework for the observed attenuation of AD-related pathology.
Microglia-specific TFEB overexpression restores ALP function in 5xFAD mice
To investigate whether TFEB overexpression in microglia modulates ALP markers in vivo, we assessed lysosomal and autophagy-related proteins in the HP and CX of WT, 5xFAD, and 5xTFEB mice. Confocal immunofluorescence revealed marked increases in LAMP1 and Iba1 immunoreactivity in 5xFAD mice compared with WT controls (Fig. 7a; fold change, HP: LAMP1, 46·51 ± 4·97, p < 0·0001 vs. WT; Iba1, 4·91 ± 0·83, p = 0·0005; LAMP1^+^Iba1^+^, 15·15 ± 0·99, p < 0·0001; CX: LAMP1, 22·95 ± 2·75, p < 0·0001; Iba1, 4·44 ± 0·22, p < 0·0001; LAMP1^+^Iba1^+^, 25·48 ± 3·10, p < 0·0001), consistent with lysosomal accumulation and microglial activation. In contrast, 5xTFEB mice exhibited significantly reduced LAMP1 intensity and attenuated Iba1 signal in both regions relative to age-matched 5xFAD mice (Fig. 7a; fold change, HP: LAMP1, 25·57 ± 7·79, p = 0·02 vs. 5xFAD; Iba1, 2·79 ± 0·41, p = 0·04; LAMP1^+^Iba1^+^, 11·91 ± 0·93, p = 0·03; CX: LAMP1, 9·76 ± 1·95, p = 0·001; Iba1, 2·32 ± 0·31, p < 0·0001; LAMP1^+^Iba1^+^, 15·02 ± 1·86, p < 0·0001).
Fig. 7. Microglia-specific TFEB overexpression restores ALP in 5xFAD mice. a Representative confocal images of hippocampus (HP) and cortex (CX) from 7-month-old WT, 5xFAD, and 5xTFEB mice stained for Iba1 and LAMP1 immunoreactivity in 5xFAD mice, which was reduced in 5xTFEB mice (WT: n = 6, 5xFAD: n = 6, 5xTFEB: n = 5, scale bar, 20 μm). b Western blot analysis of HP and CX lysates for autophagy-lysosomal markers. 5xFAD mice showed elevated p62, LC3-II, Beclin-1, and LAMP1, consistent with impaired ALP activity. These alterations were reversed in 5xTFEB mice (n = 7 per group). Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparisons test a-b. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001
To confirm the cell-type specificity, we examined LAMP1 colocalization with neuronal and astrocytic markers. Elevated LAMP1 signals were detected exclusively in Iba1⁺ microglia, with no appreciable colocalization with NeuN⁺ neurons or GFAP⁺ astrocytes, indicating that TFEB-driven lysosomal changes occurred selectively in microglia (Supplementary Fig. 7).
We next corroborated these findings by immunoblotting of autophagy-related proteins in HP and CX lysates. Compared with WT mice, 5xFAD mice exhibited significantly increased levels of p62 and LC3-II, accompanied by modest elevations in Beclin-1 and LAMP1, indicating accumulation of autophagy-associated markers (Fig. 7b; HP: p62, fold change 1·27 ± 0·04, p = 0·0002 vs. WT; LC3-II/LC3-I ratio, 1·431 ± 0·15, p = 0·03; LC3-II, 1·838 ± 0·21, p = 0·002; Beclin-1, 1·32 ± 0·09, p = 0·04; LAMP1, 2·24 ± 0·13, p = 0·0001; CX: p62, 1·30 ± 0·10, p = 0·01; LC3-II/LC3-I ratio, 3·20 ± 0·32, p = 0·0001; LC3-II, 1·78 ± 0·19, p = 0·002; Beclin-1, 1·28 ± 0·05, p = 0·008; LAMP1, 1·68 ± 0·07, p = 0·004).
In contrast, 5xTFEB mice showed significant reductions in p62 and LC3-II levels in both regions, with Becline-1 and LAMP1 levels restored toward WT values (Fig. 7b; HP: p62, 1·05 ± 0·04, p = 0·002 vs. 5xFAD; LC3-II/LC3-I ratio, 1·02 ± 0·07, p = 0·04; LC3-II, 1·11 ± 0·21, p = 0·84; Beclin-1, 0·87 ± 0·10, p = 0·004; LAMP1, 1·15 ± 0·20, p < 0·0001; CX: p62, 1·05 ± 0·06, p = 0·03; LC3-II/LC3-I ratio, 2·01 ± 0·36, p = 0·02; LC3-II, 1·23 ± 0·19, p = 0·41; Beclin-1, 1·22 ± 0·08, p = 0·72; LAMP1, 1·15 ± 0·17, p = 0·02).
Together, these results indicate that microglia-specific TFEB overexpression is associated with normalization of lysosomal and autophagy-related markers in the 5xFAD brain, by reducing lysosomal overload, re-establishing autophagic insufficiency, and suppressing microglial activation.
Microglia-specific TFEB overexpression reduces NLRP3 inflammasome activation
To determine whether TFEB overexpression modulates microglial inflammatory activation in AD, we performed immunohistochemical analyses using Iba1, a pan-microglial marker, and CD68, a marker of activated microglia, in the HP and CX of WT, 5xFAD, and 5xTFEB mice. Immunofluorescence revealed marked increases in CD68 and Iba1 immunoreactivity in 5xFAD mice compared with WT controls (Fig. 8a; HP: CD68 fold change, 15·15 ± 2·30, p = 0·0001 vs. WT; Iba1, 3·66 ± 0·15, p < 0·0001; CD68^+^Iba1^+^, 15.25 ± 1·25, p < 0·0001; CX: CD68, 25·48 ± 4·54, p = 0·0001; Iba1, 3·23 ± 0·20, p < 0·0001; CD68^+^Iba1^+^, 15·66 ± 1·17, p = 0.0001), consistent with enhanced microglial activation. In 5xTFEB mice, CD68 intensity was significantly reduced and the Iba1 signal was modestly decreased relative to 5xFAD mice (Fig. 8a; HP: CD68 fold change, 7·97 ± 2·05, p = 0·03 vs. 5xFAD; Iba1, 1·85 ± 0·25, p < 0·0001; CD68^+^Iba1^+^, 8·68 ± 1·04, p < 0·0001; CX: CD68, 9·66 ± 2·67, p = 0·01; Iba1, 2·12 ± 0·30, p < 0·0001; CD68^+^Iba1^+^, 9·04 ± 1·23, p = 0·008).
Fig. 8. Microglia-specific TFEB overexpression reduces NLRP3 inflammasome activation in 5xFAD mice. a Representative confocal images of hippocampus (HP) and cortex (CX) from 7-month-old WT, 5xFAD, and 5xTFEB mice stained for CD68 (green), Iba1 (red), and DAPI (blue). Quantification revealed significantly increased CD68 and Iba1 immunoreactivity in 5xFAD compared with WT, which was attenuated in 5xTFEB (WT: n = 6, 5xFAD: n = 6, 5xTFEB: n = 5, Scale bar, 20 μm). b Western blot analysis of HP and CX lysates showing elevated expressions of NLRP3, ASC, cleaved caspase-1, and mature IL-1β in 5xFAD mice, with significant reductions in 5xTFEB mice (WT: n = 7, 5xFAD: n = 8, 5xTFEB: n = 8). c ELISA quantification of cortical IL-1β levels showing a significant increase in 5xFAD mice compared with WT controls, whereas 5xTFEB mice exhibited a marked reduction in IL-1β levels relative to 5xFAD mice (n = 5 per group). Data are presented as mean ± SEM. Statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparisons test (a-c). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001
To further examine inflammasome-associated signalling, we performed immunoblot analyses of key pathway components. In both the HP and CX lysates from 5xFAD mice, protein levels of NLRP3, ASC, cleaved caspase-1 (C-caspase-1), and mature IL-1β were significantly elevated compared to WT mice (Fig. 8b; HP: NLRP3 fold change, 1·56 ± 0·19, p = 0·04; ASC, 2·92 ± 0·66, p = 0·01; C-caspase-1, 1·37 ± 0·11, p = 0·01; IL-1β, 1·41 ± 0·14, p = 0·02; CX: NLRP3, 1·41 ± 0·14, p < 0·0001; ASC, 7·09 ± 1·44, p < 0·001; C-caspase-1, 1.41 ± 0.04, p = 0.02; IL-1β fold change, 1.42 ± 0.12, p = 0.01). In contrast, 5xTFEB mice exhibited significantly lower expression of these inflammasome-related proteins compared with 5xFAD controls in both regions (Fig. 8b; HP: NLRP3, 0·71 ± 0·12, p = 0·002; ASC, 1·20 ± 0·20, p = 0·03; C-caspase-1, 1·03 ± 0·09, p = 0·02; IL-1β, 0·98 ± 0·09, p = 0·02; CX: NLRP3, 0·98 ± 0·08, p < 0·001; ASC, 2·95 ± 0·39, p = 0·01; C-caspase-1, 1·07 ± 0·09, p = 0·003; IL-1β, 0·95 ± 0·08, p = 0·004).
Consistent with these observations, ELISA quantification demonstrated significantly reduced IL-1β levels in 5xTFEB mice compared with 5xFAD mice (WT, 4·58 ± 2·41; 5xFAD, 14·31 ± 3·11; p < 0·0001 vs. WT; 5xTFEB, 9·67 ± 1·99; p = 0·03 vs. 5xFAD) (Fig. 8c). In addition, immunohistochemical analyses revealed reduced astrocytic gliosis in 5xTFEB mice relative to 5xFAD controls (Supplementary Fig. 8).
Together, these data indicate that microglia-specific TFEB overexpression is associated with reduced glial activation and attenuation of NLRP3 inflammasome-associated inflammatory signalling in the 5xFAD brain.
TFEB overexpression alters microglial gene expression and morphology
To investigate the effects of microglia-specific TFEB overexpression on genes associated with microglial activation, microglia were isolated from 7-month-old 5xFAD and 5xTFEB mice and analysed for mRNA expression. RT-qPCR analysis revealed that the transcript levels of Nlrp3,* Casp1*, and Tyrobp, a key adaptor protein involved in microglial immune signalling, were significantly reduced in 5xTFEB microglia compared with 5xFAD controls (Fig. 9a; Nlrp3: ΔΔCt-derived fold change, 0·53 ± 0·15, p = 0·03; Casp1: 0·51 ± 0·12, p < 0·01; Tyrobp: 0·46 ± 0·13, p = 0·04). These results indicate that TFEB overexpression is associated with reduced expression of inflammasome-related and immune activation-associated genes in disease-associated microglia (DAM). A heatmap illustrating DAM gene expression patterns across WT, 5xFAD, and 5xTFEB groups is shown in Supplementary Fig. 9.
Fig. 9. Reduced inflammasome activation and Changes in microglial morphology with TFEB overexpression in microglia. a Nlrp3,* Casp1* and Tyrobp mRNA levels were significantly decreased in microglia isolated from 5xTFEB mice compared with 5xFAD mice (n = 5 per group). b Representative confocal images of microglia from 7-month-old 5xFAD and 5xTFEB mice stained with Iba1. Shown are corresponding skeletonized images and maps highlighting branch points (red) and endpoints (green). Scale bar, 20 μm. Quantification demonstrates significantly reduced cell territory volume and cell volume in 5xTFEB microglia compared with 5xFAD. Average branch length, number of branch points, and number of end points were significantly increased in 5xTFEB microglia compared with 5xFAD. Cells analysed: 5xFAD = 154, 5xTFEB = 72, obtained from n = 5 mice per group. Data are presented as mean ± SEM. Statistical analyses were performed using Student’s t-test (a-b). * p < 0.05, ** p < 0.01
To determine whether these transcriptional changes were accompanied by alterations in microglial morphology, we performed quantitative morphological analyses of Iba1 + microglia in 7-month-old 5xFAD and 5xTFEB mice. A total of 342 microglia from 5xFAD mice and 270 microglia from 5xTFEB mice were included in the analysis (Supplementary Fig. 10). Microglia from 5xFAD mice exhibited significantly larger territory volumes (5xFAD, 5121 ± 373; 5xTFEB, 1250 ± 132; p < 0.0001) and increased cell volumes (5xFAD, 3621 ± 283; 5xTFEB, 927 ± 61; p < 0.0001) compared with those from 5xTFEB mice (Fig. 9b).
In contrast, microglia from 5xTFEB mice displayed greater morphological complexity, as evidenced by increased average branch length (5xFAD, 16.44 ± 0.44; 5xTFEB, 34.45 ± 1.27; p < 0.0001), a higher number of branch points (5xFAD, 11.64 ± 0.36; 5xTFEB, 20.05 ± 0.98; p < 0.0001), and more endpoints (5xFAD, 10.86 ± 0.36; 5xTFEB, 20.36 ± 0.92; p < 0.0001) relative to 5xFAD mice (Fig. 9b).
Together, these findings indicate that microglia in 5xTFEB mice exhibit reduced expression of inflammatory and immune activation-related genes and adopt a more complex, ramified morphology, consistent with a less activated microglia state.
Discussion
In this study, we demonstrate that microglia-specific overexpression of TFEB restores ALP-associated markers, enhances Aβ clearance, attenuates neuroinflammatory signalling, and improves cognitive performance in the 5xFAD mouse model of AD. These findings underscore the pivotal role of microglial ALP in modulating AD-related pathology and highlight TFEB as a promising molecular target for cell type-specific therapeutic intervention. Consistent with this concept, prior studies have shown that deacetylation-driven TFEB activation enhances microglial degradation of fibrillar Aβ [18], neuronal TFEB reduces BACE1-dependent Aβ generation and amyloid pathology [19], and pharmacological TFEB activators mitigate AD-like phenotypes in experimental models [20].
Our results provide convergent evidence that ALP dysfunction occurs in the 5xFAD brain and microglia, as reflected by reduced nuclear localization of TFEB, accumulation of p62 and LC3-II, and downregulation of microglial Tfeb transcripts. These alterations are consistent with previous reports showing that Aβ exposure promotes cytoplasmic retention of TFEB in microglia and that Tfeb expression is reduced in Psen1_KI/KI_ microglia [21, 22]. Using a tamoxifen-inducible, microglia-specific TFEB overexpression model, we demonstrate that TFEB activation enhances ALP-related processes and facilitates oligomeric Aβ degradation. Live-cell imaging with pH-sensitive Aβ probes further revealed accelerated phagolysosomal trafficking in TFEB-overexpressing microglia, supporting improved lysosomal engagement during Aβ processing.
In vivo, microglial TFEB overexpression was associated with robust reductions in amyloid burden across the CX, HP, and EC. Histological biochemical analyses consistently demonstrated decreases in both fibrillar and soluble Aβ species, including Aβ_1–42_. These neuropathological changes were accompanied by significant improvements in working memory, associative learning, and spatial memory in both male and female mice. Together, these data support the concept that enhancing microglial TFEB activity is linked to reduced amyloid pathology and functional recovery in AD.
To verify the cell-type specificity of TFEB activation, we examined lysosomal marker expression across major brain cell populations. Increased LAMP1 immunoreactivity in 5xTFEB mice was confined to Iba1 + microglia, with no detectable upregulation in GFAP⁺ astrocytes or NeuN⁺ neurons, confirming effective microglia-specific targeting using the Cx3cr1-CreERT2 system. While these findings indicate that TFEB-driven lysosomal enhancement occurs predominantly in microglia, indirect effect on astrocytic or neuronal function secondary to altered microglial activity cannot be excluded.
Transcriptomic analyses further revealed that TFEB overexpression is associated with a coordinated shift in gene expression toward lysosomal and autophagic pathways, alongside suppression of inflammatory and apoptotic programs. Enrichment of gene set related to lysosome organization, autophagosome maturation, and endosomal transport, coupled with downregulation of cytokine signalling pathways, aligns with the CLEAR network model of TFEB-mediated transcriptional control [23–25]. These data highlight the dual capacity of TFEB to promote proteostatic clearance while restraining aberrant immune activation in the AD brain.
Importantly, our findings suggest a mechanistic link between TFEB-mediated ALP enhancement and the regulation of inflammasome activity. Previous studies have shown that microglial NLRP3 activation and ASC speck formation contribute to amyloid pathology and cognitive decline [26–29], and that autophagy can limit Aβ-induced inflammasome signalling by facilitating degradation of extracellular fibrils and directly targeting inflammasome components [29, 30]. TFEB has been implicated in inflammasome turnover through lysosomal and chaperone-mediated autophagy, with reduced TFEB activity exacerbating NLRP3 activation and TFEB-driven lysosomal programs promoting its clearance [31]. Consistent with this framework, we observed elevated levels of NLRP3, ASC, cleaved caspase-1, and mature IL-1β in 5xFAD mice, all of which were significantly reduced in 5xTFEB mice. These findings indicate that microglial TFEB overexpression is associated with attenuation of inflammasome-related inflammatory signalling in the AD brain.
Taken together, our results demonstrate that targeted TFEB activation in microglia reprograms both degradative and inflammatory pathways, leading to enhanced Aβ clearance, reduced neuroinflammation, and cognitive improvement in an amyloid-driven AD model. Within the context of existing evidence, microglia-targeted modulation of TFEB to enhance ALP represents a compelling cell type–specific strategy that warrants further preclinical investigation.
This study has several limitations. First, transcriptomic analyses were performed using bulk hippocampal and cortical tissue, which limited the resolution of cell-type specific transcriptional changes. Because microglia-specific transcriptomic profiling was not conducted, TFEB-dependent alterations in microglial autophagy- or inflammation-related gene programmes could not be directly assessed at the transcript level. Although microglia fractions isolated from 5xTFEB brains exhibited reduced expression of inflammatory markers, including NLRP3, caspase-1, and TYROBP, single-cell or spatial transcriptomic approaches will be required to validate these changes in situ and to resolve microglial subtype-specific effects. Second, in female 5xTFEB mice, cognitive improvement occurred in the absence of statistically significant reductions in amyloid pathology, suggesting a potential sex-specific dissociation between behaviour outcomes and plaque burden. This observation indicates that microglial TFEB activation may influence cognitive function through mechanisms that are not solely dependent on amyloid load and warrants further investigation into sex-dependent responses. Third, although the 5xFAD model robustly recapitulates aggressive amyloid pathology and microglial activation, it lacks overt tauopathy. Accordingly, the impact of microglial TFEB activation on tau-related pathological processes remain unresolved and should be addressed in future studies using tauopathy-relevant models.
Conclusion
This study identified microglial TFEB as a central regulator of ALP function and a critical disease-modifying factor in AD. Using a tamoxifen-inducible, microglial-specific TFEB-overexpressing 5xFAD model, we demonstrated that targeted TFEB activation restores autophagic and lysosomal activity, promotes efficient clearance of both diffuse and fibrillar Aβ species, suppresses inflammasome-associated signalling, and rescues cognitive deficits in both male and female mice. Transcriptomic profiling further revealed broad reprogramming of microglial gene expression, characterized by upregulation of lysosomal and autophagic pathways and concomitant downregulation of inflammatory and apoptotic signalling. These molecular and cellular changes align with the observed attenuation of amyloid pathology and cognitive decline.
Collectively, our findings demonstrate that enhancing microglial TFEB restores autophagy, restrains maladaptive neuroinflammation, and mitigates AD-related neurodegeneration. Microglial TFEB modulation therefore represents a promising, cell type–specific therapeutic strategy with translational relevance for AD and related disorders. Future studies should extend beyond 5xFAD to include tauopathy, mixed-pathology, and aged models to establish generalisability.
Methods
Animals
Microglia-specific TFEB overexpression mice were generated by crossing Cx3cr1^CreERT2^ mice (Jackson Laboratory) with a Cre-dependent human TFEB overexpression line. In this transgenic construct, the hTFEB coding sequence is placed downstream of a CAG promoter but is prevented from being expressed by a loxP-flanked transcriptional STOP cassette located upstream of the hTFEB open reading frame. This STOP cassette blocks promoter-driven transcription in the absence of Cre-mediated recombination. The TFEB-CA line was kindly provided by Prof. Myung-Shik Lee (Soonchungyang University, Korea) and maintained by breeding hemizygous transgenic animals with C57BL/6J mice (DBL, Eumseong-gun, Chungbuk, Korea).
The 5xFAD line overexpresses human APP695 carrying the Swedish (K670N, M671L), Florida (I716V), and London (V717I) familial Alzheimer’s disease (FAD) mutations and human PSEN1 carrying the M146L and L286V mutations [32]. 5xFAD mice were obtained from Seoul National University and maintained on a B6SJL F1 background (Jackson Laboratory, Bar Harbor, ME, USA).
To develop 5xTFEB-CA mice, TFEB-CA (C56BL/6 background) were first crossbred with SJL/J mice (Jackson Laboratory) to obtain a B6SJL background and then crossed with hemizygous 5xFAD mice. To induce TFEB overexpression, 5-week-old mice received an intraperitoneal injection of tamoxifen (80 mg/kg body weight; 20 mg/mL stock in 100% corn oil; Sigma-Aldrich, St. Louis, MO, USA) once daily for five consecutive days [33]. Brains were harvested from 9-weeks of age, allowing at least one month after the final tamoxifen injection for the turnover of CX3CR1^+^ circulating cells [34]. Genotyping was performed by polymerase chain reaction (PCR) using DNA from ear biopsies. Age-matched littermates of each genotype were used as controls for all experiments. Mice were housed under standard conditions (22 ± 2 °C, 50 ± 10% humidity, 12-h light/dark cycle) with ad libitum access to food and water. All animal experiments were approved by the Institution Animal Care and Use Committee of the Lee Gil Ya Cancer and Diabetes Institute, Gachon University (protocol codes LCDI-2024-0019 and LCDI-2024-0036). All procedures complied with the ARRIVE guidelines and were conducted in accordance with the U.K. Animal (Scientific Procedures) Act, 1986, and associated guidelines, the EU Directive 2010/63/EU for animal experiments, and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978).
Isolation and culture of adult mouse microglia
Mice were anesthetized (Zoletil, 12·5 mg/kg; Rompun, 17·5 mg/kg) and transcardially perfused with ice-cold Dulbecco’s phosphate-buffered saline (D-PBS; Welgene, Gyeongsangbuk-do, Korea). Brains were dissected, minced, and transferred to gentleMACS C-tubes (Miltenyi Biotec, Bergisch Gladbach, Germany). Tissue was dissociated using the Adult Brain Dissociation Kit (Miltenyi Biotec) according to the manufacturer’s instructions (enzyme P with buffer X, plus enzyme A with buffer Y; 37 °C for 30 min) on a gentleMACS Octo Dissociator with heaters (Miltenyi Biotec).
Cell suspensions were centrifuged (300 × g, 10 min, 4 °C), resuspended in 90 µL of 0.5% bovine serum albumin (BSA) in D-PBS (PB buffer), and incubated with CD11b MicroBeads (10 µL per 10^7^ cells; Miltenyi Biotec) for 15 min at 2–8 °C in the dark. After washing with 1 mL PB buffer and centrifugation (300 × g, 10 min, 4 °C), CD11b^+^ microglia were isolated by magnetic-activated cell sorting (MACS) using MS columns (Miltenyi Biotec) pre-equilibrated with PB buffer.
Isolated cells were resuspended in Dulbecco’s modified eagle medium (Welgene) supplemented with 10% fetal bovine serum (FBS; HyClone, Washington, DC, USA), L-glutamine (Sigma-Aldrich), and 1% penicillin-streptomycin (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), plated on poly-L-lysine-coated plates (Sigma-Aldrich) at 2 × 10^5^ cells/well (4-well plates) or 4 × 10^4^ cells/well (96-well plates), and maintained at 37 °C in a humidified incubator with 5% CO₂.
Preparation and pHrodo labelling of oligomeric Aβ
Synthetic Aβ_1−42_ peptide (Anaspec, Fremont, CA, USA) was dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP; Sigma-Aldrich), aliquoted, and evaporated under a fume hood. Dried peptide films were stored at − 80 °C until use. For oligomerization, dried peptide was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) at 5 mM and diluted with Ham’s F12 medium (Welgene) to 100 µM, followed by incubation at 4 °C for 24 h.
For fluorescent labelling, oligomeric Aβ (oAβ; 100 µL) was centrifuged at 16,000 × g for 10 min, resuspended in 0.1 M sodium bicarbonate, and incubated with 8·9 mM pHrodo Red succinimidyl ester (Thermo Fisher Scientific) for 2 h at 37 °C in the dark. pHrodo labelled oAβ (pHrodo-oAβ) was washed with methanol, re-pelleted, resuspended in Hanks’ balanced salt solution, aliquoted, and stored at − 80 °C or used immediately at 1 µM.
Phagocytosis assay
Microglia (2 × 10^5^ cells/well; 4-well plates) were cultured for 7 days, then incubated with 1 µM pHrodo-oAβ. Live-cell imaging was performed on the Incucyte S3 system (Sartorius, Göttingen, Germany) within a 37 °C, 5% CO₂ incubator. Fluorescence and phase-contrast images were acquired hourly for 24 h under identical imaging parameters across all conditions. Each condition was run in triplicate with four fields per well. The phagocytosis index was calculated as the number of phagocytosing cells divided by the number of total cells.
Immunocytochemistry
Cells were fixed with 4% paraformaldehyde, permeabilized with 0·2% Triton X-100 in PBS, and blocked with 10% BSA (0·2% Triton X-100 in PBS) for 1 h at room temperature. Cells were incubated overnight at 4 °C with primary antibodies against human-specific TFEB (Cell Signaling Technology, Danvers, MA, USA), β-amyloid (1–16; Biolegend, San Diego, CA, USA), and LAMP1 (Abcam, Cambridge, UK). After three PBS washes, cells were incubated for 1 h at room temperature in the dark with AlexaFluor 488-conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA, USA) and an AlexaFluor 555-conjugated goat anti-mouse IgG (Invitrogen). Nuclei were counterstained with DAPI (Vector Laboratories, Burlingame, CA, USA). Images were acquired on a Zeiss LSM700 confocal microscope (Zeiss, Jena, Germany) or Nikon TS2-S-SM with a DS-Qi2 camera. RIO-based fluorescence quantification with background subtraction was performed in ImageJ (version 1.54p, NIH, Bethesda, MD, USA).
Cell viability assay (MTT)
Microglia (4 × 10^4^ cells/well; 96-well black plates) were cultured for 7 days and treated with oAβ (0·5, 1, or 5 µM) for 24 h. Cell viability was assessed by adding thiazolyl blue tetrazolium bromide (MTT, 5 mg/mL in PBS; Sigma-Aldrich) to each well, followed by incubation at 37 °C for 4 h in 5% CO_2_. Supernatants were removed and formazan crystals were dissolved in DMSO. Absorbance was measured at 570 nm using a Victor X4 Multilabel Plate Reader (PerkinElmer, Waltham, MA, USA). Each condition was replicated five times.
Behavioral assessments
Cognitive and motor functions were assessed using the Y-maze, passive avoidance test (PAT), and Morris water maze (MWM) test. All sessions were automatically recorded and analysed using EthoVision XT 17 (Noldus, Wageningen, the Netherlands). Group allocation was determined by PCR genotyping and investigators were blinded to genotype. Cages and animals were labelled with coded animal IDs that did not reveal the group. Testing order was randomized across groups each. Tracking and scoring were computed automatically with pre-specified settings in EthoVision. Where any manual checks were required, they were performed by an investigator blinded to genotype. Exclusion criteria included insufficient locomotor activity or failure to engage with the task during training.
In the Y-maze test, mice were placed in a white polyvinyl plastic maze with three arms (40 cm long, 6·8 cm wide, and 15·5 cm high) and allowed to explore for 8 min. Consecutive entries into three different arms were given one point (alternation). Spontaneous alternation was calculated using the formula: (number of alternations/[total entries-2] × 100 (%).
The PAT was performed over three continuous days in a Gemini Passive Avoidance System (San Diego, CA, USA) with illuminated and dark chambers separated by a guillotine door. On day 1, mice were habituated to both chambers. On day 2, mice entering the dark chamber received a 0·3 mA foot shock for 3 s. On day 3, latency to enter the dark chamber was recorded as a measure of memory retention.
In the MWM test, mice were trained in a circular pool (90 cm diameter, 45 cm high, water maintained at 22 °C, made opaque with non-toxic paint) to locate a hidden platform. Four training trials were conducted per day for four consecutive days, with the platform kept in a fixed location. The escape latency (time to locate the platform) was recorded. After the mice found the hidden platform, they were kept on it for 5 s. If the mice could not find the platform within 60 s, they were placed on the platform for 20 s to encode the location of the escape platform; for these trials, the escape latency was recorded as 60 s. Afterwards, the mice were removed from the pool, dried, and returned to their home cage. In the probe trial, conducted within 48 h of the final training session, the platform was removed, and the mice were allowed to swim for 60 s. The time spent in the target quadrant, swim distance, velocity, and crossing frequency were quantified as measures of spatial learning and memory.
Tissue Preparation and Immunofluorescence
Mice were anesthetized with a mixture of Zoletil (8·3 mg/kg) and Rompun (15 mg/kg), and the brains were harvested. One hemisphere was dissected into the cortex (CX) and hippocampus (HP), snap frozen in liquid nitrogen, and stored for immunoblotting. The contralateral hemisphere was fixed in 4% paraformaldehyde at 4 °C for 24 h, cryoprotected in 30% sucrose for 3 days, and embedded in optimal cutting temperature compound (Sakura, Osaka, Japan). Coronal Sect. (30 μm) were prepared using a cryomicrotome (Thermo Fisher Scientific) and stored in cryoprotectant (30% ethylene and 30% glycerol in PBS) at 4 °C. The cryostat work was performed at a Core-Facility for Cell to In-vivo Imaging.
For immunofluorescence, free-floating sections were permeabilized with 0·4% Triton X-100 in PBS, incubated in 1% BSA in 0·4% Triton X-100 in PBS for 1 h at room temperature, and incubated overnight at 4 °C with primary antibodies against human-specific TFEB (Cell Signaling Technology), Iba1 (FUJIFILM Wako, Osaka, Japan), β-amyloid (1–16; Biolegend), Lamp1 (Abcam), and CD68 (Invitrogen). After washing with PBS, sections were incubated with AlexaFluor 488-conjugated donkey anti-rabbit IgG (Invitrogen) and an AlexaFluor 555-conjugated donkey anti-goat IgG (Invitrogen) for 1 h in the dark at room temperature. Nuclei were counterstained with DAPI (Vector Laboratories, Burlingame, CA, USA). Images were acquired using an LSM900 confocal microscope (Zeiss, Jena, Germany) and a Nikon TS2-S-SM microscope (Nikon Microscopy) equipped with a Nikon DS-Qi2 camera. Morphological analysis was performed using ImageJ (NIH) with region of interest-based intensity measurements after background subtraction.
RNA extraction and real-time quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted from dissected cortical tissues using TRIzol reagent (Qiagen, Hilden, Germany). RNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific), and 1 µg of total RNA was reverse transcribed into cDNA using a cDNA synthesis kit (TOYOBO, Osaka, Japan). RT–qPCR was performed on a real-time PCR system (Applied Biosystems, Waltham, MA, USA) using SYBR Green real-time PCR master mix (TOYOBO) and gene-specific primers (Supplementary Table 1). Relative gene expression levels were calculated using the ΔΔCt method, with GAPDH used as the internal control. To screen autophagy-related gene expression profiles, including Gabarapl2 and Atg5, RT–qPCR was additionally performed using the AccuTarget™ qPCR Screening Kit (mouse autophagy pathway; SM-0000-30; Bioneer, Daejeon, Republic of Korea) according to the manufacturer’s instructions.
For microglial gene expression analysis, microglia were isolated from mouse brains using magnetic-activated cell sorting (MACS), and total RNA was extracted from the isolated microglia using the same RNA extraction and RT–qPCR procedures as described above.
Western blot
Cortical and hippocampal tissues from 6-month-old mice were homogenized in ice-cold radio-immune precipitation assay buffer (RIPA buffer; 150 mM NaCl, 1% NP-40, 0·5% sodium deoxycholate, 0·1% sodium dodecyl sulphate [SDS], 50 mM Tris, pH 8·0) or fractionated using NE-PER nuclear and cytoplasmic extraction reagents (ThermoFisher Scientific). Protease inhibitors (Roche, Basel, Switzerland) and phosphatase inhibitor (Sigma-Aldrich) were added. The protein concentration was determined with a bicinchoninic acid (BCA) protein assay kit (ThermoFisher Scientific). Samples were denatured at 95 °C for 10 min, resolved on 8–15% SDS-polyacrylamide gels (SDS-PAGE), and transferred to polyvinylidene fluoride membranes (Merck Millipore, Burlington, MA, USA). Membranes were blocked with 3% BSA in Tris-buffered saline with Tween (TBS-T) and incubated overnight at 4 °C with primary antibodies against TFEB (Bethyl Laboratories, Montgomery, TX, USA), GAPDH (Santa Cruz Biotechnology, Dallas, TX, USA), Lamin-B1 (Abcam), p62/SQSTM1 (Progen, Heidelberg, Germany), LC3 (Novus Biologicals, Littleton, CO, USA), human-specific TFEB (Cell Signaling Technology), AIF-1/Iba1 (Novus Biologicals, Englewood, CO, USA), anti-β-amyloid (1–16; Biolegend), Beclin1 (Cell Signaling Technology), Lamp1 (Abcam), NLRP3 (Adipogen, San Diego, CA, USA), ASC (Cell Signaling Technology), cleaved-caspase1 (Cell Signaling Technology), and interleukin (IL)-1β (Santa Cruz Biotechnology). After washing with TBS-T, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (goat anti-rabbit, goat-anti-mouse, or goat-anti-guinea pig IgG; Invitrogen) and visualized using the D-plus ECL PICO system (ELPIS-biotech, Daejeon, Korea) or the Immobilon Western chemiluminescent HRP substrate (Merck Millipore, Burlington, MA, USA). Band intensity was quantified using ImageJ (NIH).
Enzyme-linked immunosorbent assay (ELISA)
Brain tissues were homogenized in RIPA buffer containing protease inhibitors, and lysates were centrifuged (20,000 × g, 15 min, 4 °C) to obtain supernatants. Protein concentrations were quantified using a BCA assay (Thermo Fisher Scientific) and equal amounts of protein were used for subsequent ELISA analyses. Human \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{A}{\upbeta\:}}_{1-42}$$\end{document} levels were measured using the Amyloid beta 42 Human ELISA Kit, Ultrasensitive (Invitrogen, KHB3544), and IL-1β levels were quantified using the IL-1β ELISA assay kit (DY401, R&D Systems, Minneapolis, MN, USA), following the manufacturers’ protocol instructions. Each standard and experimental sample was run in triplicate, and concentrations were calculated based on a standard curve, and the results were averaged.
Microglial morphology analysis (3DMorph)
Iba1-labeled microglia were imaged as z-stacks (4096 × 4096; z-step 0·5 μm) on a Zeiss LSM 900 confocal microscope with a 20× objective. Images were exported as multi-page TIFF (Fiji/ImageJ) preserving channels and z-metadata and analysed with 3DMorph [35]. The spatial scale was set to 0·36 × 0·36 × 0·5 µm^3^. Cells were segmented as three dimensional (3D)-connected components, and partial cells or edge-contacting objects were excluded from further analysis. For each cell, the soma volume, territorial volume, endpoints, branch points, and mean branch lengths were calculated with 3D skeletonization (Skeleton3D). Outputs (images and CSV files) were exported for downstream statistical analysis.
Bulk RNA-Seq and analysis
Total RNA from the CX and hippocampus (HP) of 7-month-old wild-type (WT), 5xFAD, and 5xTFEB mice was extracted using TRIzol reagent (Qiagen, Hilden, Germany). The RNA concentration and purity were initially assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific), and further quality control for RNA integrity was performed by Ebiogen (Seoul, Republic of Korea) prior to RNA sequencing. RNA-Seq data were initially processed using the ExDEGA software package (EBIOGEN, Seoul, Republic of Korea) for quality control and primary expression profiling. Differential expression analysis was subsequently performed in R (version 4·5·1; R Foundation for Statistical Computing, Vienna, Austria) using DESeq2 (version 1·48·1; Bioconductor Project, Seattle, WA, USA) with normalization and variance stabilization according to the standard workflow. Differentially expressed genes (DEGs) were defined as those with a fold change ≥ 1·2 and a p-value < 0·05. Functional enrichment analysis of DEGs was conducted using Gene Ontology (GO) terms to identify overrepresented biological processes. In addition, gene–gene interaction networks were explored using the STRING database (https://string-db.org/) to investigate functional associations among the identified DEGs.
SEA-AD microglia transcriptomes analysis
Human microglial snRNA-seq data were obtained from the SEA-AD Atlas (Allen Institute; AWS S3 bucket: https://sea-ad-single-cell-profiling.s3.amazonaws.com/index.html#Microglia-and-Immune-for-AAIC/; accessed on 17.Nov.2025) and were downloaded and processed using Python (Scanpy v1.10). The dataset includes microglia and other immune-cell nuclei obtained from multiple brain regions across donors with varying degrees of Alzheimer’s disease (AD) neuropathology. Metadata containing donor-level neuropathological scores, including Braak staging, were extracted from the AnnData object. To examine disease-associated transcriptional changes, samples were stratified into two groups based on established neuropathological criteria: Control (Braak 0–2) and AD (Braak 5–6), while intermediate stages (Braak 3–4) were excluded. To examine lysosomal–autophagy pathway alterations, log1p-normalized counts per million (CPM) values were computed and raw UMI counts were library-size–normalized to CPM and transformed using log1p. Gene-level expression values were then aggregated across all brain regions to obtain group-level means for each target gene. Differential expression between Control and AD groups was assessed using the non-parametric Mann–Whitney U test. Citation followed the Allen Institute policy (https://alleninstitute.org/citation-policy/).
Statistical analysis
Statistical analyses were performed in GraphPad Prism 9·1·0, and data are reported as mean ± standard error of the mean (SEM). Two-group comparisons used two-tailed unpaired Student’s t-tests; multiple-group comparisons used one-way ANOVA with Tukey’s post-hoc test; and designs with two independent factors used two-way ANOVA with Bonferroni’s post-hoc test. Assumptions of normality and homogeneity of variance were examined using the Shapiro–Wilk and Levene tests, respectively; when these were violated, non-parametric alternatives were applied (Mann–Whitney U or Kruskal–Wallis followed by Dunn’s multiple-comparison test). Unless otherwise specified, statistical tests were two-sided with α = 0·05, and the exact n, unit of analysis, and test used are provided in the figure legends. The experimental unit was the individual mouse. Animals were assigned to genotype-defined groups (WT, 5xFAD, 5xTFEB) based on PCR genotyping. No formal a priori power calculation was undertaken; sample sizes were determined during study design from prior experience with the 5xFAD model and published reports showing robust behavioural phenotypes, and we therefore targeted ≥ 8 mice per behavioural group to accommodate variability [36]. Behavioural cohorts included both sexes (n = 9–14 per genotype), while histological, biochemical, and transcriptomic assays used independent cohorts (n = 4–7 per group), as detailed in the figure legends. Correlation analyses between cortical \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{A}{\upbeta\:}}_{1-42}$$\end{document} levels and individual behavioral outcomes were performed using Pearson’s correlation analysis, and the correlation coefficient (r) and corresponding p-value are indicated in each panel. Statistical significance was defined as p < 0.05.
Supplementary Information
Supplementary Material 1.
The reference list from the paper itself. Each links out to its DOI / PubMed record.