Microglial cGAS Deletion Preserves Intercellular Communication and Alleviates Amyloid-β-Induced Pathogenesis of Alzheimer's Disease

Abstract

Innate immune activation plays a crucial role in the pathogenesis of Alzheimer's disease (AD) and related dementias (ADRD). The cytosolic DNA sensing pathway, involving cGAMP synthase (cGAS) and Stimulator of Interferon Genes (STING), has emerged as a key mediator of neurodegenerative diseases. However, the precise mechanisms through which cGAS activation influences AD progression remain poorly understood. In this study, we observed significant up-regulation of cGAS-STING signaling pathway in AD. Notably, this increase is primarily attributed to microglia, rather than non-microglial cell types. Using an inducible, microglia-specific cGAS knockout mouse model in the 5xFAD background, we demonstrated that deleting microglial cGAS at the onset of amyloid-β (Aβ) pathology profoundly restricts plaque accumulation and protects mice from Aβ-induced cognitive impairment. Mechanistically, our study revealed cGAS promotes plaque-associated microglia accumulation and is essential for inflammasome activation. Moreover, we showed that restricting cGAS-mediated innate immunity is crucial for preserving inter-cellular communication in the brain and induces pleiotrophin, a neuroprotective factor. These findings offer novel insights into the specific roles of the innate immune system in AD employing a cell-type-specific approach. The conclusions provide a foundation for targeted interventions to modulate the microglial cGAS-STING signaling pathway, offering promising therapeutic strategy for AD treatment.

Keywords: Alzheimer's disease; cGAS; innate immune; microglia.

Figure 1

Microglia‐specific cGAS deletion protects mice from Aβ accumulation‐induced cognitive deficits. a) Western‐blot evaluation of cGAS protein levels in human postmortem brain samples. n = 7/group. b) qPCR measurement of cGAS mRNA levels from human spinal cord samples of normal and AD individuals. n = 5–6/group. c) Protein levels of cGAS and STING in mouse brain lysate. (Female, 9‐month‐old, n = 4/group). d) Cerebrum tissue mRNA levels of cGAS measured with qPCR. n = 4/genotype. e) Representative tracking, f) Latency (^## P = 0.0387, ^* P = 0.0409), and g) Percent time in the target quadrant during training of the MWM test. (Showing male data at 6 months old). h) Nest building test score. (7‐month‐old males and females, n = 8–12/group). Student's t‐test for (a,b), two‐way ANOVA followed with Šidák correction for (c,d,g, and h), repeated measures ANOVA for (f).

Figure 2

Selective microglial cGAS ablation significantly decreases plaque loads in 5xFAD mouse brains. a) Immunofluorescence staining of mouse cerebrum sections with a pan Aβ antibody to evaluate Aβ accumulation. Scale bars represent 500µm. b) Quantification of (a), n = 8–9/group, male and female combined, 7 months old. c) Evaluation of plaque loads using methoxy‐X04 staining (Male). Scale bars represent 500µm. d) Quantification of (c), n = 4/group. e) Western‐blot evaluation of APP cleavage product (ACP) levels in male cerebral TBS and TBSX lysates, n = 6/group. f) Aβ40, g) Aβ42, and h) Aβ38 levels in mouse brain lysates measured with ELISA. n = 11–12/group, male and female combined. A soluble fraction represents protein levels in TBS and TBSX combined equally, and insoluble fraction represents protein levels from GuHCl isolation. Student's t‐test for (b, d). Two‐way ANOVA followed with Šidák correction for (e–h).

Figure 3

cGAS ablation ameliorates microgliosis in 5xFAD mouse brains. a) Hippocampal IF co‐staining using Aβ and IBA1 antibodies with DAPI (blue color) in 7‐month‐old mice. Scale bars represent 200µm. b) Quantification of overall hippocampal microglial density, and c) plaque‐associated microglial density in the hippocampus area, n = 10 mice/group. d) Cortical IF co‐staining using Aβ and IBA1 antibodies with DAPI. Scale bars represent 100µm. e) Quantification of overall cortical microglial density, and f) plaque‐associated microglia density in cortex area, n = 10 mice/group. g) Analysis of plaque‐associated microglia count within a 10 µm distance to plaques using methoxy‐X04 co‐staining with PU.1 and DRAQ (DNA stain indicating nuclei). Scale bars represent 50 µm. h) Quantification of g), n = 35 plaques/genotype. i) Quantification of IBA1 IF staining in Loxp and mKO mice without plaques, n = 5 mice/group. Student's t‐test for (b, c, e, f, h, and i).

Figure 4

Loss of microglial cGAS limits Aβ‐induced transcriptomic signatures and suppresses microglial phagocytotic activity in 5xFAD mice. a) PCA analysis based on cerebrum tissue mRNA levels gathered from the NanoString neuroinflammation panel. (n = 3/group, 7‐month‐old males). b) Heatmap displaying pathway scores calculated based on cerebrum tissue NanoString data. c) Volcano plot showing DEGs by comparing 5xFAD‐Loxp to Loxp group. Red and blue color represents increased and decreased expression, respectively. The green color indicates genes of special interest. d) Volcano plot showing DEGs by comparing 5xFAD‐mKO to the 5xFAD‐Loxp group. e) Venn diagram showing overlap DEGs between plaque‐induced genes (5xFAD‐Loxp vs Loxp, blue color) and genes decreased upon cGAS deletion (5xFAD‐mKO vs 5xFAD‐Loxp, orange color) measured using NanoString (p ≤ 0.05). f) Microglial gene signature comparison among different genotypes based on NanoString counts. ns (not significant) or ^*(p < 0.05) denotes the significance level by comparing 5xFAD‐mKO to 5xFAD‐Loxp group. g) Western‐blot evaluation of CD68, LILRB4, and AXL in 7‐month‐old male cerebrum tissues. h) Quantification of g) using β‐Actin levels as endogenous control. n = 3–6/group. i) Representative histogram of flow cytometry detection of microglia population with or without Aβ phagocytosis (Aβ marked by methoxy‐X04). j) Quantification of the proportion of Aβ phagocytotic microglia among genotypes (Male, 5‐month‐old, n = 3/genotype). k) qPCR detection of TNFα and IL1β expression levels in mouse cerebrum samples (n = 3–4/group, male, 7‐month‐old). One‐way ANOVA followed with Tukey's test for (h). Two‐way ANOVA followed with Šidák correction for (j) and (k).

Figure 5

Selective microglial cGAS ablation prevents dystrophic neurites formation. a) Hippocampal IF co‐staining using Aβ and LAMP1 antibodies in 7‐months‐old mice. Scale bars represent 200µm. b) Quantification of overall hippocampal LAMP1 density, and c) percent area ratio in the hippocampus region, n = 7 mice/group. d) Cortical IF co‐staining using Aβ and LAMP1 antibody in 7‐months‐old mice. Scale bars represent 100µm. e) Quantification of overall cortical LAMP1 density, and f) percent area ratio in cortex region, n = 7 mice/group. g) IF staining, and h) Quantification of plaque‐associated LAMP1 levels, n = 7 mice/group. Scale bars represent 25µm. i) mRNA counts of Homer1 gene measured with NanoString. n = 3/genotype. Student's t‐test for (b, c, e, f, and h). Two‐way ANOVA followed with Šidák correction for i).

Figure 6

Deletion of microglial cGAS preserves Aβ accumulation‐induced loss of cell–cell communication in 5xFAD brain. a) Circle plot displaying the number of interactions among all major cell types from each genotype group. The thickness of lines represents the relative quantity of intracellular communication. b) Inferred total number of interactions and interaction strength in each group. c) Bubble plot showing the probability of ligand‐receptor interaction derived from microglia to each of the other 6 cell types (For the genotype label, L: Loxp; F: 5xFAD‐Loxp; M: 5xFAD‐mKO). d) Hierarchy plot of PTN signaling among cells in different genetic groups. e) Western‐blot evaluation and quantification of PTN protein levels in cerebrum tissues (male, 7‐month‐old, n = 3–6/genotype). f) IF evaluation and quantification of PTN protein levels and distribution in brain sections (7‐month‐old, n = 5/genotype). Scale bars represent 50µm. g) GO biological process analysis of DEGs (5xFAD‐mKO vs 5xFAD‐Loxp) from oligodendrocyte population. One‐way ANOVA followed with Tukey's test for (e). Student's t‐test for (f). Differential expression analysis g) was generated using model‐based analysis of single‐cell transcriptomics (MAST) test in the Seurat package.

Figure 7

Suppression of microglial innate immunity protects the integrity of neuronal populations in mice. a) Bar graph showing the frequency of each cell type in each genotype. b) UMAP showing sub‐clusters of inhibitory neuron population. c) Dot plot evaluating the expression of established sub type‐specific marker genes of inhibitory neurons. d) Bar graph showing DEG counts generated by indicated comparison pairs. e) Violin plot comparing the expression level of the Chat gene in cluster 8 among different genotypes. f) qPCR analysis of cholinergic neuron markers expression in cerebrum lysates (n = 6/genotype). g) WiKi pathway analysis using shared gene lists that are oppositely regulated in the following two comparisons: 5xFAD‐Loxp versus Loxp, and 5xFAD‐mKO versus 5xFAD‐Loxp. h) UMAP showing sub‐clusters from the excitatory neuron population. i) Dot plot displaying the expression of sub type‐specific marker genes from excitatory neurons. j) Syp expression in excitatory neuron population. k) Western‐blot of PSD95 and synaptophysin protein levels from mouse brain lysates, n = 6/genotype. l) Quantification of (k). Mouse semi‐cerebrum samples (n = 3) were pooled for sequencing of each genotype for snRNA‐seq. Two‐way ANOVA followed with Šidák correction for (f) and (l). p‐values for differential expression analysis between different clusters in (e) and (j) were calculated using the Mann‐Whitney U test.