Tau Accumulation Induces Microglial State Alterations in Alzheimer's Disease Model Mice

Abstract

Unique microglial states have been identified in Alzheimer's disease (AD) model mice and postmortem AD brains. Although it has been well documented that amyloid-β accumulation induces the alteration of microglial states, the relationship between tau pathology and microglial states remains incompletely understood because of a lack of suitable AD models. In the present study, we generated a novel AD model mouse by the intracerebral administration of tau purified from human brains with primary age-related tauopathy into App knock-in mice with humanized tau. Immunohistochemical analyses revealed that Dectin-1-positive disease-associated microglia were increased in the AD model mice after tau accumulation in the brain. We then performed single-nucleus RNA sequencing on the AD model mice to evaluate the differences in microglial states with and without tau propagation and accumulation. By taking advantage of spatial transcriptomics and existing single-cell RNA sequencing datasets, we showed for the first time that tau propagation and accumulation induce a disease-associated microglial phenotype at the expense of an age-related nonhomeostatic counterpart (namely, white matter-associated microglia) in an AD model mouse brain. Future work using spatial transcriptomics at single-cell resolution will pave the way for a more appropriate interpretation of microglial alterations in response to tau pathology in the AD brain.

Keywords: Alzheimer's disease; amyloid-β; microglia; single-cell RNA-seq; tau.

Figure 1.

Generation of a humanized AD mouse model with tau propagation and accumulation. A, B, Immunohistochemical analysis of phosphorylated tau in dKI mice at 6 (A) and 12 (B) months after PART tau injection. The AT8 (S202/T205) antibody was used to visualize phosphorylated tau (red). Scale bars, 2.5 mm. C–E, The area of AT8 signal was quantified by image analysis. PART tau was prepared from three subjects (#8568, #8619, #8739). Each dot represents an experimental point [n = 3 (6 m.p.i. #8568), n = 8 (12 m.p.i. #8568), n = 3 (12 m.p.i. #8619), and n = 4 (12 m.p.i. #8739), where each n reflects one mouse; *p < 0.05, t test. F, G, High-magnification images of phosphorylated tau (red) and amyloid plaques (green) in noninjected dKI (C) and Tau-dKI (D) mice. The N1D antibody (Saido et al., 1995) was used to visualize amyloid plaques. Scale bars, 100 μm. H–K, Immunohistochemical analyses using various tau antibodies against phosphorylated tau (H, CP13, S202; I, RZ3, T231; J, PHF1, S396/S404) or tau with altered structure (K, MC1). Scale bars, 2.5 mm. L, M, Immunostaining of three-repeat (L) and four-repeat (M) tau using specific antibodies (red) against each isoform and amyloid plaques (green). Note that both three- and four-repeat tau accumulated in the NP tau pathology. Scale bars represent 100 μm. N, O, Gallyas silver staining in the AD brain (K) and in Tau-dKI mice (L). Scale bars, 100 μm.

Figure 2.

Increased Dectin-1⁺ microglia are associated with tau accumulation in the hippocampus. A, Left, Schematic diagram showing the timings of tau injection and brain sampling. Right, AT8⁺ hyperphosphorylated tau accumulation (AT8) in the injected side of the hippocampus. Cell nuclei were visualized with 4′,6-diamidino-2-phenylindole (DAPI; blue). B, Confocal images of hippocampal sections stained with antibodies against Iba1 (purple), Dectin-1 (green), and galectin-3 (red). Dectin-1 and galectin-3 colocalization is shown as white. Cell nuclei were visualized with DAPI (blue). Images are taken from the dentate gyrus region. The top panels are representative images from dKI mice, and the bottom panels are representative images from Tau-dKI mice. Scale bars, 100 μm. C, Quantitative analyses of microglia in 24-month-old dKI and Tau-dKI mice. The numbers of Iba1⁺ microglia, Dectin-1⁺ microglia, Dectin-1/galectin-3 double-positive microglia, and Dectin-1/galectin-3 double-negative microglia are presented as boxplots (each black dot represents an experimental point; n = 3 mice per group; *p < 0.05, t test).

Figure 3.

Alteration of microglial states in the Tau-dKI hippocampus. A, UMAP plot based on the gene expression of 53,093 single nuclei. Ten clusters (0–9) were discriminated. Extended Data Figure 3-1 shows the expression of selected marker genes of the clusters. B, Dot plot showing average gene expression levels and the percentages of cells expressing marker genes across all clusters. Hexb, Cx3cr1, P2ry12, and Tmem119 were used as marker genes for microglia. Identified marker genes of all clusters are added as Extended Data Table 3-1. C, UMAP plot of the 11,551 single microglial nuclei that passed quality control after the removal of other border-related macrophages. Five microglial subclusters (0–4) were detected in both dKI (n = 3) and Tau-dKI (n = 3) mice. Extended Data Figure 3-2 includes information of all Hexb+ population subtracted from the whole dataset. Extended Data Figure 3-3 shows similarity scores for the known states in each microglial subcluster. D, UMAP projection plot of all microglial nuclei with color mapping to the transcriptional pseudotime, calculated using Monocle 3. E, Gene expression heatmap for the microglial clusters. The genes shown on the left are the top gene markers for each microglial subcluster. See Extended Data Figure 3-4 for Ctnna3 expression in our Tau-dKI model and other AD models, Extended Data Figure 3-5 for histological distribution of Ctnna3 mRNA in aged AppNL-G-F brain. F, Integrative analyses of spatial transcriptomics using a single-nucleus RNA-seq dataset. The likelihood of the presence of a microglial subpopulation at each spot on the injected side of a Tau-dKI brain section was estimated from single-nucleus RNA-seq data. Subcluster 0 was shown to predominantly localize to white matter (WM). The remaining two subclusters were present in gray matter (GM) at least as much as in WM. Extended Data Figure 3-6 includes basic information of the spatial transcriptomics. G, UMAP plot of microglia in dKI (5,731 nuclei) and Tau-dKI (5,820 nuclei) mice. Dot plot showing the frequencies of nuclei per cluster in dKI and Tau-dKI mice (each colored dot represents the cluster frequency from an experimental sample; n = 3 mice per group; *p < 0.05, t test, **p < 0.01, t test). The names of microglial subclusters (0–4) were changed to the previously defined states: white matter-associated microglia (WAM), intermediate disease-associated microglia (DAM1), homeostatic microglia (HM), disease-associated microglia (DAM2), and interferon response microglia (IRM). Differential expression analyses were performed between Tau-dKI and dKI (Extended Data Fig. 3-7).