Reactive microglia drive tau pathology and contribute to the spreading of pathological tau in the brain

Abstract

Pathological aggregation of tau is a hallmark of Alzheimer's disease and related tauopathies. We have previously shown that the deficiency of the microglial fractalkine receptor (CX3CR1) led to the acceleration of tau pathology and memory impairment in an hTau mouse model of tauopathy. Here, we show that microglia drive tau pathology in a cell-autonomous manner. First, tau hyperphosphorylation and aggregation occur as early as 2 months of age in hTauCx3cr1(-/-) mice. Second, CD45(+) microglial activation correlates with the spatial memory deficit and spread of tau pathology in the anatomically connected regions of the hippocampus. Third, adoptive transfer of purified microglia derived from hTauCx3cr1(-/-) mice induces tau hyperphosphorylation within the brains of non-transgenic recipient mice. Finally, inclusion of interleukin 1 receptor antagonist (Kineret®) in the adoptive transfer inoculum significantly reduces microglia-induced tau pathology. Together, our results suggest that reactive microglia are sufficient to drive tau pathology and correlate with the spread of pathological tau in the brain.

Keywords: Alzheimer’s disease; microglia; neuroinflammation; tau protein; tauopathies.

Neuroinflammation accelerates tau pathology, but the role played by microglia is uncertain. Maphis et al. provide direct evidence that reactive microglia are sufficient to drive hyperphosphorylation and aggregation of tau protein in a cell-autonomous manner, and show that microglial activation correlates with the spread of tau pathology in mice.

Figure 1

CX3CR1 deficiency accelerates tau hyperphosphorylation and p38 MAPK activation in hTau mice at early stages of development. (A) Western blot analysis of hippocampal lysates shows elevated tau phosphorylation on AT8, AT180 and PHF-1 sites and activation of p38 MAPK (pT180/pY182) in hTauCx3cr1^−/−mice at 2 months of age. No significant alterations in total tau (Tau5 antibody) or total (t) p38 MAPK levels. GAPDH was the loading control. Lysates from tau knockout mice were included in the Mapt^−/− lane and show no expression of tau. (B–E) Quantification of western blots for AT8, AT180, PHF-1 and phospho (p)-p38 MAPK reveal a statistically significant [n = 5 per group for 2, 12 and 24-month-old groups; n = 6 for 6-month-old group; mean ± SEM integrated density value ratio for respective epitopes (*P < 0.05 and **P < 0.01; two-way ANOVA Bonferroni’s multiple comparisons test)] increase in AT8 (B), AT180 (C), PHF-1 (D) and p-p38 MAPK (E) at 2-, 6- and/or 24-months of age in hTauCx3cr1^−/−mice compared to hTau mice. For p-p38 MAPK, both hTau and hTauCx3cr1^−/−mice show a biphasic response (overall increase at 2-, 12- and 24-months, reduction at 6 months).

Figure 2

Early onset and progression of tau pathology in the hippocampus of hTauCx3cr1^−/− mice. (A and B) Immunohistochemical analysis (using Sigma-Fast DAB with CoCl₂-metal enhancement therefore appears dark purple) shows presence of numerous AT8⁺ and AT180⁺ neurons in the dentate gyrus of 2- and 6-month-old hTauCx3cr1^−/− mice compared to age-matched hTau mice. By 6 months of age, hTau mice also show both AT8⁺ and AT180⁺ neurons. Scale bar = 50um (A and B). (C) Quantitative morphometric analysis shows a significant increase in AT8⁺ and AT180⁺ neurons in hTauCx3cr1^−/− mice from 2- to 12-months of age with an overall drop in AT8⁺ neurons at 12 months and AT180⁺ neurons at 24 months of age mean ± SEM immunoreactive area ratio of respective epitopes (C). (D) Gallyas silver positive pre-tangle aggregates (appeared intraneuronal) are evident in the CA1 neurons of hippocampus of 2-month-old hTauCx3cr1^−/− mice compared to hTau mice. Scale bar = 10 µm. (E) Sarkosyl insoluble tau is evident in the hippocampi of 2-month-old hTauCx3cr1^−/− mice compared to age-matched hTau mice.

Figure 3

Reduced SNAP25 levels, neuronal abnormality and impaired spatial memory in the hTauCx3cr1^−/− mice. (A and B) Western blot analysis of hippocampal lysates reveal reduced SNAP25 levels in the 6-month-old hTauCx3cr1^−/− mice. Quantification of western blots for SNAP25 and GAPDH revealed a statistically significant (*P < 0.05 for 12-month-old hTauCx3cr1^−/− mice versus 6-month-old hTauCx3cr1^−/− mice; n = 3; mean ± SEM of SNAP25/GAPDH ratio; unpaired t-test) decrease in the SNAP25 from 6 months to 12 months of age in hTauCx3cr1^−/− mice. (C and D) Thickness of the NeuN⁺CA1 layer appears reduced in the identical region of the CA1 subfield in the 2-month-old hTauCx3cr1^−/− mice. Scale bar = 20 µm. Wet weights of hippocampi were significantly (*P < 0.05; n = 5; mean ± SEM; unpaired t-test) lower in the 24-month-old hTauCx3cr1^−/− mice compared to age-matched hTau mice. (E–H) Visible and hidden platform versions of the Morris Water Maze. (E) At 6-months of age: mean latency to escape to a visible or hidden platform across training days was assessed from Day 1 through to Day 7. Note statistically significant differences (**P < 0.01 for hTauCx3cr1^−/− and *P < 0.05 for hTau mice versus non-transgenic mice on Day 2; *P < 0.05 for hTau versus non-transgenic on Day 4; **P < 0.01 for hTauCx3cr1^−/− versus non-transgenic on Day 7; two-way ANOVA with Tukey’s multiple comparison test; n = 7; mean ± SEM) in the mean escape latency on Days 2, 4 and 7. (F) At 12 months of age: platform proximity during visible and hidden training sessions are significantly (*P < 0.05 for hTau and hTauCx3cr1^−/− mice versus non-transgenic comparing mean ± SEM of three trials for each day; two-way ANOVA with Tukey’s—for the platform proximity; n = 6) altered in hTau and hTauCx3cr1^−/− mice compared to non-transgenic mice. (G and H) At 12 months of age: representative traces showing swim pattern of different groups of mice during probe test (when the hidden platform was removed and the time spent in target quadrant was assessed). Bottom right is the target quadrant. A probe test on Day 8 revealed only 12-month-old hTauCx3cr1^−/− mice spending significantly less time in the target quadrant (*P < 0.05 for hTauCx3cr1^−/− mice versus non-transgenic; one-way ANOVA with Tukey’s multiple comparison test; n = 6; mean ± SEM).

Figure 4

Microglial activation and IL1B maturation precedes spreading of tau pathology in hTauCx3cr1^−/− mice. (A) Iba1⁺microglia in the CA3 region of hippocampus of 2-month-old hTau and hTauCx3cr1^−/− mice. Scale bar = 10 µm. (B) Quantitative real-time PCR analysis of Il1b transcript levels reveal a significant (*P < 0.05; n = 5; mean ± SEM; unpaired t-test) increase in IL1B expression in the hemi-brains of 2-month-old hTauCx3cr1^−/− mice compared to age-matched non-transgenic controls. (C and D) Mature/active IL1B is present at elevated levels in hTauCx3cr1^−/− mice as early as 2 months of age. Overnight stimulation of HEK-Blue™ IL1B reporter cells (InvivoGen) with detergent (T-PER, Pierce) soluble hippocampal lysates from 2-month-old non-transgenic, hTau and hTauCx3cr1^−/− mice show significantly (**P < 0.01 one-way ANOVA with Tukey’s multiple comparison test; n = 3 mice per group; mean ± SEM) increased production of SEAP (NF-κB/AP-1-inducible SEAP reporter gene that is activated when mature IL1B in the lysate binds to IL1R1 on the surface of HEK-Blue™ IL1B cells). SEAP levels in the media were measured by QUANTI-Blue™ assay at 620 nm. (B–D) Levels of full-length and cleaved IL1B (17 kDa) are significantly higher in hTauCx3cr1^−/− mice (quantification for pro-IL1B/GAPDH in (C) C: *P < 0.05 for hTau mice and **P < 0.01 for hTauCx3cr1^−/− mice versus non-transgenic; for cleaved IL1B/GAPDH ratio in (D) D: *P < 0.01 for hTauCx3cr1^−/− versus non-transgenic mice. One-way ANOVA with Tukey’s multiple comparison; n = 3; mean ± SEM]. (E) Schematic showing hippocampal circuitry where subiculum (Sb) receives input from the CA1 and sends out projections to entorhinal cortex (EC). Shaded region (light red) shows the area where morphometric analysis was performed in subiculum. (F) CD45 reactive microglia are present in the subiculum of hTauCx3cr1^−/− mice as early as 2 months of age, however, the AT180⁺ tau is almost not detectable in the lateral subiculum at 2 months of age. More robust staining for CD45 and AT180 is apparent in the subiculum of 24-month-old hTauCx3cr1^−/− mice. Scale bars = 20 µm (all frames). (G and H) Quantification of CD45 immunoreactive area (G) and number of AT180⁺ neurons (H). Note, CD45 immunoreactive area, but not the number of AT180⁺ neurons, is significantly higher in hTauCx3cr1^−/− mice (*P < 0.05 for 2-month-old hTauCx3cr1^−/− versus non-transgenic; unpaired t-test; n = 4; mean ± SEM).

Figure 5

Adoptive transfer of purified microglia from hTauCx3cr1^−/− mice induces tau hyperphosphorylation in vivo. (A) GFP⁺ Cx3cr1^−/− microglia 14 days in vitro in primary culture. Scale bar = 20 µm. (B) A representative image showing AT8 immunoreactivity in the striatum of a naïve non-transgenic recipient mouse that received GFP⁺ Cx3cr1^−/− microglia from primary cultures. Scale bar = 20 µm. (C) Schematic showing the magnetic-based isolation of CD11b⁺ microglia and intracerebral injections of purified microglia into the recipient mouse brain. (D) A representative low magnification image showing a needle track from the brain surface through layer VI of the cortex and parts of the carpus callosum. GFP⁺ microglia from hTauCx3cr1^−/− mice are viable, appear activated and have migrated several hundreds of microns away from the injection site (green in the insets). Boxes with dashed line show representative areas where quantifications were done (for panel M). Scale bars = 20 µm (top left inset) and 10 µm (bottom right inset). (E–H) Microglia derived from 6-month-old hTauCx3cr1^−/− mice induce robust AT8⁺ staining in neurons/dystrophic neurites within the striatum of 2-month-old non-transgenic recipient mouse brain (H) compared to those derived from non-transgenic (F) or hTau (G) donor mice and RPMI media (vehicle) injected mice (E). hTauCx3cr1^−/− microglial recipients also showed several AT8⁺ pyramidal neurons around the needle track in the cortex (inset in H). Scale bar = 20 µm. (I–K) Representative confocal images showing hTauCx3cr1^−/− donor microglia within the recipient mouse brain (I) and the majority of them appear activated. In the same field, certain dystrophic neurites and cells show AT8 positivity (J and K). Scale bar = 20 µm. See also Supplementary Figs 5 and 6. (L) Inclusion of IL-1Ra with the hTauCx3cr1^−/− microglial inoculum shows no AT8 immunoreactive dystrophic structures/neurons within the recipient mice. Scale bar = 20 µm. (M) Quantification of AT8⁺ neurons/mm² revealed a statistically significant (n = 3–5 recipients per treatment; five sections per mouse; six random fields per section were scored; mean ± SD; *P < 0.05; **P < 0.001; one-way ANOVA with a Tukey’s multiple comparison test; each marker represent individual recipient mouse) increase in AT8⁺ neurons in the ipsilateral cortex of recipient mouse brain that received (i) primary Cx3cr1^−/− microglia (purple data set); and (ii) microglia from 6-month-old hTauCx3cr1^−/− donor mice (brown data set). Vehicle injected or those receiving microglia from 6-month-old non-transgenic or hTau donors did not show AT8⁺ neurons (black, blue and red data sets). Inclusion of IL-1Ra in the inoculum significantly reduced the number of AT8⁺ neurons (turquoise data set).

Figure 6

Inclusion of IL-1Ra in the microglial inoculum reduces overall reactivity for phosphorylated p38 MAPK within the recipient mouse brain. (A, B and E) Immunostained brain sections displaying overall reduced phosphorylated (p)-p38 MAPK (pT180/pY182) immunoreactivity around the needle track when the IL-1Ra was included with microglia from 6-month-old hTauCx3cr1^−/− donor mice (B) compared to those receiving hTauCx3cr1^−/− donor microglia alone (A). Scale bar = 30 µm. Quantification of percentage p-p38 MAPK immunoreactive area revealed a statistically significant (*P < 0.05; n = 3 animals per group; three sections per mice and four fields per section were used; mean ± SEM; unpaired t-test) decrease in the p-p38 MAPK⁺ area in the brains of non-transgenic mice that receiving microglia from hTauCx3cr1^−/− donor mice (E). (C and D) Triple immunofluorescence analysis revealing EGFP⁺ donor microglia within the recipient mouse brain. The area surrounding such microglia display relatively less p-p38 MAPK immunoreactivity (red in D) when the IL-1Ra was included with hTauCx3cr1^−/−microglial inoculum. Scale bar = 20 µm.