Soluble phospho-tau from Alzheimer’s disease hippocampus drives microglial degeneration

The role of microglial cells in the development and progression of Alzheimer’s disease (AD) has not been elucidated. Here, we demonstrated the existence of a weak microglial response in human AD hippocampus which is in contrast to the massive microglial activation observed in APP-based models. Most importantly, microglial cells displayed a prominent degenerative profile (dentate gyrus > CA3 > CA1 > parahippocampal gyrus), including fragmented and dystrophic processes with spheroids, a reduced numerical density, and a significant decrease in the area of surveillance (“microglial domain”). Consequently, there was a substantial decline in the area covered by microglia which may compromise immune protection and, therefore, neuronal survival. In vitro experiments demonstrated that soluble fractions (extracellular/cytosolic) from AD hippocampi were toxic for microglial cells. This toxicity was abolished by AT8 and/or AT100 immunodepletion, validating that soluble phospho-tau was the toxic agent. These results were reproduced using soluble fractions from phospho-tau-positive Thy-tau22 hippocampi. Cultured microglial cells were not viable following phagocytosis of SH-SY5Y cells expressing soluble intracellular phospho-tau. Because the phagocytic capacity of microglial cells is highly induced by apoptotic signals in the affected neurons, we postulate that accumulation of intraneuronal soluble phospho-tau might trigger microglial degeneration in the AD hippocampus. This microglial vulnerability in AD pathology provides new insights into the immunological mechanisms underlying the disease progression and highlights the need to improve or develop new animal models, as the current models do not mimic the microglial pathology observed in the hippocampus of AD patients. Electronic supplementary material The online version of this article (doi:10.1007/s00401-016-1630-5) contains supplementary material, which is available to authorized users.


Supplemental Figure 1. Microglial activation and pathological characterization of Braak II, III-IV and V-VI hippocampus.
a) The expression of microglial markers Pu.1 (a1) and c/EBPa (a2) was determined by qPCR. The expression of these two markers was increased significantly (Kruskal-Wallis and Dunn post-hoc test, see figure) only in Braak V-VI cases, as compared with Braak 0 or Braak II samples. All affected microglial markers showed significant linear correlations using tau-Kendall (a3-6), corroborating the existence of microglial activation and the implication of a single cell population. b-c) Neuropathological characterization of phospho-tau and Abeta in human hippocampal samples using immunohistochemistry with AT8 (b1-6) or 4G8 (c1-6) antibodies. Phospho-tau accumulation was clearly observed in Braak III-IV samples, mainly in CA1 field (b5) and in both CA1 (b6) and dentate gyrus (b3) in Braak V-VI stages. The Abeta plaques affected the CA1 region and molecular layer of dentate gyrus in Braak III-IV cases (c2 and c5), however the hilar region was only affected in Braak V-VI individuals (c3 and c6). Alv, alveus; g, granular layer; h, hilus; m, molecular layer; so, stratum oriens; slm, stratum lacunosum-moleculare; sp, stratum pyramidale; sr, stratum radiatum. Scale bars: b1-b6 and c1-c6, 1 mm.

Supplemental Figure 2. The expression of P2ry12 was reduced in activated microglial cells.
a) The co-localization (a3 and a6) between Iba1 (a1 and a4) and P2ry12 (a2 and a5) microglial specific markers was assessed by laser confocal microscopy in Braak V-VI samples. Co-localization (a3 and a6, arrows) between both markers was clearly observed in microglial cells not in direct contact with Abeta plaques (visualized in blue colour). The boxed regions, at higher magnification in insets, showed activated Iba1 positive and P2ry12 negative microglial cells, closely associated with Abeta plaques (in blue). Furthermore, ex vivo experiments, using isolated adult mice microglia, indicated that the P2ry12 expression was slightly reduced after LPS activation (1.05 ± 0.40 vs 0.65 ± 0.15 relative units for PBS or LPS respectively, n=4, two-tailed t-test, t=2.08, p=0.075). Thus, the expression of P2ry12 receptor was down-regulated after microglial activation g, granular layer. Scale bars: a1-a6, 50 µm (insets in a3 and a6, 20 µm) Supplemental Fig. 3 Microglial response in Abeta (APP/PS1) and tau (Thy-tau22) transgenic models. a) APP/PS1 model displayed a strong age-dependent accumulation of Abetaplaques (a1; 12-month-old mice) and total Abeta (a2, representative westernblot using 6E10+82E1) in the hippocampal formation. This Abeta accumulation was associated with a clear microglial activation. Representative images showing Iba1 (a3, panoramic image; a4, high magnification) or CD45 (a5, panoramic image; a6, high magnification) immunostained microglial cells from 12-month-old mice. The microglial activation was also assessed by determining (qPCR) the expression (a7) of microglial markers (ANOVA and Tukey post hoc test; * p<0.05). Data (n=5 animals per age) were normalized using 9-month-old WT mice (n=5; not shown) and are mean ± S.D. Quantitative Iba1-load determinations at the hilar region (a8), from 6-or 12-month-old mice, also demonstrated the existence of a significant (ANOVA and Tukey post hoc test; * p<0.05; n=5/age and genotype) increase in the area covered by Iba-1 positive cells, as compared with age-matched WT mice. b) Thy-tau22 transgenic model presented an age-dependent phospho-tau accumulation principally in the CA1 region, determined by AT8 immunohistochemistry (b1) or western-blots (b2).
This phospho-tau accumulation produced an attenuated microglial response as probed by Iba1 immunostaining, from 12-16-month-old WT (b2) and Thy-tau22 (b3) mice. Microglial activation was also assessed by qPCR (b5). Only the expression of CD45, CD68 and TREM2 was slightly but significantly increased (ANOVA and Tukey post hoc test; *p<0.05). Data (n=5 animals per age) were normalized using 9-month-old WT mice (n=5; not shown) and are mean ± S.D. a) Hippocampal samples were fractionated into: S1, soluble fraction containing extracellular/cytosolic proteins; S2, intracellular vesiculated fraction; S3, aggregated proteins, releasable by SDS treatment and P3, highly-aggregated proteins, only releasable by SDS-Urea. The presence of phospho-tau species was analyzed by western blots using AT100 (a1) or AT8 (not shown). As expected, Braak III-IV and Braak V-VI samples showed a prominent phosphotau accumulation (AT100 positive tau in the SDS releasable pool (S3). This is consistent with the presence of aggregated and phosphorylated tau protein, forming intracellular tangles. We also observed a clear accumulation of AT100 positive tau protein in the S1 soluble fraction (extracellular and/or cytosolic) from Braak V-VI samples. Total tau in the soluble S1 fractions was assessed by sandwich ELISA (a2). b) Soluble Abeta present in S1 fractions was quantitatively analyzed by ELISA (b1) or dot-blots (b2-3) using the OC antibody. and 10 μl of sarkosyl-insluble tau/ 100 μl culture media; equivalent to ten to a hundred times higher dose than S1), we did not detect any toxic effect on BV2 cells (a3). Data are mean ± SD of three independent experiments. b) Toxicity (using flow cytometry, b1) of S1 fractions isolated from Braak 0, Braak II, Braak III-IV and Braak V-VI samples was also assessed using the astrocytic cell line WJE. As shown (b2), no toxicity was observed in any case. Data are mean ± S.D. of three independent experiments.
a2) Quantitative analysis of BV2 cell toxicity after treatment with 6-or 18-monthold APP/PS1-derived S1 fractions showing percentage of viable, early apoptotic cell (Annexin V positive) or late apoptotic cells (Annexin V and PI positive cells).
Data are mean ± S.D. of three different experiments. b) S1 fractions isolated from Thy1-tau22 models. The phospho-tau (b1, upper panel) or total tau content (b1, middle panel) in these fractions was assessed by western blots using AT8 or Tau46 antibodies. As shown, there was an age-dependent increase in the soluble AT8-positive phosho-tau in the S1 fractions. (b2) Quantitative analysis of BV2 cell viability after treatment with 2-, 9-or 12-monthold Thy1-tau22-derived S1 fractions. Percentage of viable, early apoptotic cell