Fig 2: PQBP1 interacts with extrinsic Tau in microglia.a Schematic presentation of deletion mutants of tau and PQBP1 used for SPR and NMR analyses. N1/2 N-terminal domain, R1/2/3/4 C-terminal microtubule-binding domain, PRR proline-rich region, WWD WW domain, PRD polar amino acid-rich domain, IDP/IDR intrinsically disordered region. b Equilibrium binding curves of PQBP1 to Tau 410, Tau 441, Tau 410 P179A, Tau 410 P216A, and Tau 410 P179A/P216A monitored by SPR. Tau proteins were immobilized on sensor chip surfaces. The binding response at equilibrium was plotted against the concentration of PQBP1. Black, full-length PQBP1; red, PQBP1(1–94); blue, PQBP1 (94–176); green, PQBP1 (193–265). RU resonance unit. c 1H-15N HSQC spectra of Tau 1–150, 151–197, 198–243, 244–341, and 342–410 in the absence (black) and presence (red) of PQBP1(1–94). Amino acid residue numbers of tau are based on Tau 410. The molar ratio of Tau:PQBP1 (1–94) is 1:5. Residues showing significant chemical shift perturbation and line broadening of resonances are labeled in the HSQC spectra of Tau 151–197 and 198–243. Ctrl control, ppm parts per million. d Tau 410 amino acid sequence showing consensus motifs for interaction with PQBP1-WWD (red letter) and mutated prolines (underlined). e A representative western blot showing co-precipitation of tau with PQBP1 from a lysate of C57BL/6 microglia cultured for 3 days with 25 nM non-aggregated Tau-TAMRA (upper panels). Quantitative immunoprecipitation was performed with C57BL/6 microglia cultured for 3 days with 25 nM non-aggregated Tau-TAMRA. Input was the intensity of TAMRA-labeled Tau 410 in the C57BL/6 microglia lysate samples used for immunoprecipitation. Output was the intensity of TAMRA-labeled Tau 410 in samples after immunoprecipitation with anti-tau, anti-PQBP1, and human IgG. The fluorescence intensity of TAMRA-labeled Tau 410 was measured on a FLUOstar OPTIMA-6 microplate reader. n.s. not significant, AU arbitrary unit. N = 4 per group. P = 0.9963 (ctrl-IgG vs PQBP1), 0.8874 (ctrl-IgG vs tau), 0.9215 (PQBP1 vs tau), 9.63e-6 (PQBP1 vs ctrl-IgG), 3.69e-7 (tau vs ctrl-IgG). ##P < 0.01, n.s.P > 0.05 in Tukey’s HSD test. Box plots show the median, quartiles, and whiskers that represent data outside the 25th to 75th percentile range.

Fig 3: PQBP1 is essential for in vivo activation of microglia by extrinsic Tau.a Protocol of Tau 410 monomer (2.2 µg/µl, 1 µl/injection, 4.4 µg/mouse) or equivalent PBS injection to bilateral entorhinal cortices of non-induced or tamoxifen-induced Pqbp1-cKO mice. Four groups were made for analyses. i.p. intra-peritoneal injection, qPCR quantitative RT-PCR, IHC immunohistochemistry. b Immunohistochemistry of Iba1-positive microglia with PQBP1-Tau colocalization, cGAS-tau colocalization, and NF?B nuclear translocation in injected areas of non-induced Pqbp1-cKO and tamoxifen-induced Pqbp1-cKO. (Right) Quantitative analyses of PQBP1-Tau colocalized microglia, cGAS-tau colocalized microglia, and activated microglia (nuclear NF?B-positive) among Tau-incorporating microglia. N = 5. P value: [Upper] 3.55e-9 (non-induced), 5.71e-9 (Tau410). [Middle] 1.00e-7 (non-induced), 1.00e-7 (Tau410). [Lower] 2.29e-14 (non-induced), 6.98e-12 (Tau410). c RT-qPCR of induction of Tnf and Isg54 by Tau 410 in entorhinal cortex from non-induced or tamoxifen-induced Pqbp1-cKO mice. N = (4, 6, 3, 6). P value: [Upper] 0.0011 (non-induced), 0.0005 (Tau410). [Lower] 0.0031 (non-induced), 0.0080 (Tau410). d Y-maze test at 7, 10, and 14 days and Morris water maze test at 14 days after tau injection into bilateral entorhinal. Data points (Morris water maze test): mean ± SEM. Y-maze: N = (8, 8, 7, 8) for 7 days, (7, 8, 7, 8) for 10 days and (8, 8, 7, 8) for 14 days. Morris water maze: N = (7, 8, 6, 7). P value: [Y-maze; 10 days] 0.0439 (non-induced), and 0.0055 (Tau410). [14 days] 0.0014 (non-induced), and 0.0018 (Tau410). [Morris water maze, Day4] 0.0043 (non-induced), and 0.0048 (Tau410). e In vivo live imaging of microglia by two-photon microscopy with B6.129P2(Cg)-Cx3cr1tm1Litt/J (Cx3cr-1GFP) mice in which microglia were labeled by the endogenous Cx3cr1 enhancer/promoter-driven EGFP protein. f Immunohistochemistry confirmed PQBP1-KD by siRNA-PQBP1 injection. (Graph) Quantification of PQBP1 signal intensity/cell. Data points: mean ± SEM. n = 100 microglia. P = 1.71e-102, **P < 0.01 in two-sided Student’s t-test. Box plots show the median, quartiles, and whiskers that represent data outside the 25th to 75th percentile range. Tuple of N numbers indicates PBS/non-induced, Tau410/non-induced, PBS/induced Pqbp1-cKO, and Tau410/induced Pqbp1-cKO), respectively. For multiple group comparisons, Tukey’s HSD test was performed (##P < 0.01, #P < 0.05).

Fig 4: Direct and indirect pathways mediating neuronal toxicity of Tau.a Immunohistochemistry of Tau-injected mice revealed most of the non-microglia cells incorporating tau were MAP2-positive neurons but not GFAP-positive astrocytes. The ratios of Tau-incorporating cells were equivalent in non-induced and tamoxifen-induced Pqbp1-cKO mice, as the basis for analyses in (b). N = 20 images from three mice. Statistical test: Student’s t-test. b Frequencies of neuronal cell death (TUNEL-positive and NeuN-positive dying neurons) by addition of normal or PQBP1-binding mutant Tau 410 were changed in Tau-negative neurons in addition to that of Tau-positive neurons. Indirect effect of Tau 410 monomer on Tau-negative neurons was suppressed by PQBP1-binding mutation or by tamoxifen-induced Pqbp1-cKO, while the direct effect of Tau 410 polymers on Tau-positive neurons was not affected by PQBP1-binding mutation or by tamoxifen-induced Pqbp1-cKO. TUNEL terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling. Values in each group are presented as mean ± SEM. N = 4 fields. #P < 0.05, ##P < 0.01 in Tukey’s HSD test. c Direct effect of monomer and polymerized Tau 410 on primary neurons. Non-polymerized Tau 410 was not toxic to neurons in the absence of microglia. N = 5 visual fields. n.s.P > 0.05 in Tukey’s HSD test. d Direct and indirect pathways from extracellular tau to neuronal damage. Tau polymer directly damages neurons, while tau monomer is incorporated into LPL-negative non-DAM microglia (Supplementary Fig. 8), and indirectly damages neurons via cytokines. Though not included in this study, DAM around Aß plaques might play a different role to prevent tau seeding. Box plots show the median, quartiles, and whiskers that represent data outside the 25th to 75th percentile range.

Fig 5: Extrinsic Tau proteins induce nuclear translocation of NF?B and expression of pro-inflammatory genes in microglia.a Immunocytochemistry of primary microglia prepared from newborn C57BL/6 mice reveals nuclear translocation of NF?B at 48 h after addition of recombinant full-length Tau 410, Tau 441, Tau 410 P179A, Tau 410 P216A, or Tau 410 P179A/P216A protein in the non-polymerized or polymerized state to culture medium. Images were obtained by confocal microscopy. Red or white arrows indicate microglia with or without nuclear translocation of NF?B. Lower panels show high magnification of representative cells. In the presence of tau proteins in culture medium, nuclear translocation of NF?B was induced in most microglia, while in the absence of tau proteins NF?B remained in the cytoplasm. b The effects of various tau species in the polymerized and non-polymerized state on nuclear translocation of NF?B were compared. Aggregation of Tau 410/441 reduces the frequency of NF?B nuclear translocation. PQBP1-binding proline single mutants (Tau 410 P179A, Tau 410 P216A) partially, while double proline mutant (Tau 410 P179A/P216A) substantially, lost the ability to induce NF?B nuclear translocation. N = 5, n = 20 (cells). ##P < 0.01 in Tukey’s HSD test. c Isoforms, mutations, and aggregation states of tau affect induction of Tnf and Isg54 gene expression. Consistent with immunocytochemistry of NF?B, non-polymerized Tau 410 induced pro-inflammatory genes most efficiently, while Tau 441 possesses 60–80% of the induction ability. PQBP1-binding proline single mutants (Tau 410 P179A, Tau 410 P216A) partially, while double proline mutant (Tau 410 P179A/P216A) substantially, lost transactivation of Tnf and Isg54 gene expression. RT-qPCR reverse transcriptase quantitative polymerase chain reaction. N = 3. ##P < 0.01 in Tukey’s HSD test. Box plots show the median, quartiles, and whiskers that represent data outside the 25th to 75th percentile range.