Unique pathological tau conformers from Alzheimer's brains transmit tau pathology in nontransgenic mice

Abstract

Filamentous tau aggregates are hallmark lesions in numerous neurodegenerative diseases, including Alzheimer's disease (AD). Cell culture and animal studies showed that tau fibrils can undergo cell-to-cell transmission and seed aggregation of soluble tau, but this phenomenon was only robustly demonstrated in models overexpressing tau. In this study, we found that intracerebral inoculation of tau fibrils purified from AD brains (AD-tau), but not synthetic tau fibrils, resulted in the formation of abundant tau inclusions in anatomically connected brain regions in nontransgenic mice. Recombinant human tau seeded by AD-tau revealed unique conformational features that are distinct from synthetic tau fibrils, which could underlie the differential potency in seeding physiological levels of tau to aggregate. Therefore, our study establishes a mouse model of sporadic tauopathies and points to important differences between tau fibrils that are generated artificially and authentic ones that develop in AD brains.

Figure 1.

Preparation of different variants of tau fibrils. (A) Sedimentation test for passages 1–6 of repetitively self-seeded T40 fibrillization without heparin. Supernatant (S) and pellet (P) fractions were resolved on SDS-PAGE and stained by Coomassie blue. Data are representative of more than three independent fibrillization series. (B) Sedimentation test for T40 fibrils induced with and without heparin; three different preparations are shown for each category. The heparin-free fibrils were from passages 9 or 10 of repetitively self-seeded fibrillization as shown in A. (C) Negative staining EM images for T40 fibrils induced with and without heparin after sonication. Bar, 100 nm. (D) A schematic diagram summarizing the main steps of tau PHF purification from AD brains. (E and F) Different fractions from PHF purification (refer to the schematic in D) were immunoblotted with 17025 (a polyclonal pan-tau antibody) and PHF-1 (a monoclonal antibody specific for tau phosphorylated at S396/S404). (F) Ponceau S staining for the final purification steps revealed further removal of contaminants from the sarkosyl pellet. The final supernatant (fraction 3, red in D) is the fraction used in the study and referred to as AD-tau. Data are representative of at least eight independent extractions (Table S2). (G) Silver staining of AD-tau shows prominent bands between 50 and 75 kD, recognized by a panel of tau antibodies, including T14 (a monoclonal antibody specific for human tau), 17025, PHF-1, RD3 (a 3R tau-specific monoclonal antibody), and an anti-4R tau polyclonal antibody. Representative preparations from the frontal cortices of three AD cases are shown (1a and 1b are two different preparations from the same case). (H) Negative staining EM images for AD-tau with (+) and without (−) sonication. Bar, 100 nm. (A, B, and E–G) Molecular mass is indicated in kilodaltons.

Figure 2.

Different variants of tau fibrils differentially seed tau pathology in non-Tg neurons. (A) Induction of endogenous mouse tau pathology in non-Tg neurons treated with the different variants of tau fibrils (amount of tau per coverslip: 4.5 µg for Hep-T40 and X-T40; 1.5 µg for AD-tau). Neurons were fixed with methanol to remove soluble tau (Fig. 2 C) and immunostained with T49, a mouse tau–specific monoclonal antibody (green). Data are representative of more than three independent experiments. Bar, 100 µm. (B) Quantification of the area occupied by mouse tau pathology normalized to total cell count, shown as mean + SEM. For each fibril variant, three different preparations were tested across three independent sets of neurons; AD-tau preparations from three different cases were tested. (C) Immunostaining of mouse tau (mTau; T49; green) and MAP2 (polyclonal antibody 17028; red) in DPBS-treated non-Tg neurons fixed with 4% PFA or with cold methanol. In these control neurons, MAP2 immunoreactivity, which is in the neuronal cell bodies and dendrites, remains intact with methanol extraction, but the axonally located mouse tau is largely removed by methanol fixing. The results are verified in two independent experiments. Bar, 50 µm. (D) The thread-like neuritic tau aggregates induced by both X-T40 and AD-tau fibrils rarely colocalized with MAP2 staining, suggesting their axonal location. X-T40 but not AD-tau fibrils induced tau aggregation in a subset of neuronal cell bodies (shown by asterisks in the top panels). Bar, 50 µm. (E and F) Mouse tau pathology induced by AD-tau preparations after a mock immunodepletion using control mouse IgG (E) or after immunodepletion of tau using Tau 5 (F). The volume of unbound fraction added per coverslip contained 0.2 µg of AD-tau before immunodepletion. Seeding activity of AD-tau was abolished by Tau 5 but not by the control mouse IgG. (G and H) Mouse tau pathology induced by 4.5 µg Hep-T40 fibrils (G) or 1.5 µg X-T40 fibrils (H) that had been mixed with tau-immunodepleted AD-tau preparations. (E–H) Images are representative of three AD cases tested. Bar, 100 µm. (I) RIPA-extracted lysates from non-Tg neurons that were 6, 10, 15, and 20 d in vitro (DIV) were probed for 3R and 4R tau expression using isoform-specific monoclonal antibodies RD3 and RD4, respectively. Total tau was shown by K9JA (a polyclonal antibody recognizing residues 243–441 of tau) or R2295 h&mTau (a fraction of polyclonal serum recognizing both human and mouse tau). Data are representative of three independent sets of neurons tested. (J) Dephosphorylation of neuronal lysates showed that the two predominant isoforms of mouse tau expressed in culture are the shortest ones (i.e., 3R0N and 4R0N) when aligned to the six isoforms of human tau (the lane marked by an asterisk). The blot was probed with R2295 h&mTau. Because a mouse tau sequence is shorter than a human tau, 4R0N mouse tau is aligned with 3R0N human tau, and 3R0N mouse tau runs even lower. Three dephosphorylation experiments were performed on two independent sets of neuronal lysates. (K) Supernatant (S) and pellet (P) fractions from 1% sarkosyl extraction of treated non-Tg neurons were immunoblotted with mouse tau (mTau)–specific antibodies (R2295 and T49), with two preparations of fibrils tested for each variant (9 µg Hep-T40, 3 µg X-T40, and 3 µg AD-tau for each well on a 12-well plate). Black and red arrows indicate 4R and 3R mouse tau, respectively. The asterisk indicates a nonspecific band detected by R2295. Data are representative of three independent sets of non-Tg neurons tested. (I–K) Molecular mass is indicated in kilodaltons.

Figure 3.

AD-tau is a more potent seed for tau aggregation in non-Tg mice. (A) Schematics showing injection sites in the dorsal hippocampus and overlying cortex indicated by red dots (Bregma −2.54 mm, 2 mm from midline, and −1.4 mm from skull [for the cortex] and −2.4 mm from skull [for the hippocampus]). (B) Differential induction of tau pathology recognized by AT8 (a monoclonal antibody specific for tau phosphorylated at S202/T205) in non-Tg mice at 3 mo p.i. of different tau fibrils. Amount of tau injected per mouse: 9 µg for Hep-T40 (four mice) and X-T40 fibrils (six mice); 8 µg for AD-tau (four mice). Contra, contralateral; Ctx, cortex; HP, hippocampus; Ipsi, ipsilateral. Bar, 100 µm. (C) AT8 immunostaining at 24 mo p.i. of Hep-T40 fibrils (9 µg/mouse; two mice). (D) AT8 immunostaining at 9 mo p.i. of X-T40 fibrils (9 µg/mouse; three mice). Considerable variability was found among injected animals, with the one mouse developing the most abundant pathology shown here. (C and D) Bars, 200 µm. (E) AT8 immunostaining at 3, 6, and 9 mo p.i. of AD-tau (8 µg/mouse; four mice per time point). Bar, 100 µm. (F) Quantification of percent area occupied by AT8-positive tau pathology developed in the ventral hippocampal hilus with 8 µg AD-tau injection (four mice per time point; each dot represents one mouse). One-way ANOVA was performed across different time points for ipsilateral and contralateral pathology separately. Significant differences were found for ipsilateral pathology, and Tukey’s multiple comparison test identified a significant difference between 3 and 9 mo p.i. *, P < 0.05. mpi, months p.i. (G) AD-tau inoculation (8 µg/mouse) did not lead to significant neuron loss in the ventral hippocampal hilus region as compared with mice injected with control (Ctrl) brain extracts. Data are shown as mean + SEM. One-way ANOVA found no significant differences among the groups. Data shown in this figure were obtained from injections performed all on 2–3-mo-old non-Tg mice (C57BL6 or C57BL6/C3H F1). Refer to Table S4 for the details of animals analyzed per group.

Figure 4.

Propagation of tau pathology after inoculation of a lower dose of AD-tau into young and aged non-Tg mice. (A) AT8 immunostaining for various brain regions after a lower dose of AD-tau inoculation (2 µg/mouse) into the dorsal hippocampus (HP) and overlying cortex of 2–3-mo-old non-Tg mice (C57BL6). Only weak AT8 immunoreactivities were detected in the ipsilateral (Ipsi) ventral hippocampus at 1 mo p.i. (three mice). More prominent AT8-positive tau inclusions appeared in both the ipsilateral and contralateral (Contra) ventral hippocampus at 3 mo p.i. (three mice), which showed a trend of decline between 6 and 9 mo p.i. (four and three mice, respectively). Bar, 100 µm. (B) Quantification of percent area occupied by AT8-positive tau pathology developed in the ventral hippocampal hilus with 2-µg AD-tau injection (three to four mice per time point; each dot represents one mouse). (C) Quantification of NeuN⁺ cells in the ventral hippocampal hilus region with 2-µg AD-tau injection. Data are shown as mean + SEM. (B and C) One-way ANOVA showed no significant differences among groups. (D) Comparisons of AT8-positive tau pathology developed in the ipsilateral entorhinal cortex, ipsilateral fimbria, and corpus callosum at 6 mo p.i. of AD-tau into young (2–3 mo) versus aged (15–19 mo) non-Tg mice (2 µg/mouse; four young and three aged mice at 6 mo p.i. with each dot representing one mouse; all C57BL6). Bar, 100 µm. (E–G) Quantification of percent area occupied by AT8 immunoreactivities in regions shown in D followed by Student’s t tests indicated that aged mice developed significantly more pathology in the ipsilateral entorhinal cortex (*, P < 0.05) and a trend of more abundant pathology in the ipsilateral fimbria and corpus callosum. Data represent three to four mice per time point; each dot represents one mouse.

Figure 5.

Pathology developed after AD-tau inoculation is composed of endogenous mouse tau and cannot be induced by control brain extracts. (A and B) IHC of the injection site with AT8- and human tau–specific antibody HT7 at 2 and 7 d p.i. (2 µg/mouse; two mice per time point). (C and D) AT8-positive neuropil staining, presumably caused by endogenous mouse tau accumulations, was observed near the injection site at 3 and 6 mo p.i. (2 µg/mouse for C; 8 µg/mouse for D; four mice per condition except for three mice at 3 mo p.i. of 2 µg/mouse). (E) IHC with mouse tau–specific R2295 confirms recruitment of endogenous mouse tau into aggregates formed in the ipsilateral ventral hippocampus (Ventral HP Ipsi) and raphe nucleus at 3 mo p.i. (8 µg/mouse; four mice). (F) Lack of AT8 immunoreactivity throughout the brains after injections of extracts from control brains prepared the same way as AD-tau. Bars: (A–D) 200 µm; (E and F) 100 µm. Data shown in this figure were obtained from injections performed all on 2–3-mo-old C57BL6 mice.

Figure 6.

Spreading and maturation of tau pathology induced by AD-tau inoculation in non-Tg mice. (A) Heat maps showing semiquantitative analyses of tau pathology based on AT8 immunostaining (0: no pathology, gray; 3: maximum pathology, red) at 3 and 6 mo p.i. of AD-tau (8 µg/mouse; four mice per time point). Six coronal planes are shown from left to right: bregma 0.98 mm, −2.18 mm, −2.92 mm, −3.52 mm, −4.96, and −5.52 mm. The distribution of tau pathology at 9 mo p.i. is very similar to that at 6 mo p.i. (not depicted). (B) A schematic showing various brain regions connected to the dorsal hippocampus, the primary site receiving inoculums (Fig. 5, A and B), with both afferents and efferents indicated by directional arrows. Brain regions with tau aggregate formation after AD-tau injections are marked in red. The diagram was created based on data from the Allen Brain Atlas C57BL/6 mouse connectivity studies (Oh et al., 2014). (C) Brain sections were immunostained with antibodies AT180 (recognizing tau phosphorylated at T231), MC1 (recognizing a pathological conformation of tau found in the AD brains), and TG3 (recognizing conformationally altered tau phosphorylated at T231), which all demonstrated increased immunoreactivities over time after AD-tau inoculation (8 µg/mouse; four mice per time point). Double-labeling immunofluorescence using AT8 (red) and ThS (green) shows a subset of pretangles matured into β sheet–rich tangle-like aggregates starting 6 mo p.i. (D) AT8 and ThS double labeling of the ipsilateral ventral hippocampus at 9 mo p.i. of AD-tau (8 µg/mouse; four mice). Bars, 100 µm. Data shown in this figure were obtained from injections performed all on 2–3-mo-old C57BL6 mice.

Figure 7.

rTau fibrils templated by AD-tau reveal structural relationship between 4R and 3R tau. (A) Three possible models of 4R/3R tau arrangement in AD PHFs (left) and proposed patterns of recombinant T40/T39 (rT40/rT39) recruitment by AD-tau based on models 2 and 3 (right). (B) Sedimentation test for nonseeded T40 and T39 after 3-d incubation at 37°C with constant agitation at 1,000 rpm. Cofibrillization was conducted at the same total protein concentration as single-isoform fibrillization with T40 and T39 mixed at a 1:1 ratio. S, supernatant; P, pellet. (C) Sedimentation test for singly or cofibrillized T40 and T39 seeded by 10% AD-tau ([AD]T40, [AD]T39, and [AD]T40 + T39) under the same fibrillization condition as in B. Monomers were used as the controls. rTau remaining in the soluble fraction appeared to be truncated. (D) Sedimentation test for singly or cofibrillized Myc-T40 and HA-T39 seeded by 10% AD-tau ([AD]Myc-T40, [AD]HA-T39, and [AD]Myc-T40 + HA-T39) under the same fibrillization condition as in B, with monomers as the controls. Supernatant and pellet fractions were immunoblotted with anti-Myc and anti-HA antibodies. These two antibodies do not label truncated tau in the supernatant as seen in C, suggesting the C terminus of tau was cleaved together with the tag. (B–D) Data shown are representative of at least two independent experiments. (E) Quantification for the percentage of starting tau monomers recruited into the pellet fraction with AD-tau seeding as shown in C and D. Data are displayed as mean + SEM and are based on three independent experiments. (F) Double-labeling immuno-EM images for [AD]Myc-T40 and [AD]HA-T39 fibrils, stained with PHF-1 and anti-Myc or anti-HA antibody. Secondary antibodies are conjugated to colloidal gold particles of two different sizes (6 nm or 12 nm). Arrowheads point to examples of Myc or HA immunoreactivity (12 nm), and arrows point to examples of PHF-1 immunoreactivity (6 nm). (G) Immuno-EM images of [AD]Myc-T40 + HA-T39 fibrils stained with a single primary antibody (anti-Myc alone at the top and anti-HA alone at the bottom) but with both secondary antibodies showed labeling by single-size gold particles, suggesting lack of nonspecific binding by secondary antibodies. (H and I) Double-labeling immuno-EM images for [AD]Myc-T40 + HA-T39 fibrils stained with both anti-Myc and anti-HA antibodies showed frequent intermingling of Myc (12 nm) and HA (6 nm) immunoreactivities on the same filaments (H) and occasional filaments with only Myc or only HA immunoreactivities (I). Bars, 100 nm. (F–I) Images are representative of at least two independent experiments. (B–D) Molecular mass is indicated in kilodaltons.

Figure 8.

rTau fibrils templated by AD-tau acquire seeding competency in non-Tg neurons. (A) Mouse tau aggregates (methanol-resistant T49 staining in green) induced in non-Tg neurons by [AD]rTau fibrils compared with the same dose of AD-tau (100% AD-tau) and 10% AD-tau seed control (amount of tau per coverslip: 0.15 µg for 10% AD-tau and 1.5 µg for the rest). (B) Quantification of the area occupied by mouse tau pathology normalized to total cell count for the experiment shown in A, displayed as mean + SEM. Four different sets of [AD]rTau fibrils that were seeded by three different preparations of AD-tau were tested; two independent sets of non-Tg neurons were used. Pairwise Student’s t tests were performed between seeded rTau fibrils and 10% seed control. *, P < 0.05; ***, P < 0.0005; ****, P < 0.00005. (C) Immunoblotting for sarkosyl-soluble (sup) and -insoluble (pel) fractions of non-Tg neurons treated with [AD]rTau fibrils or 10% AD-tau seed control (9 µg [AD]rTau and 0.9 µg AD-tau for each well on a 12-well plate). Black and red arrows indicate 4R and 3R mouse tau, respectively. Data are representative of two independent experiments. mTau, mouse Tau. Molecular mass is indicated in kilodaltons. (D) AT8-positive mouse tau pathology induced by different isoforms of tau seeded by 10% X-T40 fibrils. (E) Quantification of tau pathology for the experiment shown in D. The same preparation of X-T40 fibrils was used to seed different batches of independently purified monomers: four batches of monomers for T40 and two batches of monomers for T34, T39, and T37. Data are shown as mean + SEM. Pairwise Student’s t tests were conducted between seeded T40 and the other isoforms. **, P < 0.005. Bars, 100 µm.

Figure 9.

Amplification of AD-tau using phosphorylated T40 and all six isoforms of tau. (A) In vitro phosphorylation of T40 mediated by SAPK4. Lane 1: nontreated T40; lanes 2 and 4: mock-treated T40 in the same buffer condition as the phosphorylation reaction; lanes 3 and 5: T40 phosphorylated by incubation with SAPK4. Concentrations of T40 in the reactions were 12 µM for lanes 2 and 3 and 45 µM for lanes 4 and 5. The differently treated T40 was immunoblotted with 17025 and PHF-1 or AT8. (B) Sedimentation test for mock-treated T40 (−) and phosphorylated T40 (+) seeded by 10% AD-tau. (C) Sedimentation test for mixed six isoforms of tau with (+) and without (−) seeding by 10% AD-tau. (B and C) Reactions were incubated at 37°C for 3 d with constant agitation at 1,000 rpm. P, pellet; S, supernatant. Samples were resolved on SDS-PAGE and stained with Coomassie blue. (A–C) Data are representative of two independent experiments. Molecular mass is indicated in kilodaltons. (D and F) Methanol-resistant mouse tau pathology (T49 staining in green) induced by mock-treated T40 (mock-T40), phosphorylated T40 (p-T40), nontreated T40 (T40 alone), and six isoforms of tau (6hTau), all seeded by 10% AD-tau (1.5 µg tau added per coverslip). Bars, 100 µm. (E and G) Quantification of mouse tau pathology for experiments shown in D and F. Pathology was normalized to that induced by seeded mock-T40 (E) or seeded T40 alone (G) in each set of experiments. Data from two independent experiments are shown as mean + SEM. Student’s t tests showed no significant differences between the two groups for both E and G.

Figure 10.

The differential seeding activities of distinct tau fibril variants are underlain by their conformational differences and influenced by the presence of heparin. (A) The different variants of tau fibrils (Hep-T40, X-T40, and [AD]T40) and control samples (equal amount of T40 monomer; 10% of AD-tau) were incubated with increasing concentrations of trypsin (0%, 0.00125%, 0.0025%, 0.005%, and 0.01%) for 30 min at 37°C. The resulted digestion products were resolved on SDS-PAGE and stained by Coomassie blue. The asterisk indicates dominant trypsin-resistant fragments for Hep-T40 and [AD]T40 fibrils. The 10% AD-tau seeds in [AD]T40 did not contribute to the 18-kD fragment, as suggested by the 10% AD-tau control. For each fibril variant, three to four independently prepared samples were tested across three independent experiments. (B) Digestion products from incubation with 0.01% trypsin were resolved on SDS-PAGE and immunoblotted with tau antibodies recognizing different epitopes. R2: second MT-binding repeat. Data are representative of three independent experiments. (C) The different variants of tau fibrils (three preparations of Hep-T40 fibrils, three preparations of X-T40 fibrils, and two preparations of [AD]T40 fibrils) were incubated with 1 µg/ml of proteinase K for 30 min at 37°C. The resulted products were resolved on SDS-PAGE and stained by Coomassie blue. (A–C) Molecular mass is indicated in kilodaltons. (D) Methanol-resistant mouse tau pathology induced by 1.5 µg X-T40 or 0.15 µg AD-tau fibrils with and without 50 nM heparin cotreatment. The amount of heparin added is equivalent to that contained in 10 µM heparin-induced 40 µM T40 fibrils added at 4.5 µg of tau per coverslip. Bar, 100 µm. (E) Quantification of mouse tau pathology for experiments shown in D. For each fibril variant, three different preparations of fibrils were tested across three independent sets of non-Tg neurons. Data are shown as mean + SEM. Pairwise Student’s t tests were performed between untreated and heparin-treated cells. *, P < 0.05; ***, P < 0.001.