Tau seeds from patients induce progressive supranuclear palsy pathology and symptoms in primates

Abstract

Progressive supranuclear palsy is a primary tauopathy affecting both neurons and glia and is responsible for both motor and cognitive symptoms. Recently, it has been suggested that progressive supranuclear palsy tauopathy may spread in the brain from cell to cell in a ‘prion-like’ manner. However, direct experimental evidence of this phenomenon, and its consequences on brain functions, is still lacking in primates.

In this study, we first derived sarkosyl-insoluble tau fractions from post-mortem brains of patients with progressive supranuclear palsy. We also isolated the same fraction from age-matched control brains. Compared to control extracts, the in vitro characterization of progressive supranuclear palsy-tau fractions demonstrated a high seeding activity in P301S-tau expressing cells, displaying after incubation abnormally phosphorylated (AT8- and AT100-positivity), misfolded, filamentous (pentameric formyl thiophene acetic acid positive) and sarkosyl-insoluble tau. We bilaterally injected two male rhesus macaques in the supranigral area with this fraction of progressive supranuclear palsy-tau proteopathic seeds, and two other macaques with the control fraction. The quantitative analysis of kinematic features revealed that progressive supranuclear palsy-tau injected macaques exhibited symptoms suggestive of parkinsonism as early as 6 months after injection, remaining present until euthanasia at 18 months. An object retrieval task showed the progressive appearance of a significant dysexecutive syndrome in progressive supranuclear palsy-tau injected macaques compared to controls. We found AT8-positive staining and 4R-tau inclusions only in progressive supranuclear palsy-tau injected macaques. Characteristic pathological hallmarks of progressive supranuclear palsy, including globose and neurofibrillary tangles, tufted astrocytes and coiled bodies, were found close to the injection sites but also in connected brain regions that are known to be affected in progressive supranuclear palsy (striatum, pallidum, thalamus). Interestingly, while glial AT8-positive lesions were the most frequent near the injection site, we found mainly neuronal inclusions in the remote brain area, consistent with a neuronal transsynaptic spreading of the disease.

Our results demonstrate that progressive supranuclear palsy patient-derived tau aggregates can induce motor and behavioural impairments in non-human primates related to the prion-like seeding and spreading of typical pathological progressive supranuclear palsy lesions. This pilot study paves the way for supporting progressive supranuclear palsy-tau injected macaque as a relevant animal model to accelerate drug development targeting this rare and fatal neurodegenerative disease.

Darricau et al. show that patient-derived PSP-tau aggregates induce typical neuropathological PSP lesions that can trigger motor and behavioural impairments in a primate host. The results support the use of PSP-tau inoculated macaques as an animal model to accelerate drug development.

Introduction

Progressive supranuclear palsy (PSP) is a primary tauopathy causing various motor and cognitive deficits, including ocular motor dysfunction, postural instability, parkinsonism and dysexecutive syndrome. The disease is pathologically characterized by neuronal and glial 4R-tau inclusions inducing neurofibrillary tangles, globose tangles, tufted astrocytes and coiled bodies.^1,2 These changes occur first in the brainstem and basal ganglia, followed by the frontal-parietal cortex (CTX) in late stages.³ The mechanisms underlying the initiation and spreading of tau pathology in patients with PSP remain poorly understood. However, recent experimental and human studies suggested cell-to-cell propagation in a ‘prion-like’ manner.^4,5

According to this pathophysiological model, misfolded tau assemblies (named proteopathic seeds⁶) affect nearby soluble tau proteins and template their polymerization. Pathological tau aggregates then amplify locally and spread in brain regions that are anatomically and functionally connected. This ‘prion-like’ hypothesis can explain the clinical and pathological heterogeneity of tauopathies by the initial existence of distinct ‘strains’ of tau protofibrils that cause for instance glial rather than neuronal inclusions.⁷

This hypothesis relies primarily on findings from experiments conducted in transgenic mice. To date, no experimental data support that proteopathic tau seeds from PSP brains can directly induce glial and neuronal tau inclusions in recipient species closer to humans (i.e. non-human primates). More importantly, it has never been tested in any species whether a small amount of patient-derived tau seeds is sufficient to induce a clinical phenotype mimicking PSP.

Therefore, investigating the prion-like hypothesis of PSP tauopathy in primates is highly needed for allowing (i) the investigation of a complex cell-to-cell anatomical propagation of tau within a ‘human-like’ cerebral connectivity; (ii) a close sequence analogy between human proteopathic seeds and host proteins, necessary for efficient seeding; and (iii) the fine-grained assessment of anthropomorphic cognitive and motor skills.

In this pilot study, we injected rhesus monkeys (Macaca mulatta) in the supranigral area with tau aggregates extracted post-mortem from brains of PSP patients (PSP-tau) or with the same fraction extracted from brains of age-matched individuals without PSP (CTL-tau). We prospectively evaluated cognitive and motor performance of injected macaques every 6 months. Eighteen months after injections, we euthanized the macaques to perform histological brain analyses.

Materials and methods

Tau extraction from human brains

Human brain samples were obtained from the Brain Bank GIE NeuroCEB (Pitié-Salpétrière Hospital, Paris, France), approved by the French Ministry of Higher Education and Research to collect and distribute brain samples (agreement AC-2013–1887). Consents were signed by patients themselves or their next of kind in their name, following French bioethical laws. We selected fresh-frozen mesencephalon (MES) and frontal CTX samples from two pathologically confirmed PSP cases⁸ (one 70-year-old male with pure PSP pathology, and one 71-year-old male with PSP and Braak stage 2 Alzheimer’s disease tauopathy) and frontal CTX from two control donors (one 52-year-old male without any significant brain pathology and one 82-year-old female with Braak stage 2 Alzheimer’s disease tauopathy and Thal phase 1 amyloid deposition). To obtain an equivalent and sufficient amount of tissue for all conditions, the two PSP frontal CTX samples (PSP-tau-CTX), two PSP MES samples (PSP-tau-MES) and the two control CTX samples (CTL-tau) were pooled to obtain 3 g of tissue for the extraction protocol.

Pathological sarkosyl-insoluble tau aggregates were extracted and purified according to an adaptation of the Guo et al. protocol.⁹ Briefly, tissues were homogenized in 9 volumes of ice-cold extraction buffer [10 mM Tris-HCl, pH 7.4, 0.8 M NaCl, 10% sucrose, 1 mM EDTA, 0.1 mM PMSF, 0.1% sarkosyl, 2 mM imidazole, 1 mM sodium orthovanadate, 1 mM sodium fluoride and complete ultra EDTA-free protease inhibitor cocktail (Fischer Scientific)] using FastPrep24 lysing matrix D tubes (MP Biomedical). Homogenates were then transferred to Ti45 tubes and rotor (Beckman Coulter) and centrifuged at 11 300 rpm for 10 min at 4°C. Pellets were re-extracted using the same conditions, and the supernatants from all two extractions were filtered and pooled. Additional sarkosyl was added to the pooled supernatant to reach 1% final concentration. After 1 h of stirring (100 rpm) at room temperature, samples were centrifuged at 43 800 rpm for 75 min at 4°C (Ti45 rotor). Supernatants were discarded, and 1% sarkosyl-insoluble pellets containing pathological tau aggregates were resuspended, washed in PBS, broken up with pipettes and centrifuged again in Ti70 tubes (59 300 rpm at 4°C for 30 min). For further purification, supernatants were discarded and pellets were resuspended in PBS and incubated overnight on a rocker. Pellets were broken up again, sonicated (Mini Chiller 300, 10 cycles of 30 s), transferred into TLA110 tubes (Beckman Coulter) and centrifugated for 40 min at 53 000 rpm at 4°C. Supernatants were discarded. Pellets were finally dislodged, resuspended in 100 µl of PBS, and broken up with 23-gauge needles. Samples of PSP-tau seeds (or control material from healthy donors: ‘CTL-tau’) were then sonicated, aliquoted and stored at −80°C (Fig. 1A).

Figure 1.

Extraction, purification and in vitro characterization of tau aggregates. (A) Schematic summary of the extraction/purification protocol. (B) Filter retardation assay probed with tau monoclonal antibody (HT7) to assess the final total tau concentration in CTL-tau and PSP-tau samples. (C) Quantitative results of the high-throughput seeding assay using the tau P301S-Venus cell line transfected with patient-derived tau seeds. ****P < 0.0001; **P = 0.0008; two-way ANOVA with Tukey’s multiple comparison test. (D) Representative fluorescent images of the seeding assay. Arrowheads represent the tau-venus-positive punctae/aggregates typically taken into account to measure the relative level of seeding 48 h after adding seeds or control material. Because this assay revealed a higher in vitro seeding activity of the PSP-tau-MES sample compared to PSP-tau-CTX, the PSP-tau-MES sample was selected for further in vitro characterization and future injections in macaques. (E) Immunofluorescence performed on PSP-tau seeded and control cells with AT8 antibody and pFTAA staining (probe of filamentous proteins). All scale bars = 150 µm. (F) Western blot of soluble and sarkosyl-insoluble proteins from PSP-tau seeded and controls cells using the pan-tau KJ9A and AT8 antibodies.

In vitro characterization of PSP-tau seeds

First, tau concentrations in PSP-tau-MES, PSP-tau-CTX and CTL-tau samples were measured with quantitative filter retardation assay on nitrocellulose membranes¹⁰ and HT7 antibody (Fischer Scientific). The standard curve for quantification was obtained using serial 2-fold dilutions of a recombinant tau solution at 1.5 mg/ml (human 6xHis-0N4R tau P301S, initially expressed in Escherichia coli BL21 DE3, New England Biolabs, as described previously¹¹) (Fig. 1B).

Second, the seeding ability of PSP-tau-MES, PSP-tau-CTX and CTL-tau were assessed with a high-throughput seeding assay using clonal 0N4R tau P301S-Venus human embryonic kidney 293 (HEK293) cells, as previously described.¹² Briefly, tau extracts were diluted in OptiMEM (Life Technologies), mixed with Lipofectamine 2000, left for 10 min at room temperature and then administered to the cells. After 1 h, c-DMEM was added to each well to stop the seed transduction. The cells were incubated at 37°C for 48 h after adding seeds and then fixed with ice-cold methanol for 3 min at room temperature. Nuclei were stained with Hoechst, and images were acquired at 405 and 488 nm with an InCell Analyser 6000 high-resolution automated microscope. Nuclear and seeded aggregates counting was performed using Fiji software.¹³ The relative level of seeding was calculated as the number of venus-positive punctae/aggregates in each field, normalized to the corresponding number of cells, and compared to the untreated control (Supplementary material and Fig. 1C and D). Because this seeding assay identified PSP-tau-MES extracts as better seeding agents than PSP-tau-CTX extracts, all the following in vitro and in vivo experiments were conducted with PSP-tau-MES (and referred to as ‘PSP-tau’).

Third, immunofluorescence experiments were performed using HEK293T and HEK293Trex cells expressing human (untagged) 0N4R tau P301S, seeded for 1 h with PSP-tau seeds or control material (CTL-tau) as described previously. Three days later, seeded cells were trypsinized to remove the residual extracellular seeds, pooled and expanded. The dish with the seeded cells alongside the untreated control was kept in culture for two passages (up to 10 days post-seeding). Cells were then fixed with ice-cold methanol, blocked with 5% FCS and incubated with pan-tau KJ9A antibody (Agilent) and/or phospho-AT8 antibody (Ser202/Thr205, MN1020, Fisher Scientific) and/or AT100 antibody (Thr212/Ser214, MN1060, ThermoFisher Scientific) and the Alexa488/568/647-conjugated secondary antibodies (ThermoFisher Scientific) appropriate for each species. To label tau fibrils, the amyloid-specific pFTAA (pentameric formyl thiophene acetic acid) staining was used (incubation of 30 min with 3 µM pFTAA in PBS). Images were acquired using a Leica DMI 4000 B microscope and Leica-LAF software (Supplementary material, Supplementary Fig. 1 and Fig. 1E).

Finally, sarkosyl-insoluble tau from seeded cells was extracted after two passages (up to 10 days post-seeding) as described previously.¹⁴ After incubation with 1% n-lauroylsarcosinate (1 h at room temperature while shaking), the cell lysate supernatant was retained as the ‘soluble’ fraction. The pellet (‘insoluble’ fraction) was resuspended in PBS (Supplementary material and Fig. 1F). Initial cell lysate and soluble and insoluble fractions were then analysed with western blot using pan-tau KJ9 antibody and AT8 antibody.

Animals and stereotactic injections

Four male rhesus macaques (M. mulatta, Xieerxin, Beijing, China, mean weight = 9.5 kg, mean age = 7 years) were housed in individual cages under controlled conditions allowing visual contact and interaction with monkeys housed in adjacent cages. Food and water were available ad libitum. Animal care was supervised daily by veterinarians skilled in the healthcare and maintenance of non-human primates. Experiments were performed in accordance with the European Union directive (2010/63/EU) on the protection of animal use for scientific purposes in the Motac facilities in Beijing, accredited by AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care) and following study design acceptance by the ethical committee of the Institute of Lab Animal Science (Chinese Academy of Science, Beijing, China, approval number LQ20001).

Macaques received two stereotactic bilateral supranigral injections (anterior commissure = −7, depth = −4, length = 3.5 and anterior commissure = −8.5, depth = −4.5, length = 4; Fig. 4A) of PSP-tau seeds (n = 2) or CTL-tau (n = 2) that were bath-sonicated during 5 min before injections. The total injected volume per hemisphere was 20 µl (10 µl per injection site). After each injection, the syringe was left in place for 5 min to prevent leakage along the needle track.

Figure 4.

Injections of patient-derived PSP-tau aggregates induced typical neuropathological PSP lesions. (A) Anatomical representation showing the schematic distribution of tau pathology in PSP macaques along the antero-posterior (rostro-caudal) axis. (B–J) Representative images of AT8 immunostaining. (B and C) Tufted astrocytes, (D) globose tangles (E) coiled bodies and (F) neurofibrillary tangles and threads were observed close to the injection sites. (G–I) Tau inclusions were also found in anterior remote brain areas, in the putamen (Pu), caudate (Cd), Globus pallidus (GPe) and antero-ventral thalamus (VA). Scale bars = 20 µm. (K) Representative double-stained immunofluorescence images confirming astrocytic (GFAP-S100), oligodendroglial (Olig2) and neuronal (NeuN) phospho-tau (AT8) inclusions. (L) Representative images of three-repeat isoform RD3 and four-repeat isoform RD4 immunostaining in PSP and control macaques. Scale bars = 20 µm.

Eighteen months post-injection, macaques were euthanized using terminal anaesthesia with an overdose of ketamine (15 mg/kg) and pentobarbital (100 mg/kg), followed by perfusion with 2000 ml of 0.9% saline solution with 1% heparin, as recommended by the European Veterinary Medical Association guidelines. Brains were quickly removed after perfusion, post-fixed in 4% PFA for 1 week, cryo-protected in sucrose PBS, frozen in a snap-frost isopentane bath at −50°C and stored immediately at −80°C. All histological analyses reported here were performed on the right hemisphere.

Behavioural analyses

Motor behaviour—kinematic analyses

We evaluated motor performance using a corridor gait platform, as previously described.¹⁵ Briefly, we trained the macaques for 1 month after injections to cross the corridor in response to placing a food reward on the opposite side. We then recorded their full-body kinematics at 6, 12 and 18 months after injections. To record the kinematics, we first attached 20 reflective markers to the right and left sides of arms, torso and legs of the macaques for motion tracking. A kinematic recording trial started on the one end of the corridor with the macaque sitting. Macaque then walked across the corridor and sat on the other side, ending the kinematic recording trial. During the trials, we captured the whole-body kinematics using the SIMI-Motion capture system (Simi-Motion Systems, Germany), combining four high-speed (100 Hz) synchronized cameras placed around the corridor. We used the SIMI-Motion software to reconstruct the 3D spatial coordinates of each marker over time (Fig. 2A). We extracted the kinematic data from each trial, and computed and analysed 57 kinematic features of locomotor pattern for each gait cycle¹⁵ (Supplementary material). First, to visualize gait differences between PSP and control macaques without a priori, we projected the extracted 57 kinematic features into an interpretable low-dimensional space using principal component analysis (PCA; Supplementary material). Second, kinematic features representing symptoms related to parkinsonism were particularly assessed (e.g. step duration, step length, limb length, articulation range of motion) because macaques were injected in the supranigral area.

Figure 2.

Injections of patient-derived PSP-tau aggregates induced parkinsonism in macaques. (A) Schematic representation of the experimental design and of the kinematics recording session. Macaques were equipped with markers for the extraction of whole-body kinematics in 3D (57 kinematic features for each gait cycle were computed and analysed). Recordings were performed in three recording sessions that occurred 6, 12 and 18 months post-injection. (B) Principal component analysis (PCA) 6 months post-injection (first session). Each dot on the left panel plot represents one gait cycle. Different colours represent different macaques. Larger dots represent the means across all gait cycles of each macaque. The right panel plot shows the same data represented by ellipsoids with the centre and principal semi-axis as the mean and SD calculated across all the gait cycles for that condition and macaque—in a space spanned by the three leading principal components (PCs). (C) We performed the PCA analysis over the concatenated gait parameter data of three recording sessions m6, m12 and m18. The plot shows the points that represent average of gait parameters for each session and each macaque projected into the spanned by the three leading PCs. Note that CTL-tau No. 2 macaque declined to perform m18 session. (D) Bar plots represent the mean values of gait features suggestive of parkinsonism. Difference limb elevation angle describes the amplitude of limb oscillation in degrees around the hip joint; Difference shank elevation angle describes the amplitude of limb oscillation in degrees around the knee joint; Difference knee angle describes the difference in degrees between maximal flexion and maximal extension of the limb (number of gait cycles: CTL-tau 1: 43, CTL-tau 2: 33, PSP-tau 1: 36 and PSP-tau 2: 36). *P < 0.05, **P < 0.01, ***P < 0.001; Kruskal–Wallis test with Tukey–Kramer correction for multiple comparisons.

Cognitive behaviour—object retrieval

Object retrieval task was previously described by Schneider et al.¹⁶ Briefly, we trained the macaques to reach for and retrieve a food reward from a clear Plexiglas box (15 × 15 × 15 cm) with one open side, presented just outside of the testing cage (Fig. 3A). The position of the box, the location of the open side of the box and the placement of the reward differed during trials (30 trials per test session), thus affecting the cognitive and motor difficulty of the trial. We divided the trials into ‘easy’ and ‘hard’ depending on the position of the reward in the box and the position of the open side. We recorded the percentage of first attempt successes, the number of motor errors (dropping reward,…), the number of cognitive errors (impulsive reach, persevered reach, frontal barrier reach,…) and the completion time. We trained the macaques 1 month after injections to obtain a 90% of successful reward retrievals. We then recorded the performance measures at 6, 12 and 18 months after injections.

Figure 3.

Injections of patient-derived PSP-tau aggregates induced the progressive occurrence of a dysexecutive syndrome in macaques. (A) Schematic representation of the longitudinal assessment using the object retrieval task that assesses inhibition, planning and perseverations. (B) Longitudinal analysis of macaques’ performance with hard items. Thin dotted lines: longitudinal performance of individual macaques. Wide and full lines: mean trajectory of PSP and CTL macaques. **P < 0.01 and ***P < 0.001; two-way ANOVA.

Histological analysis

Tau staining

Immunohistochemical staining for phosphorylated-tau (AT8, MN1020, Fisher Scientific, 1:1000, mouse) and isoform-specific tau (RD3, 8E6/C11, Fisher Scientific, 1:100, mouse and RD4, 1E1/A6, Fisher Scientific, 1:100, mouse) was performed on coronal free-floating slices, as previously described.¹⁷ Sections were incubated with the primary antibody overnight at room temperature and then with EnVision-HRP enzyme conjugate secondary antibodies (anti-mouse, Agilent DAKO) for 30 min. Sections were revealed with 3,3′ diaminobenzidine (DAB, Agilent DAKO), counterstaining with 0.1% Cresyl violet solution, and mounted on gelatinized slides. All sections were scanned in a panoramic digital slide's scanner (Panoramic Scan II, 3DHISTECH). For 3-repeat isoform RD3 and 4-repeat isoform RD4 staining, a specific additional pretreatment was necessary, as previously described¹⁸: before antibody incubation, sections were treated with 100% formic acid for 30 min and then heat retrieved with 0.01 M citrate buffer pH 6 in water bath (80°C for 30 min).

Immunofluorescence

Double staining immunofluorescence was performed to confirm both glial and neuronal tau inclusions. Staining specific for neurons (NeuN, Sigma-Aldrich, 1:1000, chicken), astrocytes [glial fibrillary acidic protein (GFAP), Agilent Dako, 1:4000, rabbit; S-100, Abnova, 1:1000, rabbit], oligodendrocytes (Olig 2, Sigma-Aldrich, 1:100, rabbit) and for phosphorylated-tau (AT8, MN1020, Fisher Scientific, 1:100, mouse), were performed on coronal free-floating slices. Sections were permeabilized and blocked using normal goat serum diluted in PBS 1× Saponine 0.2% for 1 h before incubating 90 min with primary antibodies overnight at 4°C. The next day, sections were incubated sequentially with the appropriate secondary antibodies coupled with a fluorochrome (Goat anti-mouse Alexa 568, goat anti-chicken Alexa 488 and/or goat anti-rabbit Alexa 488, Fisher Scientific, 1:400). Then, to reduce lipofuscin autofluorescence, slices were treated with a solution of Sudan Black B 0.1% (Sigma-Aldrich) in 70% ethanol for 10 min. After washing, sections were counterstained with 10 µM of Hoechst 33 342 and then mounted with Vecta-Shield without DAPI media (Vector Laboratories) on non-gelatinized slides. Images were acquired at 40× magnification using a Zeiss Axio Imager 2 (ExploraNova).

Neurodegeneration

To assess the impact of PSP-tau injections on the dopaminergic neurons of the substantia nigra, tyrosine hydroxylase (TH) immunohistochemistry was performed as previously described.¹⁹ Briefly, serial sections of substantia nigra corresponding to the whole area were incubated with a human anti-TH antibody (EP1532Y, Abcam, 1:5000, rabbit) overnight at room temperature and then revealed by EnVision-HRP enzyme conjugated secondary antibodies (anti-rabbit, Agilent DAKO) for 30 min. Sections were revealed with 3,3′ diaminobenzidine (DAB, Agilent DAKO), counterstaining with 0.1% Cresyl violet solution, and mounted on gelatinized slides.

As previously described, TH-positive cells were counted by stereology, using a Leica DM6000B motorized microscope coupled with the Mercator software (ExploraNova), blinded of the experimental conditions. The substantia nigra was delineated for each slide, and probes for stereological counting were applied to the map (size of probes 100 × 80 µm spaced by 600 × 400 µm). Finally, the optical fractionator method was used to estimate the total number of TH-positive cells in the substantia nigra of each macaque.

Statistical analysis

For seeding assay, kinematics and object retrieval task experiments, comparisons between means were carried out by using two-way ANOVA and followed, when appropriate, by post hoc multiple comparisons tests. All values are expressed as the mean ± standard error of the mean (SEM). Statistical analyses were performed with GraphPad Prism v.9.2.0 (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance was set at a P-value <0.05.

The debate about the need to move beyond P-value is raging and data must now be analysed further with estimation graphics,²⁰ emphasizing the effect size. Therefore, we used an estimation graphic called the ‘Gardner–Altman plot’ to present the TH data. This plot uses two graphs. The graph on the left presents data of CTL-tau and PSP-tau groups as scatter plots showing the observed values along with the previously defined descriptive statistics (mean ± SEM). The graph on the right displays the effect size by presenting the distribution of the difference of TH + neuron counts between the CTL-tau and PSP-tau groups using resampled distributions of observed data (blue curve). Horizontally aligned with the mean of the test group, the mean difference is indicated with the black circle. The black vertical line illustrates the 95% CI of the mean difference.

Data availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Results

In vitro characterization of proteopathic seeds

We extracted and purified sarkosyl-insoluble tau fractions from MES and cortical samples of PSP brains (PSP-tau-MES and PSP-tau-CTX). We extracted the same fraction from control cortical samples (CTL-tau) (Fig. 1A). The final total tau concentration was higher in PSP-tau-MES sample (2 µg/ml) than in PSP-tau-CTX sample (0.5 µg/ml) and CTL-tau sample (0.2 µg/ml) (Fig. 1B). The incubation of tau seeds on P301S-Venus expressing cells revealed a higher in vitro seeding activity of the PSP-tau-MES sample compared to PSP-tau-CTX and CTL-tau samples [Fig. 1C and D; two-way ANOVA: F(10,144) = 44.3, P < 0.0001; concentration: F(5,144) = 58.6, P < 0.0001; samples: F(2,144) = 232.1, P < 0.0001].

On the basis of these first in vitro results, the PSP-tau MES sample was selected for further in vitro characterization and future injections in macaques (referred to as ‘PSP-tau’). Immunostaining of P301S-HEK293T cells incubated with PSP-tau confirmed the abnormal phosphorylation of tau aggregates (AT8- and AT100-positivity) and their misfolded and filamentous nature (pFTAA positive) (Fig. 1E and Supplementary Fig. 1). Finally, we confirmed that the ‘second generation’ AT8-positive tau aggregates generated in vitro in P301S-HEK293T cells were insoluble in sarkosyl (Fig. 1F), like the original human-PSP proteopathic seeds.

Brain injections of PSP-tau induced parkinsonism and dysexecutive syndrome in macaques

Macaques were injected bilaterally in the supranigral area with PSP-tau (n = 2) or CTL-tau (n = 2). Every 6 months after surgery, macaques were tested for motor and cognitive behaviour (Fig. 2A). We assessed motor abilities by limb kinematic analyses while macaques were walking in a corridor. The PCA using all kinematic features discriminated PSP and CTL macaques 6 months after injections (Fig. 2B). PCA confirmed the behavioural difference for recording sessions 12 and 18 months after injections (Fig. 2C and Supplementary Fig. 2). Compared to controls, the two macaques injected with PSP-tau exhibited similar motor symptoms, mimicking parkinsonism, with a significant increase in step duration, a significant decline in step length, a significant decrease in limb length and a significant decrease in articulation range of motion (P < 0.05, one-way ANOVA with post hoc Tukey’s test for multiple comparisons; 6, 12 and 18 months after injection; Fig. 2D and Supplementary Fig. 2).

We assessed cognitive impairment using the object retrieval task (Fig. 3A), which measures executive function performance (inhibition and planification) for the retrieval of ‘hard’ items. We found no differences between macaques’ performance during ‘easy’ trials, which do not require executive functions. We also found no significant difference between macaques regarding the completion time or the number of motor errors, whatever the task’s difficulty. However, during ‘hard’ trials, PSP macaques made significantly more cognitive errors [two-way ANOVA, interaction time course × macaques: F(2,4) = 11.32, P = 0.023; time course: F(2,4) = 6.77, P = 0.052; macaques: F(2,4) = 110.6, P = 0.0003] and failed significantly more often to retrieve ‘hard’ rewards on the first attempt [two-way ANOVA, interaction time course × macaques: F(2,4) = 6.18, P = 0.059; time course: F(2,4) = 1.77, P = 0.28; macaques: F(2,4) = 35.6, P = 0.003]. This dysexecutive syndrome increased gradually over time (Fig. 3B).

Brain injections of PSP-tau induced typical PSP pathology

We euthanized the macaques 18 months after intracerebral injections. Histological examination of the whole brain showed AT8-positive lesions of tauopathy only in macaques injected with PSP-tau. These lesions were found close to the injection sites (supranigral area and ventral thalamus; Fig. 4A–F) and also in rostral (anterior) connected areas (putamen, caudate, globus pallidus and thalamus; Fig. 4A and G–J). We did not find AT8 staining in more caudal (posterior) regions (cerebellar pedunculus and nuclei). The morphology of AT8-positive lesions was typical of those usually observed in the brain of PSP patients, with tufted astrocytes (Fig. 4B and C), globose tangles (Fig. 4D), coiled bodies in oligodendrocytes (Fig. 4E) and neurofibrillary tangles (Fig. 4F). While astroglial AT8-positive lesions were the most commonly found close to the injection site, we found mainly neurofibrillary tangles and some coiled bodies in remote brain areas (Fig. 4G–J).

Double immunofluorescence histochemistry confirmed the presence of both glial and neuronal inclusions, as evidenced by the colocalization between tau aggregates (AT8) and astrocytes (GFAP-S100), oligodendrocytes (Olig2) and neurons (NeuN) (Fig. 4K).

Finally, three-repeat isoform RD3 and four-repeat isoform RD4 immunostaining revealed 4R-tau inclusions in macaques injected with PSP-tau, but no 3R-tau inclusion. We found neither 4R-tau nor 3R-tau deposition in CTL macaques (Fig. 4L).

Brain injections of PSP-tau induced modest dopaminergic cell loss

Finally, we assessed whether brain injections of PSP-tau affected dopaminergic neurons viability in the substantia nigra (Fig. 5). The effect size displayed on the right side of the Gardner–Altman plot shows a 19.3% loss of dopaminergic neurons in PSP-tau macaques compared to control-injected macaques (Fig. 5A and B) but neuronal loss only seemed clear in a single PSP-tau macaque. These data indicate that PSP-tau-injected macaques present modest dopaminergic neuronal death in addition to tau pathology, indicating a potential first phase of PSP-related neurodegeneration induction.

Figure 5.

Injections of patient-derived PSP-tau aggregates induce dopaminergic neuronal loss in the substantia nigra of macaques. (A) Representative images of thyroxine hydroxylase (TH) immunohistochemistry on the substantia nigra of CTL-tau (left) and PSP-tau macaques (right). (B) Stereological quantification of the number of TH+ neurons in the substantia nigra of macaques. Data are shown as mean ± SEM (horizontal lines). Each dot represents one NHP of the control (grey) and PSP-injected macaques (blue). The bootstrapped mean difference with 95% CI (error bar) is shown on the right side of this graph. Scale bars = 10 µm (inset, 20 µm).

Discussion

We showed that patient-derived PSP-tau can trigger the pathological conversion of endogenous tau in macaques with both glial and neuronal 4R-tau inclusions. The same extract from control brains was biologically inert. We also illustrated that this pathological process spread in a manner consistent with the ‘prion-like’ hypothesis: from the injection site to connected brain areas usually affected in PSP. Furthermore, the location of these pathological lesions correlated with the progressive development of a PSP-like phenotype. While the first mesencephalic lesions around the injection site explain the rapid appearance of parkinsonism 6 months post-injection, the diffusion of tau aggregates towards the striatum and the pallidum can explain the progressive appearance of a dysexecutive syndrome. Indeed, these basal ganglia nuclei reportedly support executive functions²¹ (especially inhibition, as measured here with the object retrieval task) and are associated with executive performances in PSP.^21,22

Unlike previous studies that used uncharacterized brain homogenates,^23,24 we extracted and purified proteopathic tau seeds from human-PSP brains and then carefully characterized them in vitro. We were, therefore, able to quantify the amount of tau proteins in our samples, showing in our experiments that <40 ng of pathological tau were sufficient to induce the seeding and the spreading of the disease in a primate brain (the final total tau concentration of the PSP-tau sample was 2 µg/ml, and a total of 40 µl was injected per macaque; i.e. ∼40 nM, assuming 50 kDa for tau). This concentration was far below the concentrations commonly used in mouse challenge experiments (micromolar concentrations) and closer to physiological ranges.²⁵ Furthermore, we showed before injection that PSP-tau seeds were (as expected) misfolded, filamentous (pFTAA positive) and able to induce in vitro sarkosyl-insoluble phospho-tau aggregates in cell-based assays.

Understanding tauopathies’ clinical and pathological heterogeneity is a hot topic in clinical neurosciences.²⁶ Distinct anatomical distributions, distinct filament folds and distinct cell-type specificities are currently thought to be the main explanations for this heterogeneity.^27,28 Our results strengthen and extend previous findings in both transgenic and wild-type rodents about the cell-type specificity of distinct tau ‘strains’, where brain homogenates or tau seeds from different human tauopathies induce unique cellular distributions of tau pathology (i.e. ‘pure’ neuronal inclusions in Alzheimer’s disease, glial inclusions in PSP and corticobasal degeneration, etc). Indeed, here we report all the pathological hallmarks of PSP: neurofibrillary and globose tangles in neurons, tufted astrocytes and coiled bodies in oligodendrocytes, using both classical immunohistochemistry and double staining immunofluorescence to confirm the cell-type specificity of inclusions. For example, the AT8/GFAP-S100 double-labelling in PSP macaques revealed the typical redistribution of GFAP filaments around the nuclei, which were surrounded by tufts of AT8-positive tau inclusion, precisely as in human PSP.²⁹ In contrast, we did not find small AT8-positive dots along astroglial processes that define the astrocytic plaques observed in corticobasal degeneration, another 4R tauopathy that shares similarities with PSP.

Interestingly, we found astroglial inclusions predominantly close to the injection sites. In contrast, we mainly observed neuronal inclusions in remote (but connected) brain areas (associated with some oligodendroglial coiled bodies). An inverse correlation between neuronal and astrocytic tau pathology has been previously described in mice injected with PSP-tau.³ This experimental evidence echoes a recent report combining functional MRI, tau-PET and post-mortem neuropathological assessment in PSP patients.⁵ This report showed that the association between brain connectivity and the distribution of tau pathology was strongest for neuronal tau compared to glial tau. It suggests that transsynaptic neuronal transmission is the main driver of tau spreading in PSP, according to the ‘prion-like’ theory. The mechanisms underlying glial transmission of tauopathy remain, however, elusive.

A long-standing question in the PSP field concerns clinicopathological correlations. Are intracellular tau aggregates sufficient to induce cell dysfunction and hence clinical symptoms, or must there be associated neuronal death? Do glial or neuronal tau inclusions have the greatest impact on the development of clinical symptoms? Our study provides evidence that can help answer these questions because we clearly found parkinsonian symptoms in PSP macaques without dramatic dopaminergic denervation but in the presence of a predominant astrocytic pathology in the peri-nigral regions. In non-human primate models of Parkinson’s disease, overt parkinsonism is observed only when the lesion reaches ∼45% of dopamine neurons in the substantia nigra and 80% of dopaminergic fibres in the striatum.^30,31 Non-overt symptoms such as cognitive impairment or ethologic behaviour disruption have been reported with sub-threshold lesions obtained with neurotoxins^32,33 or Lewy bodies intracerebral administration,³⁴ but the consensus stands firm about motor symptoms. A dopamine neuron degeneration by 19.3%, as we report here, is inconsistent with the parkinsonian symptoms reported in our kinematic experiments. This provides strong evidence that neuronal death is not a necessary condition for appearance of motor symptoms in PSP and that glial pathology has a direct functional impact.

Previous studies investigating the transmission of tau pathology from various types of human sample to animals have relied on the use of transgenic mice overexpressing human mutant tau^23,35 and, more rarely, on wild-type mice^23,36 or mouse lemurs^24,37 (the smallest prosimian primates, presenting some genetic and brain morphological homologies with humans). To the best of our knowledge, our work is the first reporting the consequences of injecting patient-derived tau aggregates in rhesus monkeys (M. mulatta), the animal species phylogenetically closest to humans on which we can conduct research. This experimental choice allowed us to reproduce the pathophysiology of sporadic tauopathies as closely as possible. Indeed, human and macaque tau shared strong sequence homology,³⁸ and both species expressed 3R and 4R tau isoforms (while adult rodents only expressed 4R tau), which is of the utmost importance for the templating of soluble proteins into misfolded ones.^39,40 Investigations on macaques also allowed us to study fine-grained anthropomorphic cognitive and motor skills to demonstrate that the tau-induced pathology is sufficient to cause PSP-like motor and cognitive symptoms, which was not achieved with rodents. Finally, the shape, size and connectivity of human and macaque brains are comparable. This will enable precise anatomical examination of tau spreading and future use of identical imaging technology (MRI or tau-PET) in humans and animals, thereby accelerating translational research opportunities.

The main limitation of this pilot study is its small sample size. However, a small sample size mainly impacts statistical power (large β risk), making negative results difficult to interpret. Still, in this study, we consistently induced PSP pathology and phenotype in the two injected macaques while the two controls remained unharmed. Such a clearcut picture explains that pilot studies are conducted with two or three animals per group, a number accepted in the field of macaque research⁴¹ for both scientific and ethical reasons. Such pilot studies should, however, be expanded and confirmed using larger cohorts. We also acknowledge that our experiments only recapitulate a particular type of PSP, with predominant parkinsonism and dysexecutive presentation,¹ in relation to the supra-nigral injection site. Future studies need to address how to induce other clinical presentations of PSP, usually associated with distinct spatial features of tau pathology, distinct involvement of neuronal versus glial pathology and/or higher tau load than PSP with predominant parkinsonism.^3,42

Our results demonstrate that the brain injection of patient-derived PSP-tau aggregates can induce typical neuropathological PSP lesions that can trigger motor and behaviour impairments in a primate host. In addition to the pathophysiological information about the ‘prion-like’ nature of the disease, our results support the use of PSP-tau inoculated macaques as relevant animal models to accelerate drug development targeting this rare and fatal neurodegenerative disease.

Funding

This work was supported by grants from PSP-France Foundation, Encephalia and the Bettencourt-Schueller Foundation (CCA-recherche Inserm-Bettencourt). This study received financial support from the French government in the framework of the University of Bordeaux’s IdEx ‘Investments for the Future’ program/GPR BRAIN_2030. M.D. received a PhD fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche. The in vitro part of this project received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement number 116060 (IMPRiND). This Joint Undertaking received support from the European Union’s Horizon 2020 research and innovation program and EFPIA. W.M. is a Lister Institute Fellow and supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (206248/Z/17/Z). This work was supported by the UK Dementia Research Institute, which receives its funding from DRI Ltd, funded by the UK Medical Research Council, Alzheimer’s Society and Alzheimer’s Research UK. This work was also supported by the Swiss State Secretariat for Education, Research and Innovation (SERI) under contract number 17.00038, and the Defitech Center for Interventional Neurotherapies—NeuroRestore. The opinions expressed and arguments employed herein do not necessarily reflect the official views of these funding bodies.

Competing interests

E.B. is a director and shareholder of Motac Neuroscience Ltd. G.C. is a shareholder of ONWARD Medical, a company without direct relationships with the presented work. The other authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.