Development of a novel tau propagation mouse model endogenously expressing 3 and 4 repeat tau isoforms

Abstract

The phenomenon of ‘prion-like propagation’ in which aggregates of abnormal amyloid-fibrilized protein propagate between neurons and spread pathology, is attracting attention as a new mechanism in neurodegenerative diseases. There is a strong correlation between the accumulation or spread of abnormal tau aggregates and the clinical symptoms of tauopathies. Microtubule-associated protein tau (MAPT) contains a microtubule-binding domain that consists of three or four repeats (3R/4R) due to alternative mRNA splicing of transcripts for the MAPT gene. Although a number of models for tau propagation have been reported, most use 4R human tau transgenic mice or adult wild-type mice expressing only endogenous 4R tau and these models have not been able to reproduce the pathology of Alzheimer's disease in which 3R and 4R tau accumulate simultaneously, or that of Pick’s disease in which only 3R tau is aggregated. These deficiencies may reflect differences between human and rodent tau isoforms in the brain.

To overcome this problem, we used genome editing techniques to generate mice that express an equal ratio of endogenous 3R and 4R tau, even after they become adults. We injected these mice with sarkosyl-insoluble fractions derived from the brains of human tauopathy patients such as those afflicted with Alzheimer’s disease (3R and 4R tauopathy), corticobasal degeneration (4R tauopathy) or Pick’s disease (3R tauopathy).

At 8–9 months following intracerebral injection of mice, histopathological and biochemical analyses revealed that the abnormal accumulation of tau was seed-dependent, with 3R and 4R tau in Alzheimer’s disease-injected brains, 4R tau only in corticobasal degeneration-injected brains and 3R tau only in Pick disease-injected brains, all of which contained isoforms related to those found in the injected seeds. The injected abnormal tau was seeded, and accumulated at the site of injection and at neural connections, predominantly within the same site. The abnormal tau newly accumulated was found to be endogenous in these mice and to have crossed the species barrier. Of particular importance, Pick’s body-like inclusions were observed in Pick’s disease-injected mice, and accumulations characteristic of Pick’s disease were reproduced, suggesting that we have developed the first model that recapitulates the pathology of Pick’s disease.

These models are not only useful for elucidating the mechanism of propagation of tau pathology involving both 3R and 4R isoforms, but can also reproduce the pathology of tauopathies, which should lead to the discovery of new therapeutic agents.

Introduction

The distribution of aggregated proteins such as tau in Alzheimer’s disease, α-synuclein in Parkinson’s disease and Lewy body disease or TAR-DNA binding protein of 43 kDa (TDP-43) in frontotemporal lobar degeneration and amyotrophic lateral sclerosis were shown to be strongly correlated with clinical phenotypes, and to have an effect on disease progression.¹ The normal or abnormal protein is released during normal cellular activity or on cell death and these proteins are taken up by nearby cells or spread via synaptic transmission. This process can be frequently repeated resulting in the spread of aggregated protein within the brain. Recently, this phenomenon has been called ‘prion-like propagation’.

The neurodegenerative diseases in which tau accumulates are collectively referred to as ‘tauopathies’. These include Alzheimer’s disease, argyrophilic grain disease, corticobasal degeneration, chronic traumatic encephalopathy, globular glial tauopathy, progressive supranuclear palsy, Pick’s disease and senile dementia of neurofibrillary tangle type.

Six tau isoforms are expressed in adult human brain. They range from 352 to 441 amino acids in length and are produced by alternative mRNA splicing of transcripts from the MAPT gene. The six isoforms are unfolded in the native state and differ by the absence (0N) or presence of insertions of 29 (1N) to 58 (2N) amino acids in the amino terminal half. The inclusion of the 31 amino acid repeats encoded by exon 10 results in the production of three tau isoforms with four repeats each (4R), and its exclusion results in a further three isoforms with three repeats each (3R) in the microtubule-binding domain. The isoform composition of repeats from the tau filament core can vary among diseases. Thus, in Alzheimer’s disease, chronic traumatic encephalopathy and senile dementia of the neurofibrillary tangle type, both 3R and 4R tau make up the neurofibrillary tangles. However, in Pick’s disease, 3R tau predominates in the neuronal inclusions. The assembly of 4R tau into filaments is characteristic of argyrophilic grain disease, corticobasal degeneration, globular glial tauopathy and progressive supranuclear palsy.

The first report of abnormal tau propagation in vivo was demonstrated in 2009.² Brain homogenates of P301S tau transgenic mice were injected into the hippocampus and upper cortical area of Alz17 mice that are human 2N4R tau transgenic.³ Six months later, the accumulation of phosphorylated tau or Gallyas–Braak-positive tau was observed. Next, the same group injected brain homogenates into Alz17 mice from tauopathies such as Alzheimer’s disease, senile dementia of the neurofibrillary tangle type, argyrophilic grain disease, progressive supranuclear palsy, corticobasal degeneration and Pick’s disease. Mice inoculated with these tauopathies, except for Pick’s disease, were able to reproduce the respective human pathology.⁴ Brains of mice inoculated with brain extracts of Pick’s disease showed the formation of dystrophic neurites but no evidence of Pick’s bodies. Human 4R tau was expressed in Alz17 mice and the endogenous tau expression in these brains was also 4R tau. This may be the reason why brain extracts from Pick’s disease mice, in which only 3R tau accumulates, cannot reproduce the human pathology.

Preformed fibrils were generated in vitro with recombinant tau and injected into the hippocampus of young PS19 mice.⁵ Within 1 month, preformed fibrils were found to have strongly accumulated in the vicinity of the inoculation site, and it was shown that they caused the accumulation of abnormal tau and propagation from the inoculation site to brain regions with neural connections.

Another group has also reported experiments in which the hippocampus of PS19 mice was injected with P301L-K18 fibrils.⁶ The accumulation of abnormal tau was observed from 1.5 months after inoculation, and subsequent propagation from the hippocampus to areas of neural connections was also observed. There have also been reports of the injection of PS19 mice with tau aggregates prepared in cultured cells as seeds.^7,8 In another study, the unilateral hippocampus of young PS19 mice was injected with brain extracts from patients with corticobasal degeneration or patients with Alzheimer’s disease.⁹

Brain extracts from 5–5.5-month-old P301S mice were used to inject the hippocampus and the superior cortex of 2-month-old P301S mice.¹⁰ At 2 weeks after inoculation, tau accumulation was seen in neurons of the CA1 region of the hippocampus, and 2.5 months later, tau pathology was observed to have spread from the hippocampus to brain regions with neural connections and to the contralateral hippocampus. The present study also showed that tau pathology was propagated in both an anterograde and retrograde direction.

Peeraer et al.¹¹ constructed a model in which they injected the hippocampus of P301L tau transgenic mice with P301L-K18 fibrils generated in vitro. With this model, they were able to induce AT8-positive tau pathology in the hippocampus, cerebral cortex, amygdala and thalamus on the inoculated side of the brain as early as 1 month post-injection.

A model of tau propagation using transgenic mice in which the endogenous tau had been knocked out and which expressed only the human six-isoform tau was reported.¹² The hippocampus of 3-month-old mice was injected with hyper-phosphorylated tau oligomers isolated from the cerebral cortex of Alzheimer’s disease patients. Eleven months after the injection, neurofibrillary changes and neuropil thread-like pathology were noted.

Experiments using a mouse model in which human tau is expressed exclusively in the entorhinal cortex also showed that abnormal tau propagates through synaptic circuits to the hippocampus and dentate gyrus.^13,14

In 2016, a model of tau propagation in wild-type mice was reported, but with this model, it is necessary to keep in mind that most of the tau expressed in the adult brain is 4R.¹⁵ Sarkosyl-insoluble fractions were extracted from the brains of Alzheimer’s disease patients and used to inject the hippocampal dentate gyrus of wild-type mice.¹⁶ In these experiments, argyrophilic-grain-disease-like 4R tau pathology formed in the hippocampus and fimbria at 3 months after inoculation.

The same group performed injection of the unilateral hippocampus of wild-type mice with the following three types of fibril: 2N4R fibrils produced in vitro with heparin (Hep-T40) or without heparin (X-T40), and fibrils extracted from an Alzheimer’s disease brain (Alzheimer’s disease-tau).¹⁷ Immunohistochemical staining with AT8 at 3 months after inoculation showed the most widespread tau pathology in the Alzheimer’s disease-tau-injected mice.

Sarkosyl-insoluble fractions derived from the brains of Alzheimer’s disease, corticobasal degeneration and progressive supranuclear palsy patients were used to inject the hippocampus and upper cerebral cortex of wild-type mice.¹⁸ Three months after the injection, immunohistochemical staining with AT8 revealed the accumulation of phosphorylated tau at the inoculation site and in the area with direct neural connection to the hippocampus.

Thus, all of the propagation models before 2016 used overexpression of transgenic tau, and in addition, the mice expressed endogenous mouse 4R tau, which is a unique environment. It is well known that the expression patterns of tau isoforms in the mouse brain are different from those in humans. In humans, the isoform of tau expression in the foetal brain is 3R. Subsequently, tau expression patterns change with growth, and adult human brain expresses both 3R and 4R tau. In rodents including mice, only 3R tau is expressed in the brain during foetal and juvenile stages, but it has been shown that it is mostly replaced by 4R tau at 2 weeks of age in mice.¹⁵ Therefore, since only 4R tau is expressed in the wild-type mouse model, the pathology of 4R tauopathies, such as corticobasal degeneration and progressive supranuclear palsy, can be reproduced to some extent.^4,18 On the other hand, the lower seeding efficiency of Alzheimer’s disease, a 3R+4R tauopathy, compared to that of corticobasal degeneration and progressive supranuclear palsy in wild-type brain,¹⁸ may be due to the fact that only 4R tau is expressed in the mouse brain. Models of 3R tauopathy have never been reported other than in one study.⁴ In our present study, to reproduce more human-like 3R+4R pathologies such those that as occur in Alzheimer’s disease and to create a model for the propagation of Pick’s disease, we generated a novel mouse line that endogenously expresses 6-isoform mouse tau, similar to that of humans, using a genome editing technique. Using this mouse, we developed a novel recapitulated isoform-dependent pathogenesis model of tau.

Materials and methods

Ethics statement

This study was approved by the Ethics Committee of the Tokyo Metropolitan Institute of Medical Science (Permission numbers 21–32 and 15–33) and performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All animal experiments were carried out in agreement with the Guidelines for Proper Conduct of Animal Experiments (Science Council of Japan) and the ARRIVE guidelines, and all experimental protocols were approved by the Animal Care and Use Committee of the Tokyo Metropolitan Institute of Medical Science (Permission numbers 14046, 14058, 15037, 16029, 17046, 18040, 19042 and 20-035).

Generation of an endogenous 3R/4R tau expression mouse by genome editing

Genome-edited mice (Tau 3R/4R mice) were generated using the CRISPR–Cas9 system.^19-21 The designed target sequence was 5ʹ-GGATAATATCAAACACGTCCCGG-3ʹ (the PAM region is underlined). To prepare template DNA for sgRNA synthesis, PCR amplification was performed using the synthetic gene of the guide RNA sequence²² as a template. The forward primer containing the T7 promoter and target sequence was 5ʹ-gacttaag ctaatacgactcactataGGATAATATCAAACACGTCCgttttagagctagaaatagcaagt-3ʹ, and the reverse primer was 5ʹ-aaaagcaccgactcggtgcc-3ʹ. Next, in vitro transcription for sgRNA synthesis was performed using a MEGAshortscript T7 Transcription kit (Life Technologies). Cas9 mRNA was prepared using an in vitro transcription kit (mMESSAGE mMACHINE T7 Ultra kit, Life Technologies) with Cas9 plasmid (Transposagen) as the template DNA. The synthetized RNAs were purified and eluted using a MEGAclear kit (Life Technologies) and RNase-free water.

The mixture of Cas9 mRNA (10 ng/µl), sgRNA (5 ng/µl) and donor DNA (10 or 100 ng/µl) in RNase-free water was microinjected into the pronucleus and cytoplasm of C57BL/6J (B6J) fertilized eggs (ARK Resource Co). The microinjected embryos were transferred into the oviduct of pseudo-pregnant CD1 females (Charles River Laboratories Japan). Tau 3R/3R mice were generated by crossing Tau 3R/4R mice, and the Tau 3R/4R mice used in the experiments were produced by crossing Tau 3R/3R mice with wild-type mice (Tau 4R/4R).

The mice were reared in the animal facility of Tokyo Metropolitan Institute of Medical Science under conventional conditions at 23 ± 2°C and were maintained on a commercial diet (CE-2, Nihon CLEA) ad libitum.

Genotyping

An aliquot of blood from the tail of each mouse was collected on filter paper and was used as a template for PCR. PCR amplification was carried out in 25 μl of reaction mixture containing 1× MightyAmp buffer v.2 (TaKaRa Bio), 0.625 units of MightyAmp DNA polymerase (TaKaRa Bio) and 10 pmol of forward and reverse primers. The primers synthesized were as follows: 5′-CCAGATTCCTTTTGTGACTTCCAGGGTGCCATCC-3′ (forward primer), and 5′-CCAGAGATGAGGGAAGAGGTGTCAGCC-3′ (reverse primer). The mixture was prepared before the addition of a 1.2 mm diameter disc removed from the blood sample on the filter. PCR amplification was carried out using a Veriti Thermal Cycler (Applied Biosystems). The amplification program consisted of an initial denaturation step at 98°C for 2 min. The remaining cycles were 10 s at 98°C, 15 s at 60°C and 35 s at 68°C. Thirty-two cycles were performed in all. After amplification, PCR products were separated on a 2% agarose gel and visualized by incubation for 10 min in a solution containing 10 ng/ml of ethidium bromide. Photographs of the gel were taken with a Gel Doc EZ Imager (Bio-Rad).

Preparation of the sarkosyl-insoluble fraction

Sarkosyl extraction of the human and mouse brains was performed as previously described.²³

Intracerebral stereotaxic injection

Sarkosyl-insoluble fractions derived from patients were sonicated for 60 s using an Ultra S Homogenizer VP-5S (TAITEC) before intracerebral injection. Tau 3R/4R mice aged 4–6 months were used. Animals were housed (six animals per cage) with free access to food and water. Mice were anaesthetized using isoflurane (Pfizer), and were immobilized in a stereotaxic frame (Stoelting). Five microlitres of the sarkosyl-insoluble fraction were injected unilaterally into the right striatum (anterior–posterior: 0.2 mm; medial–lateral: 2 mm; dorsal–ventral: −2.6 mm) using a LEGATO 130 syringe pump (KD Scientific) set at 2.5 μl/min.^24,25 When sampling, mice were euthanized under rapid anaesthesia with isoflurane and the brains were quickly removed. Brains of each group were fixed in 4% paraformaldehyde for 36 h for histochemical analyses or were cut along the sagittal plane, then frozen and stored at −80°C for biochemical analyses.

Histochemical analysis

For immunohistochemistry, the staining was performed as previously described.²⁶ The primary antibodies used for staining were as follows: AT8 (1:1000, Innogenetics), pS396 (1:1000, Calbiochem), T46 (1:1000, Invitrogen), 12E8 (1:1000, Dr Seubert), HT7 (1:1000, Thermo Scientific), RD3 (1:1000, Merck Millipore), anti-4R tau (1:1000, Cosmo Bio) and anti-mouse tau (1:1000).²⁷ Photographs were taken using a BX51 microscope equipped with a DP71 camera (Olympus) or a BX43 microscope (Olympus) equipped with a WRAYCAM-NOA630 camera (Wraymer).

For Gallyas–Braak silver staining, brain sections were mounted on APS-coated glass slides (Matsunami Glass Industry, Ltd) and air dried. Delipidation was achieved with increasing concentrations of ethanol and xylene, followed by several washes in distilled water. Silver staining was carried out according to the Gallyas–Braak method.^28–30

For fluorescent immunohistochemistry, free floating sections were incubated for 10 min with 0.1% Sudan black/70% ethanol solution and then incubated for 24 h with RD3 (1:500) and anti-4R tau antibody (1:500). Antibody labelling was performed as previously described.²³ Photographs were taken using a BZ-8000 (Keyence).

Dephosphorylation of tau

The sarkosyl-soluble fraction from a Tau 3R/4R mouse was reacted overnight with alkaline phosphatase in FastAP buffer (Thermo Scientific) at 37°C. Samples were then added to SDS-PAGE sample buffer and heated for 5 min at 100°C for immunoblotting.

Immunoblotting analyses

Brain extracts from the sarkosyl-insoluble-fraction-injected mice were boiled for 5 min with SDS-PAGE sample buffer (60 mM Tris-HCl, pH 6.8, containing 2% SDS, 10% glycerol, 0.025% bromophenol blue and 5% mercaptoethanol) and loaded onto a 5–20% acrylamide minigel (FUJI FILM Wako Pure Chemical Co.) or a 10% TGX FastCast minigel (Bio-Rad). Loaded samples were electrophoresed for 75 min(5-20% acrylamide minigel) or 60 min (10% minigel) at 200 V with molecular weight markers (Bio-Rad). Electrophoresed proteins were transferred onto a polyvinylidene difluoride membrane (Merck Millipore) and subjected to 2.5 A for 10 min using a Trans-Blot Turbo system (Bio-Rad). The printed membranes were blocked with 3% gelatin or PVDF Blocking Reagent for Can Get Signal (TOYOBO) for 15 min and then incubated in a primary antibody solution [T46, RD3, RD4 (Merck Millipore), anti-4R tau, pS396 or AT8] overnight at room temperature. Following incubation with secondary anti-mouse or anti-rabbit antibody (1:50 000, Bio-Rad), immunoreactivity was detected as previously described.³¹

Statistical analysis of the proportion of tau isoforms

Continuous variables were expressed as median and interquartile range data. The statistical significance of differences in mean values between two populations was assessed with the Brunner–Munzel test. P < 0.05 was considered significant.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article and its Supplementary material.

Results

Development of mice that endogenously express both three- and four-repeat tau

Using the CRISPR–Cas9 system to perform genome editing between the exon-10 and intron-10 boundary of the Mapt gene (Fig. 1A), we developed a novel mouse exhibiting tau expression patterns similar to those of humans (Fig. 1C). We obtained 26 litters after microinjection of sgRNA and Cas9 mRNA, and assays revealed that genome editing had occurred in 18 (69.2%) (data not shown). We then selected two lines (numbered 2 and 13), which were easily distinguishable by PCR, for use in genetic analyses (Fig. 1B). DNA sequencing revealed that non-homologous end-joining occurred after genome cleavage by CRISPR–Cas9, with a 108 bp deletion in Line 2 and a 211 bp insertion/4 bp deletion in Line 13 (Supplementary Fig. 1) in addition to the wild-type Mapt gene. This would eliminate the 5ʹ-splice site of intron 10 and prevent exon 10 from undergoing selective mRNA splicing. The edited genome produces exon 9/exon 11 mRNA, which is translated to produce 3R tau (Fig. 1A and Supplementary Fig. 1). On the other hand, the genome on the unaffected side of the editing produces exon 9 to exon 11 mRNAs that produce 4R tau, which is expressed in wild-type mice (Fig. 1A). The production of both mRNAs in mouse Lines 2 and 13 may have resulted in the endogenous expression of both 3R and 4R tau proteins (Fig. 1C). This mouse expressed only 3R tau immediately after birth (post-natal Day 0: P0) but began to express 4R tau at P14 (Supplementary Fig. 2). After Day 14, both 3R and 4R tau continued to be expressed.

Proteins were extracted from the brains of adult Tau 3R/4R and wild-type mice, then dephosphorylated with alkaline phosphatase and immunoblotted using a total tau antibody (T46) to detect tau expression patterns. In the wild-type mice, three isoforms of 4R tau alone were detected, whereas in Lines 2 and 13, three isoforms of 3R tau were detected in addition to three isoforms of 4R tau. Thus, these novel Tau 3R/4R mice expressed a total of six endogenous tau isoforms just as is seen in human brains (Fig. 1C).

In an experiment using 3R tau-specific antibody (RD3) and 4R tau-specific antibody (RD4), only 4R tau was detected in the wild-type, whereas both 3R and 4R tau were detected in Lines 2 and 13 (Supplementary Fig. 3). This study was mainly conducted using Line 13 mice, unless specified.

Next, the expression ratios of each isoform were compared. When the total tau in a Line 13 mouse was set to 100%, the expression ratio of each isoform was as follows: 24.25 ± 1.04% for 0N3R, 24.54 ± 0.76% for 0N4R, 10.70 ± 0.40% for 1N3R, 10.58 ± 0.86% for 1N4R, 15.45 ± 0.86% for 2N3R and 2N4R for 14.47 ± 0.93% (Fig. 1D). There was no statistically significant difference between 0N3R and 0N4R, 1N3R and 1N4R, and 2N3R and 2N4R (Fig. 1D), respectively. The ratio of 3R tau to 4R tau in Line 13 was found to be 1:1.

Injection of the sarkosyl-insoluble fraction from tauopathy patients induces the accumulation of tau in mice

To develop a novel model of propagation of tau pathology, sarkosyl-insoluble fractions (5 μl/mouse) containing abnormal tau derived from tauopathy patients with Alzheimer’s disease (AD), corticobasal degeneration (CBD) or Pick’s disease (PiD) were injected into the right striatum of Tau 3R/4R mice (bregma to anterior–posterior = +0.2 mm, medial–lateral = +2.0 mm, dorsal–ventral = −2.6 mm) since the striatum has abundant neural connections with other brain areas.

Three, 6 and 9 months after the intracerebral injection of AD and CBD, or 8 and 12 months after the intracerebral injection of PiD, the brain was removed and fixed with 4% paraformaldehyde. Sliced sections were stained with anti-phosphorylated tau antibody (AT8). Accumulation of phosphorylated tau was observed in the striatum of the AD, CBD or PiD-injected mice (Figs 2A, 3A and 4A) at the site of injection.

Tau accumulation and propagation

In the sarkosyl-insoluble fraction of AD-injected or CBD-injected mice, we observed the accumulation of AT-8-positive tau mainly in neurites of the striatum and corpus callosum of the ipsilateral side at 3 months post-injection. At 6 months, it was localized in the cell bodies, and at 9 months, AT-8 staining was denser (Figs 2A and 3A). In addition to the striatum and corpus callosum, AT-8 positive staining was widely distributed at 9 months and included the amygdala, stria terminalis, substantia nigra, thalamus and several cortical areas (Figs 2B and 3B). The temporal and spatial distribution in the AD-injected mouse brain is shown in Supplementary Fig. 4. We performed fluorescent staining using 3R- and 4R-tau-specific antibodies to examine AD-injected mouse brains (Fig. 2C). The merged photograph indicated that almost all of the 3R tau and 4R tau were co-localized.

In brains of human CBD patients, tau also accumulates in astrocytes where it assumes a characteristic form termed an astrocytic plaque. Astrocytic plaque-like tau structures were also observed in CBD-injected Tau 3R/4R mice (Fig. 3C). These results suggested that injection of the sarkosyl-insoluble fraction of CBD induced tau accumulation in neurons and astrocytes in this model.

In the sarkosyl-insoluble fraction of PiD-injected mice, the accumulation of AT-8-positive tau was also mainly in neurites of the striatum and corpus callosum at 8 months post-injection (Fig. 4A). At 12 months, it was localized in the cell bodies. In addition to the striatum and corpus callosum, AT-8-positive staining was widely distributed at 12 months and included the cingulate gyrus and stria terminalis (Fig. 4B). Pick-body-like round inclusions were found in PiD-injected mouse brains (Fig. 4C). AT8 or antimouse tau-specific antibody staining showed relatively small, round inclusions in the hypothalamus and cortex (Fig. 4C). Low magnification images are shown in Supplementary Fig. 6A and B. No AT8-positive staining was observed at 3 months (Supplementary Fig. 6C and D) or 6 months (Supplementary Fig. 6E and F) post-injection. Thus, it was thought to be the first model ever to reproduce Pick body-like inclusions.

It has been reported that Gallyas–Braak staining is negative in Pick's disease,³² and we determined whether this finding could be reproduced. Gallyas–Braak staining of AD- or CBD-injected mouse brains was positive but that of PiD-injected brains was negative (Fig. 4D), suggesting that the tau pathology induced by injection of the sarkosyl-insoluble fraction recapitulated human tau pathology in Tau 3R/4R mice brains.

It has also been reported that serine 262 (S262) of tau is not phosphorylated in PiD.³² Immunohistochemical staining of the brains of AD-, CBD- and PiD-injected mice with 12E8 (pS262) antibody showed that S262 was phosphorylated in AD- and CBD-injected brains, but not in PiD-injected brains (Fig. 4E). Similar to the results of Gallyas–Braak staining, the pattern of tau accumulation in this model reproduced that seen in PiD. These findings suggested that injection of the sarkosyl-insoluble fraction into the striatum of Tau 3R/4R mice recapitulated the pathology of the human tauopathies.

Seed-dependent tau pathology propagation and accumulation of endogenous mouse tau

Next, we investigated whether abnormal tau accumulation occurred in a seed-dependent manner. Sliced brain samples were stained with 3R- and 4R-tau-specific antibodies. AD (3R+4R tauopathy)-injected mice exhibited both 3R and 4R tau accumulation. CBD (4R tauopathy)-injected mice showed positive staining only with the 4R-tau-specific antibody. PiD (3R tauopathy)-injected mice were only positive with 3R tau staining (Fig. 5A). These results indicate that this new propagation model using our Tau 3R/4R mice recapitulated seed-dependent, isoform-specific abnormal tau accumulation.

The brains of AD, CBD and PiD-injected mice were stained with a human tau-specific antibody (HT7) and no abnormal human tau accumulation was detected, whereas the brains stained with a mouse tau-specific antibody showed abnormal tau accumulation (Fig. 5B). These results indicate that the injected human tau had degraded and was under the detection limit, and that the endogenous mouse tau was converted to the abnormal form and had accumulated. AT8 staining images of the same region are shown in Supplementary Fig. 5 for comparison. Injection of the sarkosyl-insoluble fraction induced tau accumulation with crossing of the species barrier.

Intracerebral injection of recombinant tau seeds also induced tau pathology in Tau 3R/4R mice (Supplementary Fig. 7). AT8-positive tau pathology was observed in the striatum after injection with dextran-sulphate (DS)-induced 3R tau seeds and 4R tau seeds (Supplementary Fig. 7A and B). Fluorescent double staining with RD3 and anti-4R antibodies showed that the injection of DS-induced 3R tau seeds caused both 3R and 4R tau accumulation, whereas that of DS-induced 4R tau seeds induced only 4R tau aggregation (Supplementary Fig. 7C).

Biochemical analysis

Eight months after intracerebral injection of the sarkosyl-insoluble fraction, biochemical analyses were performed on the ipsilateral brain with total tau antibody (Fig. 6A). Doublet bands were detected in AD-injected mice, with the upper and lower bands representing 4R tau and 3R tau, respectively. A single band representing 4R tau was seen in CBD-injected mice. An accumulation of a truncated tau of ∼37 kDa, characteristic of CBD³³, was also observed. A very faint band representing 3R tau was observed in PiD-injected mice. There was no tau detected in Huntington disease or in age-matched control injected mice (Fig. 6A). Accumulated tau was also detected by two anti-phosphorylated antibodies (AT8 and pS396), indicating that the abnormal tau was phosphorylated (Fig. 6B and C).

Immunoblotting with 3R- and 4R-specific tau antibodies detected bands in AD and PiD for 3R-specific antibody (Fig. 6D) and in AD and CBD, for 4R-specific antibody (Fig. 6E). These results indicated that seed-dependent accumulation of tau was induced in this model.

Injected human tau converts mouse tau into an abnormal form by crossing the species barrier

To confirm that there was no injected tau remaining, the following experiment was performed. After injection of the Alzheimer's disease sarkosyl-insoluble fraction into the striatum, the mice were decerebrated at the following time points: immediately after injection, at 7 days, at 14 days, or at 8 months. The left and right (ipsilateral side) brains were divided, and the sarkosyl-insoluble fraction was collected for immunoblotting. Immunoblotting with the pS396 antibody (Fig. 7, top) showed an AD tau-derived band in the right brain of the mice immediately after the injection (Fig. 7). Seven days later, the injected tau was almost completely degraded, and by 14 days, the injected tau had degraded to below the detection limit. At 8 months after injection, the abnormal tau was predominantly accumulated in the ipsilateral side of the brain. Immunoblots of mouse-specific tau antibodies in bottom panel of Fig. 7 show that the abnormal tau was newly accumulated and consisted of endogenous tau.

Propagation of tau pathology through the neural network

With reference to AD-inoculated mice with the most widely spread tau pathology, the relationship between the propagation of tau pathology and neural connectivity is shown in Fig. 8. The injection site was the striatum and brain regions where the abnormal tau had accumulated and propagated ultimately surrounded it. The brain regions that confirmed propagation of tau pathology showed neural connectivity, indicating that the spread or propagation of this pathology is related to neural connectivity rather than to proximity.

Discussion

In this study, we used the CRISPR–Cas9 system to perform genome editing between the exon-10 and intron-10 boundary of the Mapt gene. Two mouse lines (Lines 2 and 13) were examined in this study. Non-homologous end-joining occurred after genome cleavage by CRISPR–Cas9, with a 108 bp deletion in Line 2 and a 211 bp insertion and 4 bp deletion in Line 13 (Fig. 1A and Supplementary Fig. 1). This would have eliminated the 5ʹ-splice site of intron 10 and prevented exon 10 from undergoing selective mRNA splicing. The edited genome produces exon 9/exon 11 mRNA, which is translated to produce 3R tau. On the other hand, the genome on the unaffected side of the editing produces exon 9 to exon 11 mRNAs that produce 4R tau, which is expressed in wild-type mice. The production of both mRNAs in mice of Lines 2 and 13 may have resulted in the endogenous expression of both 3R and 4R tau proteins (Fig. 1C). A previous study on tau injection using brain homogenates of tauopathy patients induced the accumulation of 4R tau, but not 3R tau.⁴ In addition, it failed to show an accumulation of 3R tau when mice were injected with homogenates of a human Pick's disease brain. In the present study, we injected mice expressing both endogenous 3R and 4R tau with the sarkosyl-insoluble fractions of Alzheimer’s diseasepatient brains and found that they exhibited both 3R and 4R tau accumulation on immunohistochemical (Fig. 2A–C) and biochemical analyses (Fig. 6D and E). Furthermore, when we injected mice with the sarkosyl-insoluble fractions of Pick's disease brains, in which only 3R tau had accumulated (Figs 4A, B, 6D and E), or CBD brains, in which only 4R tau had accumulated, we found accumulation of only 3R tau and 4R tau, respectively (Figs 3A, B, 6D and E), with immunohistochemical and biochemical analyses. These results indicated that this injected model might be useful for demonstrating seed-dependent, isoform-specific tau aggregation. Considering that most tau injection models reported so far could only induce the accumulation of 4R tau^{2,4–6,9–12,16,18}, the present model seems to be the one best suited to reproduce the propagation of tauopathies.

The striatum of Tau 3R/4R mice was injected with a sarkosyl-insoluble fraction from an Alzheimer’s disease patient and the subsequent abnormal tau accumulation was studied in detail (Fig. 2A). At 3 months post-intracerebral injection, a slight tau accumulation was observed in neurites. At 6 months after injection, there was a moderate increase in the accumulation of AT-8-positive tau in the cell bodies, and at 9 months, the staining was denser (Fig. 2A). Furthermore, tau accumulation in AD-injected mice spread to the amygdala, substantia nigra, thalamus and several cortices (Fig. 2B), and the spread in the CBD-injected mice was similar (Fig. 3A and B). Propagation of tau pathology in the PiD-injected mice was not more widespread than in the AD- or CBD-injected animals (Fig. 4A and B). These results were probably due to the difference in the amount of tau contained in the sarkosyl-insoluble fraction. These results indicated that propagation of tau pathology might have occurred through neural connectivity and not proximity (Fig. 8).

It is known that the properties of tau accumulation are different in each tauopathy, and we investigated whether these differences can be inherited in this mouse model. We examined whether the pathological character of each tauopathy could be reproduced in mice injected with the sarkosyl-insoluble fraction. In AD-injected Tau 3R/4R mouse brain, fluorescent double staining with RD3 and anti-4R antibodies revealed that most 3R and 4R tau was co-localized (Fig. 2C). This pathology is thought to be similar to that observed in the human Alzheimer’s diseasebrain.

Pick-body-like round inclusions were observed in PiD-injected Tau 3R/4R mouse brain (Fig. 4C) and these inclusions were found in the cortex and hypothalamus, both of which have neural connections to and from the striatum. However, these inclusions were somewhat smaller than human Pick bodies. The reason for this is that the time that elapsed after injection of the PiD-insoluble fraction was shorter than the time required for Pick body formation in human Pick's disease. It was speculated that this might be a precursor state of Pick body generation.

Pick's disease is known to be undetectable on Gallyas–Braak silver staining,³² and we investigated whether this phenomenon could be reproduced in this model. AD and CBD-injected Tau 3R/4R mouse brains were positive on Gallyas–Braak silver staining; however, PiD-injected Tau 3R/4R mouse brains were negative (Fig. 4D). Next, it is also known that S262 of tau is not phosphorylated in Pick's disease.³² Similar to the results from the Gallyas–Braak staining, S262 of tau was phosphorylated in AD- and CBD-injected Tau 3R/4R mouse brains, but not in PiD-injected Tau 3R/4R mouse brains (Fig. 4E). These results indicate that the properties of tau that accumulate in human Pick's disease are retained and propagated. This PiD injection model is relatively limited but it can generate Pick body-like round inclusions and is also able to reproduce negative Gallyas–Braak and negative pS262 staining, which are characteristic of Pick's disease in humans. Thus, it represents the first valuable model for researching this disease.

Moreover, astrocytic plaque-like structures were found in the cortex of CBD-injected Tau 3R/4R mouse brain (Fig. 3C). As previously reported, the appearance of astrocytic plaques was observed when CBD-derived brain extracts were injected into Alz17 mice.⁴ By using the Tau 3R/4R mouse, we might be able to construct a model that can reproduce the characteristics of specific tauopathies.

Sarkosyl-insoluble fractions were extracted from mice injected with tauopathy brains and subjected to biochemical analysis. Doublet bands were detected in the AD-injected brain with total tau antibody (Fig. 6A). The upper and lower bands were considered to be 4R tau and 3R tau, respectively, whereas immunoblotting of the sarkosyl-insoluble fraction of human Alzheimer’s disease brain with total tau antibody showed triplet band patterns of 68, 64 and 60 kDa.²³ Bands were also detected in CBD-injected Tau 3R/4R mouse brains at the full-length tau position and at around 37 kDa, the C-terminal fragment of tau (Fig. 6A). This 37 kDa band has already been reported to be a biochemical characteristic of CBD.³³ Immunoblotting of the sarkosyl-insoluble fraction of human corticobasal degeneration brain with total tau antibody revealed doublet band patterns of 68 and 64 kDa.²³ PiD-injected Tau 3R/4R mouse brains showed a faint single band at the position of the full-length tau (Fig. 6A), while immunoblotting of the sarkosyl-insoluble fraction of human Pick's disease brain with total tau antibody showed doublet band patterns of 64 and 60 kDa. The difference in tau accumulation between humans and mice may be due to the difference in the expression ratio of each isoform in Tau 3R/4R mice (Fig. 1C and D). The accumulated tau was confirmed to be phosphorylated using AT8 (Fig. 6B) and pS396 (Fig. 6C) antibodies. Immunoblotting of isoform-specific antibodies revealed that RD3 detected abnormal tau in AD and PiD-injected Tau 3R/4R mouse brain (Fig. 6D) and that RD4 detected it in AD and CBD-injected brains (Fig. 6E). These results revealed that tau accumulation occurred in a seed-dependent manner after intracerebral injection of the sarkosyl-insoluble fraction of the tauopathies.

We examined the process by which mouse tau accumulates after injection with human tau derived from an Alzheimer’s disease patient. Injected human Alzheimer’s disease-derived tau was detected on the ipsilateral side of the brain immediately after the injection, but it became degraded with time and fell below the detection limit at 14 days after injection. Eight months later, enhanced accumulation of abnormal tau was confirmed (Fig. 7). The abnormal accumulation of tau in these mice was found to consist of endogenous tau, and it was inferred that the injected human tau had crossed the species barrier to convert mouse tau into the abnormal form that then accumulated (Figs 5B and 7). This experiment confirms that abnormal tau has ‘prion-like’ properties. Since the abnormal tau injected into the striatum propagated mainly in areas of neural connection to and from the striatum (Fig. 8), it is likely that it propagated in either an anterograde or retrograde manner within the neural circuitry. In the future, it will be necessary to examine whether the area of transmission of abnormal tau is altered by changing the inoculation site, whether behavioural abnormalities appear or whether neuronal cell death is induced. We believe that this Tau 3R/4R mouse model can be used to screen for agents that could inhibit the spread of tau.

During the preparation of this paper, a mouse model was developed that expresses the six human tau isoforms (6hTau mice).³⁴ Their Mapt knockout transgenic mouse harbours human tau, resulting in the expression of the six human tau isoforms. These investigators reported that the 0N3R and 0N4R human tau isoforms were most abundant in 6hTau mice, which is different from human brains in which 1N3R and 1N4R tau isoforms predominate.³⁵ Our Tau 3R/4R mice also abundantly expressed 0N3R and 0N4R mouse tau isoforms. These two studies revealed that the RNA splicing mechanism involved in the formation of 0N-2N on the N-terminal side of tau is different between humans and mice.

Recently, we reported that DS-induced assemblies of recombinant tau cause endogenous tau aggregation and propagation in wild-type mice.²⁷ We performed an experiment in which we injected preformed DS-induced 3R and 4R tau fibrils into the striatum of Tau 3R/4R mice (Supplementary Fig. 7). In mice injected with DS-3R tau, the accumulation of 3R tau was predominant, but some co-localization with 4R tau was observed. In contrast, only the accumulation of 4R tau was observed in DS-4R tau-injected Tau 3R/4R mice. The difference in results of experiments with the sarkosyl-insoluble fraction from a tauopathy brain and those using DS-induced fibrils requires further investigation. With a combination of DS-induced tau filaments and this Tau 3R/4R mouse model, it may be possible to build a comprehensive model for the propagation of various tauopathies, thereby overcoming ethical issues that arise when using a human post-mortem brain.

Acknowledgements

The authors would like to thank Mrs Catherine Campbell for editing of the manuscript.

Funding

This research was partially supported by the Japan Society for the Promotion of Science, a Grant-in-Aid for Scientific Research (C) (JSPS KAKENHI grant number K1807618) and by a Mitsui Sumitomo Insurance Welfare Foundation Research Grant 2016 to M.Ho.

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.

References

Abbreviations

 
• AD/CBD/PiD

  sarkosyl-insoluble fractions containing abnormal tau derived from tauopathy patients with Alzheimer's disease/corticobasal degeneration/Pick's disease

 
• DS

  dextran sulphate

Author notes