Increase in Tau Pathology in P290S Mapt Knock-In Mice Crossed with AppNL-G-F Mice

Abstract Alzheimer’s Disease (AD) is characterized by the pathologic assembly of amyloid β (Aβ) peptide, which deposits into extracellular plaques, and tau, which accumulates in intraneuronal inclusions. To investigate the link between Aβ and tau pathologies, experimental models featuring both pathologies are needed. We developed a mouse model featuring both tau and Aβ pathologies by knocking the P290S mutation into murine Mapt and crossing these MaptP290S knock-in (KI) mice with the AppNL-G-F KI line. MaptP290S KI mice developed a small number of tau inclusions, which increased with age. The amount of tau pathology was significantly larger in AppNL-G-FxMaptP290S KI mice from 18 months of age onward. Tau pathology was higher in limbic areas, including hippocampus, amygdala, and piriform/entorhinal cortex. We also observed AT100-positive and Gallyas-Braak-silver-positive dystrophic neurites containing assembled filamentous tau, as visualized by in situ electron microscopy. Using a cell-based tau seeding assay, we showed that Sarkosyl-insoluble brain extracts from both 18-month-old MaptP290S KI and AppNL-G-FxMaptP290S KI mice were seed competent, with brain extracts from double-KI mice seeding significantly more than those from the MaptP290S KI mice. Finally, we showed that AppNL-G-FxMaptP290S KI mice had neurodegeneration in the piriform cortex from 18 months of age. We suggest that AppNL-G-FxMaptP290S KI mice provide a good model for studying the interactions of aggregation-prone tau, Aβ, neuritic plaques, neurodegeneration, and aging.


Introduction
Alzheimer's disease (AD) is defined by the simultaneous presence of two different filamentous amyloid inclusions: abundant extracellular deposits of amyloid b (Ab ) and abundant intraneuronal inclusions of tau. Genetic evidence has indicated that Ab is key to the pathogenesis of Alzheimer's disease (Hardy and Higgins, 1992;Hardy and Selkoe, 2002). Multiplications of the gene encoding the Ab precursor protein (APP), as well as mutations in APP and the presenilin genes (PSEN1 and PSEN2), cause familial Alzheimer's disease with abundant Ab deposits and tau inclusions. It follows that both proteopathies are probably linked.
To investigate this link, one needs experimental systems that develop both pathologies. Although multiple transgenic mouse models with abundant Ab deposits have been produced, filamentous tau inclusions did not form (Radde et al., 2006;Saito et al., 2014). Conversely, overexpression of human mutant tau can result in the development of filamentous tau inclusions and neurodegeneration, but in the absence of Ab deposits (Allen et al., 2002). Crossing mice that overexpress human mutant APP with mice overexpressing human mutant tau has been shown to result in more assembled tau, suggesting that Ab exacerbated tau assembly (Lewis et al., 2001;Chen et al., 2016). Moreover, in 3xTg-AD mice, which express three mutations in human APP, tau, and PSEN1, a reduction in Ab deposits by immunotherapy resulted in decreased tau inclusions (Oddo et al., 2004).
While these models show a potentially significant interaction between tau and Ab pathologies, they can suffer from artifactual phenotypes as a result of uncontrolled expression and/or random genomic insertion of the transgenes. Thus, the overexpression of APP can result in elevated levels of fragments that are not overexpressed in diseased human brains. Moreover, insertion of the transgenes may lead to the interruption of genes that are vital for nerve cell function (Saito et al., 2016;Gamache et al., 2019).
Knock-in (KI) models can provide a way around these drawbacks, since expression is driven by the natural promoters. In recent years, mouse lines were produced that deposit large amounts of human Ab , without requiring overexpression of APP (Saito et al., 2014). Knock-in models, however, have so far not led to the formation of abundant filamentous tau deposits. Only models of the overexpression of multiple human wild-type tau isoforms or single-mutant human tau isoforms have given rise to abundant tau filaments and neurodegeneration (Ishihara et al., 1999;Allen et al., 2002;Sahara et al., 2002). A murine knock-in line that expresses all six humanized brain tau isoforms did not develop filamentous tau inclusions on crossing with the App NL-G-F KI line, which expresses humanized Ab , together with the Swedish (KM670/671NL), Arctic (E693G), and Beyreuther/ Iberian (I716V) mutations (Saito et al., 2019). Similarly, a mouse line with a P301L tau knockin, did not exhibit filamentous tau inclusions (Gilley et al., 2012).
Here we describe the production and characterization of a mouse model of tauopathy, where mutation P290S was knocked into exon 10 of murine Mapt, the tau gene. Murine P290S tau is equivalent to human P301S tau that causes dominantly inherited frontotemporal dementia and results in the formation of abundant filamentous tau inclusions and neurodegeneration when overexpressed in transgenic mice (Bugiani et al., 1999;Allen et al., 2002;Yoshiyama et al., 2007). Mapt P290S knock-in mice developed small numbers of tau inclusions that increased with age. To develop a model for the relationships between Ab and tau pathologies, we crossbred Mapt P290S knock-in mice with line App NL-G-F . These mice developed both Ab and filamentous tau pathologies, and exhibited a significant and age-related increase in the number of tau inclusions when compared with Mapt P290S knockin mice.

Mouse lines
Animal experiments were conducted in accordance with the UK Animals (Scientific Procedures) Act of 1986, with local ethical approval (MRC Laboratory of Molecular Biology Animal Welfare and Ethical Review Body). To generate the Mapt P290S knock-in line, a targeting construct was designed to insert S290 into exon 10 of Mapt, the murine tau gene, by homologous recombination (Fig. 1). The targeting construct was transfected into calcium channel blocker embryonic stem cells (129 S/Sv) and positive clones were identified by Southern blotting. They were injected into C57BL/6 blastocysts, and the resulting chimeras were crossed with C57BL/6 mice, to establish germline transmission. The progeny was genotyped by PCR (oligos for wild-type allele, CAGCAAAGTAGGGAG AGCAAC and CAGAGATGAGGGAAGAGGTGTC; oligos for knock-in allele, CAGCAAAGTAGGGAGAGCAAC and TTCGCCAATGACAAGACGC). The Neo-loxP cassette was removed by crossing with Stella-Cre transgenic mice that express Cre recombinase (Liu et al., 2011). The presence of murine tau with the P290S mutation (equivalent to human P301S tau) was confirmed by DNA sequencing. Following backcrossing, analysis using single nucleotide polymorphisms established that the line was on a 99.9% C57BL/6 background (The Jackson Laboratory; MAXBAX 384 SNP panel, Charles River). App NL-G-F knock-in mice express humanized Ab and carry the Swedish double mutation (KM670/671NL) in APP, the Arctic mutation in Ab (E22G) and the Beyreuther/Iberian mutation (I716F) in APP (Saito et al., 2014). The strain was generated using C57BL/6N ES cells. The line was backcrossed for eight generations to establish the C57BL/6J-based knock-in strain, which is now 99% C57BL/6J background (Takashi Saito, personal communication). To generate the App NL-G-F xMapt P290S knockin line, mice from the Mapt P290S knock-in line were crossbred with mice from the APP NL-G-F knock-in line. All lines were made homozygous by genotyping of the progeny by PCR.

Tissue preparation
Tissues were collected from mice 3, 6, 12, 18, and 22-24 months of age (n = 4-8/time point). The sexes of the species used in each experiment have been designated as male (M) or female (F) in the Materials and Methods section and figure legends below. They were perfused intracardially with 4% paraformaldehyde in 0.1 M PBS, pH 7.4. Brains were dissected and postfixed overnight at 4°C, followed by cryoprotection in 20% sucrose in PBS for at least 24 h. Coronal sections (50 mm) were cut on a freezing microtome (model SM2400, Leica Microsystems) and kept at 4°C in PBS with 0.1% sodium azide.

Immunostaining
All steps were performed on a rocker at room temperature, with the exception of primary antibody incubation, which was performed at 4°C. Wash steps were carried out in PBS containing 0.1% Triton X (PBST), three times for 10 min each. For immunohistochemistry, endogenous peroxidase activity was quenched by incubation with 0.3% hydrogen peroxide in water for 30 min, followed by washing. Sections were incubated in blocking buffer consisting of either 10% normal goat serum or horse serum in PBST for 1 h, followed by an overnight incubation with primary antibodies diluted in blocking buffer. The next day, sections were washed and incubated in biotinylated horse anti-mouse IgG or goat anti-rabbit IgG secondary antibody (1:200; Vector Laboratories) for 1 h, followed by washing. Sections were incubated with the VECTASTAIN ELITE ABC Kit (Vector Laboratories) for 1 h, followed by washing, and developed with a diaminobenzidine peroxidase substrate kit (Vector Laboratories). Immunostained sections were then washed in water, mounted onto Superfrost plus slides (Thermo Fisher Scientific), air dried, and coverslipped with Pertex mounting medium (CellPath).

Gallyas-Braak silver staining
Eight sections (1:24 series) per brain were mounted onto slides and left to air dry. Reagents were prepared according to the study of Braak and Del Tredici (2015). Slides with mounted sections were pretreated with 3% periodic acid for 10 min, followed by a 3 min wash in distilled water. They were then transferred to alkaline solution of silver iodide complexes for 2 min. Slides were immersed in developer for 30-50 min. Following this, they were immersed in the following: 0.5% acetic acid (3 min), distilled water (3 min), gold chloride (1 min), distilled water (3 min), 0.5% sodium thiosulfate (5 min), and distilled water (3 min). Silver staining was performed on brains sections from mice [at 3 and 6 months, n = 3 (2 M, 1 F); 12 and 18 months, n = 8 (4 M, 4 F); and 22-24 months, n = 4 (2 M, 2 F)].
Extraction of Sarkosyl-insoluble tau from mouse brain homogenates Fresh frozen whole mouse brains were weighed and homogenized in 10% (w/v) extraction buffer (20 mM Tris-HCl, pH 7.4, 800 mM NaCl, 5 mM EDTA, 15% sucrose, 1% Sarkosyl, and one tablet/10 ml complete protease/phosphatase inhibitor cocktail; Pierce), and total protein was normalized using the Pierce BCA protein assay kit (Thermo Fisher Scientific). Sarkosyl-insoluble tau was extracted by differential centrifugation as described previously (Goedert et al., 1992). Briefly, tissue homogenates were incubated for 1 h in 1% Sarkosyl and centrifuged at 18,000 Â g for 20 min at room temperature. Supernatants were filtered through a 45 mm cell strainer, then spun for 1 h at room temperature at 45,000 rpm using a TLA-55 rotor in an Optima Max XP ultracentrifuge. Pellets were resuspended in 150 mM NaCl, and 50 mM Tris HCl, at pH 7.4, and spun again at 45,000 rpm. The final pellets were resuspended in 30 ml of 150 mM NaCl, 50 mM Tris HCl.

Seeding assay
HEK293 cells stably expressing human 0N4R P301S tau-venus (McEwan et al., 2017) were maintained in DMEM plus GlutaMAX (catalog #31966-021, Thermo Fisher Scientific) with 10% fetal calf serum (catalog #10270-106, Thermo Fisher Scientific) and 1% penicillin/streptomycin. Black 96-well plates (catalog #3603, Costar) were pretreated with poly-D-lysine, and HEK293 P301S tau-venus cells were plated at a density of 20,000 cells/well and allowed to adhere overnight. Cells were then washed with PBS, and 50 ml of prepared tau filaments extracts, or seeds, diluted 1:50 in OptiMEM (catalog #3198502, Thermo Fisher Scientific) with 1:50 lipofectamine 2000 (catalog #11668027, Thermo Fisher Scientific), was added to each well. Cells were incubated with seeds for 1 h before adding 100 ml of complete-DMEM to block further entry of lipofectamine-seed complexes. After 48 h, cells were rinsed with PBS and fixed with 4% paraformaldehyde for 10 min at room temperature. For immunohistochemistry, cells were incubated in primary antibody overnight at 4°C, followed by washing and incubation in Alexa Fluor 647-conjugated goat antimouse IgG secondary antibody (1:500; Thermo Fisher Scientific) for 1 h at room temperature. Cells were then incubated with 1 mg/ml Hoechst 33342 diluted in 1Â PBS for 10 min. Images were taken using an inverted fluorescence microscope (model HCA, Nikon) under a 10Â lens. Image analysis was performed using NIS-Elements AR. Quantitation represents the total area of the venus puncta signal per nucleus. Immunofluorescence images were taken on an inverted confocal microscope (model 780, Zeiss) under a 20Â lens.
Extraction of Sarkosyl-insoluble tau from seeded cells HEK293 P301S tau-venus cells were plated at 500,000 cells/well in six-well plates and allowed to adhere overnight. Sarkosyl-insoluble tau filament extracts were prepared at 1:50 in 500 ml of OptiMEM containing 1:50 lipofectamine 2000 (catalog #11668027, Thermo Fisher Scientific) and added to 500 ml of cells in DMEM. After 1 h, 1 ml of DMEM was added to stop seeding. After 48 h, cells were washed gently with PBS before being resuspended by forceful pipetting in 2 ml of PBS. Resuspended cells were centrifuged at 400 Â g for 5 min, and the resultant pellets were resuspended in 500 ml of extraction buffer (10 mM Tris, pH 7.4, 0.8 M NaCl, 1 mM EGTA, at pH 7.2, and 5 mM EDTA, at pH 7.4, and 10% sucrose) and lysed on ice for 30 min. Cell homogenates were centrifuged at 18,000 relative centrifugal force for 20 min at 4°C. Protein concentrations of the resultant supernatants were determined by the Pierce BCA Protein Assay Kit (catalog #23225, Thermo Fisher Scientific) and normalized to the lowest concentration. Supernatants were then spun at 45,000 rpm for 1 h at 4°C. The pellets were resuspended in 200 ml of 50 mM Tris HCl, at pH 7.4, and spun again as before. The final pellets were resuspended in 15 ml of 50 mM Tris, pH 7.4, and stored at 4°C until immunoblot analysis.

Electron microscopy
Extracted tau filaments were diluted 10-fold to 20-fold and deposited on glow-discharged 400-mesh formvar/ carbon film-coated copper grids (model CF400-Cu, EM Sciences) for 40 s. Grids were blocked for 10 min in 0.1% cold water fish skin gelatin in PBS before a 1 h incubation with primary antibodies diluted in blocking buffer. Grids were washed with blocking buffer and incubated with gold-conjugated anti-mouse IgG (catalog #G7652, Sigma-Aldrich) or anti-rabbit (catalog #G7402, Sigma-Aldrich) secondary antibody for 1 h. Following washing, the grids were stained with 4 ml of 2% uranyl acetate for 90 s. Images were acquired at 11,000Â and 15,000Â, with a defocus value of À1.4 mm using a transmission electron microscope (Tecnai G2 Spirit) at 120 kV. For in situ electron microscopy (EM), perfusion-fixed brains (4% paraformaldehyde) were postfixed in 4% paraformaldehyde/0.1% glutaraldehyde for at least 24 h. Sections (50-200 mm) were cut on an Oxford vibratome and collected in PBS. Gallyas-Braak silver staining was performed on vibratome-cut sections. Following osmication (30 min in 1% OsO 4 in 100 mM phosphate buffer), the sections were stained for 15 min in 0.1% uranyl acetate in sodium acetate buffer at 4°C, dehydrated, cleared in propylene oxide, and embedded in Araldite resin between two sheets of Melanex (Imperial Chemical Industries). Semithin (1 mm) sections were cut with glass knives and stained with toluidine blue adjacent to thin sections (70 nm) that were cut with a diamond knife on a Reichert Ultracut ultramicrotome. Sections were collected on copper mesh grids, counterstained with lead citrate and viewed in an electron microscope (model 1400, Jeol). Photomicrographs were taken with a Gatan Rio Camera.

Stereology
Four sections (1:24 series) per brain were used for cell counting with SteroInvestigator 11 (MBF Bioscience). Section thickness was determined using Investigator software. For each section, a 200 Â 400 mm region of piriform cortex was traced under a 5Â objective, starting from the ventral side. NeuN-positive cells and their nuclei within the dissector volume were counted using the 100Â objective. The investigator was blind with respect to the nature of the groups. For wild-type mice at all ages, n = 3 M; for Mapt P290S KI mice at 12 and 18 months, n = 4 (2 M, 2 F), and at 22-24 months, n = 6 (3 M, 3 F); and for App NL-G-F xMapt P290S KI at 12 and 18 months, n = 4 (2 M, 2 F), and at 22-24 months, n = 5 (2 M, 3 F).

Statistics
Statistical analysis was performed using GraphPad Prism software. All the data are shown as the mean 6 SEM. Data were analyzed by one-way or two-way ANOVA, followed by Tukey's multiple-comparisons test.

Tau filaments in Sarkosyl-insoluble fraction from
We generated a knock-in mouse model of tau assembly (Mapt P290S KI) by incorporating a mutation equivalent to P301S, which causes an inherited form of frontotemporal dementia in humans (Bugiani et al., 1999), into exon 10 of Mapt ( Fig. 1; see Materials and Methods section). App NL-G-F KI mice were used as a model of Ab deposition. Mapt P290S KI mice were crossed with App NL-G-F KI mice, to generate a double-KI model (Mapt P290S KIxApp NL-G-F KI), with tau and Ab pathologies. Immunoblot analysis ( Fig. 2A) of brain homogenates from 3-, 6-, 12-, and 18-month-old mice from single-and double-KI lines showed no significant differences in expression levels of tau and APP between wild-type and KI mice. Sarkosyl-insoluble fractions from the brains of mice from the App NL-G-F KI, Mapt P290S KI, and App NL-G-F KIxMapt P290S KI lines were analyzed by immunoblotting. Only in the App NL-G-F KIxMapt P290S KI line did we detect Sarkosyl-insoluble tau. From 18 months of age, anti-tau antibodies T49, AT8, AT100, pS422, BR133, and BR134 detected a Sarkosyl-insoluble band of 55 kDa (Fig. 2B). Immunogold negative-stain electron microscopy showed abundant tau filaments in the Sarkosyl-insoluble fraction from the brains of mice from the App NL-G-F KIxMapt P290S KI line. Filaments were decorated by MT-1, AT8, AT100, and BR134 (Fig. 2C).
Research Article: New Research Figure 3. Characterization of brain tau pathology by immunohistochemistry and Gallyas-Braak silver staining. A, AT8 (top) and AT100 (bottom) immunoreactivity in Mapt P290S KI and App NL-G-F xMapt P290S KI mice at 24 months of age. AT8-or AT100-immunoreactive cells were quantified at 3, 6, 12, 18, and 22-24 months in 50 mm whole-brain coronal sections. Results shown are the mean 6 SEM. Significance between genotypes and adjacent time-points within a genotype is reported for two-way ANOVA with Tukey's post hoc analysis (*p , 0.05, **p , 0.01, ****p , 0.0001). Quantitation values of AT8 staining in Mapt P290S KI mice were as follows: 3 months, n = 6 (3 M, 3 F); 6 months, n = 6 (3 M, 3 F); 12 months, n = 8 (4 M, 4 F), 18 months, n = 8 (4 M, 4 F); and 22-24 months, n = 6 (3 M, 3 F). Values in App NL-G-F xMapt P290S KI mice were as follows: 3 months, n = 6 (3 M, 3 F); 6 months, n = 8 (4 M, 4 F); 12 months, n = 8 (4 M, 4 F); 18 months, n = 8 (4 M, 4 F); 22-24 months, n = 4 (2 M, 2F). Brain sections from 4 (2 M, 2 F) 18-month-old App NL-G-F xMapt P290S KI mice were not available at the time of AT100 staining. B, Representative images of Gallyas-Braak silver staining at 24 months of age in Mapt P290S and App NL-G-F xMapt P290S KI mice. C, Timeline depicting the appearance of cells the brains of Mapt P290S KI and App NL-G-F KIxMapt P290S KI mice (Fig. 3A). Immunoreactivity with AT8 and AT100 became detectable at 6 months of age in both lines and increased until 12 months, mostly along midline structures. Tau inclusions were present in nerve cell bodies and axons. There were no significant differences between lines. However, between 12 and 18 months, a significant increase in the number of AT100-immunoreactive cells was present in App NL-G-F xMapt P290S (p , 0.05), but not in Mapt P290S (p . 0.9999), KI mice. At 18 months, there were six times as many AT100-immunoreactive cells in App NL-G-F xMapt P290S mice than in Mapt P290S KI mice (p , 0.05). There was no significant difference in the number of AT8-immunoreactive cells between App NL-G-F xMapt P290S and Mapt P290S KI mice, probably because of the large variability in the amount of tau pathology at this time point in the double-KI line. At 22-24 months, the difference in the number of tau inclusions between App NL-G-F xMapt P290S and Mapt P290S KI mice reached 33-fold with AT100 (p , 0.0001) and 75-fold with AT8 (p , 0.0001; Fig. 3A). Gallyas-Braak silver-positive inclusions were first detected in both KI lines at 12 months, but increased progressively until 24 months of age only in the App NL-G-F xMapt P290S KI line (Fig. 3B,C). In situ transmission electron microscopy of Gallyas-Braak silver-stained tissues showed filamentous structures in nerve cells, as illustrated for a cell body from the piriform cortex (Fig. 3D).
continued labeled with Gallyas-Braak silver stain in Mapt P290S KI and App NL-G-F xMapt P290S KI mice. Values were as follows: at 3 and 6 months, n = 3 (2 M, 1 F); at 12 and 18 months, n = 8 (4 M, 4 F); at 22-24 months, n = 4 (2 M, 2 F). D, In situ electron microscopy of Gallyas-Braak silver-stained brain sections from 24-month-old App NL-G-F xMapt P290S KI mouse showing labeled filamentous content of a tau inclusion (i) within the cell body of a neuron, adjacent to a nucleus (n).
Region-specific increase in tau pathology in App NL-G-F KIxMapt P290S KI mice AT8 immunoreactivity was quantified in PV, PAG, amygdala, hippocampus, and piriform cortex. In Mapt P290S KI mice, AT8-immunoreactive cells were present in PV and PAG, but only rarely in hippocampus, amygdala, and piriform cortex. In App NL-G-F xMapt P290S KI mice, AT8-positive cells were also present in PV and PAG; from 18 months, immunoreactivity expanded into amygdala, hippocampus, and piriform cortex (Fig. 4A). At 18 months of age, there was a threefold increase in AT8-immunoreactive cells in the PAG (p = 0.0405), with no significant difference in the PV, and a 33-fold increase in the piriform cortex (p = 0.0086) of App NL-G-F xMapt P290S KI mice, compared with age-matched Mapt P290S KI mice. At 24 months, this increase was threefold in PAG (p = 0.0206) and PV (p = 0.0377), and at least 170-fold in the amygdala (p , 0.0001), hippocampus (p = 0.0005), and piriform cortex (p , 0.0001; Fig. 4B). Cerebellum, striatum, and lumbar spinal cord were devoid of tau immunoreactivity.

Neuritic plaques with and without tau filaments
In the brains from App NL-G-F KI mice, Ab plaques were surrounded by dystrophic neurites from 6 months of age onward; they were immunoreactive with AT8, but negative with AT100 and Gallyas-Braak silver. In contrast, in App NL-G-F xMapt P290S KI mice, neuritic plaques were immunoreactive with AT8 and AT100 from 6 months of age onward (Fig. 5A,B). They were Gallyas-Braak silver positive at 18 months (Fig. 5B,C). This shows that dystrophic neurites from App NL-G-F xMapt P290S mice contained hyperphosphorylated and filamentous tau, whereas in neuritic plaques from App NL-G-F KI mice, tau was hyperphosphorylated, but not filamentous. By in situ transmission electron microscopy of Gallyas-Braak silver-stained brain tissues from App NL-G-F xMapt P290S mice, neuritic processes with labeled filaments surrounded Ab plaques (Fig. 5C).

Seeded assembly of tau
We used a tau reporter line in HEK293 cells expressing human 0N4R P301S tau with a venus tag (McEwan et al., 2017) to investigate seeding activity, as assessed by fluorescent cytoplasmic foci, of Sarkosyl-insoluble brain extracts from Mapt P290S and App NL-G-F xMapt P290S KI mice. Sarkosylinsoluble brain extracts from 18-month-old Mapt P290S and App NL-G-F xMapt P290S KI mice seeded, as evidenced by the presence of AT8-and AT100-positive fluorescent puncta (Fig. 6A). Seeding with brain extracts from Mapt P290S KI mice was higher than with brain extracts from wild-type mice, but this difference was not statistically significant. The seeding ability of brain extracts from App NL-G-F xMapt P290S mice was 28-fold higher than that of wild-type brain extracts at 12 months (p = 0.0440), and 43-fold higher than that of wild-type brain extracts at 18 months of age (p , 0.0001). At 18 months, there was a significant threefold difference in seeding activity between App NL-G-F xMapt P290S and Mapt P290S KI mice (Fig. 6B). Immunoblotting with an antihuman tau antibody showed that small amounts of Sarkosylinsoluble tau could be extracted from cells seeded with brain Figure 5. Neuritic plaques in Mapt P290S and App NL-G-F xMapt P290S KI mice. A, Immunofluorescence staining of tau (AT8 and AT100) and Ab (D12B2) in 18-month-old App NL-G-F and App NL-G-F xMapt P290S KI mouse brains. App NL-G-F brains contain neuritic plaques (asterisk) that are positive for AT8, but not for AT100. App NL-G-F xMapt P290S KI KI brains contain neuritic plaques and tau inclusions (arrow) that are positive for both AT8 and AT100. B, Timeline of the appearance of neuritic plaque labeled with AT8, AT100, and Gallyas-Braak silver staining in App NL-G-F and App NL-G-F xMapt P290S KI lines. Values are as follows: at 3 and 6 months, n = 3 (2 M, 1 F); at 12 and 18 months, n = 8 (4 M, 4 F); at 22-24 months, n = 4 (2 M, 2 F). C, Gallyas-Braak-labeled neuritic plaque in the piriform cortex of an App NL-G-F xMapt P290S KI mouse at 24 months of age, visualized by light microscopy and in situ electron microscopy. Figure 6. Tau assemblies extracted from brains of Mapt P290S and App NL-G-F xMapt P290S KI mice induce aggregation of P301S tau expressed in HEK293 cells. A, Representative images of HEK293 P301S tau-venus cells seeded with 18-month-old Mapt P290S and App NL-G-F xMapt P290S Sarkosyl-insoluble tau assemblies labeled with AT8 and AT100. B, Quantitation of tau-venus signal per nucleus seeded with assemblies from 3-, 6-, 12-, and 18-month-old Mapt P290S and App NL-G-F xMapt P290S and 20-month-old wild-type and App NL-G-F controls (n = 3). C, Representative immunoblots of total homogenates and Sarkosyl-insoluble extracts from seeded cells using human-tau antibody HT7 (n = 3). extracts from 12-and 18-month-old Mapt P290S KI mice, indicating that low levels of seed-competent tau were present, in accordance with these mice having little tau pathology. Brain extracts from App NL-G-F xMapt P290S KI mice, on the other hand, seeded more insoluble tau, consistent with the presence of increased tau pathology (Fig. 6C).

Age-dependent neurodegeneration in
App NL-G-F xMapt P290S KI mice Nerve cell loss in the piriform cortex was quantified using unbiased stereological cell counting. Significant nerve cell loss in App NL-G-F xMapt P290S compared with age-matched wild-type (p = 0.0178) and Mapt P290S KI mice (p = 0.0002) was detected from 18 months of age onward (Fig. 7A). No significant nerve cell loss was detected in the Mapt P290S KI mice compared with age-matched wild-type mice.

Discussion
We report the production and characterization of a knock-in mouse model (Mapt P290S KI), expressing mutant murine tau at endogenous levels. In contrast to the P290L knock-in line, in which tau pathology could not be detected by AT8 or Gallyas-Braak staining until 32 months of age, the longest time point studied (Gilley et al., 2012), Mapt P290S KI mice exhibited a limited number of AT100positive tau inclusions within a restricted group of midline nuclei from 6 months of age. However, Sarkosyl-insoluble tau was not detected by immunoblotting, probably because of the small number of inclusions. In view of AT100 positivity (Delobel et al., 2008), these findings suggest that murine tau can assemble in vivo, despite differences in isoform composition and amino acid sequence with human tau. Mice express 3R tau during fetal development, but only 4R tau is found in the brains of adults (Brion et al., 1993;Gotz et al., 1995). Additional differences between murine and human tau have been described in the N terminus (Hernandez et al., 2020). Maintaining the murine Mapt gene may be a desirable alternative to the humanization of tau expressed in a murine environment, since murine binding partners may be unable to interact with human tau isoforms. Mouse lines expressing all six isoforms of humanized wild-type tau at endogenous levels have been produced (Saito et al., 2019;He et al., 2020). These mice did not develop overt tau pathology, and crossing them with App NL-G-F KI mice led to increased tau phosphorylation, but tau inclusions were not detected for up to 24 months (Saito et al., 2019). A lack of tau pathology suggests that humanization of wild-type tau is insufficient to induce tau pathology, even in the presence of Ab . Pathogenic mutations may be required to study aggregation-prone tau in mice.
By crossing our Mapt P290S KI mice with the App NL-G-F KI line (App NL-G-F xMapt P290S KI), we demonstrate that Ab pathology promoted tau aggregation in an age-dependent manner. Sarkosyl-insoluble tau was only detectable by immunoblotting in 18-month-old App NL-G-F xMapt P290S KI mice, and significantly more tau inclusions were found in 18-month-old App NL-G-F xMapt P290S KI mice compared with age-matched Mapt P290S KI mice. This is in agreement with overexpression models, which showed enhanced tau pathology in the presence of Ab (Lewis et al., 2001;Chen et al., 2016). Unlike these models, which exhibited significant tau pathology at a young age and often needed to be culled before 12 months, the small number of tau inclusions in Mapt P290S KI mice enabled us to model interactions between tau inclusions and Ab plaques as a function of age. Mice older than 18 months are considered "elderly" (Flurkey, 2007) and begin to exhibit signs of cognitive impairment at ;22 months (Yanai and Endo, 2021). Consistent with an effect of aging, we only observed a significant effect of Ab on the number of tau inclusions after 18 months, despite Ab deposition beginning as early as 3 months.
In AD, tau pathology propagates through the brain in a stereotypical manner. Tau inclusions develop in the entorhinal cortex and later progress to limbic areas, followed by the neocortex (Braak and Braak, 1991). Overexpression models rely on the expression of tau under the control of strong promoters, such as murine Thy-1 (Allen et al., 2002) and prion protein promoters (Lewis et al., 2000;2001). This leads to tau pathology developing in spinal cord, with early behavioral phenotypes affecting motor function and gait. With mutant tau under the control of its endogenous promoter, we instead observed only minimal tau pathology along midline structures in Mapt P290S KI brains, in the absence of motor dysfunction. Interestingly, in the presence of Ab , tau pathology appeared in regions reminiscent of Braak staging, such as entorhinal cortex, hippocampus, and amygdala, as well as piriform cortex. This is consistent Figure 7. Age-dependent neurodegeneration in App NL-G-F xMapt P290S KI mice. Quantitation of NeuN-positive cells in piriform cortex of wild-type, Mapt P290S KI, and App NL-G-F xMapt P290S KI mice at 12, 18, and 22-24 months of age. Results shown are the mean 6 SEM. *p , 0.05, ****p , 0.0001 for twoway ANOVA with Tukey's post hoc analysis. For wild-type mice at all ages, n = 3 (M); for Mapt P290S KI mice at 12 and 18 months, n = 4 (2 M, 2 F); at 22-24 months, n = 6 (3 M, 3 F); for App NL-G-F xMapt P290S KI at 12 and 18 months, n = 4 (2 M, 2 F); at 22-24 months n = 5 (2 M, 3 F).
with a recent longitudinal amyloid and tau PET study, which found that people with Ab plaques in the entorhinal cortex are more likely to have spreading of tau pathology on follow-up (Lee, Brown, et al., 2022).
Dystrophic neurites that are AT8-positive have been reported to surround Ab plaques in App NL-G-F KI mice, with increased AT8 signal in double-KI mice expressing all six humanized tau isoforms (Saito et al., 2019). However, assembled tau was not present in those dystrophic neurites. We report dystrophic neurites in App NL-G-F xMapt P290S KI mouse brains that were AT8, AT100, and Gallyas-Braak positive, indicating the presence of filamentous tau. In support, we observed silver-positive filaments in neurites surrounding Ab plaques in brain sections from 24-monthold App NL-G-F xMapt P290S KI mice by in situ electron microscopy. Surprisingly, we observed that amyloid plaques appeared to form intracellularly within an enclosing membrane. However, in instances where these membranes ruptured, plaques were surrounded by Gallyas-Braakpositive neuritic processes, astrocytic processes, and many microglial cells. These observations are consistent with the recent description of PANTHOS in APP transgenic mice and AD brains .
The role of neuritic plaques in the propagation of tau pathology remains unclear. Intracerebral injection of extracts from AD brains into 5xFAD mice with significant Ab pathology led to the formation of dystrophic neurites and seeded endogenous tau inclusions (Vergara et al., 2019). Studies of AD extract-injected App NL-G-F KI mice have also suggested that the environment of Ab plaques facilitates seeded assembly and propagation of endogenous tau (He et al., 2018).
We used a cell-based model to investigate the seeding ability of tau filaments extracted from Mapt P290S KI and App NL-G-F xMapt P290S KI brains (McEwan et al., 2017). We were able to detect seeded tau from cells treated with both aged Mapt P290S KI and App NL-G-F xMapt P290S KI brain extracts. This suggests that seed-competent tau is present in aged Mapt P290S KI brains, but significant propagation of tau pathology did not occur in vivo in the absence of Ab plaques. There is growing evidence that Ab pathology promotes the spreading of tau pathology. Tau propagation to the entorhinal cortex of mice overexpressing human P301L tau was accelerated when crossed with APP/PS1 mice (Pooler et al., 2015). Brain homogenates from tauopathy patients showed enhanced seeding ability in cases with both tau neurofibrillary tangles and Ab plaques, compared with cases without plaques (Bennett et al., 2017). Immunotherapy aimed at reducing Ab plaque burden also appears to have an effect on tau spreading. In 3xTg-AD mice, anti-Ab immunotherapy has been shown to reduce the amount of tau pathology (Oddo et al., 2004). A recent report of an 84-year-old woman receiving the anti-Ab antibody aducanumab for 32 months showed lower phospho-tau immunoreactivity when compared with untreated AD cases (Plowey et al., 2022). Together, these findings suggest that assembled Ab can promote, but not induce, the assembly of tau. The same conclusion was reached in studies of tau filaments from human brains by electron cryomicroscopy (Shi et al., 2021).
To look for neurodegeneration, we examined the piriform cortex of KI mice using unbiased stereological cell counting. We observed nerve cell loss in 18-to 24-month-old App NL-G-F xMapt P290S KI mouse brains, coinciding with the significant amplification of tau inclusions and the presence of tau filaments. The relationship between filamentous tau assemblies and nerve cell loss has been characterized in an overexpressing model of P301S tau pathology (Macdonald et al., 2019). Our findings extend the correlation between filamentous tau pathology and neurodegeneration to a mouse model expressing endogenous levels of tau.
We propose that App NL-G-F xMapt P290S KI mice provide a good model system for studying the interactions among aggregation-prone tau, Ab , neuritic plaques, neurodegeneration, and aging. Despite this, there are limitations to this model as a representation of sporadic AD, which is characterized by the aggregation of wild-type tau and Ab . Humans with the P301S mutation in tau, together with a mutation in APP, have not been reported. These mutations probably affect the conformation of protein aggregates, and the current model should be used bearing this in mind. Current evidence suggests, however, that mice do not develop wild-type tau aggregates without significant genetic manipulation. In the presence of Ab plaques, humanized wild-type tau does not aggregate into mature neurofibrillary tangles in aged mouse brain (Saito et al., 2019). It is thus likely that a certain degree of genetic manipulation of Mapt and App is required for modeling tau and Ab pathologies in mice. In this case, the App NL-G-F xMapt P290S KI mice offer advantages over models that rely on transgenic overexpression of mutant tau or APP. With tau and APP expressed under the control of the endogenous murine promoter, artificial phenotypes, owing to overexpression and random insertion of transgenes, are avoided (Saito et al., 2016;Gamache et al., 2019). We suggest that the Mapt P290S xApp NL-G-F KI mice are a valuable alternative model for studying tau pathologies in the context of Ab . (The mice can be obtained by contacting our research office at techtran@mrc-lmb.cam.ac.uk.)