PATHOLOGICAL TAU AND REACTIVE ASTROGLIOSIS ARE ASSOCIATED WITH DISTINCT FUNCTIONAL DEFICITS IN A MOUSE MODEL OF TAUOPATHY.

Abstract

Pathological aggregation of tau and neuroinflammatory changes mark the clinical course of Alzheimer’s disease and related tauopathies. To understand the correlation between these pathological hallmarks and functional deficits, we assessed behavioral and physiological deficits in the PS19 mouse model, a broadly utilized model of tauopathy. At 9 months, PS19 mice have characteristic hyperactive behavior, a decline in motor strength, and deterioration in physiological conditions marked by lower body temperature, reduced body weight, and an increase in measures of frailty. Correlation of these deficits with different pathological hallmarks revealed that pathological tau species, characterized by soluble p-tau species, and tau seeding bioactivity correlated with impairment in grip strength and thermal regulation. On the other hand, astrocyte reactivity showed a positive correlation with the hyperactive behavior of the PS19 mice. These results suggest that a diverse spectrum of soluble pathological tau species could be responsible for different symptoms and that neuroinflammation could contribute to functional deficits independently from tau pathology. These observations enhance the necessity of a multi-targeted approach for the treatment of neurodegenerative tauopathies.

1. INTRODUCTION

Pathological aggregation of the microtubule-associated protein tau and the preponderance of neurofibrillary tangles (NFT) or other tau-containing inclusions are defining histopathological features of Alzheimer’s disease (AD) and many other neurodegenerative diseases, collectively known as tauopathies, including Pick’s disease (PiD), Progressive Supranuclear Palsy (PSP), Corticobasal Degeneration (CBD), and Frontotemporal Lobar Degeneration (FTLD) (Alonso et al., 2008; Lee et al., 2001). At the pathological level, the correlation between NFTs and disease progression has been studied extensively with conflicting results, and the mechanisms linking the pathological aggregation of tau with synaptic dysfunction and neurodegeneration are poorly understood. An emerging view is that the NFTs themselves are not the true toxic entity in tauopathies; rather, it is the soluble tau aggregates that play a primary toxic role (Lasagna-Reeves et al., 2012b; Maeda and Takashima, 2019; Ren and Sahara, 2013). For instance, in mouse models of tauopathy, levels of soluble tau oligomers correlate with memory loss at different ages (Berger et al., 2007). Moreover, suppression of soluble tau improves cognitive function without reducing the level of insoluble NFTs in inducible mouse models of tauopathy (Santacruz et al., 2005; Sydow et al., 2011). The causal role of tau aggregation in the pathogenesis of tauopathies is supported by several tauopathy mouse models, in which expression of human wild-type or mutant tau leads to neurodegeneration and the development of behavioral abnormalities (Lewis et al., 2000; Santacruz et al., 2005). These tauopathy models have been extremely useful for studying the contributions of synaptic pathology and gliosis to human tauopathies (Jankowsky and Zheng, 2017; Santacruz et al., 2005; Yoshiyama et al., 2007). However, the translational relevance of these mouse models is limited without a correlation between the appearance of these pathological hallmarks and behavioral impairment, since tau pathology, synaptic loss, and neuroinflammatory changes, such as activation of microglia and astrocytes, generally delineate the clinical course of cognitive impairment in AD patients (Long and Holtzman, 2019).

This led us to behaviorally characterize the PS19 mouse model, a broadly utilized model of tauopathy, expressing human tau (1N4R) with the P301S mutation under the control of the mouse prion promoter (Yoshiyama et al., 2007). Since this model is characterized by progressive tau accumulation and major neuroinflammation, we evaluated if any of these pathological hallmarks were associated with functional impairments and if this association was brain-region-specific. We determined that by 9 months of age, PS19 mice exhibit a distinct hyperactivity-associated increase in locomotor activity and a decline in motor strength when there is significant tau pathology and glial reactivity in the brain. Furthermore, these animals have a decline in physiological conditions marked by lower body temperature, weight loss, and an increase in measures of frailty. Our results show that the tau pathology, quantified by soluble p-tau levels and tau seeding activity, correlates with the functional impairment of motor strength and temperature homeostasis. The hyperactivity associated to enhanced locomotor activity showed a correlation with astrocyte reactivity, a common pathological hallmark of human tauopathies that is present in the PS19 mouse model. The results of our study indicate that there is possibly a differential role of multiple pathologies that underlie the broad spectrum of functional deficits in a mouse model of tauopathy, suggesting the necessity to develop novel disease-modifying therapies that target different pathologies associated with neurodegenerative tauopathies rather than using a monotherapy approach.

2. MATERIALS AND METHODS

2.1. Transgenic mouse model

The PS19 mouse model (Yoshiyama et al., 2007), which overexpresses human 1N4R tau with the P301S mutation, was directly purchased from The Jackson Laboratory (stock number 008169). 30 PS19 (14 females and 16 males) and 26 wild-type (WT) (10 females and 16 males) mice were used. All mice were on C57BL/6J background. Mice were bred and housed at the Indiana University School of Medicine (IUSM) animal care facility and were maintained according to USDA standards (12 hr light/dark cycle, food, and water ad libitum), in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD). For all described experiments, 9-month-old PS19 and WT littermates of both sexes were utilized. Animals were anesthetized and euthanized according to IUSM Institutional Animal Care and Use Committee-approved procedures. Brains were extracted, the left hemisphere was stored at −80°C for biochemical analysis, and the right hemisphere was formalin-fixed for the preparation of paraffin blocks as previously described (You et al., 2019).

2.2. Behavioral and physiological assays

Spontaneous alternation Y-maze:

Mice were subjected to the Y-Maze (MazeEngineers) test described in (Cisternas et al., 2020), with modifications. Mice were allowed to freely explore the maze for 10 min. The number of arm entries, re-entries, and the distance traveled were recorded using an automatic monitoring system (ANY-Maze). A spontaneous alternation occurred when a mouse entered a different arm of the Y-Maze in each of three consecutive arm entries. The spontaneous alternation percentage was calculated as: ((# spontaneous alternation)/(Net arm entries)) × 100.

Body temperature, body weight, and frailty:

Mice were weighed, and the resting core body temperature was measured by inserting a lubricated rectal probe (Bioseb), ~1 cm into the rectum for 10 sec. For the frailty examination, mice were assessed for the presence or absence of 26 different characteristics as described in (Whitehead et al., 2014), with modifications, (Supplementary File 1). A score of 0 was given if the mouse had no sign of the deficit, 0.5 if there was a mild deficit, and 1 if there was a severe deficit.

Grip Strength:

Grip strength of the forelimbs (front two paws) and all limbs (four paws) was evaluated using the Grip Strength Meter (Bioseb, BIO-GS3). As per the manufacturer’s protocol, mice were held by the tail and lowered towards the apparatus and allowed to grab the metal grid using two or four paws. The mice were pulled backward horizontally, and the force applied to the grid just before they lost their grip was recorded as the peak tension (converted to grams by the transducer). Peak force was measured two times in succession for each mouse for the front two paws and all four paws. The mean value of both trials was used for analysis. Mice were given a minimum break of 5 min between trials.

Open Field test:

Mice were tested with the open field test described in (Cisternas et al., 2020), with modifications. Briefly, mice were placed in the center of a dimly lit (20–30 lux) open field arena (45 × 45 × 45 cm) for 60 min and were allowed to freely explore the arena. Animal movements were tracked by an automatic monitoring system (ANY-Maze). The area was virtually divided into a center (28 cm edge lengths), four wall corridors (7 cm along the walls), and four corner squares (7 cm edge lengths). The time and distance traveled were automatically measured for each quadrant.

Fear Conditioning:

Fear conditioning was performed as previously described (Takeuchi et al., 2011) with modifications. Each mouse was placed in a sound-attenuated chamber, with clear acrylic walls, a shock grid floor, and vanilla odor and allowed to freely explore for 5 min. The conditioned stimulus (CS), a 90-dB white noise, was presented for 30 sec followed by the unconditioned stimulus (US), a mild (1 sec, 0.75 mA) foot shock. Four more CS-US pairings were presented at 5 min intervals. Context testing was conducted one day after conditioning in the same chamber for 10 min in absence of the CS. Cued testing with an altered context was conducted the next day in a chamber with black and white striped walls, a smooth black Plexiglass surface instead of a shock-grid floor, and peppermint odor. Data acquisition and stimuli (i.e., tones and shocks) analyses were performed automatically using Fusion software (Omnitech).

Active Place Avoidance:

The active place avoidance assay was conducted with mice using an apparatus developed by MazeEngineers, consisting of a rotating arena (diameter 78 cm) placed on a metal grid floor (80 × 80 cm) and enclosed by a 24 cm high clear acrylic fence. The whole apparatus is raised 35 cm above the ground by a metal stand. ANY-Maze was linked to an overhead camera, and the ConductMaze program controlled the rotation (1 RPM) and shock (0.4 mA). Distinct visual cues were placed in four coordinate directions, and a gray screen was used to hide the experimenter from the view. Mice were allowed to explore the arena for 10 min while avoiding the shock zone, which was constant with respect to the room’s coordinates, using visual cues. Acquisition learning was performed for three days and reversal-learning on the fourth day.

2.3. Mouse brain sample preparation and immunoblot analysis

22 randomly selected PS19 brains (11 females and 11 males) were homogenized at a 1:10 (w/vol) ratio of the brain in 1X TBS supplemented with a complete protease inhibitor cocktail (Roche) using a Bead Beater (1 min) and then sonicated for 1 min (50% amplitude, pulse on 12 sec, pulse off 3 sec). Samples were then centrifuged at 15,000 rpm for 10 min at 4°C. TBS soluble fractions were aliquoted into fresh tubes. The insoluble pellets were resuspended in 88% formic acid (FA) at one fourth volume of their brain homogenates and then incubated for 1 hr at RT. Samples were then diluted with distilled water to obtain the same volume used for brain homogenates. Samples were then lyophilized for 24 hr. Freeze-dried samples were reconstituted in PBS using the same volume as brain homogenates, then sonicated for 30 sec. For immunoblot analysis, TBS soluble fractions and TBS insoluble fractions were run on a 24-well NuPAGE 4–12% Bis-Tris protein gel (Invitrogen) and transferred to a nitrocellulose membrane. The primary antibodies used against tau species were PHF1 (1:1000, from P. Davies, AB_2315150) (Greenberg et al., 1992; Otvos et al., 1994), anti-pS214 tau (1:1000, ab170892, Abcam), anti-pT217 tau (1:1000, 44–744, ThermoFisher), total human tau, HT7 (1:1000, MN1000, ThermoFisher) and anti-Vinculin (1:1000, V9131, Sigma). The secondary antibodies used were goat anti-mouse HRP IgG (1:5000, A16066, Invitrogen) and goat anti-rabbit HRP IgG (1:5000, PI31460, Invitrogen). For each analysis, the relative level of each tau antibody and Vinculin was quantified from Western blot images using ImageJ software (NIH). The level of each tau species was normalized to HT7 to determine the relative abundance of each tau species with respect to total tau levels.

2.4. NFT immunostaining and quantification

Paraffin sections from the 22 randomly selected 9-month-old PS19 mice were deparaffinized, rehydrated, and washed in 1X TBS three times for 5 min each. Gallyas silver staining of NFTs was done as previously described (Berger et al., 2007), with modifications. Briefly, slides were incubated in 5% periodic acid for 10 min and then washed twice with distilled water for 5 min each. Slides were then incubated in alkaline silver iodide solution for 1 min and washed in 0.5% acetic acid for 10 min. Slides were immediately placed in freshly made developer solution and developed for 10 min. Two washes were performed in 0.5% acetic acid for 3 min per wash. Slides were incubated in 1% sodium thiosulphate solution for 5 min and then washed in running tap water for approximately 5 min. Finally, the slides were counterstained in 0.1% nuclear fast read for 2 min, and washed in tap water, dehydrated, and cover slipped. The hippocampus, cortex, hypothalamus, and pons region of the brain stem were examined using a Leica DFC7000 microscope. To count the number of NFTs, three pictures of each area per animal were analyzed using ImageJ Fiji software. A cell with positive black staining in the soma and at the beginning of the neurite was counted as one NFT. The average number of NFT counts in the three pictures was normalized to the total area quantified to obtain the final measurement of NFT/mm².

2.5. Flow cytometry and analysis of seeding activity

Seeding assay was performed as previously described (Holmes et al., 2014), with modifications. TauRD P301S FRET Biosensor cells (ATCC #CRL-3275) were plated at 35,000 cells per well in 130 μL media in a 96-well plate and incubated at 37°C overnight. The next day, cells were transfected with brain lysate at a concentration of 20 μg total protein per well using Lipofectamine 2000 and incubated at 37°C for 2 days. Cells were harvested by trypsinization. Flow cytometry was conducted with a BD LSRFortessa^™ X-20 instrument with a High Throughput Sampler. The BV421 channel was used to detect CFP (405nm Ex, 450/50 Em filter), and the BV510 channel was used to detect the FRET signal (405nm Ex, 525/50 + 505LP Em filter), with compensation to remove the CFP emission from the FRET signal. FlowJo software was used to gate cells with a positive FRET signal and seeding was quantified by measuring the percentage of positive cells and the median fluorescent intensity (MFI) of the compensated BV510 channel. Integrated FRET density was calculated by multiplying the percentage of positive cells with the MFI.

2.6. Astrocyte and microglia immunostaining and quantification

Immunostaining of astrocytes and microglia was performed as previously described (Taylor et al., 2020) using anti-GFAP (G3893, Sigma-Aldrich) and anti-IBA1 (019-19741, Wako) antibodies respectively. The hippocampus, cortex, hypothalamus, and pons region of the brain stem of the 22 selected PS19 mice were examined using a Nikon A1-R laser scanning confocal microscope. For analysis, images were imported into ImageJ software (NIH) to create one index that represented changes in both astrocytes and microglia. GFAP (+) or IBA1 (+) mean gray intensity was normalized to the area of the specific brain region using the area fraction function for each mouse and expressed as GFAP or IBA1 intensity, as previously described (Cernak et al., 2014). Slides used for quantification were blind-coded prior to analysis.

2.7. Correlation analysis

Pearson’s correlation coefficient was used to analyze the correlation between behavioral and physiological deficits with levels of tau species abundance, NFT counts, tau seeding levels, and glia intensity in the PS19 mouse brains (N=22). The number of NFTs and glial intensity across the hippocampus, cortex, hypothalamus, and brain stem was used to correlate pathology and deficits. Average values across four regions were used to represent the whole brain.

2.8. Statistical analysis

All behavioral and pathological analyses and data collection were performed in a randomized and blinded manner. For behavioral analyses and sex differences, an unpaired t-test or the Mann-Whitney U test was used where appropriate. Two-way ANOVA was used to analyze active place avoidance. Correlation analysis was performed using Pearson’s correlation coefficient as described above. GraphPad Prism 9 was used to perform all the statistical analysis and to plot the graphs. Complete statistical reports can be found in the Supplementary File 2. We used an alpha level of .05 for all statistical tests. All correlations were performed after removing any outliers from each data set (measured using ROUT analysis, FDR Q=1%). The statistical analyses before and after removing outliers are provided in Supplementary File 3.

3. RESULTS

3.1. Behavioral and physiological deficits in PS19 mice

It has been well documented that at 9 months, PS19 mice exhibit major tau pathology, functional deficits, and a decline in physiological condition (Lasagna-Reeves et al., 2016; Takeuchi et al., 2011). Therefore, we subjected 9-month-old PS19 mice and age-matched WT littermates to a battery of behavioral and physiological tests to measure memory and cognition, locomotor activity, motor strength, and general body condition (Fig. 1). We evaluated memory and cognition using the Y-Maze spontaneous alternation assay, fear conditioning, and active place avoidance assay (Fig. 1A–C). In the Y-Maze test, the PS19 mice exhibited a greater number of arm entries (t(54)=2.9, p = .005), total alternations (t(54)=2.768, p = .007), and distance traveled (t(54)=4.8, p = <.0001), indicating hyperactive behavior (Fig. 1A). However, the percentage of spontaneous alternations of PS19 mice did not differ from WT mice, suggesting no difference in working memory between two groups at 9 months of age (Fig. 1A). Both female and male PS19 mice displayed a hyperactive behavior but no differences in working memory compared to WT mice (Supplementary Fig. 1A). In addition, we did not observe any difference between the two groups in the contextual and cued fear conditioning and active place avoidance task, which rely on the hippocampus and amygdala (Takeuchi et al., 2011), suggesting that there are no major cognitive impairments at this age in the PS19 mice (Fig. 1B–C), at least in the context of these assays and the specific settings tested. The hyperactive behavior of the PS19 mice, as observed in the Y-Maze test, was supported by their locomotor activity during the 60 min Open Field test. Compared to WT mice, the PS19 mice exhibited greater locomotor activity (t(54)=3.8, p = .0003) in the Open Field arena (Fig. 1D). When we analyzed sex differences in this behavioral test, both female and male PS19 mice displayed increased activity levels in the Open Field arena (Supplementary Fig. 1B). We did not observe any thigmotaxis (wall-hugging) or an increase in the time spent in the corners of the arena by the PS19 mice (Figure 1D). These results suggested that the PS19 mouse model does not have any apparent anxiety-related behavior but rather considerable hyperactivity, that is consistent across both sexes. To assess motor strength, we compared the grip strength of PS19 and WT mice. The grip strength of the front two paws and all four paws were tested. PS19 mice exhibited reduced 2-paw (t(54) = 4.17, p = .0001) and 4-paw (t(54) = 3.65, p= .0006) grip strength compared to that of WT mice (Fig. 1E). The decrease in grip strength was significant in both female and male PS19 mice (Supplementary Fig. 1C), suggesting that there is a decline in motor strength in the PS19 mice by 9 months of age. No obvious walking abnormalities were observed in these mice. Nevertheless, further quantitative analysis should be performed to establish walking impairment in this tauopathy model. We also evaluated the physiological characteristics of these mice by measuring the basal core body temperature, body weight, and markers of frailty. A comparison between the basal body temperature of PS19 and WT mice revealed that the PS19 mice had lower basal body temperatures (t(54)=2.8, p = .006) than those of the WT mice (Fig. 1F). Furthermore, reduced body temperatures were observed in both sexes (Supplementary Fig. 1D). Literature evidence suggests that lower body temperatures contribute to the development and exacerbation of AD pathology (Whittington et al., 2010). We next measured the body weight of these animals. Although PS19 mice exhibited normal gross morphology, they had significantly lower body weight (t(54)=3.4, p = .0009) compared to that of WT mice (Fig. 1G). Both female and male PS19 mice had significantly lower body weight than that of the age-matched and sex-matched WT mice (Supplementary Fig. 1E). Finally, we performed a clinical exam to assess the 26 frailty parameters (see methods). PS19 mice exhibited greater measures of frailty (Mann Whitney U = 202, p = .001) than the WT mice (Fig. 1H), suggesting a decline in their general health condition. Female PS19 mice had significantly greater signs of frailty than WT controls (Supplementary Fig. 1F). Male PS19 mice had a trend towards higher frailty scores than WT mice but failed to reach statistical significance (p = 0.05) (Supplementary Fig. 1F). Overall, these results indicate that the PS19 mice have a hyperactive phenotype as marked by increased locomotor activities, a decrease in motor strengths, and a decline in physiological health conditions as marked by a decrease in body temperature, weight loss, and increase in measures of frailty.

Figure 1. Behavioral and physiological deficits in the PS19 mouse model of tauopathy.

PS19 mice were subjected to different behavioral and physiological tests. (A) In the Y-Maze, PS19 mice had significantly increased total entries and total alternations, and increased distance traveled. No difference in the percentage of spontaneous alternations was observed, suggesting no impairment in working memory in PS19 mice. (B–C) PS19 mice exhibited no impairments in cognition as measured by (B) fear conditioning test and (C) active place avoidance test. (D) In the Open Field test, PS19 mice traveled longer distances and did not exhibit preferences for corners or along the walls. (E) PS19 mice had lower grip strength of both the front 2-Paw and 4-Paw compared to WT mice. (F–H) PS19 mice had significantly lower (F) core body temperature, (G) lower body weights, and (H) increase in measures of frailty. Black bars = WT, red bars = PS19. Data represented as the mean +SEM, p indicated on each behavioral test using unpaired Student t-test, Mann-Whitney Test for frailty, and two-way ANOVA for Active Place avoidance test. *p<0.05, **p<0.01, ***p<0.001, ****p ≤0.0001. WT (N=26), PS19 (N=30); fear conditioning, WT (N=10) PS19 (N=8); active place avoidance WT (N=16) PS19 (N=22).

To understand if the observed behavioral and physiological deficits in 9-month-old PS19 mice are dependent on each other, we performed a correlation analysis between the deficits. There was no significant correlation observed between the functional deficits; only a negative correlation between hyperactivity and body weight was observed (Table 1). This suggests that the functional deficiencies might not depend on each other and that having one deficit might not necessarily imply the presence of other deficits. Furthermore, it is tempting to hypothesize that there could be differential pathologies associated with each of the deficits observed.

Table 1. No major association between functional deficits in the PS19 mouse model.

Functional deficits observed in the PS19 mice were correlated with each other using Pearson’s correlation coefficient. A significant correlation was only observed between hyperactivity and lower body weights.

Measure	Open Field activity (m)	2-Paw Grip Strength (g)	4-Paw Grip Strength (g)	Body Temperature (°C)	Body Weight (g)
Open Field activity (m)	--
2-Paw Grip Strength (g)	0.39	--
4-Paw Grip Strength (g)	0.18	0.21	--
Body Temperature (°C)	−0.31	−0.27	0.26	--
Body Weight (g)	−0.53**	−0.15	−0.09	0.32	--

^**p<0.001

3.2. Distinct pathological tau species correlate with a decrease in motor strength and a decrease in core body temperature in PS19 mice

Growing evidence suggests that the NFTs are not the primary toxic entities in tauopathies (Berger et al., 2007; Castillo-Carranza et al., 2014; Santacruz et al., 2005; Yoshiyama et al., 2007). Rather, it has been suggested that soluble tau species are primary mediators of neurotoxicity and that hyperphosphorylation of tau could play a critical role in neurotoxic events (Haroutunian et al., 2007; Lasagna-Reeves et al., 2011; Lasagna-Reeves et al., 2012b; Maeda and Takashima, 2019; Takashima, 2008; Yoshiyama et al., 2007). Therefore, we first performed a biochemical analysis to measure a set of soluble phosphorylated pathological tau species in the brain tissues of the 9-month-old PS19 mice that were tested for functional deficits. The TBS soluble brain fraction from each mouse was used for a Western blot analysis to quantify the levels of soluble phosphorylated tau using the following antibodies: for late-stage hyperphosphorylation, PHF1 against p-tau S396/S404, for early-stage phosphorylation, anti-p-S214 tau (Augustinack et al., 2002), and for the recently identified potent blood biomarker, anti-p-T217 tau (Janelidze et al., 2020). We then correlated the levels of each tau species with the behavioral and physiological deficits that were observed for each corresponding mouse. We specifically correlated soluble phosphorylated tau levels with distance traveled (Open Field), grip strength (2-paw and 4-paw assay), body temperature, and body weight (Fig. 2). Interestingly, the levels of p-tau S214 (r = −0.44, p = 0.04) and p-tau T217 (r = −0.48, p = 0.02) species showed a correlation with the decline in the 2-paw grip strength (Fig. 2B–C). Additionally, we also observed a negative relation between levels of p-tau S214 species and the 4-paw grip strength (r = −0.48, p = 0.02) (Fig. 2B). These results suggest a correlation between distinct soluble p-tau species and motor strength impairment in the PS19 mice. We then measured the levels of these phosphorylated tau species in the insoluble brain fractions. There was no correlation observed between the levels of insoluble phospho-tau species and behavioral or physiological deficits (Supplementary Fig. 2). These results lend support to the notion that the soluble tau species are the primary toxic entities.

Figure 2. Soluble p-tau species correlates with the decrease in motor strength.

Levels of soluble p-tau species in PS19 mouse brains (left) and correlations with behavioral and physiological deficits (right). (A) No correlation was observed between levels of soluble PHF1 in the brain and functional deficits. (B) Levels of soluble p-tau S214 showed a correlation with the decline in 2-Paw and 4-Paw grip strengths, but no correlation was observed with hyperactivity and physiological deficits, temperature, and weight. (C) Levels of soluble p-tau T217 showed correlation with decline in 2-paw grip strength. Y-axis represents the quantification of a specific parameter on each behavioral or physiological test. Each black dot corresponds to a PS19 mouse.

The NFT pathology has been shown to have a consistent correlation with memory deficits in human AD patients (Arriagada et al., 1992; Guillozet et al., 2003). Therefore, we used Gallyas Silver staining to detect mature NFTs in the brain sections of the PS19 mice and quantified the number of NFTs in the whole brain (Fig. 3A). The NFT counts in the brain were then correlated with the functional deficits for each mouse. The tangle pathology measured by NFT counts showed no correlation with the functional decline of PS19 mice (Fig. 3). When we correlated NFT pathology from individual brain regions with functional deficits, only NFT pathology in the hippocampus showed correlation with decline in body weight (Supplementary Fig. 3). The absence of significant correlation of NFT pathology with functional deficits extends the previously reported observations that the NFTs formation is not sufficient for the onset of functional deficits and that the soluble tau species are likely the main mediators of toxicity.

Figure 3. NFT counts do not correlate with the functional deficits in PS19 mice.

(A) Tangle pathology measured by Gallyas Silver stain (black). Arrows point to tangles. (B–F) NFT counts did not correlate with (B) locomotor activity in the Open Field, (C-D) grip strength, (E) body temperature or (F) body weights. Each black dot corresponds to a PS19 mouse.

Several studies have suggested that misfolded tau aggregates can act as a template or “seed” for intracellular monomeric tau aggregation. This tau species has also been shown to mediate the propagation of pathological tau between cells (Clavaguera et al., 2009; Lasagna-Reeves et al., 2012a; Liu et al., 2012). We thus measured the levels of tau seeding activity in the brains of the PS19 mice using the widely used biosensor cell-based assay that uses flow cytometric detection of FRET (Holmes et al., 2014). Briefly, the biosensor cell lines stably express tau repeat domains (TauRDs) fused with a cyan fluorescent protein (TauRD-CFP) or a yellow fluorescent protein (TauRD-YFP). Intracellular aggregation of these endogenous TauRDs occurs in the presence of a “seed”, which results in a FRET signal (Fig. 4A). We applied TBS soluble brain lysates from each PS19 mouse to these biosensor cells to quantify seeding activity in each brain. Tau seeding showed a wide range of bioactivity (Fig. 4B), which is consistent with results from human AD brain samples and can be attributed to the diversity of tau molecular species (Dujardin et al., 2020). Our correlation analysis between the amount of seeding activity in the PS19 mouse brains and functional deficits showed that seeding activity correlated with a decline in body temperature (r = −0.43, p = 0.04) in PS19 mice (Fig. 4C). None of the soluble or insoluble p-tau species that we measured correlate with tau seeding activity (Supplementary Fig. 4). Overall, these results suggest that a spectrum of soluble pathological tau species could be associated with diverse deficits, such as a decrease in motor strength, and a decrease in core body temperature in PS19 mice by 9 months of age.

Figure 4. Tau seeding correlates with the decrease in body temperature.

(A) Schematic representation of in vitro tau seeding assay. (B) Heterogeneity in the levels of tau seeding activity in the PS19 mouse brains. (C) Tau seeding activity in PS19 mice solely correlates with the decrease in body temperatures.

3.3. Reactive astrogliosis correlates to hyperactive behavior in PS19 mice

Glial activation is one of the pathological hallmarks of AD and other tauopathies, which closely parallels tau pathology (Leyns and Holtzman, 2017). The astrocytic and microglial activation in the PS19 mouse model has been well described (Gratuze et al., 2020; Leyns et al., 2017; Yoshiyama et al., 2007). After confirming major astrogliosis and microgliosis in the 9-month-old PS19 mice vs. the WT littermates (Supplementary Fig. 5), we investigated whether glial activation correlates with the functional deficits observed in the PS19 mouse model. We performed a correlation analysis between GFAP reactivity (to measure reactive astrogliosis) in the brains of the PS19 mice and the functional deficit observed for each mouse. Interestingly, astrogliosis showed a positive correlation (r = 0.51, p = 0.01) with the hyperactivity observed in the Open Field test (Fig. 5B). We also correlated reactive astrogliosis in individual brain regions, the hippocampus, cortex, hypothalamus, and the pons region of the brain stem, with functional deficits for each PS19 mouse. Astrogliosis in the cortex showed a correlation (r = 0.52, p = 0.02) with the hyperactive behavior of the PS19 mice observed in the Open Field test and with decline in body weight (r = −0.46, p = 0.04) (Supplementary Fig. 6). Furthermore, astrogliosis showed no correlation with the motor strength function and body temperature, features with statistically significant correlation with pathological soluble tau species.

Figure 5. Reactive astrogliosis correlates with the hyperactive behavior of PS19 mice.

(A) GFAP (red) intensity was used to measure levels of reactive astrogliosis in the brain. (DAPI, blue) (B) Reactive astrogliosis correlates with hyperactivity associated increase in locomotor activity in the Open Field. (C–F) No correlation observed between reactive astrogliosis in whole brain with (C-D) grip strength, (E) body temperature, and (F) body weight. Y-axis represents the quantification of a specific parameter on each behavioral or physiological test. The X-axis represents GFAP intensity.

We measured microglial reactivity by staining the PS19 brain sections with an anti-Iba1 antibody and correlated microgliosis in the brain with the observed functional deficits; microgliosis was more abundant in the PS19 mice than in the WT mice (Supplementary Fig. 5). However, there was no correlation between microgliosis and the behavioral or physiological deficits in 9-month-old PS19 mice (Fig. 6). When we correlated individual brain regions with functional deficits, we did not observe a significant correlation between levels of microgliosis in the cortex, hippocampus, hypothalamus, and pons region of brain stem with any functional deficits (Supplementary Fig. 7). To determine if gliosis is secondary to tau pathology in the PS19 mouse model, we correlated soluble and insoluble p-tau levels with astrogliosis and microgliosis. We did not observe any correlation between soluble and insoluble p-tau levels with gliosis (Supplementary Fig. 8).

Figure 6. Microgliosis does not show a correlation with functional deficits in 9-month-old PS19 mice.

(A) Iba1 (green) intensity was used to measure levels of microglia reactivity in the brain. (DAPI, blue) (B–F) No correlation was observed between microgliosis and (B) hyperactivity, (C-D) grip strength, (E) body temperature, and (F) body weight. Y-axis represents the quantification of a specific parameter on each behavioral or physiological test. The X-axis represents Iba1 intensity.

Taken together, these results suggest that glial activation, one of the major pathological hallmarks of tauopathies, could contribute to distinct functional impairments that are different from those associated with pathological tau.

4. DISCUSSION

While several animal models have enabled the elucidation of novel pathophysiological characteristics and functional deficits of tauopathies (Denk and Wade-Martins, 2009), there is still limited knowledge regarding which pathological features correlate with the observed deficits. The data presented here demonstrates that soluble tau species and astrogliosis, two of the major pathologies of tauopathies, correlate with distinct functional deficits in the PS19 mouse model. A detailed analysis of functional deficits in 9-month-old PS19 mice revealed that, at this age, there is a characteristic hyperactive behavior, decline in motor strength, and a deterioration in the physiological conditions marked by lower body temperatures, weight loss, and increase in the measures of frailty. Furthermore, these deficits do not show a correlation with each other, which suggests that these deficits are either independent of each other or that more than one pathology could be associated with the deficits independently. We identified that, in PS19 mice, tau pathology marked by a spectrum of soluble p-tau species and tau seeding bioactivity correlates with impairment in motor strength and thermal regulation. Astrocyte reactivity, on the other hand, showed a correlation with the hyperactive behavior of the PS19 mice. Our results suggest that two major pathologies of tauopathies, tau and gliosis, could possibly have an independent association with distinct functional deficits and that the spectrum of deficits manifested could be a result of the severity of different pathologies.

In Alzheimer’s disease, tau pathology initially appears in circumscribed regions and subsequently progresses throughout the brain in a fairly stereotypical pattern. This hierarchical pattern of progressive involvement of brain regions provides the basis for a neuropathological staging system proposed by Braak and Braak (1991), which is based on the detection of NFTs by silver impregnation (Braak and Braak, 1995). However, growing evidence suggests that NFT formation is neither necessary nor sufficient for neurodegeneration. For instance, in the brains of AD patients, neuronal loss in regions such as the superior temporal sulcus exceeds the number of NFTs more than seven-fold, suggesting that the majority of neurons probably die without having developed NFTs (Gomez-Isla et al., 1997). Moreover, in several lines of transgenic mice expressing WT or mutant human tau, including the PS19 mouse model, impaired synaptic function and cognitive deficits occur before or without any eventual NFT formation (Jankowsky and Zheng, 2017; Santacruz et al., 2005; Yoshiyama et al., 2007), suggesting that soluble tau species, rather than NFTs, are the toxic entities. Our results demonstrated that, in the PS19 mouse model, the levels of soluble tau species phosphorylated at Serine 214, an early event of phosphorylation (Augustinack et al., 2002; Wesseling et al., 2020), and phosphorylated at Threonine 217, a recently identified potent blood biomarker (Janelidze et al., 2020), correlate with motor strength impairments, suggesting that soluble tau species could play a role in the development of certain functional deficits in this model of tauopathy. Notably, soluble tau species were also associated with the development of functional deficits in the Tg4510 and JNPL3 mouse models of tauopathy (Berger et al., 2007). We did not observe a robust correlation between NFT pathology and functional deficits in the PS19 mice, suggesting the NFTs are not the primary mediators of toxicity in this model. A similar lack of correlation has been observed between NFTs and behavioral deficits in the Tg4510 model during later stages of tau pathology (Berger et al., 2007).

Several studies have proposed that tau pathology progresses through the brain in a prion-like manner of self-propagation (Aguzzi and Rajendran, 2009; Aguzzi et al., 2008; Prusiner, 1982). In this concept, physiological tau is recruited to be transformed into tau aggregates in a continuous process driven by template-directed misfolding or seeding (De Calignon et al., 2012; Lasagna-Reeves et al., 2012a). Despite the extent of the current knowledge (Devos et al., 2018), the exact cellular mechanism involved in tau propagation and the nature of the tau species involved in the spreading remain unclear. As reported by Holmes and colleagues (2014), we confirmed that the brain lysates from 9-month-old PS19 mice have strong seeding activity. Interestingly, previous studies demonstrated that seeding occurs as early as 2 months in PS19 mice (Holmes et al., 2014), a considerable amount of time before the appearance of synaptic impairments, glial activation, NFT pathology, and behavioral impairments (Yoshiyama et al., 2007). This early seeding suggests that the tau species with seeding activity may not be the species responsible for behavioral impairments or that the tau-seed at early stages could be structurally different from the seed present at later stages of pathology in the PS19 mouse model. Our analysis demonstrated that tau seeding correlates with the decrease in body temperature, suggesting a possible involvement of this pathological soluble tau species in a mechanism responsible for the impairment of thermal homeostasis in this model, which is a phenotype also observed in normal aging and is associated with AD (Whittington et al., 2010).

Overall, the results of our studies, correlating the behavioral deficits and different tau pathological species suggest that each of the tau species analyzed (soluble p-tau Ser214, soluble p-tau T217, and tau-seed) could be responsible for a variety of behavioral deficits in the PS19 mouse model. These interpretations are based on studies that support the notion that the structural diversity of tau aggregates could underlie some of the differences in symptoms and pathologies of neurodegenerative tauopathies (Gerson et al., 2016). One of the largest barriers to conclusive mechanistic investigations of tau proteins is the inconsistency in the tau species being studied (Cowan and Mudher, 2013), which is an important consideration since the aggregation state of tau is critical to its pathological function. Furthermore, it has also been demonstrated that tau can exhibit conformational differences within the same aggregation state, which could exert diverse downstream effects (Hyman, 2014; Sanders et al., 2014). This concept has been further supported by the results of a novel study that showed that different conformers arising from the same MAPT-P301L mutation drive distinct phenotypes in FTLD patients (Daude et al., 2020). Therefore, understanding the characteristics of different pathological tau species could be critical for the investigation of the toxicity of tau and for determining how each of these tau species may be responsible for distinct symptoms observed in neurodegenerative tauopathies.

Neuroinflammation is a typical pathology observed in several human tauopathies. Genome-wide association studies (GWAS) have enabled systemic identification of genes associated with the risk for developing tauopathies, many of which have immune-related functions (Didonna, 2020). The roles of microglia and astrocytes as major drivers of the neuroinflammatory response in tauopathies have been validated by several studies. For instance, reactive microglia are observed around NFTs in the AD brain as well as in PiD, PSP, and CBD (Henkel et al., 2004; Serrano-Pozo et al., 2011). Similarly, reactive astrogliosis, the process whereby astrocytes respond to pathology (Escartin et al., 2021), is observed in neurodegenerative tauopathies (Kovacs et al., 2016). Many of these diseases exhibit neuronal tau pathology parallel to astrocytic tau inclusions (Kovacs et al., 2016). Furthermore, the results of a recent study strongly suggested that tau pathology in astrocytes may significantly contribute to clinical symptoms (Mate De Gerando et al., 2021). These observations support the notion that neuroinflammation and the glial response may contribute significantly to the damage and degeneration of neurons and neurites, leading to dementia (Mcgeer and Mcgeer, 1995; Mcgeer and Rogers, 1992). Furthermore, a novel in vivo PET imaging study concluded that tau pathology and neuroinflammation could predict cognitive decline in patients with symptomatic Alzheimer’s disease pathology (Malpetti et al., 2020). These findings enhance the relevance of glial activation as a major contributor to dementia and support the idea of targeting tau and neuroinflammation with disease-modifying therapies for AD. In a neuropathological study, Vehmas and colleagues (2003) examined the association between microglia, astrocytes, Aβ, and tau and cognitive changes in clinically characterized subjects with pathological diagnoses of definite AD, possible AD, those defined as cognitively normal but with neocortical neuritic senile plaques, and age-matched controls. They observed that microglial activation occurred during the early stage of the pathogenesis of AD, whereas astrocytic activation significantly correlated with dementia, suggesting that astrocytes play a role during the later stages of the disease. Tau immunoreactivity was also a strong morphological correlate of dementia. Similarly, our neuropathological correlation studies demonstrated an association between astrogliosis and tau pathology with the behavioral deficits in the PS19 mouse model.

Notably, our studies demonstrated that astrogliosis correlates with the hyperactive phenotype, while pathological tau shows correlation mainly with the impairment of motor strength and thermal homeostasis, suggesting that glial activation could be responsible for a distinct set of symptoms from those triggered by tau aggregation in the PS19 tauopathy mouse model. Furthermore, in our study, glial activation did not show a correlation with tau pathology. The notion that tau and neuroinflammation can exert behavioral abnormalities and neurodegeneration independently of each other is supported by previous studies utilizing the PS19 mouse model, in which neurodegenerative phenotypes were rescued solely by decreasing tau pathology or neuroinflammation but not both (Largo-Barrientos et al., 2021; Leyns et al., 2017). In one interesting study using the PS19 mouse model, downregulation of TREM2, which is expressed predominantly in microglia, reduced microgliosis, lowered the levels of inflammatory cytokines, and decreased astrogliosis along with a significant decrease in brain atrophy (Leyns et al., 2017). Strikingly, no significant changes in tau accumulation were observed suggesting that glial activation and neuroinflammation can contribute to the neurodegenerative process in tauopathy without altering tau aggregation (Leyns et al., 2017). Unfortunately, no behavioral analysis was performed in this study to determine the sole contribution of neuroinflammation to the behavioral deficits in the PS19 mouse model. In yet another study utilizing the PS19 mouse model, the authors arrived at a different conclusion. In this study, a 50% downregulation of the expression of presynaptic protein synaptogyrin-3 decreased tau pathology in the PS19 mouse and strongly rescued defects in synaptic plasticity and working memory. Nevertheless, neither gliosis nor neuroinflammation was alleviated (Largo-Barrientos et al., 2021), suggesting that tau-induced neurodegeneration and cognitive impairment can be rescued even in the presence of major glial activation. The effect of synaptogyrin-3 downregulation was evaluated by Morris water maze, an assay that we did not perform in our study. It would be interesting to determine in future studies if downregulation of synaptogyrin-3 expression rescues other behavioral abnormalities known to be present in the PS19 mouse model, such as hyperactivity (Takeuchi et al., 2011; Wu et al., 2019), which correlates with astrogliosis according to our current findings. Overall, there are several lines of evidence that indicate that tau aggregation in animal models induces neuroinflammation, including microglial and astrocyte activation, and that inflammation enhances tau aggregation through a feed-forward mechanism, amplifying the neurotoxic insults (Bellucci et al., 2004; Didonna, 2020; Shi et al., 2017). Nevertheless, future studies will be necessary to determine if distinct behavioral and physiological deficits associated with neurodegenerative tauopathies are directly associated with tau aggregation or if they are independently associated with glial activation or specific neuroinflammatory responses.

In conclusion, in this study, we demonstrated that the PS19 tauopathy mouse model has a broad range of behavioral and physiological impairments. By performing correlation analysis, we demonstrated that an individual mouse could perform relatively poorly in one assay but fairly better in another, suggesting that each mouse has a unique behavioral and physiological profile. Our correlation analysis also demonstrated that a diverse set of pathological forms of tau, such as soluble p-tau aggregates and soluble tau species with seeding activity, correlate with the impairment of motor strength or temperature homeostasis, while reactive astrogliosis showed a correlation with the hyperactive phenotype. These results suggest that a diverse spectrum of pathological tau species could be responsible for the distinct symptoms that are observed in neurodegenerative tauopathies and that neuroinflammation could contribute to impairment independently from tau pathology. These observations enhance the necessity of a multi-targeted approach for the treatment of these neurodegenerative tauopathies that are characterized by the presence of a broad spectrum of symptoms and support the development of precision medicine treatments based on the specific symptoms of each patient.

Highlights.

• Pathological tau and astrogliosis correlate with diverse deficits in a tauopathy model

• A diverse spectrum of pathological tau species may underlie different symptoms

• Glial activation can independently contribute to functional deficits