Abstract
Tauopathies, such as Alzheimer’s disease (AD) and frontotemporal dementia (FTD), are characterized by formation of neurofibrillary tangles consisting of hyperphosphorylated tau. In addition to memory loss, patients experience behavioral symptoms such as agitation, aggression, depression, and insomnia. We explored the behavioral phenotype of a mouse model (rTg4510) carrying the human tau P301L mutation found in a familial form of FTD. We tested these mice in locomotor activity assays as well as in the Morris water maze to access spatial memory. In addition to cognitive impairments, rTg4510 mice exhibited a hyperactivity phenotype which correlated with progression of tau pathology and was dependent on P301L tau transgene expression. The hyperactive phenotype was characterized by significantly increased locomotor activity in a novel and in a simulated home cage environment together with a disturbed day/night cycle. The P301L-tau-dependent hyperactivity and agitative-like phenotype suggests that these mice may form a correlate to some of the behavioral disturbances observed in advanced AD and FTD.
INTRODUCTION
In Alzheimer’s disease (AD), neurofibrillary tangles (NFTs) consisting of hyperphosphorylated tau form part of a complex pathology together with amyloid-beta (Aβ) plaques. In other tauopathies such as frontotemporal dementia (FTD), NFTs are the primary pathological hallmark [1]. AD and FTD are progressive neurodegenerative disorders, which in addition to memory loss give rise to a variety of behavioral symptoms such as agitation, aggression, depression, and insomnia. These symptoms are debilitating and often the reason for institutionalization of dementia patients [2]. 47–85% of AD patients suffer from agitation, increasing with severity of the disease [3, 4]. It is not known how this progressive increase inagitated behavior is linked to Aβ plaques, NFTs, and the progression of pathology.
In transgenic amyloid-β protein precursor (AβPP) mouse models, behavioral phenotypes such as changed activity levels have been linked to Aβ pathology [5–7]. Although no direct association has been established between the pathology of amyloidogenic proteins and the behavioral phenotypes, it is believed that individuals that carry a higher load of frontal pathology have a higher incidence of agitation and aggressionsymptoms [8].
In a subset of patients with frontotemporal dementia with Parkinsonism linked to chromosome 17 (FTDP-17), a high frequency of agitation symptoms have been observed as well as in some advanced AD cases. These FTDP-17 patients are classified as behavioral variant, or bvFTD [9], and carry mutations in the MAPT gene [10]. Non-mnemonic behavioral changes have been reported in transgenic mouse models carrying human FTDP-17 MAPT mutations such as mutant P301L tau [11, 12]. The rTg4510 mice conditionally express mutant P301L tau driven by the Calcium/Calmodulin dependent kinase II alpha (CaMKIIα) promoter. rTg4510 mice express approximately 13 times more human mutant tau than endogenous murine tau predominately in the forebrain, resulting in a rapid progression of tau pathology with hyperphosphorylated tau, NFTs, cognitive deficits, and neurodegeneration [13, 14]. The rTg4510 strain is created by crossing two transgenic parental strains. One contains P301L tau downstream of an inducible tetracycline-operon-responder (TRE) promoter. The second contains a tetracycline-responsive transcriptional activator (tTA) driven by the CaMKIIα promoter. Therefore, the transcription of mutant P301L tau in rTg4510 mice can be switched off by doxycycline [14]. When transgene tau expression is suppressed hyperphosphorylated tau is prominently reduced, and the severe cognitive deficits reversed [14, 15]. In addition to cognitive impairment, some recent studies reported non-mnemonic behavioral changes such as increased exploratory activity in the rTg4510 mice [16, 17]. However, others reported normal exploratory activity [18].
We identified a non-mnemonic behavioral defect in rTg4510 mice and explored hyperactivity dependent on P301L tau expression as a model for agitation. To address the increase in locomotor activity, we tested mice at different ages in a novel and in a simulated home cage environment. In the latter, we investigated changes in the locomotor activity pattern during the light-dark phases. We investigated the relationship between tau pathology and behavioral defects by suppressing tau transgene expression. In addition to the non-mnemonic behavior, spatial reference memory was accessed using the Morris water maze (MWM) and tau pathology was evaluated by histological and biochemical analysis at the end of the behavioralstudies.
MATERIALS AND METHODS
Mice
The rTg4510 mice express the TauP301L mutation in the MAPT gene associated with FTDP-17. The generation of rTg4510 mice was described previously [14]. Briefly, a human tau cDNA with the P301L mutation (4R0N TauP301L) was placed downstream of a TRE construct. To activate the transgene, the responder had to be co-expressed with an activator construct, consisting of the tetracycline conditional gene expression system (tTA). The tTA activatorsystem was placed downstream of the CaMKIIα pro-moter thus restricting the expression of TRE mainly to forebrain structures. The tau transgene responder was expressed in the FVB/N (Taconic) mouse strain, and the tTA activator system was maintained on 129S6 (Taconic) mouse strain. Their F1 progeny carried responder and activator transgenes (rTg4510), necessary for the expression of the tau transgene, or thetau responder transgene alone (tau only) or the tTAactivator transgene alone (tTA), along with non-transgenic (FVB/129) littermate mice (Wt). Only F1mice were used for the experiments. All mice werebred at Taconic, Denmark and genotyped by theanalysis of tail DNA using the primer pair’s 5′-GATTAACAGCGCATTAGAGCTG-3′ and 5′-GCATATGATCAATTCAAGGCCGATAAG-3′ for the tTA activator transgene and 5′-TGAACCAGGATGGCTGAGCC-3′ and 5′-TTGTCATCGCTTCCAGTCCCCG-3′ for the mutant tau responder transgene. Mice were group-housed and received water and food (Brogaarden, Denmark) ad libitum as well as enrichment materials. To suppress transgene expression, doxycycline (dox) (200 ppm) enriched chow (Harlan, Germany) was administered to mice ad libitum during the course of the study. In addition, dox (1.5 mg/ml) was administered for the first 2 days in drinking water in 4% sucrose (controls were given 4% sucrose only). The light/dark cycle was 12 h; room temperature was 21 ± 2°C and a relative humidity of 55% ± 5% . Experiments were run in accordance with Danish legislation on experimental animals (license no. 2009/561-1596).
Experimental design
We conducted three separate studies; one for age dependent characterization of cognition and locomotor function of the rTg4510 mice and two for doxycycline dependent prevention of the observed deficits. In the characterization study mice were tested in the 3-h locomotor activity assays at 6–10 weeks (rTg4510 n = 89, Wt n = 88), 16 weeks (rTg4510 n = 74, Wt n = 77), 20 weeks (rTg4510 n = 58, Wt n = 66), 24 weeks (rTg4510 n = 32, Wt n = 26), 32 weeks (rTg4510 n = 9, Wt n = 11), and 48 weeks (tTA n = 25). At 32-week old, 25 tTA and 18 rTg4510 mice were tested in a 72-h locomotor activity assay. Following the 3-h locomotor activity test, 2 mice from each time point were euthanized and brains removed and processed for tau analysis. At 16-week old, male mice (rTg4510 n = 10, tTA n = 10, Wt n = 10) were tested in the MWM.
For the first doxycycline prevention study, half the mice (rTg4510: n = 55, Wt: n = 60) received dox-enriched chow from 6 weeks of age, while the other half (rTg4510: n = 55, Wt: n = 57) received normal chow. These mice were subjected to the 3-h locomotor activity test at 24 weeks of age, 8 weeks after a subset of the mice had been tested in the MWM at 16 weeks of age (rTg4510 on dox: n = 23, rTg4510 normal chow: n = 21, Wt on dox: n = 16, Wt normal chow: n = 17). Following hyperactivity testing, 4 mice from each group were euthanized and brains harvested for tau analysis. In the second doxycycline study, mice received dox enriched chow (rTg4510: n = 10, Wt: n = 8) and normal chow (rTg4510: n = 19, Wt: n = 8) from 6 to 32 weeks of age and were subjected to the 3-h locomotor activity test at 32 weeks of age. 5 mice from each group were euthanized and brains harvested for tau analysis of the amygdala.
Tissue collection
Mice were euthanized by cervical dislocation. Following decapitation, brains were quickly removed and divided sagittally down the midline to yield two hemispheres. One hemisphere was parted from the cerebellum and hind brain and the forebrain containing cerebral cortex and hippocampus was snap-frozen on dry ice and stored at –80°C until use for biochemical analysis. The other hemisphere was immersion-fixed in 4% paraformaldehyde for 36 hours and processed for histology.
Western blotting
Tissue extraction and western blotting was performed as described previously [19]. Briefly, tissues were homogenized in 10 volumes of Tris-buffered saline containing protease and phosphatase inhibitors as follows: 50 mM Tris/HCl (pH 7.4); 274 mM NaCl; 5 mM KCl; 1% protease inhibitor mixture (Roche); 1% phosphatase inhibitor cocktail I & II (Sigma); and 1 mM phenylmethylsulfonyl fluoride (PMSF). The homogenates were centrifuged at 27,000 × g for 20 min at 4°C to obtain supernatant (S1) and pellet fractions. Pellets were re-homogenized in 5 volumes of high salt/sucrose buffer (0.8 M NaCl, 10% sucrose, 10 mM Tris/HCl, [pH 7.4], 1 mM EGTA, 1 mM PMSF) and centrifuged as above. The supernatants were collected and incubated with sarkosyl (1% final concentration; Sigma) for 1 h at 37°C, followed by centrifugation at 150,000 × g for 1 h at 4°C to obtain sarkosyl-insoluble pellets, referred to as P3 fraction. The P3 pellet was resuspended in TE buffer (10 mM Tris/HCl [pH 8.0], 1 mM EDTA) to a volume equivalent to half of the original volume used for the brain homogenates. Fractionated tissue extracts S1 and P3 were dissolved in SDS-sample buffer containing DTT (100 mM). The heat-treated samples (95°C for 10 min) were separated by gel electrophoresis on 4–12% Bis-Tris SDS-PAGE gels (Invitrogen) and transferred onto PVDF membranes (BioRad Laboratories, Hercules, CA). After blocking with a solution containing 5% nonfat milk and 0.1% Triton-X100 in TBS, the membranes were incubated with various antibodies. Mouse monoclonal tau5 antibody (Abcam, Cambridge, UK), polyclonal tau antibodies E1 (raised against amino acids 19–33 (GLGDRKDQGGYTMHQ) of the longest isoform of human tau 2N4R), and pS396 tau (Invitrogen) and GAPDH (Meridian Life Science Inc., Memphis, TN) for normalization were used. Tau5 is a pan tau antibody and recognizes human and murine tau at non-phosphorylated and phosphorylated epitopes. E1 antibody recognizes exclusively human tau at non-phosphorylated and phosphorylated epitopes, while pS396 tau antibody recognizes the phosphorylated epitope of murine and human tau. Membranes were washed and incubated with peroxidase-conjugated anti-mouse IgG and anti-rabbit antibodies (1:5000; Jackson ImmunoResearch, West Grove, PA). Bound antibodies were detected using an enhanced chemiluminescence system (ECL PLUS kit; PerkinElmer). Quantitation and visual analysis of Western blot immunoreactivity was performed with a computer-linked LAS-4000 BioImaging Analyzer System (Fujifilm, Tokyo, Japan) and Multi Gauge v3.1 software (Fujifilm).
Histology
Brain tissues were embedded in paraffin and sectioned at 4μm. Antigen retrieval was applied by boiling in 10 mM Citrate buffer, pH 6, in the microwave oven. Immunohistochemistry was performed using standard procedure in a DAKO Universal Immunostainer with p-S202/T205 antibody (clone AT8; ThermoScientific). Further, tissue sections were stained with Gallyas silver stain for tangle pathology and with thionin stain for hippocampal morphology. Quantitative analysis of tau burden in the amygdala was measured as AT8-immunoreactivity using Image-Pro Plus 7.0 software (media cybernetics, MD, USA). Using manual color cube RGB a mask was defined and the specific optical density of the brown chromogen as a percentage of burden within the annotated region of interest was measured. Throughout procedures, the investigator of the specimen was blind to the genotype, age, and treatment of the animals.
Morris water maze
The MWM study was performed with 16-week-old male mice (rTg4510, tTA, and Wt) and male and female mice from the doxycycline reversal study. The water maze consisted of a circular tank of 120 cm in diameter and 40 cm deep with white walls. The white platform was 8 cm in diameter and adjustable in height. The water was made opaque by adding 500 ml optic white (E308). Mice were tracked with Ethovision 3.0.15 software (Noldus Technologies, Netherlands). The maze was divided into four quadrants and a circular outer zone (encompassing the area 10 cm from the maze wall) for further analysis. The release points changed between trials. Following 5 days of handling (60 s, including 20 s exposure to water), the animals were subjected to a 10-day water maze protocol. It consisted of 6 days of acquisition (hidden platform) with 4 trials per day with an inter-trial-interval of 40 min. During the acquisition, the platform was hidden 1.2 cm below water surface. Along with the acquisition trials, the animals received 5 probe trials (24 h after acquisition trials 8, 12, 16, 24, and 72 h after trial 24). The trials lasted 30 s and the animals were always released from the same point. 24 h after the last probe trial, the animals were subjected to 6 visual trials within one day (day 10). The platform was placed in a new position and made visible with a flag in the middle [13].
3-h locomotor activity test in a novel environment
The activity assay consisted of a large cabinet with 16 Makrolon activity cages (20 × 35 × 18 cm). The activity cages contained no bedding and were equipped with 5 × 8 infrared light sources and photocells placed 1.8 cm above the cage floor. Mice were housed in the experimental room 24 h prior to testing. On the day of testing they were placed individually in Makrolon cages, and were left for 3 h without any disturbances. After the test, the mice were returned to their original home cages. The frequency with which the animal crossed the photo beams was used as a measure of activity (pooled in 5 min bins). Total activity counts representing mean + 3xSD from the Wt population during the 3 h activity test period (>9,873 counts) were defined as hyperactivity. Data were collected using UMOTWin (Ellegaard systems, Denmark).
72-h locomotor activity test in home cage environment
For the home cage simulation, the mice were placed in the same equipment as described above and monitored for 72 h. After the first hours of habituation to the cage, this setup resembles the home cage behavior of the individual mice. Each Makrolon cage was filled with sawdust in the bottom and the mice had access to food and water throughout the study. The light/dark cycle was 12 h. Activity counts were pooled in 1-h intervals. Total activity counts representing mean + 3xSD from the tTA population during the 72-h activity test period (>97,411 counts) were defined as hyperactivity. Data were collected and analyzed with UMOTWin.
Statistical analysis
All data was analyzed with Sigmaplot version 11 (Systat Software, San Jose, CA). All graphs are presented as mean ± SEM. Water maze data were analyzed by RM ANOVA and Bonferroni post hoc tests. The difference in the number of hyperactive mice between the groups was analyzed with Chi2 test.
RESULTS
Age-dependent increase in tau pathology in rTg4510 mice
rTg4510 mice are known to display tau pathology with age-dependent increase in intra-neuronal hyperphosphorylated tau and NFT formation at 16 weeks of age [13, 14]. Here, histological and biochemical analysis were performed on brains from rTg4510 mice aged 6, 16, 24, and 32 weeks (Figs. 1 and 2). Paraffin sections of the CA1 subregion of the hippocampus, cortex, and amygdala from rTg4510 and non-transgenic littermates were analyzed for the presence of phospho-tau (P-tau) species using the AT8 antibody (Ser202/Thr205) and NFT by Gallyas silver staining (Fig. 1). In agreement with our previous study in rTg4510 mice [20], we observed only very limited AT8-immunoreactivity in the CA1 at 6 weeks while P-tau presence was pronounced in neurites and cell bodies at 16 and 24 weeks of age both in the CA1 and cortex (Fig. 1A). P-tau was detected only in neurites (axons and dendrites) at 6 weeks of age in cortex (Fig. 1A). There was no staining for P-tau in 24-week-old non-transgenic littermates (data not shown). NFTs were observed in the CA1 and cortex from 16-week-old rTg4510, not in 6-week-old mice (Fig. 1A). Progression of tau pathology was in line with previous reports in rTg4510 mice [13, 14]. Similar to CA1 and cortex, we detected in the central and basolateral amygdala pronounced AT8-immunoreactivity in neurites and cell bodies and NFTs in 32-week-old rTg4510 mice (Fig. 1C). Pathological tau burden measured as AT8-immunoreactivity in the basolateral amygdala of 5 individual 32-week-old rTg4510 mice averaged 19.6 ± 10.2% of area (data not shown), a level which was comparable to rTg4510 mice in a recent report [16].
Forebrain homogenates were isolated into a soluble fraction (S1) and a sarkosyl-insoluble fraction (P3) as characterized before [19] and examined for expression of pan tau (endogenous murine tau and transgenic human tau) using the tau-5 antibody, transgene human tau using the human tau specific E1 antibody, and for P-tau species at the S396 epitope. At all ages, 55 kDa tau species were observed by western blotting, representing the human 4R0N transgenic tau in the soluble fraction (Fig. 2A). As expected, the 55 kDa human tau species were detected by the pan tau(tau5), human tau (E1), and the pS396 tau antibody. Additionally, we observed endogenous murine tau in the non-transgenic mice detected with the pan tau5 and pS396 antibody (Fig. 2A), indicating physiological phosphorylation at the S396 residue of both murine tau and 55 kDa human tau species. We observed that hyperphosphorylated 4R0N tau with P301L mutant was displayed as mobility shifted tau of 64 kDa and 70 kDa in the soluble and the sarkosyl-insoluble fractions on SDS page (Fig. 2A). The sarkosyl-insoluble 64 kDa and 70 kDa tau species isolated in P3 are regarded as the biochemical NFT equivalent [13, 14]. The soluble 64 kDa tau species in S1 are regarded as the biochemical pre-tangle tau equivalent [13, 19]. At 6 weeks of age, no 64 and 70 kDa tau species were detected in any of the fractions. In line with our histological findings at 16 weeks of age, 64 and 70 kDa tau species were observed in P3 detected by pan tau (tau5), human tau (E1), and the pS396 tau antibody, indicating increase in NFTs. As tau pathology became more dominant at 24 weeks of age, a transition from 55 kDa to 64 kDa was observed in S1 (Fig. 2A). When tau transgene expression was switched off by doxycycline treatment in 6-week-old animals, which were kept on doxycycline chow for 18 weeks until they reached 24 weeks of age, we observed a significant reduction in human tau levels both by histology (Fig. 1B) and biochemical analysis (Fig. 2B). In the soluble fraction, the amount of human transgene 55 kDa tau was reduced by approximately 75% detected with the tau5, E1, and the pS396 antibody (Fig. 2B) in line with previous reports [14, 21]. In the soluble fraction, we observed a stronger pS396-tau signal in rTg4510 mice treated with doxycycline than in non-transgenic mice which can be explained by the 2-3 fold overexpression of the human transgene in the presence of doxycycline [14]. Importantly, in doxycycline-treated rTg4510 mice, no soluble and sarkosyl-insoluble 64 and 70 kDa tau species were detected by western blot analysis using tau5, E1, and pS396 antibodies (Fig. 2B). P-tau measured by AT8-immunoreactivity was strongly suppressed and no NFTs were detected in the CA1, cortex, and amygdala in 24 and 32-week-old rTg4510 mice, respectively, which received doxycycline from 6 weeks of age (Fig. 1B, C). Together, these findings indicate that hyperphosphorylated tau was abolished in 32-week-old rTg4510 mice when tau transgene suppression was initiated from 6 weeks of age.
Tau-dependent cognitive deficits in rTg4510 mice
rTg4510 mice displayed impaired performance in acquisition and spatial reference memory in the MWM at 16 weeks of age (Fig. 3) in agreement with previous reports [13, 22]. During the acquisition phase of the MWM, the latencies to reach the hidden platform were significantly longer for the rTg4510 mice than for non-transgenic littermates (Two-way RM ANOVA: F(genotype) = 123.371, p < 0.001, Fig. 3A). Furthermore, female rTg4510 mice had significantly longer latencies to reach the platform compared with male rTg4510s on the last test days (RM ANOVA: F(sex*day) = 5.508, p < 0.001, Post hoc bonferroni corrected tests: tsex within day5 = 2.481, p = 0.014, tsex within day6 = 5.518, p < 0.001), indicating an additional learning impairment in female rTg4510 mice as previously reported [22]. As the gender difference was only observed on the last 2 days and not in the Wt population, we continued analyzing the rTg4510 mice as one group. To control for possible differences in swim speed, we also analyzed distance to hidden platform. Results from this parameter compared well with latency. The swim distance to reach the hidden platform was significantly longer in rTg4510 mice compared to non-transgenic littermates (Two-way RM ANOVA: F(genotype) = 81.799, p < 0.001, Fig. 3B). Thigmotaxis (time spent in the outer zone of the maze) decreased for all groups during the acquisition phase. However, rTg4510 mice showed increased thigmotaxis compared with the non-transgenic littermates (Two-way RM ANOVA: F(genotype) = 37.876, p < 0.001, data not shown). In the retention phase of spatial reference memory during the probe trial we observed a significant interaction between the effect of genotype and quadrant (Two-way RM-ANOVA: F(genotype*quadrant) = 22.960, p < 0.001). Non-transgenic mice preferred the target quadrant, while rTg4510 mice showed no preference for the target quadrant compared to the other three quadrants (Fig. 3D). In addition, non-transgenic littermates spent significantly more time in the target quadrant (t-test: tWt vs . Tg within N = 7.247, p < 0.001), while rTg4510 mice spent significantly more time in the outer rim of the maze displaying increased thigmotaxis (Fig. 3C). Littermate tTA mice have recently been reported to display behavioral deficits [23]. Therefore, we investigated the performance of 16-week-old tTA and non-transgenic littermates in acquisition and spatial reference memory in the MWM. We did not observe any difference between both genotypes (T-test: Acquisition: t(genotype) = 0.289, p = 1.000, Probe trial: t(genotype in north) = 0.413, p = 1.000, data not shown), indicating that tTA littermates are not impaired in learning and memory in the MWM.
The cognitive deficits of the rTg4510 mice were previously reported to be dependent on the presence of hyperphosphorylated tau [13–15]. We suppressed tau transgene expression by doxycycline and observed improved performance in acquisition and spatial reference memory in the MWM in rTg4510 mice. 16 non-transgenic littermates and 23 rTg4510 mice fed doxycycline from 6 weeks of age were included in the MWM at 16 weeks. During the training phase, the rTg4510 mice on doxycycline performed significantly better than the untreated rTg4510 mice (Two-way RM-ANOVA: F(doxycycline*day) = 4.027, p = 0.002, Fig. 3A). However, they were still significantly impaired compared to the non-transgenic mice (Two-way RM-ANOVA: F(genotype within dox) = 48.225, p < 0.001). In the retention phase of spatial reference memory, we observed that rTg4510 mice on doxycycline chow spent significantly more time in the target quadrant compared to untreated rTg4510 mice (Two-way RM ANOVA: F(doxycycline*quadrant) = 14.336, p < 0.001; Fig. 3D). Furthermore, doxycycline treated rTg4510 mice showed less thigmotaxis during the probe trial (Fig. 3C). As expected, doxycycline treatment did not affect the performance of the non-transgenic mice (Two-way RM ANOVA: F(doxycycline) = 1.021, p = 0.320).
Age-dependent increase in hyperactivity in rTg4510 mice
In addition to the recognized cognitive deficits, we identified hyperactivity, a non-mnemonic behavioral change in the rTg4510 mice. When tested for locomotor activity, we observed that some rTg4510 mice displayed an abnormal increase in activity levels with an average of 54,468 total activity counts in a 3-h locomotor activity test in a novel cage environment. For comparison, non-transgenic littermates displayed between 102 and 9,485 total activity counts in this setup. A similar range of activity levels was observed for tTA mice, which did not display increased locomotor activity (Fig. 4A, B). We defined hyperactivity as total activity counts above mean + 3xSD of the non-transgenic population which was >9,873 total activity counts in the 3-h test in a novel cage environment. The number of rTg4510 animals with a hyperactive phenotype increased with age. We observed the first hyperactive animals at 16 weeks of age where 15% displayed hyperactivity when placed in a novelenvironment, increasing to 24% at 20 weeks of age and to 62% in the 24 weeks of age and older group (Fig. 4A). We measured locomotor activity in rTg4510 up to 32 weeks of age where we observed 74% hyperactive animals (Fig. 4A). The absolute level of hyperactivity, measured by the total activity counts, could vary for individual animals with increasing age, but a hyperactive animal maintained its increased locomotor activity through the entire study length (data not shown). There was no statistical difference between the numbers of hyperactive female and male rTg4510 mice (t-test: t = 0.08059, p = 0.9384). Unexpectedly, the hyperactivity phenotype did not correlate with the performance of individual animals in the MWM test. At 16 weeks of age when we accessed the cognitive performance of the rTg4510, both hyperactive and non-hyperactive rTg4510 animals were equally impaired in learning and memory (Fig. 3, and data not shown). Mortality was not increased in hyperactive rTg4510 mice (data not shown).
We were interested whether the abnormal locomotor activity levels were aligned with the normal day/night cycle of the animals. Therefore, 32-week-old mice were tested in a simulated home cage environment with a light/dark cycle of 12 h continuously for 72 h. Hyperactivity in the simulated home cage environment was defined as >97,411 total activity counts (mean + 3xSD of the tTA population) over the 72-h period. As expected, tTA mice had normal locomotor activity during the 72 h with an average of 10,266 total counts during the light phases increasing to an average of 13,811 total counts during the dark phases, the active phases of the mice (Fig. 4B, zoom). 81% of the rTg4510 at 32 weeks of age presented hyperactivity (Fig. 4B). Mice showed the highest activity levels during the dark phase with >232,218 total counts and maintained hyperactive during all three dark phases (Fig. 4B). Interestingly, the activity patterns of the hyperactive animals were abnormal as they also included bursts of heightened activity during the light phase, the normally inactive phase. We observed more than 53,634 total counts in the light phase of the hyperactive rTg4510 (Fig. 4B), indicating a disturbed day/night cycle.
Tau-dependent hyperactivity phenotype in rTg4510 mice
Next, we investigated whether hyperactivity similar to the cognitive deficits was dependent on hyperphosphorylated tau. Therefore, we tested 117 non-transgenic littermates and 110 rTg4510 mice on normal or doxycycline chow at 24 weeks of age in the 3-h locomotor activity test in the novel environment. 60 non-transgenic littermates and 55 rTg4510 mice fed doxycycline from 6 weeks of age were included in the locomotor activity testing. From our previous findings, we expected that 59% of the rTg4510 mice on normal chow to be hyperactive at 24 weeks of age. As anticipated, we did not observe hyperactive animals in the non-transgenic groups, and there was no effect of doxycycline on activity levels of the non-transgenic animals (Fisher’s exact test: p = 1.000). We observed 46% hyperactive animals in the rTg4510 group on normal chow at 24 weeks of age, confirming our previous results. In the doxycycline fed rTg4510 group, significantly fewer (only 5) mice displayed hyperactivity (Fisher’s exact test: p < 0.001; Fig. 4C), suggesting that the hyperactivity phenotype is linked to expression levels of P301L tau. Human tau expression was reduced by approximately 75% (Fig. 2B), and we detected no significant difference in the levels of human 55 kDa tau from the soluble fraction in the 5 hyperactive and 5 non-hyperactive doxycycline treated rTg4510 (data not shown). Hyperphosphorylated 64 and 70 kDa tau species were not present in brain fractions from doxycycline treated rTg4510 (Fig. 2B) neither in the brain fractions from the 5 hyperactive mice (data not shown). Based on these findings, we conclude that doxycycline treatment was effective in all animals and that hyperactivity is linked to the presence of hyperphosphorylated tau in rTg4510 mice. Though, other factors could possibly also be involved in the hyperactive phenotype in rTg4510 mice since a few animals still displayed hyperactivity under effective transgene tausuppression.
DISCUSSION
We identified a non-mnemonic behavioral defectin the rTg4510 mice in addition to the well-knowncognitive deficits. Hyperactivity was dependent on P301L tau expression and included abnormal day/night activity pattern in rTg4510 mice.
We characterized progressing tau pathology which was evident at 16 weeks of age with hyperphosphorylated tau and NFT formation (Figs. 1A and 2A) as we reported previously [20]. We determined pre-tangle and tangle tau by histology and their biochemical equivalents in the soluble and sarkosyl-insoluble fractions by western blot analysis. Hyperphosphorylated 64 kDa tau species detected in the soluble fraction represent pre-tangle tau while hyperphosphorylated 64 and 70 kDa tau species isolated in the sarkosyl-insoluble fraction represent NFT-tau [13, 19]. At 6 weeks of age, rTg4510 mice expressed human tau, but did not present hyperphosphorylated 64 and 70 kDa tau species or tangles (Figs. 1A and 2A). We characterized this age group as devoid of hyperphosphorylated tau. At 16 weeks of age, hyperphosphorylated 64 and 70 kDa tau species and tangle tau appeared, increasing in the 24-week-old animals, which displayed mature tangle pathology (Figs. 1A and 2A). The observed appearance and progression of tau pathology was fully in agreement with previous reports investigating the rTg4510 model [13–15, 24]. When tau transgene expression was suppressed with doxycycline initiated at 6 weeks of age, P-tau presence was highly reduced in 24-week-old mice (Figs. 1B and 2B) and neither 64 and 70 kDa tau species (Fig. 2B) nor tangle tau (Fig. 1B) was detected. These findings indicate that doxycycline treatment was effective in preventing appearance of hyperphosphorylated tau and in agreement with previous reports [13, 19].
In association with the presence of hyperphosphorylated tau, we observed an impaired performance in spatial reference memory test in the MWM in rTg4510 mice at 16 weeks of age (Fig. 3). Our mice obtained environmental enrichments which have been reported to improve cognitive function in rodents [25]. Despite environmental enrichments, rTg4510 mice displayed impairments in all aspects of the MWM (Fig. 3). These findings are in agreement with most previous reports showing that rTg4510 mice display deficits in both learning and memory without impairments in visual capabilities [13, 22]. One study reported only deficits in memory and not in learning in rTg4510 mice [15]. Our data additionally confirm the previously reported severe learning impairment in female compared to male rTg4510 mice [22]. We observed increased thigmotaxis in the rTg4510 mice (Fig. 3C), swimming along the wall of the maze, as also reported by others [22]. Thigmotaxis can indicateincreased anxiety and/or deficits in acquiring new search strategies [26]. We hypothesize that the increased thigmotaxis was connected to the inability to progress the search strategy in the maze, since we found an inverse correlation between learning and thigmotaxis in rTg4510 mice (Fig. 3A–C). We confirmed previous reports that the cognitive deficits were reversed by doxycycline treatment and dependent on hyperphosphorylated tau [14, 15]. We observed significant improvements in learning and in spatial reference memory, and decreased thigmotaxis in 16-week-old rTg4510 mice which received doxycycline from 6 weeks of age (Fig. 3). Under our doxycycline regimen, transgene tau expression was reduced to 75% of the expression level in untreated rTg4510 mice, and similar to a previous report neither hyperphosphorylated 64 and 70 kDa tau nor tangles were detected (Figs. 1B and 2B) [14]. However, complete reversal of the cognitive impairment to the level of non-transgenic littermates was not achieved (Fig. 3). In addition to the 2-3 fold tau overexpression in the rTg4510 mice in the presence of doxycycline [14], leakiness in the TRE controlled P301L tau expression has been described recently [21, 27]. In the tau transgene responder mice containing only the TRE-P301L-Tau gene, which was not dependent on the tTA activator system, P301L tau expression has been observed [21, 27]. The levels of P301L tau expressed in the tau transgene responder mice was only 2-3% of that of the untreated rTg4510 mice [21], and mice did not display pathological tau [21, 27] or cognitive impairments [27]. However, leaky expression of the TRE-P301L-Tau can lead to a continuous low tau transgene expression in the rTg4510 mice independent of the tTA activator and not suppressible by doxycycline. Theoretically, such constant tau transgene expression could lead to accumulation of intermediate tau other than the detected hyperphosphorylated 64 and 70 kDa tau species in rTg4510 mice, which also could be involved in the manifestation of behavioral deficits. A doxycycline-independent leaky tau expression would be one possible explanation why full reversal of the cognitive deficits was not reachable in rTg4510 mice.
We identified a non-mnemonic behavioral defect in the rTg4510 mice, hyperactivity, which was characterized by abnormally increased locomotor activity both in a novel and in a simulated home cage environment (Fig. 4A, B). We observed significantly increased activity levels in rTg4510 mice, which were 5–100 folds above the level of non-transgenic littermates in the novel cage environment (Fig. 4A). To investigate activity pattern during the normal 12-h light/dark cycles, locomotor activity was accessed continuously for 72 h in the simulated home cage environment. We recognized prominent changes in the day/night activity pattern in hyperactive rTg4510 mice (Fig. 4B). Hyperactivity was most pronounced during the active phase (dark phase) but extended into the inactive phase (light phase) of the animals. Animals showed bursts of abnormally increased activity during the inactive phase, which strongly indicates a disturbed day/night cycle in the hyperactive rTg4510 mice. We suggest that this hyperactivity phenotype with a distinct day/night pattern has some similarities to an agitation behavior observed in some AD patients [28, 29]. A subset of AD patients displays symptoms of agitation which can occur alone or in combination with aggression [28, 30]. The non-aggressive agitation behavior is characterized by abnormal wandering and shifted day/night activities [28] bearing some resemblance to the hyperactivity phenotype we observed in the rTg4510 mice.
Hyperactivity appeared equally in male and female rTg4510 mice and was both dependent on age and expression of P301L tau in the rTg4510 mice. The hyperactivity phenotype appeared in a subset of mice that was increasing in parallel with the progression of tau pathology. Hyperactivity was developing from 16 weeks of age and was pronounced at 20–24 weeks of age (Fig. 4A), coinciding with mature tangle pathology (Fig. 1A). We investigated animals up to 32 weeks of age where 81% displayed the hyperactive phenotype (Fig. 4B) and mature tau pathology in CA1 and cortex (data not shown) and amygdala (Fig. 1C). Hyperactivity was not observed in tTA littermates at 48 weeks of age (Fig. 4A), indicating that this defect is not caused by the background strain but dependent on P301L tau expression. Not all rTg4510 mice developed a hyperactive phenotype at the same time. Additional factors are likely to influence the susceptibility of the individual animal to develop hyperactivity and could determine when during the progression of tau pathology this particular behavioral phenotype will manifest in the animal.
Switching off transgene expression by doxycycline prevented hyperphosphorylated tau (Figs. 1B, C and 2B) and significantly reduced the number of rTg4510 mice with the hyperactive phenotype (Fig. 4C). However, 5 rTg4510 mice treated with doxycycline still displayed the hyperactive phenotype. Doxycycline treatment was measured to be effective in all animals, since we did not detect any 64 and 70 kDa tau species in the 5 hyperactive animals nor did we find any significant difference in the 55 kDa tau levels in the 5 hyperactive and 5 non-hyperactive animals (data not shown). Both behavioral deficits, impaired cognition and hyperactivity, were significantly improved in rTg4510 mice under doxycycline treatment but not completely reversed to non-transgenic littermate levels. We speculate that the doxycycline-independent leaky tau expression in the rTg4510 model could play a role in this phenomenon. However, other factors independent of tau transgene expression could also be involved in the manifestation of the hyperactivity phenotype in rTg4510 mice, since we observed a few animals displaying hyperactivity in the absence of hyperphosphorylated tau.
Increased exploratory activity has been reported in some [16, 17], but not all [18], rTg4510 mouse studies. In our rTg4510 mice, both parent background strains FVB/N and 129S6 were bred at Taconic which deviates from the rTg4510 mice in previous studies where the FVB/N strain was bred at Charles River [13–19, 22]. A change in breeder can theoretically cause a shift in phenotype in genetically identical animals. However, both in terms of tau pathology and MWM behavior, our rTg4510 were very similar to rTg4510 derived from the original colony containing the FVB/N background strain from Charles River [13–15, 22]. Recently, increased exploratory activity was reported in rTg4510 mice containing the Charles River FVB/N background strain [16, 17]. Based on these findings we argue that rTg4510 mice can display non-mnemonic behavioral changes, irrespective whether the FVB/N background strain originates from Charles River or Taconic.
An increase in locomotor activity has been observed in other transgenic mice overexpressing tau or AβPP [5–7, 12], indicating that these behavioral changes are likely linked to neuropathology rather than to a direct effect of the protein overexpression. This is further supported by our finding that all rTg4510 mice overexpressed transgene tau but not all displayed the hyperactive phenotype simultaneously. Similar to our findings in the rTg4510 mice, different mouse models that carry the P301L mutation have shown changes in locomotor activity levels [11, 12]. One study reported both increased activity levels along with impaired cognitive abilities [11]. Taken together, we suggest that common tau dependent mechanisms may underlie the development of hyperactivity in tau transgenic animals at least with the P301L mutation.
These tau dependent mechanisms might be similar in some tauopathies and could help explain the abnormal non-mnemonic behaviors observed in these patients. In tauopathies, a link between pathological changes in the frontal lobe and agitation behavior in several dementia states has been demonstrated in several studies [4]. In particular, the orbitofrontal cortex (OFC) has been implied in agitation and aberrant motor behavior in AD patients. It has been shown that levels of NFTs in this specific area correlates with both agitation and aberrant motor behavior [8]. Furthermore, functional evaluations through fMRIs have shown that dysfunction in the OFC was associated with agitation in AD and FTD patients [8, 31]. Similar neuroanatomical pathways are thought to be involved in non-mnemonic behavior in rodents [32]. It is hypothesized that abnormal non-mnemonic behavior such as anxiety, agitation, and aggression can be mediated through loss of OFC regulation on the medial amygdala [33]. The amygdala plays an important role in several aspects of non-mnemonic behavior, including attention and perception, and inhibition/disinhibition [34]. We speculate that in rTg4510 mice the amygdala or the particular feedback pathway from the OFC to the amygdala could be disturbed in hyperactive animals. Disturbance could arise either directly through neuronal death or by synaptic dysfunction induced by hyperphosphorylated tau. In line with this, another group recently reported that rTg4510 mice display amygdala-dependent behavioral dysfunctions such as increased locomotor activity [16]. The CaMKIIα promotor can drive transgene expression in the amygdala [35] and tau transgene expression and tau pathology in the amygdala has been confirmed in rTg4510 mice (Fig. 1C) and [16]. Age-dependent progression of tau pathology was not only found in the hippocampal CA1 but also to a similar degree in the amygdala of rTg4510 mice [16], suggesting that hyperphosphorylated tau in the latter could account for behavioral dysfunctions such as hyperactivity in rTg4510 mice. Pathological tau burden measured by AT8 immunoreactivity averaged to approximately 20% of area in the basolateral amygdala of 20 and 32-week-old hyperactive rTg4510 mice (data not shown). During this time frame, an increase in the number of hyperactive animals occurred (Fig. 4A), but tau burden determined in the basolateral amygdala remained stable, suggesting that saturation of tau pathology is reached at 20 weeks of age. However, we observed considerable variations in tau burden in the basolateral amygdala between individual rTg4510 mice from both age groups. Consequently, more than 100 hyperactive and non-hyperactive rTg4510 mice would be required for a significant analysis associating hyperactivity and pathological tau load in the amygdala.
In conclusion, the rTg4510 mouse strain showed a robust and abnormal increase in locomotoractivity linked to P301L tau transgene expression. As this phenotype correlated well with the presence of hyperphosphorylated tau, hyperactivity in the rTg4510 mice might be used as a functional measure for tau targeting therapies, although the precise relationship between pathological tau and the hyperactive phenotype warrants further investigation. The present data suggest that the hyperactivity phenotype can be used as a model for agitation without aggression observed in AD and other dementias, underscoring the usefulness of the rTg4510 mice in studying non-mnemonic changes in behavior related to dementia.
Footnotes
ACKNOWLEDGMENTS
The excellent technical assistance of A. Bredal Christiansen, K. Jørgensen, P. Møller Carstensen and A. Ravn Neve is gratefully acknowledged. We thank Dr. J. Lewis, University of Florida, Gainesville, USA and Dr. N. Sahara, National Institute of Radiological Sciences, Chiba, Japan for providing valuable advice in regards to testing rTg4510 animals in the Morris water maze and performing the biochemical tau analysis, respectively.
