Homozygous 3xTg-AD transgenic mice (3xTg-AD) containing Psen1, APPSwe, and tauP301L mutations and non-transgenic controls (Ntg) with a B6;129 genetic background were generated from breeders purchased from the Jackson Laboratories (Bar Harbor, ME, USA). The mice were group-housed in a quiet facility and maintained on a 12-hour light/dark cycle (lights on at 7 am), with ad libitum access to food and water. Experiment procedures were carried out according to the National Institutes of Health Guide for Care and Use of Laboratory Animals and approved by the Bioethics Committees of the Yangtze University and Guangzhou Medical University. Male mice were subjected to a 5-month voluntary wheel running initiated at 8 weeks of age and compared with their age-matched sedentary controls. Two cohorts of mice were used, and each contained 3xTg-AD runner, 3xTg-AD sedentary control, Ntg runner, and Ntg sedentary control. At the end of running, one cohort of mice was subjected to a relatively ‘stress-free’ object location task, followed by analyses of spines and synapses, and the other was subjected to a classic water maze task, followed by measures of Aβ pathology.
Physical training was performed as described previously [31, 32]. Briefly, mice were individually placed in vertically revolving activity wheels (16 cm in diameter and 5 cm in depth) and trained for 2 weeks (6–10 pm), followed by 5 months wheel running (7–10 pm). In the training session, mice learned to run 10 min on the first day, and running time progressively increased 10 min per day until they ran 1 h per day. After the training, mice that had learned to run voluntarily in wheels were chosen and assigned to running and sedentary groups. The mice in running group continued to run for 1 h per day during the dark cycle, 5 days per week. The sedentary controls were also placed in running wheels for 1 h per day, but the wheels were immobile. Considering the wheel running is voluntary, we set a criterion for effective training in the runners such that the heart-to-body weight ratio is at least 2 standard deviations (SD) above the mean in sedentary controls (Mean + 2 SD) [31, 32]. 12 in 18 Ntg runners and 13 in 18 3xTg-AD runners in one cohort of mice fulfilled the standard, and 10 in 16 Ntg runners and 11 in 16 3xTg-AD runners in the other cohort were selected for data analyses.
Behavioral tests of memory function.
Object location memory test was performed as described previously . Prior to training, mice were handled 2 min and then habituated to experimental apparatus for 10 min per day for 5 days in absence of objects. In training phase, two identical objects were presented for 10 min exploration. One day later, object exploration was tested for 5 min. The training and testing were performed without knowledge of groups. A video tracking system (EthoVision, Noldus, the Netherlands) was used to record both training and testing phases. Exploration was scored when a mouse's head was oriented toward the object within 1 cm or when the nose was touching the object. Exploration times of object in familiar and novel locations were recorded and expressed as a ratio of novel vs. familiar. Total exploration times were calculated, and the novel/familiar ratio was used as an index of memory function.
Water maze test was conducted with the aid of a video tracking system (EthoVision) as described [31, 32]. The maze consists of a circular tub (120 cm in diameter) and a transparent non-slip platform (9 cm in diameter) which was placed in a constant position for each set of trials. The platform was placed in the middle of a quadrant and 1 cm under the surface of room temperature water. During a 5-day training (3 trials per day), mice explored the platform and learned to associate its location to visual cues in the room. If a mouse cannot locate the platform within 60 seconds, they were gently guided to it. One day after the last trial on day 5, probe trials (1 min each) were performed, in which the platform was removed. During the probe testing, mice were placed in the water facing the pool wall, the number of crossing the original location of the platform and the duration of staying in the platform quadrant and all other quadrants were recorded.
Tissue handling, Golgi staining, and immunohistochemistry.
At the end of object location test, mice were anesthetized with sodium pentobarbital (80 mg/kg) and perfused via the aorta with 0.9% saline solution for 2 min to flush out blood. The brains were rapidly removed from the skull and blocked to two hemispheres. All left hemispheres were harvested for Golgi staining and right hemispheres were collected for immunohistochemistry.
Golgi staining was performed using a superGolgi kit (Bioenno Tech LLC, Santa Ana, CA). Briefly, after 2 days of impregnation in a Golgi-Cox solution (22 ± 1°C, in the dark) provided in the kit, the solution was renewed and the impregnation was continued for another 7 days (22 ± 1°C), total 9 days. The tissue blocks were sectioned coronally at 200 µm (Leica VT1200). Series sections from dorsal hippocampus (1 in 3, 6–8 sections per brain, Bregma − 1.06 mm to -2.54 mm) were mounted on gelatin-coated slides, subjected to staining (15 min, 22 ± 1°C) and post-staining (18 min, 22 ± 1°C) in a parallel manner. Sections were dehydrated in ethanol, cleared in xylene, and covered with Permount® mounting medium.
Immunohistochemistry of postsynaptic density protein-95 (PSD-95) was performed on free-floating sections . Briefly, brain tissues were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) (pH7.4) overnight at 4°C. Fixed tissue blocks were cryoprotected in 30% sucrose in 0.1 M PB and sectioned coronally (20 µm) using a Leica cryostat. Series of sections (every sixth section from dorsal hippocampus, Bregma − 1.22 mm to -2.54 mm) was subjected to fluorescent immunostaining. The sections were incubated with mouse monoclonal anti-PSD-95 (1:2,000, MA1-25629, clone 7E3-1B8, Affinity BioReagents, Golden, CO) for 3 days at 4°C in a parallel manner. After washing in PBS-T (3 × 5 min), antibody binding was visualized with anti-mouse IgG conjugated to Alexa Fluor 488 (1:200, Molecular Probes).
Quantitative analyses of dendritic spines and synapses.
Dendritic spines: Two approaches were applied to evaluate the number and density of dendritic spines in Golgi-stained Sect. 1) Spines in area CA1 were counted with the aid of Stereo Investigator (MBF Bioscience, Williston, VT, USA) using stereological fractionator method [39, 40]. Section series (every third section) from dorsal hippocampus was subjected to counting. Strata radiatum (SR) and lacunosum-moleculare (SLM) in area CA1 were defined using a 5× objective and spines were counted using a 100×/1.4 objective. We used a counting frame of 25 × 25µm, a sampling grid of 200 × 200 µm, a guard zone of 10 µm and disector height of 50 µm. 2) Spine density (number of spines per 20 µm of dendrite length) was calculated on individual CA1 pyramidal cells (3–4 cells per brain, 28–36 cells per group). Well impregnated cells were captured with a Nikon E400 equipped with a CCD camera (DS-Fi3) and motorized stage and reconstructed using Imaris (v7.1.0) and Adobe Photoshop (v6). Cells in each group included equal representation of long- and short-shaft populations . High-magnification images (100×/1.4 oil lens) permitted all spines of a given dendritic segment to be visualized. The sides of dendritic branch were carefully examined for vertical protrusions stretching upward and downward off the branch. Branches including basal dendrites in the oriens (SO), apical obliques in SR, and apical distal branches in SLM were analyzed. The apical trunk was not analyzed because some spines that extended vertically toward the observer were difficult to classify from an aerial perspective. Density of spines on each branch was separately analyzed with the aid of concentric circles at an interval of 20 µm. The data were grouped by cell and then by animal. Spines were classified by size and shape as mushroom-type, thin, and stubby [33, 41].
Synapses: The number of synapses represented by PSD-95 immunoreactive (ir) puncta were quantitatively analyzed as described [32, 33]. Briefly, 10 sections (90 µm interval each other) per animal were used and z-stack images were taken from the regions of interest using a Zeiss 510 confocal microscope with a 63× objective. The region of interest was defined using a 4× objective and images were then captured using a 63×/0.8 objective. We used a counting frame of 25 × 25 µm, a sampling grid of 200 × 200 µm, a guard zone of 10 µm and a disector height of 5 µm. The precision of the study was estimated by calculating the coefficient of error (CE). The CE value for each individual animal ranged between 0.02 and 0.05. Confocal three-dimensional image stacks were processed for iterative deconvolution at 99% confidence (Volocity 6.3). Counts (per 7,500 µm3) of labeled puncta from each section were averaged to obtain a value for each brain. Sections from four groups were processes concurrently and analyzed without knowledge of treatment group.
Preparation of protein extracts and Western blot analyses of synaptic proteins.
Dorsal hippocampus was quickly dissected and area CA1 was further isolated. Dissected tissue was immediately frozen in dry ice and then homogenized in T-PER Tissue Protein Extraction Reagent (ThermoFisher Scientific, Rockford, IL) (150 mg/ml) containing protease and phosphatase inhibitor cocktail (1:100, Sigma-Aldrich, St. Louis, MO). The sample was centrifuged at 100,000× g for 1 hour. The pellet was re-suspended with 70% formic acid, followed by centrifugation at 100,000× g for another hour. Protein concentration in the supernatant was determined using the Bradford assay.
Equal amounts of protein (20 µg) were diluted in Laemmli buffer, separated on 4–12% Bis-Tris gel (Invitrogen, Carlsbad, CA), and transferred to nitrocellulose membranes. The membranes were blocked with 5% nonfat milk in 1× TBS-T overnight at 4°C, followed by incubation in primary antibodies overnight at 4°C. The antibodies included mouse anti-PSD95 (1:5,000, clone 7E3-1B8), rabbit anti-GluR1 (1:2,000; PC246, Calbiochem), and mouse anti-synaptophysin (1:10,000, Sigma). The membranes were washed in TBS-T (3 ×5 min) and incubated in appropriate secondary antibody (anti-rabbit IgG or anti-mouse IgG conjugated to HRP) at a dilution of 1:10,000 (Pierce Biotech) for 1 hr at RT. Membranes were then washed in TBS-T (3 × 5 min) and the blots were developed using SuperSignal (Thermo Scientific). Hippocampal extracts of individual mice of different groups were run concurrently on the same gel. Specificity of signal was verified by pre-adsorbing primary antibodies with their respective antigens as well as by excluding the primary antibodies in the presence of the secondary antibodies. These treatments resulted in no immunoreactive bands.
Analyses of Aβ peptides, oligomers, and APP secretases.
ELISA of Aβ peptides: Aβ1-38, Aβ1-40, and Aβ1-42 were measured using the 6E10 Abeta Peptide Ultra-Sensitive Kits (K151FSE, K151FTE, and K151FUE, respectively, MSD, Rockville, MD, USA) . T-PER soluble fractions from dorsal hippocampus were loaded directly onto the ELISA plate, and the formic acid supernatants (insoluble fractions) were diluted 1:2 in neutralization buffer (1M Tris base and 0.5M NaH2PO4) before loading. After incubation in blocking solution, samples and standard peptides including Aβ1-38, Aβ1-40, and Aβ1-42 were added to the 96-well plate and incubated overnight (4°C), followed by washes in 1× Tris wash buffer (3 × 5 min). After 1 hr incubation in detection solution (RT), the plate was washed with 1× Tris buffer, and read in a Sector Imager plate reader (MSD) immediately after addition of the 1× read buffer. The concentrations of Aβ peptides were calculated with reference to the standard curves and expressed as picograms per milligrams of proteins.
Dot-blot of Aβ oligomers and amyloid fibril: Equal amounts of protein (3 µg) were transferred to nitrocellulose membranes, and the membranes were blocked with 5% (w/v) nonfat milk in 1× Tris-buffered saline containing 0.2% Tween 20 (TBS-T, pH 7.5) for 1 hr at RT. The membranes were then incubated overnight at 4°C with one of the following primary antibodies: rabbit anti-oligomer (A11) polyclonal (AHB0052, 1:1,000, ThermoFisher) and rabbit anti-amyloid fibrils OC (AB2286, 1:3,000, Sigma-Aldrich). A11 reacts with soluble AB40 oligomers, but not with soluble low molecular weight AB40 or AB40 fibrils. The membranes were washed in TBS-T (3 × 5 min) and incubated in goat anti-rabbit IgG secondary antibody (1:10,000; ThermoFisher) for 1 hr at RT. The blots were developed using SuperSignal chemiluminescent substrates (ThermoFisher).
Western blot analyses of APP and secretases: Equal amounts of protein (20 µg) were diluted and transferred to nitrocellulose membranes as described above. The membranes were blocked with 5% nonfat milk in 1× TBS-T overnight at 4°C, followed by incubation in primary antibodies overnight at 4°C. The antibodies included rabbit anti-APP-CT20 for C99 and C83 (1:1,000, Calbiochem, San Diego, CA), rabbit anti-ADAM10 (1:1,000, MA5-32616, ThermoFisher), rabbit anti-ADAM17 (1:1,000, PA5-11572, ThermoFisher), rabbit anti-BACE1 (1:1,000, MA5-35126, ThermoFisher), and mouse anti-GAPDH (1:5,000, sc-47724, Santa Cruz Biotech, Santa Cruz, CA). The membranes were incubated in appropriate secondary antibody, and the blots were developed using SuperSignal as described.
Data were analyzed using Prism 9 (GraphPad Software Inc.) or SPSS 22.0 (SPSS Inc.). Analyses of variance (ANOVA) including two-way and three-way were used to detect differences in the escape latencies, path lengths, and swimming speeds in water maze task with groups (Ntg vs. 3xTg-AD, sedentary vs. running) and time (day) as factors, followed by Tukey’s or Bonferroni’s post hoc test. Two-way ANOVA was employed to compare the exploration times and novel/familiar ratios among groups. Spines, synapses, and synaptic proteins were analyzed by three-way or two-way ANOVA with groups and branch segment and/or subregion/subtype as factors, followed by Tukey’s or Bonferroni’s post hoc test. One-sample t-test was also performed to distinguish whether exploration times of objects were different from those predicted by chance. Pearson test was used for the correlation analysis. Data were expressed as mean ± SEM, significance was set at 95% confidence.