Breeding pairs of Amyloid precursor protein (APP)/presenilin-1 (PS1) double transgenic male mice (B6C3-TgAPPswe, PSEN1dE985DboJNju; abbreviated to AD mice) were obtained from Guangdong Medical Laboratory Animal Center (license number SCXK [Yue] 2018-0002). The transgenic mice express a chimeric mouse/human APP (Mo/HuAPP695swe) and a mutant human PS1-dE9 directed to central nervous system (CNS) neurons. Genotypes were confirmed by polymerase chain reaction (PCR) genotyping using genomic DNA extracted from tail tissue samples. The strain of mice develops beta-amyloid (Aβ) deposits in the brain by 6–7 months of age. Both male and female mice were used in this study, and they were 7–8 months old. During the experimental period, mice were housed in groups of 1–3 animals in plastic cages (30 cm × 18 cm × 14 cm) under constant temperature (23 ± 1°C) and stable humidity with a natural light-dark cycle. They had free access to food and water. The experimental and animal care procedures were performed according to the guidelines for the National Care and Use of Animals and approved by the National Animal Research Authority. Furthermore, all experiments were approved by the Animal Care and Ethics Committee of Kunming University of Science and Technology (approval No. 20190001).

The AD mice were randomly divided into three groups: (1) atDCS group with anodal tDCS; (2) ctDCS group with cathodal tDCS; and (3) sham group with sham stimulation (Figure 1A). We used a protocol of tDCS similar to that described previously (Cambiaghi et al., 2010). Mice were implanted with an epicranial plastic tube for filling with saline as active electrode prior to stimulation. In the present study, animals were meanwhile implanted with metal electrodes for EEG recording (Figure 1B). After the surgery of electrode implantation, mice were allowed at least one week for recovery. In addition, all mice were allowed 3 days for habituation in recording and stimulation conditions before the beginning of the experimental routine.

The experimental routine was diagrammed in the figure (Figure 1A). tDCS was given for five consecutive days, with one day prior for adaption for the whole stimulation protocol (with no current stimulation). Spontaneous EEG activity was acquired for three times: one time before (Spon-Pre) and two times after (Spon-Post1 and Spon-Post2) tDCS. EEG activity during Y-maze performance (Y-EEG) was acquired for one time after tDCS. In addition, behavioral performance was assessed using the following tasks: Barnes maze (BM), open field (OPF), and T-maze. Mice were finally sacrificed and the brains were removed for further histological examination (right brains) and enzyme-linked immunosorbent assay (ELISA) procedure (left brains). All experiments were completed within 1 month after the first tDCS. Animal numbers for each group were detailed in the figure (Figure 1A). It is noted that for spontaneous EEG recording, the electrodes from one animal could not collect EEG signals during the third recording, so data from the first and second recordings were excluded out for this animal.

Surgery was performed under pentobarbital anesthesia (80 mg/kg, i.p.; dissolved in saline, 10 mg/ml, Merck, Darmstadt, Germany). After a midline scalp incision, the scalp and underlying tissues were removed. An epicranial plastic tube (inner diameter: 3.5 mm) was first implanted with its center positioned over the left PFC (AP: +2.95 mm, ML: –1.5 mm). The tube was tightly glued onto the skull by modified acrylate adhesive. In addition, four burr holes were drilled in the skull for implanting EEG electrodes. One twisted pairs of perfluoroalkoxy (PFA)-coated stainless steel wires (diameter: 0.002″, A-M systems, WA, United States) were implanted in the brain through one hole, serving as an EEG recording electrode, to record the right PFC (AP: +2.80 mm, ML: +1.0 mm, DV: –0.5 mm from dura). A stainless-steel watch screw (M1.0 × L2.0 mm, RWD) was placed in contact with the dura through one hole, serving as the left cortex (Ctx) recording electrode (AP: –3.8 mm, ML: –2.5 mm). Through the other two holes, two stainless-steel watch screws were also placed in contact with the dura above the left olfactory bulb and central cerebellum, serving as reference and ground electrodes, respectively. All electrodes were attached to male pins that were secured in a rectangular pin array. Finally, all electrodes together with the epicranial plastic tube were secured with dental acrylic.

Before tDCS, the animal was restrained on a self-made fixed table in an awake state, keeping its head and torso still (Figure 1B). The animal remained consciousness during the stimulation to prevent the interaction between current stimulation effect and anesthetic drugs. For tDCS, the epicranial plastic tube was filled with saline, serving as active electrode (round, diameter: 3.5 mm). The counter electrode was a saline-soaked sponge (round, diameter: 12 mm) applied over the ventral thorax. Here, the electrodes of saline and saline-soaked sponge were chosen instead of metal electrodes, mainly because the latter can be polarized by the direct current of tDCS. Both atDCS and ctDCS were applied at a current intensity of 300 μA for 20 min by a constant current stimulator (Cerebooster, Droian, Hangzhou Zhuo An Zi network technology Co. Ltd.). The current intensity corresponds to a density of 3.12 mA/cm² (300 μA/0.096 cm²). The current flow changed linearly within 10 s at the beginning and the end of tDCS to avoid a stimulation break effect from switching it on and off directly. In addition, the operation of sham group was the same as that of tDCS groups but with no current applied. tDCS was conducted for five consecutive days.

EEG signals were collected by EEG acquisition system, which consisted of an RHD2132 amplifier, RHD2000 USB interface board, and RHD2000 Interface GUI Software (Intan Technologies, Los Angeles, CA, United States). Data were acquired at a sampling rate of 1,000 Hz. The rectangular pin array of electrodes on each animal was connected by a cable to the amplifier, then to the interface board, and finally to the computer. The cable was suspended by a helium balloon to allow the animal free movement. EEG recording was performed in a shielding cage (80 cm × 70 cm × 100 cm, width × length × height). The animal behaviors in the cage were monitored with a ceiling-mounted camera. The video signals were displayed and saved by video-recording software.

Spontaneous and task-related EEG signals were collected in the shielding cage. Spontaneous EEG signals were acquired three times for each animal. Each time lasted for 30 min and the animal was under spontaneous and awake state. A new home cage was placed in the shielding cage, and the floor of home cage was covered with sawdust. This condition is similar to their conditions of daily living. Task-related EEG signals were acquired one time for each animal during Y-maze performance. The task consisted of two trials: 10-min training trial and 5-min testing trial. EEG signals were acquired for both trials. For this task, the shielding cage was decorated with a curtain inside. Several visual spatial cues, such as black square, yellow circular grating, and white triangle, were suspended on the curtain. In addition, the camera, amplifier, and exposed wood of the cage were used as spatial cues. A Y-maze was placed in the shielding cage, and the floor of the maze was covered with sawdust, which was mixed after each trial to equate differential olfactory stimuli. Please refer to the section of Behavioral tests for details of the Y-maze task.

Spatial learning and memory were assessed in animals after tDCS (Figure 1A). Short-term recognition memory was evaluated by two-trial Y-maze test, long-term learning and memory were by Barnes maze (BM) test, and working memory was by T-maze test. In addition, general locomotor activity and anxiety-like behavior of animals were assessed by open field (OPF) test. During behavioral tests, a video-monitoring system (KEmaze, China KEW Basis) was simultaneously used to obtain and analyze the trajectories of animal.

The Y-maze was made of black PVC board, and consisted of three arms with an angle of 120° between adjacent arms. There was 1–2 differential visual cues placed on the walls of each arm. Each arm was 8 cm × 30 cm × 15 cm (width × length × height). The three identical arms were randomly designated for each animal: (1) start arm, in which the animal started to explore (always open); (2) novel arm, which was blocked during the training trial, but open during the testing trial; and (3) other arm (always open). The Y-maze task consisted of 10-min training trial, 5-min testing trial (retrieval test), and 1-h inter-trial interval (ITI) phase. The animal was returned to its home cage during the ITI phase. During training, the animal was allowed to explore the start and other arms, with the novel arm blocked. During testing, the animal was allowed to explore all three arms. The percentage of time spent in the novel and other arms and the percentage of number of arm visits were calculated for each animal. Data in the start arm were excluded to avoid the stimuli of placing the animal in the maze. The first choice for novel arm was also recorded. These data in testing trial were as spatial recognition memory indices. In addition, the total number of arm visits in the maze was counted for each animal as a locomotor activity index for both training and testing trials.

The BM maze consisted of a round platform and a stainless steel bracket. The platform was 90 cm high from floor. It was made of white PVC board and the diameter was 90 cm. The platform contained 20 holes, each 5 cm in diameter, equally distributing around the platform. During the experiment, one hole was randomly assigned to be the target hole for each animal. There was an escape box in the target hole, which was communicated with the platform through transparent plastic tunnels. The escape box cannot be seen from the platform.

The BM task included habituation, acquisition and probe phases. During habituation (day 1), the animal was placed on the center of the platform, and two bright lights were turning on as an aversive stimulus. The animal was guided to the target hole, and once it was in the escape box, the lights were turned off and the animal was kept inside for 2 min. The acquisition (days 2–5) phase lasted for 4 days and each day included four trials. Each trial lasted for 3 min with an ITI of 15 min. In each trial, the animal was placed on the center of the platform and the lights were turned on. The latency to find the target hole was recorded for each animal. If the animal did not find the target hole within 3 min, it was placed at the entrance of the target hole for 1 min and was then returned to its home cage. The probe test (day 6) was conducted 24 h after the last acquisition trial. The escape box was taken out, and the animal was allowed to explore the maze for 90 s. The latency to find the target hole in the first place was recorded. In addition, the total ambulation and speed on the platform were calculated automatically by the software. During the task, the platform was rotated after each trial for each animal. The location of the target hole for each animal was kept in the same orientation. The platform was cleaned with 75% alcohol and then with dry paper towel after each trial.

The OPF box was made of PVC board, with gray walls and white floor. The box was square with 50 cm × 50 cm × 30 cm (width × length × height). The animal was placed into the box with its head toward one wall and then allowed to explore the box for 5 min. The total ambulation and speed in the box were calculated automatically by the software. The ambulation and time spent in the central area (25 cm × 25 cm, width × length) of the box were also calculated. In addition, the times of rearing and the number of fecal droppings (defecation rate) were recorded for each animal. The box was cleaned after each test.

The T-shaped maze was made of black PVC board and consisted of a start arm and two choice arms. The start arm was 40 cm × 8 cm × 15 cm (length × width × height) and the choice arm was 30 cm × 8 cm × 15 cm (length × width × height). A recessed black plastic food cup (4 cm in diameter, 1 cm in depth) was placed on the floor at the end of each choice arm. The maze was enclosed by a curtain with some visual spatial cues suspended on it. The animal in the maze could see the cues outside the maze.

The T-maze task included habituation, training and testing phases. During habituation (day 1), there were two pieces of peanuts in both food cups, and the animal was placed in the start arm and allowed to explore the maze for 10 min. The animal was returned to its home cage after habituation. During training (days 2–10), there was two pieces of peanuts only in one choice arm designed as correct arm for each animal. The training phase lasted for 9 days and each day included five trials. Each trial began by placing the animal in the start arm with its head toward the wall, and ended when the animal entered the correct arm and ate the peanut. If the animal did not choose either arm or correct arm within 2 min, it was returned to its home cage. The ITI varied randomly within 40–60 s. The testing phase (day 11) included only one trial and there was nothing in both food cups. The trial began by placing the animal in the start arm and ended when the animal entered one choice arm. During the experiment, one choice arm was randomly assigned to be correct arm for each animal, but the correct arm and orientation of the maze were kept stable for each animal. Correct rate for every day and mean correct rate for the total training phase were calculated for each animal. Correct rate for the testing day was also calculated for the stimulation group. The floor of the maze was covered with sawdust, which was mixed after each individual trial to equate differential olfactory stimuli.

After all experiments, animals were anesthetized with pentobarbital sodium (80 mg/kg, i.p.). Electrolytic lesions were first made by applying an anode direct current (60 mA, 2.5 min) to the PFC electrode to mark the EEG recording location. The animals were then transcardially perfused with saline. Brains of mice were removed after decapitation and dissected on an ice-cold plate to isolate: (1) the right brains for histologically confirming Aβ plagues and EEG recording locations; and (2) the left hippocampal tissues for ELISA experiment. Followed (1), the right brains were post-fixed in 4% paraformaldehyde (PFA) at 4°C for 48 h, dehydrated twice in 30% sucrose with PBS at 4°C, and then stored in 4°C refrigerator until further use. Followed (2), the left hippocampal tissues were initially stored in a liquid nitrogen tank and then in –80°C refrigerator until further use.

As described earlier (Fu et al., 2019), EEG signals were off-line analyzed by MATLAB. EEGs were recorded by the Intan system as rhd data files. For frequency band power analysis, each rhd data file was first separated into segments, with each segment comprised of 1,024 sample points. The segments were filtered for the following EEG frequency bands (with no 50 Hz notch filter): (1) delta: 2–4 Hz, (2) theta: 4–8 Hz, (3) alpha: 8–12 Hz, (4) beta: 12–20 Hz, and (5) gamma: 20–100 Hz (Figure 1C). There is large variability for the definition of frequency bands for the EEG rhythms (Newson and Thiagarajan, 2018). In this study, the frequency range for each band was in keeping with our previous works on aging (Fu et al., 2008, 2019; Zhou et al., 2021). For each frequency band, the absolute power of each segment was calculated as: P = Σχ²/1,024. The relative power (RP) was calculated as the percentage of power relative to the total power of all frequency bands.

Data with normal distribution were expressed as means ± SEM, and data with non-normal distribution were shown as medians and range (min to max). For normally distributed data, paired-samples T test was used for paired samples; one-way ANOVA and repeated-measure ANOVA were used for independent samples or repeat measurements, respectively. For non-normal distribution data, non-parametric analyses were performed. Wilcoxon matched-pairs signed rank test was used for paired samples and Friedman test were used for repeat measurements, respectively. P values smaller than 0.05, 0.01, and 0.001 were considered statistically significant, highly significant, and very highly significant, respectively.

Spontaneous EEG activity was recoded once before and twice after the stimulation. The most obvious EEG changes were found in the PFC (Figure 2). For both atDCS and ctDCS, the PFC delta EEG activity showed significant decrease after the stimulation (Friedman test; χ² = 11.14 and 12.67 for atDCS and ctDCS, respectively, P < 0.01 for both) (Figures 2B,C). The decrease could be observed in a short time (Post1, 1–2 days after tDCS) and/or in a long time (Post2, 10–11 days after tDCS) (please refer to the figure for the detailed significances from post-hoc pairwise comparisons; the same below). Specifically, atDCS significantly increased EEG activity in the alpha band in a long time after the stimulation (χ² = 12.29, P < 0.01) (Figure 2B); and ctDCS significantly increased EEG activity in the theta band (χ² = 6.89, P < 0.05) and decreased EEG activity in the beta band (χ² = 18.00, P < 0.01) in a long time after the stimulation (Figure 2C). PFC EEG changes in five frequency bands were not significant for sham (Figure 2A).

Ctx theta EEG activity showed significant decrease in sham and ctDCS groups after the stimulation (χ² = 7.00 and 6.89 for sham and ctDCS, respectively, P < 0.05 for both) (Figures 2D,F). As significant change was found in sham group, it may because theta EEG activity in the Ctx was affected by conditions of the recording day. In addition, EEG changes in the other frequency bands were not significant in this region (Figures 2D–F).

Behaviorally, the AD animals in ctDCS group could differentiate the novel arm from the other arm in the Y-maze task. They showed significantly higher percentage time and number in the novel arm than in the other arm [Paired t-test; t(8) = 3.03 and 6.34, P < 0.05 and 0.001 for time and number, respectively] (Figures 3A,B). In contrast, animals in sham and atDCS groups showed similar percentage time and number in both arms (P > 0.05). It was noted that the first choice for the novel arm was higher for atDCS (85.71%, 6/7) and ctDCS (77.78%, 7/9) groups than sham group (57.14%, 4/7) (Figure 3C). In addition, the total number of arm visits either in training or testing trials showed no significant difference among three groups (P > 0.05) (Figures 3D,E).

EEG activities in five frequency bands were first compared between training and testing trials for each stimulation group. It is interesting that there was much EEG difference for ctDCS group (Figure 4). PFC EEG activities in the alpha-gamma bands were significantly higher in testing trials than those in training trials (Wilcoxon test; Z = 2.67, 2.19 and 2.43, P < 0.01, 0.05 and 0.05 for alpha-gamma, respectively; the same below) (Figure 4C). Ctx EEG activities in the alpha and gamma bands were significantly higher and that in the delta band was significantly lower in testing trials than those in training trials (Z = 2.19, 2.19, and –2.07 for alpha, gamma, and delta, respectively, P < 0.05 for all) (Figure 4F). For atDCS group, PFC EEG activity only in the theta band was significantly lower in testing trials than that in training trials (Z = –2.20, P < 0.05) (Figures 4B,E). In contrast, for sham group, there was no significant EEG difference between the two trials in any frequency band (Figures 4A,D).

In addition, EEG activities in five frequency bands were compared between novel and other arms in testing trials (Figure 5). Similarly, there was much EEG difference for ctDCS group. PFC EEG activity in the delta band was significantly lower (Z = –2.43, P < 0.05) and those in the alpha-gamma bands were significantly higher (Z = 2.67, 2.31, and 2.55, P < 0.01, 0.05, and 0.05 for alpha-gamma, respectively) in the novel arm than those in the other arm (Figure 5C). In addition, for atDCS group, PFC EEG activities in the delta (Z = –2.20, P < 0.05) and gamma (Z = 2.37, P < 0.05) bands showed similar differences as those in ctDCS group (Figure 5B). For sham group, Ctx EEG activity in the beta band was significantly higher in the novel arm than that in the other arm (Z = 1.99, P < 0.05) (Figure 5D). EEG changes in five frequency bands were not significant for PFC in sham group (Figure 5A) and for Ctx in both atDCS and ctDCS groups (Figures 5E,F).

The animals in atDCS group showed improved spatial learning ability during the 4 days training session [ANOVA-R and ANOVA-1; combined analysis: F(3,60) = 4.37, P < 0.01 for main effect of day and F(6,60) = 3.69, P < 0.01 for interaction effect of day and group; atDCS: F(3,18) = 8.20, P < 0.01; sham: F(3,18) = 1.61, P = 0.22; ctDCS: F(3,24) = 0.55, P = 0.65] (Figure 6A). They showed significantly lower latency to the target hole on the fourth day than that on the first day (post-hoc LSD, P < 0.05). In addition, on the fourth day, the atDCS group showed significantly lower latency to the target hole than both sham and ctDCS groups [ANOVA-1; F(2,20) = 4.43, P < 0.05; post-hoc LSD, vs sham: P < 0.05; vs ctDCS: P < 0.01].

During the testing session, atDCS group also showed improved spatial memory when compared with the other two groups [F(2,20) = 6.16, P < 0.01; post-hoc LSD, vs sham: P < 0.01; vs ctDCS: P < 0.05] (Figure 6B). Total ambulation distance and speed showed no significant difference among three groups (P > 0.05) (Figures 6C,D).

Total ambulation distance and speed in the OPF showed no significant difference among the three groups (P > 0.05) (Figures 7A,D). The percentages of ambulation and time in the center area also showed no significant difference among groups (P > 0.05) (Figures 7B,C). In addition, total rearing times and fecal dropping showed no significant difference among groups (P > 0.05) (Figures 7E,F).

Combine ANOVA-R revealed no significant change for the correct rate with the training days [F(8,152) = 1.56, P = 0.14] (Figure 8). However, when compared with the chance level (50%), both atDCS and ctDCS showed significant difference on some days [One-Sample T-test; atDCS: t(6) = 2.47 and 2.96 for day 5 and 8, respectively, P < 0.05 for both; ctDCS: t(8) = 2.50, P < 0.05 for day 8] (Figure 8A). The difference from the chance level was not significant for any day in sham group (P > 0.05 for all days). For atDCS group, the mean correct rate was slightly higher than the chance level [t(6) = 2.08, P = 0.08] (Figure 8B). During testing day, the ctDCS group (77.8%, 7/9) showed higher correct rate than the sham group (50%, 3/6) (Figure 8C).

The AD mice used in the present study were with Aβ plaques, which distributed throughout cortex and hippocampus as well (Figure 9A). However, total Aβ concentrations showed no significant difference among atDCS, ctDCS, and sham groups (Kruskal–Wallis test; χ² = 1.23, P = 0.54) (Figure 9B).