During our studies, we used 4-week-old adolescent male C57BL/6J mice weighing 15–18 g (Beijing Vital River Laboratory Animal Technology, Beijing, China). Mice were randomly divided into groups of four individuals per cage under controlled conditions (12-h light/dark cycle from 7:00 to 19:00, 50 ± 10% humidity, and 23 ± 2°C temperature control) with food and water ad libitum. Mice were accustomed to this experimental environment for 7 days and allowed to adapt prior to each individual experiment. There were 36 mice in the saline (Sal) group and 36 mice in the METH group. Animal procedures were conducted in accordance with the United Kingdom Animals (Scientific Procedures) Act and Institutional Animal Care Committee at Xi’an Jiaotong University.

Methamphetamine hydrochloride (National Institute for the Control of Pharmaceutical and Biological Products, Beijing, China) was dissolved in 0.9% physiological saline to a concentration of 0.2 mg/ml and injected intraperitoneally at a volume of 10 ml/kg (i.p.). All mice were consecutively exposed to corresponding concentrations of METH (1, 2 mg/kg) or saline one time a day from 9:00 a.m. to 11:00 a.m. for 14 days after adaptation (35–48 postnatal days). The rationale for the dosing regimen was based on a previous study that demonstrated that the treatment of mice with METH for 14 days led to a clear impairment in recognition memory (North et al., 2013).

The Morris Water Maze (MWM) protocol was based on a previous study with some modifications (Cao et al., 2018). In brief, the surface of the water was artificially divided into four quadrants, each with an entry point. Signs of the same size and color, but in different shapes, were pasted in the middle of the water tank wall in each quadrant. During the spatial learning phase, a plexiglass platform with a diameter of 10 cm was hidden 1–1.5 cm below the water surface in the second quadrant; this platform was subsequently removed until the probe test phase. A camera was placed above the tank and was used to automatically track each animal’s performance with a video-computerized tracking system (SMART, Panlab SL, Barcelona, Spain).

The MWM was performed during adolescence or adulthood; there were eight mice in the METH group or Sal group in different periods (adolescence, adulthood) and at different doses (1, 2 mg/kg), respectively. The experiment was divided into a spatial learning phase (four trials a day for 4 or 5 days) and a probe test phase (24 h after the last trial). During the learning phase, the mice were permitted to acclimatize to the environment for 1 day before the learning phase; they were then placed randomly in the water facing the tank wall at one of four fixed entry points. If the mice climbed onto the platform within 60 s, they were allowed to stay there for 10 s and the latency was recorded; if not, the mice were manually placed on the platform for 15 s and the latency was recorded as 60 s. After each trail, the mice were wiped with a towel and placed in their home cage for at least 10–15 min until the next training session. The mean latency and speed in four trails per day were set as the key indices for the learning ability of the mice. During the probe test phase, the mice were placed in water from the point farthest away from the platform and recorded for 60 s without the platform. During this phase, the spatial memory retention ability was represented by the number of platform site crossings, the time in the target quadrant, and the time in the target.

The procedure used for the New Object Recognition Test (NORT) was described previously (Antunes and Biala, 2012). The 4-day experiment was carried out in adulthood after exposure to 2 mg/kg of METH during adolescence (n = 8). The mice were adapted to the empty box for 10 min a day for the first 2 days of the experiment. During the familiarization phase on day 3, two identical objects were placed at an equal distance from the box, and the mice were allowed to explore freely for 5 min. Each mouse was required to perform three training sessions at 15-min intervals. A short-term memory test (test 1) was performed 3 h after the last training session, in which one object (A) was replaced by another object (B) of the same size but a different shape. The long-term memory test (test 2) was administered 24 h after the last training session and was the same as the short-term memory test, except that object (B) was replaced by another new object (C); the time was recorded for 5 min. We also determined the recognition index (%). The top of the box was recorded by the camera and finally analyzed by software (SMART, Panlab SL, Barcelona, Spain).

The Y-maze spontaneous alternation test was carried out as described previously (Yan et al., 2019). The experiment was conducted 1 week after the NORT (n = 8). During this experiment, each mouse was placed at the central point and allowed to explore freely for 5 min; we recorded the order in which the animals entered each arm. Alternation was defined as continuous entry into three different arms, such as (A, B, C or C, A, B). Next, we calculated the alternation triplet (%), as follows: Alternation triplet (%) = actual alternative/maximum alternative × 100%. Mice with less than eight total arm entries were excluded. The behavioral indicators for each mouse were recorded by specific software (SMART, Panlab SL, Barcelona, Spain).

After the behavioral experiments, the mice were sacrificed (n = 6) and their brains were quickly removed. Next, we prepared tissue sections from the dorsal hippocampus (Franklin and Paxinos, 2001) using a cryostat; these sections were stored at –80°C until treatment. Then, total RNA was isolated using an E.Z.N.A. DNA/RNA/Protein Kit (Omega Bio-Tek, Norcross, GA, United States). The concentration and mass of the extracted RNAs were determined by a Nano Drop spectrophotometer (Thermo Scientific, Waltham, MA, United States). Next, 500 ng of total RNA was reverse transcribed-into 10 μl of cDNAs with a PrimeScript™ RT Master Mix (Takara Biomedical Technology, Beijing, China) at 37°C for 15 min, 85°C for 5 s, and 4°C for 5 min. Then, we determined the relative expression levels of Arc, c-fos, and Bdnf mRNA in dorsal hippocampus of mice. qPCR was performed with SYBR Premix Ex Taq II (Takara Biomedical Technology, Beijing, China) and a Bio-Rad iQ5 detection instrument (Bio-Rad, Hercules, CA, United States) under the following conditions: 95°C for 30 s, followed by 40 cycles of 95°C for 10 s, 60°C for 30 s, and 72°C for 30 s. After the completion of the reaction, specificity was verified by melting curve analysis. The relative mRNA values were normalized to the control values of the Gapdh gene and calculated using the 2^–△△Ct method (Livak and Schmittgen, 2001). The sequences of the forward (F) and reverse (R) primers are shown in Table 1.

To prepare tissue for Transmission Electron Microscope (TEM), mice (n = 3) were exposed to 2 mg/kg of METH and then anesthetized after completing the final behavioral experiment in adulthood. The brain tissue was fixed by left ventricular vascular perfusion; we injected the following in sequence: 0.9% physiological saline, 0.1 M phosphate buffer (PB), and a mixed solution of 4% paraformaldehyde and 0.25% glutaraldehyde (pH 7.4). Then, the brain was removed and the bilateral dorsal hippocampus were extracted and dissected into tissue blocks that were approximately 1 mm³ in size. These small tissue blocks were then fixed in 0.1 M PB containing 4% paraformaldehyde and 2.5% glutaraldehyde (pH 7.4) at 4°C for at least 2 h. Next, the samples were immersed in 0.1 M PB for more than 30 min, followed by fixation in a fresh 1% solution of osmium tetroxide for 120 min and immersed in 0.1 M PB for 10 min. Next, the tissues were dehydrated in different concentrations of ethanol and then embedded in Epon-Araldite resin. Next, the brain tissue was cut into 1-μm-semi-thin sections, stained with methylene blue, and then cut into 50-nm-ultra-thin sections. A total of three ultra-thin sections of each mouse were randomly selected for photography, and 3–5 images of each ultra-thin section were taken for statistical analysis. Images were acquired by an electron microscope (H-7650, Hitachi Limited, Tokyo, Japan), so that we could analyze the number of synapses (magnification 20,000×; 30 photographs) and the synaptic ultrastructure (magnification 40,000×; 38 photographs). For analysis, synapses needed to contain three or more synaptic vesicles in their presynaptic elements and obvious dense layers in the postsynaptic elements (Wang et al., 2020). Image Pro Plus 6.0 (Media Cybernetics, Rockville, MD, United States) was used to measure the thickness of postsynaptic density (PSD) at the thickest part, the width of the synaptic cleft, the length of the presynaptic active zone, and the total number of synapses. All analyses were performed in a blinded manner.

Mice (n = 3) were deeply anesthetized and then fixed by brain perfusion with 0.9% physiological saline and 4% paraformaldehyde via the left ventricular vessels. Then, the brain tissue was fixed with 4% paraformaldehyde at room temperature for 4 h and dehydrated with 30% sucrose solution. When completely dehydrated, the tissues were embedded with optimal cutting temperature compound and frozen in liquid nitrogen. Coronal sections (20 μm) at the hippocampal level were prepared using a freezing microtome (Leica CM1850, LEICA, Wetzlar, Germany). Subsequently, sections were blocked with 5% bovine serum albumin at 37°C for 60 min. Subsequently, brain tissue sections were incubated with chicken-anti GFAP (1:200; Abcam Cat# ab134436, RRID:AB_2818977) and rabbit-anti NeuN (1:100; Abcam Cat# ab177487, RRID:AB_2532109) specific primary antibody mixed solution at 4°C for 24–48 h. In the following morning, the sections were washed with phosphate buffer saline (PBS) and incubated with goat anti-chicken Alexa fluor 488 (1:500; Abcam Cat# ab150173, RRID:AB_2827653) and goat anti-rabbit coralite594 conjugate (1:500; Proteintech Cat# SA00013-4, RRID:AB_2810984) for 90 min at room temperature, respectively. Finally, after 5 min of incubation with 4′,6-diamidino-2-phenylindole (DAPI), stained sections were examined under a fluorescence microscope (Carl Zeiss, Axio scope A1, Germany); 10–15 regions were randomly selected from each group. The positive cell density was determined and the NeuN/GFAP ratio was calculated. All stained sections were analyzed by observers who were unaware of the experimental cohort.

Statistical analyses were performed using SPSS 18.0 (IBM, Armonk, NY, United States) or Prism 8.0 (Blue Prism, London, United Kingdom). Data collection and analysis were not performed blinded to the experimental conditions. The mice were randomly assigned into study groups. All data are presented as the mean ± standard error of the mean (SEM). The two-way repeated measures analysis of variance (ANOVA) was used to analyze the latency to platform in the MWM, and a post hoc multiple comparison (Bonferroni test) was used to analyze the differences between different days. Other data were analyzed by the Student’s t-test according to the homogeneity of variance. p-Values < 0.05 were considered statistically significant.

To investigate the effect of METH on learning and memory in adolescent mice (49–54 postnatal days), we first exposed mice to 1 mg/kg of METH for 14 days (35–48 postnatal days). MWM was performed immediately after drug exposure to identify the changes in the learning and memory abilities of mice (Figure 1A). During the spatial learning phase, as the number of training days increased, the latency of the mice to the platform shortened; however, there was no significant difference between the two groups with this respect (Figure 1C, when compared to day 16 of the same group, p < 0.05). We also tested the swimming speed to exclude potential interference effects; there was no significant difference between the groups (Figure 1B). In addition, probe test 1 showed that there was no significant difference between the groups in terms of platform site crossings (Figure 1D), the time in the target quadrant (Figure 1E), and the time in the target (Figure 1F).

Next, we changed the dose of METH. Adolescent mice were exposed to 2 mg/kg of METH for 14 days, and the MWM was also performed in adolescence (Figure 1G). The results were very similar to those seen with the 1 mg/kg treatment (Figures 1H–L). During the spatial learning process, the latency to the platform in all groups of mice decreased as the number of training days increased, although the drug had no significant effect (Figure 1I, compared with day 16 of the same group, p < 0.05). We also found that speed had no significant effect on latency (Figure 1H). In probe test 1, there was no significant difference between the Sal group and the METH group in terms of platform site crossings (Figure 1J), the time in the target quadrant (Figure 1K), and the time in the target (Figure 1L). These results showed that long-term exposure to METH during adolescence did not affect the learning and memory abilities of mice in their adolescence.

Next, we investigated whether drug exposure in adolescence would cause changes in learning and memory abilities at adulthood (63–84 postnatal days). Mice were exposed to 14 days of METH (1 or 2 mg/kg). The MWM experiment was performed in adulthood to verify whether the learning and memory abilities had changed. As shown in Figures 2A–F, during the phase of spatial learning, all mice learned to find the platform faster as training continued (Figure 2C, when compared to day 46 of the same group, p < 0.05). Swimming speed had no effect on latency to the platform (Figure 2B). There were no significant differences arising from probe test 2 in terms of platform site crossings (Figure 2D), the time in the target quadrant (Figure 2E), and the time in the target (Figure 2F).

In contrast, exposure to 2 mg/kg of METH during adolescence impaired spatial memory retention but not acquisition in their adulthood. As shown in Figures 2G–L, during the training stage of the MWM (Figure 2I), the latency of mice climbing the platform was not affected by the drug after 5 days of training when the groups were compared (Figure 2I, when compared to day 45 of the same group, p < 0.05). Mean swimming speed did not have a significant effect on latency to the platform (Figure 2H). In probe test 2, the mice exposed to METH during adolescence underwent fewer platform site crossings [Figure 2J, t₍₁₄₎=2.849, p < 0.05], less time in the target quadrant [Figure 2K, t₍₁₄₎=2.233, p < 0.05], and less time in the target [Figure 2L, t₍₁₄₎=2.158, p < 0.05]. We compared the mean METH/Sal ratio between different doses of exposure in probe test 1 (Figure 2M) and probe test 2 (Figure 2N), including platform site crossings, the time in the target quadrant, and the time in the target. This allowed us to investigate the dose effects on changes in spatial memory in adolescence (probe test 1) and adulthood (probe test 2). The smaller the index, the more serious damage was revealed. As expected, the damage caused by the 2 mg/kg dose was more serious than that caused by the 1 mg/kg dose in adulthood [t₍₄₎=3.161, p < 0.05].

In addition, we conducted the NORT and Y-maze to investigate the cognitive ability of mice during adulthood (Figure 2G). The NORT showed that the short-term memory recognition ability of mice was not significantly affected (Figure 2O). However, during the long-term memory recognition stage (Figure 2P), the recognition index of the METH group was significantly lower than that of the Sal group [t₍₁₄₎=2.547, p < 0.05]. The Y-maze was performed 7 days after the end of the NORT and revealed no change in the number of spontaneous alternations (Figure 2Q). Collectively, these results suggested that adolescent mice exposed to 2 mg/kg METH suffered from impaired spatial memory and long-term recognition memory in their adulthood.

To explore the neural plasticity mechanism of memory impairment in adulthood, we first examined the changes in the mRNA expression of genes related to neuroplasticity, including Arc, c-fos, and Bdnf in the dorsal hippocampus of mice exposed to METH (2 mg/kg) during adolescence. The expression levels of Arc [Figure 3A, t₍₁₀₎=3.776, p < 0.01], c-fos [Figure 3B, t₍₁₀₎=8.220, p < 0.001], and Bdnf [Figure 3C, t₍₁₀₎=4.267, p < 0.01] in the METH group were significantly reduced when compared to controls.

We then investigated the changes in the ultrastructure of synapses in the dorsal hippocampus (Figure 4A) of adult mice that had been exposed to METH during their adolescence (2 mg/kg). Using TEM, we observed the number of synapses (the left side of Figure 4B, 20,000×) and the ultrastructure of the synapses (the right side of Figure 4B, 40,000×), including the thickness of the PSD, synaptic cleft width, and the length of the presynaptic activity zone. The statistical results are shown in Figures 4C–F; compared with the Sal group, the number of synapses [Figure 4C, t₍₅₈₎=5.135, p < 0.001] and the thickness of the PSD [Figure 4D, t₍₇₄₎=2.72, p < 0.01] in the METH group were significantly higher, whereas the width of the synaptic cleft was significantly reduced in the METH group [Figure 4E, t₍₇₄₎=3.176, p < 0.01]. In addition, the length of the presynaptic activity zone was also reduced significantly in the METH group when compared to the Sal group [Figure 4F, t₍₇₄₎=4.751, p < 0.001]. These results indicated that exposure to METH during adolescence changed synaptic ultrastructure in the dorsal hippocampus during adulthood; this may be the structural basis for spatial memory impairment.

Structural plasticity includes not only synaptic changes but also the number of neurons and glial cells (Allen and Eroglu, 2017). Astrocytes are the most abundant glial cells in the central nervous system and have been repeatedly shown to be involved in neuroplasticity; many studies have reported that astrocytes are vital for learning and memory (Allen and Eroglu, 2017; Robin et al., 2018). Therefore, we labeled GFAP and NeuN to detect the changes in the number of astrocytes and neurons in the CA and dentate gyrus (DG) regions of the dorsal hippocampus in adult mice that had been exposed to METH during their adolescence (2 mg/kg) (Figure 5). There was no significant difference in terms of cell density of astrocytes in the CA1 region (Figures 5A,B) when compared between the METH group and the Sal group (Figure 5C); However, we identified a significant reduction in the cell density of neurons [Figure 5C, t_(21.175)=4.025, p < 0.01], and the ratio of NeuN to GFAP [Figure 5C, t₍₂₅₎=4.078, p < 0.001] in the METH group. Similarly, in the CA3 region (Figures 5A,B), there was no significant difference in the number of GFAP-positive cells in the METH group when compared to the Sal group (Figure 5D); however, the density of NeuN-positive cells was significantly lower [Figure 5D, t₍₁₉₎=2.427, p < 0.05]. The ratio of NeuN to GFAP was also significantly reduced in the METH group [Figure 5D, t₍₁₉₎=2.67, p < 0.05]. Surprisingly, there was no change in the DG region (Figure 5E). In summary, exposure to METH during adolescence can reduce the number of neurons in the CA region in adulthood.