Male APPswe/PS1dE9 (APP/PS1) transgenic mice and age-matched wild-type C57BL/6 mice were obtained from Huachuang Xinnuo (Jiangsu, China). All animals were housed in standard cages under controlled conditions (temperature 22–24 °C, humidity 50–60%, and a 12-h light/dark cycle) with ad libitum access to food and water. At 3 months of age, APP/PS1 mice were randomly assigned to either a sedentary group (AD-SED, n = 40) or an exercise group (AD-EXE, n = 40). Age-matched wild-type C57BL/6 mice served as the control group (WT-SED, n = 40). All experimental procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals (Ministry of Health, People’s Republic of China) and were approved by the Animal Use and Ethics Committee of Gannan Normal University (Approval No. gnnu2024-0624).

Mice in the exercise group underwent a three-month treadmill running intervention, following a protocol adapted from previous studies (Zhang et al., 2019; Liang et al., 2022). The intervention consisted of a familiarization phase followed by a formal training phase. The 6-day familiarization phase involved running at incremental speeds of 5 m/min, 8 m/min, and 12 m/min for 2 days at each speed, for 15 min per day. The subsequent formal training phase lasted 12 weeks, with sessions conducted 5 days per week for 45 min daily. Each session began with 5 min at 5 m/min, followed by 5 min at 8 m/min, 30 min at 12 m/min, and a final 5-min period at 5 m/min. All training sessions were conducted between 18:00 and 20:30. Mice in the WT and AD control groups were placed on a stationary treadmill for the same duration to control for environmental stimulation.

After the exercise intervention, animals were assigned to behavioral, biochemical, and metabolic analyses. Specifically, a subset of 12 animals per group (n = 12) were assigned to conduct the Morris Water Maze test, and one animal in AD-EXE group was excluded due to its persistent floating behavior without active swimming in the visible platform trial. Six animals per group (n = 6) were assigned to perform Biochemical analyses (e.g., Elisa, Western blotting). The remaining animals per group were assigned to obtain microglia samples (WT-SED and AD-SED: n = 34, AD-EXE: n = 33). Six samples per group (n = 6) were used for Seahorse assays. For metabolomics analyses, in order to meet the minimum sample size requirement, every 6–7 samples per group were pooled, yielding 4 pooled samples (n = 4).

The Morris water maze test was conducted over 7 days to assess spatial learning and memory. On day 1, a visible platform trial was performed with the platform raised 1 cm above the water surface. The average swimming speed was recorded to exclude potential confounding effects of visual or motor impairments on subsequent results. From days 2 to 6, the place navigation trial was conducted. Mice were introduced into the pool from four different quadrant entry points and given 60 s to locate a submerged platform (1 cm below the water surface). Any mouse that failed to find the platform within 60 s was manually guided to it and remained there for 10 s. Escape latency was recorded to evaluate spatial learning ability. On day 7, the spatial probe trial was performed by removing the platform and allowing each mouse to swim freely for 60 s. Swimming trajectory, number of platform crossings, and time spent in the platform quadrant were recorded. All data were automatically collected and analyzed using a Morris water maze video tracking system (Noldus, Netherlands). Throughout the testing period, consistent environmental conditions (lighting, temperature, and noise) were maintained for all animals to minimize external influences.

Soluble and insoluble hippocampal protein fractions were prepared according to an established protocol (Velazquez et al., 2016). Briefly, six mice per group were anesthetized by intraperitoneal injection of 10% chloral hydrate (3.5 mL/kg body weight) and no typical symptoms of peritonitis were appeared until the sample preparation. Hippocampal tissues were rapidly dissected on ice and homogenized in 10 volumes of RIPA lysis buffer containing protease and phosphatase inhibitors. The homogenate was centrifuged at 100,000 × g for 1 h at 4 °C to obtain the soluble fraction (supernatant). The resulting pellet was resuspended in 70% formic acid and centrifuged again under identical conditions to collect the insoluble fraction (supernatant). Protein concentrations were normalized to 1 mg/mL using a BCA assay kit. The soluble fraction was used for Western blot and ELISA detection of soluble Aβ, while the insoluble fraction was reserved for ELISA detection of insoluble Aβ.

Soluble and insoluble Aβ40 and Aβ42 levels were quantified using commercial ELISA kits (Aβ40: MEXN-H0951; Aβ42: MEXN-H0921; Meixuan). Following the manufacturer’s instructions, equal amounts of soluble or insoluble samples were added to a 96-well microplate. After sequential incubation, washing, color development, and termination steps, the optical density was measured at 450 nm using a microplate reader.

Western blot analysis was performed using standard procedures, including gel preparation, sample loading, electrophoresis, membrane transfer, antibody incubation, and chemiluminescent detection. The following primary antibodies were used: NLRP3 (19771-1-AP, Proteintech), ASC (10500-1-AP, Proteintech), caspase-1 p20 (341,030, Zenbio), IL-1β (A20527, Abclonal), and GAPDH (D190090, BBI).

Following deep anesthesia, the remaining mice from each group were transcardially perfused with pre-cooled PBS. The hippocampus was rapidly dissected, and single-cell suspensions were prepared through sequential mechanical dissociation, enzymatic digestion, filtration, centrifugation, and washing steps. Cells were resuspended in staining buffer and incubated with an Fc receptor blocker for 10 min at 4 °C. Subsequently, cells were stained with CD45-FITC (E-AB-F1136C, Elabscience) and CD11b-PECY7 (E-AB-F1018HC, Elabscience) antibodies for 30 min at 4 °C in the dark. After washing and filtration, cells were resuspended in FACS buffer and sorted using a BD FACSAria III flow cytometer. The collected microglia were prepared for subsequent LC–MS/MS targeted metabolomics and Seahorse assays.

For each cell sample (10⁶ cells), 1,000 μL of pre-cooled 80% methanol aqueous solution was added. Samples were vortex-mixed, sonicated in an ice bath for 20 min, and incubated at −20 °C for 1 h. The mixtures were centrifuged at 16,000 × g for 20 min at 4 °C. The supernatant was evaporated using a high-speed vacuum concentrator. The dried samples were reconstituted in pre-cooled 50% methanol aqueous solution and centrifuged at 20,000 g 4 °C for 15 min. The supernatant was collected for LC–MS/MS analysis. Chromatographic separation was performed using a Shimadzu Nexera X2 LC-30 AD system. Mass spectrometric detection was conducted on a QTRAP 6500 + instrument (AB SCIEX) in positive and negative ion modes with multiple reaction monitoring (MRM) for metabolite detection. MultiQuant software was used to extract chromatographic peak areas and retention times.

The Cell Mitochondrial Stress Test and Glycolysis Rate Test were performed using a Seahorse XF96 Analyzer (Agilent, Santa Clara, US) following established protocols (Holland et al., 2018; Mela et al., 2020). Microglia were seeded into Seahorse XF culture plates at a density of 9 × 10³ cells per well and cultured overnight prior to the assays. For the mitochondrial stress test, the oxygen consumption rate (OCR) was measured under basal conditions and following the sequential injection of oligomycin (1.5 μmol/L, #495455, Sigma-Aldrich, UK), FCCP (0.5 μmol/L, #C2920, Sigma-Aldrich, UK), and a mixture of antimycin A and rotenone (0.5 μmol/L, #A8674, Sigma-Aldrich, UK). For the glycolysis rate test, the extracellular acidification rate (ECAR) was measured under basal conditions and after sequential injections of oligomycin (1.5 μmol/L) and 2-deoxy-D-glucose (50 mmol/L, #D8375, Sigma-Aldrich, UK). Data were recorded at 10-min intervals and analyzed using Seahorse Wave software.

Statistical analysis and visualization were performed using GraphPad Prism 9.3.0. All data are presented as mean ± standard error of the mean (SEM). Data from the place navigation trial were analyzed using two-way repeated measures analysis of variance (ANOVA) with training day and group as factors, followed by Tukey’s post-hoc test. For other datasets, normality and homogeneity of variance were first assessed. Data meeting both assumptions were analyzed by one-way ANOVA followed by Tukey’s test. Data with normal distribution but heterogeneous variance were analyzed using Welch’s ANOVA followed by Dunnett’s T3 test. Data not normally distributed were analyzed with the Kruskal-Wallis test followed by Dunn’s post-hoc test. A p-value of less than 0.05 was considered statistically significant.

Impaired spatial learning and memory is one of the most prominent behavioral performance of AD. To evaluate the effects of treadmill exercise on these cognitive deficits, the classic Morris Water Maze test were employed (Figure 1). The visible platform trial was conducted firstly as a critical baseline assessment, and the results showed that there was no significant difference in the average swimming speed among the three groups (p > 0.05, Figure 1A), eliminating the potential influence of vision and motor function on subsequent results in place navigation trial and spatial probe trial.

Subsequently, the place navigation trial was conducted and the escape latency was analyzed to evaluate the spatial learning ability (Figure 1B). Two-way repeated measures ANOVA revealed a significant main effect of training day (F = 10.8, p < 0.0001), but no significant main effect of group (F = 1.265, p = 0.296) or interaction effect between training day and group (F = 1.839, p = 0.091). The result is highly consistent with the biological logic of spatial learning, which argues that with the increase of training day, mice gradually establish memory of the platform’s spatial location, resulting in progressive reductions in escape latency. To further elucidate the spatial learning differences, post-hoc analyses were performed. Within-group comparisons indicated that compared to their respective Day 1 performance, the escape latency of the WT-SED mice was significantly shortened on Day 4 (p = 0.0448) and continued to shorten on Day 5 (p = 0.0036), while AD-EXE mice only showed a significant shorten on Day 5 (p = 0.0202). Between-group comparisons on the same day revealed that on Day 5, AD-SED mice exhibited longer escape latency than WT-SED (p = 0.0102), while the escape latency in AD-EXE mice was significantly shortened compared to AD-SED mice (p = 0.0475). The results indicate that 6-month-old APP/PS1 mice exhibit impaired spatial learning ability, which can be partially restored following a three-month treadmill exercise intervention.

Then, the spatial probe trial was conducted to evaluate the spatial memory ability, with the representative swimming trajectories and heat maps depicted in Figure 1C. Analysis of platform crossings (Figure 1D) demonstrated that AD-SED mice had significantly fewer platform crossings than both WT-SED (p < 0.0001) and AD-EXE (p = 0.0334) mice. However, no significant differences were observed in the percentage of time spent in the target quadrant among the three groups (p > 0.05, Figure 1E). These results indicate that spatial memory were impaired in 6-month-old APP/PS1 mice, and that treadmill exercise ameliorated this deficit.

Collectively, the results from Morris Water Maze indicate that 6-month-old APP/PS1 mice exhibit deficits in spatial learning and memory abilities, while the three-month treadmill exercise effectively improves these cognitive deficits.

Aβ accumulation is a core pathological hallmark of AD and contributes to cognitive deficits through exerting neurotoxic effects, including disrupting synaptic function, triggering neuroinflammation, etc. To examine whether treadmill exercise modulates this core pathology, ELISA was used to quantify soluble and insoluble Aβ40 and Aβ42 levels in the hippocampus (Figures 2A–D). AD-SED mice exhibited significantly elevated hippocampal levels of soluble Aβ40 (p < 0.0001, Figure 2A), insoluble Aβ40 (p < 0.0001, Figure 2B), soluble Aβ42 (p < 0.0001, Figure 2C), and insoluble Aβ42 (p < 0.0001, Figure 2D) compared with the WT-SED group, indicating a widespread Aβ accumulation. Following exercise intervention, hippocampal levels of soluble Aβ40 (p < 0.0001, Figure 2A), insoluble Aβ40 (p = 0.0052, Figure 2B), soluble Aβ42 (p < 0.0001, Figure 2C), and insoluble Aβ42 (p < 0.0001, Figure 2D) in AD-EXE mice were all significantly decreased compared with the AD-SED mice. These results indicate that treadmill exercise significantly reduces hippocampal Aβ accumulation, a key contributor that may underlie the observed cognitive improvements.

In addition to Aβ accumulation, chronic neuroinflammation is also a key pathological driver of AD progression. A line of evidence has confirmed that microglial NLRP3 inflammasome activation is a central mediator in this process, which triggers the cleavage of pro-caspase-1 into mature p20 and subsequently promoting the production of interleukin-1β (IL-1β) (Chen et al., 2023). To examine whether treadmill exercise regulates this inflammatory pathway, western blotting was used to evaluate the expression of NLRP3 inflammasome-associated proteins in the hippocampus(Figure 2E). Results showed that AD-SED mice displayed increased activation of NLRP3 inflammasome pathway compared with WT-SED mice, with significantly increased protein levels of NLRP3 (p < 0.0001, Figure 2F), ASC (p < 0.0001, Figure 2G), caspase-1 p20 (p < 0.0001, Figure 2H), and IL-1β (p < 0.0001, Figure 2I) in the hippocampus. Following treadmill exercise, AD-EXE mice showed significant reduction in protein levels of NLRP3 (p < 0.0001, Figure 2F), ASC (p < 0.0001, Figure 2G), caspase-1 p20 (p < 0.0001, Figure 2H), and IL-1β (p = 0.0003, Figure 2I) compared with AD-SED mice. These findings indicate that treadmill exercise effectively mitigates NLRP3 inflammasome-triggered neuroinflammation in APP/PS1 mice.

Emerging evidence highlights that microglial glucose metabolic reprogramming from oxidative phosphorylation to glycolysis leads to their functional impairments, manifested as reduced Aβ clearance and increased inflammatory activation (McIntosh et al., 2019; Cheng et al., 2021). Both of these factors contribute to increased Aβ accumulation and chronic neuroinflammation. Given the critical role of microglial glucose metabolism and the observed exercise-induced reduction in Aβ accumulation and chronic neuroinflammation, we further examined whether treadmill exercise modulates microglial glucose metabolism. To address this, Seahorse assays were performed to measure ECAR and OCR, which indicative of glycolysis and OXPHOS, respectively. ECAR and OCR profiles was generated and presented in Figures 3A,B. Compared with WT-SED microglia, AD-SED microglia exhibited significant reductions in glycolysis (p = 0.0007, Figure 3C), glycolytic capacity (p < 0.0001, Figure 3D), basal respiration (p < 0.0001, Figure 3E), and maximal respiration (p < 0.0001, Figure 3F), indicating the impairments in both glycolysis and OXPHOS. Moreover, the OCR/ECAR ratio was also significantly decreased in AD-SED microglia compared with WT-SED microglia (p = 0.0169, Figure 3G), indicating a shift toward a less efficient glycolysis-dominant metabolic phenotype. Following treadmill exercise, AD-EXE microglia showed significant increases in glycolysis (p = 0.0095, Figure 3C), glycolytic capacity (p < 0.0001, Figure 3D), basal respiration (p = 0.001, Figure 3E), and maximal respiration (p < 0.0001, Figure 3F) compared with AD-SED microglia, with a concurrent increase in the OCR/ECAR ratio (p = 0.0439, Figure 3G). These results demonstrate that treadmill exercise restores both glycolytic and OXPHOS in microglia, while balances their contribution ratio.

To validate these metabolic findings observed in Seahorse assays, LC–MS/MS targeted metabolomics was further used to quantify key glucose metabolites abundance in microglia (Figure 4). These results showed that compared with WT-SED microglia, AD-SED microglia exhibited significantly decreased levels of lactate (a glycolytic end product, p = 0.0246, Figure 4F), fumarate (a TCA cycle intermediate, p = 0.0489, Figure 4I), and malate (another TCA cycle intermediate, p = 0.0031, Figure 4J). The results corroborated the impairment of glycolysis and OXPHOS observed in the Seahorse assays. Moreover, AD-SED microglia also exhibited increased oxaloacetate (p = 0.0464, Figure 4K), ADP (p = 0.036, Figure 4L), AMP (p = 0.0401, Figure 4M), ADP/ATP ratio (p = 0.036, Figure 4O) and AMP/ATP ratio (p = 0.0306, Figure 4P), indicating a potential energy deficit. Following treadmill exercise, AD-EXE microglia showed significantly decreased fumarate (p = 0.0336, Figure 4I), AMP (p = 0.0078, Figure 4M) and AMP/ATP ratio (p = 0.0091, Figure 4P) compared with AD-SED microglia, indicating partial normalization of metabolite perturbations.

Collectively, these metabolic findings demonstrate that microglia from 6-month-old APP/PS1 mice exhibit impairments in both glycolysis and OXPHOS, with a metabolic profile predominantly reliant on glycolysis. Treadmill exercise partially corrects AD-induced metabolic abnormalities by enhancing both glycolysis and OXPHOS while increasing the relative contribution of OXPHOS to energy production.