Abstract
Aerobic exercise (AE) has been shown to offer significant benefits for Alzheimer's disease (AD), potentially influencing the gut microbiota. However, the impact of changes in intestinal flora in early Alzheimer's disease induced by aerobic exercise on metabolic pathways and metabolites is not well understood. In this study, 3‐month‐old APP/PS1 and C57BL/6 mice were divided into two groups each: a control group (ADC for APP/PS1 and WTC for C57BL/6) and an aerobic exercise group (ADE for APP/PS1 and WTE for C57BL/6). The exercise groups underwent a 20‐week aerobic training program on a motorized treadmill before the behavioral test (both the Morris water maze experiment (MWM) and the eight‐arm maze test). Fecal samples were collected to analyze gut microbiota profiles via 16S rRNA gene sequencing. At the same time, the metabolic pathway analysis and the detection of metabolites were carried out. At the phylum level, the ADE group exhibited a significant reduced in the relative abundance of Bacteroidetes compared to the ADC group. At the genus level, both Ileibacterium and Faecalibaculum were found to be more abundant in the ADE group than in the ADC group. Additionally, PICRUSt analysis revealed that lipid metabolism and bile acid metabolism pathways were significantly enriched in the cecal microbiota of mice in the ADE group. The metabolites detected further confirmed the changes in the metabolic pathways mentioned above. Aerobic exercise may modify gut microbiota profiles and metabolites in APP/PS1 mice, thereby potentially playing a beneficial role in delaying cognitive impairment associated with early‐stage Alzheimer's disease.
Keywords: Alzheimer's disease, exercise, gut microbiota, KEGG pathways, metabolites
1. INTRODUCTION
Alzheimer's disease (AD) is an irreversible neurodegenerative disorder characterized by a progressive decline in function, cognition, and behavior. 1 The accumulation of amyloid plaques and neurofibrillary tangles represents the cardinal pathological changes associated with the disease. Currently, AD presents a high rate of disability and lacks effective curative treatments, imposing a significant burden on patients and their families. 1 Low levels of physical activity are recognized as a risk factor for AD. 2 Exercise has been shown to delay the progression of the disease, 3 , 4 with benefits persisting even after exercise cessation. 2 Research indicates that exercise induces changes in the brain at not only the anatomical level but also the cellular and molecular levels. 2 The mechanisms through which exercise enhances cognitive function in AD patients are complex. Dysregulation of the “microbiota–gut–brain axis” and metabolomic alterations have been observed in both animal studies and human patients with AD. 5 , 6 , 7 , 8 Notably, several metabolic pathways related to the pathophysiology of AD have been identified, including glycerophospholipid metabolism, linoleic acid metabolism, bile secretion, and phenylalanine metabolism. In AD patients who engage in exercise, increased levels of polyunsaturated free fatty acids (PUFAs) and reductions in ceramides, sphingolipids, and phospholipids have been documented. 5
Dysregulation of the “microbiota–gut–brain axis” has increasingly been implicated in the pathophysiology of AD. 9 , 10 Evidence suggests that amyloid may function as an antimicrobial peptide within the brain. 11 Alterations in microbiota have been reported in AD mouse models, indicating that dysbiosis of gut microbiota may contribute to the disease. 12 , 13 Improving microbiome balance could potentially aid in the prevention of AD. 7 , 14 , 15 A recent study has indicated that changes in gut microbiota may play a role in human aging. 16 Numerous studies have demonstrated that the composition of gut microbiota can be modified by exercise, particularly concerning probiotic taxa. 17 , 18 , 19 , 20 While gut microbiota changes have been identified as a significant mechanism through which exercise may enhance cognition in AD and influence human aging, results have been inconsistent. 16 , 21 , 22 Some studies report that exercise downregulates Bacteroidetes abundance, whereas other studies found that Bacteroidetes, identified as probiotics, are more abundant in younger individuals. 19 , 22 Some previous studies have demonstrated that exercise can enhance the cognitive function of AD mice. 21 , 23 However, other research suggests that combining exercise with probiotics may be necessary to achieve significant improvements in cognition. 22 Additionally, the effects of exercise on the gut microbiota of AD mice have been inconsistent across different studies. Previous studies have primarily focused on either exercise‐induced alterations in gut microbiota or changes in gut metabolites separately, failing to directly elucidate the interrelationship between the two. 21 , 22 , 23 Therefore, we conducted a 20‐week aerobic exercise regimen for AD mice, monitoring changes in their swimming speed before and after training, followed by assessments of cognitive function, gut microbiota, and metabolite.
We hypothesized that aerobic exercise (AE) could enhance cognitive function and induce beneficial changes in the gut microbiota of AD mice when applied at appropriate intensity and duration, while these changes may induce alterations in gut metabolism. This study aimed to investigate the effects of AE on gut microbiota and gut metabolism in APP/PS1 mice using 16S rRNA gene sequencing.
2. MATERIALS AND METHODS
2.1. Animal grouping
The timeline for this experiment is shown in Figure 1. All protocols were approved by the Animal Studies Committee of Chengdu Sport University. Three‐month‐old APP/PS1 mice (AD mice, n = 12 per group) and C57BL/6 mice (wild‐type, WT mice, n = 12 per group) were obtained from the Laboratory Animal Center of Beijing Huafukang Biotechnology Co., Ltd. [License: SCXK (Beijing) 2019‐0008]. The mice were housed in pathogen‐free conditions with a 12:12 h light–dark cycle and had ad libitum access to food and water. They were divided into control groups (CON; nADC = 12, nWTC = 12) and aerobic exercise groups (AE; nADE = 12, nWTE = 12). Sixteen mice (four per group) were randomly selected for gut microbiota analysis (16S), and all mice completed behavioral assessments and twenty‐four mice (six per group) were randomly selected for fecal metabolomics evaluations.
FIGURE 1.
Timeline of the experimental process.
2.2. Exercise protocol and fecal sample collection
Mice in the AE groups underwent a 20‐week training regimen on a motorized treadmill, starting at a speed of 12 m/min for 10 min and progressing to 15 m/min for 20 min (one session per day, five days per week). The CON groups were placed in the exercise environment but did not undergo training. Following the intervention, Morris water maze (MWM) tests and Eight‐arm maze test were conducted, and a 12‐h fast was implemented prior to the 16S rRNA gene analysis of cecal feces.
2.3. Behavioral tests
2.3.1. Morris water maze test
The MWM test, conducted 20 weeks after the exercise intervention, evaluated spatial learning and memory. This test included a spatial acquisition phase followed by a detection phase, both carried out in a controlled environment. 24 The circular pool (122 cm in diameter, 50 cm high, and 30 cm deep) was divided into quadrants, with a hidden platform located in the northeast quadrant. The water temperature was maintained between 20 and 22°C. The testing period spanned seven days, beginning with an adaptive swim on Day 1, followed by five days of spatial acquisition trials (four trials per day). Escape latency, defined as the time taken to reach the platform, was recorded. On Day 7, the platform was removed, and the mice swam for 120 s. Metrics including time spent in the target quadrant, distance traveled, and number of platform crossings were recorded using ANY‐maze software (Stoelting Co., IL, USA).
2.3.2. Eight‐arm maze test
This experiment aimed to evaluate working and reference memory in mice. The mice were introduced to the eight‐arm maze environment for one week. The elevated eight‐arm radial maze was positioned 60 cm above the laboratory floor. Each arm measures 35 cm in length and 5 cm in width, surrounded by 25 cm high black plexiglass plate walls (Sans Bio SA203). At the end of each arm, there is a ceramic food dish embedded in the maze, with a diameter of 3.0 cm. The entrance to each arm features transparent black plexiglass plate doors that can be opened and closed remotely. Before testing, each group of mice was weighed and fasted for 24 h. After each day's experiment, they received a restricted diet of 2–3 grams to maintain their weight at 80%–85% of that of normally fed mice. On Days 1 and 2, food pellets (4–5 per mouse, each 3–4 mm in diameter) were placed in each arm of the maze as well as in the central area. Four mice were simultaneously positioned in the center of the maze, and the gates to each arm were opened to allow free access to the food for 10 min. On Days 3 and 4, mice were trained individually, with food pellets placed in each arm for 10 min or until consumed. On Days 5 and 6, food pellets were placed in arms 1, 3, 5, and 7, with retrieval occurring after the food was consumed or after 10 min had passed. The first six days included testing twice daily, with intervals exceeding one hour. On Day 7, food was again placed in arms 1, 3, 5, and 7 with the gates closed. Mice were positioned in the center of the maze and released after a 30‐s delay to access the food freely, with retrieval occurring once the food was finished or after 10 min. Working memory errors (WME), reference memory errors (RME), and the total number of working and reference memory errors per mouse were recorded. After each mouse's experiment during both the learning and testing phases, the maze was cleaned with a cloth soaked in disinfectant alcohol.
2.3.3. Statistical analysis
All quantitative data were expressed as mean ± standard deviation (M ± SD) and analyzed using SPSS statistical analysis software. In the Morris water maze experiment, the spatial navigation test employed a three‐factor mixed design with a 2 (genotype) × 2 (intervention method) × 6 (time) repeated‐measures ANOVA. The spatial exploration experiment utilized a two‐factor between‐subjects ANOVA with a 2 (genotype) × 2 (intervention method) design. The remaining experiments also included a two‐factor between‐subjects ANOVA with a 2 (genotype) × 2 (intervention method). Throughout the analysis, non‐sphericity was corrected using the Greenhouse–Geisser method, and post‐hoc comparisons were conducted using the Bonferroni method. The significance level was set at p < .05. Image processing was performed using GraphPad Prism 8 and Photoshop software.
2.4. 16S rRNA gene sequencing and microbiota analysis
Genomic DNA was extracted from fecal samples using the MagAtract PowerSoil Pro DNA Kit (Qiagen, Hilden, Germany). Sixteen samples (n = 4 per subgroup) were subjected to microbiota analysis. PCR amplification targeted the V4 region of bacterial 16S rRNA genes, with sequencing conducted on the Illumina HiSeq3000 platform. Data analysis was performed using QIIME II software, which included quality filtering, OTU clustering, and the calculation of diversity indices (Chao1, ACE, Shannon, and Simpson). Bacterial abundance at the phylum and genus levels was analyzed using Metastats software. Principal coordinates analysis (PCoA) based on Bray‐Curtis distances was utilized to assess microbial community similarity, while PERMANOVA was employed to analyze group differences. Linear discriminant analysis Effect Size (LefSe) identified biomarkers with adjustments for the Kruskal–Wallis test. PICRUSt2 software (http://huttenhower.sph.harvard.edu/galaxy) was used to predict the function of 16S amplicon sequencing. Initially, the OTU abundance table was standardized using PICRUSt, which stores COG and KO information corresponding to Greengenes IDs, thereby eliminating the influence of 16S marker gene copy number variations within species genomes. Subsequently, COG family information and KEGG Ortholog (KO) data associated with each OTU were retrieved based on their respective Greengenes IDs. The abundances and KO abundances for each COG were then calculated. Utilizing the COG database information (evolutionary genealogy of genes: Non‐supervised Orthologous Groups, http://eggnog.embl.de/), descriptive details regarding each COG along with its functional attributes were analyzed through the eggNOG database, resulting in a functional abundance spectrum. Furthermore, leveraging KEGG database resources (Kyoto Encyclopedia of Genes and Genomes, Kyoto Encyclopedia of Genes and Genomes, http://www.genome.jp/kegg/) allowed for the extraction of KO, pathway, and EC information, thus enabling calculation of abundance across various functional categories based on OTU abundances. Additionally, concerning pathways, three levels of metabolic pathway information while generating separate abundance tables for each level were accessed by PICRUSt2.
2.5. Microbiota mass spectrometry conditions and data analysis
The analysis of metabolic signatures was performed using the Waters ACQUITY UPLC system in conjunction with an AB Triple TOF 5600 mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). This setup utilized a heated electrospray ionization (ESI) source to operate in both ESI‐positive and ESI‐negative ion modes. The high‐energy collision dissociation (HCD) MS/MS resolution was set at 35 000, while the overall mass spectrometry scan resolution was 70 000. A collision energy of 10 eV was selected for the experiments. The raw HPLC‐MS data were analyzed using Progenesis QI V2.3 software (Nonlinear Dynamics, Newcastle, UK). The product tolerance was set at 10 ppm, with a product ion threshold of 5% and a precursor tolerance of 5 ppm. The resulting three‐dimensional dataset, saved in an Excel file, included peak identification, peak matching, and peak alignment. Data processing was further enhanced with Native QI software (Waters Corporation, Milford, CT, USA), which utilized public databases such as HMDB (http://www.hmdb.ca) and METLIN (http://www.lipidmaps.org/METLIN) to identify metabolites. The normalized data were then imported into R for orthogonal partial least squares discriminant analysis (O)PLS‐DA, allowing for an intuitive visualization of metabolic changes between the experimental groups. To determine the significance of variables affecting projections generated from the OPLS‐DA model, a threshold value known as Variable Importance in Projections (VIP) was applied. Substances with a VIP greater than 1.5 were selected as candidate biomarkers if they also met the criteria of a p‐value less than .05 following one‐way ANOVA (SPSS 22.0, Amenk, NY, USA) and t‐tests. Additionally, the random forest approach was employed to analyze feature importance ranking between the ADC and ADE groups, using Mean Decrease Accuracy as the assessment indicator.
3. RESULTS
3.1. Behavioral test
3.1.1. Morris water maze experiment
The mean escape latency decreased significantly with training in the WTE and ADE groups compared to the WTC and ADC groups (Tables 1, 2, 3, 4 and Figure 2A–E). Post‐training, the time spent in the target quadrant and the number of platform crossings were significantly higher in the exercise groups compared to the non‐exercise groups (both WT and AD groups). Notably, swimming speed was significantly greater in the ADE group following training (Table 2 and Figure 2B), with the ADE group exhibiting the longest swimming distance before reaching the platform (SDBS) (Table 4 and Figure 2E).
TABLE 1.
Behavioral test—Morris water maze experiment. The mean escape latency (MWM) of mice.
| WTC (n = 12) | WTE (n = 12) | ADC (n = 12) | ADE (n = 12) | |
|---|---|---|---|---|
| Day 1 | 106.95 ± 17.08 | 106.58 ± 12.08 | 118.39 ± 15.25 ✱ | 115.22 ± 9.55 |
| Day 2 | 93.113 ± 18.45 | 88.71 ± 18.07 | 100.15 ± 16.86 | 95.45 ± 22.68 |
| Day 3 | 70.41 ± 15.71 | 43.22 ± 16.55 | 90.43 ± 19.10 | 80.59 ± 11.66 |
| Day 4 | 54.28 ± 18.17 | 33.08 ± 10.51 ✱ | 74.26 ± 12.36 ✱ | 56.98 ± 11.37 ● , ▼ |
| Day 5 | 40.93 ± 11.63 | 19.47 ± 6.82 ✱ | 68.27 ± 22.36 ✱ | 44.39 ± 11.37 ● , ▼ |
| Day 6 | 25.73 ± 12.36 | 16.17 ± 7.87 ✱✱ | 64.47 ± 17.84 ✱ | 31.03 ± 4.47 ● , ▼ |
p < .05 compared with WTC group.
p < .001 compared with WTC group.
p < .05 compared with WTE group.
p < .05 compared with ADC group.
TABLE 2.
The mean average speed of mice.
| WTC (n = 12) | WTE (n = 12) | ADC (n = 12) | ADE (n = 12) | |
|---|---|---|---|---|
| Day 1 | 11.52 ± 1.17 | 12.68 ± 2.03 | 9.92 ± 1.18 ✱ | 13.91 ± 2.94 ▼▼ |
| Day 2 | 10.22 ± 0.97 | 11.15 ± 1.75 | 9.48 ± 2.09 | 15.43 ± 2.96 ●● , ▼▼ |
| Day 3 | 9.85 ± 2.21 | 11.42 ± 1.73 | 8.31 ± 1.54 ✱ | 15.84 ± 3.37 ●● , ▼▼ |
| Day 4 | 10.01 ± 1.27 | 11.03 ± 1.70 | 7.56 ± 1.04 ✱✱ | 16.04 ± 3.14 ●● , ▼▼ |
| Day 5 | 9.81 ± 1.52 | 10.23 ± 1.26 | 7.71 ± 0.89 ✱✱ | 14.94 ± 2.04 ●● , ▼▼ |
| Day 6 | 9.26 ± 1.68 | 10.40 ± 0.87 | 7.30 ± 1.58 ✱✱ | 13.72 ± 3.38 ●● , ▼▼ |
p < .05 compared with WTC group.
p < .001 compared with WTC group.
p < .05 compared with WTE group.
p < .001 compared with WTE group.
p < .05 compared with ADC group.
p < .001 compared with ADC group.
TABLE 3.
The times in target quadrant and number of platform crossing.
| Times in target quadrant (n = 12) | Number of platform crossing (n = 12) | |
|---|---|---|
| WTC | 44.26 ± 8.23 | 3.17 ± 0.75 |
| WTE | 57.63 ± 3.60 ✱ | 5.83 ± 1.04 ✱✱ |
| ADC | 20.81 ± 4.90 ✱ | 0.83 ± 0.15 ✱✱ |
| ADE | 36.22 ± 3.18 ▼▼ | 2.62 ± 1.21 ●● , ▼ |
p < .05 compared with WTC group.
p < .001 compared with WTC group.
p < .05 compared with WTE group.
p < .001 compared with WTE group.
p < .05 compared with ADC group.
p < .001 compared with ADC group.
TABLE 4.
Swimming distance before going on stage.
| Day/groups | WTC (cm) (n = 12) | WTE (cm) (n = 12) | ADC (cm) (n = 12) | ADE (cm) (n = 12) |
|---|---|---|---|---|
| Day 1 | 1245.58 ± 198.05 | 1608.42 ± 180.14 ✱ | 1292.02 ± 198.57 | 1957.10 ± 337.08 ▼▼ |
| Day 2 | 1046.67 ± 304.01 | 1439.17 ± 424.68 ✱ | 1028.58 ± 347.35 | 1909.88 ± 421.67 ● , ▼▼ |
| Day 3 | 759.27 ± 135.98 | 787.10 ± 175.90 | 936.23 ± 176.44 ✱✱ | 1912.39 ± 453.43 ●● , ▼▼ |
| Day 4 | 724.32 ± 129.36 | 586.92 ± 148.39 | 909.42 ± 155.45 ✱ | 1193.27 ± 345.40 ● |
| Day 5 | 614.87 ± 141.33 | 426.83 ± 192.78 ✱ | 763.57 ± 87.94 ✱✱ | 1063.23 ± 308.11 ●● , ▼ |
| Day 6 | 553.01 ± 102.08 | 402.785 ± 104.73 ✱ | 771.64 ± 107.87 ✱✱ | 1004.44 ± 164.05 ●● , ▼▼ |
p < .05 compared with WTC group.
p < .001 compared with WTC group.
p < .05 compared with WTE group.
p < .001 compared with WTE group.
p < .05 compared with ADC group.
p < .001 compared with ADC group.
FIGURE 2.
Results of the behavioral test (both the Morris water maze experiment (MWM) and the eight‐arm maze test) in each group. (A–E) MWM test (n = 12 per group). (A) Escape time of the mice in each group during the 6‐day learning period. (B) Average swimming speed during the 6‐day learning period. (C) Time spent in target quadrant at sixth day. (D) Number of crossings of the platform at sixth day. (E) Path of navigational activity in MWM at sixth day after platform's removal. The area enclosed by the blue line indicates the platform quadrant, and the small blue circle in this quadrant indicates the platform. The round point is the swimming start point, the square point is the end point, and the red line between both points is the swimming path. (F–J) The eight‐arm maze test (n = 12 per group). (F) Number of working memory errors at sixth day. (G) Rate of working memory errors at sixth day. (H) Number of reference memory errors at sixth day. (I) Rate of reference memory errors at sixth day. (J) Activity trajectory in the eight‐arm maze at sixth day after platform's removal. The area enclosed by the blue line indicates the platform quadrant. The square point is the swimming start point, the round point is the end point, and the red line between both points is the mouse movement trajectories. Data are presented as the mean ± SD (n = 12 per group). *p < .05 compared with WTC group; **p < .001 compared with WTC group; ● p < .05 compared with WTE group; ●● p < .001 compared with WTE group; ▼ p < .05 compared with ADC group; ▼▼ p < .001 compared with ADC group.
3.1.2. Eight‐arm maze experiment
The number of working memory errors observed is summarized in Table 5. Mice in the ADE group exhibited significantly fewer working memory errors compared to those in the ADC group (p < .05). In contrast, the ADC group showed significantly more errors than the WTC group (p < .05), while the ADE group made significantly more errors than the WTE group (p < .05) (Figure 2F,G). Regarding reference memory, the WTE group had significantly fewer errors than the WTC group (p < .05). The ADE group recorded significantly fewer errors than the ADC group (p < .05) but significantly more than the WTE group (p < .05). Additionally, the ADC group made significantly more errors than the WTC group (p < .05) (Figure 2H,I).
TABLE 5.
Results of the eight‐arm maze test.
| Number of working memory errors | Rate of working memory errors | Number of reference memory errors | Rate of reference memory errors | |
|---|---|---|---|---|
| WTC (n = 12) | 1.58 ± 0.51 | 12.18 ± 2.63 | 3.83 ± 1.28 | 39.17 ± 9.98 |
| WTE (n = 12) | 0.67 ± 0.49 ✱✱ | 7.17 ± 5.31 ✱✱ | 1.50 ± 1.00 ✱ | 18.73 ± 5.35 ✱✱ |
| ADC (n = 12) | 3.50 ± 1.08 ✱✱ | 21.59 ± 4.67 ✱✱ | 8.17 ± 2.52 ✱✱ | 51.78 ± 7.91 ✱ |
| ADE (n = 12) | 2.08 ± 0.79 ● , ▼▼ | 11.99 ± 8.15 ● , ▼▼ | 5.25 ± 1.60 ●● , ▼▼ | 39.46 ± 11.96 ●● , ▼▼ |
p < .05 compared with WTC group.
p < .001 compared with WTC group.
p < .05 compared with WTE group.
p < .001 compared with WTE group.
p < .05 compared with ADC group.
p < .001 compared with ADC group.
3.2. 16S rRNA gene sequencing and microbiota analysis
3.2.1. Sequence splicing, OTU cluster, and alpha diversity index analysis of gut microbiota
Over 150 000 valid tags were obtained from the intestinal bacteria sequencing data of each group. No significant differences were observed in the Chao1, ACE, Simpson, or Shannon indices among the groups. The relative abundance distribution of species at the phylum level was visually represented using stacked histograms. A total of 516 OTU clusters were identified across 16 samples, with 98 OTUs shared among all groups. Unique OTUs included 8 OTUs in the WTC group, 4 OTUs in the ADC group, 7 OTUs in the ADE group, and 9 OTUs in the WTE group (Figure 3A).
FIGURE 3.
Effects of aerobic exercise on gut microbes. (A) Venn diagram of OUT in each group (n = 4 per group) at the species level. (B) PCoA on OUT level; PCoA cluster analysis showed that there were differences in intestinal microbiota between the exercise group and the non‐exercise group. (C–E) Species abundance of gut microbiota in four groups. (C) Species abundance at the phylum level in each group. (D) Species abundance at the class level in each group. (E) Species abundance at the genera level in each group. (F) Difference of gut microbiota at general level between ADC group and ADE group. (G) Linear discriminant analysis (LefSe), Bar plot from LefSe analysis indicating enriched bacterial genera associated with four groups. Those taxa in each level are colored for which it is more abundant (p < .05; LDA score 2).
3.2.2. Altered gut microbial taxa at the OTU levels
The relative abundance of gut flora at the phylum, class, and genus levels is illustrated in Figure 3C–E. At the phylum level, the predominant microbiota in all groups included Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes, and Patescibacteria, collectively comprising over 90% of the microbiota. At the class level, the dominant microbiota consisted of Bacilli, Erysipelotrichia, Gammaproteobacteria, Actinobacteria, Clostridia, Bacteroidia, and Deltaproteobacteria, also accounting for over 90%. At the genus level, the five most prevalent microbial taxa varied, with Psychrobacter and Staphylococcus being predominant across all groups.
Further analysis revealed differences in gut microbiota based on disease status and exercise. At the phylum level, compared to WT groups (WTE + WTC), the relative abundance of Proteobacteria was significantly higher in the AD groups (ADE + ADC). Additionally, the relative abundance of Actinomycota was significantly increased in the exercise groups (ADE + WTE) compared to the non‐exercise groups (ADC + WTC). Notably, the abundance of Bacteroidetes in the ADC group was higher than that in both WT groups. ADE group demonstrated significantly decreased levels of Bacteroidetes compared to ADC group, although there were no differences in Bacteroidetes abundance between the WTE and WTC group (Figure 3C).
At the class level, the relative abundance of Saccharimonadia was decreased in AD groups compared to WT groups, while Gammaproteobacteria abundance was significantly higher in AD groups, with no differences observed between the ADC and ADE groups (Figure 3D).
At the genus level, AD groups demonstrated significantly increased levels of Psychrobacter compared to WT groups, although there were no differences in Psychrobacter abundance between the ADE and ADC groups. Furthermore, Ileibacterium and Faecalibaculum were found to be more abundant in the ADE group than in the ADC group (Figure 3E,F).
LefSe analysis (Figure 3G) identified several gut microbial taxa that were specifically elevated in each group, with Faecalibaculum being significantly increased in the WTE group.
3.2.3. Beta diversity of gut microbiota
Beta diversity analysis revealed that the composition of fecal microbiomes differed more significantly between the exercise and non‐exercise groups than within each group. Our study also found that the difference in fecal microbial community composition between the ADC and WTE groups was greater than that between the ADC and WTC groups (Figure 3B).
3.2.4. Predicted microbial functions altered in the ADC group via PICRUSt analysis
A total of 36 KEGG pathways at level 3 were generated from the 16S rRNA sequencing data comparing the ADC and ADE groups. The top 10 KEGG pathways generated from the sequencing data are shown in Figure 4A. The cecal microbiota of the ADC group exhibited different pathways associated with glycerophospholipid metabolism, primary bile acid biosynthesis, and others compared to the ADE group (Figure 4B,C). The significant differences of various metabolites are shown in Figure 4. Notably, diethanolamine, PE (14:0/22:2(13Z,16Z)), LysoPC (22:2(13Z,16Z)), and 7 alpha,26‐dihydroxy‐4‐cholesten‐3‐one were upregulated by exercise, while 5beta‐cyprinolsulfate was downregulated in the ADE group (Figure 4D).
FIGURE 4.
Predicted KEGG functional pathway and changed metabolites induced by aerobic exercise in AD groups. (A–C) Predicted KEGG functional pathway differences at level 3 inferred from 16S rRNA gene sequences using PICRUSt in ADC group (compared to ADE group). (A) Predicted KEGG functional pathway differences at level 3 were found between ADC and ADE group (n = 4 per group). (B) Metabolic pathway of glycerophospholipid metabolism in ADC group (compared to ADE group, n = 4 per group). (C) Metabolic pathway of primary bile acid biosynthesis pathway in ADC group (compared to ADE group, n = 4 per group). Red denotes an increase in the expression of the corresponding metabolite, whereas blue indicates a decrease. (D) Heat map of metabolites between ADC and ADE groups. Six samples were randomly selected from both the ADC group and ADE group for analysis (n = 6 per group). The scale reflects the level of metabolite expression, with red indicating elevated levels and green signifying reduced levels of metabolite expression; each row corresponds to a specific metabolite. Values of “1” and “2” denote high expression, while “−1” and “−2” indicate low expression.
4. DISCUSSION
Previous studies have confirmed that exercise can delay cognitive impairment in AD. 2 , 3 , 25 However, Abraham et al. found that exercise training alone was insufficient to improve cognition in AD mice, suggesting that combining exercise with probiotics might be necessary for significant cognitive improvements. 22 In this experiment, we employed two behavioral assessment methods. In the Morris water maze test, the ADE group exhibited a shorter escape latency compared to the ADC group, with no significant differences observed between the ADE and WTC groups. Similarly, in the Eight‐Arm Maze experiment, mice in the ADE group made significantly fewer working memory errors and showed better reference memory performance compared to those in the ADC group. These findings indicate that aerobic exercise may help preserve cognitive function in AD mice. The differing conclusions in Abraham et al.'s study could be attributed to variations in the training protocols for AD mice, suggesting that different exercise intensities and patterns may exert distinct effects on cognitive improvement in AD.
While mice in the WTC group, which did not undergo aerobic training, swam more slowly than those in the ADE group, their time spent in the target quadrant and the number of platform crossings did not significantly differ from the ADE group. These results indicate that the shorter escape latency observed in the ADE group likely reflects an improvement in cognitive function rather than simply increased swimming speed. We hypothesize that the heightened motor activity in the ADE group may stem from a stress response linked to the increased effort required to locate the target platform. In contrast, the ADC group not only exhibited longer escape latencies but also significantly reduced swimming speeds and distances traveled before reaching the platform, implying a more severe decline in cognitive function and possibly the presence of depressive symptoms, which could contribute to a lack of motivation. Previous research has indicated that AD is often associated with both cognitive dysfunction and emotional disorders, such as depression. 26 , 27 Our findings support the notion that exercise can positively influence both cognitive function and mood in AD mice.
The study incorporated various alpha diversity indices, specifically Chao1, ACE, Shannon, and Simpson indices. Chao1 and ACE indices served as indicators of gut microbiota richness, while the Shannon and Simpson indices assessed diversity; notably, the Shannon index was more reflective of microbiota evenness compared to the Simpson index. 28 Some studies indicate a correlation between reduced microbiota diversity and neurological disorders, including AD. 29 , 30 Conversely, additional research also supports this association. 31 , 32 In our investigation, no significant differences in Chao1, ACE, Shannon, or Simpson indices were found between the exercise and non‐exercise groups, suggesting similar gut microbiota richness across both conditions.
Exercise has been shown to enhance gut microbiota diversity, a finding supported by numerous studies. 18 , 33 , 34 , 35 , 36 This increased diversity is linked to improved health outcomes. 16 , 37 , 38 Our study did not observe an overall increase in intestinal flora abundance in the exercise group, which may be attributed to the small sample size. Nonetheless, we identified that 20 weeks of aerobic exercise positively influenced the relative abundance of various gut microbial species in both wild‐type (WT) and AD mice, particularly in the AD cohort. A significant increase in Proteobacteria was noted in the AD group, aligning with previous findings. 30 Some previous literature has reported an increased Bacteroidetes to Firmicutes ratio following exercise 20 , 39 ; another study showed the opposite trend. 8 Our result supports a decreased Bacteroidetes ratio after exercise in the AD groups. Different results may be related to different disease models and exercise strategies.
The PCoA analysis revealed that the composition of fecal microbiomes differed more significantly between the exercise and non‐exercise groups than within each group, aligning with findings from a previous study. 21 Additionally, our study found that the difference in fecal microbial community composition between the ADC and WTE groups was greater than that between the ADC and WTC groups. This suggests that exercise may amplify the differences in fecal microbiota between wild‐type mice and AD mice. Combined with the results of cognitive behavioral tests in our experiments, this microbial difference may have a protective effect on cognitive function.
At the phylum level, we observed a significant increase in Bacteroidetes abundance in the ADC group compared to the WTC group, corroborating previous findings. 30 Additionally, our study revealed that the abundance of Bacteroidetes in the ADE group was lower than in the ADC group. A recent publication indicates that Bacteroidetes are more prevalent in younger individuals than in older ones, 16 suggesting their potential role as probiotics with heightened reactivity in response to aging‐related lesions. However, since aerobic exercise mitigates the severity of these lesions, the increased reactivity of Bacteroidetes may also be diminished. Previous studies have linked changes in Bacteroidetes abundance to alterations in lipid and glucose metabolism, 16 which is supported by our findings of altered lipid metabolism pathways in the ADE group compared to the ADC group. Abraham et al. found in previous studies that levels of B. thetaiotaomicron were correlated with poorer performance in the Morris maze test (p < .05). This group of bacteria was significantly elevated in the microbiome of all APP/PS1TG mice compared to wild‐type mice; this finding was confirmed in our trial. Further verification with a larger sample size is needed. 22
At the genus level, we noted a significant increase in the abundance of the short‐chain fatty acid‐producing gut microbiota, Faecalibacterium, in the ADE group relative to the ADC group. Previous literature presents inconsistent findings regarding Faecalibacterium abundance in subjects with cognitive impairment 40 , 41 , 42 ; some studies report decreased levels in those with cognitive decline, while others indicate increased levels. Our findings align with the former, demonstrating a reduction in Faecalibacterium abundance in AD mice, which aerobic exercise subsequently elevated. Faecalibacterium species are known to produce short‐chain fatty acids, which are vital for maintaining the normal function of the large intestine and the integrity of colonic epithelial cells. 43 Short‐chain fatty acids, particularly butyrate, exert protective anti‐inflammatory effects and influence amyloid plaque levels in the brains of AD patients. 44 Previous research also identified Faecalibacterium as a butyrate producer, 45 , 46 with butyrate regulating gut microbiome composition by modulating intestinal pH and preventing the overgrowth of pathogenic bacteria. Additionally, butyrate can alleviate intestinal mucosal inflammation by inhibiting NF‐κB activation and upregulating PPARγ while suppressing interferon‐γ. Thus, increased Faecalibacterium abundance may attenuate the inflammatory response in AD mice, potentially slowing disease progression, though this requires validation in larger studies. Our results indicate that exercise enhances Faecalibacterium abundance, supported by LefSe analysis, which showed a significant elevation in Faecalibaculum within the WTE group. This suggests that pathophysiological changes in AD contribute to a decline in Faecalibaculum abundance, which exercise may help restore. Several mechanisms may underlie the alterations in gut microbiota composition due to aerobic exercise: it may improve intestinal permeability, 47 thereby affecting gut microbiota homeostasis, and it may modulate the gut‐brain axis through the release of myokines, metabolites, and neuroendocrine hormones, including alterations in nitric oxide pathways. 48 Further research is necessary to elucidate the precise mechanisms by which aerobic exercise modifies intestinal microbiome profiles.
In our study, we observed a significant increase in the abundance of Ileibacterium in the exercise group, particularly in the ADE cohort. Previous research has demonstrated that an increased abundance of Ileibacterium occurs in mice fed trans‐10, cis‐12 conjugated linoleic acid (t10, c12‐CLA), a dietary supplement that promotes weight loss by enhancing fat oxidation and energy expenditure. 49 Changes in Ileibacterium abundance may be linked to purine and bile acid metabolism, specifically cholic acid (CA) metabolism. 50 , 51 Notably, when blood lipid levels decrease, the abundance of Ileibacterium is upregulated, 52 , 53 , 54 suggesting that exercise could modify its levels. Our findings revealed that the abundance of Ileibacterium in the WTE group was significantly higher than that in the WTC group, with a more pronounced increase in the ADE group compared to the ADC group. Thus, we hypothesize that exercise enhances Ileibacterium abundance, potentially influencing fat and bile acid metabolism, which together may delay the progression of AD. Increased Ileibacterium abundance could serve as a protective factor within the pathophysiological mechanisms of AD.
Furthermore, in this study, we conducted a KEGG analysis of metabolic pathways associated with the intestinal flora, and differential metabolites across various pathways were detected, confirming that the ADE group differed from the ADC group in both the lipid metabolism and bile acid metabolism pathways. However, the precise mechanisms through which exercise induces changes in intestinal flora and metabolism require further investigation.
The top 10 KEGG pathways at level 3, derived from the 16S rRNA sequencing data, are illustrated in Figure 4A. Focusing on the metabolic pathways altered by exercise in the ADE group compared to the ADC group, we identified significant changes related to lipid metabolism (glycerophospholipid metabolism and glycosylphosphatidylinositol (GPI)—anchor biosynthesis), bile acid metabolism (primary bile acid biosynthesis), cell growth and apoptosis (Autophagy—other, mTOR signaling pathway and Autophagy animal), inflammation (Kaposi sarcoma—associated herpesvirus infection and PPAR signaling pathway), Carbohydrate metabolism (N‐glycan biosynthesis), and disease‐related (Alzheimer's disease). The changes in these related metabolic pathways indicate that exercise‐induced alterations in gut flora can affect the metabolism of AD mice through multiple pathways and aspects, especially for lipid metabolism. Three upregulated metabolites were observed in the glycerophospholipid metabolism pathway, supporting the notion that exercise enhances lipid metabolism in AD subjects. Lipid dyshomeostasis has become a prominent focus in AD research in recent decades. 55 The interplay between changes in gut microbiota and metabolic alterations due to exercise may contribute to the pathophysiological mechanisms that delay AD progression. Additionally, in the bile acid metabolism pathway, our study identified one upregulated and one downregulated metabolite resulting from exercise. Previous studies have reported significantly lower serum concentrations of primary bile acids, such as cholic acid, in AD patients compared to cognitively normal older adults, 56 corroborating our findings. We propose that Faecalibacterium and Ileibacterium may influence the pathophysiological course of AD through their effects on lipid and bile acid metabolism. At the same time, the detection results of metabolites in our study also confirmed the conclusion of our KEGG pathway analysis. Accordingly, our results suggest that aerobic exercise may modulate AD pathways by altering gut flora, which plays a crucial role in metabolic processes. Meanwhile, combined with our findings, we observed that the effects of exercise on the gut microbiome varied across different investigations in AD mice, 21 , 22 likely due to individual differences and environmental factors. Despite these inconsistencies, KEGG analysis in our study revealed that exercise‐induced changes in the gut microbiota of AD mice had the most significant impact on lipid metabolism. As a non‐pharmacological intervention, exercise appears to regulate AD progression through multiple pathways, 2 potentially exerting a broad modulatory effect on the gut microbiota and positively influencing metabolic processes. This finding suggests that future therapeutic strategies may benefit from targeting the overall gut microbiome rather than focusing on supplementing a single bacterial strain.
4.1. Strengths and limitations
This study is the first to comprehensively examine both gut microbiota and metabolites in APP/PS1 mice simultaneously under aerobic exercise conditions, allowing for meaningful inter‐group comparisons. Changes in swimming speed were meticulously measured and analyzed during cognitive assessments of APP/PS1 mice. Through KEGG pathway analysis, we identified target differential metabolites and validated alterations within the KEGG pathways via detection of these metabolites. Nonetheless, certain limitations must be acknowledged. Firstly, the small sample size imposes significant constraints, potentially explaining why Ileibacterium did not emerge as a characteristic bacterium in LDA analysis. Secondly, this cross‐sectional study does not capture exercise‐induced changes in gut microbiota throughout disease progression. Future research with larger sample sizes is essential to corroborate our findings. Third, there were no further studies on a single metabolite in this study (such as diethanolamine). The role of these metabolites in disease, etc., needs to be further clarified in future studies.
5. CONCLUSION
In conclusion, our study demonstrates that aerobic exercise can delay cognitive decline by altering the gut microbiome in AD mice. Ileibacterium and Faecalibaculum may play pivotal roles in the underlying pathophysiological mechanisms, which may cause changes in a variety of intestinal metabolites.
AUTHOR CONTRIBUTIONS
Conceptualization: Cuilan Wei and Rui Huang; Methodology: Cuilan Wei and Rui Huang; Software: Chuikun Li; Validation: Yeting Zhang; Formal Analysis: Xiaojing Wu; Investigation: Cuilan Wei; Resources: Cuilan Wei; Data Curation: Rui Huang; Writing—Original Draft Preparation: Cuilan Wei and Xiaojing Wu; Writing—Review and Editing: Rui Huang and Qiongjia Yuan; Visualization: Xiaojing Wu; Supervision: Rui Huang; Project Administration: Rui Huang and Qiongjia Yuan. All authors were involved in drafting and revising the manuscript.
ACKNOWLEDGEMENTS
Thanks for revision from the editors. Authors acknowledge the assistance of Professor Xue Li in the preparation of the manuscript.
FUNDING INFORMATION
This research received no external funding.
DISCLOSURES
The authors declare no conflicts of interest.
ETHICS STATEMENT
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of the Animal Studies Committees of Chengdu Sport University(protocol code—(2021)21 and date of approval—May 12, 2021).
Wei C, Wu X, Li C, Zhang Y, Yuan Q, Huang R. Aerobic exercise regulates gut microbiota profiles and metabolite in the early stage of Alzheimer's disease. The FASEB Journal. 2025;39:e70327. doi: 10.1096/fj.202402572R
Cuilan Wei, Xiaojing Wu, Qiongjia Yuan, and Rui Huang contributed equally to this work.
Contributor Information
Qiongjia Yuan, Email: yqj1225@163.com.
Rui Huang, Email: huangruisn@med.uestc.edu.cn.
DATA AVAILABILITY STATEMENT
Data are contained within the article.
REFERENCES
- 1. Scheltens P, de Strooper B, Kivipelto M, et al. Alzheimer's disease. Lancet. 2021;397(10284):1577‐1590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. De la Rosa A, Olaso‐Gonzalez G, Arc‐Chagnaud C, et al. Physical exercise in the prevention and treatment of Alzheimer's disease. J Sport Health Sci. 2020;9(5):394‐404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cui MY, Lin Y, Sheng JY, Zhang X, Cui RJ. Exercise intervention associated with cognitive improvement in Alzheimer's disease. Neural Plast. 2018;2018:9234105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Yang L, Wu C, Li Y, et al. Long‐term exercise pre‐training attenuates Alzheimer's disease‐related pathology in a transgenic rat model of Alzheimer's disease. Geroscience. 2022;44(3):1457‐1477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Gaitán JM, Moon HY, Stremlau M, et al. Effects of aerobic exercise training on systemic biomarkers and cognition in late middle‐aged adults at risk for Alzheimer's disease. Front Endocrinol (Lausanne). 2021;12:660181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Grabrucker S, Marizzoni M, Silajdžić E, et al. Microbiota from Alzheimer's patients induce deficits in cognition and hippocampal neurogenesis. Brain. 2023;146(12):4916‐4934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Jemimah S, Chabib CMM, Hadjileontiadis L, AlShehhi A. Gut microbiome dysbiosis in Alzheimer's disease and mild cognitive impairment: a systematic review and meta‐analysis. PLoS One. 2023;18(5):e0285346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Haran JP, Bhattarai SK, Foley SE, et al. Alzheimer's disease microbiome is associated with dysregulation of the anti‐inflammatory P‐glycoprotein pathway. MBio. 2019;10(3):e00632‐19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Cattaneo A, Cattane N, Galluzzi S, et al. Association of brain amyloidosis with pro‐inflammatory gut bacterial taxa and peripheral inflammation markers in cognitively impaired elderly. Neurobiol Aging. 2017;49:60‐68. [DOI] [PubMed] [Google Scholar]
- 10. Friedland RP. Mechanisms of molecular mimicry involving the microbiota in neurodegeneration. J Alzheimers Dis. 2015;45(2):349‐362. [DOI] [PubMed] [Google Scholar]
- 11. Soscia SJ, Kirby JE, Washicosky KJ, et al. The Alzheimer's disease‐associated amyloid beta‐protein is an antimicrobial peptide. PLoS One. 2010;5(3):e9505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Harach T, Marungruang N, Duthilleul N, et al. Reduction of Abeta amyloid pathology in APPPS1 transgenic mice in the absence of gut microbiota. Sci Rep. 2017;7:41802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Zhang L, Wang Y, Xiayu X, et al. Altered gut microbiota in a mouse model of Alzheimer's disease. J Alzheimers Dis. 2017;60(4):1241‐1257. [DOI] [PubMed] [Google Scholar]
- 14. He B, Sheng C, Yu X, Zhang L, Chen F, Han Y. Alterations of gut microbiota are associated with brain structural changes in the spectrum of Alzheimer's disease: the SILCODE study in Hainan cohort. Front Aging Neurosci. 2023;15:1216509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Wang J, Fasina OB, Manzoor M, et al. A new gentiopicroside derivative improves cognitive deficits of AD mice via activation of Wnt signaling pathway and regulation of gut microbiota homeostasis. Phytomedicine. 2023;113:154730. [DOI] [PubMed] [Google Scholar]
- 16. Wang T, Shi Z, Ren H, et al. Divergent age‐associated and metabolism‐associated gut microbiome signatures modulate cardiovascular disease risk. Nat Med. 2024;30:1722‐1731. [DOI] [PubMed] [Google Scholar]
- 17. Clark A, Mach N. Exercise‐induced stress behavior, gut‐microbiota‐brain axis and diet: a systematic review for athletes. J Int Soc Sports Nutr. 2016;13:43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Dohnalová L, Lundgren P, Carty JRE, et al. A microbiome‐dependent gut‐brain pathway regulates motivation for exercise. Nature. 2022;612(7941):739‐747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Gubert C, Kong G, Renoir T, Hannan AJ. Exercise, diet and stress as modulators of gut microbiota: implications for neurodegenerative diseases. Neurobiol Dis. 2020;134:104621. [DOI] [PubMed] [Google Scholar]
- 20. Motiani KK, Collado MC, Eskelinen JJ, et al. Exercise training modulates gut microbiota profile and improves endotoxemia. Med Sci Sports Exerc. 2020;52(1):94‐104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Yuan S, Yang J, Jian Y, et al. Treadmill exercise modulates intestinal microbes and suppresses LPS displacement to alleviate neuroinflammation in the brains of APP/PS1 mice. Nutrients. 2022;14(19):4134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Abraham D, Feher J, Scuderi GL, et al. Exercise and probiotics attenuate the development of Alzheimer's disease in transgenic mice: role of microbiome. Exp Gerontol. 2019;115:122‐131. [DOI] [PubMed] [Google Scholar]
- 23. Li X, Zhang Y, Ding X, Jin Y, Wei C, Xu J. Mass spectrometry chromatography‐based metabolomics: the effect of long‐term aerobic exercise on learning ability and the metabolism of intestinal contents in mice with Alzheimer's disease. Metabolites. 2023;13(11):1150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Morris R. Developments of a water‐maze procedure for studying spatial learning in the rat. J Neurosci Methods. 1984;11(1):47‐60. [DOI] [PubMed] [Google Scholar]
- 25. Ai ZL, Zhang X, Ge W, et al. Salvia miltiorrhiza extract may exert an anti‐obesity effect in rats with high‐fat diet‐induced obesity by modulating gut microbiome and lipid metabolism. World J Gastroenterol. 2022;28(43):6131‐6156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lyketsos CG, Carrillo MC, Ryan JM, et al. Neuropsychiatric symptoms in Alzheimer's disease. Alzheimers Dement. 2011;7(5):532‐539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Talebi V, Alamdari KA, Patel DI. Simple and complex wheel running effect on depression, memory, neuroinflammation, and neurogenesis in Alzheimer's rat model. Med Sci Sports Exerc. 2024;56(6):1159‐1167. [DOI] [PubMed] [Google Scholar]
- 28. Wu FZ, Wang XZ, Pan K. [Effects of wheat and soybean stubbles on soil microbial ecological characteristics in cucumber field]. Ying Yong Sheng Tai Xue Bao. 2008;19(4):794‐798. [PubMed] [Google Scholar]
- 29. Cani PD, Delzenne NM. Gut microflora as a target for energy and metabolic homeostasis. Curr Opin Clin Nutr Metab Care. 2007;10(6):729‐734. [DOI] [PubMed] [Google Scholar]
- 30. Hung CC, Chang CC, Huang CW, Nouchi R, Cheng CH. Gut microbiota in patients with Alzheimer's disease spectrum: a systematic review and meta‐analysis. Aging (Albany NY). 2022;14(1):477‐496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Manichanh C, Rigottier‐Gois L, Bonnaud E, et al. Reduced diversity of faecal microbiota in Crohn's disease revealed by a metagenomic approach. Gut. 2006;55(2):205‐211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Monreal MT, Pereira PC, de Magalhães Lopes CA. Intestinal microbiota of patients with bacterial infection of the respiratory tract treated with amoxicillin. Braz J Infect Dis. 2005;9(4):292‐300. [DOI] [PubMed] [Google Scholar]
- 33. Donati Zeppa S, Agostini D, Gervasi M, et al. Mutual interactions among exercise, sport supplements and microbiota. Nutrients. 2019;12(1):17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Kessoku T, Imajo K, Kobayashi T, et al. [Gut microbiota and nonalcoholic steatohepatitis]. Nihon Yakurigaku Zasshi. 2018;152(4):187‐193. [DOI] [PubMed] [Google Scholar]
- 35. Monda V, Villano I, Messina A, et al. Exercise modifies the gut microbiota with positive health effects. Oxidative Med Cell Longev. 2017;2017:3831972. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Plissonneau C, Capel F, Chassaing B, et al. High‐intensity interval training and α‐linolenic acid supplementation improve DHA conversion and increase the abundance of gut mucosa‐associated Oscillospira bacteria. Nutrients. 2021;13(3):788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Lallès JP, Bosi P, Smidt H, Stokes CR. Nutritional management of gut health in pigs around weaning. Proc Nutr Soc. 2007;66(2):260‐268. [DOI] [PubMed] [Google Scholar]
- 38. Sanz Y, Nadal I, Sánchez E. Probiotics as drugs against human gastrointestinal infections. Recent Pat Antiinfect Drug Discov. 2007;2(2):148‐156. [DOI] [PubMed] [Google Scholar]
- 39. Aragón‐Vela J, Solis‐Urra P, Ruiz‐Ojeda FJ, Álvarez‐Mercado AI, Olivares‐Arancibia J, Plaza‐Diaz J. Impact of exercise on gut microbiota in obesity. Nutrients. 2021;13(11):3999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Sheng C, Lin L, Lin H, Wang X, Han Y, Liu SL. Altered gut microbiota in adults with subjective cognitive decline: the SILCODE study. J Alzheimers Dis. 2021;82(2):513‐526. [DOI] [PubMed] [Google Scholar]
- 41. Ueda A, Shinkai S, Shiroma H, et al. Identification of Faecalibacterium prausnitzii strains for gut microbiome‐based intervention in Alzheimer's‐type dementia. Cell Rep Med. 2021;2(9):100398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Xi J, Ding D, Zhu H, et al. Disturbed microbial ecology in Alzheimer's disease: evidence from the gut microbiota and fecal metabolome. BMC Microbiol. 2021;21(1):226. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Zagato E, Pozzi C, Bertocchi A, et al. Endogenous murine microbiota member Faecalibaculum rodentium and its human homologue protect from intestinal tumour growth. Nat Microbiol. 2020;5(3):511‐524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Qian XH, Xie RY, Liu XL, Chen SD, Tang HD. Mechanisms of short‐chain fatty acids derived from gut microbiota in Alzheimer's disease. Aging Dis. 2022;13(4):1252‐1266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Heinken A, Khan MT, Paglia G, Rodionov DA, Harmsen HJM, Thiele I. Functional metabolic map of Faecalibacterium prausnitzii, a beneficial human gut microbe. J Bacteriol. 2014;196(18):3289‐3302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Rivière A, Selak M, Lantin D, Leroy F, De Vuyst L. Bifidobacteria and butyrate‐producing colon bacteria: importance and strategies for their stimulation in the human gut. Front Microbiol. 2016;7:979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Keirns BH, Koemel NA, Sciarrillo CM, Anderson KL, Emerson SR. Exercise and intestinal permeability: another form of exercise‐induced hormesis? Am J Physiol Gastrointest Liver Physiol. 2020;319(4):G512‐G518. [DOI] [PubMed] [Google Scholar]
- 48. Tse JKY. Gut microbiota, nitric oxide, and microglia as prerequisites for neurodegenerative disorders. ACS Chem Neurosci. 2017;8(7):1438‐1447. [DOI] [PubMed] [Google Scholar]
- 49. Mailhe M, Ricaboni D, Vitton V, Lagier JC, Fournier PE, Raoult D. ‘Ileibacterium massiliense’ gen. nov., sp. nov., a new bacterial species isolated from human ileum of a patient with Crohn disease. New Microbes New Infect. 2017;17:25‐26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Fu T, Huan T, Rahman G, et al. Paired microbiome and metabolome analyses associate bile acid changes with colorectal cancer progression. Cell Rep. 2023;42(8):112997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Ji S, Han S, Yu L, et al. Jia Wei Xiao Yao San ameliorates chronic stress‐induced depression‐like behaviors in mice by regulating the gut microbiome and brain metabolome in relation to purine metabolism. Phytomedicine. 2022;98:153940. [DOI] [PubMed] [Google Scholar]
- 52. Gu Y, Zhang Y, Li M, et al. Ferulic acid ameliorates atherosclerotic injury by modulating gut microbiota and lipid metabolism. Front Pharmacol. 2021;12:621339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Hu W, Gao W, Liu Z, et al. Specific strains of Faecalibacterium prausnitzii ameliorate nonalcoholic fatty liver disease in mice in association with gut microbiota regulation. Nutrients. 2022;14(14):2945. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Xie B, Zu X, Wang Z, Xu X, Liu G, Liu R. Ginsenoside Rc ameliorated atherosclerosis via regulating gut microbiota and fecal metabolites. Front Pharmacol. 2022;13:990476. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Yin F. Lipid metabolism and Alzheimer's disease: clinical evidence, mechanistic link and therapeutic promise. FEBS J. 2023;290(6):1420‐1453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. MahmoudianDehkordi S, Arnold M, Nho K, et al. Altered bile acid profile associates with cognitive impairment in Alzheimer's disease—an emerging role for gut microbiome. Alzheimers Dement. 2019;15(1):76‐92. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data are contained within the article.