What this is
- This research investigates how exercise influences autism-like behaviors through gut microbiota modulation.
- A rat model of autism was used, with groups undergoing exercise or ().
- Key findings include improved social interactions and altered gut microbiota composition following exercise.
Essence
- Exercise significantly alleviates autism-like behaviors in a rat model by modulating gut microbiota, (), and neurotransmitter levels. from exercising rats replicates these benefits.
Key takeaways
- Exercise improved social interactions in rats, indicated by increased time spent with novel conspecifics during behavioral tests.
- Gut microbiota composition shifted with exercise, increasing beneficial genera like Limosilactobacillus and Lactobacillus while decreasing Allobaculum.
- from exercised rats replicated behavioral improvements in rats, validating the link between exercise-induced microbiota changes and enhanced social behavior.
Caveats
- The sample size was relatively small, which may limit the generalizability of the findings.
- The study focused on short-term interventions, leaving long-term effects of exercise and microbiota modulation unclear.
- 16S rRNA sequencing primarily identifies microbial genera, potentially missing important species-level details.
Definitions
- Autism Spectrum Disorder (ASD): A neurodevelopmental disorder characterized by impaired social communication and repetitive behaviors.
- Fecal Microbiota Transplantation (FMT): A procedure that involves transferring fecal bacteria from a healthy donor to a recipient to restore gut microbiota balance.
- Short-chain fatty acids (SCFAs): Fatty acids with fewer than six carbon atoms, produced by gut bacteria during fermentation, important for gut health.
Simplified
Background
Autism spectrum disorder (ASD) is a serious neurodevelopmental disorder characterized by impaired social communication as well as the occurrence of narrow interests and repetitive behaviors [1]. According to data from the Centers for Disease Control and Prevention Morbidity and Mortality Weekly Report, the overall ASD incidence is one in thirty-six children aged 8 years in the USA [2]. The pathology of ASD remains unclear, and there is no specific treatment available. Therefore, finding effective ways to improve and alleviate the core symptoms of ASD has become a focal point in autism rehabilitation research. In recent years, interest in the relationship between the gut microbiota and human health has increased, providing new perspectives for understanding the pathogenesis of various diseases [3].
Accumulating evidence indicates that the gut microbiota plays a significant role in the onset and progression of ASD in both humans and animals [4–8]. Studies have also shown that modulating the gut microbiota through interventions such as probiotic supplementation and fecal microbiota transplantation (FMT) can alleviate symptoms in individuals with ASD [9, 10]. The bidirectional communication between the gut microbiota and the brain involves multiple pathways, primarily the immune system, neuroendocrine pathways, metabolism, and the vagal nerve pathway [11–13]. As an endocrine-like organ, the microbiota not only produces many metabolic products, such as tryptophan and short-chain fatty acids (SCFAs), which are involved in energy balance and metabolism but also synthesizes and releases several neurotransmitters, similar to those in the brain [14]. Moreover, neurotransmitters are the most important intermediaries in the communication of neurons, and dysfunction of neurotransmitters may explain the mechanism of excitatory or inhibitory imbalance in ASD patients. Therefore, regulating the microbiota and its metabolism, as well as microbiota transplantation, has consistently been regarded as a potential therapeutic target and intervention strategy for ASD [11, 15].
Furthermore, research has shown that the microbiota can be influenced by exercise, as can antibiotics, probiotics, nutrients, and microbiota transplantation. Additionally, research indicates that exercise can improve symptoms related to various diseases, including colitis, diabetes, and Alzheimer’s disease, by altering the gut microbiota [16–19]. In addition, extensive studies have demonstrated the positive effects of exercise on the core symptoms of ASD [20–27], as well as improvements in quality of life [24], but the underlying mechanism of the ameliorative effects of exercise on ASD remains unclear. Moreover, the regulatory effects of exercise on central neurotransmitters have also been confirmed by numerous studies [28–31]. However, there is currently a lack of evidence regarding whether the ameliorating effects of exercise on ASD are related to alterations in the microbiota and their connection to neurotransmitter levels.
In this study, we successfully established a rat model of ASD through prenatal exposure to valproic acid (VPA). The rats were then randomly divided into four groups: the ASD group, the six-week voluntary wheel-running exercise intervention (E-ASD) group, the FMT group, in which the feces of the rats in the E-ASD group were transplanted, and the saline sham transplant (sFMT) group. This study reveals how exercise modulates the gut microbiota to influence metabolic changes and ameliorate autism-like behaviors by comparing behavioral performance, gut microbiota composition, SCFAs, and neurotransmitter changes in rats receiving exercise interventions and FMT. These findings highlight the potential of exercise interventions as a therapeutic strategy for modulating ASD and offer new perspectives on the use of microbiota transplantation as a treatment approach.
Methods
Animals and housing conditions
Pregnant Sprague–Dawley rats were obtained from the Laboratory Animal Center of Southern Medical University (P. R. China). All experimental procedures were approved by the Ethics Committee of Guangzhou Sport University and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23). On gestational day 12.5, the pregnant rats received an intraperitoneal injection of either 400 mg/kg valproic acid sodium salt (VPA; P4543, Sigma, USA) to induce autism-like phenotypes in the offspring, or an equivalent volume of normal saline as a control, as previously described [32]. The pregnant rats were housed individually and provided with ad libitum access to food and water. They were allowed to rear their litters without intervention, except for those designated for growth and behavioral development assessments. Offspring were weaned on postnatal day 23. All groups were maintained under identical housing conditions to minimize any potential variability in social behaviors caused by differences in housing environments. The light/dark cycle was carefully controlled and set to 12 h of light and 12 h of darkness each day to simulate a natural diurnal rhythm, and the ambient temperature in the housing area was consistently maintained at 23 °C ± 2 °C to ensure a stable and comfortable environment for the animals. The gut microbiota composition, SCFAs, and neurotransmitter levels of all rats were assessed both before and after the intervention, along with behavioral testing. The methods used for behavioral testing (including the open field test and three-chamber social interaction test), 16S rRNA gene sequencing, SCFA analysis, and neurotransmitter analysis were consistent with those described in previous publications [33]. For more detailed descriptions of the experimental protocols and analytical methods, please refer to the Supplementary Materials.
Exercise intervention protocol
Rats in the E_ASD group were provided with a voluntary running wheel (35 cm diameter) to encourage physical activity. The intervention began on postnatal day 28 (P28) and lasted 42 days. During this period, the rats were allowed to use the running wheel continuously in their home cages and their running distances were automatically recorded by a computerized system. The system tracked the total running distance of each rat on a daily basis to accurately monitor their physical activity levels during the intervention. Running distance was calculated by multiplying the total number of wheel revolutions by the circumference of the 35-cm wheel. Daily running distance was then derived over each 24-hour cycle, providing a high-precision, high-resolution measure of voluntary activity while minimizing disturbance to the animals. The resulting running-distance data were further used as a quantitative indicator of exercise intensity and were incorporated as a covariate in the behavioral statistical analyses.
FMT procedure
To investigate whether the increase in behavioral performance is due to changes in the gut microbiome, fecal samples from the rats in the exercise group were transplanted into the rats in the non-exercise group. This was accomplished via gavage starting in the third week of the exercise intervention. The method of preparation followed that of Sampson et al. [34]. Fresh fecal pellets from exercised group rats were collected at 9:00 a.m. and diluted in sterile phosphate-buffered saline (PBS) at a 1:5 ratio. The mixture was thoroughly stirred to form a uniform suspension. Next, the suspension was passed through a 0.5 mm mesh filter to remove food debris, yielding a secondary suspension. This was then centrifuged at 6000 rpm for 15 min at 4 °C. After discarding the supernatant, the resulting pellet was resuspended in the same volume of sterile PBS used initially, generating the final fecal bacterial filtrate. This suspension was administered to recipient rats by oral gavage at a dosage of 1 mL per 100 g of body weight, once daily for 4 weeks.
Bioinformatics analysis
The raw 16S rRNA gene sequencing data were initially processed via the standard QIIME2 pipeline, which included quality filtering, adapter trimming, and chimera removal to ensure the overall accuracy and reliability of the dataset. The DADA2 algorithm was subsequently applied for sequence dereplication and quality control, resulting in the generation of amplicon sequence variants with 100% sequence similarity. Low-abundance taxa present in less than 10% of the samples were filtered out to ensure data reliability. The microbial composition was visualized by selecting the top ten most abundant genera. Alpha diversity indices, including Shannon, Simpson, Chao1, and ACE indices, were calculated at the genus level, and intergroup differences in within-sample diversity were assessed using the Wilcoxon rank-sum test. Beta diversity was analyzed using principal coordinate analysis (PCoA) based on the Bray-Curtis distance matrix, and intergroup differences were calculated using permutational multivariate analysis of variance (PERMANOVA) with 999 permutations. Differentially abundant taxa between groups were identified via linear discriminant analysis effect size (LEfSe), with statistical significance determined via Kruskal‒Wallis and Wilcoxon tests (p < 0.05) and a linear discriminant analysis (LDA) score threshold set at > 4. To investigate the potential interactions among gut bacterial taxa, SCFAs, and neurotransmitters, a correlation-based network analysis was performed. Pairwise Spearman correlation coefficients were calculated based on the relative abundance or concentration levels of significantly different bacterial genera, SCFAs, and neurotransmitters. Only statistically significant correlations with FDR-corrected p values < 0.05 and absolute correlation coefficients |r| ≥ 0.3 were retained for network construction. The resulting correlation network was visualized using Cytoscape software (version 3.9.1). In the network, each node represents a bacterial genus, SCFA, or neurotransmitter, while edges indicate significant correlations between these variables. The direction and strength of the correlations are denoted by the color and thickness of the edges: red edges represent positive correlations, whereas blue edges indicate negative correlations.
Statistical analysis
All statistical analyses were performed using SPSS software (version 19.0, IBM Corp., Armonk, NY, USA), GraphPad Prism (version 8.0, GraphPad Software Inc., San Diego, CA, USA), and R software (version 4.3.2). Prior to statistical testing, data normality was assessed using the Shapiro-Wilk test, and homogeneity of variance was evaluated using Levene’s test. For normally distributed data with equal variances, independent t-tests were used for comparisons between two groups. For data that did not meet the assumptions of parametric tests, the Mann-Whitney U test was applied for comparison. For comparisons among the three groups, one-way ANOVA followed by Tukey’s post hoc test was performed when the data met the assumptions of normality and homogeneity of variance; when these assumptions were not satisfied, the Kruskal–Wallis test followed by Dunn’s multiple comparison correction was applied. Spearman’s rank correlation analysis was conducted to identify significant associations among differentially abundant bacterial genera, SCFAs, and neurotransmitters. To reduce the risk of type I errors due to multiple comparisons, the Benjamini–Hochberg procedure was applied to control the false discovery rate (FDR), with statistical significance set at p < 0.05. All statistical results were visualized using a range of analytical tools: box plots and violin plots were generated using GraphPad Prism (v9.0); correlation heatmaps and network analyses were carried out with R packages including ggplot2, pheatmap, and igraph. Microbial co-occurrence networks and metabolite interaction networks were further refined and visualized using Cytoscape software (v3.9.1).
Results
Voluntary wheel running alleviates autism-like behaviors
In our previous study, we measured behavioral performance, gut microbiota composition, SCFAs, and neurotransmitter levels prior to the intervention, thereby confirming the successful establishment of the ASD-like rat model [33]. To assess whether exercise could ameliorate autism-like behaviors in VPA rats, voluntary wheel-running was initiated at postweaning (P28) and continued for six weeks (Fig. S1a). Throughout the intervention, running distance increased steadily across individuals, indicating sustained and regular engagement in voluntary exercise (Fig. S2). In the three-chamber social test, during the sociability phase, both the control and E_ASD groups spent significantly more time in the stranger-1 (S1) chamber than in their respective empty chambers (Block), whereas the ASD group showed no clear preference between S1 and Block (Fig. 1a). This pattern indicates an impairment in basic social interest in the ASD group, while early voluntary wheel-running partially restored this preference for conspecific interaction. During the social novelty preference phase, the control and E_ASD groups spent more time in the stranger-2 (S2) chamber than in their corresponding S1 chambers, whereas the ASD group displayed no difference between S1 and S2 (Fig. 1b). Consistently, the social novelty index in the ASD group was significantly lower than that in both the control and E_ASD groups, whereas no significant difference was observed between the control and E_ASD groups (Fig. 1c), suggesting that exercise intervention partially ameliorated the VPA-induced deficit in social novelty. Although no significant differences in the overall sociability index were observed among the three groups (Fig. S3a), the E_ASD group showed an improving trend compared with the ASD group. In the open-field test (OFT), the ASD group exhibited more pronounced anxiety-like behavior, reflected by a greater average distance from the center, shorter time spent in the central area, and fewer center entries relative to the control group (Fig. 1d–f), whereas the E_ASD group showed partial improvement but did not completely return to control levels. Meanwhile, no significant differences were found among the groups in total distance traveled, average locomotor speed, or central-zone movement speed (Fig. S3b–d), indicating that the intervention did not fully restore the initial behavioral state and that longer-term exercise or multimodal therapeutic approaches may be required for more complete recovery.
Behavioral assessments across three groups: control, ASD, and E_ASD.. Time spent in the stranger-1 (S1) chamber versus the empty chamber (Block) during the sociability phase.. Time spent in the stranger-1 (S1) versus stranger-2 (S2) chamber during the social novelty preference phase.. Social novelty index calculated as (S2 − S1) / (S2 + S1). d. Distance from the center during the open field test (OFT), reflecting anxiety-like behavior. e. Duration of time spent in the center of the OFT. f. Number of entries into the center of the OFT. *< 0.05, **< 0.01 indicate significant differences between groups.= 8 (male = 4, female = 4) a b c p p N
Early voluntary wheel running alters the gut microbiota, SCFAs, and neurotransmitters in ASD rats
To investigate whether exercise affects ASD behavior through changes in the gut microbiota, SCFAs, and neurotransmitters, we measured these three indicators. In the control group (Fig. 2a), the gut microbial community was characterized by a dominant taxa profile enriched with typical beneficial genera such as Ligilactobacillus and Lactobacillus, reflecting a stable microbial structure associated with healthy gut physiology. In contrast, the ASD group (Fig. 2b) exhibited a distinct microbial signature, with Bacteroides and Prevotella constituting the major dominant taxa, representing a characteristic ASD-associated microbial configuration. The microbiota composition of the E_ASD group (Fig. 2c) showed a shift away from the ASD-like pattern, indicating partial improvement in microbial imbalance. The α diversity results revealed no significant differences between the Shannon and Simpson indices. However, consistent patterns were observed for the ACE and Chao1 indices, indicating that the species diversity in the ASD group was significantly greater than that in both the E_ASD and control groups (Fig. 3a). Notably, no significant changes were observed between the E_ASD and control groups. For β-diversity analysis, PCoA based on Bray-Curtis distances revealed distinct clustering patterns among the three groups, further supported by PERMANOVA, which confirmed significant differences in the gut microbiota composition (Fig. 3b). Lefse analysis provided insights into specific taxonomic enrichments within each group (Fig. 3c, d). Notably, the E_ASD group was enriched with Lactobacillus and Limosilactobacillus at the genus level. Interestingly, the ASD group presented an overabundance of Allobaculum.
Furthermore, we found that butyrate levels were significantly lower in the ASD group compared with both the E_ASD and control groups, while no significant difference was observed between the E_ASD and control groups (Fig. 4b). The levels of acetic acid and hexanoic acid were significantly lower in the ASD group than in the control group, while the levels in the E_ASD group tended to increase, although the difference was not statistically significant (Fig. 4a, c).
Importantly, neurotransmitter analysis of the prefrontal cortex (PFC) of the E_ASD rats revealed significant reductions in threonine (Fig. 4d), kynurenine (Fig. 4e), 3-hydroxytyramine (Fig. 4f), tryptophan (Fig. 4g), and gamma-aminobutyric acid (Fig. 4h) levels compared with those in the ASD group, with no significant differences compared with those in the control group. Epinephrine levels were significantly greater in the E_ASD group than in the ASD group (Fig. 4i), approaching those of the control group but still showing a notable difference.
Comparative composition of dominant gut microbial genera across groups.. Relative abundance of gut microbiota at the genus level in the Control group.. Relative abundance of gut microbiota at the genus level in the ASD group.. Relative abundance of gut microbiota at the genus level in the E_ASD group.. Relative abundance of gut microbiota at the genus level in the FMT group.. Relative abundance of gut microbiota at the genus level in the sFMT group. N =8 (male = 4, female = 4) a b c d e
Comparative analysis of the gut microbiome across the control, ASD, and E_ASD groups.. Alpha diversity indices, including the Shannon, Simpson, ACE, and Chao1 indices, for the control, ASD, and E_ASD groups;. PCoA plot based on Bray-Curtis distances, showing the separation of the gut microbiome profiles among the control, ASD, and E_ASD groups, with R= 0.21 andvalue = 0.002 indicating the significance of the group separation;. LEfSe of the gut microbiota, displaying taxa with significant differences in relative abundance between groups, with LDA scores highlighting the importance of each taxon in distinguishing between the groups;. Cladogram illustrating the phylogenetic relationships between differentially abundant taxa identified by LEfSe analysis, with branch tips colored according to group abundance. N = 8 (male= 4, female = 4) a b c d 2 p
Metabolic profile comparisons among the control, ASD, and E_ASD groups.. Concentrations of acetic acid;. Concentrations of butyric acid;. Concentrations of hexanoic acid;. Concentrations of threonine;. Concentrations of kynurenine;. Concentrations of 3-hydroxykynurenine;. Concentrations of tryptophan;. Concentrations of gamma-aminobutyric acid;. Concentrations of epinephrine. *<0.05, **<0.01 indicate significant differences between groups. N= 8 (male = 4, female = 4), N= 5 a b c d e f g h i p p a-c d-i
FMT from E_ASD group rats alleviates autism-like behavior
Exercise-induced gut microbiota involvement in behavioral improvement was assessed by performing fecal microbiota transplantation from the E_ASD group to ASD rats, with sFMT used as the control (Fig. S1b). After four weeks of intervention, the FMT group exhibited clear behavioral improvements in the three-chamber social test (Fig. 5a–c). During the sociability phase (S1 vs. Block), rats in the FMT group spent significantly more time in the stranger-1 (S1) chamber than in the empty chamber (Fig. 5a), demonstrating a clear restoration of social interest. In contrast, the sFMT group did not show a significant difference between the two chambers. Consistently, the sociability index of the FMT group was significantly higher than that of the sFMT group (Fig. 5b), further supporting the beneficial effect of FMT on sociability. During the social novelty preference phase (S1 vs. S2), the FMT group spent significantly more time in the stranger-2 (S2) chamber than in the S1 chamber, demonstrating a distinct preference for social novelty, whereas the sFMT group showed no such preference (Fig. 5c).
Behavioral assessments across three groups: the E_ASD, FMT and sFMT groups.. Time spent in the stranger-1 (S1) chamber versus the empty chamber (Block) during the sociability phase.. Sociability index calculated as (S1 − Block) / (S1 + Block).. Time spent in the stranger-1 (S1) and stranger-2 (S2) chambers during the social novelty preference phase. *< 0.05, **< 0.01 indicate significant differences between groups.=8 (male = 4, female = 4) a b c p p N
FMT alters the gut microbiota, SCFAs, and neurotransmitters in ASD rats
After four weeks of FMT, significant changes were observed in the gut microbiota, SCFA levels, and neurotransmitter profiles of the ASD rats. Notably, the sFMT group presented minimal differences compared with the previous ASD group (Fig. 2e), whereas the FMT group presented increases in the Turicibacter and Clostridium genera (Fig. 2d). Moreover, significant alterations in the α diversity of the gut microbiota were detected across all groups (Fig. 6a). Notably, compared with both the E_ASD and FMT groups, the sFMT group presented significant differences in gut microbial diversity, as observed in both the Chao1 and ACE indices. However, no significant differences in α diversity were found between the E_ASD and FMT groups. Additionally, β diversity analysis revealed differences in the gut microbiota composition among the three groups (Fig. 6b). Importantly, Bray–Curtis PCoA revealed a pronounced separation between the FMT and ASD groups (Fig. S4), indicating that exercise-derived microbiota effectively reshaped the recipient microbial community. In contrast, no separation was observed between the sFMT and ASD groups (Fig. S5), confirming that saline gavage alone did not alter the native microbial structure. Lefse analysis revealed distinct microbial taxa between these groups (Fig. 6c, d). The E_ASD group exhibited continued enrichment of Lactobacillus and Limosilactobacillus, whereas the sFMT group presented increased abundances of the Prevotella and Onthenecus genera, and the FMT group was enriched in Treponema.
In terms of SCFAs, we observed significant reductions in the levels of isovaleric acid (Fig. 7a), butyric acid (Fig. 7b), hexanoic acid (Fig. 7c), valeric acid (Fig. 7d), and isobutyric acid (Fig. 7e) in the sFMT group compared with those in the E_ASD group. However, no significant differences were found between the FMT and E_ASD groups. Although an increasing trend in SCFA levels was observed in the FMT group compared with the sFMT group, this difference was not statistically significant.
Interestingly, neurotransmitter analysis revealed a significant increase in serine (Fig. 7f) in the FMT group compared with the sFMT group. In addition, glycine, γ-aminobutyric acid, ethanolamine, and kynurenine levels in the FMT group closely resembled those in the E_ASD group, with no significant differences between the two. Notably, both groups exhibited significantly lower levels than the sFMT group, indicating that FMT restored these neurotransmitters abnormalities toward the profile observed after exercise intervention (Fig. 7g-j).
Comparative analysis of the gut microbiome across the E_ASD, FMT, and sFMT groups.. Alpha diversity indices, including the Shannon, Simpson, ACE, and Chao1 indices, for the E_ASD, FMT, and sFMT groups;. PCoA plot based on Bray-Curtis distances, showing the separation of the gut microbiome profiles among the E_ASD, FMT, and sFMT groups, with R= 0.26 andvalue = 0.001 indicating the significance of the group separation;. LEfSe of the gut microbiota, displaying taxa with significant differences in relative abundance between groups, with LDA scores highlighting the importance of each taxon in distinguishing between the groups;. Cladogram illustrating the phylogenetic relationships between differentially abundant taxa identified by LEfSe analysis, with branch tips colored according to group abundance. N = 8 (male = 4, female = 4) a b c d 2 p
Metabolic profile comparisons among the E_ASD, FMT, and sFMT groups.. Concentrations of isovaleric acid;. Concentrations of butyric acid;. Concentrations of hexanoic acid;. Concentrations of valeric acid;. Concentrations of isobutyric acid;. Concentrations of serine;. Concentrations of glycine;. Concentrations of gamma-aminobutyric acid;. Concentrations of ethanolamine;. Concentrations of kynurenine. *< 0.05, **< 0.01 indicate significant differences between groups, with specific comparisons detailed in the figure. N= 8 (male = 4, female = 4), N= 5 a b c d e f g h i j p p a-e f-j
Correlations between the differential SCFA, neurotransmitter and bacterial taxa
A correlation network was constructed to examine the interactions among differential bacterial taxa, SCFAs, and neurotransmitters (Fig. 8). The adjusted network contained 19 nodes and 38 significant edges, with nodes representing specific taxa or metabolites and edges indicating statistically significant correlations. Among the bacterial taxa, Prevotella, Lactobacillus, and Limosilactobacillus exhibited high connectivity, each forming multiple associations with SCFAs and neurotransmitters, suggesting their prominent roles within the microbial–metabolic interaction landscape.
Among the metabolites, epinephrine emerged as the most highly connected node, displaying strong positive correlations with several bacterial taxa, whereas kynurenine showed predominantly negative correlations with multiple neurotransmitters, indicating its distinct regulatory pattern within the metabolic network. These interaction patterns highlight key microbial and metabolic nodes that may contribute to the regulation of the gut-brain metabolic axis.
Network analysis of microbiota‒metabolite interactions in the E_ASD group. This network diagram illustrates the interactions between the gut microbiota and metabolites. The blue nodes represent bacterial genera, the green nodes represent neurotransmitters and other related compounds, and the orange nodes represent SCFAs and other metabolites. The connections between nodes indicate correlations or interactions, with the thickness of the lines reflecting the strength of the relationships. Red edges represent positive correlations, whereas blue edges indicate negative correlations. AA, Acetic acid; IBA, Isobutyric acid; BA, Butyric acid; IVA, Isovaleric acid; VA, Valeric acid; HA, Hexanoic acid
Discussion
Over the past years, research has increasingly illuminated the impact of exercise on the gut microbiota [35–40], with an increasing number of studies confirming the link between gut microbes and autism. This study investigated the effects of six weeks of voluntary wheel running on autism-like behaviors and its association with changes in the gut microbiota. The results demonstrated that this six-week exercise intervention not only ameliorated autism-like behaviors but also significantly modulated the gut microbiota, SCFAs, and neurotransmitters in the PFC. Moreover, transplantation of gut microbiota from the E_ASD group markedly improved behavioral performance in ASD rats, accompanied by alterations in SCFAs and neurotransmitter levels. These findings suggest that exercise intervention improves autism-like behaviors by modulating the gut microbiota, which in turn influences SCFAs and neurotransmitters.
From a behavioral perspective, early-stage exercise primarily improved social interest and preference for social novelty in VPA rats, while having no significant effect on anxiety-like behavior in the open field. This pattern of “social behavior improvement taking precedence” aligns with previous research, indicating that exercise more readily modulates prefrontal-limbic circuits associated with social motivation and reward processing, while exerting relatively limited influence on anxiety- or emotion-related amygdala pathways [41–44]. Previous research suggests exercise enhances synaptic plasticity in the PFC and boosts signaling of reward-related neurotransmitters [45]. Notably, gut microbiota transplantation in the exercise group partially replicated the improvement in social behavior, whereas no similar changes were observed in the sFMT group. These findings indicate that exercise-induced microbial changes can transfer social behavioral benefits, consistent with the gut–brain axis as a key regulator of social function.
Our findings indicated that six weeks of voluntary wheel running led to alterations in the gut bacterial profiles, notably increasing the abundance of the genera Limosilactobacillus and Lactobacillus, which are recognized as probiotics involved in carbohydrate fermentation into lactic acid and butyric acid. Recent studies have shown that Limosilactobacillus reuteri significantly improves social function in children with ASD, with positive effects observed across various measurement indicators. However, it does not significantly affect the overall severity of autism or other behavioral domains [46]. In addition, Lactobacillus plantarum PS128 has also demonstrated significant improvements in children with ASD, particularly in social communication, with fewer side effects. Research indicates that younger children benefit more prominently, highlighting the specific effects of these strains in alleviating ASD symptoms [47], particularly in the improvement of social behaviors. These findings support the notion that exercise may enhance digestion and overall gut health by promoting the growth of beneficial bacteria [48–52], suggesting that exercise plays a role in fostering a healthier gut microbiome.
Our study also revealed the positive impact of exercise on the production of SCFAs by the microbiota. Previous studies have demonstrated that exercise can modulate gut microbiota and their metabolites, particularly short-chain fatty acids, and exert effects on central nervous system function through multiple gut-brain pathways-including immune signaling, the vagus nerve, and the HPA axis-acting in concert. This mechanistic framework provides a crucial foundation for understanding how exercise improves brain function [53]. Specifically, exercise was found to increase the levels of butyric acid, acetic acid, and hexanoic acid, which are associated with various health benefits, including enhanced immune function, improved colonic epithelial cell integrity, and improved brain function [54]. These findings align with those of previous studies, such as the work of Matsumoto et al. [55], who reported an increase in butyric acid levels in the cecal contents of rats following 5 weeks of voluntary wheel running. Similarly, Erlandson et al. [56] reported that sedentary elderly individuals presented increased butyric acid levels in the gut following exercise intervention. Additionally, Barton et al. [54] reported that athletes had higher levels of SCFAs than did sedentary individuals. Notably, our previous research revealed that, compared with control rats, rats with VPA-induced ASD presented lower SCFA levels in their feces [33]. However, in the present study, we found that six weeks of voluntary wheel running significantly increased SCFA levels in the feces of autistic rats, restoring them to within the normal range. SCFAs play a critical role in the growth and development of both the intestinal and central systems, serving as essential energy substrates [57]. Notably, accumulating evidence also supports the view that gut microbes influence central neurochemistry [14].
Moreover, previous studies have suggested that a neuronal excitation/inhibition imbalance caused by dysfunctional neurotransmitters is an important etiology of autism [57–59] and that gut microbes are capable of producing most neurotransmitters found in the human brain [14]. A cohort study revealed that multiple neurotransmitter biosynthesis-related pathways in the gut microbiome were depleted in children with ASD compared with TD children [60]. Therefore, in this study, we further measured 55 transmitters in the PFC through neurotransmitter-targeting technology via LC‒MS. We found that, in the early stage of improvement during adolescence, several neurotransmitters in the PFC, such as threonine, kynurenine, tryptophan, 5-hydroxyindoleacetic acid, and betaine aldehyde chloride [33], were present at lower levels in the ASD group than in the control group. However, at adulthood, after 10 weeks, the levels of these neurotransmitters were greater than those in the control group. Notably, the trend of neurotransmitter changes in the exercised group was consistent with that in the control group, indicating a restorative effect. These findings support existing research suggesting that brain development may be associated with the characteristics of ASD. In summary, this study demonstrated that exercise can ameliorate the behavioral and synaptic abnormalities in offspring induced by VPA exposure, providing potential insights for interventions in ASD.
At the molecular level, exercise-induced changes in SCFAs and neurotransmitters may jointly influence ASD-like behaviors through multiple intersecting pathways. As a prototypical anti-inflammatory metabolite, butyrate not only enhances gene expression of neuroplasticity-related genes in the PFC and hippocampus but also improves intestinal barrier function and reduces peripheral inflammation, thereby diminishing inflammatory signaling interference with the central nervous system [61–63]. Acetic acid and propionic acid, meanwhile, transmit gut signals to the brainstem and limbic system via the vagus nerve, modulating neural circuits associated with social motivation [64]– [65]. At the same time, exercise-induced reshaping of tryptophan metabolism may reduce the production of neurotoxic metabolites such as 3-hydroxykynurenine and enhance serine-related signaling, a process typically associated with reduced oxidative stress and alleviation of autism-like symptoms [66]– [67]. Collectively, exercise synergistically optimizes gut-brain axis function across multiple levels by modulating microbial metabolites and brain neurotransmitters, thereby producing comprehensive improvements in ASD-like behaviors.
More importantly, to investigate the causal relationship between behavioral changes and the effects of exercise on the microbiota-gut-brain axis, we transplanted the fecal microbiota from the E_ASD group rats into the ASD group rats via intragastric administration. The results indicated that after four weeks of FMT, the FMT group rats presented significant behavioral improvements, along with alterations in the gut microbiota, SCFAs, and central neurotransmitters, which began to resemble the profiles observed in the E_ASD group. However, no significant changes were observed in the sFMT group. At present, the commonly used methods are beneficial bacterial supplementation, diet structure adjustment, antibiotic treatment and fecal bacterial transplantation. Although some small-scale studies have suggested that probiotic supplementation may help alleviate certain complications of ASD, such as constipation, or reduce its core symptoms, the benefits observed are generally modest and short-lived. This may be because supplementation with a certain probiotic alone has too little effect on the intestinal microecology. For example, a systematic review by Kristensen et al. [68] revealed that supplementation with probiotics did not seem to affect the composition of the fecal microbiota in healthy people. Microecotherapy is considered a very promising protocol for targeting related neurodevelopmental diseases such as ASD [69, 70]. The results of the population experiment by Lin et al. [71] also showed that transplanting the gut microbiota of normally developing children into ASD patients could significantly improve their gut microbiota properties. In 2017, Kang et al. [72] reported the transplantation of the gut microbiota in children with ASD complicated with gastrointestinal problems and reported that the intestinal problems of patients were not only reduced but also that the core symptoms of ASD improved. Moreover, the results of the follow-up two years later showed that this improvement effect still persisted [10]. Recently, Chen et al. [73] transplanted fecal bacteria from healthy people into ASD mice after in vitro culture (gavage once every other day, a total of 7 times) and reported that the core symptoms of ASD model mice could be significantly improved. These findings suggest that supplementing with the body’s native gut microbiota, as a source of regulation, may offer more durable and effective results compared to supplementation with exogenous microbes, such as most current probiotics, which are not isolated from the gut. The adjustment of the gut microbiota by exercise is driven by exercise and does not produce exogenous microorganisms; thus, the gut microbiota can be well colonized and function after transplantation into the same type of body. Therefore, in the present study, we transplanted the whole gut microbiota of the exercise group. Notably, after 4 weeks of FMT, incomplete alignment was observed in the FMT group and the E_ASD group after transplantation. This discrepancy can be attributed to two factors: the interactions among the gut microbiota itself and the interactions among neurotransmitters. However, further studies are needed to clarify the specific mechanisms involved.
Notably, while both exercise intervention and FMT improve ASD-like behaviors, the alterations they induce in gut microbiota composition, SCFA levels, and neurotransmitter profiles are not entirely consistent. This suggests their effects may not stem from a simple “species replication” of donor microbes, but rather represent a function-centered “selective expression” process [74]. Despite differing taxonomic compositions, both interventions were accompanied by restorative regulation of key SCFAs like butyrate and multiple neurotransmitter pathways. This function-priority expression pattern aligns more closely with the homeostasis principles of the gut ecosystem and explains why FMT can only partially replicate the neurochemical effects triggered by exercise intervention. Overall, this phenomenon further underscores the high complexity and multi-pathway synergistic mechanisms of the gut-brain axis across various diseases [75]– [76].
Although this study provides preliminary evidence suggesting that exercise can regulate various central neurotransmitters in ASD rats through the gut microbiota, several limitations should be considered. One limitation of this study is the relatively small sample size, which may affect the generalizability of the results. Additionally, although 16S rRNA gene sequencing provides valuable information about the composition of the gut microbiota, it primarily identifies microorganisms at the genus level and may not fully capture the specific microbial species involved in modulating ASD-like behaviors. Furthermore, this study focused on short-term interventions (6 weeks of exercise and 4 weeks of FMT), and the long-term effects of exercise and microbiota modulation remain unclear. The lack of follow-up behavioral assessments after the interventions also limits our understanding of the sustained impact of exercise on ASD-like behaviors. Finally, while the results support the role of gut microbiota and short-chain fatty acids in behavior modulation, the specific molecular mechanisms underlying these effects require further investigation.
Conclusions
In conclusion, our study demonstrated that voluntary wheel running significantly improved the behavioral performance of rats with ASD. It not only increases the abundance of Limosilactobacillus and Lactobacillus but also modulates the levels of SCFAs and neurotransmitters, bringing them closer to the levels observed in normal rats. When the gut microbiota from the exercise group was transplanted into ASD rats, similar changes were observed as those in the exercise group. These findings suggest that physical exercise and FMT may influence SCFAs and neurotransmitters by modulating the gut microbiota, providing a potential mechanism for the improvement of ASD-like behaviors.
Supplementary Information
Acknowledgements
We would like to thank MetWare Co. Ltd for its contribution to the measurement of 16S rRNA gene sequencing, and the quantification of SCFAs and neurotransmitters.
Authors’ contributions
Xiaohui Hou and Kai Wu designed the study. Jiugen Zhong, Baoyuan Zhu, Yinhua Li, and Yanqing Feng performed the experiments, collected the data, and conducted the analysis. Jiugen Zhong drafted the manuscript. Xiaohui Hou and Zhi Zou critically revised the manuscript. All authors contributed to the article and approved the final version.
Funding
This study was supported by the Key-Area Research and Development Program of Guangdong Province (grant no. 2023B0303020001), the Provincial Significant Scientific Research Projects for General Universities in Guangdong Province (grant no. 2021ZDJS021) and the Guangzhou Municipal Science and Technology Bureau, Basic Research Program (grant no. 2025A04J4356).
Data availability
All data generated and/or analyzed in this study have been deposited in the Genome Sequence Archive (GSA). The raw sequencing data can be retrieved using the GSA accession number CRA029953, and the project-level information corresponds to the BioProject accession PRJCA046171.
Declarations
Ethics approval and consent to participate
The experimental protocol was approved by the Ethics Research Committee of Guangzhou Sport University and was conducted in accordance with the Guide for the Care and Use of Laboratory Animals.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Contributor Information
Kai Wu, Email: kaiwu@scut.edu.cn.
Xiaohui Hou, Email: houxh@gzsport.edu.cn.
References
Associated Data
Supplementary Materials
Data Availability Statement
All data generated and/or analyzed in this study have been deposited in the Genome Sequence Archive (GSA). The raw sequencing data can be retrieved using the GSA accession number CRA029953, and the project-level information corresponds to the BioProject accession PRJCA046171.
Gut-Brain Axis