Age-dependent effects of exercise on gut microbiota-mitochondria axis and cognitive function in aging mice
Article information
Abstract
Aging is accompanied by progressive impairments in mitochondrial bioenergetics, apoptosis regulation, and gut microbiota homeostasis, all of which contribute to cognitive decline. In this study, we investigated whether the effects of treadmill exercise on the gut microbiota-mitochondrion-neuronal plasticity axis differed between young (15 months) and old (28 months) mice. Male C57BL/6 mice were randomly assigned to the following groups: early sedentary, early exercise, late sedentary, or late exercise groups and completed an 8-week treadmill training protocol. Cognitive function was assessed using the passive avoidance test and the Morris water maze test. Hippocampal mitochondrial respiration, Ca2+ retention capacity, and Bax/Bcl-2 expression were quantified, and the gut microbiota composition was analyzed using 16S ribosomal RNA sequencing. Mice that did not exercise in old age exhibited memory impairment, decreased mitochondrial oxidative respiration, reduced Ca2+ retention, increased Bax expression, decreased Bcl-2 levels, and decreased abundance of Lactobacillus, Bifidobacterium, and Akkermansia. Exercise significantly improved behavioral performance, mitochondrial function, and apoptosis balance, while also increasing beneficial gut microbiota. Notably, these effects were significantly greater in late-aged compared to early-aged mice. These results demonstrate that the efficacy of exercise in modulating the microbiota-mitochondrion-brain axis varies with age. Early-aged appears to represent a more responsive biological period during which exercise is more effective in improving mitochondrial integrity, microbiota composition, and cognitive resilience. These results suggest that initiating exercise early in the aging process may maximize neuroprotective effects and delay age-related functional decline.
INTRODUCTION
Aging is a progressive biological process characterized by a decline in the function of multiple physiological systems, including the central nervous system. Mitochondria are essential regulators of neuronal survival, as they are involved in adenosine triphosphate production, calcium buffering, and redox balance. Age-related mitochondrial dysfunction contributes to increased oxidative stress, mitochondrial permeability transition pore (mPTP) opening, and activation of apoptotic signaling pathways via Bcl-2 family regulation and caspase cascades (Endres and Friedland, 2023; Salminen et al., 2012).
The gut microbiota influences brain function through microbial metabolites, endocrine signaling, immune pathways, and neural networks such as the vagus nerve (Hashim and Makpol, 2022; Loh et al., 2024). Importantly, age-related dysbiosis is characterized by a decrease in beneficial taxa, including Lactobacillus and Bifidobacterium, and an increase in potentially harmful microbial communities, such as Proteobacteria and Desulfovibrio (Mishra and Thakur, 2022). This gut microbiota imbalance has been proposed to directly influence mitochondrial activity through metabolites such as short-chain fatty acids (SCFAs) and tryptophan derivatives, which regulate mitochondrial respiration, inflammatory status, and apoptosis vulnerability (Jackson and Theiss, 2020; Zachos et al., 2024). These findings have led to the emergence of the “microbiota-mitochondria axis” hypothesis, which links peripheral microbiota status to neuroaging.
Importantly, aging is not a single physiological state but rather a staged process. Early aging is often characterized by mild mitochondrial inefficiency and subclinical changes in microbiota diversity, whereas mid-aging is characterized by marked mitochondrial decline, chronic inflammation, and accelerated apoptotic susceptibility (Mattson and Arumugam, 2018; Mitnitski et al., 2017). These stage-specific biological differences suggest that responses to therapeutic stimuli such as exercise may be inconsistent throughout the aging process. Nevertheless, direct comparative studies examining differences in exercise effects between early and late aging are still limited.
Physical exercise is one of the most effective non-pharmacological interventions known to mitigate age-related physiological decline. Exercise has been shown to promote mitochondrial biogenesis, modulate apoptotic signaling, and promote synaptic plasticity in aged animal models (Prajapati et al., 2023; Sun et al., 2022). Furthermore, exercise alters the gut microbiome, increasing beneficial systemic and metabolic outputs associated with neuroprotection (Cintado et al., 2025; Nicolas et al., 2024).
Therefore, in line with the aim of this study to elucidate the stage-dependent effects of exercise on the microbiota-mitochondrial neuroplasticity axis during aging, we investigated whether an 8-week treadmill exercise intervention differentially affected gut microbiota composition, hippocampal mitochondrial function, apoptosis-related protein expression, and cognitive abilities in early (15 months) and late (28 months) mice.
MATERIALS AND METHODS
Experiment animals and groups
All procedures involving animals followed the guidelines of the National Institutes of Health and the Korean Academy of Medical Sciences. The study protocol was reviewed and approved by the Institutional Animal Care and Use Committee of Kyung Hee University (approval number: KHUASP [SE]-20-573). Male C57BL/6 mice were used in this study and housed under controlled environmental conditions (temperature: 25°C±1°C, 12-hr light/dark cycle, lights on from 7:00 a.m. to 19:00 p.m.). After acclimation, the mice were randomly assigned to one of four experimental groups based on age and exercise condition (n=10 per group): early sedentary group (15 months), early exercise group (15 months), late sedentary group (28 months), and late exercise group (28 months).
Exercise protocol
Exercise training was performed as previously described (Park et al., 2024). Mice assigned to the exercise group underwent a 1-week habituation period prior to intervention to reduce handling and environmental stress. The exercise protocol was performed 5 days a week for 8 weeks. Each session consisted of a 5-min warm-up at 3 m/min, followed by 20 min of treadmill running at 10 m/min, and a final 5-min cool-down at 3 m/min at a 0° incline. Sessions were performed once daily, and the mice were continuously monitored to ensure appropriate running behavior. Sedentary control animals were handled identically and were kept on a stationary treadmill for the same amount of time without running.
Step-through passive avoidance test
Short-term associative memory was assessed using a modified step-though passive avoidance test, a modified version of a previously established procedure (Park et al., 2022). Briefly, mice were placed in a two-chamber apparatus consisting of a light chamber and a dark chamber. During the first two familiarization trials, mice were allowed to freely enter the dark chamber and remain there for approximately 20 sec. In the third exposure, a scrambled footshock (1 mA, 2 sec) was delivered immediately after entering the dark chamber. Memory retention was assessed by measuring the latency to re-enter the dark chamber 24 hours later, with a maximum of 300 sec.
Morris water maze test
Spatial learning and memory were assessed using the Morris water maze test, as previously described (Park et al., 2022). Mice completed a 60-sec habituation swim in a pool without a platform the day before training. Training was conducted for five consecutive days, consisting of three trials per day. On each trial, mice were released from a pseudorandom starting position and given up to 60 sec to find a hidden escape platform. If they found the platform, they remained on it for 30 sec; if they failed to find it, they were gently guided to the platform. Twenty-four hours after the final acquisition day, the platform was removed and a 60-sec free swim was allowed for an exploration test. Behavioral performance, including time spent in the target quadrant and exploration patterns, was quantified using automated tracking software (SMART, Spain).
Western blot analysis for Bax and Bcl-2
Western blot analysis was performed using a slightly modified method described previously (Park et al., 2022). Hippocampal tissue samples were homogenized in ice-cold lysis buffer containing protease inhibitors and centrifuged at 12,000×g for 15 min at 4°C. Protein concentration was measured using a colorimetric assay. Equal amounts of protein (25–30 μg) were denatured and transferred to nitrocellulose membranes. After blocking with 5% nonfat milk, the membranes were incubated overnight at 4°C with primary antibodies against Bax (1:1,000), Bcl-2 (1:1,000), and β-actin (1:3,000). Horseradish peroxidase-conjugated secondary antibodies were applied for 1 hr, and proteins were detected using enhanced chemiluminescence. Band intensities were quantified using ImageJ software (National Institute of Health, USA) and normalized to β-actin.
Measurement of mitochondrial oxygen consumption
Mitochondrial respiration was quantified using a high-resolution respirometer at 37°C, based on previously reported methodology (Park et al., 2019). After mechanical homogenization and permeabilization, hippocampal samples were transferred to the respirometer chamber containing assay buffer. Complex I-coupled respiration was initiated with glutamate (2 mM) and malate (1 mM), followed by adenosine diphosphate (ADP) (2 mM) to stimulate oxidative phosphorylation. Succinate (3 mM) was then added to assess maximal coupled respiration via Complex II. Cytochrome-c was applied to confirm mitochondrial outer membrane integrity. Oxygen consumption was continuously monitored and normalized to tissue mass, reported as pmol O2/min/mg. This procedure was modified from a validated hippocampal mitochondrial protocol.
Measurement of mitochondrial calcium retention capacity
Mitochondrial permeability transition sensitivity was assessed by assessing Ca2+ retention capacity using calcium green-5N fluorescence at 37°C, following a previously described procedure (Park et al., 2019). After homogenization, samples were equilibrated in a buffer supporting state 4 respiration, and baseline fluorescence was recorded. Gradual Ca2+ pulses were applied while continuously monitoring fluorescence (excitation 506 nm, emission 532 nm). A rapid increase in fluorescence indicated permeability transition pore opening and Ca2+ release. Total Ca2+ uptake before pore opening was quantified and expressed as pmol Ca2+/mg tissue.
Measurement of gut microbiome composition
Microbial community profiling was performed as previously described (Park et al., 2024). Fresh fecal samples were collected aseptically and stored at −80°C until DNA extraction. Approximately 200 mg of the sample was processed using a silica membrane-based genomic DNA isolation method. The bacterial 16S rRNA V3–V4 region was amplified using universal primers in a polymerase chain reaction (PCR) containing template DNA, buffer, deoxynucleotide triphosphates, and a high-fidelity polymerase. Cycle parameters included denaturation, annealing, and extension steps, followed by a final extension step. Amplicons were purified using magnetic beads and indexed using PCR to add dual barcodes and sequencing adapters. The final libraries were cleaned, quantified by quantitative PCR, and fragment integrity assessed prior to paired-end sequencing on an Illumina MiSeq platform. Reads underwent quality filtering, trimming, sequence assembly, chimera removal, and operational taxonomic units clustering at 97% similarity. Taxonomic identities were assigned using curated reference datasets, and relative abundance profiles were generated to assess microbial community changes associated with aging and exercise.
Data analysis
Statistical analyses were performed using IBM SPSS Statistics ver. 21.0 (IBM Co., USA). Before hypothesis testing, all variables were assessed for normality and homogeneity of variance. Data are presented as the mean±standard error of the mean. Differences between groups were analyzed using one-way analysis of variance. If a significant main effect was found, post hoc comparisons were performed to confirm differences between groups. Effect sizes and 95% confidence intervals were reviewed to aid in the interpretation of statistical significance. In all analyses, the statistical significance level was set at P<0.05.
RESULTS
The change in behavioral performance
Fig. 1 summarizes the behavioral outcomes assessed using the step-through avoidance test and the Morris water maze test. In the passive avoidance task (left panel), short-term memory performance was significantly decreased in the late-aged sedentary group compared to the early-aged control (P<0.05). Exercise enhanced avoidance latency in both early- and late-aged groups, with significant increases observed relative to their respective sedentary counterparts (P<0.05). In the Morris water maze probe test (right panel), the percentage of time spent in the target quadrant was significantly reduced in the late-aged sedentary group (P<0.05), indicating impaired spatial memory. Exercise significantly improved spatial memory performance in both age groups compared to the sedentary condition (P<0.05), with more pronounced effects observed in the early-aged exercise group.
Effects of aging and exercise on cognitive performance assessed by behavioral tasks. The left panel shows step-through latency in the passive avoidance test. The right panel displays time spent in the target quadrant during the Morris water maze trial. A, aging early sedentary group; B, aging early exercise group; C, aging late sedentary group; D, aging late exercise group. Data are expressed as the mean±standard error of the mean. *P<0.05 compared to aging early sedentary group. #P<0.05 compared to aging late sedentary group.
The change in apoptosis-related protein expression
Fig. 2 presents hippocampal protein expression levels of Bcl-2 and Bax. Relative Bcl-2 expression was significantly decreased in the late-aged sedentary group compared to the early-aged control (P<0.05). Exercise elevated Bcl-2 expression in both early- and late-aged animals, with significant increases observed compared to the respective sedentary groups (P<0.05). In contrast, Bax expression was significantly elevated in the late-aged sedentary group compared to the early-aged control (P<0.05). Exercise markedly reduced Bax expression in both age groups relative to their sedentary counterparts (P<0.05), although Bax levels remained higher in late-aged animals than in the early-aged groups.
Effects of aging and exercise on hippocampal apoptosis-related protein expression. The upper panel shows representative Western blot bands for Bcl-2, Bax, and β-actin. The lower left bar graph presents relative Bcl-2 expression and the lower right bar graph shows relative Bax expression. Data are expressed as the mean±standard error of the mean. A, aging early sedentary group; B, aging early exercise group; C, aging late sedentary group; D, aging late exercise group. *P<0.05 compared to aging early sedentary group. #P<0.05 compared to aging late sedentary group.
The change in mitochondrial respiration and Ca2+ retention capacity
Fig. 3 summarizes the mitochondrial functional outcomes. Mitochondrial oxygen respiration was significantly lower in the late-aged sedentary group under all substrate conditions (glutamate/malate, ADP-stimulated, and succinate-supported respiration) compared to the early-aged control (P<0.05). Exercise enhanced oxygen consumption in both early- and late-aged groups relative to their respective sedentary controls (P<0.05), with the greatest improvement observed in the early-aged exercise group. Similarly, mitochondrial Ca2+ retention capacity was reduced in the late-aged sedentary group compared to the early-aged control (P<0.05). Exercise significantly increased Ca2+ retention capacity in both age groups compared to sedentary animals (P<0.05), although levels remained lower in late-aged mice than in their early-aged counterparts.
Effects of aging and exercise on mitochondrial respiration and Ca2+ retention capacity. The left panel shows mitochondrial O2 consumption under glutamate/malate (GM), adenosine diphosphate (ADP)-stimulated, and succinate (SUCC)-supported respiration conditions. The right panel displays mitochondrial Ca2+ retention capacity. A, aging early sedentary group; B, aging early exercise group; C, aging late sedentary group; D, aging late exercise group. Data are expressed as the mean±standard error of the mean. *P<0.05 compared to aging early sedentary group. #P<0.05 compared to aging late sedentary group.
The change in gut microbiome composition
Fig. 4 illustrates the relative abundance of key gut microbial taxa. The abundance of Lactobacillus and Bifidobacterium was significantly reduced in the late-aged sedentary group compared to the early-aged control (P<0.05). Exercise increased the abundance of both taxa in early- and late-aged animals, with significant improvements observed compared to the respective sedentary groups (P<0.05). A similar trend was observed for Akkermansia muciniphila, where the late-aged sedentary group exhibited lower abundance relative to the early-aged control (P<0.05). Exercise significantly increased Akkermansia abundance in both age groups compared to sedentary animals (P<0.05), although levels remained lower in late-aged mice than in their early-aged counterparts.
Effects of aging and exercise on gut microbial composition. The left, middle, and right panels represent the relative abundance of Lactobacillus, Bifidobacterium, and Akkermansia muciniphila, respectively, based on 16S ribosomal ribonucleic acid sequencing. A, aging early sedentary group; B, aging early exercise group; C, aging late sedentary group; D, aging late exercise group. Data are expressed as the mean±standard error of the mean. *P<0.05 compared to aging early sedentary group. #P<0.05 compared to aging late sedentary group.
DISCUSSION
In this study, we investigated whether the effects of treadmill exercise on mitochondrial function, apoptosis signaling, gut microbiota composition, and cognitive function differed between early aging (15 months) and late aging (28 months). Key findings include: (1) Late-aged mice exhibited significantly impaired mitochondrial respiration, reduced Ca2+ storage capacity, increased Bax/Bcl-2 ratio, and cognitive impairment. (2) Aging was associated with gut microbiota dysbiosis, specifically characterized by a decrease in Lactobacillus, Bifidobacterium, and Akkermansia. (3) Treadmill exercise significantly mitigated these impairments, with more pronounced effects observed in early-aged mice compared to late-aged mice.
Consistent with previous findings, aged mice exhibited significantly reduced mitochondrial respiratory capacity and Ca2+ storage, along with increased Bax expression and decreased Bcl-2 expression, hallmarks of age-related mitochondrial fragility and apoptosis. These results are consistent with previous studies showing that mPTP opening is exacerbated with age, promoting neuronal death and functional decline in memory-related brain regions (Chabi et al., 2008; Toman and Fiskum, 2011). This age-related mitochondrial functional vulnerability has also been well documented in skeletal muscle studies, further supporting the systemic nature of mitochondrial decline with age (Seo et al., 2016).
The exercise-induced improvements in mitochondrial bioenergetics observed in this study are consistent with previous reports demonstrating enhanced oxidative phosphorylation capacity, improved mitochondrial dynamics, and enhanced antioxidant defenses after endurance-based exercise interventions (Hood et al., 2019; Short et al., 2004). The partial restoration of mitochondrial function in aged mice supports the hypothesis that the mitochondrial system retains adaptive plasticity throughout aging, but appears to be attenuated in response to early-life stages (Short et al., 2005). Furthermore, the enhanced Bax/Bcl-2 ratio after exercise further supports the antiapoptotic role of physical activity, which is consistent with evidence that exercise attenuates mitochondrial-mediated apoptosis and modulates intrinsic apoptotic signaling axis activation and mitochondrial stabilization (Kwak et al., 2006).
Aging was associated with significant decreases in the abundance of Lactobacillus, Bifidobacterium, and Akkermansia, genera closely linked to gut barrier function, SCFA biosynthesis, synaptic plasticity, and metabolic homeostasis (Cintado et al., 2025; Mishra and Thakur, 2022). In particular, the observed decrease in Akkermansia muciniphila in aged mice may be particularly relevant, as reductions in mucin-degrading bacteria are associated with increased metabolic inflammation and decreased neurotrophic signaling. This suggests a potential mechanistic role in amplifying vulnerability to the aged phenotype (Park et al., 2024). Treadmill exercise mitigated these age-related microbial changes, particularly in early-life mice, suggesting that microbiota plasticity declines with aging. Microbial toxins have been shown to directly impair mitochondrial integrity in brain regions such as the midbrain, highlighting the pathological relevance of the gut-mitochondrial axis (Esteves et al., 2023). These findings support the hypothesis that beneficial microbial taxa modulate the physiological effects of exercise through lactate-related microbial signaling, vagal regulation, and modulation of SCFA-based mitochondrial activity (Clark and Mach, 2017).
These results are consistent with experimental studies demonstrating that exercise modulates gut microbiota ecology, thereby improving neurogenesis, systemic metabolic signaling, and behavioral outcomes (Allen et al., 2018; Nicolas et al., 2024). This supports the broader framework in which oxidative stress, microbiota composition, and brain function dynamically interact within the gut-brain axis (Dumitrescu et al., 2018). These interactions are also known to be implicated in the pathogenesis of neurodegenerative diseases, further highlighting the need to consider microbiota-mitochondrial interactions in aging research (Borbolis et al., 2023).
This interpretation is consistent with emerging experimental evidence that exercise-induced cognitive improvements depend, at least in part, on hormetic adaptations mediated by the microbiota, which decline in aging societies (Cintado et al., 2025).
These results suggest that the benefits of exercise are mediated through the microbiota-mitochondrial neuroplasticity axis, and that strengthening this axis early in the aging process may be more effective in preventing functional decline.
Notes
CONFLICT OF INTEREST
No potential conflict of interest relevant to this article was reported.
ACKNOWLEDGMENTS
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Ministry of Science and ICT (or other relevant ministry) (Grant No. NRF-2020R1I1A1A01072455).