Comparative study of treadmill and swimming exercises on spinal regeneration and reactive astrocytes in the brain after spinal cord injury Authors: Min-Jun Kim, Yeong-Jik Kim, Tae-Beom Seo Published: 2025 ABSTRACT AbstractThe purpose of this study is to compare the effects of treadmill and swimming exercise on cell survival and reactive astrogliosis in the spinal cord and brain after spinal cord injury (SCI). Thirty-two male rats were randomly assigned to four groups: the normal group (Norm, n=8), the SCI+sedentary group (Seden, n=8), the SCI+treadmill exercise group (Tread, n=8), and the SCI+swimming exercise group (Swim, n=8). All rats in the SCI groups were allowed to recover for 2 weeks after injury, and then underwent 4 weeks of low-intensity treadmill and swimming exercise. In the spinal cord, both exercise interventions decreased the cavity formation around the injury, and significantly increased induction levels of growth associated protein-43, with greater improvements in the Tread group. In the motor cortex and hippocampus of the brain, the Tread group upregulated B-cell leukemia/lymphoma 2 and downregulated cleaved-caspase-3 and glial fibrillary acidic protein relative to the Seden group, indicating enhanced cell survival and reduced reactive astrogliosis in the injury site. Compared with swimming, treadmill exercise further improved tissue preservation and neural circuit-level recovery across the spinal cord, motor cortex, and hippocampus after SCI. Thus, our findings suggested evidence that treadmill exercise might be one of the rehabilitation strategies to support the spinal cord and brain after SCI.Keywords: Spinal cord injury, Treadmill, Swimming, Motor cortex, Hippocampus, Reactive astrocyte INTRODUCTION Spinal cord injury (SCI) initiates a complex cascade that begins with vascular rupture, intraparenchymal hemorrhage within the gray matter, and rapid neuronal loss. In the subacute phase, normally quiescent astrocytes, microglia, and other glial cells become reactive, releasing pro-inflammatory cytokines that amplify leukocyte infiltration and secondary damage (Bhatt et al., 2024). In parallel, the mitochondrial apoptosis pathway becomes activated, with B-cell leukemia/lymphoma 2 (Bcl-2) downregulated and executioner caspase-3 activated, thereby exacerbating neuronal loss. As a result of this process, a glial scar is formed by reactive astrocytes at the margin of the cystic cavity, creating an inhibitory milieu that limits axonal repair and impedes neural regeneration. SCI not only causes local changes within the spinal cord, but also has widespread effects on the neural circuits of the brain and skeletal muscles. Following SCI, synaptic plasticity in the motor cortex and hippocampus is markedly constrained, along with reduced efficiency of synaptic transmission, diminished cerebral perfusion, and dysregulated hormone secretion. In addition, disturbances in the hypothalamic–pituitary–adrenal axis emerge, further contributing to aberrant neural signaling (Solstrand Dahlberg et al., 2018). In skeletal muscle, satellite cell death, fiber-type shifts, atrophy, and reduced neurotransmitter expression further degrade motor function and complicate recovery trajectories (Kok et al., 2024; Verdijk et al., 2012). A wide array of experimental therapies—stem cell transplantation, exogenous neurotrophic factor delivery, and blockade of inhibitory molecules—seeks to reopen developmental growth programs or neutralize extrinsic brakes on regeneration. However, most are invasive and face unresolved questions regarding durability, safety, scalability, and real-world effectiveness (Chambel and Cruz, 2023; Farid et al., 2022). These limitations have intensified interest in noninvasive, system-level interventions that can reshape the postinjury environment. Accordingly, recent studies report that exercise, as a noninvasive intervention, effectively improves the neuroregenerative environment after SCI (Ying et al., 2025). Among aerobic modalities, treadmill and swimming paradigms are widely used. Both have been reported to increase regeneration- related protein expression within the injured spinal cord and to improve a cortical and hippocampal environment conducive to remodeling, in part by restraining aberrant non-neuronal proliferation (Davaa et al., 2021; Kumar et al., 2021; Li et al., 2024; Liao et al., 2017; Wu et al., 2025). Yet the literature is not uniform: benefits may depend on injury severity, timing, and dosing, and some studies suggest that initiating swimming too early can exacerbate deficits and delay recovery (Lee et al., 2025; Smith et al., 2009). Crucially, direct comparisons of treadmill and swimming on cell survival and astrocyte reactivity in the spinal cord and brain are limited. Accordingly, the present study seeks to systematically compare and delineate the effects of treadmill and swimming exercise on cell survival and reactive astrocytes in the spinal and brain following SCI. MATERIALS AND METHODS Experimental animals Sprague-Dawley male rats, 4 weeks of age, maintained under controlled environmental conditions (room temperature 22°C–24°C; relative humidity 60%) with a 12:12-hr light-dark cycle. Commercial rat chow (Samyang Co., Korea) and water were available ad libitum. All experimental procedures conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee of Jeju National University (2022-0030). The animals randomly assigned to four groups: the normal control group (Norm, n=8), the SCI+sedentary group (Seden, n=8), the SCI+treadmill exercise group (Tread, n=8), and the SCI+swimming exercise group (Swim, n=8). Spinal cord injury For induction of spinal cord contusion, all rats were anesthetized with inhaled isoflurane using an inhalation anesthesia system (Jeungdo Bio and Plant, Korea), with the isoflurane concentration maintained between 1.5% and 2.5%. After the spinal cord was exposed at the T9–10 level without dura injury, a contusion was produced by dropping a 10-g weight from a height of 2.5 cm onto the dorsal surface of the cord using the New York University Impactor System (NYU Impactor, USA) (Yu and Seo, 2024). The muscles and skin were then sutured, and the animals were kept on a heating pad (rectal temperature 36°C–37°C) until full recovery from anesthesia. All animals were euthanized 24 hr after the final exercise session by CO2 inhalation for tissue collection. The harvested samples included the thoracic spinal cord (T5–12) and cerebrum, which were stored at −80°C until further analysis. Treadmill and swimming exercise program Two weeks after SCI, animals performed treadmill and swimming exercise 5 days per week for four consecutive weeks. The treadmill exercise, adapted from Jung et al. (2016), was conducted using a motorized treadmill (CAST-10R, CASSCALE, Korea) at 0° incline and 6-m/min speed. Training consisted of 4 cycles of 3-min walking bouts during the first week, followed by 6 cycles of 5-min bouts from the second to the fourth week, with 5-min rest intervals between bouts. Swimming exercise was carried out in a custom-built acrylic tank (80 cm×50 cm×50 cm). To prevent forelimb support against the wall, a 10 g was weight attached to the tail, thereby enforcing active swimming. Water temperature maintained at 32°C–35°C, with ambient room temperature set to 25°C. The protocol was a modification of Harman et al. (2021) and consisted of 4 cycles of 3-min swimming bouts in the first week and 6 cycles of 5-min bouts from the second to the fourth week, with 5-min rest intervals between each trial. Hematoxylin and eosin staining For cavity size measurement in the injured spinal cord, prepared tissues were embedded in frozen section compound (Leica Biosystems Richmond, Inc., USA) and sagittal sectioned at 20-μm thickness using a cryostat (Leica CM1860, Leica, Germany). The sections were stained with hematoxylin and eosin (H&E; Sigma-Aldrich, USA). Slides were immersed in Mayer’s hematoxylin for 30 sec, rinsed with tap water until clear, counterstained with eosin for 30 sec, and rinsed again. Dehydration was performed by sequential immersion twice in 95% ethanol, twice in 100% ethanol, and twice in xylene. Finally, the slides were cover-slipped and examined under a light microscope (E-600, Nikon, Japan). Immunofluorescence staining The injured spinal cord and cerebrum tissues embedded in frozen section compound (Leica Biosystems) and sectioned using a cryostat (Leica). Double immunofluorescence staining was conducted to visualize neuronal survival and growth associated protein-43 (GAP-43) in the spinal cord, and astrocyte proliferation in the motor cortex and hippocampus of the brain. Sagittal sections of the injured spinal cord were cut at 20 μm, and frontal sections of the cerebrum at 30 μm. Sections were fixed in 4% paraformaldehyde with 4% sucrose in phosphate-buffered saline (PBS) for 40 min, permeabilized with 0.5% Nonidet P-40 in PBS, and blocked in 2.5% bovine serum albumin (BSA, Sigma-Aldrich) for 4 hr. Tissues were then incubated with the following primary antibodies: anti-rabbit neuronal nuclei (NeuN, 1:200, Cell Signaling Technology, USA), anti-mouse glial fibrillary acidic protein (GFAP, 1:200, Cell Signaling Technology), anti-mouse GAP-43 (1:500, Santa Cruz Biotechnology, USA). After primary incubation, rhodamine-goat anti-rabbit secondary antibody (1:600, Molecular Probes, USA) and fluorescein-goat anti-mouse secondary antibody (1:400, Molecular Probes) were treated as secondary antibodies for 1 hr at room temperature. Cell nuclei were counterstained with Hoechst 33258 dye (2.5 μg/mL, Sigma-Aldrich) for 10 min prior to the final wash. Finally, sections were coverslipped with gelatin mounting medium and imaged under a fluorescence microscope (E-600, Nikon). All digital images were captured and processed using Adobe Photoshop software (CS6, Adobe, USA). Western blot analysis The extracted tissue rinsed with PBS and lysed in Triton lysis buffer. Protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and subsequently transferred to a polyvinylidene difluoride membrane for 2 hr. Membranes were blocked with 5% BSA (Sigma-Aldrich) in tris-buffered saline containing 0.1% Tween-20 at room temperature for 1 hr and then incubated with primary antibodies at 4°C for 24 hr. The primary antibodies used were anti-rabbit Bcl-2 (1:2,000, Abcam, UK), anti-rabbit cleaved caspase-3 (1:1,000, Cell Signaling Technology), anti-mouse GFAP (1:1,000, Cell Signaling Technology), anti-mouse GAP-43 (1:1,000, Santa Cruz Biotechnology), and anti-mouse β-actin (1:2,000, Santa Cruz Biotechnology). After washing, membranes were incubated with horseradish peroxidase conjugated anti-mouse or anti-rabbit IgG secondary antibodies (1:5,000, GeneTex, USA) for 1 hr at room temperature. Protein bands were visualized using Westar enhanced chemiluminescence detection reagent (Cyanagen, Italy) and analyzed with the ChemiDoc imaging system (Bio-Rad, USA). Statistical analysis All data are expressed as the mean±standard error of the mean. Statistical comparisons were carried out using one-way analysis of variance followed by Duncan post hoc test, based on the statistical procedures used in previous studies (Yu and Seo, 2024). A significance threshold was set at P<0.05. Data processing and graphical representations were performed using Prism 6 software (GraphPad Software, USA). RESULTS Cell survival and cavity size in the injured spinal cord To compare exercise-dependent changes in proapoptotic signaling and histopathology after SCI, we performed Western blotting and H&E staining. As shown in Fig. 1A, the prosurvival marker Bcl-2 did not differ across interventions (P>0.05), whereas cleaved-caspase-3 expression was significantly lower in the Tread (P<0.001) and Swim (P<0.006) groups than the Seden group. In specific, Tread group exhibited a more pronounced reduction compared to the Swim (P<0.001) group. Consistent with these molecular findings, Fig. 1B shows that the lesion cavity size was reduced in both Tread (P<0.004) and Swim (P<0.030) groups relative to the Seden group, respectively. Changes in cell survival-related proteins and cavity formation at 4 weeks after spinal cord injury (SCI). (A) Western blot analysis on expression levels of B-cell leukemia/lymphoma 2 (Bcl-2) and cleaved-caspase-3 in the injured spinal cord. Upper panel: representative expression of Bcl-2 and cleaved-caspase-3. Lower panel: the density ratio of target protein to actin band. **P<0.01, ***P<0.001 vs. Seden; ###P<0.001 for Tread vs. Swim. (B) Hematoxylin and Eosin staining for examining cavity size at the margin of the cystic cavity in the injured spinal cord. Upper panel: image of cavity in the injured spinal cord. Lower panel: Quantitative graph on cavity size. Norm, normal control group; Seden, SCI+sedentary group; Tread, SCI+treadmill exercise group; Swim, SCI+swimming exercise group. *P<0.05, **P<0.01 vs. Seden. Reactive astrocyte and regeneration marker in the injured spinal cord We next assessed expression of GFAP and GAP-43 in the injured spinal cord by Western blotting (Fig. 2A). GFAP was increased in the Seden group post SCI, but Tread (P<0.036) and Swim (P<0.029) groups reduced expression levels of them. Conversely, GAP-43, which declined after SCI, was significantly upregulated by the Tread (P<0.042) group. As shown in Fig. 2B, NeuN–GAP-43 colocalization was higher in the Tread versus the Seden (P<0.002), consistent with improved neuronal integrity and proregenerative activity under treadmill training. Reactive astrocytes and regeneration protein at 4 weeks after spinal cord injury (SCI). (A) Quantitative analysis on expression levels of glial fibrillary acidic protein (GFAP) and growth-associated protein-43 (GAP-43) in the injured spinal cord. Upper panel: representative expression of GFAP and GAP-43. Lower panel: the density ratio of target protein to actin band. *P<0.05 vs. Seden; #P<0.05 for Tread vs. Swim. (B) Double immunofluorescence staining to investigate GAP-43 induction in neurons around the glial scar after SCI. Left panel: image of GAP-43-expressed neurons in the injured spinal cord. Right panel: quantitative graph on intensity of GAP-43. Norm, normal control group; Seden, SCI+sedentary group; Tread, SCI+treadmill exercise group; Swim, SCI+swimming exercise group. **P<0.01 vs. Seden; ##P<0.01 for Tread vs. Swim. Reactive astrocyte in the motor cortex Western blots of the motor cortex (Fig. 3A) revealed that Bcl-2 was decreased in the Seden group but it was upregulated by both the Tread (P<0.033) and Swim (P<0.035) groups. Caspase-3 was significantly reduced in the Tread (P<0.001) group. GFAP also rose following SCI and declined with the Tread (P<0.018) group. Immunofluorescence at the motor cortex (Fig. 3B) indicated lesser GFAP signals in the Tread compared to the Seden (P<0.028) group, consistent with a moderated reactive gliosis profile under treadmill exercise intervention. Cell survival and reactive astrocytes in the motor cortex 4 weeks after spinal cord injury (SCI). (A) Western blot analysis on B-cell leukemia/lymphoma 2 (Bcl-2), cleaved-caspase-3 and glial fibrillary acidic protein (GFAP) in the motor cortex post injury. Top left panel: representative expression of Bcl-2 and cleaved-caspase-3. Top right and bottom panel: the density ratio of target protein to actin band. *P<0.05, ***P<0.001 vs. Seden; #P<0.05, ##P<0.01 for Tread vs. Swim. (B) Immunostaining to examine reactive astrocyte in the motor cortex post injury. Top left panel: Hoechst-stained image of cortex and hippocampus of the brain, Bottom left panel: quantitative graph on intensity of GFAP. Right panel: image of GFAP-stained astrocytes in the motor cortex after SCI. DG, dentate gyrus; Norm, normal control group; Seden, SCI+sedentary group; Tread, SCI+treadmill exercise group; Swim, SCI+Swimming exercise group. *P<0.05 vs. Seden. Reactive astrocyte in the hippocampus In the hippocampus (Fig. 4A), Bcl-2 expression was higher in the Tread (P<0.038) and Swim (P<0.043) groups than the Seden group, whereas Caspase-3 was lower in the exercise intervention groups (P<0.010 in the Tread group; P<0.028 in the Swim group). GFAP was increased after SCI and showed a significant reduction in the Tread (P<0.010) group. In Fig. 4B, immunofluorescence quantification in the dentate gyrus of the hippocampus revealed higher reactive astrocyte levels in the Seden group relative to the Tread group (P<0.048), and both the Tread and Swim groups further decreased GFAP-positive cells in the hilus than those in the Seden group (P<0.002 and P<0.005, respectively). Cell survival and reactive astrocytes in the hippocampus 4 weeks after spinal cord injury (SCI). (A) Western blot analysis on B-cell leukemia/lymphoma 2 (Bcl-2), cleaved-caspase-3 and glial fibrillary acidic protein (GFAP) in the hippocampus post injury. Left panel: representative expression of Bcl-2 and cleaved-caspase-3. Three panels on the right: the density ratio of target protein to actin band. *P<0.05 vs. Seden; #P<0.05 for Tread vs. Swim. (B) Immunostaining to examine reactive astrocyte in various regions (dentate gyrus, hilus and molecular layer) of the hippocampus post injury. Left panel: image of GFAP-stained astrocytes in the hippocampus after SCI. Right panel: quantitative graph on intensity of reactive astrocytes. Norm, normal control group; Seden, SCI+sedentary group; Tread, SCI+treadmill exercise group; Swim, SCI+Swimming exercise group. *P<0.05, **P<0.01 vs. Seden. DISCUSSION Mitigating the pathophysiological cascade and limiting secondary injury after SCI is essential to achieve neuroprotection and axonal regeneration in the spinal cord and brain, and exercise rehabilitation is an effective and noninvasive approach to modulating these processes (Ding and Chen, 2022). Accordingly, this study investigated how treadmill and swimming exercise after SCI influence molecular indices of apoptosis, cell survival, and reactive gliosis—together with corresponding histopathological changes —across the spinal cord, motor cortex, and hippocampus. In the injured spinal cord, the caspase-3 expression level and the cavity formation were lower in the Tread group than in the other groups. During early stage of regeneration after SCI, activation of caspase-3 in the injured spinal cord precipitates apoptosis of neurons and oligodendrocytes across white and gray matter (Emery et al., 1998), and sustained apoptosis fosters secondary injury through circuit disconnection and demyelination. Accordingly, reducing caspase-3 expression may expect to mitigate secondary damage (Caprariello et al., 2012). In line with this mechanism, Jung et al. (2014) reported a significant decrease in caspase-3 in the injured cord with treadmill exercise. In present study, swimming also lowered caspase-3 relative to the Seden group following SCI, but the Tread group produced a greater reduction than swimming, indicating a modality-specific advantage of treadmill exercise in suppressing caspase-3–mediated apoptosis after SCI. In the spinal cord, we observed that GFAP expression decreased in the two exercised groups after SCI, whereas GAP-43 increased specifically with the Tread group. GFAP is a canonical marker of reactive astrocytes; its upregulation after SCI reflects astrocytic hypertrophy/proliferation and contributes to glial scar formation. By contrast, GAP-43 is a prototypical axon growth-associated protein, expressed at high levels in growth cones after peripheral and central nerve injury, and implicated in axonal elongation, regeneration, and de novo synaptogenesis (Kawasaki et al., 2018). Consistent with previous reports, exercise interventions after SCI can suppress reactive astrocyte that form glial scar around the injury site and enhance GAP-43 in neuronal cells to increase axonal regeneration, thereby fostering a proregenerative environment within the injured spinal cord (Coyoy-Salgado et al., 2023; Loy et al., 2018; Ying et al., 2005). Although multiple studies applying treadmill and swimming exercises have reported similar trends (Cai et al., 2025; Sun et al., 2013), our findings indicate that treadmill training may be comparatively more effective than swimming in attenuating glial scar formation and promoting axonal regeneration after SCI. In the motor cortex and hippocampus, protein profiling revealed a broadly similar pattern across both exercise groups: Bcl-2 was generally upregulated, whereas caspase-3 and GFAP were downregulated, with the most pronounced effects in the treadmill group (Figs. 3 and 4). In other words, treadmill exercise appeared to establish a more anti-apoptotic milieu (via increased Bcl-2), while suppressing apoptotic signaling (caspase-3) and attenuating reactive astrogliosis (GFAP), thereby easing the secondary injury environment in both regions. This pattern aligns with prior evidence that post-SCI exercise recalibrates cortical and hippocampal tissue toward anti-apoptotic and anti-scarring states. In specific, treadmill interventions have been shown to enhance PI3K/Akt activity, increase the Bcl-2/Bax ratio, and suppress cleaved caspase-3 (Jung et al., 2014), while body-weight-supported treadmill training reduces reactive astrocyte markers and glial-scar overgrowth for facilitating functional recovery after SCI (Cai et al., 2025). Although swimming exercise consistently improves locomotor outcomes after SCI (Smith et al., 2006; Yu and Seo, 2024), direct quantitative evidence for its effects on apoptosis- and gliosis-related biomarkers remains comparatively limited. Taken together, our biochemical and histological data indicate that treadmill training provides larger and more consistent improvements than swimming across spinal, cortical, and hippocampal compartments after SCI, suggesting a relative advantage for tissue preservation and circuit-level repair. Footnotes CONFLICT OF INTEREST No potential conflict of interest relevant to this article was re ported. ACKNOWLEDGMENTS This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT)(NRF-2022R1A2C1011162). REFERENCES