Introduction

Parkinson’s disease (PD) is a chronic, progressive neurodegenerative disorder that primarily affects motor control through the loss of dopaminergic neurons in the substantia nigra (SN)1. Although drug treatments such as levodopa can relieve symptoms, there is currently no evidence that any treatment can halt PD progression2. Consequently, researchers have increasingly focused on non-pharmacological interventions to complement existing therapies and address disease pathophysiology at the molecular level.

Exercise has emerged as a promising adjunctive therapy for PD3. Over the past decade, epidemiological evidence has established an association between physical activity and PD risk, demonstrated exercise’s role in mitigating disease progression and relieving symptoms, collectively underscoring the critical importance of exercise4,5. Various types of physical exercise effectively alleviate PD symptoms, potentially through coordinated modulation of cytokine networks and associated signaling pathways6. Animal studies demonstrated that exercise can counteract the neurodegenerative processes associated with PD by promoting neuroplasticity, reducing neuroinflammation, and enhancing mitochondrial function7,8,9. In 2016, “exerkines” were first proposed and defined as peptides and nucleic acids released by skeletal muscle and other organs induced by exercise10. Exerkines include a number of neurotrophic factors, such as brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), growth/differentiation factors (GDFs), and bone morphogenetic proteins (BMPs)11,12,13,14, which play critical roles in neuronal survival and synaptic plasticity. These factors support the growth and maintenance of neurons and are essential for the repair of damaged dopaminergic systems, providing a neuroprotective effect in PD.

Mitochondrial dysfunction and chronic neuroinflammation are important hallmarks of PD pathology. Exercise promotes mitochondrial biogenesis and improves mitochondrial function, which is crucial in mitigating the oxidative stress and energy deficits that contribute to neuronal death in PD15,16. Exercise has been shown to reduce systemic inflammation and modulate microglial activation, which in turn reduces neuroinflammation in the brain17.

Despite robust evidence supporting exercise efficacy in PD, the molecular mechanisms underlying its benefits remain incompletely elucidated. This review synthesizes symptom-alleviating exercise modalities across PD models and patients, along with the underlying molecular mechanisms (Fig. 1): (1) upregulation of neuroprotective factors (BMP/GDNF/BDNF), (2) neuroinflammation attenuation via reduced pro-inflammatory mediators and enhanced neuroplasticity, (3) mitochondrial functional restoration, and (4) irisin-mediated regulation—acting as a central hub integrating protective factor synthesis, anti-inflammatory responses, and mitochondrial oxidative stress mitigation. Beyond these primary axes, exercise engages complex molecular networks with extensive crosstalk. We further evaluate current exercise-based therapies and propose future research directions and possible approaches.

Fig. 1: Molecular mechanisms underlying exercise-induced benefits in Parkinson’s disease.
Fig. 1: Molecular mechanisms underlying exercise-induced benefits in Parkinson’s disease.
Full size image

Various forms of exercise that alleviate symptoms in patients with Parkinson’s disease (PD) include treadmill training, overground walking, active and passive cycling, endurance training, resistance training, Tai Chi, and rotational walking exercise (RWE). The beneficial molecular mechanisms of exercise involve the stimulation of neurotrophic factor release, anti-inflammatory effects, regulation of mitochondrial oxidative stress, production of irisin, and other interconnected pathways. These mechanisms interact to contribute to the neuroprotective effects of physical activity in PD. Image created with BioRender.com.

Exercise-induced improvement in PD

Many studies have found that different forms of exercise play critical roles in improving PD symptoms, and identifying the optimal exercise mode has become a research focus. To date, three randomized controlled trials demonstrate that moderate-to-vigorous intensity aerobic exercise may slow PD progression and ameliorate symptomatology. Notably, these trials revealed a dose-response relationship between exercise intensity and motor function improvement, with higher-intensity aerobic exercise resulting in greater alleviation of motor symptoms18,19,20. In a rat PD model induced by 6-OHDA, treadmill training improved gait and walking speed by enhancing dopamine (DA) transmission and corticostriatal synaptic plasticity21. A six-month brisk walking and balance program similarly relieved motor symptoms, including gait and dynamic balance, in PD patients18. Studies further reveal differential brain activation patterns between treadmill versus ground walking and between active and passive cycling. Significantly, incorporating dynamic balance challenges into treadmill training yields superior improvements in gait performance and postural stability compared to conventional treadmill protocols22,23,24. Epidemiological findings suggest that physical activity may lower PD risk by protecting dopaminergic neurons, repairing motor control circuits, and promoting endogenous neurotrophic factors25. A recent study by De Laat et al. demonstrated that six months of high-intensity exercise reversed the anticipated decline in dopamine transporter availability in early-stage PD patients. ¹⁸F-FE-PE2I PET imaging revealed significant uptake increases in both the substantia nigra and putamen26. Mechanistically, these therapeutic benefits are attributed to enhanced cortical-striatal plasticity and dopaminergic neurotransmission27,28.

Crucially, aerobic exercise can be done at home by patients with PD with mild disease severity, and it attenuates off-state motor signs19. Aerobic exercise augments neuroplasticity in PD through three core elements: exercise intensity, movement repetition, and cognitive engagement. A prospective observational cohort study innovatively quantified aerobic exercise intensity using metabolic equivalent of task, providing a direction for the exploration of exercise forms in patients with PD29. High-intensity regimens specifically strengthen basal ganglia-cortical connectivity, thereby ameliorating motor symptoms7,19,30. Tai Chi training, a traditional Chinese martial art, improves balance, reduces falls, and outperforms stretching and resistance training in PD patients31. A 3.5-year follow-up cohort study indicates that Tai Chi training reduced the annual changes in the deterioration of the Unified Parkinson’s Disease Rating Scale (UPDRS) and delayed the need for increasing antiparkinsonian therapies32. The benefits of Tai Chi training on PD extend to non-motor symptoms, including cognitive function, sleep, and quality of life, with evidence supporting its long-term beneficial effectiveness32,33,34. A recent study demonstrates that Yi Jin Jing – a traditional mind-body practice – improves cognition and physical activity in PD patients, showing promise as a non-pharmacological home-based adjunct therapy35. Complementary evidence reveals yoga’s superior efficacy over proprioceptive training for enhancing balance and proprioceptive sensitivity36, and dance therapy confers significant motor/non-motor benefits, including improved balance control, gait parameters, executive function, and quality-of-life indices37. Nevertheless, the neurobiological mechanisms underlying these exercise modalities remain incompletely characterized.

Notably, a network meta-analysis reveals distinct effects of exercise modalities on PD symptoms38. For movement improvement, dance shows moderate benefits, while aquatic training, gait/balance/function training, and multidomain training likely offer moderate benefits. Mind-body training (e.g., Tai Chi, yoga) and endurance training demonstrate small benefits, whereas flexibility training shows minimal effects. Regarding quality of life, aquatic training may provide large benefits, endurance training moderate benefits, and gait/balance/function/multidomain training small benefits. The effects of mind-body training, gaming, strength training, dance, “Lee Silverman Voice training BIG” (LSVT BIG), and flexibility training on quality of life remain uncertain. The authors made a conclusion that there does not appear to be a clear difference in the benefits of different forms of exercise38. For PD patients, most types of physical activity can reduce the severity of motor symptoms and improve quality of life, while the exact exercise type might be secondary. Both forced and voluntary exercise can exert anti-PD effects by upregulating neurotrophic factors expression39. Therefore, the beneficial effects of physical activity seem to be certain, and little evidence of differences exists between these interventions, though it may be possible that specific motor symptoms may be treated most effectively by PD-specific programs. The lack of consensus regarding optimal exercise types, intensity, frequency, and duration poses challenges for standardizing recommendations across diverse PD patients.

Safety and feasibility of exercise in PD

Extensive evidence from systematic reviews and meta-analyses confirms that exercise demonstrates a favorable safety profile and high feasibility in PD patients5,40,41. Across 48 randomized trials, no serious adverse events were reported in studies monitoring safety, while non-serious events (e.g., transient musculoskeletal pain, occasional falls, or dizziness) occurred infrequently and were rarely directly attributable to interventions. Reinforced by a network meta-analysis of 85 studies—where adverse events were documented in only 28 trials (predominantly non-serious falls and pain) without any serious events—the collective data reveal consistently minimal risks. Feasibility is further substantiated by low dropout rates in exercise groups (8%) comparable to non-exercising controls (11%), coupled with high adherence rates averaging 91% of planned sessions completed. Nevertheless, while exercise confers significant therapeutic benefits, potential risks, including injuries or falls, underscore the critical need for professional supervision and individualized programming. Successful long-term implementation requires careful consideration of patient-specific factors such as disease stage, adherence, accessibility, comorbidities, financial constraints, motivational barriers, and fluctuating symptom severity. Consequently, optimizing outcomes necessitates collaborative development of patient-centered regimens through coordinated efforts of neurologists (for medical profiling), physiotherapists (for movement competency assessment), and families (for psychosocial support and adherence monitoring). This multidisciplinary approach ensures risk mitigation while maximizing therapeutic efficacy, positioning exercise as both safe and clinically viable despite current GRADE evidence certainty remaining very low—representing a paradigm shift in PD management that balances robust benefits with manageable risks through personalized supervision.

Exercise-induced neurotrophic factors in PD

Recent studies found that three types of exercise—aerobic, anaerobic, and resistance exercise—had a positive impact on brain plasticity and cognitive function, but the mechanisms seemed to vary between different training styles. The effects of aerobic and anaerobic exercise are mainly mediated by changes in the expression of BDNF, lactate, vascular endothelial growth factor (VEGF), and several other proteins in the brain6. Resistance exercise, however, affects brain plasticity through myogenic factors such as irisin, insulin-like growth factor-1 (IGF-1), and BDNF42. These are involved in tissue-brain crosstalk to mediate the positive benefits of exercise. An interesting study combining voluntary running and blueberry juice (BBJ) intake found that the combination significantly reduced substantia-striatal DA neurodegeneration compared with running alone, possibly through differential modulation of GDNF43. BBJ supplementation alone failed to replicate these beneficial effects, suggesting that polyphenols may potentiate exercise-induced neuroprotection. This observation provides critical insight: combining exercise with complementary oxidative stress modulators may amplify therapeutic outcomes, potentially explaining conflicting efficacy reports across exercise interventions due to undetected confounding variables. Also, chronic aerobic exercise, like treadmill exercise, has been shown to increase GDNF levels in PD models13. However, clinical evidence regarding exercise-induced alterations in circulating GDNF levels remains limited.

Clinical evidence indicates that serum IGF-1 concentrations are elevated during the initial phase of PD, possibly reflecting a compensatory neuroprotective response, but progressively decline with disease progression44. An 8-week multimodal exercise intervention (combining aerobic, resistance, and balance/dual-task training) induced a marginal increase in serum IGF-1 concentrations in PD patients45. As members of the transforming growth factor-β (TGF-β) superfamily, BMPs drive osteoblast differentiation and bone formation. Exercise upregulates BMP production46, with BMP2 exhibiting dose-dependent neurorestorative effects: low doses promote dopaminergic neuron differentiation while high doses induce glial lineage commitment, suggesting therapeutic potential for PD47. Furthermore, BMP2 enhances neurite outgrowth in 6-OHDA- or MPP⁺-treated SH-SY5Y cells and primary dopaminergic neurons48. Smad transcription factors (Smad1/5/8), key effectors of BMPs signaling, regulate central nervous system development, and their knockdown leads to neurodevelopmental abnormalities49. BMP6/BMP2 and TGF-β2 regulate dopaminergic synapse development in nigrostriatal and midbrain limbic neurons, respectively, through the activation of Smad1 and Smad2 pathways. Both pathways are critical for dopaminergic signaling and function50.

BDNF, expressed throughout the central nervous system, is essential for neuronal survival, adult hippocampal neurogenesis, and neural plasticity51,52. Exercise enhances neuroplasticity in PD patients, increasing cortical motor excitation, gray matter volume, and BDNF expression53. Both exercise intensity and cumulative exercise volume demonstrated positive correlations with alterations in BDNF levels54. A recent study indicated that voluntary wheel-running promotes DA release through BDNF expression and provides mechanistic insight into the beneficial effects of exercise in PD55. Acute exercise-induced stress increases BDNF levels in peripheral blood monocytes, highlighting the interplay between the nervous and immune systems, with exercise-induced immune responses protecting the nervous system56. Physical exercise induces BDNF expression in the motor cortex of adult mice and reverses early-life stress, reducing susceptibility to neuropsychiatric disorders57,58. In the LPS-induced PD mouse model, four weeks of running exercise restored BDNF-TrkB signaling, preventing DA neuron loss and motor impairment, while blocking BDNF signaling eliminated this protective effect59. Emerging exercise methods such as rotational walking exercise (RWE) improve MPTP-induced motor deficits and neurodegeneration by activating AMPK phosphorylation and inducing BDNF expression in the subventricular zone, subgranular zone, substantia nigra, and striatum60.

Exercise-induced anti-inflammatory effects in PD

Exercise mediates significant downregulation of glial fibrillary acidic protein (GFAP) and complement component 3 (C3) expression, along with reduced IL-1β and TNF-α mRNA levels61,62. Specifically, low-intensity rotational exercise attenuates hippocampal C3 protein expression—a biomarker of pro-inflammatory astrocytes63. C-reactive protein (CRP), an acute-phase protein produced by the liver, is a marker of inflammation. A comprehensive study compared CRP levels in peripheral blood and cerebrospinal fluid (CSF) between PD patients and healthy controls, revealing significantly elevated CRP in both the peripheral circulation and CSF of PD patients relative to matched controls64. Furthermore, continuous aerobic exercise upregulates hippocampal irisin in post-stroke depression models, concurrently suppressing NF-κB activation and reducing NLRP3/caspase-1/IL-1β inflammasome signaling, thereby inhibiting neuroinflammatory cascades65.

The benefits of Tai Chi in PD involve peripheral cytokine responses, further emphasizing the role of inflammation in PD pathology66. One study explored the effects of long-term Tai Chi training on PD motor symptoms and reported that it improved neuroplasticity, reduced inflammation, and positively influenced amino acid and neurotransmitter metabolism67. Specifically, Tai Chi reduces L-malic acid and 3-phosphoglyceric acid levels, increases adenosine, and enhances pathways such as the urea cycle, tricarboxylic acid cycle, and β-oxidation of very long-chain fatty acids, suggesting its potential as a holistic therapy for PD67. Similarly, the disease-alleviating effects of RWE stem not only from increased neurotrophic factors but also from the inhibition of microglial activation, the reduction of pro-inflammatory markers, and increases in the anti-inflammatory cytokines IL-10 and TGF-β in MPTP-induced mouse models. RWE further reduces α-syn phosphorylation by suppressing matrix metalloproteinase-3 and p-GSK3β expression68. Endurance exercise also has anti-neuroinflammatory effects on MPTP-induced PD mouse models by reducing α-syn, pro-inflammatory cytokines, and Toll-like receptor (TLR) signaling activation, thereby attenuating motor impairments and neuronal cell death69. Notably, studies demonstrate that high-intensity exercise reverses early synaptic plasticity impairments induced by α-syn aggregates and curbs the spread of toxic α-syn species to vulnerable brain regions70. Treadmill exercise could effectively alleviate neuronal damage via inhibition of NLRP3 inflammasome via downregulation of TLR4/MyD88/NF-κB pathway in MPTP-induced PD mouse model71. Further research reveals that physical activity attenuates α-syn-triggered neuroinflammation and synaptic loss via the CD55-complement pathway8. This implies that complement may be a key element through which exercise suppresses microglial synapse elimination and slows the PD progression. Arachidonic acid lipoxygenase-12 (ALOX12), a key member of the arachidonic acid lipoxygenase family, has been identified as crucial for ferroptosis in PD. Xu et al. elucidated the molecular mechanism by which physical exercise attenuates microglial ferroptosis via the SLC7A11/ALOX12 axis72.

Exercise regulates the mitochondrial oxidative stress pathway in PD

Mitochondria are the central hub for cellular respiration, oxidative phosphorylation, and apoptosis regulation. Mitochondrial dysfunction is intricately linked to PD pathology73. PINK1 and parkin mutations act in a shared pathway to regulate mitochondrial function and trigger mitophagy74. Mitochondrial dysfunction manifests through damage to electron transport chain complex I, increased oxidative stress, disruptions in mitochondrial quality control, and impaired cellular energy metabolism75. DJ-1, a key gene involved in cellular stress responses and antioxidant protection, is also recognized as a causative gene in PD. Notably, exercise in enriched environments upregulates DJ-1 expression, enhancing cognitive and motor function. This beneficial effect occurs through pathways involving the receptor for advanced glycation end products (RAGE) and TLR signaling, indicating that targeting DJ-1 could offer protective benefits for the nervous system and ameliorate non-motor PD symptoms76,77.

Exercise serves as a potent intervention to enhance mitochondrial biogenesis and optimize mitochondrial structure and function. By increasing mitochondrial respiratory chain activity, exercise boosts cellular energy production, reduces oxidative stress, and limits reactive oxygen species generation, ultimately improving cellular health78. Exercise-induced improvements in mitochondrial function help restore the integrity of the dopaminergic system and enhance the overall health of neurons in the brain15,16. While α-syn monomers increase ATP synthase efficiency under normal conditions, pathological α-syn aggregation in mitochondria induces selective oxidation of ATP synthase and lipid peroxidation, leading to mitochondrial swelling, permeability transition pore opening, and cell death79. The evidence consistently supports exercise as a tool to restore mitochondrial function.

Treadmill training enhances mitochondrial function in both muscle and brain tissues, mitigating dopaminergic neurodegeneration80. In PD models, treadmill training elevates expression of Sirtuin 1 (SIRT1), resulting in increased mitochondrial biogenesis and reduced oxidative stress81,82. Similarly, Sirtuin 3 (SIRT3) confers anti-aging benefits by regulating energy metabolism and mitochondrial biogenesis, reducing dopaminergic neuron vulnerability. Regular exercise upregulates SIRT3 expression in mammals83,84. Supporting this, treadmill running in aged rat models significantly increased levels of IGF-1, SIRT1, SIRT3, and VEGF compared to sedentary controls. This upregulation modulates renin-angiotensin system activity in the substantia nigra, helping to inhibit oxidative stress and inflammation, protect dopaminergic neurons, and mitigate PD risk84. In an MPTP-induced PD mouse model, treadmill exercise upregulated the mitochondrial import machinery proteins TOM-40, TOM-20, and TIM-23 while reducing α-syn expression85. Furthermore, treadmill exercise maintains mitochondrial network integrity by increasing the expression of antiapoptotic proteins (MCL-1 and BCL-2), reducing the expression of pro-apoptotic markers (e.g., apoptosis-inducing factor), and upregulating the expression of autophagy-related proteins, with the involvement of SIRT186,87.

Clinical studies demonstrate that high-intensity exercise training significantly improves motor function in patients with moderate-to-advanced PD. Mechanistically, these improvements are associated with skeletal muscle adaptations—including fiber hypertrophy and transitions toward fatigue-resistant fiber types. Enhanced mitochondrial function within sarcolemmal and myogenic regions contributed to these improvements88. Thus, mitochondrial dysfunction is a pivotal factor in PD pathology, and exercise provides neuroprotective benefits by restoring mitochondrial function, reducing oxidative stress, and inhibiting α-syn aggregation72,89,90,91. This mechanistic understanding underscores the importance of incorporating exercise as a therapeutic strategy for PD management.

Exercise-produced irisin for the improvement of PD

Irisin, a myokine proteolytically cleaved from fibronectin type III domain-containing protein 5 (FNDC5), functions as a pivotal exercise mediator regulating metabolic homeostasis, inflammatory responses, and neurodegenerative processes. Beyond its musculoskeletal actions, irisin demonstrates multi-system protective effects spanning cardiovascular, metabolic, and neurological disorders92. Critically, irisin is associated with inflammation, mitochondrial dysfunction, and α-syn aggregation, underscoring its therapeutic potential as a brain-muscle crosstalk mediator92,93,94.

Irisin levels decrease with age, contributing to the progression of aging-related diseases95. It mitigates age-related pathologies through coordinated modulation of cellular energy metabolism, enhancement of proteostasis via autophagy optimization, promotion of mitochondrial quality control, reduction of reactive oxygen species (ROS), and attenuation of inflammatory processes96. These mechanisms also address skeletal diseases such as osteoporosis97. Irisin has been detected in human plasma and CSF, with levels modifiable through exercise or exogenous administration. This suggests potential blood-brain barrier (BBB) penetration and does not exclude the possibility of central in situ synthesis95,98,99. Research indicates that locally produced irisin in the brain exerts significant neuroprotective effects through multiple mechanisms, notably upregulating BDNF expression, reducing neuroinflammation, and counteracting oxidative stress100. The primary source of cerebral irisin remains unclear—whether derived from centrally expressed FNDC5 precursor or peripherally originated blood-borne irisin. Research assessing paired CSF-plasma samples from healthy individuals suggests that irisin may cross the BBB via a saturable transport system. Mechanistically, integrin αVβ5 receptors on endothelial cells are proposed as key mediators for this BBB crosstalk101, while extracellular vesicles (EVs) may facilitate transport102. Nevertheless, the specific mechanisms, receptor interactions, and signaling pathways require further elucidation.

The neuroprotective potentials of irisin in PD are multifaceted. First, it suppresses pathological α-syn aggregation and phosphorylation in cortical neurons. In α-syn PFFs models, irisin ameliorates motor deficits and neurodegeneration by enhancing endolysosomal degradation90. Clinically, plasma irisin levels inversely correlate with α-syn burden and UPDRS-scores while positively associating with cognitive function in PD patients, suggesting its therapeutic potential103. Mechanistically, irisin enhances mitochondrial biogenesis and respiratory capacity while reducing oxidative stress via Akt/ERK1/2 signaling104. Furthermore, irisin upregulates nuclear factor erythroid 2-related factor 2 (Nrf2) and enhances mitochondrial integrity by restoring balanced expression of fusion/fission proteins89. Exercise-induced PGC-1α activates FNDC5/irisin release, forming a regulatory axis that inhibits NF-κB signaling, improves mitochondrial function, and prevents α-syn neurotoxicity89,105. Additionally, irisin attenuates neuroinflammation by downregulating microglia-mediated neuroinflammation101,106. Experimental models of PD demonstrated that exogenous irisin administration alleviated α-syn PFFs-induced NLRP3 inflammasome activation107. Finally, exercise-induced irisin upregulates hippocampal BDNF expression, promoting neurogenesis and synaptic plasticity108. The positive correlation between FNDC5/irisin and BDNF levels in hippocampal tissue and serum confirms irisin’s role as a key mediator of exercise-induced neuroplasticity109,110,111,112.

Quantitative mass spectrometry confirms that exercise induces irisin production in humans and mice, with levels positively correlating with improved balance function89,113,114. Resistance training, aerobic exercise, and high-intensity interval training effectively elevate irisin levels and ameliorate PD pathology115. Notably, an observational cross-sectional study demonstrated circadian rhythmicity in human circulating irisin, peaking at 21:00116. The time-dependent effects of resistance training observed in human studies were further validated in murine models: compared to FNDC5-knockout groups, wild-type mice exercising during the late circadian phase (endogenous irisin peak) exhibited significantly enhanced muscle mass and metabolic health. Mechanistically, the core clock protein BMAL1 cooperates with the exercise-responsive factor PGC-1α4 to regulate FNDC5/irisin transcriptional rhythms, mediating downstream signaling through muscle αV integrin receptors117. This suggests that optimizing exercise timing may maximize irisin-mediated benefits for PD patients. Furthermore, exercise-induced irisin in mice reduces obesity and insulin resistance, implying its potential anti-PD effects through shared mechanisms bridging diabetes and neurodegeneration113,118. Also, recent studies have unveiled the potential neuroprotective mechanisms of irisin, which may modulate the gut-muscle-brain axis119 and enhance neurogenesis and synaptic plasticity in the hippocampus120. Although numerous studies suggest irisin’s potential neuroprotective effects in PD, current evidence remains largely confined to animal models, with the critical translational gap to human PD pathology inadequately addressed.

Other molecular mechanisms

Treadmill exercise has been shown to increase the expression of synaptophysin, postsynaptic density protein 95, tyrosine hydroxylase, and dendritic spines in nigrostriatal dopaminergic neurons in PD animal models121. Exercise also enhances cognitive and motor function in rotenone-induced PD models by reducing inflammatory cytokines and inhibiting adenosine 2A receptor activity122. In addition, treadmill exercise was shown to reverse the increased expression of Cav1.3 and p-calmodulin-dependent protein kinase (CaMK) II (Thr286) in the striatum, inhibiting glutamatergic hyperactivity and alleviating Parkinsonian symptoms123,124. Exercise also impacts autophagy and apoptosis regulation in PD. In striatal neurons, regular aerobic exercise increases the expression of Ca2+/CAMKIIα, activates the CAMK signaling pathway, and promotes autophagy marker expression, maintaining striatal apoptosis and autophagic homeostasis125. Additionally, exercise improves insulin sensitivity and has been shown to reduce insulin resistance, a common factor in neurodegenerative diseases and diabetes126. Since PD is closely related to insulin resistance, the effects of exercise under both conditions may involve common molecular pathways, offering insight into potential therapeutic strategies for both PD and metabolic disorders126. A retrospective analysis of human studies reveals that long-term combined endurance and resistance training mitigates neurodegeneration by downregulating EV-transported miRNAs (e.g., miR-142-3p, miR-23a)127. Nevertheless, the precise molecular mechanisms and dose-response effects of exercise parameters (intensity/duration) remain inadequately characterized.

Exercise-based therapies targeting molecular mechanisms in PD

There are still no effective treatments for PD to prevent its pathological progression. Research has focused primarily on protecting the remaining dopaminergic neurons by delivering neurotrophic factors to the SN or striatum. Growth differentiation factor 5 (GDF5) has emerged as a potential PD disease-modifying therapy128. However, producing and maintaining the stability of these neurotrophic factors, especially BDNF, remains a significant challenge. To investigate the molecular mechanisms of neurotrophic factors, BDNF-luciferase transgenic mice have been developed, allowing in vivo monitoring of BDNF levels via luminescent signals after d-luciferin injection. Although promising for real-time tracking, this technique still requires refinement to effectively trace BDNF within the brain129. Notably, the cystine knot in the BDNF structure, which is essential for its biological function, is difficult to preserve during industrial and pharmaceutical production130. Although BDNF shows promise in preclinical studies, challenges related to delivery across the BBB persist. Currently, the BDNF administration requires either direct injection, viral constructs, implanted protein-secreting cells, or active transport methods into the brain. Small-molecule agonists targeting BDNF receptors offer another potential route, although they also present known and unknown challenges131.

An investigational therapeutic approach involves using exogenous irisin to mimic exercise benefits, particularly when physical activity is infeasible. While skeletal muscle-derived irisin can cross the BBB, its transport pathways and efficiency remain uncharacterized90. Crucially, irisin may exert neuroprotective effects without BBB penetration: by reducing free radical production, alleviating oxidative stress, and preserving BBB integrity during neurodegenerative cognitive impairment, thereby ameliorating PD pathology132,133. These effects are partially mediated through the modulation of matrix metalloproteinase-9 expression and activity99,134. Notably, endurance exercise, resistance exercise, and irisin administration each elicit proteomic changes associated with the beneficial effects of exercise. Exogenous irisin administration promoted the expression of proteins associated with exercise-induced benefits, indicating its potential as a substitute for physical activity in PD patients unable to exercise135. Additionally, irisin has been shown to combat age-related metabolic diseases; improve muscle strength, mass, and function; and reduce insulin resistance136. Collectively, irisin likely exerts multifaceted effects, and molecular therapies utilizing peripherally administered irisin represent a viable approach for PD management. However, key translational challenges—including irisin’s pharmacokinetic profile, potential long-term side effects of modulation, and the absence of clinical trials in PD patients—require systematic clarification and warrant further investigation.

Although the molecular targets (Table 1) reviewed herein (e.g., irisin, BDNF, VEGF, DJ-1) have demonstrated neuroprotective and pathology-clearing potential in human studies, their clinical translation remains at a critical bottleneck. Future research should prioritize three frontiers: (1) developing targeted delivery systems to overcome blood-brain barrier constraints, (2) establishing biomarker-guided personalized dosing paradigms, and (3) validating long-term efficacy through multimodal clinical trials. Only by bridging the “bench-to-bedside” gap can these mechanistic discoveries evolve into tangible therapies that enhance the quality of life for PD patients.

Table 1 Molecular targets of exercise-induced neuroprotection in Parkinson’s disease: mechanisms of action, modulation profiles, and cross-target interactions
Full size table

Future perspectives and translational imperatives

Chronic aerobic exercise induces systemic adaptations by modulating multi-pathway mechanisms underlying PD pathogenesis and progression, with BDNF, GDNF, VEGF, and irisin identified as exercise-responsive biomarkers potentially mediating neuroprotection and symptom alleviation. Critical research gaps persist, necessitating: (1) objective activity monitoring to strengthen epidemiological links between physical activity and reduced PD risk; (2) urgent investigation into exercise’s dynamic effects on pathological biomarkers (α-synuclein, NfL) in prodromal/early PD and their disease progression relationships; and (3) validation of exercise as a core non-pharmacological intervention through multicenter trials with precise progression endpoints. Key priorities include comparative efficacy studies of exercise modalities (aerobic/resistance/balance) with dose optimization, integrated mechanistic investigations combining neuroimaging (PET/fMRI) and neuroplasticity biomarkers, symptom-stratified rehabilitation approaches, cost-effectiveness analyses acknowledging exercise-medication synergy, digital adherence solutions bridging the research-to-real-world gap, and subgroup-specific protocols for early-onset PD. Concurrent clinical implementation requires: initiating exercise with pharmacotherapy at diagnosis, preventive interventions for high-risk populations (REM sleep behavior disorder/genetic carriers), and dynamic prescription systems integrated through multidisciplinary collaboration. Circadian rhythms regulate various physiological processes, including the expression of FNDC5 and the secretion of irisin. Studies have shown that irisin levels exhibit diurnal variation, with higher circulating concentrations typically observed during the early active phase of the day. This rhythmic fluctuation may influence the efficacy of exercise interventions, as the timing of physical activity could modulate FNDC5 expression and downstream neuroprotective effects. Therefore, considering the time of day when prescribing exercise may optimize irisin-mediated benefits in PD. Exercise ameliorates PD via synergistic modulation of neuroprotective factor induction, neuroinflammation regulation, mitochondrial optimization, and enhanced neuronal resilience. These molecular insights are derived from preclinical models, and their translation into meaningful clinical outcomes remains an ongoing challenge. Thus, clinical translation must prioritize overcoming barriers, including human-specific dosing for exercise-mimetic therapies, pharmacokinetic challenges (BBB penetration, long-term safety), and developing individualized prescriptions for PD subtypes and comorbidities to improve feasibility.