Aerobic-Strength Exercise Improves Metabolism and Clinical State in Parkinson’s Disease Patients

Regular exercise ameliorates motor symptoms in Parkinson’s disease (PD). Here, we aimed to provide evidence that exercise brings additional benefits to the whole-body metabolism and skeletal muscle molecular and functional characteristics, which might help to explain exercise-induced improvements in the clinical state. 3-months supervised endurance/strength training was performed in early/mid-stage PD patients and age/gender-matched individuals (n = 11/11). The effects of exercise on resting energy expenditure (REE), glucose metabolism, adiposity, and muscle energy metabolism (31P-MRS) were evaluated and compared to non-exercising PD patients. Two muscle biopsies were taken to determine intervention-induced changes in fiber type, mitochondrial content, and expression of genes related to muscle energy metabolism, as well as proliferative and regenerative capacity. Exercise improved the clinical disability score (MDS-UPDRS), bradykinesia, balance, walking speed, REE, and glucose metabolism and increased muscle expression of energy sensors (AMPK). However, the exercise-induced increase in muscle mass/strength, mitochondrial content, type II fiber size, and postexercise phosphocreatine (PCr) recovery (31P-MRS) were found only in controls. Nevertheless, MDS-UPDRS was associated with muscle AMPK and mechano-growth factor (MGF) expression. Improvements in fasting glycemia were positively associated with muscle function and the expression of Sirt1 and Cox7a1, and the parameters of fitness/strength were positively associated with the expression of MyHC2, MyHC7, and MGF. Moreover, reduced bradykinesia was associated with better muscle metabolism (maximal oxidative capacity and postexercise PCr recovery; 31P-MRS). Exercise training improved the clinical state in early/mid-stage Parkinson’s disease patients, including motor functions and whole-body metabolism. Although the adaptive response to exercise in PD was different from that of controls, exercise-induced improvements in the PD clinical state were associated with specific adaptive changes in muscle functional, metabolic, and molecular characteristics. Clinical Trial Registration www.ClinicalTrials.gov, identifier NCT02253732.

inTrODUcTiOn Parkinson's disease (PD) is a chronic neurodegenerative disorder that affects ~1% of the population >60 years of age (1). The clinical profile includes a variety of motor and non-motor symptoms, with increased risk of falls and a dramatic impact on the quality of life and functional independence (2,3). The progressive nature of the disease and generally symptomatic treatment require new strategies in early stage disease management (4). Mounting evidence shows that physical activity has unequivocal benefits for PD patients (5)(6)(7), with a potential to lower the disability score (MDS-UPDRS) (8). Regular exercise has the potential to improve underlying metabolic derangements, including inflammation, mitochondrial dysfunction, and glucose metabolism. A higher incidence of glucose intolerance and type 2 diabetes (>50%) was found among PD patients (9)(10)(11) and the presence of glucose intolerance has been shown to accelerate the progression of PD (12), with a sedentary lifestyle considered one of the common denominators of neurodegeneration and metabolic dysfunction (13). Skeletal muscle is the organ of motion and the largest organ in our body, corresponding to 35-40% of the whole-body mass. Muscle is a major player in the whole-body energy metabolism, with the ability to communicate with other cells, tissues, and organs to maintain functional integrity and energy homeostasis. Improving muscle functional state in PD by regular exercise could, therefore, improve the whole-body functional capacity, slowing down disease progression. It is known that the dopaminergic deficit, central to the pathophysiology of PD, leads to increased tonic inhibition of the thalamus, thus reducing the excitatory drive for the motor cortex (14). This, in turn, may affect the cortical activation of muscles (15,16), which may be manifested as muscle weakness. Such a reduction in the force generated during muscle contraction is indicative of strength and/or movement speed. There is a proposed relationship between muscle weakness and bradykinesia (17,18), one of the primary motor symptoms of PD (19).
In this work, muscle 31 P-MRS, as well as measures of balance and muscle power, were used to evaluate the exercise-related changes in muscle metabolic and functional state. When combined with the assessment of muscle fiber type, mitochondrial content, and the expression of key metabolic genes, they provided a comprehensive view of the exercise-induced adaptive changes that are associated with improvements of whole-body metabolism and motor disability in Parkinson's disease patients.
In our study, we evaluated effects of 3-month supervised aerobic-strength training intervention on the whole-body and muscle metabolism, clinical disabilities, physical fitness and muscle functional, morphological and molecular characteristics in patients with PD, and age/gender/BMI-matched controls.

MaTerials anD MeThODs
The study population consisted of 13 sedentary seniors and 12 sedentary patients with Parkinson's disease (duration: 7.1 ± 3.9 years, Hoehn-Yahr 1-3). All PD patients received standard care from a neurologist and were on appropriate PD medication (l-DOPA/carbidopa, dopamine agonists, MAO inhibitors). All volunteers underwent medical examination, including blood tests, complex metabolic phenotyping, aerobic physical fitness and muscle strength assessments, nutritional profiling, free-living physical activity assessment, and motoric/balance testing before/ after training intervention. The capacity to undergo intervention was assessed/approved by a cardiologist, and patients with known uncontrolled or late-stage cardiac, renal, liver, oncologic, or other chronic diseases were excluded. Parkinson's disease patients and age/gender/BMI-matched controls (n = 11/11) completed the 3-month-supervised aerobic-strength exercise intervention. The study population was complemented by Parkinson's disease patients who did not undergo training intervention (n = 5, disease duration 7.8 ± 4.8 years, Hoehn-Yahr 2-3, age 62.4 ± 9.8 years, M/F 4/1). The clinical study flow chart is shown in (Figure S1 in Supplementary Material).
The protocol was approved by the Ethics Committee of the University Hospital Bratislava and conformed to the ethical guidelines of the Helsinki declaration of 1964 (2000 revision). All individuals signed a written, informed consent prior to the study. The small patient sample is a major limitation of this study, which was attributable to the complex nature of the study protocol. (Omron-Healthcare Co., Japan) for a minimum of three consecutive days with at least 12 h active-time recording. A standardized acute bicycling exercise was performed before and after training, using stationary bicycles and heart rate monitoring (Polar RS300X, 40 min at 70% HRmax).

Unified Parkinson's Disease rating scale and Motor Function Testing
The severity of Parkinson's disease was evaluated by the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS). All patients underwent examination in the "ON" state, after taking appropriate medication, with the difference between the "ON" and the "OFF" state being >30% (MDS-UPDRS). Specific subscores reflecting posture and gait, rigidity, tremor, and bradykinesia were calculated (Table S1 in Supplementary Material). The Berg Balance Scale (BBS) was used to assess a relevant change in balance (23) and the risk of falls.

skeletal Muscle Biopsy
Samples of the m.vastus lateralis were obtained by Bergström needle biopsy under local anesthesia in the fasted state, before/ after the 3-month exercise program, as previously described (20). Muscle samples were immediately cleaned and frozen/stored in liquid nitrogen. A small, well-defined part of the muscle was embedded in TissueTek, frozen in 3-methylbuthane chilled by liquid nitrogen, and stored at −80°C for immunohistochemistry.

Determination of Muscle Fiber Type in native Tissue sections
Transversal 6 µm cryosections were prepared. A myofibrillar ATPase activity assay was performed following preincubation with an acid (pH~4) solution that predominantly inhibited the myosin ATPase activity in fast glycolytic/fast oxidative (type 2B/2X/2A) fibers. The method is described in Ref. (24). Fiber type-specific fiber size and relative quantity were evaluated.
Muscle Metabolism by 31

statistical analysis
Statistical analyses were performed using SAS Jump Statistics Software (USA) and G*Power software ver. 3.1.9.2. Data were tested for normal distribution. Paired t-test was used to assess the difference between baseline and postexercise variables. Unpaired t-test and general linear model were used to assess the differences between the intervention effects (delta follow-up baseline). More than two group differences were evaluated by ANOVA with the Tukey post hoc test. Results are given as means ± SD (unless indicated otherwise). Pearson correlation and a stepwise regression model were used to determine the association state between variables. Statistical significance was considered at p < 0.05, and for trends, p < 0.09.

characteristics of the study Population
Eleven sedentary PD patients (MDS-UPDRS score 42.0 ± 6.5) were compared to age/gender/BMI-matched healthy controls. Key phenotypic characteristics are summarized in Table 1.
A small group of PD patients who did not undergo exercise training intervention had a higher MDS-UPDRS disability score (52.0 ± 11.9, p = 0.045). However, obesity and metabolic phenotypes [BMI, body fat, subcutaneous, and visceral adiposity (MRI), muscle mass, HOMA-IR, 2-h glucose, circulating lipids, metabolic substrate preference (RQ), and REE] of this nonexercising control group were comparable to the pre-intervention state of PD patients who underwent exercise intervention. No physical fitness, muscle functional parameters, or muscle biopsies were obtained from non-exercising PD controls.
exercise Training and the clinical state of Patients with Parkinson's Disease A 3-month, strength-endurance training program improved the total MDS-UPDRS score ( Figure 1A), motor score (Part III), and specifically, the bradykinesia subscore ( Figures 1B,C), while no significant improvements in the MDS-UPDRS were found in non-exercising PD patients (Table S1 in Supplementary Material). Exercise benefits on motor functions were also documented by improved static and dynamic balance (BBS), as well as by a better performance in the 10 m preferred speedwalking test, chair-stand-test, and 1-mile Rockport walking test (Figures 1D-F; Table 1). The parameters of balance (BBS) were positively associated with muscle strength (R = 0.540, p = 0.0036) and VO2max (R = 0.581, p = 0.0008). Exercise training for PD patients had a stronger effect on performance in the maximal speed-walking test (p < 0.05). These exercise-induced benefits were present in PD patients despite the fact that the rate of force development was increased only in the control population ( Table 1).
exercise Training, Metabolic health, and Physical Fitness As expected, 3-month exercise training did not induce significant changes in body weight or total body fat mass. However, a trend toward a greater decline in body fat was found in a healthy control population ( Table 1). More importantly, the traininginduced increase in muscle mass and strength was higher in the control population than in PD patients, who exhibited lower muscle strength compared to controls, both before and after training ( Table 1; Figures 2A,B). Exercise significantly increased whole-body REE ( Figure 3C) and had a small but consistent lowering effect on RQ, indicating higher exercise-induced metabolic substrate preference for lipids in PD patients than in controls ( Table 1). Intervention-induced changes were absent in nonexercising PD patients (p = 0.54). Moreover, REE was positively associated with muscle mass (Figure 3D). Compared to controls, PD patients had similar fasting glycemia, but displayed reduced insulin sensitivity (HOMA-IR) in the baseline pre-exercise state ( Table 1). Exercise intervention had a greater effect on glucose metabolism in PD patients compared to controls (  Figure 2I). Exercise training decreased fasting serum lactate in both PD and healthy control populations (Table 1), and, in PD patients, it reduced serum lactate response to an acute bout of bicycling exercise ( Table 1). PD patients tended to be less physically active, with lower levels of aerobic fitness and muscle strength ( Table 1)

effects of exercise on Muscle Morphologic, Metabolic, and Molecular characteristics
There were no differences in body composition across the groups, but a small, consistent training-induced increase of muscle mass was found in all control individuals ( Table 1). Dynamic 31 P-MRS showed that τPCr (the time constant of postexercise PCr recovery) was regulated by training in the muscle of healthy controls, but not in PD patients (Table 1). Moreover, positive associations were found between τPCr and (i) the bradykinesia subscore ( Figure 2G), (ii) the performance in the 10 m preferred speed-walking test (R = 0.79; p = 0.004), (iii) the 2-h glycemia   IIb fibers with a larger diameter and their relative content was positively associated with REE (Figures 2D-F). As expected, the glycolytic (type IIb) muscle fiber cross-sectional area was positively associated with MyHC1 mRNA (Figure 4). Regular exercise increased the mRNA of AMPKα1 (PRKAA1) in the skeletal muscle of both controls and PD patients ( Table 1), while muscle mitochondrial DNA content and MyHC2/7 ratio (indicative of type II fiber prevalence) were upregulated only in the control population ( Table 1). Exercise-related improvements in the PD clinical state (MDS-UPDRS) were associated with increased muscle expression of PRKAA1 and mechano-growth factor (MGF) (Figure 4). Moreover, improvements in fasting glycemia were positively associated with muscle expression of Sirt1 and Cox7a1 (Figure 4). Muscle functional parameters were positively associated with the expression of MyHC2, MyHC7, and MGF (Figure 4).

Determinants of exercise-induced effects on Fitness and clinical state in PD
In PD patients, VO2max was associated with better performance in the chair-stand test (

DiscUssiOn
Many clinical studies point to a link between Parkinson's disease (PD), the neurodegenerative disease with salient motor system manifestations, and metabolic disease, demonstrating that >50% of patients with PD are either glucose-intolerant or have type 2 diabetes (T2D) (9,11). A prospective study with >36,000 T2D patients and >108,000 controls, with a 7.3-year follow-up, showed that the presence of T2D increases the risk of PD by 36% (27). It has been shown that PD patients with T2D have more severe neurological symptoms (28). Treatment of early-stage PD is generally symptomatic, although therapy that offers neuroprotection is likely to be beneficial. Neuroprotective strategies under study include the enhancement of mitochondrial function, insulin sensitivity, anti-inflammatory, and anti-AGE mechanisms (29). Intensive supervised endurance/strength exercise training was used as a physiological therapeutic modality, with the potential to stimulate many of the above mentioned neuroprotective mechanisms, and to ultimately improve the clinical state of patients with early and mid-stage PD. This strategy provides a unique opportunity to improve our understanding of the role of exercise in PD, especially when parallel changes in the whole-body metabolism and muscle metabolic, functional, and molecular state are being examined.  This was exemplified by a >16% decrease in the MDS-UPDRS disability score, mostly due to improvements in motor functions, specifically bradykinesia, and corroborates the results of several previous reports (30,31). Bradykinesia contributes to impaired balance and increases the risk of falls. Training-induced improvement in the Berg balance test in PD patients was similar to a previously reported improvement after a 6-month complex exercise program (32). Furthermore, PD patients displayed better performance in the 10-m preferred-pace walking test, and in walking speed on a 1-mile track (Rockport test). Others have shown that 5 weeks/10 sessions of robotic treadmill training had comparable effects in a similar patient population (30).   This was substantiated by a small but uniform increase of muscle strength in elderly controls, which was not observed in PD patients, who, in fact, were inferior to their healthy counterparts in MVC and rate of force development both before and after 3 months of intervention. Ni et al. showed that 3-month, low-load/ high-velocity resistance training improved muscle strength in PD patients (31). Combined aerobic/resistance training increased REE, which paralleled a change in muscle mass. Combined exercise caused a small but consistent increase of VO2max in all tested individuals, assessed indirectly by the Rockport test, which enabled us to determine the functional capacity of PD patients to walk. This, however, introduced variability related to differences in balance, gait efficiency, muscle strength, and pain tolerance, which could have prevented us from seeing a more robust exercise-response in VO2max. O'Leary et al. reported a 14% increase in VO2max, assessed by treadmill ergometry in obese elderly individuals, after a 3-month supervised aerobic training (33). An exercise-induced increase in cardiovascular fitness in PD patients was described (34,35). While Bergen et al. measured VO2max directly, as oxygen consumption at a peak exercise load (bicycle ergometry), we and others examined functional performance using the Rockport walking test (34)(35)(36). Exercise intervention reduced fasting and 2-h glycemia and the area under the post-load glycemic curve in PD patients. Moreover, the insulin-sensitizing effect of exercise was more pronounced in PD patients, who were more insulin-resistant prior to exercise intervention. Exercise-induced improvement in insulin resistance tended to correlate with VO2max, as well as with muscle maximal oxidative capacity (Qmax, 31 P-MRS). Associations between exercise-induced changes in glucose metabolism and VO2max were previously reported in a small population of 16 sedentary seniors (33), but information on exercise-related changes in muscle energy metabolism obtained by 31 P-MRS in PD patients is unprecedented (26,37). Exercise intervention led to consistent, progressive lowering of fasting serum lactate in both controls and a PD population, which was paralleled by increased physical fitness. It is plausible to think that trained muscles remove lactate from the systemic circulation with higher efficiency (38).

effects of exercise on Muscle Metabolism, Morphology, and Molecular characteristics
Postexercise PCr recovery (τPCr) was regulated by exercise intervention in the muscle of healthy controls, but not in PD patients. Moreover, exercise-induced changes in τPCr, as well as in Qmax (muscle maximal oxidative capacity), were positively associated with improvements in bradykinesia and negatively associated with circulating HDL-cholesterol. τPCr was also positively related to walking performance and 2-h glycemia, suggesting that the ability to replenish energy during repeated contractions could also be an important determinant of muscle motoric and metabolic state elicited by regular exercise in PD patients. Training increased the mRNA expression of AMPKα1 (PRKAA1) in the skeletal muscle of controls and PD patients. However, muscle mitochondrial DNA content, type IIa fiber size, MVC force, and muscle mass only increased in the control population, indica ting differences in the magnitude of exercise-induced adaptive changes in the muscle of PD patients. Moreover, exercise-related improvements in the PD disability score (MDS-UPDRS) were associated with the increased muscle expression of AMPKα1 and MGF, indicating the importance of both the metabolic and the mitogenic response. There is a line of evidence that dysregulation of mitochondrial function and energy metabolism are the key determinants of PD pathology, as the maintenance of mitochondrial homeostasis is crucial for neuronal development (39). The induction of PGC-1α, a regulator of mitochondrial biogenesis, via activation of PPARs (rosiglitazone, bezafibrate), or the modulation of energy metabolism by activating AMPK (AICAR, metformin, resveratrol) or Sirt1 (SRT1720, isoflavone-derived compounds), could be highlighted as a potential strategy with which to modulate mitochondrial biogenesis, with effects similar to those of exercise (39). Various exercise benefits had, in our hands, distinct gene predictors. Reduced fasting glycemia was positively associated with the expression of Sirt1 and Cox7a1 and muscle functional parameters were positively associated with the expression of MyHC2, MyHC7, and MGF, indicating the importance of mechanical sensing and fiber type shift in response to regular muscle use. Our results are in agreement with an observation that a single bout of sprint-induced mitochondrial activity was associated with the increased expression of Sirt1, the upstreamdeacetylase of PGC-1α, and the induction of Thr(172) AMPKα phosphorylation (40). We observed that type IIb and IIa fibers with an enlarged cross-sectional area represented the major morphological characteristic of muscle in our PD patients, while the size of type I fibers did not differ between the groups. This is in agreement with a higher expression of MyHC1 and MyHC2 genes (markers of type IIa and IIb fibers) and a lack of change in MyHC7 (type I fiber marker). Regular exercise increased the expression of MyHC1 in controls, but not in a PD population, with MyHC7 expression unregulated. The shifts in muscle morphology/phenotype that occur in PD may come about as a consequence of the modified pattern of motor unit activation and rigidity. Previous histochemical studies demonstrated both atrophic and hypertrophic fibers and fiber type grouping in the limb muscles of PD patients (41,42). Histological changes in muscle fiber composition appear to be muscle-specific and related to PD subtypes (41). It is plausible to think that shifts in muscle fiber composition in PD are related to duration, severity, and the specific clinical character of the disease. Our results showed that exercise lowered the type 1/ type 2 ratio in both controls and the PD group, likely resulting from either type 2 fiber hypertrophy or hyperplasia, induced by endurance/strength training. the interplay between neurodegeneration and metabolism are unexplored. A sedentary lifestyle and chronic energy overload are common denominators of neurodegeneration and metabolic impairment (43), promoting chronic inflammation and insulin resistance, both at the whole-body level and in the brain (13,44). In our study, we observed increased peripheral IR in PD patients, pointing at systemic metabolic dysfunction in PD compared to age/BMI-matched controls, which was alleviated by exercise. It can be assumed that systemic insulin resistance is mirrored by brain insulin resistance, which has been linked to compromised cholesterol synthesis and mitochondrial function, decreased brain plasticity, and apoptosis (45). Our preliminary data, indicating a positive association between serum and CSF glucose (2PD/6controls), support the notion that exercise modulates both peripheral and central glucose metabolism, which has been thought to counteract the neurodegenerative process (46).

Determinants of exercise-induced effects on Physical Fitness and the clinical state of PD
Moreover, muscle maximal oxidative capacity and/or the time needed for postexercise PCr recovery were associated with muscle functional parameters, REE, and whole-body glucose metabolism. Recent advances in MR spectroscopy have enabled us to study both muscle and brain function and metabolism in vivo, providing detailed compartmentalized information about metabolic interactions between neurons (47) and about the levels of important neurotransmitters (GABA, glutamate) (48).
We and others have shown that reduced muscle mitochondrial content/function is linked to insulin resistance and T2D (49,50). However, regular exercise is a well-recognized stimulus of mitochondrial biogenesis, not only at the level of skeletal muscle (51), but also at the systemic level (52). In this study, we have shown that exercise failed to upregulate muscle mitochondrial DNA content in PD patients compared to controls. However, an increase in the expression of AMPKα1 indicates that signaling pathways involved in mitochondrial biogenesis, as well as glucose uptake, contributed to the adaptive response to exercise (53). Further research is necessary to evaluate the impact of exercise on muscle metabolism and mitochondrial content/function in PD.
In conclusion, complex intensive, although relatively short endurance-strength exercise intervention, improved the clinical state of early/mid-stage PD patients. This was associated with improvements in muscle and whole-body metabolism, suggesting that life-long maintenance of muscle metabolic and functional capacity by regular exercise is essential to counteract the pathophysiology of PD and should become a part of the management for Parkinson's disease.

eThics sTaTeMenT
This study was carried out in accordance with the Declaration of Helsinky. All subjects provided written informed consent prior entering the study. The protocol was approved by the "Ethics Committee of Faculty of Medicine, Comenius University and Univesrity Hospital Bratislava. " aUThOr cOnTriBUTiOns