Mechanisms of Muscle Denervation in Aging: Insights from a Mouse Model of Amyotrophic Lateral Sclerosis

: Muscle denervation at the neuromuscular junction (NMJ) is thought to be a contributing factor in age-related muscle weakness. Therefore, understanding the mechanisms that modulate NMJ innervation is a key to developing therapies to combat age-related muscle weakness affecting the elderly. Two mouse models, one lacking the Cu/Zn superoxide dismutase (SOD1) gene and another harboring the transgenic mutant human SOD1 gene, display progressive changes at the NMJ, including muscle endplate fragmentation, nerve terminal sprouting, and denervation. These changes at the NMJ share many of the common features observed in the NMJs of aged mice. In this review, research findings demonstrating the effects of PGC-1α, IGF-1, GDNF, MyoD, myogenin, and miR-206 on NMJ innervation patterns in the G93A SOD1 mice will be highlighted in the context of age-related muscle denervation.

Many retrospective and prospective studies have shown that age-related muscle weakness (dynapenia) is associated with decreased muscle function and increased risk for mortality in the elderly [1][2][3][4]. Dynapenia is a main cause of muscular instability, contributing to falls and subsequent fractures, ultimately impacting quality of life [5,6]. Both diet and exercise are practical interventions demonstrated to delay many devastating and debilitating age-associated pathologies including type 2 diabetes, cardiovascular disease, cerebrovascular disease, cognitive dysfunction, and even some cancers [7][8][9][10][11][12][13][14]. However, physical activity is the most effective means for increasing the quality of life in the elderly [15,16] and gait speed is a strong predictor of survival rate in elderly subjects [17]. Dynapenia represents a dramatic and inevitable decline in skeletal muscle function, which inevitably poses challenges for implementing a physical activity program to benefit the quality of life in the elderly. It is thought that complex degeneration of the neuromuscular system contributes to dynapenia [18][19][20][21][22].
It has been proposed that age-associated changes occur in both the nervous system and the muscular system, leading to an overall loss in muscular strength [23]. In humans, a significant decrease in the number of motor units and muscle strength is observed with aging [24]. Studies using rodent models of aging show extensive alteration in NMJ morphology. Findings from the Balice-Gordon lab demonstrated significant instability in the NMJ of aged rats using a longitudinal visualization technique for monitoring the NMJ in vivo [25]. Examination of 12-to 20-month-old rats demonstrated a progressive loss of synaptic areas and fragmentation of muscle endplates, leading to significant losses in pre-and post-synaptic structures over time [26]. Subsequent studies in mice have shown similar NMJ denervation with aging [27][28][29].

Muscle denervation in a mouse model of oxidative stress
One of the leading theories on mechanisms underlying age-related muscle denervation points to oxidative stress [30][31][32]. Reactive oxygen species (ROS) are natural byproducts of mitochondrial activity involved in respiration and energy production. ROS-mediated oxidative damages to DNAs, proteins, and lipids are normally kept in check by antioxidants. However, excessive ROS production can overwhelm the antioxidant defense, leading to increased oxidative damage of cellular machinery. Cu/Zn superoxide dismutase (SOD1) is a cytoplasmic antioxidant enzyme involved in the scavenging of superoxide free radicals. Mice lacking SOD1 enzyme (SOD1-/-mice) show increased oxidative damages to proteins, lipids, and DNAs [33]. In addition, these mice display progressive muscle denervation, weakness, and loss; changes seen despite the absence of spinal cord motor neuron and ventral root axon loss [34][35][36][37][38]. NMJ denervation and sprouting are observed in these mice between one and four months of age and precedes muscle loss [35,36], which is observed between three and four months of age [33]. Furthermore, muscle denervation and loss are greater in the gastrocnemius and tiabialis anterior compared to the soleus [33,35,36].
The tibialis anterior and gastrocnemius muscles have higher proportion of fast muscle fibers, whereas the soleus muscle has higher proportion of slow muscle fibers. Therefore, fast muscle seems to be more vulnerable to muscle denervation and loss in these animals. These changes are similar to what has been shown in human aging studies. It has been shown that type II (fast) muscle fibers are more affected than type I (slow) fibers during age-related muscle deterioration and loss in elderly adults [23,39]. A recent study by Chai and colleagues in older mice also showed age-related muscle denervation that was more pronounced in fast muscles [27]. However, findings by Valdez and colleagues did not demonstrate this fiber type-dependent sensitivity to age-related muscle denervation in older mice [28]. It is worth noting that their data indicate that NMJs in the soleus muscle as a whole seem to be less affected than NMJs in gastrocnemius, extensor digitorum longus (EDL), and tibialis anterior muscles [28] .

Mutant SOD1 transgenic mouse model of fALS
Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder affecting the upper and lower motor neurons. One of the first identified gene mutations associated with familial ALS (fALS) are those in the SOD1 gene [40]. This discovery led to the development of the G93A mutant SOD1 transgenic mouse model for fALS (G93A SOD1 mice) [41].
G93A SOD1 mice recapitulate many of the pathological hallmarks of ALS, such as progressive muscle weakness and denervation, motor neuron loss and paralysis [42][43][44]. It has been demonstrated that muscle denervation is observed as early as 47 days of age in these mice [45][46][47][48] and precedes both motor neuron loss [42,43] and muscle atrophy [49,50]. These characteristics are similar to those observed in the rodent models of aging [27,28] as well as in SOD1-/-mice [35,36,51]. In addition, these mice display preferential denervation of fast-twitch muscles [52][53][54][55]. It has been suggested that both fast muscle and motor units are selectively vulnerable to the disease process in ALS patients [56,57]. This increased sensitivity to muscle denervation in fast muscles has also been demonstrated in aged mice [27,28] and SOD1-/mice [35,36].
A potential mechanism mediating muscle denervation in the G93A SOD1 mice may be increased oxidative stress resulting from transgenic expression of mutant SOD1 gene. Examination of the gastrocnemius muscle showed increases in malondialdehyde and protein carbonyl levels which are indicative of oxidative damage to lipids and proteins, respectively [58]. In a related study, Muller and colleagues compared ROS production in response to sciatic nerve transection in muscles of three different models of mice that show muscle denervation: SOD1-/-, G93A SOD1, and aged mice [59]. Examination of gastrocnemius and tibialis anterior muscles following nerve injury showed varying levels of increase in ROS. The authors reported three-fold increase in ROS in 28-to 32-month-old wildtype mice compared to 10-month-old wildtype mice (aged vs. young wildtype mice) [59]. In SOD1-/-mice, ROS production is increased by 30% and 100% in 5-month-old and 20-month-old mice, respectively, compared with age-matched wildtype mice [59]. Lastly, G93A SOD1 mice showed 10-fold increase in ROS generation compared to age-matched wildtype mice [59].
These findings suggest that many features of the muscle denervation observed in the G93A SOD1 mice are similar to features of muscle denervation observed in aged mice and SOD1-/-mice, and may be mediated by oxidative stress. Since G93A SOD1 mice represent the most frequently studied fALS mouse model, there have been a large number of studies examining the physiological mechanisms modulating the disease progression in these mice. In the following sections, research findings demonstrating the effects of PGC-1α, IGF-1, GDNF, MyoD, myogenin, and miR-206 on NMJ innervation patterns in the G93A SOD1 mice will be highlighted in the context of age-related muscle denervation.

Role of PGC-1α as a mediator of exercise and caloric restriction effects in NMJ maintenance
The effect of voluntary exercise and caloric restriction (CR) on age-related muscle denervation has been directly examined by Valdez and colleagues [29]. In their study, they examined mice that were either subjected to 20 months of caloric restriction (from 4 to 24 months of age) or 1 month of voluntary wheel running (from 21 to 22 months of age) [29]. Their findings show that many of the age-related changes (e.g., fragmentation, denervation, and sprouting) in NMJs are reduced in response to the exercise and diet regimen used in the study [29].
Although the effects of exercise and CR on the disease progression in G93A SOD1 have been examined, the effects on muscle denervation phenotype have not been directly examined to date. A number of studies in G93A SOD1 mice have examined the role of physical activity on the survival rate and neuroprotection of spinal cord motor neurons. Kaspar and colleagues showed that wheel running exercise increases spinal cord motor neuron survival and improves Rota-rod performance [60]. Treadmill running in these mice have also shown benefits on motor neuron survival and Rota-rod performance [61]. However, additional studies seem to suggest that the intensity of the exercise may be an important factor in moderating the disease progression; moderate exercise may impart benefit while high intensity exercise hastens disease progression in G93A SOD1 mice [61,62]. On the other hand, caloric restriction studies to date in G93A SOD1 mice have only shown a detrimental effect on disease progression and motor function [63][64][65].
Peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) is a transcription coactivator involved in both mitochondrial biogenesis and oxidative metabolism in skeletal muscle [66]. PGC1-α overexpression in muscle increases mitochondrial biogenesis and enhancement of muscle oxidative phenotype [67]. Research findings suggest that the effects of caloric restriction may be in part mediated by PGC-1α [68] and AMP-activated protein kinase α (AMPK) [69]. The critical role of AMPK and PGC-1α in mediating exercise effects has also been demonstrated by Narkar and colleagues, where they showed that pharmacological activation of AMPK and PGC-1α imparts an endurance phenotype in sedentary mice [70].
Two of the currently approved antidiabetic drugs, metoformin and pioglitzone, have been shown to activate AMPK and PGC-1α, respectively [71][72][73][74]. Studies examining the effect of metformin demonstrated that the treatment using multiple doses in drinking water (0.5mg/ml, 2mg/ml, and 5mg/ml) did not improve the survival rate or disease progresssion in G93A SOD1 mice, but did show a moderate increase in the motor unit number in EDL of male mice [75]. On the other hand, G93A SOD1 mice treated with pioglitzone, a PPARagonist, showed improved motor function, muscle fiber integrity, and motor neuron survival [76]. This finding has been confirmed using a genetic approach; transgenic overexpression of PGC-1α improved Rota-rod performance and increased motor neuron survival in G93A SOD1 mice [77,78]. Interestingly, despite the motor function improvement in these mice, Liang and colleagues did not observe an increase in AChRα subunit expression [78].
This result is surprising since regulation of NMJ genes and AChR clustering by PGC-1α has been demonstrated using muscle-specific PGC-1α transgenic and knockout mice [79]. Furthermore, findings suggest that muscle activity or state of innervation may be a better predictor of the PGC-1α level. Active older adults (70±5 years) who maintained greater muscle strength than sedentary older individuals (63±10 years) displayed much higher PGC-1α levels in their muscle, comparable to levels observed in the muscle of young individuals (22±2 years) [80]. In rats, chronic unilateral sciatic or peroneal nerve denervation reduces the level of PGC-1α by 70% in ipsilateral muscles [81]. Experimental findings showing decreased PGC-1α in the muscles of both elderly humans and aged rats may reflect age-related muscle denervation [82,83].
Despite these findings, studies targeting the transgenic PGC-1α expression to skeletal muscle in the G93A SOD1 mice suggest that PGC-1α is not sufficient for halting muscle denervation in these mice. A study by Da Cruz and colleagues demonstrated that transgenic overexpression of PGC-1α in the muscle of G37R mutant SOD1 mice did not modulate NMJ denervation in these mice, even though PGC-1α overexpression in the muscle increased mitochondrial biogenesis, AChR clustering, and improved muscle endurance [84]. A similar study in G93A SOD1 mice resulted in enhanced oxidative phenotype in the skeletal muscle, but did not improve Rota-rod performance or muscle strength [85].

Role of IGF-1 in NMJ maintenance
Insulin-like growth factor-1 (IGF-1) is a neuroprotective hormone that has also shown to mediate exercise effect on neurogenesis and neural activity [86][87][88]. Kaspar and colleagues examined the effect of IGF-1's neurotrophic property in ALS using the G93A SOD1 mice. They utilized retrograde uptake property of adeno-associated virus (AAV) at presynaptic terminals to deliver IGF-1 to the motor neurons via intramuscular injections of AAV-IGF-1 [89]. Intramuscular injections of AAV-IGF-1 in adult G93A SOD1 improved both the muscle function and motor neuron survival [89]. These identical changes were also observed in exercised G93A SOD1 mice [60].

Aging and Disease • Volume 6, Number 5, October 2015 383
Intramuscular injections of AAV-IGF-1 increased IGF-1 expression in both the muscle and spinal cord motor neurons [89]. When IGF-1 expression was restricted to the muscle using a lentiviral vector that was not retrogradely transported to the spinal cord, Kaspar and colleagues observed in these mice a moderate but statistically significant increase in survival that was less than that for AAV-IGF-1 treatment [89].
A number of studies have addressed the question of central versus local action of IGF-1 on the survival of G93A SOD1 mice. It has been shown that continuous intrathecal infusion of IGF-1 into the spinal cord is neuroprotective and improves survival rate [90]. However, in another study where the IGF-1 level in the spinal cord was elevated via intrathecal delivery of IGF-1:tetanus toxin fragment C fusion protein, there was no effect on the survival rate [91]. Similarly, spinal cordtargeted transgenic expression of IGF-1 in G93A SOD1 mice did not improve motor neuron degeneration and muscle function decline [92].
It has been demonstrated that muscle-targeted expression of IGF-1 improves muscle function and reduces age-related muscle atrophy [93], and enhances nerve branching at NMJ and increases muscle endplate size in older mice (22-24 month old), resulting in measures that are similar to those observed in young mice (2-6 month old) [94]. Muscle expression of IGF-1 also enhances nerve regeneration following nerve injury in mice, as demonstrated by increased markers of muscle repair and nerve regeneration as well as accelerated recovery of nerve conduction in the damaged nerve [95].
Dobrowolny and colleagues directly examined the effect of muscle-directed expression of muscle isoform of IGF-1 (mIGF-1) on NMJ changes in G93A mutant SOD1 mice [96]. The findings demonstrate that skeletal muscle expression of mIGF-1 extends survival and protects motor neurons in G93A SOD1 mice [96]. In addition, mIGF-1 expression increased agrin expression and preserved AChR clustering [96]. However, in another study, muscle-directed expression of IGF-1 in G93A SOD1 mice did not improve motor neuron survival or motor function [92].
These equivocal findings are surprising given the fact that positive effects of localized IGF-1 on nerve injurymediated muscle denervation have been demonstrated. Caroni and colleagues showed that intramuscular injections of IGF-1 increased GAP-43, which suggests axon growth, but did not affect acetylcholine receptor (AChR) clustering [97]. In a follow-up study, the authors blocked the IGF-1 activity in the muscle using IGF-1 binding proteins [98]. In this study, they induced paralysis using Botulinum toxin A to induce nerve sprouting. Interference with IGF-1 action through subcutaneous delivery of IGF-BP4 to the paralyzed muscle using osmotic minipump prevents nerve sprouting, suggesting that IGF-1 mediates nerve sprouting at the NMJ [98]. However, a double-blind placebo-controlled phase III randomized clinical trial has proven that a two-year treatment of twice-daily subcutaneous injections of IGF-1 is clinically ineffective in ALS patients [99].

Role of GDNF in NMJ maintenance
Glial cell line-derived neurotrophic factor (GDNF) is also a neurotrophic factor found to have a strong pro-survival effect on motor neurons [100]. GDNF mRNA is detectable in the central nervous system as well as in the peripheral organs, including skeletal muscle [101][102][103]. The role of GDNF in NMJ innervation was demonstrated by studies showing that skeletal muscle-targeted expression of GDNF in transgenic mice increases the number of muscle of endplates and multiply-innervated NMJs [104,105]. However, when subcutaneously injected the exogenous GDNF's ability to maintain synaptic remodeling in mice is observed only if chronic exposure is initiated within the developmentally-critical postnatal time window between postnatal day zero and eight [106].
It has been shown that the GDNF level in the postnatal muscle decreases with age. In rats, the GDNF mRNA level in the skeletal muscle is reduced three months after birth; however, constant high levels of GDNF protein are detected in postnatal muscle [107]. Examination of muscle samples from human adults 20-43 years of age also demonstrated the presence of GDNF in the skeletal muscle, and its expression in the vicinity of neuromuscular junctions as indicated by co-localization of GDNF and α-bungarotoxin staining [103]. Research evidence from adult rats suggests that GDNF activity in muscle may modulate NMJ morphology. It has been shown that when rats are subjected to two weeks of involuntary exercise, the GDNF level is increased in soleus but decreased in EDL [108]. Examination of muscle endplates revealed a decrease in the total area per endplate in EDL while an increase was observed in the soleus [108]. This correlation between GDNF levels and muscle endplate sizes has been further documented in subsequent studies [109,110].
In G93A SOD1 mice, intramuscular injections of AAV expressing GDNF in adult mice improved Rota-rod performance and spinal cord motor neuron survival [111]. The effect observed in this study is likely mediated by both central and local mechanisms, since GDNF levels were upregulated in both the muscle and spinal cord [111]. Although immunoreactivity for transgene-derived GDNF was shown to be highly co-localized with αbungarotoxin staining, the effect on muscle denervation was not directly assessed in this study [111].
Aging and Disease • Volume 6, Number 5, October 2015 384 In order to directly test the role of central versus local effect of GDNF on the neuropathology, Li and colleagues transgenically targeted GDNF expression to either neurons or myofibers in G93A SOD1 mice [112]. Direct measures of muscle denervation demonstrated that transgenic GDNF expression targeted to the muscle increased the number of innervated NMJs, whereas neuronally targeted transgenic GDNF expression did not produce such an effect [112]. Therefore, muscle-directed transgenic expression of GDNF is important for NMJ maintenance in these mice.
To determine the therapeutic potential of GDNF treatment, Suzuki and colleagues have used human stem cells to examine the effect of chronic GDNF delivered to either muscle or spinal cord on the muscle denervation phenotype in the adult rat model of fALS (G93 SOD1 rats) [113,114]. In one study, GDNF-secreting human neural progenitor cells were intrathecally transplanted into the spinal cord at 70 days of age. Although the transplantation imparted local protection on spinal cord motor neurons, there was no change in the number of innervated NMJs [114]. In another study, the group implanted human mesenchymal stem cells engineered to express GDNF (hMSC-GDNF) intramuscularly at 80 days age, with a goal of delivering GDNF to NMJs locally. Although their findings showed that transplantations of either hMSC-GDNF or hMSC expressing GFP increased the number of innervated NMJs, only the hMSC-GDNF transplant showed statistically significant increase when compared to nontransplanted G93A SOD1 rats [113]. These results suggest that in addition to the GDNF effect, the transplanted hMSCs themselves are able to modulate muscle denervation by providing additional trophic support to the muscle and NMJs.

Role of MyoD, myogenin, and miR-206 in NMJ maintenance
The findings from various growth factor studies using the G93A SOD1 mice and G93A SOD1 rats suggest that muscle may indeed play a role in mediating muscle denervation process. MyoD and its related basic helixloop-helix (Myf5, myogenin, and MRF4) are muscle specific regulatory factors. During development, these myogenic regulatory factors have critical roles in skeletal muscle development [115][116][117]; For example, MyoD is involved in determining skeletal muscle lineage [118], whereas myogenin is important for muscle differentiation [119,120]. In adult muscle, myogenin is expressed preferentially in slow muscle (composed of predominantly type I muscle fibers) [121][122][123] and MyoD is enriched in fast muscle (composed of predominantly type II muscle fibers) [121,123].
A number of studies suggest that fast muscle is more sensitive to age-related muscle denervation process as compared to slow muscle [27,39]. Similarly, it has also been suggested that the fast muscles and motor units in both ALS patients and mouse models are preferentially denervated [52,[54][55][56][57]. It has been shown that expression of either exogenous MyoD or myogenin in skeletal muscle can shift the muscle phenotype. MyoD expression has been shown to impart type IIx muscle fiber phenotype under denervation condition [124], while myogenin expression has been shown to enhance the oxidative phenotype in muscle fibers [122,123].
Recently, my colleagues and I used adenovirus constructs to express either human MyoD or myogenin in muscles of adult G93A SOD1 mice to evaluate their effects on disease progression [85]. Adenovirus expressing either human MyoD, human myogenin, or GFP was injected into hind limb muscles bilaterally at 30 days of age. Treatment effects on the NMJ innervation patterns were examined using stereology on gastrocnemius samples taken at 100 days of age. Our findings show that postnatal muscle expression of MyoD exacerbates both muscle denervation and Rota-rod performance decline, while postnatal muscle expression of myogenin attenuates muscle denervation and improves Rota-rod performance when compared to the virus control G93A SOD1 group [85]. Therefore, MyoD and myogenin differentially modulate muscle denervation in these mice.
It has been shown that myogenic transcription factors can regulate the expression of microRNAs (miRs) in muscle [125,126]. miRs are short non-coding mRNAs that are post-transcriptional regulators of gene expression. Of the many miRs that have been identified in the muscle, it has been shown that miR-206 can be regulated by both MyoD and myogenin [125,[127][128][129]. miR-206 shows skeletal muscle-specific expression and is thought to regulate many aspects of skeletal muscle function, including NMJ formation [126]. Although its expression has been reported to be enriched in slow muscle [127,130], its upregulation following muscle denervation is enhanced in fast muscles [127]. Although NMJ development and innervation is unaffected in miR-206 knockout mice, NMJ reinnervation is delayed following nerve transection or injury [127]. In the absence of miR-206, muscle denervation is exacerbated in the skeletal muscle of G93A SOD1 mice [127]. Upon further investigation, the authors determined that the effect in their study is in part mediated via repression of histone deacetylase 4 (HDAC4) translation by miR-206. The HDAC4 protein level was increased in miR-206 knockout mice, and when HDAC4 is selectively knocked out in the skeletal muscle of wildtype mice, NMJ reinnervation following nerve injury is enhanced [127]. However, the Aging and Disease • Volume 6, Number 5, October 2015 385 effect of HDAC4 knockout on muscle denervation in G93A SOD1 mice was not directly investigated.

Conclusion
The similarities in muscle denervation profiles between SOD1-/-, G93A SOD1, and aged mice suggest that both SOD1-/-and G93A SOD1 mice represent an accelerated muscle-aging model for oxidative stress-mediated muscle denervation. Much of the past studies in G93A SOD1 mice have focused on neuroprotective effects on spinal cord motor neurons. However, investigation of effects on muscle denervation in G93A SOD1 increasingly provides valuable insight into potential mechanisms of progressive muscle denervation. Recent findings have highlighted muscle-specific factors such as MyoD, myogenin, and miR-206 as mediators of muscle denervation. Furthermore, evidence of muscle-targeted genetic manipulations demonstrating greater impact on muscle denervation highlights the importance of the muscle in NMJ innervation regulation.