Introduction
Spinal and bulbar muscular atrophy (SBMA) is an X-linked motor neuron disease characterized by late-onset degeneration of lower motor neurons and skeletal muscle atrophy [
41]. In addition, SBMA patients can develop non-motor symptoms, such as sexual dysfunctions, diabetes, and severe urinary tract dysfunction, which are related to androgen insensitivity [
75]. SBMA is caused by the expansion of a polyglutamine tract in the gene encoding AR [
43]. Polyglutamine expansions in specific genes are responsible for eight other neurodegenerative diseases, namely Huntington’s disease (HD), dentatorubral–pallidoluysian atrophy, and spinocerebellar ataxia (SCA) type 1, 2, 3, 6, 7, and 17 [
67]. A unique feature of SBMA in the family of polyglutamine diseases is sex specificity: full manifestations are restricted to males. Sex specificity is due to the conversion of polyglutamine-expanded AR to a toxic species that occurs upon binding to its natural ligand, testosterone. In its inactive state, AR localizes to cytosol. Binding of AR to testosterone results in nuclear translocation, DNA binding, interaction with transcriptional co-regulators, and regulation (activation and repression) of expression of androgen-responsive genes; these post-translational events have all been associated with pathogenesis [
71]. Although several experimental and clinical strategies to modify disease have been tested to date [
81], no effective therapy for SBMA has yet been developed.
Skeletal muscle (dys)function is emerging as a key component of SBMA pathogenesis [
84]. In addition to selective degeneration of motor neurons, SBMA patients develop progressive skeletal muscle weakness, fasciculations, and atrophy. SBMA patients show signs of muscle denervation, such as nerve sprouting, muscle fiber atrophy, and fiber-type grouping, together with signs of muscle degeneration, such as splitting, presence of central nuclei, and degeneration of fibers [
96]. Neurogenic and myopathic processes have both been described in mouse models of SBMA [
72,
84]. Notably, knock-in SBMA mice show muscle pathology that precedes spinal cord pathology [
106]. Consistent with the idea that muscle plays a key role in pathogenesis, targeted overexpression of insulin-like growth factor 1 (IGF-1) selectively in muscle ameliorated disease manifestations in transgenic mice [
70]. Unequivocal genetic evidence for a primary role of muscle in SBMA pathogenesis was recently provided by the observation that expression of polyglutamine-expanded AR in all tissues except skeletal muscle was sufficient to prevent disease manifestations in transgenic mice [
17]. Supporting these observations were concurrent studies demonstrating that knockdown of polyglutamine-expanded AR expression only in peripheral tissues by antisense oligonucleotides was sufficient to attenuate disease manifestations in SBMA knock-in mice [
45]. However, how polyglutamine-expanded AR causes muscle atrophy remains to be clarified.
Alterations of metabolic pathways and of the nutrient and energy status at the organism and cell levels are emerging as key players in neurodegenerative diseases [
11]. Skeletal muscle is a central nutrient- and stress-sensor and has a primary role in age-related neurodegenerative diseases [
24]. Skeletal muscle regulates systemic aging by influencing metabolism and responses of the organism to dietary restriction and oxidative stress; as such, it affects both life span and the progression of age-related diseases [
3,
25]. The homeostasis of skeletal muscle is maintained by a fine balance between anabolic and catabolic processes, which co-exist in a dynamic equilibrium to regulate the rate of protein synthesis and degradation (i.e., protein turnover) as well as muscle fiber size. Muscle hypertrophy is characterized by an increase in muscle mass, which results from increased protein synthesis rate and muscle fiber size. On the other hand, muscle atrophy is characterized by reduced muscle mass, which results from reduced fiber size, increased protein degradation, and decreased force production and fatigue resistance. This equilibrium is maintained by several signaling pathways, including the evolutionarily conserved IGF-1/Akt pathway [
91]. Based on the central role of skeletal muscle in energy metabolism and the fact that muscle is an easier therapeutic target compared to the nervous system for neurodegenerative diseases, here we investigated pathological processes occurring in SBMA muscles.
Discussion
That primary cell-autonomous degenerative processes occur in SBMA muscle is supported by multiple lines of evidence [
84], including the observations that abnormalities were detected in cultured myotubes isolated from the quadriceps of SBMA patients [
52], modulation of expression of non-expanded as well as polyglutamine-expanded AR in muscle modifies disease [
17,
59,
77], and intervention to reduce the accumulation of polyglutamine-expanded AR selectively in muscle is beneficial in SBMA mice [
70]. What is less obvious is how expression of polyglutamine-expanded AR causes damage to muscle, and how this in turn contributes to the metabolic alterations found in SBMA patients [
75]. Skeletal muscle plays a central role in the regulation of glucose and fatty acid metabolism [
25]. Skeletal muscle metabolism is dependent on fiber-type composition, with slow-twitch oxidative fibers using lipids as their main source of energy, and fast-twitch glycolytic fibers using glucose. We observed metabolic changes characterized by a glycolytic-to-oxidative switch in muscle in both knock-in mice and SBMA patients. In SBMA knock-in mice, this fiber-type switch preceded atrophy. A similar glycolytic-to-oxidative shift in fiber-type composition was overt in the quadriceps muscle of transgenic mice ubiquitously expressing polyglutamine-expanded AR, and it was attenuated by knockdown of mutant AR solely in muscle [
45]. A glycolytic-to-oxidative fiber-type switch has also been observed in AR knock-out mice, suggesting that polyglutamine expansion causes this phenomenon in muscle mainly through a loss of function mechanism [
50]. Notably, in the knock-in SBMA mice the degree of atrophy of glycolytic fibers was higher than that of oxidative fibers. Because similar changes in muscle homeostasis occur during aging [
39], these observations suggest that SBMA muscle may undergo premature aging, and that a genetic interaction between aging and expression of polyglutamine-expanded AR may contribute to SBMA pathogenesis. In addition to the loss of function, there is evidence that also gain of native AR function contributes to SBMA pathogenesis [
63,
89]. Indeed, overexpression of non-expanded AR solely in muscle caused severe muscle atrophy and premature death [
59], and overexpression of polyglutamine-expanded AR in either skeletal muscle or neurons reduced the size of fast-glycolytic fibers and induced an oxidative-to-glycolytic fiber-type shift in extensor digitorum longus [
77]. Interestingly, the levels of expression of polyglutamine-expanded AR correlated with the degree of atrophy, supporting the idea that expression of polyglutamine-expanded AR in muscle exerts cell-autonomous, dose-dependent toxic effects in muscle. Moreover, our findings suggest that dysregulation of androgen signaling in muscle alters muscle and body metabolism, thereby contributing to SBMA pathogenesis.
Our lipidomic analysis revealed increased levels of lysophosphatidylcholines, lysophosphatidylethanolamines, ceramides, diglycerides, and polyunsaturated fatty acids, and decreased levels of species enriched in polyunsaturated lipids and phosphatidylglycerol. By microarray analysis, we observed augmented expression of genes involved in lipid biosynthesis and lipolysis. Consistent with the observation that glycolytic muscles undergo a shift towards oxidative metabolism, these findings suggest that lipid synthesis and turnover are enhanced in SBMA glycolytic, but not oxidative muscles. Another interesting feature that characterizes SBMA muscle is the reduction of expression and activity of glycolytic enzymes, which also occurs during aging [
4]. Notably, changes in expression of glycolytic genes preceded those in lipid genes, suggesting that altered glycolysis contributes to the shift towards oxidative phosphorylation. This finding raises the question of whether polyglutamine expansion in AR directly hampers the expression of key glycolytic genes, such as
HkII and
Pfkfb, which are direct target of AR [
60]. Moreover, the expression of the glucose transporter GLUT3 was decreased in the muscle of AR113Q mice. GLUT3 has a higher affinity for glucose and at least a fivefold greater transport capacity than other glucose transporter family members, such as GLUT1, GLUT2 and GLUT4. The decrease in GLUT3 expression is consistent with the glycolytic-to-oxidative fiber-type switch detected in SBMA muscle and suggests that glucose uptake is defective in SBMA glycolytic muscles. Interestingly, the expression of GAPDH was decreased at late stages of disease also in soleus, which does not develop atrophy and does not show signs of denervation. It is possible that polyglutamine expansion in the AR directly alters the expression of glycolytic genes, which is detrimental in muscles that are mostly composed of glycolytic fibers, such as quadriceps and gastrocnemius. This effect also occurs to a lower extent in muscles that are mostly composed of oxidative fibers, but in this case with limited consequences on muscle metabolism. Further investigation is required to address whether altered glycolysis underlies or contributes to the metabolic shift in glycolytic muscles toward oxidative phosphorylation.
Glycolytic-to-oxidative fiber-type switch has also been reported in amyotrophic lateral sclerosis (ALS) [
68]. Similar to SBMA, glycolytic muscles are more severely affected in transgenic mouse models of superoxide dismutase 1 (SOD1)-linked ALS [
68,
74] and in ALS patients [
26], and this is likely due to selective vulnerability of fast-fatiguable motor neurons [
74]. Notably, the phenotype of mutant SOD1-expressing mice is different from that of SBMA mice. Indeed, while mutant SOD1-expressing mice develop paralysis of posterior limbs, SBMA mice do not develop paralysis even at late stages of disease, indicating that the degenerating motor neurons maintain their ability to control muscle contraction to some extent. In SBMA muscle, upregulation of denervation markers was detected after metabolic changes, indicating that, in contrast to ALS mice, in SBMA mice these alterations precede denervation. Importantly, the motor neuron number and morphology of neuromuscular junctions are preserved in knock-in SBMA mice [
40], further supporting that the metabolic changes detected in SBMA glycolytic muscles are not induced by denervation at early stages of disease.
SBMA muscle was characterized by increased rates of protein synthesis and degradation in glycolytic muscles. Protein synthesis was induced by enhanced activation of mTOR signaling, which occurred selectively in atrophic muscles, was an early event, and was also observed in the muscle of SBMA patients. Dysregulation of mTOR signaling is emerging as a key component of neurodegenerative disease pathogenesis. mTOR signaling is upregulated in the muscle of an HD mouse model [
95] and in the cortex of mice overexpressing TAR DNA-binding protein 43 (TDP43) [
102]; in contrast, it is downregulated in the muscle of a mouse model of valosin-containing protein-associated inclusion body myopathy [
14], in the striatum of HD patients and mice [
44], and in the spinal cord of transgenic mice expressing mutant SOD1 [
109]. mTOR activation can be elicited in response to denervation and excessive accumulation of amino acids released from increased protein breakdown. However, in SBMA muscle mTOR activation preceded denervation. It remains to be elucidated why mTOR is upregulated in SBMA muscle. We noticed that mTOR activation correlated with the physiological increase of serum testosterone levels that occurs at approximately 30 days of age in mice. This observation suggests that activation of AR is required for mTOR induction, and the crosstalk existing between androgen and mTOR signaling may lead to enhancement of mTOR function and muscle hypertrophy in physiological conditions [
5]. In SBMA, polyglutamine expansion in the AR may result in aberrant and sustained activation of mTOR that in turn can contribute to muscle atrophy during disease progression. Prolonged activation of mTORC1 causes a late-onset myopathy characterized by altered autophagy [
12]. As a matter of fact, autophagy was induced in SBMA muscle and altered at very late stages of disease. Although further experimental evidence is required to assess the role of mTOR in SBMA muscle, it is possible that enhancement of mTOR signaling at early stages of disease is compensatory to the increased need of protein synthesis and breakdown occurring in response to the glycolytic-to-oxidative metabolic switch in muscle, but prolonged activation of mTOR may turn to be detrimental at later stages of disease. mTOR forms two complexes, mTOR complex 1 (mTORC1), which regulates cell metabolism, protein synthesis and autophagy, and mTOR complex 2 (mTORC2), which mainly controls cell proliferation and survival. In SBMA muscle, mTORC1 is activated very early, before induction of denervation, and it increases the rate of new protein synthesis. mTORC1 is sensitive to rapamycin. Rapamycin exerts beneficial effects in several neurodegenerative diseases. In a pharmacological model of Parkinson’s disease (PD), rapamycin treatment decreased dopaminergic neuron loss by inhibiting the expression of mTORC1-induced pro-apoptotic genes [
51], and by activating autophagy [
22]. Induction of autophagy and clearance of aggregates underlie the beneficial effects of rapamycin in mouse models of Alzheimer’s disease (AD) [
10,
98], HD [
80], SCA3 [
55], and PD [
19], and in mice overexpressing TDP43 modeling frontotemporal dementia [
102]. However, rapamycin worsened the phenotype of transgenic mice expressing either mutant SOD1 [
109] or mutant valosin-containing protein (VCP) [
14]. Rapamycin treatment resulted in exacerbation of body weight loss in SBMA knock-in mice, similar to what has been observed in other pathological conditions (reviewed by [
53]). It is noteworthy that prolonged treatment with rapamycin inhibits mTORC2 and Akt in myoblast cells [
86], and as shown here in SBMA muscle. Because Akt-mediated phosphorylation of polyglutamine-expanded AR promotes degradation and reduces toxicity, it is possible that rapamycin-induced deterioration of phenotype in SBMA mice results from Akt inactivation in muscle [
69,
70,
89].
mTOR controls cell metabolism through regulation of PGC1α function [
20]. PGC1α is upregulated selectively in SBMA glycolytic muscles. Different from SBMA, PGC1α is downregulated in the muscle of mutant SOD1-linked ALS mice [
68], and its overexpression selectively in muscle attenuated motor dysfunction, even if it had no effect on survival [
21]. While in ALS muscle the glycolytic-to-oxidative fiber-type switch is independent of PGC1α [
68], our results show that in SBMA muscle PGC1α is progressively upregulated selectively in atrophic muscles following disease progression, thereby suggesting a role for this nuclear receptor co-regulator in SBMA pathogenesis. Several transcriptional co-regulators that interact with AR and other nuclear receptors play an important role in controlling muscle fiber-type composition. For instance, muscle-specific ablation of nuclear receptor co-activators, such as Sox6 and Mediator 1, or co-repressors, such as RIP140, lead to a glycolytic-to-oxidative fiber-type switch [
13,
47,
76,
93]. The upregulation of PGC1α detected in SBMA muscle was decreased by treatment of the mice with rapamycin, indicating that PGC1α is induced by mTOR. Muscle-specific overexpression of PGC1α leads to a glycolytic-to-oxidative fiber-type switch [
47], whereas loss of PGC1α function has the opposite effect [
37]. Moreover, PGC1α controls the expression of genes involved in oxidative phosphorylation [
61]. These observations suggest a model whereby polyglutamine-expanded AR leads to mTOR activation, which in turn leads to PGC1α induction and glycolytic-to-oxidative fiber-type switch. PGC1α levels and mTOR activation can be modulated by a diet enriched with lipids [
6,
97]. Importantly, feeding SBMA mice an HFD decreased mTOR signaling activation, reduced the expression of PGC1α as well as of genes involved in oxidative phosphorylation, and attenuated the glycolytic-to-oxidative metabolic fiber-type switch.
PGC1α promotes mitochondrial biogenesis [
104]. Consistent with the glycolytic-to-oxidative fiber-type switch and the upregulation of PGC1α, genes involved in mitochondrial dynamics were induced, and the activity of citrate synthase and the mitochondrial respiratory complexes was increased in SBMA muscle. Mitochondrial homeostasis is altered in SBMA [
79], and mitochondrial membrane potential was decreased by oligomycin in glycolytic SBMA muscles. While the HFD did not affect mitochondrial complex activity, it decreased mitochondrial membrane depolarization induced by oligomycin, indicating that the HFD improves mitochondrial quality control. Mitochondrial dysfunction is often associated with induction of autophagy [
90]. Autophagy is induced in the muscle of SBMA knock-in mice [
16,
108], and our results show that autophagy flux is altered at late stages of disease. Notably, TFEB is induced in SBMA muscle, leading to upregulation of expression of several autophagy target genes [
16]. In neurons, TFEB has been shown to work as a co-factor of AR, whose function is diminished by polyglutamine expansion [
18]. It is possible that in muscle polyglutamine expansion in AR leads to a gain of TFEB function, resulting in enhanced autophagy. Although TFEB is negatively regulated by mTOR [
94], we detected concurrent mTOR and TFEB activation in SBMA glycolytic, but not oxidative muscles, providing evidence that in disease conditions the two signaling pathways can be simultaneously activated. Importantly, the HFD reduced the expression of TFEB, linking TFEB to metabolism. Interestingly, an HFD attenuated autophagy defects and muscle atrophy caused by muscle-specific knock-out of histone deacetylases I and II, supporting the idea that HFD rescues autophagy alterations and muscle degeneration [
62]. Consistent with this observation, the HFD decreased the levels of LC3I while increasing the levels of LC3II, suggesting a positive effect on autophagosome formation and autophagy flux.
Progressive muscle atrophy and body weight loss are prognostic factors for several neurodegenerative diseases, such as AD [
54], HD [
85], PD [
1,
7], and ALS [
27]. Body weight loss can result from reduced energy intake, increased energy expenditure, or both. Weight loss cannot be explained by decreased food intake in these conditions [
85,
101]. Rather, neurodegenerative disorders are often associated with a hypermetabolic state [
28,
58], suggesting that altered energy homeostasis contributes to pathogenesis [
73]. These systemic metabolic deficits are characterized by altered lipid (cholesterol and fatty acid) metabolism, as reported in HD [
9], and AD [
29,
46]. Interestingly, male ALS patients have hypolipidemia [
105], and decreased levels of low-density lipoprotein have been shown to correlate with respiratory impairment [
15]. High levels of serum cholesterol and triglycerides correlate with longer survival, further highlighting the relevance of lipid metabolism in ALS [
32]. Moreover, epidemiological studies showed that a higher body fat content is associated with a lower risk for ALS [
36,
48], whereas a high-carbohydrate/low-fat diet is associated with an increased risk for ALS [
66]. A beneficial effect of high-calorie diet has been proven in patients suffering from HD and ALS [
31,
85,
101,
103], as well as in mice expressing either mutant SOD1 [
33] or mutant VCP [
49]. Notably, an HFD has been shown to stabilize body weight loss in ALS patients [
31]. Recently, in a double-blind, placebo-controlled clinical trial, a high-carbohydrate diet, but not an HFD, ameliorated some aspects of disease in ALS patients [
103]. In this trial, treatment was started after about 20 % body weight loss, and it remains to be established whether an HFD can be beneficial if treatment is started at earlier stages of disease. Different from ALS patients, in SBMA patients the average serum lipid levels are on the upper range to normal, and low-density lipoprotein levels are elevated in some patients [
23,
34,
75]. In knock-in SBMA mice, we did not detect serum lipid alterations. Feeding normal mice an HFD results in development of metabolic syndrome characterized by glucose intolerance, insulin resistance and obesity [
35]. AR113Q mice neither developed signs of glucose intolerance and insulin resistance nor they became obese when fed the HFD. Moreover, the HFD increased the weight of fat pad in AR113Q mice to a lesser extent compared to control mice. These results support the idea that SBMA mice require a higher caloric intake compared to control mice. Indeed, skeletal muscle represents 30 % of body weight, and a metabolic change in muscle towards oxidative phosphorylation is expected to impact body metabolism with increased need for fatty acids. Therefore, the effect of chronic exposure to HFD on metabolism leads to development of metabolic syndrome in control mice, but not SBMA mice, in which fatty acids are used by glycolytic muscles that have undergone a shift towards oxidative metabolism. However, before this approach can be translated to clinic, it is important to consider that an HFD may have undesired side effects in patients with serum hyperlipidemia.
Our results provide proof-of-principle that an HFD ameliorates muscle pathology, improves motor function, and extends life span of a mouse model of SBMA. The HFD ameliorated SBMA mouse phenotype by possibly affecting several tissues and metabolic pathways. In addition to muscle, the effect of diet may also come from adipose tissue. AR is highly expressed in adipose tissue, and the effect of polyglutamine expansion in AR on this peripheral tissue remains to be addressed. Our findings suggest an unprecedented role for adipose tissue in SBMA. Notably, androgen signaling mediated by the AR inhibits adipocyte differentiation and promotes lipolysis [
65]. Adipose tissue controls whole body insulin sensitivity. Interestingly, pan-tissue knock-out of AR in mouse results in increased fat pad mass with normal insulin sensitivity and reduced body weight [
78]. On the other hand, adipose tissue-specific knock-out of AR resulted in hyperinsulinemia in the absence of obesity, and when fed an HFD, the adipose tissue-specific AR knock-out mice developed obesity, hyperglycemia, and impaired insulin secretion. Another important aspect concerning adipose tissue and metabolism is the interplay between androgens and adipokine signaling [
65]. High molecular weight adiponectin is expressed in white adipose tissue, increases insulin sensitivity, and is downregulated in obesity. Androgens inhibit the release of high molecular weight adiponectin, and castration or AR knock-out increased the serum levels of adiponectin. Further analysis will establish whether polyglutamine expansion in the AR alters adipokine signaling in SBMA. Our findings are consistent with a model whereby polyglutamine expansion in AR, in addition to conferring toxic gain of functions, leads to loss of AR function in muscle and possibly other tissues, such as adipose tissue. Some of these aspects were attenuated by the HFD, suggesting that this as a novel strategy in support to SBMA patients to modify disease onset, progression, and outcome.