Reactive oxygen species (ROS) in ageing

A factor clearly associated with loss of function during ageing in numerous tissues is oxidative damage and experimental evidence from humans and rodents indicates that skeletal muscles and other musculoskeletal tissues show age-dependent increases in the products of oxidative damage to biomolecules including proteins, lipids and nucleic acids^{(Reference Broome, Kayani and Palomero17–Reference Mecocci, Fano and Fulle20)}. Various reports have attributed the positive correlation between age and oxidative damage to age-related changes in reactive oxygen species (ROS) production, with skeletal muscles from old mice exhibiting a higher intracellular ROS generation in comparison to muscles from young mice^{(Reference Palomero, Vasilaki and Pye21,Reference Vasilaki, Mansouri and Remmen22)} . The loss of muscle that occurs with ageing occurs in parallel with loss of motor units in both humans and rodents^{(Reference Campbell, McComas and Petito23,Reference Sheth, Iyer and Wier24)} . A 25–50% reduction in the number of motor neurons occurs in both man and rodents with ageing^{(Reference Rowan, Rygiel and Purves-Smith25,Reference Tomlinson and Irving26)} . Loss of innervation of individual fibres occurs in muscles of aged humans and animals and our study which indicated that ∼15% of muscle fibres from old mice are completely denervated and ∼80% of neuromuscular junctions (NMJs) showed some disruption^{(Reference Vasilaki, Pollock and Giakoumaki27)}. Recent data indicate that this loss of innervation may play a fundamental role in in the changes in ROS generation that occur in ageing skeletal muscle^{(Reference Pollock, Staunton and Vasilaki28,Reference Staunton, Owen and Pollock29)} providing evidence for important inter-tissue interactions affecting muscle viability during ageing^{(Reference Staunton, Owen and Hemmings30)}.

Roles of ROS in physiology of the musculoskeletal system

The term oxidative stress as it related to oxidative damage to cells and tissues was coined by Helmut Sies and colleagues in 1985^{(Reference Cadenas and Sies31)} and is defined as “a disturbance of the pro-oxidant-antioxidant environment in favour of the former”. The implications of this definition were originally that oxidative stress was potentially deleterious to tissues and cells and that inhibition or reversal of the stress on cells and tissues would generally be beneficial. This could be potentially achieved by a reduction in the promotors of oxidation (usually free radicals or ROS), or an increase in substances or pathways that decrease oxidation (antioxidant substances or regulatory proteins). These assumptions underlined many of the original studies to investigate the roles of ROS and antioxidants in skeletal muscle and in exercise^{(Reference Brady, Brady and Ullrey32–Reference Jackson, Jones and Edwards35)}. Particularly prominent in these studies was the assumption that nutritional antioxidants would be beneficial and many of the early studies included a component to examine the possibility that antioxidant supplementation would suppress effects to demonstrate the possible negative role of free radicals or ROS^{(Reference Brady, Brady and Ullrey32,Reference Dillard, Litov and Savin34)} . As further studies were undertaken, it rapidly became clear that skeletal muscle could not only generate ROS, but also that it could respond to that generation by upregulation of regulatory pathways^{(Reference Jackson, Papa and Bolanos36,Reference McArdle, Pattwell and Vasilaki37)} which prevented the potential for subsequent oxidative damage to the tissue. Thus, ROS in this situation were not necessarily damaging but inducing adaptive changes in tissues. These apparent contrasting roles of ROS have subsequently been described as redox signalling effects compared with oxidative stress and damage and described more specifically by Helmut Sies and colleagues as oxidative eustress and oxidative distress^{(Reference Sies, Berndt and Jones38)}).

Redox signalling in skeletal muscle

Signalling by ROS is mainly achieved by targeted modifications of specific residues in proteins^{(Reference Janssen-Heininger, Mossman and Heintz39,Reference Sobotta, Liou and Stocker40)} . Muscle fibres respond to contractions by an increase in the intracellular generation of superoxide and nitric oxide (NO) with the formation of secondary ROS and reactive nitrogen species^{(Reference Palomero, Pye and Kabayo41–Reference Pye, Palomero and Kabayo43)}. This leads to activation of a number of transcription factors, including NF-κB, AP-1 and HSF-1^{(Reference Jackson, Papa and Bolanos36,Reference Ji, Gomez-Cabrera and Steinhafel44–Reference Vasilaki, McArdle and Iwanejko46)} and an increased expression of regulatory enzymes and cytoprotective proteins^{(Reference McArdle, Pattwell and Vasilaki37,Reference Hollander, Lin and Scott47,Reference McArdle, Spiers and Aldemir48)} . Redox-regulation is also apparent for genes associated with catabolism^{(Reference Bar-Shai, Carmeli and Reznick49–Reference Van Gammeren, Damrauer and Jackman51)} and mitochondrial biogenesis^{(Reference Bakkar, Wang and Ladner52,Reference Irrcher, Ljubicic and Hood53)} . Identification of the specific redox mediated steps in adaptive pathways to exercise has proven difficult to define in skeletal muscle. Studies in humans and animals using very high levels of antioxidants have provided evidence that these interventions inhibited cytoprotective responses (e.g., exercise-induced increase in heat shock and other stress proteins)^{(Reference Venditti, Napolitano and Barone54)}, reduced mitochondrial biogenesis^{(Reference Ristow, Zarse and Oberbach55–Reference Gomez-Cabrera, Domenech and Romagnoli57)}, prevented an increase in muscle insulin sensitivity^{(Reference Ristow, Zarse and Oberbach55)} and inhibit the release of cytokines and inflammatory mediators^{(Reference Wuyts, Vanaudenaerde and Dupont58)}. These antioxidant supplementation studies have been controversial^{(Reference Gomez-Cabrera, Ristow and Viña59,Reference Higashida, Kim and Higuchi60)} , but additional adaptations potentially activated by ROS have been identified in genetic knockout mouse models, designed to delete ROS-generating enzymes. For instance, NADPH oxidase 2 (Nox2) knockout mice show reductions in post-exercise glucose uptake via impaired GLUT4 translocation^{(Reference Henriquez-Olguin, Renani and Arab-Ceschia61,Reference Henríquez-Olguin, Knudsen and Raun62)} , Nox4 knockout in mice was found to lead to development of insulin resistance^{(Reference Xirouchaki, Jia and McGrath63)} and specific endothelial Nox4 knockout leads to impaired metabolic adaptations to chronic exercise^{(Reference Specht, Kant and Addington64)}. Thus, together these studies indicate that the range of adaptive pathways activated during exercise and regulated by redox pathways is likely to be extensive. Key processes involved in muscle adaptations to exercise have been intensively studied for a number of years and among these, multiple pathways have been identified where redox regulation appears important including the key pathways leading to the adaptations described above^{(Reference Jackson65)}.

Is redox signalling disrupted during ageing in the musculoskeletal system?

It now seems clear that the level of ROS generation and oxidative damage is not a fundamental determinant of lifespan although some authors have argued that the age-related changes in ROS activities and oxidative damage are important mediators of age-related disorders^{(Reference Hamilton, Walsh and Van Remmen66)}. Several of the ROS-stimulated responses to exercise are attenuated in old mice including increased stress responses^{(Reference Vasilaki, McArdle and Iwanejko46)} and mitochondrial biogenesis^{(Reference Viña, Gomez-Cabrera and Borras67,Reference Cobley, Moult and Burniston68)} . Mitochondrial peroxide generation has also been repeatedly reported to be increased in skeletal muscle during ageing^{(Reference Vasilaki, Mansouri and Van Remmen69,Reference Jang and Van Remmen70)} . In order to decipher the effects of ROS in musculoskeletal ageing, a number of studies have examined the effects of deletion of regulatory enzymes for ROS in mammalian models. Despite frequent observations of increased oxidative damage in these models of dysregulated ROS homeostasis, no clear relationship with skeletal muscle ageing was seen^{(Reference Jang and Van Remmen70)}. Studies of muscle ageing in mice predominantly use the C57Bl strain of laboratory mice which reach maturity at 4–6 months of age and in many laboratory animal facilities they show age-related loss of muscle force production and loss of muscle mass from approximately 22 months of age. In the relevant studies described here, mice were examined at 6–8 months of age (adult mice) and 22–26 months of age (old mice).

Mice with a whole body deletion of SOD1 (Cu, Zn superoxide dismutase) differed from all of the other models studied and showed an increase in tissue oxidative damage associated with neuromuscular changes with ageing. This was described by the discoverer as “accelerated age-related loss of muscle mass”^{(Reference Muller, Song and Liu71,Reference Deepa, Van Remmen and Brooks72)} . Adult Sod1KO mice show a decline in skeletal muscle mass, loss of muscle fibres and a decline in the number of motor units, loss of motor function and contractility, partial denervation and mitochondrial dysfunction by 8 months old^{(Reference Jang, Lustgarten and Liu73,Reference Larkin, Davis and Sims-Robinson74)(Reference Vasilaki, van der Meulen and Larkin75)} . The fibre loss in Sod1KO mice is accompanied by degeneration of NMJs^{(Reference Jang, Lustgarten and Liu73)}. These changes are also seen in old control wild type (WT) mice, but not until after 22 months of age. These mice also show the attenuation of redox-mediated responses to contractile activity that is seen in ageing mice^{(Reference Vasilaki, McArdle and Iwanejko46)}. Hence, Sod1KO mice have been proposed as a useful model to examine the potential role of ROS in skeletal muscle ageing^{(Reference Jackson76)}. The only known function of Sod1 is to catalyse the dismutation of superoxide to hydrogen peroxide, a reaction that also occurs chemically in the absence of Sod1 but at a much slower rate^{(Reference Sakellariou, Pye and Vasilaki77)}. Superoxide also reacts rapidly with nitric oxide (NO) to generate peroxynitrite, a reaction that is approximately 3 times faster than the chemical dismutation of superoxide to hydrogen peroxide (Fig. 1). Furthermore, the muscle cytosolic concentration of NO is many fold higher than superoxide. We have demonstrated increased generation of peroxynitrite in muscles of Sod1KO mice providing a potential mechanism by which the lack of this protein specifically leads to accelerated muscle loss^{(Reference Sakellariou, Pye and Vasilaki77)}.

Fig. 1. Schematic illustrating the reactions of superoxide to generate hydrogen peroxide via Sod1-catalysed dismutation and the chemical reaction with NO to generation peroxynitrite. Data presented in Sakellariou et al., (2011) indicate increased peroxynitrite generation occurs in muscle fibres in the Sod1KO mice^{(Reference Sakellariou, Pye and Vasilaki77)}.

The Sod1KO mouse shows many of the muscle phenotypes of old WT mice at a much earlier age and the major effect of the lack of Sod1 appears to be through effects at the level of the motor neuron. This model has also been proposed as a useful experimental model of frailty since Sod1KO mice exhibit four characteristics that have been used to define human frailty: weight loss, weakness, low physical activity and exhaustion. In addition, Sod1KO mice show increased inflammation and sarcopenia, which are strongly associated with human frailty^{(Reference Deepa, Bhaskaran and Espinoza78)}. A series of tissue-specific Sod1KO mice have been generated to establish the key tissue and cellular locations at which the lack of Sod1 exerts an effect to lead to skeletal muscle loss. In recent studies, we have examined in detail the changes in motor neurons and the NMJ, which occur in inducible neuron-specific Sod1KO mice (i-mnSod1KO mice) which present with an early onset of muscle loss^{(Reference Pollock, Macpherson and Staunton79)}. Surprisingly no specific effect of a lack of neuronal Sod1 was seen, but rather all of the changes seen in ageing were accelerated. We concluded that neuronal deletion of Sod1 induced exaggerated loss of muscle in old mice and this deletion leads to a reduced axonal area, increased proportion of denervated NMJ and reduced acetylcholine receptor complexity and other changes in nerve and NMJ structure that are also seen in WT mice at a more advanced age^{(Reference Pollock, Macpherson and Staunton79)}. Thus, the Sod1KO mouse model and its tissue specific derivatives have provided a great deal of valuable information on the tissue interactions and mechanisms that lead to muscle loss in ageing. It is clear that a simple lack of Sod1 does not occur during ageing in WT animals or humans but the detailed analogies in phenotype and mechanisms seen in ageing WT mice and Sod1KO mice provide confidence in the relevance and utility of this model.

The current data therefore suggest that aberrant ROS generation and subsequent defective redox signalling and is a feature of ageing in skeletal muscle and contributes to attenuated responses to contractile activity and diminished efficacy of adaptations to contractile activity. This appears to be an important component in maintenance of muscle mass and function during ageing since studies in mice have shown that restoration of some stress responses helps maintain muscle mass and function in aged cohorts^{(Reference McArdle, Dillmann and Mestril80,Reference Kayani, Close and Dillmann81)} although comparable human studies have not been undertaken.

Potential role of skeletal muscle mitochondria in aberrant redox signalling in ageing

Studies in ageing models suggest that early in the ageing process mitochondria show a change to a phenotype reflecting modified fusion, together with a change in orientation more perpendicular to the fibre axis^{(Reference Del Campo, Contreras-Hernandez and Castro-Sepulveda82)} in association with other changes in mitochondrial dynamics^{(Reference Sharma, Smith and Yao83)}. Mitochondria play a central role in regulation of muscle protein synthesis and degradation through multiple signalling pathways, including energy production, generation of ROS, modified calcium handling and cytochrome C release initiating apoptotic pathways^{(Reference Romanello, Guadagnin and Gomes84,Reference Hyatt and Powers85)} . Mitochondrial signalling to activate these various pathways has been linked to failure of protein homeostasis in muscle atrophy through, for example, altered mitochondrial ATP generation leading to energy dependent dephosphorylation of AMPK^{(Reference Shenkman86)}, increased ubiquitination due to increased ROS generation^{(Reference Dodd, Gagnon and Senf87,Reference Hyatt, Deminice and Yoshihara88)} and activation of apoptotic pathways due to increased cytochrome c release^{(Reference Hyatt and Powers89,Reference Siu, Pistilli and Alway90)} . The role of mitochondria as a potential master regulator of muscle mass and function in a variety of different models seems clear, but due to the varying aetiologies of the onset of different conditions leading to muscle loss, common initiating factors that lead to mitochondrial disruption have not been recognised. The loss of muscle with ageing occurs with loss of motor units in both humans and rodents^{(Reference Campbell, McComas and Petito23,Reference Sheth, Iyer and Wier24)} , and loss of innervation of individual fibres has been reported in aged muscles. We found that ∼15 % of muscle fibres in old mice were completely denervated and ∼80 % of NMJs showed disruption^{(Reference Vasilaki, Pollock and Giakoumaki27)}. Studies of mice lacking Sod1 (a model of accelerated skeletal muscle ageing)^{(Reference Muller, Song and Jang91–Reference Zhang, Davis and Sakellariou95)} have highlighted the role of disruption of neuromuscular integrity in regulation of muscle mitochondrial ROS generation. These data, combined with studies of transection of the innervating nerve, which also caused a large increase in muscle mitochondrial peroxide generation^{(Reference Muller, Song and Jang91)}, identified a key role for motor neuron and NMJ integrity in regulation of muscle mitochondrial ROS generation in old mice. We examined the effect of partial denervation of the mouse tibialis anterior (TA) muscle and found a substantial increase in mitochondrial peroxide generation in the denervated fibres and also in neighbouring innervated fibres (Fig. 2)^{(Reference Pollock, Staunton and Vasilaki28)}. These data suggest that loss of innervation in fibres contributes to increased mitochondrial ROS generation^{(Reference Pollock, Staunton and Vasilaki28)} and associated mitochondrial degeneration^{(Reference Scalabrin, Pollock and Staunton96)} in ageing.

Fig. 2. (a) Effect of partial denervation of the Tibialis Anterior (TA) muscle on peroxide generation by mitochondria from different regions of the muscle. The four regions identified contained fibres that were either fully innervated (Region R1), fully denervated (Region R3) or partially denervated (Regions R2 and R4) and small bundles of fibres were obtained from each region. (b) These fibres were permeabilised and state 1 mitochondrial peroxide generation examined at 7 d post-surgery in comparison with sham-operated control muscles; *P < 0.05 compared with fibres from the same region of sham-operated muscles. Modified from reference^{(Reference Pollock, Staunton and Vasilaki28)}.

We have linked the attenuation of responses to contractile activity seen in both aging and the Sod1KO mice to an increase in muscle mitochondrial hydrogen peroxide production in both situations. We speculated that the increase in mitochondrial hydrogen peroxide would lead to an increased expression of regulatory enzymes for reactive oxygen species (Prx, GPx, TrX etc) which would suppress the likelihood of oxidation of critical cysteines in signalling proteins during contractions^{(Reference Jackson65)}. Furthermore since cycles of localised denervation and re-innervation appear to occur throughout life and may contribute to disrupted mitochondrial peroxide generation^{(Reference Jackson65)} and mitochondrial structure and function^{(Reference Scalabrin, Pollock and Staunton96)}, we speculated that the focal denervation seen in both adult Sod1KO and old WT mice leads to the increased mitochondrial peroxide production in both the denervated and neighbouring innervated muscle fibres which would drive the attenuation of redox-regulated adaptive mechanisms^{(Reference Jackson65)} (Fig. 3). This provides a testable mechanism by which focal and intermittent denervation during aging have a deleterious effect in suppressing key responses of muscle to exercise.

Fig. 3. Schematic illustrating the outline mechanism by with contractile activity in skeletal muscle leads to increased cytosolic hydrogen peroxide, thiol oxidation and activation of adaptive signalling pathways. In ageing or denervation, a chronic increase in mitochondrial hydrogen peroxide generation leads to attenuation of these responses by inducing a chronic upregulation of expression of various regulatory proteins for hydrogen peroxide. See text for further details.

Implications for nutritional interventions

Most of the current interest in nutritional interventions to ameliorate or prevent age-related loss of muscle mass and function is concerned with the protein content or composition of the diet and potential protein supplements^{(Reference Campbell, Deutz and Volpi97)}. Such approaches offer considerable promise in helping maintain muscle bulk in the elderly^{(Reference Mathewson, Azevedo and Gordon98)}. Interventions that affect the ageing process per se also appear a potential route to preservation of muscle, but currently these are focussed on experimental models and primarily involve pharmacological approaches^{(Reference Moskalev, Guvatova and Lopes99)}. There has been considerable speculation and preliminary studies examining whether “antioxidant” nutrients may be beneficial in prevention or treatment of sarcopenia, but intervention studies have been disappointing (e.g. see recent reviews on vitamins E and C^{(Reference Khor, Abdul Karim and Ngah100–Reference Liu, Zhang and Li102)}). Epidemiological data also support a potential beneficial effect of a Mediterranean diet high in antioxidants as protective against sarcopenia. A recent systematic review concluded that Mediterranean diet adherence had a positive effect in maintaining muscle mass and muscle function in older subjects, although the results were less clear with regard to muscle strength^{(Reference Papadopoulou, Detopoulou and Voulgaridou103)}.

The mechanistic data described above also indicate that aberrant mitochondrial ROS generation and defective redox signalling are features of ageing in skeletal muscle and contribute to attenuated responses of skeletal muscle to contractile activity and diminished adaptations to exercise^{(Reference Jackson, Pollock and Staunton104)}. The scheme shown in Fig. 3 suggests multiple sites at which pharmacological or nutritional interventions may interact to prevent or reverse the changes in redox signalling mechanisms that occur with ageing in skeletal muscle.