Journal of Applied Physiology

The production of reactive oxygen and nitrogen species by skeletal muscle

Malcolm J. Jackson, Deborah Pye, Jesus Palomero


Skeletal muscle has been recognized as a potential source for generation of reactive oxygen and nitrogen species for more than 20 years. Initial investigations concentrated on the potential role of mitochondria as a major source for generation of superoxide as a “by-product” of normal oxidative metabolism, but recent studies have identified multiple subcellular sites, where superoxide or nitric oxide are generated in regulated and controlled systems in response to cellular stimuli. Full evaluation of the factors regulating these processes and the functions of the reactive oxygen species generated are important in understanding the redox biology of skeletal muscle.

  • free radicals
  • mitochondria
  • reduced nicotinamide adenine dinucleotide phosphate oxidase

although there was some recognition in the 1950s that skeletal muscle could generate free radical intermediates (10), the first substantive demonstrations that exercise was associated with an increase in the formation of the products of the reactions of free radicals, reactive oxygen species (ROS) or reactive nitrogen species (RNS), with other biomolecules occurred in the late 1970s (5, 16), and these data were subsequently expanded and confirmed by Kelvin Davies and colleagues in Lester Packer's laboratory at Berkeley (13, 55). Many of the assumptions that remain prevalent about free radicals generated during exercise, such as skeletal muscle mitochondria being the predominant source for formation of free radicals and that the species formed were potentially damaging, appear to have been formulated and expounded in those early studies.

The techniques used to examine free radical activity in these early studies utilized the approaches available at that time, mainly measurements of the products of free radical reaction with lipids (5, 16), although Davies also utilized electron spin resonance spectroscopy to examine the relatively stable species that is seen by direct analysis of tissues (13). Subsequent studies examined other measures of free radical activity, but the demonstration of specific species that are generated by skeletal muscle during contractions occurred only after a further ∼10 years of research with Reid's description of superoxide release from the contracting diaphragm (56), the demonstration of nitric oxide (NO) generation by skeletal muscle (2), and the detection of hydroxyl radical formation by contracting skeletal muscle tissue (51).

Recent studies have clarified a number of the uncertainties relating to ROS generation in skeletal muscle, with greater recognition of the nature of the species formed, the identification of multiple sites for their generation, and a further understanding of their potential roles. Our current understanding of the sites of generation of ROS within skeletal muscle fibers is shown in schematic form in Fig. 1.

Fig. 1.

Schematic representation of potential mechanisms for generation of reactive oxygen species (ROS) in skeletal muscle. CAT, catalase; CuZnSOD, copper, zinc superoxide dismutase (SOD1); ecSOD, extracelluar superoxide dismutase (SOD3); GPx, glutathione peroxidase 1; eNOS, endothelial nitric oxide synthase (NOS3); NO, nitric oxide; MnSOD, manganese superoxide dismutase (SOD2); nNOS, neuronal nitric oxide synthase (NOS1); PM oxido-reductase, plasma membrane-located oxido-reductase; PLA2, phospholipase A2; VDAC, voltage-dependent anion channel; GPx, glutathione peroxidase. [Modified from Ref. 28.]


The primary free radicals generated by skeletal muscle, both at rest and during activity, are NO and superoxide, the latter of which dismutates rapidly to form hydrogen peroxide (40). These substances provide the precursors for generation of peroxynitrite (9, 53) and hydroxyl radicals in the presence of catalytic transition metals (53). The limitations of the techniques available have meant that few studies have clearly identified the nature of the ROS or RNS present within muscle fibers, and most work has examined species released or formed on the extracellular surface of muscle cells (53, 56). This approach is clearly suitable for substances such as hydrogen peroxide or NO, which can cross the plasma membrane but has limited application for highly reactive substances, such as the hydroxyl radical or the superoxide anion.

Some studies have also identified secondary free radical species that are generated during exercise, such as lipid radicals that may originate in muscle tissue (1). The source of these species has not been identified, although some data suggest they may be generated by reaction of hydroxyl radical with lipids released from muscle tissue or by enzymatic oxidation of lipid species (52).

To ensure that these species do not inevitably injure skeletal muscle, the tissue has a well-developed system to regulate ROS and prevent potentially deleterious effects. These protective systems include both mitochondrial and cytosolic isoforms of superoxide dismutase (MnSOD and CuZnSOD, respectively), catalase and glutathione peroxidase enzymes, and a number of direct scavengers of ROS, including glutathione, vitamin E, and ascorbate. In general, slow-twitch, mitochondria-rich (type I) fibers have an increased content of protective systems compared with fast (type II) fibers.




Standard texts cite mitochondria as the major site of superoxide generation in tissues (22), and many authors have reiterated early reports that 2–5% of the total oxygen consumed by mitochondria may undergo one electron reduction with the generation of superoxide (4, 38). There has been considerable debate about the site(s) of superoxide generation within mitochondria, but most data now indicate that complexes I and III of the electron transport chain are the main sites of mitochondrial superoxide production (3, 48). In complex I, the main site of electron leakage to oxygen appears to be the iron-sulphur clusters, and in complex III it appears to be the Qo semiquinone (48). Recent data also indicate that complex III releases superoxide to both sides of the inner mitochondrial membrane, providing a potential source for cytosolic superoxide generation, although CuZnSOD has also been reported to be partially located in the mitochondrial intermembrane space where it may minimize this possibility (48). When related to exercise, many researchers have assumed that the increased ROS generation that occurs during contractile activity is directly related to the elevated oxygen consumption that occurs with increased mitochondrial activity, implying potentially a 50- or 100-fold increase in superoxide generation by skeletal muscle during aerobic contractions (e.g., see Refs. 31, 64).

A number of recent findings have argued against a major formation of superoxide within mitochondria. Brand and colleagues have reassessed the rate of production of ROS by mitochondria and indicated that the upper estimate of the proportion of the electron flow giving rise to ROS was ∼0.15%, or <10% of the original minimum estimate (60), and it is becoming increasingly clear that even this low rate of production may be further reduced by intrinsic control mechanisms. Thus this same research group has highlighted the potential role of uncoupling proteins (specifically uncoupling protein-3 in skeletal muscle) as regulators of mitochondrial production of ROS (6, 7), acting to protect mitochondria against oxidative damage. In addition, there has been considerable debate about the effect of changes in the stage of mitochondrial respiration on ROS generation by mitochondria (17, 25, 35), and some evidence indicates that, during aerobic contractile activity, skeletal muscle mitochondria are predominantly in state III, and this limits their generation of ROS during contractions (17, 25, 35).

Measurements of intracellular ROS generation by contracting skeletal muscle have only been rarely undertaken, but where these are available for skeletal muscle cells, the data have indicated that intracellular ROS activity only increased by a modest two- to fourfold during contractions (43, 65), which appears to support the more recent assessments of the likely magnitude of mitochondrial ROS generation. As a final twist to this evolving picture of the role of mitochondria in ROS generation, Kozlov and colleagues (35) have recently used new spin probe approaches to examine ROS generation by skeletal muscle mitochondria and concluded that no ROS were released from isolated mitochondria of young rats (although a significant release was detected from isolated mitochondria from the tissues of old rats).

Sarcoplasmic reticulum.

Studies have identified NAD(P)H oxidase enzymes associated with the sarcoplasmic reticulum (SR) of cardiac (8) and skeletal muscle (67). The superoxide generated by these enzymes appears to influence calcium release by the SR through oxidation of the ryanodyne receptor (8). The skeletal muscle enzyme described appears to preferentially use NADH as substrate (67). Some inhibitor studies have indicated that extracellular superoxide release from stimulated myotubes was reduced by treatment with diphenyleneiodonium, a nonspecific inhibitor of NAD(P)H oxidases (53), although the NADH oxidase described by Xia et al. (67) is localized to the SR and hence seems unlikely to contribute to the extracellular release.

Plasma membrane.

Numerous studies have indicated that skeletal muscle cells release superoxide into the extracellular space (e.g., see Refs. 41, 53, 56, 69). All cells contain plasma membrane redox systems capable of undertaking electron transfer across the plasma membrane. Only the NADPH oxidase system of phagocytic and nonphagocytic cells has been extensively described. An NAD(P)H oxidase complex has been reported to be constitutively expressed in diaphragm and limb muscles of the rat and localized to the region of the plasma membrane (30). The enzyme contains four of the subunits that are found in the enzyme in phagocytic cells (gp91phox, p22phox, p47phox, and p67phox), all of which were associated with the cell membranes (37). Whether this complex predominantly releases superoxide to the inside or the outside of the plasma membrane cannot be ascertained from the experiments reported (30).

Other plasma membrane redox systems are less well characterized but are capable of transferring electrons from intracellular reductants to appropriate extracellular electron acceptors. No such system has been described in skeletal muscle, but other cell types contain networks of enzymes that perform this function (59). Morré (47) has described external NADH oxidase (ECTO-NOX) proteins that exhibit a hydroquinone (NADH) oxidase activity and a protein disulphide-thiol exchange activity. The current understanding of these systems is that they accept electrons from the hydroquinones of the plasma membrane, such as reduced coenzyme Q10 (45), and can reduce a number of nonphysiological (e.g.. ferricyanide and WST-1) and physiological (e.g., protein thiols or oxygen) electron acceptors outside the cell, although oxygen is likely to be a major acceptor in vivo (15). Transfer of electrons from cytosolic NAD(P)H to the plasma membranes has been proposed to occur through either NADH-cytochrome b5 oxidoreductase or NAD(P)H quinone oxidoreductase (15). Thus, through a series of linked steps, intracellular NAD(P)H can act as substrate for superoxide generation on the cell surface. de Grey (14) and Morré et al. (46) have hypothesized that such systems may generate extracellular superoxide to help maintain NADH homeostasis in glycolytic tissues and to help maintain cell viability where mitochondria become defective during ageing of tissues, such as skeletal muscle.

The relevance of these processes to skeletal muscle contractions has not been established, but it is feasible that such systems are activated during contractile activity or that the substrate level rises to increase electron transfer across the membrane through these systems. During the initial contractions of muscle and where muscle contractions are maintained at a high intensity, NADH levels are reported to rise acutely to levels >200 μM (24, 58), and it may be that this stimulates activity of the putative NADH-dependent plasma membrane oxido-reductase system, allowing regeneration of intracellular NAD+, with the resulting one electron reduction of molecular oxygen on the outside of the plasma membrane. The characteristics of the release of superoxide from skeletal muscle are compatible with the involvement of such a system.

Other endogenous generation systems.

A number of other potential systems have been described within skeletal muscle. Reid and colleagues have recently described phospholipase (PL) A2-dependent, intracellular superoxide generation (21). The enzyme involved was calcium independent and appeared distinct to the calcium-dependent PLA2 that was also reported to stimulate ROS production by Nethery et al. (49) and Zuo et al. (69). Reid and colleagues rationalized these apparently disparate findings by hypothesizing that the calcium-independent PLA2 was a major determinant of ROS activity under resting conditions, whereas, during contractions, heat stress, or other processes elevating intracellular calcium, the calcium-dependent PLA2 was activated and stimulates ROS production at supranormal rates. The subcellular location of these processes appears relatively undefined, although the calcium-dependent PLA2 has been reported to contribute both to mitochondrial ROS formation (50) and to the extracellular release of superoxide through a lipoxygenase-dependent system (70).

There has also been considerable speculation concerning a role for muscle-associated xanthine oxidase in superoxide generation by skeletal muscle, data that are primarily based on the effects of the xanthine oxidase inhibitors, allopurinol or oxypurinol (e.g., see Ref. 20). Skeletal muscle cells per se have been reported not to contain significant amounts of xanthine dehydrogenase or oxidase (23), although these enzymes will inevitably be present in associated endothelial cells.


NO is generated continuously by skeletal muscle, a production that is increased by contractions (2). Skeletal muscle normally expresses the neuronal (type I or nNOS) and the endothelial (type III or eNOS) isoforms of NO synthase (NOS). nNOS is strongly expressed in fast-twitch muscle fibers and localized to the muscle sarcolemma, where it is associated with the dystrophin-glycoprotein complex. eNOS is localized to the muscle mitochondria (34). Inducible NOS (type II) is also expressed in skeletal muscle in some inflammatory conditions, but it does not play a significant role in normal muscle (61). Release of NO was originally demonstrated from isolated muscles in vitro (2), although the cellular source of the NO released was unclear. Our recent analysis of myotubes in culture has confirmed that skeletal muscle cells per se release increased amounts of NO during contractile activity (53), a release that was greatly reduced by the NOS inhibitor NG-nitro-l-arginine methyl ester. nNOS appears to be the prime source of the NO released from skeletal muscle (26). Passive stretching of muscle has also been shown to increase NO release from rat skeletal muscle in vitro (63), and nNOS expression is increased by repeated exposure of muscle to contractile activity or passive stretching (56, 63).


Nonmuscle sources may play a major role in modifying muscle redox state under certain conditions. Most notable appears to be the role played by phagocytic white cells following insult to skeletal muscle. It is clear that substantial injury to muscle fibers is accompanied by invasion of the area with macrophages and other phagocytic cells (e.g., see Ref. 42). This process appears to be essential for preparation of the tissue to allow effective regeneration to occur, but involves the release of substantial amounts of ROS from the phagocytic cells (22, 39). The magnitude of this release can be such that damage to previously undamaged muscle cells may result (68).


Interactions of NO and Superoxide

NO reacts with superoxide to generate peroxynitrite (ONOO), a reaction that is approximately three times more efficient than SOD in scavenging superoxide. Hence peroxynitrite formation is preferred where both radical species are present (22). The presence of one of these species can therefore affect the “bioavailability” of the other: NO can reduce superoxide toxicity, and conversely superoxide can decrease NO availability and inhibit effects such as vasodilation. Such effects have been studied in endothelial cells (11, 27), but not in skeletal muscle. There is also evidence for a coordinated regulation of pathways influencing superoxide and NO levels. Exercise increases muscle NO and superoxide generation and leads to increases in muscle nNOS and intracellular and extracellular SOD activities (19, 41, 57). NO or superoxide also appears to influence the regulatory pathways for the other compound: an increase in hydrogen peroxide content is reported to stimulate NOS expression in endothelial cells (18), while elevated levels of NO decrease extracellular SOD activity in smooth muscle cells (62). Peroxynitrite is also reported to decrease MnSOD by nitration of the active site (54). We tested the hypothesis that NO and superoxide react in the muscle extracellular space following increased release from contracting myotubes by measuring superoxide release from wild-type cells when NOS were inhibited by NG-nitro-l-arginine methyl ester treatment and by measuring superoxide activity in muscle microdialysates following inhibition of NOS activities in vivo (Fig. 2). NOS inhibition increased the superoxide anions detected in both situations (9, 53), demonstrating indirectly that peroxynitrite was formed where both species were present. We have also examined the possibility that peroxynitrite is formed following contractile activity in muscles by analysis of nitrotyrosine residues in muscle proteins. Nitrotyrosines were increased following a single period of contractile activity (data not shown in detail). We conclude that peroxynitrite may be formed from NO and superoxide generated by muscle cells, reducing the effective bioavailability of both. This appears to provide a system whereby relatively small changes in generation of one of the primary ROS (i.e., NO or superoxide) leads to sensitive changes in several other substances that potentially influence cellular responses. Thus, for example, a decrease in NO generation will lead to a decrease in peroxynitrite formation and an increase in superoxide and hydrogen peroxide content. In other cell types, an imbalance of NO and superoxide has been recognized to cause functional changes (11, 12), but this possibility does not appear to have been examined in skeletal muscle.

Fig. 2.

Effect of NG-nitro-l-arginine methyl ester (l-NAME) treatment on microdialysate measurements of extracellular superoxide release from the resting gastrocnemius muscle of mice. Data from groups of untreated (solid bars) and l-NAME-treated (open bars) mice are shown. The additional superoxide detected following inhibition of NO synthesis appears to be due the lack of formation of peroxynitrite. *P < 0.05 compared with untreated mice. [Data redrawn from Close et al. (9).]

Transmembrane Signaling of ROS and RNS Activities

A number of pieces of evidence indicate that changes in the intracellular ROS generation or redox environment in skeletal muscle can influence ROS activities outside of the muscle cell and vice versa. While this process partially reflects the ability of skeletal muscle to generate relatively long-lived species that can cross membranes such as hydrogen peroxide, there is also evidence that more subtle effects occur. Thus Reid and coworkers (56) reported that the addition of scavenger enzymes to the medium surrounding diaphragm fibers could influence intracellular ROS activity, despite the lack of access of these proteins into the muscle cell, and more recently genetic manipulation of intracellular regulatory enzymes for ROS (e.g., MnSOD) has been shown to result in modification of the activity of hydroxyl radical activity in the muscle extracellular space (42).

The importance of these observations may lie in the potential roles of superoxide/hydrogen peroxide or NO in signaling adaptive responses (see Ref. 29 for a review). Some changes in cell signaling following contractile activity in skeletal muscle are reported to be mediated by ROS or RNS (66), and we have obtained evidence that contraction-induced ROS or RNS modulate at least some of the adaptive and stress responses in skeletal muscle following contractile activity. A single period of contractile activity in mouse muscle increased muscle SOD and catalase activities, together with heat shock protein 60 and heat shock protein 70 content (41), changes that were replicated in human muscle (32). Presupplementation with vitamin C reduced these responses, supporting the possibility that they are regulated by ROS or RNS (33). It is argued that the ability to transmit fluctuations in redox status across the plasma membrane is essential to facilitate coordinated responses to stress in multiple fibers of whole skeletal muscles and surrounding tissues.


Skeletal muscle clearly produces a variety of ROS and RNS in controlled and regulated processes, both at rest and during contractile activity. Although initial data suggested the major generation of superoxide occurred “by accident” or as a “by-product,” due to leakage from the mitochondrial electron transport chain, recent data have cast considerable doubt on this theory. In contrast, evidence is emerging of a closely controlled generation of ROS and RNS at multiple subcellular sites in response to a variety of mechanical and metabolic stimuli. Understanding the factors regulating this production, the processes underlying it, and the functions of the species generated may potentially provide a means of optimizing skeletal muscle function in a variety of disparate physiological and disease states.


We thank the BBSRC, European Commission, Wellcome Trust, Research into Ageing, and the US National Institute on Aging for financial support of this work.


The authors acknowledge the collaboration of many colleagues.


View Abstract