Exerkines and redox homeostasis

Abstract

Exercise physiology has gained increasing interest due to its wide effects to promote health. Recent years have seen a growth in this research field also due to the finding of several circulating factors that mediate the effects of exercise. These factors, termed exerkines, are metabolites, growth factors, and cytokines secreted by main metabolic organs during exercise to regulate exercise systemic and tissue-specific effects. The metabolic effects of exerkines have been broadly explored and entail a promising target to modulate beneficial effects of exercise in health and disease. However, exerkines also have broad effects to modulate redox signaling and homeostasis in several cellular processes to improve stress response. Since redox biology is central to exercise physiology, this review summarizes current evidence for the cross-talk between redox biology and exerkines actions. The role of exerkines in redox biology entails a response to oxidative stress-induced pathological cues to improve health outcomes and to modulate exercise adaptations that integrate redox signaling.

The importance of physical activity and exercise training in the prevention of chronic disease and the reduction of mortality is well-established [1,2]. Indeed, regular exercise elicits a number of beneficial adaptive responses that contribute to its whole-body health benefits [3]. While many of these adaptations are observed in skeletal muscle, exercise also elicits positive effects in the liver, adipose tissue, and brain, as well as in the respiratory, cardiovascular, neuroendocrine, and immune systems [[4], [5], [6], [7]].

Redox biology can be considered a central mediator of exercise-induced adaptations as it contributes to modulation of muscle contraction, glucose uptake, blood flow, and cell bioenergetics, as well as exercise-induced adaptations such as mitochondrial biogenesis, muscle hypertrophy, and angiogenesis [8]. The role of redox biology in exercise physiology comprises the main regulators of redox homeostasis including the nicotinamide adenine dinucleotide (NAD⁺/NADH and NADP⁺/NADPH) and glutathione (GSH/GSSG) systems, the antioxidant enzyme superoxide dismutase (SOD), and the redox signalers reactive oxygen and nitrogen species [8]. However, co-regulation of the redox status by several transcription factors, transcription factor coactivators, and protein signaling entails a more complex network to maintain redox homeostasis that can also be modulated by exercise [8].

Redox biology adaptations to exercise are, like the effects of exercise, both tissue-specific and systemic [8,9]. Importantly, inter-organ cross-talk also contributes to exercise-induced adaptations, resulting from secretory functions of the skeletal muscle, heart, brain, liver and adipose tissue [5]. During exercise, a broad group of metabolites, hormones and cytokines – termed exerkines - are secreted to induce tissue-specific and systemic adaptations resulting from their ample endocrine, autocrine and paracrine effects [5]. Several of these exerkines have been identified in different biofluids and tissues after acute and chronic exercise in both mice and humans [5]. This review will discuss the roles of exerkines in target tissues to induce health benefits modulated by their effects on redox homeostasis. For this purpose, we include discussion on exerkines that mediate the effects of exercise by acting on, or that have a clear interaction with, the redox system.

1. Exerkines mediate the effects of exercise via redox homeostasis

1.1. Irisin

Irisin was discovered in 2012 as a small peptide myokine cleaved from its parent FNDC5 in response to exercise and the transcription factor for mitochondrial biogenesis PGC-1α [10]. It was identified to induce a brown adipose tissue-like phenotype in mouse white adipose tissue, which is characterized by higher thermogenic activity, mitochondrial biogenesis, and oxygen consumption [10]. Since then, gain and loss of function studies have confirmed that irisin mediates this brown-like phenotype in mouse, but not human inguinal adipose tissue [11,12]. Irisin also has described effects on mouse bone, pancreas, and brain [[13], [14], [15], [16]], and exerts biological actions that lowered blood pressure and improved cardiac hypertrophy [17]. Metabolic functions of irisin include anti-inflammatory and anti-metastasis effects, improvement of glucose uptake in muscle cells, and facilitation and proliferation in endothelial cells [17]. Importantly, many of these actions are mediated by mitogen-activated protein kinases (MAPKs) p38 and ERK, which play pivotal roles in the response to cellular stress [17].

Several studies have shown a role for irisin to modulate the effects of the antioxidant transcription factor nuclear erythroid factor (Nrf)-2 in hepatocytes, endothelial cells, and cardiomyocytes, to regulate nitric oxide synthase (NOS) signaling in endothelial cells, and to modulate reactive species of oxygen (ROS) signaling to increase superoxide dismutase (SOD) in the brain (reviewed by Louzada et al. [9]). Recent proteomic studies have identified common redox-regulating pathways for irisin and exercise by comparing the effects of 8 weeks of aerobic exercise, resistance exercise, or 3 times/week intraperitoneal irisin injections in muscle, sciatic nerve and brains of mice [[18], [19], [20]]. Aerobic and resistance exercise, as well as irisin injections, induced redox-regulating proteins (disulfide isomerase) in sciatic nerve but not in muscle or brain, while redox proteins (transmembrane, SOD and periredoxin) increased only in the exercised groups [[18], [19], [20]]. More experiments are needed to clarify if the lack of effects of irisin in muscle indicate tissue, exercise, or dose-dependent effects, but these are the first studies to show similar redox pathways for exercise and irisin.

Since the redox effects of irisin have been described in neurons, cardiac and endothelial cells, and given the broad neuroprotective and cardioprotective effects of irisin [[21], [22], [23]], research investigating the role of exercise-induced irisin in the control of redox homeostasis has been performed primarily under conditions derived from restricted blood flow to the heart or the brain, i.e. myocardial infarction (MI) [[24], [25], [26], [27]], ischemia [[28], [29], [30]], and cardiac toxicity [31,32]. Wild type (WT) or Fndc5 knockout mice underwent an MI by ligation of the left anterior descending coronary artery and were then trained for six weeks at moderate-intensity on a treadmill [[24], [25], [26], [27]]. The exercise protocol negated tissue-specific MI insults in WT mice, which included a decrease in FNDC5 or irisin and in antioxidant activity, together with an increase in oxidative stress, inflammation, fibrosis, apoptosis, mitophagy, mitochondrial fusion, protein degradation and the marker of abnormal mitochondrial remodeling ALCAT1 [[24], [25], [26], [27]]. Conversely, the loss of Fndc5 intensified the effects of MI and prevented the beneficial effects of exercise in muscle and liver [24,25]. Collectively, these studies showed that exercise-induced irisin plays a role in the recovery of heart, skeletal muscle, liver and kidney injury induced by MI [[24], [25], [26], [27]].

Irisin also mediates the effects of exercise on cardiotoxicity induced by chemotherapy (doxorubicin) [31] or radiation [32] by modulating autophagy and mitophagy. Exercise and intraperitoneal irisin injections rescued irisin levels in serum and heart and improved cardiac function, reduced perivascular fibrosis, altered the structure of heart capillaries, and attenuated immune infiltration in the heart induced by the cardiotoxicity [31]. Moreover, mice with radiation-induced heart disease (RIHD) showed that exercising mice after the disease improved cardiac function, aerobic fitness, ATP production and mitochondrial protein content, while it also decreased mitochondrial length and increased the formation of mitophagosomes compared with sedentary RIHD mice. These changes were accompanied by elevated expression of FNDC5/irisin, and markers of mitochondrial fission (DRP1) and mitophagy (PINK1 and LC3B) in the exercised RIHD group compared to the sedentary RIHD group. Mechanistic studies in mouse isolated cardiomyocytes and cardiac microvascular endothelial cells showed that irisin ameliorated pathogenic endothelial differentiation into mesenchymal cells by attenuating ROS-induced nuclear factor k-β signaling (NF-kB, pro-oxidant transcription factor) thanks to the promotion of autophagy [31].

The exercise-induced increase in irisin shows protection against worsened cognitive performance after brain ischemia or trauma by broader mechanisms than autophagy and mitophagy. In mice with middle cerebral artery occlusion, two to four weeks of exercise or irisin injection 30 min prior to surgery increased hippocampal irisin and cognitive function while decreased oxidative stress and the infarcted area [29,33]. Mice with controlled cortical impact-induced brain trauma showed increased irisin and reduced edema when exercised before the trauma or treated with intraperitoneal irisin immediately after it [30]. Signaling studies showed that irisin stimulates ERK and Akt signaling in the infarcted area, which are responsible for improving cognitive tests results and decreasing inflammatory gene expression in the brain of mice treated with irisin and chemical blockade of these pathways [33]. Moreover, loss of function studies showed that UCP2 and Klotho, an anti-aging gene, could contribute to the molecular effects of irisin in both conditions, since both were higher in hippocampus of the mice treated with irisin and neglected irisin-induced reduction in ROS and induction of SOD in null mice [29,30]. There is one study investigating the role for exercise-induced irisin in muscle ischemia. However, the exercise protocol was performed after the ischemia and therefore, using a recovery approach [28]. Hence, the recovery in muscle mitochondrial dynamics occurred together with an increase in muscle PGC1α and FNDC5, and higher serum irisin [28]. In vitro studies showed that hypoxic and fasted C2C12 cells stimulated with electrical pulses had elevated Pgc1a/Fndc5, irisin secretion, and genes for mitochondrial fission, autophagy and mitophagy than the non-stimulated hypoxic, fasted cells. However, these effects were inhibited by silencing PGC1α instead of FNDC5 [28], which does not allow to conclude for an effect for irisin to mediate mitophagy/autophagy in muscle ischemia.

In the liver, signaling studies with H₂O₂-induced oxidative stress or LPS-induced inflammation suggested that irisin mediates the anti-apoptotic and anti-inflammatory effects of exercise by PI3K/Akt signaling in mouse primary hepatocytes [25], AMPK mediated pathways in muscle and rat kidney cells [24,26], and the induction of the irisin-PINK1/Parkin-LC3/P62 mitophagy pathway in the myocardium [27]. A role of ALCAT1 was also suggested in muscle cells with silencing studies that induced the recovery of the apoptosis and the AMPK pathway [24]. All these studies require experiments with an integrative approach to establish if all pathways are mediated in various cells/tissues and to what extent, and the specific actions of exercise. However, they are indicative of the exercise-induced effects of irisin to modulate oxidative stress, mitophagy and autophagy to reduce inflammation and apoptosis after an MI by several pathways and by inhibition of ALCAT1 in main organs.

In summary, these collective studies demonstrate an important role for exercise-induced irisin point to regulate autophagy and mitophagy processes induced by oxidative stress in multiple tissues, although more studies are needed to fully elucidate all the pathways that lead to this regulation.

1.2. Fibroblast growth-factor 21 (FGF21)

The main target tissues of FGF21 are liver, white adipose tissue, and muscle, where it improves glucose and lipid metabolism, and insulin sensitivity [34]. Therefore, most studies investigating the beneficial effects of exercise-induced FGF21 involve metabolic disease in which exercise restores FGF21 sensitivity in adipose tissue to ameliorate systemic insulin resistance, lipid metabolism and inflammation, and improve muscle and liver metabolism [35]. During exercise, FGF21 is produced and secreted in circulation in humans and mice mainly by the liver [[36], [37], [38]]. Immediately after exercise, increases in circulating FGF21 occur together with enhanced lipolysis and glucose uptake in mice. Enhanced lipolysis and reduced insulin also go hand by hand with increased FGF21 in humans 1 h post-exercise [38].

The majority of the beneficial metabolic effects of FGF21 on adipose tissue and liver occur together with an increase in mitochondrial activity and biogenesis. Conversely, several conditions that aggravate oxidative stress and disrupt the redox balance induce FGF21 expression [39]. The mechanism by which FGF21 responds to oxidative stress relies, mainly, on its associations to mitochondrial and endoplasmic reticulum (ER) stress [39,40]. The unfolded protein response activated by both processes leads to the promotion of FGF21 expression, directly or through a signaling cascade [39,40]. Other mechanisms by which FGF21 expression relies on the cellular redox status are mediated by Nrf1 and Nrf2, or the ER-stress regulator activated transcription factors (ATF)-3, −4 and −5 [[39], [41], [42]]. Most evidence shows that stress and exercise induce FGF21 to reduce oxidative stress and protect against disease, especially in main tissues that are fueled by fatty acids which are heart, muscle, liver, and adipose tissue.

Research to date on the antioxidant effects of FGF21 shows that FGF21 is protective against cardiovascular disease due to its antioxidant effect on the heart [43] that protects against the development of diabetic cardiomyopathy [44,45], high-fat diet-induced cardiac steatosis [46], myocardial infarction [47,48], and atherosclerosis [49]. Importantly, some of these cardioprotective effects can also be achieved by exercise. Recent studies by Ma et al. [48] and Bo et al. [47] showed that aerobic treadmill exercise for 6 weeks post-MI alleviated markers and pathways of cardiac fibrosis (TGF-β1-Smad2/3-MMP2/9 and apoptosis (IREα/JNK) as well as abnormal mitochondrial remodeling mediated by cardiolipin [47,48]. These were all induced by oxidative stress in vitro in isolated mouse cardiac fibroblasts and H2C9 cells by H₂O₂-induced oxidative stress, and negated by treatment with recombinant human FGF21 [47,48]. Exercise also improved cardiac function, reduced ER stress and oxidative stress, and elevated cardiac FGF21 [47,48], but these beneficial effects of exercise were all abrogated in Fgf21 null mice [47,48], indicating that FGF21 mediates the cardioprotective effects of exercise by modulating oxidative stress-induced fibrosis, apoptosis, and abnormal mitochondrial remodeling.

Treadmill exercise has numerous cardioprotective effects in mice with diabetic cardiomyopathy by reducing cardiac dysfunction and mitochondrial damage. One mechanism for this improvement was enhancing FGF21 via AMPK/FOXO3/Sirtuin (SIRT)-3 to recover the function of a cluster of mitochondrial enzymes including ATP synthase subunits, succinate dehydrogenases (SDHs), long-chain Acyl Co-A dehydrogenase (LCAD), and SOD2 [44]. In cardiomyocytes derived from human induced pluripotent stem cell with palmitic acid-induced lipotoxicity, FGF21 also restored this pathway and increased mitochondrial activity and redox status [44]. Interestingly, selective ablation of hepatocyte Fgf21 or cardiomyocyte β-klotho (FGF21 receptor) negated nearly all the effects of treadmill exercise on cardiac function and mitochondria [44]. These findings reinforce exercise-derived hepatic FGF21 as the main effector of the cardioprotective actions and highlight the relevance of tissue-specific Fgf21 sensitivity in the heart.

With regard to skeletal muscle, the effects of exercise-derived FGF21 have been studied mainly when FGF21 is secreted from the liver and in conditions of metabolic disease. In this instance, FGF21 drives exercise-induced muscle adaptations such as the AMPK-mediated increase in glucose uptake and the incremental increases in fatty acid (FA) oxidation and mitochondrial biogenesis [35,50]. However, the regulation of redox signaling by FGF21 in muscle is better described for its autocrine effects in animal models of muscle mitochondrial disease [40,51], in which FGF21 has been proposed as a biomarker [52]. In this context, local expression of FGF21 is induced by mitochondrial ROS and ER stress to regulate specific stages of the mitochondrial integrated stress response (ISR^mt). This includes FGF21-dependent regulation of crucial redox homeostasis steps like serine biosynthesis, trans-sulfuration, and induction of ER stress transcription factors ATF-3 and ATF-5, while increasing mitochondrial replication by the PGC1α/SIRT1 axis [40]. Interestingly, the ISR^mt-induced FGF21 is regulated by the master protein synthesis regulator mTORC1 [51]. Oost et al. [53] recently identified FGF21 to play a crucial role in muscle mass by governing catabolic/anabolic processes through the regulation of mitophagy in an ER stress-dependent manner, which is not observed in muscle-Fgf21^−/− mice [53]. Muscle FGF21 also signals to other metabolic tissues, likely to increase muscle fuel supply in animal models of muscle mitochondrial myopathy [54], dysfunction [55,56], or energy inefficiency [57] that run with mitochondrial or ER stress-induced FGF21 and show a lean phenotype that is hypermetabolic and resistant to diet-induced obesity. Hence, the interplay of FGF21 with local stress, muscle integrity, and local and systemic metabolism deserves exploration in the context of exercise training.

As the liver is the main production site for FGF21, several pathways have been described for FGF21 to reduce oxidative stress in the liver. This includes the induction of Nrf2 antioxidant pathways and the increase in mitochondrial activity via AMPK/PGC1α/SIRT1 observed in mice with different models of dietary or pharmacological oxidative stress [[58], [59], [60], [61]]. Oxidative stress-induced FGF21 also modulates apoptosis by reducing ER stress signaling (eIF2α/ATF4/CHOP) in livers of mice with non-alcoholic fatty liver disease (NAFLD) [62], and induces autophagy pathways via AMPK/mTOR in models of pharmacological-induced liver injury [63]. Despite these findings, only a few studies have explored these pathways after exercise training. Eight weeks of voluntary wheel running restores the impaired liver and systemic metabolic phenotype observed in Fgf21-KO mice as compared to WT mice [64]. However, exercise did not restore master regulators of mitochondrial biogenesis, antioxidant cascades, FA oxidation and glycogen synthesis in the liver including protein levels of nuclear PGC1α and PPARα, β-HAD, and PEPCK, which remained reduced compared to WT mice [64]. These findings suggest that exercise-induced FGF21 modulates alternative pathways in the liver to regulate antioxidant activity and mitochondrial dynamics. Indeed, FGF21 also mediates selective autophagy that involves lipid droplets (lipophagy) by an AMPK-dependent pathway that regulates the initiation of autophagy by ULK1 to reduce lipid droplets in HepG2 cellular models of NAFLD [65]. This pathway likely contributes to the observed reduction of steatosis and oxidative stress markers (malondialdeyde (MDA), lipofuscin) in HFD-fed mice exercised 5 days/week on a treadmill for 8 weeks [65]. Similar antioxidant effects with higher liver Fgf21 have been observed in mice fed a non-alcoholic steatohepatitis-inducing diet and exercised on a treadmill 5 days/week for 7 weeks [66]. Altogether, exercise-derived FGF21 could contribute to the antioxidant effects of exercise in the liver, while more studies are needed to explore if exercise could induce FGF21-mediated lipophagy.

Finally, different reports show that experimental approaches to manipulate redox homeostasis induce the production of FGF21 in adipose tissue, including Nrf2 deletion, alteration of the NAD⁺/NADH ratio by nicotinamide ribosamide supplementation, or transgenic models of catalase [[67], [68], [69]]. It has been proposed that the production of FGF21 in adipose tissue contributes to an improved redox homeostasis due to the induction of a phenotype that is more thermogenic and energetically efficient [70]. Interestingly, this effect can also be achieved by the main exercise-derived metabolite, lactate, the end product of glycolysis [70].

In summary, current evidence on FGF21 modulation of redox homeostasis shows the ability of this growth factor to protect from injury and disease by reducing and potentially regulating ER, mitochondrial and oxidative stress in several organs and tissues. Recent studies have expanded our understanding of the ability of exercise-induced FGF21 to exert such effects, with exciting recent findings in the heart that will undoubtedly extend to the rest of organs in the years to come.

1.3. Brain-derived neurotrofic factor (BDNF)

BDNF is a member of the neurotrophin family of growth factors that has the most widespread expression in the developing and adult mammalian brain [71]. Since its discovery, a vast number of studies have reported on numerous aspects of BDNF signaling in the central nervous system, as well as BDNF's role in neuronal development, synaptic plasticity, and the pathology and treatment of psychiatric disorders [72]. Multiple studies have shown that BDNF is induced by exercise mainly in the hippocampus, where it induces neurogenesis, neural plasticity and neural survival to mediate the improvements in memory and learning that are induced by exercise [73]. As reviewed by Chow et al. [5], rodent studies on the effects of exercise-induced BDNF in the brain show increments in response to acute but not chronic exercise. Nevertheless, the results observed in animal models for exercise-induced BDNF correlate to those observed in humans [5]. In addition to influencing brain BDNF concentration, exercise has also been shown to modulate BDNF canonical signaling pathway via tropomyosin receptor tyrosine kinase B (TrkB) and by exerkines or exercise-derived metabolites irisin, lactate, cathepsin B, and beta-hydroxybutyrate that induce its expression [74,75].

Redox signaling is crucial for physiological brain function. Therefore, brain BDNF expression and signaling are influenced by ROS at several levels in healthy rodents and animal models of brain disease, as extensively reviewed by Radak et al. [[76], [77], [78]]. The influence of ROS levels on BDNF differ and seem to be dependent on the oxidative stress and pathological status, e.g. H₂O₂ downregulates the inducer of BDNF expression and reduces BDNF, but lead to higher BDNF in an animal model of Alzheimer's Disease (AD) [[76], [77], [78]]. A recent study has suggested that ROS levels also mediate BDNF effects, since BDNF induces neural calcium release by ryanodine receptor, which plays an important role in spatial memory, by incrementing ROS production through NOS/NADPH oxidase activity in primary rat hippocampal culture [79]. Moreover, BDNF has antioxidant effects in the brain by the modulation of neural Nrf2, MAPK ERK, antioxidant genes and oxidant proteins [76,77,80]. Altogether, current evidence shows a complex cross-talk for ROS levels and redox homeostasis to regulate BDNF and its effects in the brain.

However, some studies have shown ROS-levels induce BDNF as a protective mechanism to cope with oxidative stress in the brain [[76], [77], [78]]. It makes sense then that both aerobic and resistance exercise protocols induce higher hippocampal BDNF, CREB and TrkB, as well as higher neural density and cognition in middle-aged (9, 13 and 19 months old) and aged (22 months) male and female rodents together with lower oxidative status (ROS production and MDA levels) and higher antioxidant activity (SOD, catalase, glutathione peroxidase) [[81], [82], [83], [84], [85]]. These effects are also observed in animal models of Parkinson's Disease (PD), AD, and traumatic brain or spinal cord injury [[86], [87], [88], [89]]. Two weeks of daily treadmill running for only 30 min starting 24 h after the surgery initiating PD pathology (6-OHDA injection by stereotactic surgery) protect against the drop in BDNF, catecholamine synthesis and dopamine transporter while reducing behavioral alterations, dopaminergic neuron loss, nitrite content and markers of lipid peroxidation in rat striata [86]. In a model of AD, 8 weeks of high-intensity interval training starting 2 weeks after initiating AD pathology (Aβ1-42 injection by stereotactic surgery) ameliorated the impairments in physical capacity, locomotor activity, learning and memory together with higher hippocampal BDNF/TrkB, antioxidant capacity (SOD, catalase), and lower MDA and neuritic plaques in mice [87].

Exercising in rat models of traumatic brain injury also resulted in increased hippocampal proteins for synaptic activity and BDNF, with lower oxidized proteins [88], and protected against impairments in glutamate uptake, inhibition of SOD activity and increased ROS production, motor activity and learning while inducing BDNF and Nrf2 in the hippocampus [89]. Despite the descriptive nature of these studies, it has been shown that the effects of exercise to recover axonal conduction and synaptic activity after spinal cord injury are reduced (from 75% to 25%) or completely blunted when treating mice with TrkB neutralizing antibody [90]. Therefore, the hypothesis of BDNF signaling through TrkB to mediate the rest of the effects of exercise in brain antioxidant activity, synaptic activity, and cognition cannot be discarded.

Interestingly, several studies in rodents and humans show that BDNF is also produced by skeletal muscle after exercise [91]. Despite initial studies showed no increment in muscle-derived BDNF after exercising [92], later studies show that muscle secretes BDNF into the bloodstream, since the effects of pancreatic TrkB to induce insulin release are negated both in muscle-specific Bdnf null mice and β-cell deletion of TrkB [93]. Importantly, studies have shown that during exercise the brain releases BDNF instead of taking it up [94,95], but the main source for increased BDNF during exercise has not been described. Nevertheless, exercise also stimulates skeletal muscle TrkB [[92], [96]], which could bind to circulating BDNF. Moreover, exercise-derived BDNF in skeletal muscle exerts autocrine functions to stimulate fat oxidation by AMPK [92]. It is very likely that exercise-derived BDNF also stimulates mitochondrial dynamics in muscle in response to exercise, since muscle-specific Bdnf null mice show impaired mitochondrial biogenesis, fusion, and mitophagy in skeletal muscle [97]. Furthermore, BDNF^+/− mice also show lower skeletal muscle mitochondrial DNA, AMPK signaling, PGC1α, and exercise capacity [98]. In C2C12, a battery of gene expression for fatty acid metabolism and mitochondrial activity is stimulated by BDNF, mimicking the effects of exercise, but negated by treatment with anti-BDNF receptor [98]. Interestingly, muscle BDNF is colocalized with mitochondria in slow-twitch fibers [98], but also determinates fiber fate towards a fast-twitch type [99]. Moreover, BDNF also controls muscle regeneration [100] and motor neuron activity [101]. Therefore, further research to fully determine the roles of BDNF during exercise in muscle structure, function and mitochondrial dynamics is warranted.

One study in aged rodents supports this role for BDNF in the regulation of mitochondrial dynamics and included redox modulation in skeletal muscle of lifelong exercised rats (26 months old, exercised for 18 months). Compared to age-matched sedentary rats, lifelong exercise induced an increase in power grip and handling time, with higher muscle BDNF and SOD activity and lower MDA [102]. This occurs together with changes in the proteome that show increased pathways for protein processing in the ER, antioxidation, apoptosis and regeneration in skeletal muscle of lifelong exercised rats compared to sedentary controls [102]. The proteomic results were confirmed with the enhanced expression of a battery of proteins for BDNF/TrkB signaling, autophagy, mitochondrial function, and redox regulation in skeletal muscle of the exercised, aged rats [102], suggesting co-regulation of these processes.

Finally, BDNF in muscle also contributes to the effects of exercise in MI mice [98]. In this model, injection of recombinant human BDNF recovers MI-induced reduction in exercise capacity and muscle mitochondrial content, citrate synthase activity, and fatty acid oxidation [98]. More studies have shown effects for exercise to improve cardiac function and exercise capacity together with concomitant increases in cardiac BDNF, TrkB, and serum BDNF and a reduction in oxidant activity [103]. Importantly, it has been described that TrkB blocker injection partially inhibits the cardioprotective effect of exercise in MI [104]. BDNF exerts more anti-apoptotic, anti-fibrotic, and anti-oxidant effects in the heart by modulating angiogenesis and inducing NOS signaling after MI [105,106], inducing SIRT1 signaling in heart failure [107], or recovering AKT signaling after doxorubin-induced cardiotoxicity [108]. Whether exercise-derived BDNF can modulate these pathways remains unknown and requires further research.

In brief, the effects of exercise mediated by BDNF can be divided in the ones occurring inside or outside the central nervous system. In the brain, BDNF enhances synaptic activity and decreases oxidative stress. In muscle, BDNF regulates mitochondrial dynamics that mediate exercise adaptations and result in enhanced pathways for handling cellular stress, as observed in aged rodents. While the effects of exercise-derived BDNF in heart are preliminary, the amount of evidence for its beneficial effects in several conditions is promising for future studies.

1.4. 12,13-diHOME

Multi-omic studies [109,110] and systematic reviews [6,111] show that most robust changes in the human metabolome in response to exercise involve lipid metabolism. Besides the expected elevation in lactate and some TCA metabolites, there are large increases in various fatty acids (FAs) and acylcarnitines that have been consistently reported after both endurance and resistance exercise [[109], [110], [111]].

FAs can undergo oxidative reactions during exercise by cytochrome P-450 (CYP), cyclooxygenases (COXs), lipoxygenases (LOXs) and epoxygenases (EPHXs). Through these enzymatic reactions, long-chain FAs are further metabolized to bioactive lipid mediators called oxylipins that are released after exercise to mediate several effects on metabolism, inflammation, and cardiac, vascular and muscle physiology [6,[112], [113], [114], [115], [116]].

Among these oxylipins, increases in linoleic acid (LA) derivatives after exercise have been reported in mice and humans. Deriving from oxygenation reactions of linoleic acid by CYP and EPHX, 12,13-diHOME is the most significantly increased bioactive lipid after acute or chronic exercise in mice and humans [112,115,117]. Accordingly, 12,13-diHOME also shows the strongest association to physical activity in large population human studies [118], and is increased in exercise-trained subjects and mice [112]. In vitro metabolic effects of 12,13-diHOME include an increase in skeletal muscle, cardiomyocyte, and brown adipocyte FA uptake, as well as higher respiration in muscle, cardiomyocytes, and brown adipose tissue [112,115,119]. Higher respiration is usually associated to higher FA oxidation and accordingly, basal lipid oxidation is higher in mice injected with 12,13-diHOME [112]. Despite no effect has been described in liver fatty acid oxidation yet, this is consistent with human studies showing positive correlations for 12,13-diHOME with the main liver FA oxidation product, beta-hydroxybutyrate, after adjusting for sex and age (n = 2248) [113]. The increase in FA oxidation results in elevated oxidative stress due to an increase in oxidized products. Interestingly, an upregulation in the main marker of oxidative stress in exercise, F2-isoprostanes, correlates to the increase observed in LOX-derived 9- and 13-HODEs, but not to that reported for 12,13-diHOME, after a 75 km cycling bout in young, trained men [120]. This suggests that 12,13-diHOME-induced FA oxidation does not contribute to the higher oxidative stress during exercise, while LOX-derived HODEs do.

12,13-diHOME also affects the heart; it increases ejection fraction, systolic and diastolic function, and cardiomyocyte kinetics and respiration [115]. Since nitric oxide (NO) is needed for exercise to improve cardiac function [121], it was hypothesized that the cardiac effects of exercise-derived 12,13-diHOME would require NO. Interestingly, the effects of 12,13-diHOME to improve systolic and diastolic function, and cardiomyocyte kinetics and respiration were not observed in mice genetically deficient for neuronal nitric oxide synthase (NOS1), suggesting that 12,13-diHOME signaling through NO is needed to exert its cardiac effects [115]. Therefore, 12,13-diHOME likely leads to increased redox signaling due to increased bioenergetics, and directly interacts with the redox system through NOS signaling.

More studies have shown interaction between 12,13-diHOME, NOS activity and other elements relevant to the endothelial function. Park et al. [116] recently showed that increasing circulating 12,13-diHOME can directly reduce atherosclerosis by stimulating endothelial NOS activity and rescuing NO levels in HFD-fed ApoE deficient mice. Moreover, the oxidized LDL that accumulate in the endothelium and characterize the progression of atherosclerosis show a switch in the LA-derived oxylipin profile, with lower production of the CYP-derived 9,10- and 12,13-diHOME and higher LOX-derived HODEs, oxoHODES and triHOMES [122]. Interestingly, LA LOX-derived plasma triHOMEs already increase with aging in healthy men and women (6/4, 22 years and 25.5 ± 1.3 kg/m²; 5/5, 53 years and 24.3 ± 0.7 kg/m²) [123]. It is very likely that LA-derived oxylipins compete for the same receptor, and that their synthesis pathways (LOX or CYP) also compete or regulate one another, as described for other long-chain FA-derived oxylipins and their synthesis pathways [124,125]. Therefore, these data suggest that LA-CYP derived 12,13-diHOME is needed for a normal endothelial function, as it contributes to NOS functioning and could lead to a lower production/activity of the LOX-derived oxylipins that associate to atherosclerosis and aging.

Despite this, 12,13-diHOME was not associated to the risk of cardiovascular disease events after cross-validating the physical activity lipidome in two different large-population clinical trials (VITAL, n = 589 and JUPITER, n = 1032) [118]. 12,13-diHOME was, however, lower in subjects with cardiovascular disease [115] and the 12,13-diHOME decreased with the progression of coronary artery disease [126]. These contradictory data may be due to the assessment of physical activity levels by self-reported questionnaires and scales [118] and suggest that 12,13-diHOME does not associate to the appearance of the disease but plays a role in its pathogenesis. Indeed, 12,13-diHOME was lower in other physiological and pathogenic conditions that aggravate the risk of cardiovascular disease, such as hyperlipidemia [127] and aging [128]; and is inversely correlated to BMI, total and visceral fat percentages, diastolic blood pressure, glucose, insulin and HOMA-IR, triglycerides and total-cholesterol, together with the positive correlation to HDL-cholesterol, after adjusting for age and sex [113].

This experimental and epidemiological data highlight the potential for exercise-derived 12,13-diHOME to regulate cardiac and vascular activity. The regulation occurs both directly through NOS signaling in mice, and indirectly by improving lipid metabolism in mice and humans. The specific effects of 12,13-diHOME in the modulation of endothelial function by NO signaling are promising, and its putative effects on redox homeostasis call for further exploration.

2. Muscle-derived exerkines that interact with the redox system

2.1. Interleukin-6 (IL-6)

In muscle, exercise increases IL-6 expression and secretion in a manner that is directly proportional to muscle contraction. Therefore, muscle-derived IL-6 relies on the duration and amount of muscle engaged in the exercise [129]. Acting to promote whole-body glucose uptake and lipolysis, and muscle FA oxidation, mRNA expression and protein release levels of IL-6 increase when muscle glycogen reserves are depleted [129]. Because exercise-induced IL-6 increases fuel supply by stimulating gluconeogenesis in the liver and lipolysis in adipose tissue, IL-6 has been suggested to act as an energy sensor for the muscle [130]. Initially, the increments in plasma IL-6 after exercise were associated with tissue damage due to its proinflammatory nature [129]. In this sense, IL-6 levels increase after an acute bout of exercise but decrease after chronic exercise training, and are lower in trained than sedentary subjects [131]. This, in turn, suggests IL-6 has a major role in exercise-induced metabolic adaptations.

IL-6 expression is also linked to redox homeostasis, since elements that regulate the promoter region of IL-6 are embedded in an antioxidant response element that is present in human isoforms of NADPH-oxidase, subunits of rodent glutathione-S-transferase, and rat glutathione-S-transferase-P. Therefore, ROS sensors NF-kB and the antioxidant transcription factor Nrf2 can both bind to Il6 promoter region and increase its expression, and are in fact necessary for IL-6 expression in myotubes in vitro and liver in vivo, respectively [132,133]. Conversely, IL-6 stimulates Nrf2 signaling in cardiomyocytes to increase their antioxidant capacity and protect them from the oxidative damage induced by LPS injections [134]. While IL-6-Nrf2 signaling has not been explored in exercise training or in skeletal muscle, more studies support IL-6 role as an antioxidant mediator in other cell types such as pancreatic β-cells [135]. There, IL-6 protects the cell from oxidative damage by regulating mitophagy and selective autophagy by stimulating Nrf2 removal of the antioxidant repressor Keap 1 [135]. Some studies support a role for exercise-induced IL-6 to protect pancreatic β-cells from oxidative stress, since serum from exercised but not from Il6-KO animals improve cell viability by reducing nitrite and NOS signaling [136]. Exercise-derived IL-6 also protects from insults that lead to alcoholic fatty liver disease, acting both against disease development and contributing to the healing process by decreasing oxidative and inflammatory insults through the modulation of the IL-6–p47^phox oxidative–stress axis [137]. More research is needed to understand the mechanisms of IL-6 as a repressor of pro-oxidant insults, specifically in disease development and healing, as well as the impact of exercise in both contexts.

Interestingly, some evidence shows the effects of IL-6 in the redox homeostasis in skeletal muscle. One study shows that, during exhaustive exercise, IL-6 modulates the redox balance by regulating ER stress homeostasis [138]. In this study, Il6-KO mice displayed lower time to exhaustion, and higher levels of ER stress proteins in soleus and EDL, than WT mice [138]. These findings support the role of this cytokine to mediate exercise adaptations in skeletal muscle through oxidative stress signaling. Other studies involving IL-6 and redox signaling in skeletal muscle have shown that extracellular IL-6 governs mitochondria physiology in cultured myotubes in a time and dose-dependent manner via NF-kB [139]. These effects result in increased stimulated mitochondrial ROS production that amplifies IL-6 signaling and activates a cascade of events leading to acute mitochondrial biogenesis, fusion, and increased respiration [139]. However, this is followed by a decline in respiration and further increments in mitochondrial ROS and oxidative stress when IL-6 exposure is sustained [139]. Finally, recent studies have also revealed that deregulated amounts of IL-6 contribute to the alteration of muscle and systemic redox signaling markers in mice with transgenic expression of human IL-6 [140]. These mice present an increase in the catalytic subunit of NADPH oxidase and in ROS levels, a reduction in the expression of genes associated to the activity of Nrf2, and wider atrophic areas in EDL and diaphragm [140]. These results show that IL-6 redox signaling in muscle is tightly controlled by different cellular processes and compartments and is needed not only for nutrient sensing during exercise training, but also for the control of oxidative stress to regulate exercise adaptations by influencing mitochondrial and myofiber activity and integrity.

2.2. Myostatin

Myostatin or growth/differentiator factor-8 is a member of the transforming growth factor-β (TGF-β) family that is expressed in a variety of tissues, but mainly in developing and adult skeletal muscle [141]. Myostatin inhibits muscle growth, which has been demonstrated in mice deficient in myostatin [141], and in transgenic mice overexpressing the myostatin propeptide, a truncated form of its receptors (Ac RIIB), or an inhibitor of the receptor (follistatin) [142]. These mice are larger and have skeletal muscle hypertrophy and hyperplasia. Conversely, transgenic overexpression of myostatin in skeletal muscle [143] or administration of myostatin [144] result in cachexia. The phenotype of myostatin deficient mice is the most well-studied and show hypertrophy due to an increase in the number and size of muscle fiber, not an increase in the number or activity of myofiber precursors [145]. Despite this, Mstn null mice show an impaired contractile profile compared to WT mice that is associated with lower force and power generation [[146], [147], [148]]. While this could be partly explained by a higher ratio of fast glycolytic to slow oxidative myofibers, which increase fatigue [147,149,150], Mstn null mice also have a lower mitochondrial density and activity, with reduced cardiolipin content and ratio of mitochondrial to nuclear DNA [147,148,151]. Therefore, mice and human studies show that circulating myostatin and/or muscle myostatin mRNA levels are lower after an acute bout of exercise and after chronic exercise protocols (reviewed by Allen et al. [152]). Overall, studies on null mice suggest that myostatin activity is crucial to couple exercise adaptations to myofiber size, type, and mitochondrial content. Importantly, the Mstn null mice also show an increase in bone formation that occurs mainly by loading effects, but also by the regulation of osteoclastogenesis, the main step in bone remodeling [153], which suggest a role for this myokine in the structural adaptations to exercise.

The mechanisms that lead to muscle hypertrophy and hyperplasia in Mstn null mice include several signaling pathways that involve Smad signaling, main receptor of the TGF-β family [154]. TGF-β/Smad signaling has been shown to be regulated in a feed-forward manner by different redox pathways, as reviewed by Jiang et al. [155]. Moreover, myostatin signals through MAPKs p38 and ERK pathways to inhibit myoblast proliferation and differentiation in C2C12 cells [156,157] and increase ROS production [158]. The induction of ROS in vitro leads to cytokine signaling through TNF-α to NF-kB and NADPH oxidase, and in turn higher H₂O₂ levels induce myostatin, which suggests a feed-forward loop for myostatin/ROS signaling [159]. Studies in Smad null mice have also shown ROS modulation of myostatin signaling. Smad3 null mice display higher level of ROS in skeletal muscle that are partially negated when myostatin is also deficient. Moreover, in Smad3 null mice myostatin signals via MAPKs p38 and ERK that lead to higher oxidant activity and ROS levels in C2C12 cells [154]. Smad3 null mice also show stimulation of MuRF1 signaling and muscle atrophy that is negated in Smad3 null mice deficient of myostatin, pointing to ROS signaling modulation of myostatin effects in muscle atrophy [[154], [160]]. Moreover, muscle myostatin mRNA increases along the muscle regeneration process after muscle injury in Smad3 null mice, which exhibit more ROS production and less mitochondrial biogenesis, suggesting that myostatin could signal again via the alternate pathway of MAPKs p38 and ERK to modulate mitochondrial biogenesis [160]. In this regard, myostatin also increments protein levels of main regulators of mitochondrial fission and mitophagy (Drp1, Fis1, Parkin) in C2C12 cells [161].

The negative effects of myostatin in the redox system appear reflected in the enhanced antioxidant activity observed in Mstn null mice by increased GSH and glutathione peroxidase, and lower levels of lipid peroxidation in gastrocnemius [148]. However, Mstn null mice also show reduced antioxidant enzymes catalase, SOD, and glutaredoxin in gastrocnemius [148] together with lower NADH reductase in EDL muscle [147]. Therefore, it could be hypothesized that myostatin regulates mitochondrial recycling and myofiber growth in response to exercise, which also result in the regulation of redox homeostasis. Evidence to date has focused on genetic ablation of myostatin and its signaling transducers or in vitro studies that don't allow further conclusions. Considering the described pathways also mediate myostatin regulation of bone loss [153], it is crucial to bridge the gap between redox biology and myostatin in exercise adaptations.

2.3. Lactate

Lactate is the major end product of glycolysis, and therefore lactate exhibits prominent changes during and immediately after endurance and resistance exercise in human multi-omic studies [109,110]. Discovered in the 19th century, lactate was considered responsible for the exercise-induced muscle fatigue due to an increase in acidosis [162]. However, since the 1970s many studies have shown that lactate is an exerkine that regulates adipose tissue metabolism, cardiac output, and mitochondrial FA uptake, stimulates angiogenesis, mitochondrial biogenesis, and the release of BDNF, and is the major fuel to the heart [163]. More recently, a new metabolite derived from lactate and named lactate-phenylalanine (Lac-Phe) was found to be the most changed in the metabolome of animals and humans after exercising, acting to suppress feeding and obesity [164].

In vitro, the increase in lactate levels (20 mM) lead to the production of H₂O₂ and results in an increased expression of Pgc1a and mitochondrial proteins (COX) in L6 myoblast cells, which originate from rat skeletal muscle [165]. This is consistent with in vivo studies showing that intraperitoneal lactate administration increases mitochondrial genes and content in mouse liver, brain, and gastrocnemius [166,167]. Conversely, similar lactate concentrations (30 mM) act as a ROS scavenger by reducing the levels of superoxide, hydroxyl anion, and MDA in cultured rat hepatocytes [168]. These results highlight the complex interactions between lactate and the redox system, which occur at the systemic level. Specific associations of lactate with the redox system in the exercised muscle rely on the ratio of lactate to pyruvate, which increases more than 2 orders of magnitude during exercise, and leads to an increase in the ratio of the redox couple NAD⁺/NADH. During exercise, lactate is also oxidized to pyruvate for gluconeogenesis in muscle and liver, which requires mitochondrial transport of lactate by lactate shuttles that is driven by a specific redox state regulated by NAD⁺/NADH [169]. Considering the role of NAD in cellular bioenergetics, metabolism, and redox homeostasis, exercise-derived lactate indirectly affects numerous processes via changing redox signaling [9,170].

The association between lactate metabolism and NAD⁺/NADH is also relevant to exercise when the role of NAD⁺ as cofactor for sirtuins is considered. The increase in the expression of SIRT1 relies on NAD⁺/NADH free levels in the sarcolemma [171], and exercise-induced changes in lactate have been reported to increase mitochondrial biogenesis in the brain by NAD⁺/NADH signaling through lactate shuttles [172]. Importantly, lactate signaling in the brain leads to the synthesis of BDNF, which mediates the effects of exercise in the brain [75]. Recent studies have shown that lactate regulates redox signaling in anorexigenic neurons to regulate feeding [173]. Moreover, lactate leads to the production of FGF21 in adipose tissue and muscle cells [174,175]. The effects on adipose tissue have been demonstrated in vivo in a MAPK p38-dependent manner [174], which plays central roles in cellular stress. Moreover, the changes in NAD⁺/NADH also induce FGF21 and therefore could entail a redox-dependent mechanism for lactate to induce these effects [70]. These presumed effects of lactate in adipose tissue could lead to a loop for increased redox homeostasis, since FGF21 expression in adipose tissue induces a thermogenic phenotype that is energetically more efficient [70].

All in all, several mechanisms regulated by lactate act to regulate redox homeostasis and redox-regulated metabolic homeostasis. Despite lactate now being considered a redox-signaling metabolite, only a few studies have explored this role in the context of exercise training. Recent findings for lactate in the modulation of feeding by neural redox signaling, and the recently discovered feeding Lac-Phe are thrilling and warrant further research to study how exercise training modulates the effects of both metabolites via redox biology.

3. Conclusions

The present review summarizes the effects of exerkines that depend on or modulate redox signaling and homeostasis. Here, the current evidence pictures a landscape where exerkines exert a complex regulation of redox biology, since their signaling roles mediate inter-organ cross-talk and tissue-specific effects on redox pathways by mediating ER and mitochondrial stress, ROS signaling, and metabolic efficiency that result in lower oxidative stress. The studies presented in this review strongly indicate that the redox biology of these exerkines acts to mediate the beneficial effects of exercise by improving the adaptations to exercise and health outcomes in disease.

In this sense, all the exerkines reviewed mediate mitochondrial dynamics in skeletal muscle by pathways that involve redox signaling. Among them, myostatin and BDNF are the most studied, and show tight regulation of the muscle phenotype including the mitochondrial density, fiber type, contractile profile, and hypertrophy. BDNF has been recently been shown to modulate muscle mitochondrial function, density, and exercise capacity by injection of recombinant human BDNF to MI mice and co-treatment with a blocker of the receptor in myotubes [98]. Similar studies are needed on the effects of myostatin and the rest of the exerkines that have shown strong roles for the maintenance of muscle mitochondrial stress response and fiber integrity (IL-6, FGF21) in rodent models of muscle dystrophy and gain and loss of function studies [40,140]. A complete picture for the roles of the redox biology of these exerkines in the exercise-induced muscle adaptations is essential also for a healthy aging, considering the main differences between the lifelong exercised and the sedentary, aged muscle entail pathways for antioxidant activity and handling of cellular stress [102]. Moreover, it is important to note that the exerkines effects associated to redox biology have been demonstrated mainly for muscle-derived exerkines that act in an autocrine and endocrine manner (Fig. 1). Therefore, maintaining exerkines functions will largely depend on the maintenance of a healthy skeletal muscle.

Fig. 1.

Exercise effects are mediated by the action of exerkines. Exerkines are dependent on, and modulate, redox biology. The left side of the figure shows the proposed regulatory circuits for the production of exerkines and their effects on different tissues to modulate redox biology. The right side of the figure indicates the effects of each exerkine (blue panels), the target tissue affected (green panels), and if exercise is necessary for the effect to take place (Ex, yellow pannels).♢ indicates effects that interact with or modulate redox homeostasis, ♦ indicates effects that modulate redox homeostasis to reduce oxidative stress.

On the other hand, most of the exerkines reviewed show cardioprotective actions that mediate the beneficial effects of exercise by modulating redox homeostasis and reducing oxidative stress-induced cardiac impairments. This has been proven in several cellular processes involved in the pathology of different models of cardiac disease using ample and rigorous experimental approaches, and consolidates the major role for exercise to improve cardiovascular health also by exerkines. To date, exerkines have proven these actions by exerting autocrine effects and when secreted from liver and adipose tissue. It is very likely that future studies will show a role for brain and muscle-derived exerkines to have a cardioprotective role by modulating redox biology.

The functions of exerkines to communicate with the brain and induce BDNF in an organ-central nervous system cross-talk has been poorly studied. Only a few studies have shown that exerkines like lactate can cross the blood-brain barrier and induce BDNF in the brain [74,75]. Moreover, despite these exerkines modulate redox signaling and neural activity, and despite the role of redox signaling in the brain, the role of exerkines in the central nervous system has been studied only in a descriptive manner. Future research should shed light on this organ-central nervous system signaling and on the role of redox biology in it, since it will likely govern the rest of the responses in exerkines synthesis and release.

Declaration of competing interest

The authors declare no conflicts of interest.

Data availability

Data will be made available on request.