Exercise and the Cisd2 Prolongevity Gene: Two Promising Strategies to Delay the Aging of Skeletal Muscle

Aging is an evolutionally conserved process that limits life activity. Cellular aging is the result of accumulated genetic damage, epigenetic damage and molecular exhaustion, as well as altered inter-cellular communication; these lead to impaired organ function and increased vulnerability to death. Skeletal muscle constitutes ~40% of the human body’s mass. In addition to maintaining skeletal structure and allowing locomotion, which enables essential daily activities to be completed, skeletal muscle also plays major roles in thermogenesis, metabolism and the functioning of the endocrine system. Unlike many other organs that have a defined size once adulthood is reached, skeletal muscle is able to alter its structural and functional properties in response to changes in environmental conditions. Muscle mass usually remains stable during early life; however, it begins to decline at a rate of ~1% year in men and ~0.5% in women after the age of 50 years. On the other hand, different exercise training regimens are able to restore muscle homeostasis at the molecular, cellular and organismal levels, thereby improving systemic health. Here we give an overview of the molecular factors that contribute to lifespan and healthspan, and discuss the effects of the longevity gene Cisd2 and middle-to-old age exercise on muscle metabolism and changes in the muscle transcriptome in mice during very old age.


Introduction
Aging is inevitable and affects all living organisms. According to the first World Report on Aging and Health published in 2015, "At a biological level, ageing is associated with the gradual accumulation of a wide variety of molecular and cellular damage. Over time, this damage leads to a gradual decrease in physiological reserves, an increased risk of many diseases, and a general decline in the capacity of the individual. Ultimately, it will result in death." Due to the complexity of these biological changes that lead to the functional decline of various organs, the biological changes associated with aging are "neither linear nor consistent, and they are only loosely associated with age in years". In addition to functional deterioration, aging is the primary risk factor for a range of human pathogenic diseases, including the development of cancer, diabetes, cardiovascular disorders and various neurodegenerative diseases [1].

Resting Metabolism
Resting energy expenditure (REE) is the amount of energy expended by an organism at rest. REE accounts for between 60% and 70% of total energy expenditure (TEE). Historically, obese individuals were believed to have lower energy expenditure (EE) rates than non-obese individuals (normal and overweight). It was observed that REE is related most directly to the amount of lean mass, because lean mass is more metabolically active than fat tissue. Nevertheless, current evidence has revealed that obesity is unlikely to be a result of a consistently lower daily EE or REE [16], because obese persons tend to have increased amounts of fat-free mass to support their increased fat mass. Recent studies have indicated that REE, adjusted for fat-free mass, which mainly contains skeletal muscles, is normal in both obese persons as well as normal-weight persons who are predisposed to obesity [17,18]. On the other hand, sustained weight loss may not always result in a substantial and disproportionately low REE [19]. Even among those of similar weight, age and gender, REE varies markedly. However, men generally have higher REE values than women, even after adjusting for body weight and fat-free mass. In addition to the gender effect, increasing age is associated with a decreasing REE; this is due to at least in part age-dependent loss of fat-free mass [20].

Exercise Metabolism
Exercise alters a variety of cell types and is involved in multiple signaling pathways that regulate systemic metabolism. During exercise, the relative levels of oxidation of fat and carbohydrate may be influenced by a range of factors, such as diet, muscle glycogen content, exercise intensity/duration, type of exercise and fitness status [21]. An acute bout of exercise increases the glucose uptake of skeletal muscle, while chronic exercise training improves mitochondrial function, promotes mitochondrial biogenesis and upregulates the expression of glucose transporter proteins and numerous other metabolic genes [22].
During low-intensity, steady-state exercise, RER values are typically between 0.80 and 0.88, meaning that fatty acids are the primary fuel. As the intensity of the exercise increases, carbohydrates become the dominant or primary fuel and the RER increases to between 0.9 and 1.0. However, strenuous exercise gives rise to hyperventilation and increased a blood lactate level, which is released from skeletal muscles, resulting in an elevation of CO 2 production. Under this circumstance, the RER would exceed 1.0 and no longer reflect systemic substrate utilization [23]. RER reproducibly increases during exercise and therefore it has been identified as a parameter that can document maximal effort. The rate at which oxygen consumption adjusts to and recovers from a bout of exercise is associated with certain metabolic processes, including muscle storage of phosphocreatine and ATP, the replenishing of myoglobin and hemoglobin as well as the removal of metabolic waste. The ability of an individual to utilize these processes determines the speed at which oxygen consumption rises and falls, and thus these have been found to be associated with the health condition of a given individual [24,25].

Impact of Middle-to-Old Age Exercise on Metabolism
We sought to determine the impact of middle-to-old age regular exercise on energy metabolism in very old mice (>24 months, which is equivalent to >80 years old in human) [26]. We adopted a forced treadmill exercise protocol that has been shown to produce an overt effect on the cardiorespiratory and metabolic benefits similar to those observed in humans while exercising [27,28]. Male and female C57BL/6 mice at 15-month-old of age (equivalent to a 50-year-old in humans) were subjected to two cycles of 12-weeks forced treadmill exercise on a 5-on-2-off schedule. Mice were rested for three months before the initiation of a similar second cycle of 12-weeks exercise. The exercise intensity was reduced after the third week, and was further modified according to the physical performance of the mice on treadmill ( Figure 1A). The energy metabolism of the exercise mice and the sedentary mice were monitored using a TSE metabolic cage system, and the metabolic parameters of 3-months old (i.e., young) mice that had never been on a treadmill were also measured as a comparison. Exercise did not seem to result in a significant impact on the oxygen consumption and CO 2 production of the female mice. On the other hand, the effect of the animal's circadian rhythm on oxygen consumption and heat production, when the dark and light hours of male mice are compared, appears to have a more prominent effect on the exercise group than the sedentary group ( Figure 1B). In addition, the VO 2 of exercised mice (i.e., Old-Exe) is closer to the young group than the sedentary mice (Old-Sed) when male mice were considered. On the other hand, the RER values of the male Old-Exe group are higher than the male Old-Sed group during the active hours; however this observation was not seen for the female mice ( Figure 1B). These findings suggest that regular exercise is able to prompt the male mouse body to use carbohydrates as a fuel source during the active hours, but this phenomenon does not seem to occur in female mice. and CO2 production of the female mice. On the other hand, the effect of the animal's circadian rhythm on oxygen consumption and heat production, when the dark and light hours of male mice are compared, appears to have a more prominent effect on the exercise group than the sedentary group ( Figure 1B). In addition, the VO2 of exercised mice (i.e., Old-Exe) is closer to the young group than the sedentary mice (Old-Sed) when male mice were considered. On the other hand, the RER values of the male Old-Exe group are higher than the male Old-Sed group during the active hours; however this observation was not seen for the female mice ( Figure 1B). These findings suggest that regular exercise is able to prompt the male mouse body to use carbohydrates as a fuel source during the active hours, but this phenomenon does not seem to occur in female mice.

Body Composition Measurement Techniques
At the molecular level, the human body is mainly composed of four components: water, fat, proteins and minerals [29]. A number of different techniques, both invasive and non-invasive, have been developed to allow body composition to be assessed. These techniques range from simple indirect measures, such as the waist-to-hip ratio, to more accurate direct volumetric measurements based on three-dimensional imaging [30]. The most commonly used methods for body composition measurement are bioelectrical impedance analysis, dilution techniques, air displacement plethysmography, dual energy X-ray absorptiometry (DEXA) and magnetic resonance imaging (MRI). These techniques allow us to measure fat (white and brown adipose tissues), lean mass, bone mineral content, total body water, extracellular water and various adipose tissue sub-depots (namely visceral, subcutaneous and intermuscular), skeletal muscle, selected organs and ectopic fat depots [31,32]. The measurements from DEXA and quantitative MRIs showed good agreement; they have linear correlations ranging from 0.97 and 0.99, and coefficients of variation ranging from 4.5% and 4.6% for fat (computed from adipose tissue, AT) and lean tissue (LT), based on a UK Biobank imaging cohort study of 4753 subjects. Quantitative MRIs also are able to measure muscle volumes and adipose tissues that ectopically infiltrated into skeletal muscles. When combined with rapid scanning protocols and efficient image analysis, quantitative MRI becomes a powerful tool for advanced body composition assessment [30].

Body Composition and Aging
Human body composition plays an important role in quantifying health and nutritional status, assessing the impact of disease, and measuring changes due to nutritional, therapeutic or behavioral interventions. Body composition during aging is characterized by an increase in fat mass and decrease in lean tissues, including skeletal muscle mass, which in older adults results in reduced muscle strength and poorer functional capabilities, as well as greater morbidity and mortality [32].
Using a Bruker Minispec LF50 TD (time-domain)-NMR Body Composition Analyzer, we assessed the impact of exercise at middle-to-old age on the body composition of very old (25-26 months) mice. Three components of the mouse body were obtained; these were fat, free body fluid and lean tissue. Compare to the Young group, the reduction in percent lean mass was significant for the Old-Sed male mice (p < 0.01, Figure 1C), and this was paralleled by a significant increase in fat mass (p < 0.05, Figure 1D). Exercise during middle-to-old age among male mice was able to maintain the percentage of lean mass well, and the percentage of fat mass remained similar to that of the young group. Intriguingly, net lean mass and net fat mass both increased among the Old-Sed male mice. Among female mice, middle-to-old age slightly reduced the percentage of fat mass (p < 0.08) compared to their sedentary cohorts ( Figure 1D), whereas the percent lean mass did not differ much between the young, Old-Exe and Old-Sed groups ( Figure 1C). These findings indicate that exercise during middle-to-old age has different physiological consequences on the body composition of male and female mice.

Exercise and Sexual Dimorphism
Sexual dimorphism during exercise has been demonstrated and results in higher glycerol and free fatty acid (FFA) responses among women compared to men, together with a higher carbohydrate oxidation rate among men. Women seem to have a greater reliance on lipid metabolism during exercise, while men exhibit a preference for carbohydrate utilization [33]. The mechanisms that are able to increase the lipolytic rates of women include a higher total fat mass, an enhanced lipolytic sensitivity to epinephrine and an increased activation of the β-adrenergic receptors [34].
In a study that evaluated the impact on C57BL/6 mice of sexual dimorphism and of the estrous cycle on muscle strength and running power using Ergospirometry (cardiopulmonary exercise testing) [35], the results indicated that thermoregulation was more efficient in male mice than in female mice. On the other hand, exercise-induced increase of O 2 consumption and CO 2 production were similar between the sexes. In females, pro-estrus impaired economy during running and estrus impaired exercise heat loss [36]. This study also indicated that these sex-associated phenomena that respond to exercise disappeared after further normalization of exercise performance by body weight. Previous studies have revealed that body weight, lean mass and muscle strength are secondary sexual characteristics and therefore justification factors need to be considered when comparing the differences in the running performance between the sexes [36].
Gender is a key determinant for resting lipid metabolism. Additionally, there is heterogeneity of post-exercise resting metabolism in humans. A meta-analysis has indicated that men experience a greater increase in resting metabolic rate (RMR) than women after a bout of endurance exercise [37]. It would seem that women are better able to resume normal resting metabolic parameters after exercise, whereas, metabolism of men remains changed to a significant degree for longer. It is possible that men undergo a higher degree of respiratory uncoupling or a higher metabolic burden when processes such as lipolysis and gluconeogenesis take place after exercise. This could be viewed as there being a higher level of precise metabolic control in women than in men, as women reach the resting rates of lipolysis and hepatic glucose production after exercise much quicker than men [33,37,38].
In our studies using 25-26 months old mice that had undertaken regular treadmill exercise during middle-to-old age, a significant gender difference in resting metabolism after exercise was observed ( Figure 1B). As noted above, exercised male mice tend to use carbohydrates as a fuel source during the dark (active) hours. By way of contrast, in female mice, RER was relatively unaffected during the active hours; however, exercised female mice do tend to consume more fat as an energy source (RER < 0.85, Figure 1B) during the light (sleep) hours. These findings clearly show that there is a greater reliance among female mice than among male mice on fat as an energy substrate during rest post exercise [33].

Long-Lived Mouse Models
A long-lived phenotype was first observed in the growth-retarded Ames dwarf mice and Snell dwarf mice, which have mutations of the Prop1 and Prou1f1 (Pit1) genes, respectively [39][40][41]. Later, these observations were able to link lifespan control and the best studied aging signaling pathway, the GH (growth hormone)/IGF-1(insulin-like growth factor-1)/Insulin cascade. Furthermore, extended lifespan has been observed in genetic engineered mouse models where there was loss of genes involved in insulin signaling, examples being GHR/BP (growth hormone receptor/GH binding protein) knockout (KO) mice [42], Igf1r (insulin-like growth factor type 1 receptor) heterozygous KO mice [43][44][45], liver-specific Igf1 KO mice [46], Irs1 (Insulin receptor substrate 1) KO mice [47], neuron-specific Irs2 (Insulin receptor substrate 2) KO mice [48] and adipocyte-specific insulin receptor KO mice [49]. Moreover, an indirect suppression of Irs1 and Irs2 via inactivation of S6K1 (ribosomal protein S6 kinase 1) in mice was found to lead to an increased lifespan [50]. Another long-lived mouse model is the Sirt6 transgenic mouse model, where lower serum levels of Igf1 are found, and there are alterations in the phosphorylation levels of major components of the IGF1 signaling pathway [51]. In addition, upstream hormone-like factors, such as Fgf21 (fibroblast growth factor 21) overexpression [52], hNAG-1/Gdf15 (growth differentiation factor 15) overexpression [53] and MIF (macrophage migration inhibitory factor) loss of function [54], have been found to extend the lifespan of mice by reducing the serum level of Igf1. Despite of these observations, it remains under debate whether insulin resistance or insulin sensitivity is better linked to longevity [43,50,52,53,[55][56][57].
In addition to the GH/IGF/insulin pathway, a growing number of long-lived mouse models have provided further clues to the metabolic alterations that occur during aging, though the molecular mechanisms of these networks are tightly connected. Overexpression of a thermogenic gene Ucp1 (uncoupling protein 1) in muscle is able to reduce body weight and body fat mass; this is brought about by creating an increase in body temperature. The ectopic overexpression of Ucp1 mimics nutritional deprivation by bringing about activation of AMPK, as well as inactivation of the mTOR signaling cascade; these lead to an increase of medial lifespan of 3 months [58]. Another report showed that a deficiency in cAMP-dependent protein kinase A (PKA) in mice leads to an extended lifespan, increased metabolic activity and a higher body temperature [59]. Pten transgenic mice are also long-lived, and show an elevated resting metabolic rate, as well as reduced fat mass. The overexpression of Pten activates the expression of Ucp-1 and Pgc-1a, which result in a concomitant increase in glucose uptake and oxygen consumption by brown adipose tissue [60]. Another example of a long-lived mouse occurs when there is brain-specific overexpression of Sirt1 (i.e., BRASTO). Aged BRASTO mice display significantly enhanced physical activity, a higher body temperature and increased oxygen consumption; these effects have been attributed to Sirt1-mediated orexin type 2 receptor (Ox2r) upregulation and the colocalization of Ox2r with Sirt1 in the dorsomedial and lateral hypothalamic nuclei [61]. In the genetically modified mice described above, one common feature is a decreased body mass that is accompanied by reduced adiposity [46,49,50,53,58,59,62]. The casual relationship between metabolism and lifespan is still not fully understood and this is despite the observation that mice with a high metabolic activity show greater mitochondrial uncoupling at a young age and live longer [63]. In addition to genetic modification, an evolutionarily-conserved and highly reproducible intervention that is able to extend lifespan is calorie restriction. In mice, diet restriction at either 1-month-old or 1-year-old changes the systemic metabolism of the mice, leading to a prolonged lifespan [54,64]. Exercise also alters metabolism in an organism [65,66]. Pharmaceutical research has shown that a mitochondrial uncoupler, namely 2,4-dinitrophenol, decreases the efficiency of energy conversion in a similar manner to caloric restriction and this can promote in various tissues enhanced tissue respiratory rates, improved serological glucose levels, better triglyceride levels, better insulin levels, decreased reactive oxygen species levels, reduced DNA and reduced protein oxidation, as well as reducing body weight. Importantly, 2,4-dinitrophenol-treated animals also presented with enhanced longevity [67].
Removal of harmful intracellular material has also been found to be as a key factor when trying to improve cellular and organismal health, thereby preventing aging. The p66 Shc adaptor protein is a newly recognized mitochondrial protein that oxidizes cytochrome c and forms reactive oxygen species (ROS) by utilizing electrons during respiratory chain dysfunction [68]. Mice that have a lower level of p66 shc are long-lived, and display increased resistance when subjected to various cellular stresses, including hydrogen peroxide, UV irradiation and paraquat [69]. Mitochondrial catalases are well studied enzymes that play various critical roles in protecting cells against the toxic effect of hydrogen peroxide. A transgenic mouse overexpressing a human mitochondrially-targeted catalase (MCAT) was generated to test the free radical theory of aging. Indeed, the MCAT transgenic mice showed increased resistance against ROS stress and there was also less accumulation of oxidative damage, as well as extended lifespan [70]. Adenylyl cyclase type 5 (AC5) is expressed in striatal medium spiny neurons, and is known to mediate G protein-coupled receptor signaling through the production of the second messenger cAMP [71]. AC5 KO mice are long-lived, are resistant to cardiac stress and have an increased median lifespan of approximately 30% [72]. AC5 KO mice are protected from reduced bone density, are less susceptible to fractures caused by aging and show less age-dependent decline in cardiac function [72]. Activation of autophagy is an alternative signaling pathway for extending lifespan in mice. The autophagy related 5 (ATG5) transgenic mice that have enhanced autophagy show an extension of their median lifespans by 17.2% and are metabolically healthier than the control cohort mice at 18-month old [62]. By expressing the Beclin1 F121A mutation, the interaction between Beclin1 and Bcl2 is disrupted and autophagy is stimulated. This activation of autophagy results in improved age-associated pathology within the kidneys and heart. The medial survival of both male and female Beclin1 F121A mutant mice is extended by 3 months, compared to gender-matched controls [73].
Taking all of the above genetic, behavioral and chemical intervention studies that use mouse models together, their findings indicate that insulin signaling, energy metabolism, the removal of harmful oxidative stress and a reduction in damaged organelles are the keys to longevity.

Cisd2 in Aging
Cisd2 (CDGSH Iron Sulfur Domain 2) is an oxidative stress-sensitive gene, the expression of which is able to prolong the lifespan in mice [74]. Cisd2 loss-of-function mice exhibit premature aging phenotypes and have a shortened lifespan. Conversely, Cisd2 transgenic mice not only are longer lived (both males and females), they also have a healthier physical condition, such as better fur function, increased muscle strength and improved cardiac function [75,76]. The Cisd2 protein has been localized to mitochondria, mitochondria-associated membrane (MAM) and the endoplasmic reticulum (ER), and is involved in calcium homeostasis [77]. We have demonstrated using aged mice that maintenance of the expression level of Cisd2 sustains metabolic activity, ameliorates aging-associated mitochondrial dysregulation, reduces DNA damages and improves the calcium imbalance within skeletal muscles [75], liver [77] and heart [76]. Taken together, Cisd2 is a lifespan regulator and its expression level seems to be a critical factor in relation to prolong healthspan.

Cisd2 and Exercise
Cisd2 and exercise have both been reported to contribute to an extended lifespan and to improve healthspan. A recent report has shown that the protein levels of Cisd1 and Cisd2 in skeletal muscle and white adipose tissue are increased approximately 1.5-fold and 1.2-fold after 4 weeks of voluntary excise in mice [78]. The authors also observed that there were significant increases in the levels of multiple mitochondrial proteins, which agrees with our previous discovery of increased mitochondrial number in the muscle of Cisd2 transgenic mice compared to their wild type cohorts [75]. To examine the transcription of Cisd2 in real-time, here we have generated a Cisd2 reporter transgenic mouse that carries luciferase as the reporter, which is under the control of the mouse Cisd2 endogenous upstream 31.3-kb promoter sequence (i.e., a Cisd2-Luc mouse). In addition to the endogenous promoter, a 46.9-kb region downstream of Cisd2 coding region was also included in the transgene construct in order to preserve the regulatory elements of the Cisd2 gene. The Cisd2-Luc reporter mice were trained on a treadmill for 56 days. It was found that a drastic enhancement in Cisd2 transcription was able to be observed. The most intense signal was observed at the abdomen, with the thymus showing the next largest increase in signal. A moderately increase in signal was also observed in forelimbs and hindlimbs (Figure 2A).
We further analyzed the histology of the femoris muscle of male and female Old-Sed and Old-Exe mice as described in Figure 1, and compare these to the Cisd2-transgenic (Cisd2-TG) mice at 24-26 months old as reported in Shen et al. [77]. Quadriceps femoris is composed of four individual skeletal muscles, rectus femoris, vastus medialis, vastus lateralis and vastus intermedius. The quadriceps femoris was chosen for this study because of its mass, the fiber type composition and our recent use in other studies. The quadriceps femoris is the largest muscle present in the mouse hindlimb by mass [79]. Furthermore, the quadriceps femoris muscle in terms of fiber types represent the majority of skeletal muscle. The fiber type of the quadriceps femoris muscle is IIx/IIb in rodents and IA/IIA and IIA/IIX in humans, which are commonly referred to as fast-twitch fibers. Most other skeletal muscle in limbs, other than the soleus muscle, have a similar fiber type composition both in humans and other mammals [80]. Finally, the quadricep femoris has often been used to study exercise [80] and the effects of aging [77,81,82] in mouse and humans. In agreement with the results on systemic metabolism ( Figure 1B), we found a gender difference in the histology of skeletal muscle, particularly in the Old-Sed mice. In the male mice, multiple vacuolated fibers were observed in the skeletal muscles of these old sedentary mice. This pathology was obviously attenuated in old Cisd2-TG male mice [75] and the Old-Exe male mice ( Figure 2B). Additionally, central nucleation was seen in the femoris muscle of Cisd2-TG (27.3 ± 9.15 % of total fiber, p = 0.07) and Old-Exe (33.6 ± 8.50 % of total fiber, p = 0.01) male mice ( Figure 2B, arrows). Central nucleation in muscle cells is an indicator of degeneration followed by regeneration. Accordingly, these findings suggest that exercise and an elevated level of Cisd2 are both able to enhance the repair of muscle fibers. However, in female mice, the age-related pathohistological alterations to the skeletal muscles is less obvious in the Old-Sed group and thus there appears to be no differences in pathology across the three groups of female mice ( Figure 2B). Notably, quantification of central nucleation in the female mice reveals that Cisd2-TG mice exhibited a significantly higher ratio of central nuclei in their muscle fibers (6.63 ± 1.0 % of total fiber) ( Figure 2B). This suggests that Cisd2 indeed has a beneficial effect on muscle regeneration within both males and females during aging.
To better understand the alternations in the transcriptome that are brought about by long-term exercise and/or Cisd2 overexpression, as well as which signaling pathways are potentially involved during muscle aging, we performed RNA-seq analyses by DEseq2 [83] on femoris muscles obtained from the various groups of mice, namely Old-Exe, Old-Sed and Cisd2-TG (24-26 months old) mice. Due to the substantial individual variation often observed in tissue samples collected from aged mice, a less stringent criteria [log2 fold change (LFC) > 0.14 and false discovery rate (FRD) < 0.1] were set in order to identify the differentially expressed (DE) genes. Comparing Old-Exe and Old-Sed femoris muscle, 19 DE genes and two DE genes were identified in males and females, respectively. On the other hand, only one DE gene, namely 4930539J05Rik, was found when Cisd2-TG and Old-Sed mice were compared among males, and no DE gene was identified among females. These findings correlate with the macroscopic observations obtained by histopathological analysis ( Figure 2B).
reporter mice were trained on a treadmill for 56 days. It was found that a drastic enhancement in Cisd2 transcription was able to be observed. The most intense signal was observed at the abdomen, with the thymus showing the next largest increase in signal. A moderately increase in signal was also observed in forelimbs and hindlimbs (Figure 2A). The data are presented as means ± SD. Statistical analysis used the Student's t-test. *, p < 0.05; **, p < 0.01. TG, Cisd2-TG; Sed, sedentary; Exe, exercise. (C) A heatmap showed the differentially expressed gene (RNA-Seq) clusters for the male and female femoris muscle that have been changed by longterm treadmill exercise. The RNA-seq data was analyzed by "DEseq2" and the DE genes were defined The treadmill protocol was 6 m/min for 5 min, 9 m/min for 5 min, 12 m/min for 20 min, 15 m/day for 5 min and 12 m/day for 5 min, for a total of 450 m/day. The Cisd2-Luc reporter mice ran on treadmills at a frequency of 5 days/week for 8 weeks (56 days). The quantification of luciferase intensity in each of the mice in the figure is shown at top of each image. Increased (fold change) Cisd2-Luc luciferase signals are numbered below the images; these represent mice after the exercise was completed. The photos were taken ventrally. (B) H&E staining of the femoris muscles of Cisd2-TG, Old-Sed and Old-Exe male and female mice. The arrows indicate myofibers that contain central nuclei. Bars: 30 µm. Percentage of central nucleation in each group of mice is presented on the right. The data are presented as means ± SD. Statistical analysis used the Student's t-test. *, p < 0.05; **, p < 0.01. TG, Cisd2-TG; Sed, sedentary; Exe, exercise. (C) A heatmap showed the differentially expressed gene (RNA-Seq) clusters for the male and female femoris muscle that have been changed by long-term treadmill exercise. The RNA-seq data was analyzed by "DEseq2" and the DE genes were defined as LFC > 0.14 and FDR < 0.1. The heatmap was generated using the "pheatmap" R package (Raivo Kolde (2019). pheatmap: Pretty Heatmaps. R package version 1.0.12. https://CRAN.R-project.org/package=pheatmap) according to rlog transformed count calculated by "DEseq2". A heatmap of the DE genes was plotted to explore the expression pattern of these genes across the three groups (i.e., Cisd2-TG, Old-Sed and Old-Exe, Figure 2C). There were 21 DE genes in total and these were able to be grouped into five clusters. Cluster I, II and III consisted of the DE genes that are significantly down-regulated in the Old-Exe group, compared to the Old-Sed group, and most of those DE genes encode sarcomere proteins. Of these, Myl3 (LFC = −5. 48 [84][85][86]; Myoz2 is a gene mediating fiber type switching, and Myoz2 deficiency has been reported to cause an increase in slow-twitch muscle fibers [87]; Ankrd2 encodes a I-band structure protein that also participates in embryonic myoblast differentiation [88][89][90]. The expression profile of Mup10, the only gene in cluster IV, is particularly interesting. Compared to male Old-Sed mice, Mup10 is significantly up-regulated in Old-Exe (LFC = 3.54, FDR = 0.008) and Cisd2-TG (LFC = 2.98, FDR = 0.201) male femoris muscle ( Figure 2C, blue frame). Mups (major urinary proteins) form a highly polymorphic gene family that are expressed in the liver. They are secreted into the bloodstream in order to regulate pheromone stability, and urine Mup profiles are able to function as individual identity signatures of the owners. Circulating MUPs serve as a metabolic signal that links the regulation of glucose and lipid metabolism [91]. Whereas the protein function of Mup10 has not been extensively explored, recombinant MUP1 has been shown to elevate energy expenditure markedly, ameliorate hyperglycemia and reduce glucose intolerance in mice with type 2 diabetes [91]. Intriguingly, the level of Mup10 transcripts in females is higher overall than in males ( Figure 2C, red frame), whereas many other reported MUPs show much higher transcription levels in males than in females [91]. These findings suggest that gender dimorphism of Mup10 in rodents is significantly associated with exercise.
Differences of the 13 DE genes in cluster V were moderate between the three experimental groups. Among these genes, Islr2 Ctxn3 (LFC = 1.66, FDR < 0.001) are all up-regulated in Old-Exe femoris muscle samples compared to the Old-Sed group. Of the last group of genes, it is known that Chst11 (carbohydrate sulfotransferase), a Wnt/β-cateinin down-stream target [92], participates in chondrocyte development and cartilage morphogenesis [93]. Collectively, our findings reveal that at middle-to-old age treadmill exercise creates a clear gender dimorphism in the transcriptome of the femoris muscle, and that this specifically bring about and causes expression changes in fast-twitch myosins, slow-twitch myosins and the secreted/excreted protein Mup10.

Conclusions and Perspectives
It has been well demonstrated in multicellular organisms that damage and other deleterious changes accumulate in evolutionarily conserved manner over time and these lead to aging. These changes are often irreversible; however, using genetically-modified mouse models, it has been discovered that a number of major molecular pathways, including insulin signaling, energy metabolism and removal of harmful oxidative stress, seem to govern longevity. Regular exercise has been well demonstrated to improve insulin sensitivity, increase metabolic function and induce autophagy across multiple organs. Here we have further demonstrated that metabolic function is improved in very old male and female mice that undertake treadmill exercise at middle-to-old age, although significant gender dimorphism is also obvious. These exciting discoveries, which relate exercise, metabolic regulation and healthspan, will help us to find new avenues that can be used as actionable interventions against aging (Figure 3).

Conflicts of Interest:
The authors declare no conflict of interest.

Conflicts of Interest:
The authors declare no conflict of interest. Transmembrane protein 40 Ucp1