Endurance Exercise as a Countermeasure for Aging

A total of 22 healthy young (18–30 years old) and 20 healthy older (59–76 years old) subjects gave written informed consent, as approved by the Mayo Foundation Institutional Review Board. Subjects were further divided into the following groups: young sedentary (YS) (5 women, 6 men), young trained (YT) (5 women, 6 men), older sedentary (OS) (4 women, 6 men), and older trained (OT) (4 women, 6 men). Sedentary subjects exercised less than 30 min per day, twice per week. Trained subjects performed at least 1 h of cycling or running 6 days per week over the past 4 years or longer by self-report. Activity levels were confirmed with a leisure-time activity questionnaire. Subjects were screened by medical history, physical exam, graded treadmill test, and comprehensive blood tests, including blood lipids from standard photometric methods, hormones, and glucose. Magnetic resonance spectroscopy was used to measure lipoprotein particle concentrations in plasma (20). Briefly, an EDTA plasma sample diluted 1 to 1 was injected into the flow probe of a Bruker-Biospin 400-MHz ¹H-nuclear magnetic resonance (NMR). Lipana software analyzed the data to determine LDL and HDL particle concentration. Exclusion criteria included history of metabolic or cardiovascular disease, plasma glucose >99 mg/dl, BMI >28 kg/m², medications affecting outcome measures, anemia, pregnancy, and substance abuse.

Dual X-ray absorptiometry (Lunar DPX-L; Lunar Radiation, Madison, WI) was used to measure fat and fat-free mass (FFM). Abdominal and visceral fat area were measured with single-slice computed tomography scans (Imatron C-150; Imatron, San Francisco, CA), as described previously (21). Vo₂ peak was determined from expired gas analysis during a graded bicycle test. To ensure that subjects achieved Vo₂ peak, at least two of the following criteria were confirmed: a plateau in Vo₂, respiratory exchange ratio ≥1.1, and attainment of age-predicted maximal heart rate. At least 7 days after the Vo₂ peak test, subjects were given a meat-free weight-maintaining diet for 3 days before inpatient testing and refrained from exercise. At 1700 h on day 3 of the diet, subjects were admitted to the Mayo Clinic General Clinical Research Center for 48 h. Muscle mass was estimated from 48-h urinary creatinine excretion. Subjects were given a standard meal and snack at 1800 and 2200 h of the second day, after which they remained fasted until the end of the study. Following baseline blood sampling at 0330 h the following morning, a primed (2.4 mg/kg FFM), continuous (0.04 mg · kg FFM⁻¹ · min⁻¹) infusion of [6,6-²H₂]-d-glucose was initiated. After 0530 h, one hand was placed in a heated box (120°F) with a retrograde IV catheter. At 0600 h, arterialized blood samples were taken every 10 min until 0630 h for [6,6-²H₂]-d-glucose isotopic enrichment measurements. At 0630 h, a percutaneous needle muscle biopsy (350–400 mg) was obtained from the vastus lateralis muscle under local anesthesia (22). Approximately 40 mg of muscle was immediately used for measurement of MAPR, as previously described (5), and the remaining tissue was frozen in liquid nitrogen and stored at −80°F for other analyses.

At 0630 h, following the muscle biopsy, an 8-h euglycemic-hyperinsulinemic clamp was started with insulin infused at 1.5 mU · kg FFM⁻¹ · min⁻¹. The infusion rate of [6,6-²H₂]-d-glucose was varied to mimic the anticipated changes in endogenous glucose production, as described previously (23). Plasma glucose was maintained at ∼90 mg/dl during insulin infusion with a variable infusion of 40% dextrose containing 2.15% [6,6-²H₂]-d-glucose to minimize the changes in glucose enrichments and maintain constant plasma specific activity (23). The glucose infusion rate was adjusted as needed based on blood samples taken at 10-min intervals; samples were measured with a Beckman glucose analyzer (Beckman Coulter, Fullerton, CA) (24). [6,6-²H₂]-d-glucose enrichment of blood samples and infusates were measured using mass spectrometry. As described previously (23), the steady-state equations of Steele (25) were used to calculate glucose appearance (R_a) and disappearance (R_d): R_a = R_d = F_Glu/SA of [6,6-²H₂] glucose (1), where F_Glu is the infusion rate of [6,6-²H₂]-d-glucose and specific activity (SA) of [6,6-²H₂] glucose is the plasma specific activity of [6,6-²H₂] glucose. Endogenous glucose production (EGP) was calculated as the difference between total glucose rate of appearance and exogenous glucose infusion rate (23,26). Peripheral insulin sensitivity was assessed from the rate of glucose infusion required to maintain euglycemia during the clamp, while hepatic insulin sensitivity was assessed by the extent to which EGP was suppressed during hyperinsulinemia.

MAPR was measured using a bioluminescent technique, as described previously (5). Briefly, ∼40 mg of fresh muscle was homogenized and centrifuged to isolate mitochondria. The mitochondria were suspended in buffer containing 35 μmol/l ADP and a luciferin-luciferase ATP-monitoring reagent (BioThelma, Haninge, Sweden). A BioOrbit 1251 luminometer (BioOrbit Oy, Turku, Finland) was used to monitor bioluminescence in response to the addition of glutamate plus malate, palmitoyl-l-carnitine plus malate, and succinate plus rotenone. All reactions were monitored for 25 min at 25°C, followed by calibration with an ATP standard. CS activity was measured using spectrophotometric analyses in the mitochondrial pellet (7).

DNA were extracted from frozen muscle samples using a QIAamp DNA mini kit (QIAGEN, Chatworth, CA). The abundances of mtDNA-encoded NADH dehydrogenase 1 and 4 genes were measured using a real-time quantitative PCR system (PE Biosystems, Foster City, CA). Samples were run in duplicate and normalized to S28 ribosomal DNA, as described previously (5).

Muscle samples were homogenized in cold lysing buffer with protease inhibitors. Following centrifugation for 15 min at 1,000g, the supernatants were diluted to 2 μg/μl, based on protein concentrations determined using the Lowry procedure (DC Protein Assay, BioRad). Next, 10-μl quantities were loaded and separated on SDS-PAGE (Invitrogen) and transferred onto polyvinylidene difluouride membranes (BioRad). Membranes were blocked with Tris-buffered saline with 5% nonfat milk and then washed with primary antibodies for PGC-1α (516557; Calbiochem, San Diego, CA), TFAM (PA1–24435; Affinity Bioreagents, Golden, CO), NRF-1 (33771; Santa Cruz Biotechnology, Santa Cruz, CA), and SIRT3 (ARP32389; Aviva Systems Biology, San Diego, CA). Following incubation with secondary antibody conjugated to horseradish peroxidase, membranes were exposed to enhanced chemiluminescence reagents for 5 min. Densitometric analyses were conducted using a Kodak Image Station 1000.

To determine the abundance of several individual proteins relevant to cellular ATP production, muscle samples from each subject were prepared as previously described, and the relative abundance of single peptides was determined using the iTRAQ approach (27). Samples from YT, YS, OT, and OS subjects were labeled with iTRAQ reagents 114, 115, 116, and 117, separated, and identified using liquid chromatography/tandem mass spectrometry.

Data are presented as means ± SEM. Two-way (age and activity level) ANOVA was used to examine the effects of age, activity level, and the interaction between the two on all outcome variables. When significant interactions were found, post hoc pairwise comparisons were conducted using Tukey's procedure. Individual t tests were used for a priori planned comparisons between OS and YS and between OT and YT. Regression analyses were conducted to explore the relationships between insulin sensitivity and predictors such as age, fat mass, activity score, BMI, and abdominal adiposity. Statistical analyses were conducted using SAS software (SAS, Cary, NC).

Age was associated with decreased muscle mass, increased adiposity, and decreased Vo₂ peak (Table 1). In OT and YT subjects, Vo₂ peak was higher but percent fat and abdominal and visceral fat content were lower than in OS and YS subjects. Total and LDL cholesterol were elevated in OS compared with YS adults, and HDL cholesterol was elevated in endurance-trained individuals in both age-groups. Although LDL cholesterol did not differ between sedentary and endurance-trained individuals, both total and small LDL particle concentrations measured by NMR were significantly lower in individuals who exercise.

Average plasma insulin levels during the clamp were similarly high in YS (44.7 ± 1.9 μU/ml), OS (43.1 ± 3.2 μU/ml), YT (45.7 ± 2.1 μU/ml), and OT (47.3 ± 2.3 μU/ml) subjects. The glucose infusion rate required to maintain euglycemia during the clamp was similar in young and old subjects, but higher in trained compared with sedentary subjects (Fig. 1A). The area under the glucose infusion rate curve (Fig. 1A, insets) during 0 through 480 min and 240 through 480 min (plateau region) indicate greater glucose infusion per unit FFM in trained compared with sedentary subjects, with no effects of age in either activity group. Likewise, total R_d during hyperinsulinemia was higher in trained compared with sedentary subjects, with no effects of age (Fig. 1B). EGP was lower in trained compared with sedentary individuals at baseline (Fig. 1B). During the clamp, EGP was suppressed in all groups (YS, 69 ± 7%; YT, 100 ± 11%; OS, 78 ± 7%; and OT, 97 ± 10%), although the magnitude of EGP suppression by insulin was greater in trained compared with sedentary subjects (Fig. 1B). No effects of age on EGP or relative suppression of EGP were noted. The relationship between R_d and age was nonsignificant (R = 0.11, P = 0.53; data not shown). However, R_d during the clamp was inversely related to BMI (R = −0.37, P = 0.026), fat mass (R = −0.53, P = 0.001), and abdominal fat (R = −0.37, P = 0.027) (data not shown).

MAPR was reduced in older versus younger people using substrates glutamate plus malate, succinate plus rotenone, and palmotoyl-l-carnitine plus malate but was higher in endurance-trained compared with sedentary subjects (P_age ≤ 0.004) (Fig. 2). Furthermore, paired comparisons revealed age-related declines in MAPR in sedentary (P ≤ 0.006) but not in trained subjects. Similar results were found for CS activity (Fig. 3A).

The abundance of mtDNA based on both the NADH dehydrogenase 1 and 4 gene probes was lower in older compared with young subjects (P_age ≤ 0.0004, Fig. 3B and C). Although mtDNA abundance was significantly higher in trained compared with sedentary subjects (P_training < 0.05), mtDNA remained significantly lower in OT compared with YT subjects.

Protein expression of PGC-1α, NRF-1, and TFAM were similar in YS and OS subjects and higher in trained than in sedentary (P_training < 0.001, Fig. 4) subjects. PGC-1α and TFAM protein content was significantly lower in OY than in YT subjects (P < 0.0001); however, NRF-1 expression was similar in YT and OT subjects.

Figure 5A–D depicts the relative abundance of several proteins involved in key metabolic processes in skeletal muscle. Comparison between OS and YS subjects revealed 27 oxidative and glycolytic proteins with significantly lower relative concentrations in older adults (Fig. 5A). Endurance-trained young and older individuals exhibited higher expression of numerous proteins involved in oxidative ATP production relative to their sedentary counterparts (Fig. 5B and C). A comparison of YT and OT subjects revealed that, with the exception of three subunits of cytochrome c oxidase and an aminotransferase enzyme, the age differences in protein abundance that were observed in sedentary individuals were no longer apparent in endurance-trained individuals (Fig. 5D).

Protein expression of SIRT3 was lower with age in sedentary adults and significantly higher in trained than in sedentary subjects, with no effect of age in trained adults (Fig. 6).

The current study examined the effects of aging and endurance exercise on insulin sensitivity and mitochondrial function to further explore whether age-related mitochondrial dysfunction and lowering of insulin sensitivity are inevitable consequences of chronological age. The main findings of the present study are, first, that YT and OT subjects had substantially higher insulin sensitivity than sedentary subjects and that no differences were observed between young and older subjects in either sedentary or exercise-trained groups. Second, in contrast, we found age-related declines in various markers of mitochondrial function in sedentary groups, but these age-related differences were partly (but not completely) abolished in people who practice regular endurance exercise. Finally, we show that endurance exercise may exert similar potentially lifespan-enhancing effects to those of caloric restriction through elevated SIRT3 expression in both young and older adults.

In agreement with previous reports (4,5), the capacity for mitochondrial ATP production declined with age in sedentary adults. As expected, MAPR and CS activity were higher in endurance-trained than in sedentary individuals. Although direct comparisons revealed ∼10–15% lower CS activity and MAPR in OT compared with YT subjects, these differences were not statistically significant. Thus, our data underscore the effectiveness of regular endurance exercise in largely preventing age-related declines in mitochondrial oxidative capacity. Similarly, histochemical and enzymatic measurements in young (27 ± 3 years old) and older (63 ± 6 years old) endurance runners showed that despite 11% lower Vo₂ peak compared with young runners, master's level runners exhibited no signs of decreased mitochondrial enzyme activities or capillary density (28). In vitro measures of oxidative capacity, such as those employed in the present study, provide more distinct information regarding mitochondrial function than in vivo measurements of resting mitochondrial ATP production, such as those measured using magnetic resonance spectroscopy (4). Given that skeletal muscle is relatively metabolically inert in the postabsorptive resting state and that endurance training increases Vo₂ peak with minimal effects on resting Vo₂, it is our belief that the maximal capacity for mitochondrial ATP synthesis is more relevant to the study of endurance exercise training and aging.

Several additional mitochondrial markers were also assessed in an effort to determine molecular and cellular mechanisms that would explain our MAPR findings. Using mass spectrometry methods, we demonstrated that numerous proteins involved in the citric acid cycle (isocitrate dehydrogenase, aspartate aminotransferase, malate dehydrogenase, and oxoglutarate dehydrogenase) and electron transport (cytochrome c oxidase, ubiquinol-cytochrome c reductase, NADH dehydrogenase, and ATP synthase) were expressed at lower levels in OS compared with YS subjects. Regardless of age, endurance-trained individuals exhibited higher expression of these proteins. Furthermore, the age-related differences in protein expression evident in sedentary individuals were absent in endurance-trained individuals, with the exception of three subunits of cytochrome c oxidase and aspartate aminotransferase, which remained significantly lower in OT compared with YT subjects. Our proteomic data closely parallel our MAPR data, suggesting that the expression level of key mitochondrial proteins may be a primary determinant of age-related declines in oxidative capacity and of the beneficial effects of regular endurance exercise. An important point is that regular endurance exercise could not entirely eliminate the effects of aging on protein expression, consistent with the ∼15% lower ATP production capacity in OT versus YT subjects.

The abundance of mtDNA plays a role in the expression level of several proteins that were measured using mass spectrometry: specifically, subunits of cytochrome c oxidase and NADH dehydrogenase. Consistent with previous reports (5,8), we found that mtDNA abundance was significantly reduced with age in sedentary individuals and that endurance-trained individuals in both age groups exhibited higher mtDNA abundance. Of interest, the regular endurance exercise did not completely normalize age-related declines in mtDNA abundance, which we are tempted to posit as a mechanism explaining persisting age-related deficits in the protein expression of mitochondrial-encoded cytochrome c oxidase subunits in endurance-trained individuals.

The replication, maintenance, and transcription of mitochondrial DNA are controlled by TFAM. TFAM expression is controlled by upstream nuclear transcription factors (NRF1, NRF2), which also regulate the expression of nuclear genes encoding mitochondrial proteins. PGC-1α plays a key role in activating these downstream transcription factors and, therefore, acts as a control point regulating mitochondrial biogenesis (29). The protein expression levels of PGC-1α, NRF1, and TFAM were similar in YS and OS subjects and cannot explain our observations of decreased mtDNA abundance in OS subjects. Age-related accumulation of DNA oxidative damage (5) could explain reduced mtDNA in OS subjects in spite of similar expression of mitochondrial transcription factors. PGC-1α, NRF1, and TFAM were all expressed at higher levels in endurance-trained individuals; however, a key finding is that significant effects of age were evident for PGC-1α and TFAM, but not NRF1. Thus, the persisting effects of age on mtDNA abundance despite endurance exercise training may be explained by blunted expression of upstream transcription factors involved in mitochondrial biogenesis.

A key finding of the current study is related to the effects of aging and endurance exercise on insulin sensitivity. Although decreased insulin sensitivity is reported to occur with age, we found that both hepatic and peripheral insulin sensitivity were similar in YS and OS subjects. While the absence of age-related declines in insulin sensitivity is in contrast with numerous previous reports (4,21,30,31), our results are consistent with a growing body of literature indicating that adiposity and physical activity levels, rather than chronological age, are primary determinants of age-related declines in insulin sensitivity (32–36). In support of this notion, we find that fat mass, abdominal adiposity, and BMI were significant predictors of R_d during the clamp. The OS subjects in our cohort were healthy, relatively lean people with substantially less visceral and abdominal adiposity than what has been reported in older adults with decreased insulin sensitivity (21). Our findings are in keeping with earlier studies that also revealed similarities in insulin sensitivity in relatively lean sedentary young and older adults (34,37). Peripheral and hepatic insulin sensitivity were significantly higher in endurance-trained than in sedentary individuals, and these effects were independent of age. These data are supported by previous investigators who have demonstrated that endurance exercise confers substantial improvements in peripheral and hepatic insulin sensitivity, regardless of age (14,38,39). The results from the current study strongly support the notion that decreased insulin sensitivity may not be an inevitable consequence of aging and that age-related changes in body composition are a key factor contributing to the previously reported decline in insulin sensitivity with aging. Furthermore, in addition to maintenance of normal body weight, endurance exercise should be considered a viable, effective intervention to reverse, delay, or prevent the onset of age-related declines in insulin sensitivity that is likely to occur at more advanced age.

Mitochondrial dysfunction and decreased sensitivity to insulin are often observed concurrently in type 2 diabetic (9), elderly (4,5), and obese individuals (10), prompting two opposing hypotheses concerning this relationship. It has been proposed that mitochondrial dysfunction plays an etiological role in the development of insulin resistance (11,40), while some evidence supports an alternative hypothesis that mitochondrial dysfunction may result from insulin resistance (41,42). Other studies find a dissociation between mitochondrial capacity and insulin sensitivity (21,43,44). In the present study, OS subjects were as insulin sensitive as YS subjects despite significantly reduced MAPR, casting further doubt on the causal role of mitochondrial dysfunction in decreasing insulin sensitivity. However, we observed significant elevations in both oxidative capacity and insulin sensitivity in endurance-trained individuals regardless of age. These data underscore the fact that, although mitochondrial function has been extensively studied in insulin-resistant populations, our understanding of its role in the development of insulin resistance is far from complete.

We assessed other outcomes relevant to age-related comorbidities that may impact insulin sensitivity and mitochondrial function. Percent body fat, abdominal fat, and visceral fat were elevated in older adults, and although these measures were lower in endurance-trained individuals, the effects of age on body composition were not completely normalized in endurance-trained adults. Similarly, the age-related decline in muscle mass observed in sedentary individuals was also apparent in endurance-trained individuals, suggesting that endurance exercise can partially prevent age-related changes in adiposity but has limited utility in preventing sarcopenia, which is prevented by resistance training. Both the traditional photometric-based clinical lipid panel and NMR-based method of measuring lipoprotein particle concentrations revealed the expected elevations of total and LDL cholesterol in OS compared with YS subjects, as well as the beneficial increase in HDL cholesterol in both age-groups. However, while exercise was not related to total LDL cholesterol when measured photometrically, both total and small LDL particle concentrations measured by NMR were significantly lower in exercising individuals.

We also provide data to link endurance exercise with the lifespan-enhancing effects of caloric restriction (45). Mitochondrial-localized SIRT3 is linked to longevity, possibly by interfering with the release of apoptosis-inducing factor (19,46). In a manner similar to that by which nutrient restriction increases lifespan (19), we found that SIRT3 expression was higher in endurance-trained than in sedentary individuals. Furthermore, SIRT3 expression was lower with age in OS but not OT individuals. Thus, our results support a hypothesis that exercise may confer lifespan-extending effects similar to those of caloric restriction through the actions of mitochondrial SIRT3.

It is important to consider that many detrimental effects of old age are compounded after 80 years of age. It is unclear if similar effects of regular endurance exercise would be evident in individuals at more advanced ages than were studied in the present study. It is theoretically possible that the cross-sectional design of the study may introduce a sampling bias wherein individuals who exercise across their lifespan do so because their mitochondrial function is inherently higher than that of their more sedentary peers. Notwithstanding, it is becoming increasingly apparent that waning physical activity levels are a primary determinant of age-related mitochondrial dysfunction and may accelerate the aging process through adverse effects on many biochemical and molecular factors that require ATP.

In conclusion, we find that age-related declines in oxidative capacity can be largely ameliorated by regular endurance exercise, highlighting the fact that physical inactivity plays an important role in age-related dysfunctions and underscoring the need for better control of this variable in the literature. However, certain mitochondrial markers, specifically mtDNA, transcription factors, and mitochondrial-encoded proteins, remain depressed with increased age despite endurance exercise. Neither hepatic nor peripheral insulin sensitivity was impaired with age in these healthy, relatively lean individuals, which was in keeping with the notion that adiposity is a primary determinant of age-related reductions in insulin sensitivity. Notwithstanding, we illustrate that compared with sedentary individuals, endurance-trained individuals exhibit elevated insulin sensitivity in a manner independent of age. Finally, we observed an age-related decline in muscle expression of SIRT3 in sedentary but not endurance-trained individuals, suggesting that endurance exercise may exert similar potentially lifespan-enhancing effects to those demonstrated with caloric restriction in other organisms.

The current study is supported by grants from the National Institutes of Health (RO1 AG 09531 and UL1 RR024150-01) and the National Center for Research Resources, training grant DK07198 from the Mayo Foundation, and the David Murdock Dole Professorship.

We are grateful for the assistance of the Mayo Clinic CTSA Clinical Research Unit nursing and nutrition staff, superior nursing coordination by Maureen Bigelow, and the skillful technical assistance of Jill Schimke, Lawrence E. Ward, Dawn Morse, Kate Klaus, and Jane Kahl. Special thanks are expressed to all of the subjects for their participation.