Skeletal muscle mitochondrial dysfunction is hypothesized to contribute to the pathophysiology of insulin resistance and Type 2 diabetes. Whether thiazolidinedione therapy enhances skeletal muscle mitochondrial function as a component of its insulin-sensitizing effect is unknown. To test this, we evaluated skeletal muscle mitochondria and exercise capacity in Type 2 diabetic subjects with otherwise normal cardiopulmonary function in response to rosiglitazone therapy. Twenty-three subjects were treated for 12 wk and underwent pre- and posttherapy metabolic stress testing and skeletal muscle biopsies. Rosiglitazone significantly ameliorated fasting glucose, insulin, and free fatty acid levels but did not augment the subjects' maximal oxygen consumption (V̇o2max) or their skeletal muscle mitochondrial copy number. The baseline V̇o2max correlated strongly with muscle mitochondrial copy number (r = 0.56, P = 0.018, n = 17) and inversely with the duration of diabetes (r = −0.67, P = 0.004, n = 23). Despite the global lack of effect of rosiglitazone-mediated insulin sensitization on skeletal muscle mitochondria, subjects with the most preserved functional capacity demonstrated some plasticity in their mitochondria biology as evidenced by an upregulation of electron transfer chain proteins and in citrate synthase activity. This study demonstrates that the augmentation of skeletal muscle mitochondrial electron transfer chain content and/or bioenergetics is not a prerequisite for rosiglitazone-mediated improved insulin sensitivity. Moreover, in diabetic subjects, V̇o2max reflects the duration of diabetes and skeletal muscle mitochondrial content. It remains to be determined whether longer-term insulin sensitization therapy with rosiglitazone will augment skeletal muscle mitochondrial bioenergetics in those diabetic subjects with relatively preserved basal aerobic capacity.
- exercise capacity
- citrate synthase activity
- peroxisome proliferator-activated receptor-γ agonist
diminished skeletal muscle glucose uptake and attenuated skeletal muscle insulin sensitivity are important precursors to the development of Type 2 diabetes mellitus (19). Skeletal muscle of insulin-resistant subjects has diminished mitochondrial density (24) and reduced mitochondrial oxidative phosphorylation (27). Diabetic individuals also show coordinate downregulation of genes encoding the oxidative phosphorylation enzymes (22) and diminished skeletal muscle mitochondrial respiration (23), and evidence of reduced skeletal muscle bioenergetic capacity as illustrated by slower recovery of skeletal muscle high-energy phosphate stores in response to exercise compared with nondiabetic controls (32). The proposed mechanism whereby disruption in skeletal muscle mitochondria exacerbates insulin resistance has been recently reviewed and suggests that the resultant reduction in fatty acid β-oxidation increases intracellular fatty acyl-CoA and diacylglycerol, which phosphorylates and inactivates insulin signal transduction intermediates via protein kinase C signaling (18). Collectively, these data suggest that disruption of the skeletal muscle mitochondrial biogenesis regulatory program and bioenergetics are integral to insulin resistance in skeletal muscle and raise the question as to whether these perturbations exacerbate the progressive pathophysiology of Type 2 diabetes mellitus. Interestingly, the thiazolidinedione insulin sensitizers delay the onset of Type 2 diabetes (7) and have been shown to restore mitochondrial respiratory function and biogenesis in adipose tissue (4, 39). In light of the importance of skeletal muscle insulin sensitivity in the development of insulin resistance and diabetes, we evaluated whether the augmentation of mitochondrial bioenergetics in skeletal muscle may be an important component in rosiglitazone-mediated improvement in insulin sensitivity and glycemic control. Furthermore, if this biology is operational, we would expect an improvement in exercise capacity in parallel with an improvement in skeletal muscle mitochondrial function.
To establish the effect of thiazolidinedione therapy on skeletal muscle biology, we performed studies on skeletal muscle biopsies and compared these with metabolic stress tests in diabetic subjects in response to 12 wk of rosiglitazone therapy. Although improvement in glycemic control was a uniform finding, there was no augmentation in the study subjects' aerobic capacity. On direct assessment of mitochondrial proteins, an induction of mitochondrial oxidative metabolic protein content and activity was evident in the subset of study subjects with most robust baseline aerobic capacity. Furthermore, baseline V̇o2max was shown to be a strong indicator of the duration of diabetes and of underlying skeletal muscle mitochondrial genomic content. Taken together, these data show that the improvement in skeletal muscle mitochondrial function is not mechanistically linked to rosiglitazone-mediated insulin sensitization. A hypothesis generated by this study is that the capacity for skeletal muscle mitochondrial bioenergetic augmentation in diabetes subjects may be dependent on the duration of diabetes and on a subject's baseline maximal exercise capacity.
This was a prospective open-labeled study to evaluate the skeletal muscle mitochondrial response to the addition of rosiglitazone to the therapeutic regimen of diabetic subjects naïve to thiazolidinedione therapy. Subjects were excluded from participation if they had abnormal cardiopulmonary function or an inability to exercise on a treadmill. Subjects were maintained on their prestudy therapy, and rosiglitazone was introduced and titrated up from 2 mg daily to a maximum of 4 mg two times daily, or to the highest dose tolerated, over a 6-wk period. Therapy was continued for an additional 6 wk. All subjects had an initial visit with the dietician, but no further attempt was made to modify their exercise regimen or dietary habits, since the objective of this study was to assess the exclusive effect of rosiglitazone on maximal effort tolerance and on skeletal muscle biology. This study was approved by the Institutional Review Board of the National Heart, Lung, and Blood Institute, and all subjects provided informed written consent.
Subjects were seen biweekly to adjust dosing and then monthly once they were on their maximal tolerated dose. Additional baseline studies included metabolic stress testing and a gastrocnemius muscle biopsy that was performed >5 days after the maximal stress test. Baseline and final insulin sensitivity was calculated using the QuickI formula, which incorporates fasting insulin and glucose values and correlates with the oral glucose tolerance test (29). Finally, brachial artery reactivity was assessed at baseline and at 12 wk to determine effects of the thiazolidinediones on endothelial nitric oxide bioactivity. Endothelium-dependent flow-mediated vasodilation was assessed by Doppler imaging, as previously reported (9).
Skeletal muscle analyses.
Symptom-limited cardiopulmonary treadmill exercise testing was performed 1) at baseline and post study. Ventilatory expired gas analysis was collected using a metabolic cart (Sensormedics; Viasys Healthcare, Conshohoken, PA). Oxygen uptake and carbon dioxide output were acquired, and the maximal oxygen consumption (V̇o2max) and peak respiratory exchange ratio (RER) were expressed as the highest 10-s averaged samples obtained during the last stage of the exercise test.
Genomic DNA was extracted from a random selection of 17 subjects to use for analysis of mitochondrial copy number. Mitochondrial DNA copy number was assessed as described previously (15). In brief, DNA was collected from whole tissue lysates, and quantitative real-time PCR was performed for mitochondrial-encoded cytochrome c oxidase (COX) subunit III and for nuclear-encoded 18S using SYBR green PCR Master Mix (Applied Biosystems, Foster City, CA) and the MJ Research DNA Engine Opticon 2 fluorescence detection system. The ratio of COXIII DNA copies to 18S DNA represents the relative mitochondrial copy number.
Since proteomic analysis is a nonbiased technique to identify regulation at the protein expression level, a random sample of six subject's muscle samples were employed for proteomic analytical screening pre- and posttherapy using light chromatography/mass spectroscopy/mass spectroscopy analysis as previously described (10). Residual protein available from 12 subjects was used to perform Western blot analysis. Briefly, 20–45 μg of protein were subjected to Western blot immunoblot assays using antibodies directed against tubulin (Sigma, St. Louis, MO), peroxisome proliferator-activated receptor (PPAR)-γ (Upstate, Chicago, IL), and the mitochondrial proteins, including ND6 from complex I, Fe-S subunit of complex II, Core2 from complex III, COXII from complex IV, and ATP synthase F1α from the FI/F0 ATPase and polyclonal antibodies to COXI and COXII complexes (Mitosciences, Eugene, OR). Bands were visualized using horseradish peroxidase-conjugated secondary antibodies (GE Healthcare, Piscataway, NJ) and the SuperSignal West pico chemoluminescence substrate kit (Pierce Biotechnology, Rockford, IL).
Citrate synthase activity was measured in all skeletal muscle samples (n = 23) using 10–25 μg of protein from whole tissue skeletal muscle. Activity level, expressed as micromoles per gram total protein per minute, was measured spectrophotometrically using a commercially available kit (Sigma).
Differences between subjects were evaluated for significance using the paired (pre and post rosiglitazone) and unpaired (between subgroups) two-tailed Student's t-test. Regression analysis was used to assess relationships between V̇o2max and percent change in citrate synthase activity, duration of diabetes, mitochondrial copy number, and endothelial function. Since citrate synthase activity reflects mitochondrial oxidative metabolism, all tissue samples were used to measure this activity. The regression analysis comparing V̇o2max with the change in citrate synthase activity in response to rosiglitazone therapy was considered the primary comparison. The limited biopsy material was then randomly designated for proteomic analysis or for the determination of mitochondrial copy number and Western blot analyses. In light of the small sample sizes, these subgroup analyses are viewed as exploratory data to determine whether these biological readouts support the primary comparison. Receiver operating characteristic curve analysis determined the optimal V̇o2max threshold at baseline to predict improvement in citrate synthase activity. A two-sided Fisher exact test was used to determine the relative effects of the use of other antidiabetic drugs in subgroup analysis. All data are expressed as means ± SE. P < 0.05 was considered statistically significant in all analyses.
Twenty-five individuals entered the study (age 60 ± 2 yr, 19 males and 6 females). Two subjects did not complete the study because one individual declined the second biopsy and one developed cellulitis following the first biopsy and we elected not to repeat the procedure. Overall, the therapy was well tolerated, with 86% of subjects reaching the maximal dose, the average dose being 7.5 ± 1 mg daily. The baseline body mass index of subjects enrolled in the study was 31 ± 0.9 kg/m2, and the most common side effect was further weight gain, with 31% of study subjects gaining more than 2 kg over the 12-wk study. The efficacy of rosiglitazone is shown in Table 1. These changes are consistent with previous studies showing significant reductions in fasting glucose, fasting insulin, in free fatty acids, and in the highly sensitive C-reactive protein. Interestingly, there was no change in the glycosylated hemoglobin levels, which may reflect the good control already evident in these subjects with a baseline HbA1c of 6.5 ± 0.16%. Despite the improvement in serological parameters, there was no significant change in V̇o2max (from 24.6 ± 1.3 to 24.7 ± 1.2 ml·kg−1·min−1, P = 0.89) after 12 wk of rosiglitazone. Furthermore, peak RER values were >1.10 during both the baseline and postintervention exercise tests, indicating that the subjects put forth a maximal, comparable effort (baseline 1.17 ± 0.08, postintervention 1.17 ± 0.09, P = 0.96).
Maximal exercise capacity and diabetes.
To assess whether V̇o2max could be used to study skeletal muscle mitochondrial function, the relative mitochondrial to nuclear DNA copy number was determined in 17 randomly selected diabetic subjects. The mitochondrial genomic copies of COXIII correlated strongly with V̇o2max (r = 0.56, P = 0.0184, n = 17; Fig. 1A). The association between the severity of diabetes with aerobic capacity was demonstrated by inverse correlations between baseline V̇o2max and the duration of diabetes (r = −0.67, P = 0.004, n = 23; Fig. 1B). A correlation between insulin sensitivity and V̇o2max did not reach statistical significance (r = 0.30, P = 0.069), and age did not correlate with V̇o2max.
Electron transfer chain protein modulation in response to rosiglitazone.
Since the V̇o2max was not changed in the group as a whole, we employed unbiased proteomic analysis to determine whether regulatory changes were evident with direct assessment of mitochondrial oxidative phosphorylation proteins. A randomly chosen subset of six study subjects' whole muscle samples were employed for this analysis. In duplicate samples, 305 peptide sequences were identified as high priority, defined as a high confidence that the peptide sequences are correct and that multiple peptide sequences of the same protein are obtained. Of these 305 high-priority sequences, 9 proteins were identified as involved in mitochondrial oxidative phosphorylation. In this small subject sample size, only the subject with the highest V̇o2max displayed a coordinate induction of the oxidative phosphorylation proteins (Fig. 2). Albeit preliminary and limited in scope, this screening test suggested that skeletal muscle mitochondrial plasticity in response to insulin sensitization therapy may be achievable in those subjects with higher basal aerobic capacity. We then evaluated the change in citrate synthase activity, a measure of mitochondrial oxidative metabolism (36) and/or of mitochondrial content (12), in all 23 subjects compared with basal V̇o2max. The change in citrate synthase activity was correlated with the baseline V̇o2max (r = 0.54, P = 0.01). Because this correlation elicited greater significance employing an exponential fit (r = 0.47, P = 0.001), we calculated the threshold of V̇o2max delineating an improvement, or not, in citrate synthase activity in response to rosiglitazone therapy. The threshold of ≥28.1 ml·kg−1·min−1 provided a sensitivity of 70% and a specificity of 92% between responders and nonresponders.
All subsequent subgroup analysis was performed comparing subjects with a basal V̇o2max of <28.1 ml·kg−1·min−1 with those with values ≥28.1 ml·kg−1·min−1. The basal mitochondrial copy number was significantly higher in subjects with a V̇o2max of ≥28.1 ml·kg−1·min−1 (Fig. 3A), and the duration of diabetes was significantly shorter in these subjects, with a mean duration of 30 ± 6.6 mo vs. 84 ± 11 mo (P = 0.0196) in subjects with reduced V̇o2max and lower mitochondrial copy number (Fig. 3B). There was no difference in age between subjects with V̇o2max of ≥ 28.1 ml·kg−1·min−1 (61.1 ± 3 vs. 58.1 ± 2.4 yr, respectively, P = 0.48; Fig. 3C). Interestingly, baseline citrate synthase activity did not correlate with V̇o2max. The association between aerobic capacity and the upregulation in oxidative phosphorylation proteins in response to rosiglitazone in this subset of diabetic subjects was further supported by the parallel increase in citrate synthase activity by 53 ± 35.8% compared with −24 ± 10.6% (P = 0.03) in subjects with lower aerobic capacity and modest attenuation in oxidative phosphorylation protein levels (n = 12; Fig. 3D). Rosiglitazone therapy also did not augment aerobic capacity in the subgroup with relatively preserved aerobic capacity (data not shown).
In parallel, steady-state protein levels of PPARγ and proteins of the electron transfer chain demonstrated coordinate upregulation in response to rosiglitazone in subjects with a baseline V̇o2max ≥28.1 ml·kg−1·min−1 (n = 12; Fig. 4, A and B). Steady-state protein levels that reached statistical significance in this group included the Fe-S subunit of complex II of the electron transfer chain, COXI of complex IV, and the ATP synthase F1α subunit of the F1/Fo ATPase.
Vascular reactivity and exercise capacity.
Because thiazolidinediones improve endothelial-dependent blood flow (5) and the improvement of vascular supply to skeletal muscle may augment mitochondrial bioenergetics, we employed shear stress-induced flow-mediated dilation (FMD) as a measure of vascular endothelial nitric oxide bioactivity. As a group, the FMD was only modestly improved from 6.8 ± 0.6 to 7.4 ± 0.6% (P = 0.16). This modest improvement in endothelial-dependent vasodilation is unlikely to have contributed to the augmentation in mitochondrial content and/or bioenergetics, since neither baseline nor change in FMD correlated to V̇o2max and FMD effects were similar between subjects irrespective of whether their V̇o2max was relatively preserved or not (Table 2).
Discordance between glycemic control and change in mitochondrial content/activity.
Under the hypothesis that mitochondrial function would be central to the pathophysiology of insulin resistance, we expected a greater augmentation of insulin sensitivity in those subjects with relatively preserved V̇o2max, higher mitochondrial copy number, increased oxidative phosphorylation proteins, and an induction of citrate synthase activity. This was not evident, since rosiglitazone had similar insulin-sensitizing effects independent of V̇o2max (Table 2). Moreover, additional antiglycemic agents being administered were unlikely to have affected the mitochondrial response in that a similar proportion of subjects in both subgroups were on metformin and on a sulfonylurea agent (61 vs. 40% and 15 vs. 30% respectively, comparing subjects with lower and higher V̇o2max, P = 0.7 and P = 0.6 respectively).
Insulin resistance is the best predictor of the future development of Type 2 diabetes (19). This is hypothesized to result from, in part, reduced skeletal muscle mitochondrial oxidative capacity-evoked fatty acid metabolite accumulation with the subsequent disruption in insulin signal transduction (24). It has recently been demonstrated that skeletal muscle mitochondrial substrate oxidation is impaired in insulin-resistant offspring of Type 2 diabetic patients (2). Whether the augmentation of skeletal muscle mitochondrial function is indeed required to improve insulin sensitivity has not been established. The PPARγ agonist rosiglitazone has been shown to improve mitochondrial respiration in adipocytes of ob/ob mice (39) and to upregulate the adipocyte mitochondrial biogenesis regulatory program in two mouse models of obesity/diabetes (31). Because skeletal muscle is a predominant reservoir for postprandial glucose uptake (21) and skeletal muscle insulin resistance is proposed to be pivotal in the development of diabetes, the question arising is whether restoration of mitochondrial biology in skeletal muscle is central to the insulin-sensitizing effects of the thiazolidinediones. This question formed the central objective of this study.
Because V̇o2max represents a composite measure of cardiopulmonary, skeletal muscle, and vascular function (37), we reasoned that the evaluation of the modulation in V̇o2max in diabetic subjects with a normal cardiopulmonary profile could be employed as an indirect measure of skeletal muscle bioenergetic changes in response to rosiglitazone. Furthermore, the direct association of V̇o2max with skeletal muscle bioenergetics is supported by the strong linear correlation between hyperinsulinemia-mediated skeletal muscle glucose uptake and V̇o2max in insulin-resistant individuals (25). At the cellular level, this association is more directly linked to the mitochondria in that a relative reduction in type I oxidative fibers (16) and an increased ratio of glycolytic to oxidative metabolic enzymatic activity (33) are evident in insulin-resistant subjects. Finally, the use of V̇o2max as an indicator of future risk for developing diabetes has been shown (3, 25, 38). In our study, where all subjects had unimpaired cardiopulmonary function, the direct correlation between baseline V̇o2max and mitochondrial copy number and the inverse correlation between baseline V̇o2max and duration of diabetes support the notion that the skeletal muscle bioenergetic perturbations associated with diabetes are progressive. However, the lack of improvement in V̇o2max in our study suggests that the selective change in skeletal muscle bioenergetics, identified in the subset of subjects with relatively preserved V̇o2max, in response to rosiglitazone is either of minor functional significance or that a longer duration of treatment may be necessary to elicit an enhancement of maximal exercise capacity. Our data contrast with an improvement in V̇o2max in response to rosiglitazone by Regensteiner and colleagues (30). The major difference between our study and the Regensteiner study is that they treated subjects for 4 mo vs. the 3 mo in our study. This modestly longer duration of therapy would support additional improvement with an extension in the duration of therapy. However, while our manuscript was under review, a study was published evaluating the effect of rosiglitazone on skeletal muscle mitochondria in high-fat-fed rats (14). In parallel with our study, Lessard et al. (14) demonstrated that a short duration of treatment did not augment skeletal muscle mitochondrial function but did ameliorate the adverse effects of the high-fat diet on adipose tissue and in the liver.
The more direct evaluation of skeletal muscle mitochondria included quantification of the relative change in skeletal muscle mitochondria genomic content, the change in activity of citrate synthase [a measure that reflects mitochondrial content (12) and/or oxidative activity (36)], and the steady-state protein levels of electron transfer proteins. The lack of change in mitochondrial genomic copy number is consistent with our V̇o2max data. Because citrate synthase activity has been shown to represent oxidative metabolism in skeletal muscle, the lack of correlation between baseline citrate synthase activity and V̇o2max is not readily interpretable. However, sedentary lifestyle, obesity, and insulin resistance all contribute to suppression of citrate synthase activity (8, 34), and because our subjects have various combinations of these phenotypes, the number of subjects investigated may have been insufficient to clearly demonstrate a correlation between baseline citrate synthase activity and V̇o2max. Inversely, the potential to reverse the skeletal muscle mitochondrial defects in diabetes is evident as shown by the coordinate induction of citrate synthase activity and the upregulation of mitochondrial oxidative metabolic proteins exclusively in the subset of subjects with relatively preserved maximal aerobic capacity. On a cautionary note, these ameliorative changes were only present in a minority of the diabetic subjects studied. Hence, whether a mitochondrial genomic content threshold predicates the capacity to augment mitochondrial biogenesis and improve mitochondrial function can only be posited as an intriguing concept that needs to be evaluated further. In addition, because peroxisome proliferator-activated receptor γ-1α (PGC-1α) levels were not measured due to insufficient material, we are not in the position to comment on any relationship between PGC-1α expression and change in citrate synthase activity in the individuals with increased basal V̇o2max.
Disruption of skeletal muscle mitochondrial function is evident early in the development of diabetes (24, 26–28) and may exacerbate insulin resistance via the accumulation of intracellular fatty acid metabolites (6, 18). Although it is likely secondary to insulin resistance, first-degree relatives of Type 2 diabetics show perturbations in insulin signaling and glucose transport preceding the dysregulation of mitochondrial biogenesis regulatory genes (11). In this study, the uniformly improved glycemic control and insulin sensitivity independent of skeletal muscle mitochondrial content/activity supports that, at least as an acute response, improvement in skeletal muscle mitochondrial biology is not a prerequisite for rosiglitazone-mediated glycemic control. This does not exclude the involvement of adipocyte mitochondrial biology or the possibility that more prolonged therapy may further enhance insulin sensitivity, in part by induction of skeletal muscle mitochondrial bioenergetics. In fact, the difference in plasticity seen in our study compared with the robust induction of mitochondrial biogenesis in adipocytes in response to rosiglitazone (31) may reflect the relative lower abundance of PPARγ in skeletal muscle compared with expression in adipose tissue (17, 35).
From a clinical perspective, we propose that these data support the determination of V̇o2max in diabetic individuals with otherwise normal cardiopulmonary function as a functional test to differentiate those earlier and later in the course of the disease. Interestingly, the biological importance of the measurement of V̇o2max in diabetic subjects has been further underscored by the recent observation that the maximal exercise capacity predicts mortality in Type 2 diabetes (20). In light of the accumulating evidence supporting interventions to delay the progression of diabetes (7, 13), the finding of a relatively maintained V̇o2max may have added motivational benefit for these individuals identified as being in the early stages of this disease. Moreover, if skeletal muscle bioenergetic reversibility is dependent on early intervention in disease progression, this would support metabolic testing as a screening tool. Prior to this, however, a more comprehensive study in diabetic subjects with preserved aerobic capacity would be required to evaluate whether early treatment or more robust insulin sensitization can indeed augment skeletal muscle mitochondrial oxidative capacity and/or improve functional capacity in parallel with a delay in disease progression.
Overall, this study illustrates new mechanistic insight into the temporal pathophysiology in skeletal muscle of diabetes with respect to mitochondrial plasticity and the effects of rosiglitazone. We conclude that the major finding in this study is that rosiglitazone augments insulin sensitivity independent of V̇o2max, demonstrating that the improvement in skeletal mitochondrial function is not required for rosiglitazone-mediated insulin sensitization. Furthermore, this study shows that the measurement of V̇o2max in Type 2 diabetics can predict the extent and duration of their disease and reflects the underlying skeletal muscle mitochondrial content.
This study was funded by the National Heart, Lung, and Blood Institute, Division of Intramural Research of the National Institutes of Health.
We thank the National Cancer Institute Surgical team for performing the muscle biopsies. Additional thanks to Kevin Smith for performing the metabolic stress tests, Gloria Zalos for FMD studies, Brenda Holbrook and Carole Malone for participation in patient care, and Dr. Vandana Sachdev and the echocardiography laboratory. Mass spectrometry was performed at the Indiana Centers for Applied Protein Sciences. We thank Dr. Toren Finkel for critical review of this manuscript.
↵* I. Pagel-Langenickel and D. R. Schwartz contributed equally to this manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society