Type 2 diabetes mellitus is becoming increasingly common in both adults and in children. Uncontrolled hyperglycemia, the cardinal feature of type 2 diabetes, can cause serious microvascular and macrovascular complications, which include lower-limb amputation, kidney failure, heart disease, periodontal disease, and blindness. Although family-based studies indicate that genetic factors increase type 2 diabetes susceptibility, the allelic structure of most populations has not changed sufficiently during the past century to explain the corresponding rise in disease prevalence. By contrast, the declining physical activity levels and the changes in dietary patterns during this period are likely to be major etiological factors that underlie this phenomenon. However, not all those who are physically active have healthy metabolic profiles (5). In lifestyle intervention studies, considerable heterogeneity exists in the metabolic response of participants (2,3), yet the variability among family members is often less than among unrelated individuals (2). Taken together, these lines of evidence indicate that inherited factors, such as genes, interact with diet and physical activity to modify the risk of type 2 diabetes.
Of the many plausible physical activity interaction genes for type 2 diabetes, the peroxisome proliferator-activated receptor γ coactivator 1α (PPARGC1A) and the protein it encodes (PGC-1α) stand out as particularly interesting candidates. This is because PPARGC1A gene expression increases in response to exercise, the coding regions of the gene are polymorphic, and the PGC-1α protein coactivates a large array of transcription factors that are involved in cellular energy metabolism.
Although transcription factors directly bind to specific DNA sequences, they are typically incapable of forming functioning protein complexes by themselves because they lack the specific enzymes necessary to manipulate chromatin, unravel DNA, and sequester RNA polymerase II. Transcriptional coactivators, such as PGC-1α, are multiprotein complexes residing within the cell's nucleus that can be recruited by transcription factors via cellular signals to facilitate transcription. Thus, in combination, transcription factors and transcriptional coactivators powerfully control many aspects of metabolism.
PGC1α was the first of three PGC-1 homologues to be discovered, the others being PGC-1β and PGC-related coactivator 1. PGC1α and PGC-1β share high sequence homology, whereas PGC-related coactivator 1 is more distinct. PGC1α mRNA is expressed predominantly in tissues with high metabolic activity, most of which are rich in mitochondria. These include heart, exercising high-oxidative (type 1) skeletal muscle fibers, brown fat, kidney, liver, brain, and other tissue such as white adipose. PGC1α was initially identified as a cold-inducible coactivator in brown adipose tissue and skeletal muscle, where it controls adaptive thermogenesis via oxidative metabolism and mitochondrial biogenesis. However, it is now known that all three PGC-1 homologues are highly versatile, with the ability to dock to a wide variety of transcription factors in various tissue types and, by consequence, facilitate a broad and complex array of metabolic processes, which include oxidative metabolism, muscle fiber-type formation and switching, glucose and lipid transport, and hepatic glucose production.
Silencing of the PGC-1α signaling cascades results in reduced oxidative capacity and metabolic derangements that manifests as an insulin resistance-like syndrome. This includes hepatic steatosis, lower mitochondrial volume density in type 1 fiber, reduced mitochondrial respiration in muscle and liver, and growth retardation of heart and soleus muscle. Furthermore, PGC-1α null mice frequently die during the postnatal period, and those that survive respond poorly to the physiological stressors associated with cold and with exercise. In humans, the expression of the PGC-1α-coactivated genes is lower in the skeletal muscle of patients with type 2 diabetes compared with the skeletal muscle of healthy, glucose-tolerant individuals. Thus, it has become apparent that PGC-1α is a master regulator of many diabetes-related molecular pathways and regulates both basal and exercise-induced control of energy metabolism.
This review aims to describe the ways in which many important beneficial effects of physical activity on energy homeostasis are associated with the activation of the PGC-1α gene and to outline how functional variation at this locus could modify the effects of physical activity on the phenotypic antecedents of type 2 diabetes.
EXERCISE, PGC-1α, AND DIABETES
The inverse association between physical activity and type 2 diabetes risk is well established, and lifestyle intervention studies show that exercise intervention dramatically reduces the risk of developing the disease (7). The reduced risk of type 2 diabetes after exercise intervention may be partially mediated by the PGC-1α gene.
Exercise is positively associated with PGC-1α activity. A single bout of exercise leads to a marked increase in PGC-1α mRNA levels during the hours after strenuous lower-limb exercise, which return to pre-exercise levels within 24 h (14). By restricting blood flow to the exercising limb, PGC-1α mRNA overexpression is enhanced (13), indicating a role for hypoxia in exercise-induced PGC-1α activity. Structured exercise training lasting several weeks also increases PGC-1α mRNA content (15,16,18) and protein levels (9) independently of the short-term effects of exercise.
Despite evidence of association between exercise and PGC-1α expression, the mechanisms that underlie this relationship are not fully determined. Several of the metabolic effects of PGC-1α may involve previous activation of AMP kinase (AMPK), which acts as a sensor of cellular energy levels and is sensitive to a variety of hormones secreted during exercise (6). In obese Zucker rats, insulin resistance is caused by defects in the LKB1-AMPK-PGC-1α pathway, but these effects are reversible with exercise training (17). However, studies in humans have shown increased PGC-1α protein formation after exercise training in the absence of changes in AMPK transcription (9). Thus, although strenuous exercise may improve insulin signaling by AMPK-dependent activation of PGC-1α, PGC-1α is also activated by AMPK-independent mechanisms. One such mechanism involves phosphorylation of PGC-1α at threonine 262, serine 265, and threonine 298 by p38 mitogen-activated protein kinase (p38 MAPK; the product of MAPK14). Induction of p38 MAPK occurs in response to cellular stress, muscle stretch, and increased cytokine production, all of which are characteristics of prolonged strenuous exercise. Lastly, increased motor nerve activity during sustained contraction elevates the intracellular free Ca2+ levels after each wave of sarcolemmal depolarization. These elevations induce Ca2+ calmodulin-dependent protein kinase and calcineurin activity, both of which transcriptionally activate PGC-1α (8).
Considerably more is known of the downstream effects ofPGC-1α on diabetes-related traits. These effects can be broadly categorized as changes in oxidative capacity, insulin signaling and glucose transport, glucose and lipid metabolism, and hepatic glucose production. The following section summarizes the literature relating to these effects (also see Fig. 1).
In eucaryotes, the synthesis of the chemical endpoint ATP can occur after the oxidation of NADH as part of the mitochondrial respiratory chain. This process results in an ATP yield from glucose 15- to 18-fold higher than that via other mechanisms and is essential in complex organisms, including man, that endure sudden and sizable surges in energy requirements. There are five inner mitochondrial enzyme complexes (NADH dehydrogenase I, succinate dehydrogenase II, cytochrome c reductase III, cytochrome c oxidase [COX] IV, and ATP synthase). The electron transport chain is comprised of three of these complexes (NADH-I, cytochrome c reductase III, and COX-IV) and transports protons across the inner mitochondrial membrane, which results in a proton gradient that generates energy necessary to transform ADP into ATP. COX-IV is the terminal enzyme of the respiratory chain and reduces molecular O2 to H2O. PGC1α coactivates the genes encoding this enzyme and other genes involved in oxidative phosphorylation. These include monoamine oxidase B, cytochrome c, nuclear respiratory transcription factor 1, nuclear respiratory transcription factor 2, ATP synthase β, mitofusin 1 and mitofusin 2, mitochondrial transcription factor α, mitochondrial transcription factor β1, and mitochondrial transcription factor β2.
Muscle-specific transgenic overexpression of PGC-1α in mice is associated with enrichment of mitochondria and oxidative enzymes and with a higher ratio of slow (oxidative)- to fast (glycolytic)-twitch muscle fiber. The transgenic overexpression of peroxisome proliferator-activated receptor δ (PPAR-δ) corresponds with a similar phenotype and may be a molecular target of PGC-1α. In PGC-1α and PPAR-δ transgenic mice, the oxidative capacity of skeletal muscles is enhanced and the time to fatigue during forced muscle contraction is prolonged compared with wild-type littermates.
Importantly, long-duration, low-intensity aerobic exercise also increases the expression of PGC-1α mRNA activity in healthy wild-type rodents and in man. These observations suggest that the induction of PGC-1α through exercise facilitates beneficial health adaptations that involve changes in skeletal muscle content of oxidative enzymes and mitochondria count, which, in turn, improves the capacity for oxidative substrate metabolism, ATP synthesis, and exercise. This process is consistent with the beneficial adaptations that can occur with aerobic exercise training.
Glucose Transport and Insulin Signaling
Although insulin enhances glucose transport and metabolism, muscle contraction initiates these effects independently of insulin. Both insulin and exercise increase the translocation of the glucose-trafficking molecule, glucose transporter isoform 4 (GLUT4), from the intracellular pools to the cell surface, resulting in increased glucose transport across the sarcolemma. However, the intracellular pools from which GLUT4 is recruited during exercise are different from those used during insulin-stimulated GLUT4 translocation. The expression of the gene encoding the GLUT4 enzyme (SLC2A4) is under the direct control of the MADS (MCM1, agamous, deficiens, and serum response factor) box transcription enhancer factor 2 (MEF2) and GLUT4 enhancer factor (GEF) genes. Exercise may increase the binding affinity of the MEF2 and GEF proteins to the GLUT4 promoter after direct physical interaction with PGC-1α (12).
In insulin-deficient individuals, the capacity to translocate GLUT4 to the cells in response to exercise is retained; by consequence, exercise training can markedly improve glucose tolerance. In those who are glucose intolerant but not insulin deficient, which is common in prediabetic obesity, exercise also augments insulin action via phosphoinositide 3 kinase and MAP kinases, such as p38, the latter of which acts on PGC-1α to enhance glucose transportation and metabolism. These effects are particularly evident within skeletal muscle and liver and persist well after exercise has ceased to an extent where intracellular fuel stores can be replenished and systemic glucose homeostasis is maintained.
Glucose and Lipid Metabolism
A positive energy balance, brought about by habitual physical inactivity and caloric excess, manifests as obesity that frequently precedes the onset of type 2 diabetes. This is partly because ectopic lipid accumulation, a feature of obesity, inhibits insulin signaling. Although the pancreatic β cells compensate by secreting more insulin in the short-term, this stressful process can eventually cause β-cell failure. At this point, the endogenous insulin concentrations rapidly decline and the glucose levels increase, which, without treatment with exogenous insulin, can cause serious damage to the vasculature of the kidneys, heart, lower limbs, and the eyes.
Physical activity increases energy expenditure, which activates numerous molecular cascades that enhance lipid and glucose oxidation. However, functional defects in the genes, which encode these cascades, increase the susceptibility to type 2 diabetes and may modify the effects of physical activity on metabolism. The orphan nuclear receptors, estrogen-related receptors (ERRs), are activated by PGC-1α and play important roles in cellular energy metabolism. Three isoforms of ERR exist (α, δ, and γ), all of which are overexpressed in tissues with high oxidative capacities, such as exercising skeletal muscle, brown fat, kidney, and heart. PGC1α directly controls ERR-α function and ERR-α-dependent transcription by binding to the C terminus AF2 domain of ERR-α. Estrogen-related receptor α cooperates with PGC-1α to induce the expression of genes involved in the transportation and the oxidation of cellular fatty acids, glucose, and glycogen in mitochondrial respiration, fusion, and biogenesis, and in contractile responses. Estrogen-related receptor α mRNA activity increases markedly after intense aerobic activity, peaking at around 2 h after exercise and remains above basal levels for up to 24 h.
Reduced Hepatic Glucose Production
Energy deficit induced by prolonged strenuous exercise or fasting increases the requirement for glucose and stimulates a molecular cascade that promotes hepatic glucose production. PGC1α coactivates many of the hepatic glucogenic genes, which include cyclic AMP-responsive element-binding protein, hepatocyte nuclear factor 4 α (HNF4a), forkhead box O1A (FOXO1), liver carnitine palmitoyltransferase I, cyclic AMP, phosphoenolpyruvate carboxykinase 2 (via ERR-α), and glucocorticoid receptor.
Glucose-6-phosphatase (G6Pase) is primarily expressed in liver and kidney and hydrolyses glucose-6-phosphate to glucose and inorganic phosphate, which constitutes the final step of the gluconeogenic pathway. Deficiency of G6Pase causes glycogen storage disease type 1a (also known as von Gierke disease), which manifests as a complex phenotype characterized by chronic hypoglycemia, hyperlipidemia, and organ enlargement. The overexpression of G6Pase is thought to contribute to excessive hepatic glucose production that occurs in type 1 and type 2 diabetes. PGC1α stimulates G6Pase gene expression via a highly conserved region of the G6Pase promoter that binds HNF4a. The abolition of HNF4a completely attenuates the ability of PGC-1α to coactivate gluconeogenic genes, such as G6Pase and phosphoenolpyruvate carboxykinase. The peroxisome proliferator-activated receptor-γ/retinoid-X receptor α heterodimer is also likely to be an important mediator of PGC-1α-induced expression of G6Pase.
In mice, insulin suppresses the PGC-1α-initiated glucose production in liver, possibly via a feedback mechanism mediated by FOXO1. This is suggested by studies showing that the coexpression of a mutant insulin-insensitive FOXO1 allele completely reverses the hepatic PGC-1α suppression by insulin. Therefore, lowering the hepatic insulin requirements in insulin-resistant individuals through exercise training could indirectly influence hepatic PGC-1α activity.
PGC-1α NUCLEOTIDE SEQUENCE VARIATION AND DIABETES-RELATED TRAITS
Although PGC-1α is a highly polymorphic gene, almost all of the multiple epidemiological studies that have assessed the association between PGC-1α sequence variants and type 2 diabetes have focused on a single nonsynonymous variant (Gly482Ser). By pooling the data from these studies, we showed that Gly482Ser is associated with a modest elevation in risk of type 2 diabetes; for each copy of the minor Ser482 allele, the odds of having type 2 diabetes increases roughly by 11% (1). Ling et al. (10) reported lower PGC-1α mRNA levels in carriers of the Ser482 allele compared with those of Gly482 homozygotes. However, there is very little evidence to show that Gly482Ser disrupts PGC-1α function. Thus, the associations reported for Gly482Ser with diabetes and mRNA levels may be due to high linkage disequilibrium with a hitherto unknown functional variant.
Only a few epidemiological studies have reported on PGC-1α sequence variants in the context of physical activity. In the study by Ling et al. (10), the Gly482Ser genotype associated with the level of cardiorespiratory fitness (VO2max) in such a way that VO2max was lower in Ser482 carriers in comparison with that of Gly482 homozygotes, which is consistent with our own observations (11). Furthermore, we also found that the Gly482 allele frequency is approximately 10% higher in elite endurance athletes compared with unfit but nondiabetic population controls (11). Indeed, the frequency of the Ser482 allele in unfit individuals is similar to that reported in individuals with type 2 diabetes of the same ethnic origin (11). Thus, because a low level of cardiorespiratory fitness is a strong and independent risk factor for type 2 diabetes, it is plausible that the association between PGC-1α genotype and type 2 diabetes is mediated by cardiorespiratory fitness level (Fig. 1).
Although cardiorespiratory fitness may be determined to an extent by variation at the PGC-1α gene, it is possible that the risk of being unfit and developing diabetes conveyed by these genetic variants can be modified by physical activity. Testing this hypothesis adequately requires suitably designed clinical trials, of which none has been reported yet. However, we have shown in observational data that physical activity and Gly482Ser genotype interact to modify the level of cardiorespiratory fitness. In this study, individuals with the Ser482Ser genotype tended to be less fit than Gly482 allele carriers when sedentary; however, when physically active, there was no difference in fitness levels between genotype groups (4). Thus, although the carriers of the Ser482 allele may be particularly susceptible to the adverse effects of sedentary behavior, a physically active lifestyle may defer the genetic predisposition toward a low level of fitness in Ser482 allele carriers.
Despite limited empirical supporting evidence, many believe that gene-lifestyle interactions underlie the increasing prevalence of type 2 diabetes in most industrialized populations. The ability to identify true and clinically meaningful gene-lifestyle interactions will depend on the appropriate choice of candidate genes. We assert that a biologically plausible candidate gene must encode a protein important for the regulation of a relevant phenotype (or one that regulates the expression of other such genes), be responsive to changes in the lifestyle factor, and be characterized by functional variation within its nucleotide sequence. Genes that control the activity of subsets of genes involved in diabetes, such as transcriptional coactivators, are especially relevant because functional variability at these loci may have the widest consequences on metabolism. The PGC-1α gene, therefore, stands out as a very plausible candidate for gene-lifestyle interaction in type 2 diabetes.
Epidemiological studies indicate a role for PGC-1α sequence variation in type 2 diabetes risk directly and via interaction with physical activity. Furthermore, the molecular evidence showing that PGC-1α is responsive to exercise and mediates several key metabolic pathways involved in diabetes is strong. However, no clinical trials that test whether PGC-1α sequence variability influences an individual's response to lifestyle intervention have yet been reported; such studies are necessary to conclude whether PGC-1α is a viable target for diabetes prevention through lifestyle modification.
Because PGC-1α coactivates a large number of key metabolic genes, epidemiological and intervention studies that attempt to assess whether PGC-1α sequence variants interact with lifestyle factors may need to consider the combined effects of genetic variation across the molecular pathway that PGC-1α heads. Furthermore, because tests of gene-lifestyle interaction are especially power intensive, studies of this nature will benefit from the precise measurement of lifestyle exposures and the disease phenotypes, and will require sample sizes severalfold larger than those of existing studies. These demands are not insurmountable, but they will require specifically designed studies and coordinated efforts between groups of scientists.
Establishing a firm evidence base for gene-lifestyle interactions is important because this information may eventually be used to inform preventive strategies that seek to treat, through personalized lifestyle modification, individuals who are particularly susceptible to the adverse effects of sedentary behavior. On a similar basis, targeted therapy with sulfonylureas in individuals with maturity onset diabetes of the young who carry specific functional mutations at the HNF-1α locus has proven highly effective as a treatment strategy. The hope is that genetically guided treatment with tailored lifestyle intervention may have similar success for preventing and treating other forms of diabetes and complex diseases. In the meantime, the identification of gene-lifestyle interactions will help elucidate the molecular pathways that link sedentary behavior with disease and will help improve the estimates of the risk of disease attributable to specific behaviors in genetically determined subgroups of the population.
The authors thank Dr. Bert Boyer (University of Alaska, Fairbanks), the reviewers, and the editor for their helpful comments on the manuscript.
1. Barroso, I., J. Luan, M. Sandhu, P.W. Franks, V. Crowley, A. Schafer, S. O'Rahilly, and N. Wareham. Meta-analysis of the Gly482Ser variant in PPARGC1A
in type 2 diabetes
and related phenotypes. Diabetologia
2. Bouchard, C., and T. Rankinen. Individual differences in response to regular physical activity. Med. Sci. Sports Exerc.
3. Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes
with lifestyle intervention or metformin. N. Engl. J. Med.
4. Franks, P.W., I. Barroso, J.A. Luan, U. Ekelund, V.E. Crowley, S. Brage, M. Sandhu, R.W. Jakes, R.P. Middelberg, A.H. Harding, A.J. Schafer, S.O'Rahilly, and N.J. Wareham. PGC-1alpha genotype modifies the association of volitional energy expenditure with Vo2
max. Med. Sci. Sports Exerc.
5. Franks, P.W., U. Ekelund, S. Brage, M.Y. Wong, and N.J. Wareham. Does the association of habitual physical activity with the metabolic syndrome differ by level of cardiorespiratory fitness? Diabetes Care
6. Kahn, B.B., T. Alquier, D. Carling, and D.G. Hardie. AMP-activated protein kinase: ancient energy gauge provides clues to modern understanding of metabolism. Cell Metab.
7. Knowler, W.C., E. Barrett-Connor, S.E. Fowler, R.F. Hamman, J.M.Lachin, E.A. Walker, and D.M. Nathan. Reduction in the incidence of type 2 diabetes
with lifestyle intervention or metformin. N. Engl. J. Med.
8. Koulmann, N., and A.X. Bigard. Interaction between signalling pathways involved in skeletal muscle responses to endurance exercise. Pflugers Arch.
9. Kuhl, J.E., N.B. Ruderman, N. Musi, L.J. Goodyear, M.E. Patti, S.Crunkhorn, D. Dronamraju, A. Thorell, J. Nygren, O. Ljungkvist, M.Degerblad, A. Stahle, T.B. Brismar, A.K. Saha, S. Efendic, and P.N.Bavenholm. Exercise training decreases the concentration of malonyl-CoA and increases the expression and activity of malonyl-CoA decarboxylase in human muscle. Am. J. Physiol. Endocrinol. Metab.
10. Ling, C., P. Poulsen, E. Carlsson, M. Ridderstrale, P. Almgren, J.Wojtaszewski, H. Beck-Nielsen, L. Groop, and A. Vaag. Multiple environmental and genetic factors influence skeletal muscle PGC-1alpha and PGC-1beta gene expression in twins. J. Clin. Invest.
11. Lucia, A., F. Gomez-Gallego, I. Barroso, M. Rabadan, F. Bandres, F. San Juan, J.L. Chicharro, U. Ekelund, S. Brage, C.P. Earnest, N.J. Wareham, and P.W. Franks. PPARGC1A
genotype (Gly482Ser) predicts exceptional endurance capacity in European men. J. Appl. Physiol.
12. McGee, S.L., D. Sparling, A.L. Olson, and M. Hargreaves. Exercise increases MEF2- and GEF DNA-binding activity in human skeletal muscle. FASEB J.
13. Norrbom, J., C.J. Sundberg, H. Ameln, W.E. Kraus, E. Jansson, and T.Gustafsson. PGC-1alpha mRNA expression is influenced by metabolic perturbation in exercising human skeletal muscle. J. Appl. Physiol.
14. Pilegaard, H., B. Saltin, and P.D. Neufer. Exercise induces transient transcriptional activation of the PGC-1alpha gene in human skeletal muscle. J. Physiol.
15. Russell, A.P., J. Feilchenfeldt, S. Schreiber, M. Praz, A. Crettenand, C.Gobelet, C.A. Meier, D.R. Bell, A. Kralli, J.P. Giacobino, and O.Deriaz. Endurance training in humans leads to fiber type-specific increases in levels of peroxisome proliferator-activated receptor-gamma coactivator-1 and peroxisome proliferator-activated receptor-alpha in skeletal muscle. Diabetes
16. Short, K.R., J.L. Vittone, M.L. Bigelow, D.N. Proctor, R.A. Rizza, J.M.Coenen-Schimke, and K.S. Nair. Impact of aerobic exercise training on age-related changes in insulin sensitivity and muscle oxidative capacity. Diabetes
17. Sriwijitkamol, A., J.L. Ivy, C. Christ-Roberts, R.A. Defronzo, L.J.Mandarino, and N. Musi. LKB1-AMPK signaling in muscle from obese insulin-resistant Zucker rats and effects of training. Am. J. Physiol. Endocrinol. Metab.
18. Timmons, J.A., J. Norrbom, C. Scheele, H. Thonberg, C. Wahlestedt, and P. Tesch. Expression profiling following local muscle inactivity in humans provides new perspective on diabetes-related genes. Genomics