The major disease epidemics of modern society are not those of contagion, but are the result of lifestyle imposed upon our genetic pre-disposition. Unrestricted access to calorie-dense food, along with a reduction in physical activity, has resulted in a rapid rise in metabolic disorders. One such condition, type 2 diabetes (T2D), has increased dramatically in recent times, with the International Diabetes Foundation estimating that 371 million people worldwide have T2D, with this number expected to increase to greater than 550 million by 2030 (http://www.idf.org/diabetesatlas/5e/Update2012). T2D is characterized by fasting blood glucose levels higher than 7.0 mM or two-hour blood glucose levels higher than 11.1 mM after a glucose tolerance test. T2D rarely occurs in isolation and is frequently associated with a number of comorbidities, including obesity, dyslipidemia, cardiovascular disease, and inflammation, collectively referred to as the metabolic syndrome.
A central aspect of the disorders comprising the metabolic syndrome is insulin resistance; defined as an impaired ability for insulin to regulate fuel metabolism in target tissues. With respect to glucose homeostasis the main insulin-responsive tissues involved are skeletal muscle, liver and adipose tissue. Under normal physiological conditions, insulin is released into the circulation from the beta cells in the islets of Langerhans in the pancreas in response to the ingestion of a meal. Upon binding to its receptor, insulin stimulates a well-described signaling cascade  involving the phosphorylation, docking and translocation of a series of signaling molecules, ultimately leading to alterations in specific endpoints of glucose and lipid metabolism (Figure 1):
In skeletal muscle, insulin promotes the translocation of the glucose transporter GLUT4 to the plasma membrane to increase glucose uptake and also stimulates glycogen synthesis.
The major hepatic actions of insulin are the promotion of glycogen and lipid synthesis and the suppression of gluconeogenesis.
In adipose tissue, insulin stimulates GLUT4-mediated glucose uptake and lipid synthesis, and additionally represses lipolysis, leading to net lipid accumulation.
In the insulin resistant state, the effect of insulin on the above pathways is compromised, leading to insufficient uptake of glucose into tissues and an impaired suppression of hepatic glucose output. To overcome the diminished effectiveness of insulin, the pancreatic beta cells secrete more insulin. The ensuing hyperinsulinemia can adequately compensate for the insulin resistance in most of the population, however in genetically susceptible individuals, the beta cells ultimately fail in the face of the increased workload and this leads to elevated blood glucose levels and T2D. Thus insulin resistance can be considered a very early and important player in the pathogenesis of T2D.
At the molecular level, the precise mechanisms responsible for insulin resistance are not fully elucidated. Studies have reported overactivation of stress-related and inflammatory pathways in tissues of insulin resistant humans and rodents. For example, ER stress was shown by the Hotamisligil lab to be present in the liver of obese mice and subsequent studies using chaperones that reduce ER stress revealed improvements in metabolic homeostatsis [2,3]. Oxidative stress has also been implicated in the development of insulin resistance, with studies showing elevated reactive oxygen species generation in insulin resistant cell models, rodents and humans [4-6]. Finally, inflammation in adipose tissue and liver (and to some extent muscle) has been reported in obese, insulin-resistant humans and rodents [7,8]. While the above factors are often described as causative players in the development of insulin resistance, it still remains unresolved whether they are the primary factors leading to diminished insulin action, or if they arise as a consequence of insulin resistance.
One factor that is one of the earliest defects associated with insulin resistance and T2D is lipid accumulation in non-adipose tissues [9-13]. Under conditions of excess nutrient supply, fatty acids and their metabolites inappropriately spillover into tissues such as skeletal muscle, liver and the heart, precipitating defects in insulin action. More specifically, while elevated triglycerides are frequently reported in tissues of insulin resistant humans and rodents, the accumulation of metabolically active long chain acyl-CoAs (LCACoAs) and other cytosolic lipid metabolites, such as ceramides and diacylglycerol (DAG), are considered to be more directly linked with insulin resistance [9,10]. In support of this, the above lipid metabolites can activate many pathways and factors (e.g. protein kinase C, c-jun N-terminal kinase (JNK), reactive oxygen species, the nuclear factor κB (NFκB) pathway, protein phosphatase A2 (PPA2) and cytokines) that directly antagonize insulin signal transduction and glucose metabolism pathways [9,10].
The extent of lipid accumulation within any given tissue is determined by several factors. Under conditions of elevated lipid availability, enhanced uptake of fat into tissues contributes to greater lipid deposition [14,15]. This increased uptake is associated with greater expression and/or translocation of fatty acid transport proteins (e.g. CD36). Any impairment in the utilization (oxidation) of lipids would also be predicted to increase partitioning of lipids into storage pools. Indeed, over the last decade a popular theory has emerged suggesting that defects in mitochondrial oxidative metabolism, particularly in skeletal muscle, lead to obesity and lipid accumulation and thus may play an important role in the pathogenesis of insulin resistance and T2D .
2. Mitochondrial structure and function
The mitochondrion is the key site for energy production in cells, providing a platform for the oxidation of fuel substrates to produce ATP. During the oxidative metabolism of nutrients (primarily glucose and fatty acids under normal circumstances), reducing equivalents (NADH or FADH2) are generated from glycolysis, the TCA cycle and β-oxidation. When NADH and FADH2 are oxidized to NAD+ or FAD, electrons pass along the mitochondrial electron transport chain coupled to the pumping of protons into the intermembrane space through complex I, III and IV. The electrons are transferred to oxygen at complex IV to produce H2O. The pumped protons generate an electrochemical gradient across the inner mitochondrial membrane, which is used as the driving force for the ATP synthase (complex V) to produce ATP. The electrochemical gradient may also dissipate through uncoupling proteins (UCP), producing heat in a process referred to as thermogenesis.
2.1. Mitochondrial biogenesis
Mitochondrial function within a given tissues is regulated at a number of different levels, including the number or density of mitochondria. The biogenesis of new mitochondria involves a coordinated interaction between the nuclear and mitochondrial genomes . The mitochondrial genome encodes for 13 protein subunits of the mitochondrial respiratory complexes, as well tRNAs and rRNAs necessary for the translation of the mitochondrial-encoded proteins. The nuclear genome therefore encodes the vast majority of mitochondrial proteins and also encodes the transcription factor responsible for controlling mitochondrial transcription, namely TFAM. Proteins encoded by the nucleus are translated in the cytosol and imported into the appropriate mitochondrial compartments via a suite of import complexes . Thus it is obvious that mitochondrial biogenesis is an extremely complex process, reliant upon the exquisite orchestration of separate genomes and multiple cellular processes.
The master regulators of the mitochondrial biogenic program are the peroxisome proliferator-activated receptor gamma (PPARγ) coactivator (PGC-1) family of transcriptional coactivators. The PGC-1 proteins are promiscuous coactivators that interact with and promote transcriptional activity in the key transcription factors (described below) that regulate the expression of genes involved in mitochondrial substrate oxidation, fibre-type determination, mitochondrial biogenesis and mitochondrial function [17,19]. The PGC-1 proteins do not bind directly to DNA, but instead recruit a wide array of chromatin-remodelling cofactors to transcriptional complexes. PGC-1α was the first described member of this family, initially identified in a screen for activators of PPARγ in brown adipocytes . The other members of the family, PGC-1β and PRC were identified based on sequence homology to PGC-1α [21,22]. Overexpression and knockout studies for PGC-1 proteins have provided clear evidence that these coactivator proteins induce increases in mitochondrial oxidative capacity and promote a switch to a more oxidative fibre type in muscle [23-27].
PGC-1α appears to be the most responsive member of this family, with PGC-1β proposed to be important in the regulation of basal mitochondrial content . Environmental stimuli such as exercise, fasting and cold exposure can induce a rapid increase in PGC-1α expression and activity . PGC-1α also promotes its own expression through a feed-forward mechanism . The activity of PGC-1α is regulated by a number of post-translational mechanisms including acetylation and phosphorylation. The relative level of PGC-1α acetylation is determined by the balance between the activity of the acetyltransferase GCN5 and the NAD+ dependent deacetylase SIRT1 [30,31]. With respect to phosphorylation, the energy sensing kinase AMPK has been shown to directly phosphorylate PGC-1α and alter its transcriptional activity . Thus in responsive to specific stimuli, changes in the concentrations of key molecules within the intracellular milieu (e.g. NAD+, adenine nucleotides) results in stimulation of upstream regulators of PGC-1α activity and initiaition of the mitochondrial biogenic cascade. While the role of PGC-1α in regulating mitochondrial biogenesis is well established, recent work from the Spiegelman lab has described a novel splice isoform of PGC-1α that also regulates skeletal muscle hypertrophy , indicating there is still much to learn about the biology of the PGC-1 transcriptional coactivator proteins.
2.2. Nuclear transcription factors involved in mitochondrial biogenesis
PGC-1 proteins orchestrate the mitochondrial biogenic program by promoting transcriptional activity through a variety of transcription factors. Nuclear transcription factors bind to specific sequences in gene promoter regions to regulate transcription of a subset of specific genes. The key transcription factors that regulate many of the genes involved in the respiratory chain and other mitochondrial pathways are described below.
2.2.1. Nuclear Respiratory Factor 1 (NRF-1)
NRF-1 plays a crucial role in coordinating nuclear and mitochondrial gene expression. It induces the expression of TFAM, as well as other components of the mitochondrial transcriptional machinery (e.g. TFB1M and TFB2M) [34,35]. NRF-1 also promotes the expression of mitochondrial import proteins that are involved in transporting nuclear-encoded proteins into mitochondria. Forced overexpression of NRF-1 in skeletal muscle in mice, results in increased expression of a subset of mitochondrial proteins, but no net increase in mitochondrial oxidative capacity . Deletion of NRF-1 in mice on the other hand has been shown to be embryonic lethal, due to disruption of mitochondrial function .
2.2.2. Nuclear Respiratory Factor 2 (NRF-2/GABP)
Nuclear respiratory factor-2 (NRF-2 humans/GABP mice) is a second critical transcription factor that regulates the expression of proteins involved in mitochondrial function and biogenesis. The target genes of NRF-2 include all respiratory complex IV subunits, TFAM and a range of other proteins involved in mitochondrial transcription and replication . Consistent with the findings in NRF-1 knockout animals, NRF-2 deletion also causes a lethal phenotype, underlying the crucial importance of this transcription factor .
2.2.3. Estrogen-Related Receptors (ERR)
The estrogen related receptor family contains 3 members, ERRα, ERRβ, ERRγ. ERRs regulate the expression of a wide array of genes involved in substrate uptake, the TCA cycle, fatty acid oxidation (FAO), oxidative phosphorylation and mitochondrial fusion [40,41]. ERRα knockout mice only display a mild phenotype , however deletion of the other two isoforms results in a lethal phenotype [43,44].
2.2.4. Peroxisome-Proliferator Activated Receptors (PPAR)
Similar to the ERR family, PPARs nuclear receptors that exist as 3 separate isoforms PPARα, PPARδ, PPARγ. The expression of PPARs varies markedly across different tissues, with PPARα being highly expressed in liver, PPARδ in skeletal muscle and PPARγ in adipose tissue . PPARs are activated by long-chain polyunsaturated fatty acids and a range of lipid derivatives, and several mitochondrial genes, particularly those involved in fatty acid oxidation, are amongst their gene targets [46,47]. Several lipid lowering (e.g. fibrates) and anti-diabetic drugs (e.g. thiazolidinediones) target the PPAR proteins.
2.2.5. Yin Yang 1 (YY1)
A range of mitochondrial genes are also regulated by the transcription factor YY1. Puigserver’s group have shown that the regulation of mitochondrial oxidative function by mTOR is regulated through YY1  and recently it has been shown that mitochondrial function and morphology are abnormal in mice with muscle-specific YY1 knockout, highlighting an important role for this transcription factor in regulating mitochondrial function .
In addition to the biogenesis of new organelles, mitochondrial content is also partly determined by the rate of degradation. Indeed, mitochondrial autophagy (or mitophagy) is now recognized as a key quality control process regulating mitochondrial homeostasis . Autophagy is a conserved cellular event in which damaged organelles and proteins are degraded in a two-step process, that first involves the formation of a double membrane structure called the ‘autophagosome’, followed by the fusion of the autophagosome with lysosomes and the subsequent degradation of the enveloped contents. Mitophagy can be inititiated by a number of events that signal stress within mitochondria, such as opening of the permeability transition pore or fragmentation of mitochondria [51,52]. The list of proteins that are thought to be involved in the mitophagy process (e.g. PINK1, Parkin, Nix) is increasing rapidly . From a physiological perspective, mitophagy plays important roles in several developmental processes, such as red blood cell maturation and the removal of paternal mitochondria following fertilization of the oocyte [53-55]. However, recent evidence also suggests that abnormal mitophagic activity may contribute to the aging process, as well as a number of different diseases, such as neurodegenerative disorders .
2.4. Intrinsic factors regulating mitochondrial function
While the number of mitochondria is obviously an important determinant of the oxidative capacity of different tissues, variations in the intrinsic properties of mitochondria are also critical. Mitochondria from different sites in the body can have different capacities for the same process. For example, mitochondria from red slow-twitch and white fast-twitch muscle display very different rates of fatty acid oxidation . This difference is in line with the functional requirements of these muscles, and is likely related to the differences in protein expression of key enzymes in this pathway . In addition to differential expression of proteins within specific pathways, another emerging factor that may influence mitochondrial oxidative capacity is post-translational modifications of mitochondrial enzymes. Following translation, many different facets of protein function (activity, subcellular localization, protein-protein interactions) can be altered by the addition of functional groups to specific residues in the protein. The most well-described PTM is likely phosphorylation, and proteomics screens have revealed widespread phosphorylation of mitochondrial proteins . The activity of specific mitochondrial proteins, such as uncoupling protein 3, has been shown to be directly regulated by phosphorylation . Other recent work has shown that perhaps the most abundant PTM in mitochondria is lysine acetylation. Acetylation involves the transfer of an acetyl group from acetyl-CoA to a lysine residue in specific proteins. More than 30% of mitochondrial proteins have been shown to be acetylated, and reversible lysine acetylation/deacetylation has been shown to impact on the activity of a large range of mitochondrial enzymes involved in virtually all metabolic pathways within this organelle . In addition to phosphorylation and acetylation, numerous other PTMs (e.g. glutathionylation, nitrosylation, succinylation) have been described as being present in mitochondria [62-64], and their functional importance remains to be determined.
2.5. Mitochondrial dynamics
Mitochondria are not static organelles, but exist largely as a reticular network. Mitochondria are constantly engaged in the process of fusion and fission, providing morphological plasticity to allow adjustments in response to the prevailing cellular stresses and metabolic requirements . Mitochondrial fusion is mediated by the mammalian GTPases mitofusin 1 and mitofusin 2, as well as optic atrophy protein 1 (Opa1). Fusion occurs in a two-step process, which initially involves fusion of the outer membrane (mediated by mitofusins), followed by subsequent fusion of the inner membrane (driven by Opa1) [66,67]. Fission is regulated by another GTPase, dynamin-related protein 1 (Drp1), which resides in the cytosol and is recruited to the mitochondrial surface to engage other key components of the fission machinery (e.g. Fis 1) [68,69]. The fusion process is thought to allow two mitochondria to functionally complement each other through the exchange and repartitioning of their respective components (e.g. copies of the genome, metabolic enzymes and metabolites). Fission on the other hand is important both in the separation of the organelle into daughter cells during cell division and also in isolating and targeting damaged mitochondria for degradation. Collectively the balance of fusion and fission allows mitochondria to form a spectrum of shapes from small individual units to elongated interconnected networks.
In muscle cells, the mitochondrial network is arranged into two discrete, but interconnected pools – the subsarcolemmal (SS) pool near the cell surface, and the intermyofibrillar (IMF) pool in the interior of the cell between myofibres [70-72]. These two pools of mitochondria have been reported to display some differences in their metabolic characteristics, with SS mitochondria appearing to be more responsive to increase their oxidative capacity following an exercise stimuli than IMF mitochondria [57,73]. Despite the differences between mitochondrial pools, it has been proposed that the arrangement of mitochondria may important for efficient mitochondrial function; SS mitochondria have greater access to oxygen and metabolic substrates, and the proton gradient generated through substrate oxidation in the SS pool may potentially contribute fuel ATP synthesis in the IMF pool, where energy demands are highest during contraction .
3. Mitochondrial dysfunction in muscle and its association with insulin resistance
As detailed above, mitochondria represent complex organelles and perturbations in any aspect of mitochondrial regulation and function, could impact on metabolic homeostasis. The mitochondrial theory of insulin resistance has developed over the last 10-15 years and is based on the notion that defective mitochondrial metabolism will result in inadequate substrate oxidation, leading to a buildup of lipid metabolites and the subsequent development of insulin resistance. Support for this theory comes from many studies in humans and rodents, which have largely examined skeletal muscle and are reviewed below.
In the late 1990’s and early part of last decade, several groups published studies showing that muscle from obese and insulin resistant subjects displayed reduced oxidative enzyme activity [74-76]. Some of these studies also examined lipid oxidation either in muscle homogenates, or by making RQ measurements across the leg, and it was shown that fatty acid oxidation was also decreased in obese, insulin resistant subjects compared to age-matched controls, potentially suggesting that defects in mitochondrial metabolism may be involved in the development of obesity and insulin resistance [74,75]. In 2002, Kelley’s group showed that there was lower NADH:O2 oxidoreductase activity and reduced mitochondrial size, as determined by electron microscopy, in muscle of obese subjects with insulin resistance and/or T2D compared to controls . A year later, two influential microarray studies were published, reporting a coordinated downregulation of genes involved in mitochondrial biogenesis and oxidative phosphorylation in subjects with T2D and non-diabetic individuals with a family history (FH+) of T2D [78,79]. These microarray studies were considered particularly important, as they documented a reduction in the master regulator of mitochondrial biogenesis, PGC-1α, and thus they provided a mechanism for the reduced oxidative gene expression. They were also important, as they showed that abnormal mitochondrial gene expression could be observed in insulin resistant relatives of patients with T2D and thus may be a pathogenic factor in the ‘pre-diabetic’ state. Overall the conclusion from these studies was that depressed PGC-1α levels, due to genetic predisposition, physical inactivity, or excessive caloric intake, could lead to a reduction in mitochondrial content, predisposing individuals to develop insulin resistance and T2D.
In the ensuing decade since these landmark studies were published, there has been intense research into this field and one issue that has arisen is how to best measure mitochondrial function/dysfunction. Numerous approaches have been employed, including measurements of parameters in frozen muscle samples (e.g. mRNA, protein content, oxidative enzyme activity and mtDNA), functional assessment of substrate oxidation in fresh samples (e.g. radiolabelled fatty acid oxidation, mitochondrial respiration measurements) and non-invasive magnetic resonance spectroscopy (MRS) with 31P or 13C to determine
In line with the microarray studies noted above, mRNA levels for a variety of mitochondrial genes have been shown to be reduced in muscle biopsies obtained from various insulin resistant populations, including lean insulin-resistant offspring of patients with T2D , obese subjects , patients with polycystic ovarian syndrome  and subjects with established T2DM [83,84]. The level of mtDNA was also shown to be lower in both obese, insulin resistant subjects and obese subjects with T2D [85,86]. Heilbronn et al.  demonstrated reduced protein expression of respiratory chain subunits in obese insulin-resistant subjects and consistent with these findings, a recent proteomics study comparing lean, obese and T2D subjects, showed patterns of reduced mitochondrial proteins in the insulin-resistant subjects . The activity of specific enzymes involved in oxidative pathways have been reported to be lower in various insulin-resistant populations [81,86,88,89] and additionally electron microscopy studies have reported reduced mitochondrial size and density in insulin-resistant muscle [77,80,86]. Interestingly, in the studies reporting mitochondrial deficiencies, there has been disparate results regarding which population of mitochondria may underlie the functional defects. Ritov et al.  reported that the number and functional activity of subsarcolemmal mitochondria was reduced in obesity and T2D, while a more recent study found similar subsarcolemmel mitochondrial content in lean controls, lean insulin-resistant non-diabetic subjects and insulin-resistant T2D subjects, however intermyofibrillar mitochondrial content was reduced in the latter two groups . Differences in mitochondrial function may not only be present within different intramuscular populations, but also between different muscles across the body. Rabol and colleagues used high resolution respirometry to measure mitochondrial function in saponin-permeabilised fibres from m. deltoideus and m. vastus lateralis and observed reduced respiratory capacity only in the leg muscles of type 2 diabetic subjects compared to lean controls .
In addition to the above studies, several investigators have also measured
A number of investigations have sought to determine if there is an intrinsic difference in the functional capacity per mitochondrion that may underlie the reductions in mitochondrial function reported with MRS. Some studies examining respiration or fatty acid oxidation in isolated mitochondria or permeabilised muscle fibres, have reported that the functional capacity per mitochondrion in insulin resistant and/or type 2 diabetic subjects is similar or only very mildly reduced [85,88,91,101-103] in insulin-resistant individuals, but when normalized to muscle mass, a substantial reduction is seen in insulin-resistant subjects [85,88,101]. These studies therefore only see marked differences when mitochondrial capacity is expressed per unit mass of skeletal muscle and thus indicate that
One limitation of the aforementioned studies is that they only provide static measurements of different populations at a given time and are unable to delineate whether the observed defects in mitochondrial metabolism are primary drivers of insulin resistance or arise as a consequence of decreases in insulin action. In this regard, intervention studies in rodents and humans have provided some experimental evidence that manipulations which result in declines in insulin action, are also associated with mitochondrial dysfunction. For example, infusion of fatty acids into humans for 6–48h to mimic the effects of chronic lipid overload resulted in a robust induction of whole-body insulin resistance and reduced insulin-stimulated ATP synthesis rates and expression of mRNA encoding PGC1α and other mitochondrial genes in muscle [104-106]. In healthy male subjects, high-fat feeding for 3 days was sufficient to reduce mRNA levels of PGC1α, PGC-1β and several other mitochondrial genes in skeletal muscle . Similarly, genetic, or high-fat diet-induced obesity and insulin resistance in rodents has been reported by several groups to reduce mitochondrial gene expression, protein expression and mitochondrial respiration in skeletal muscle [107-111]. Providing additional evidence of a link between mitochondrial dysfunction and insulin resistance is the fact that antiretroviral therapy used to suppress human immunodeficiency virus infection causes insulin resistance in association with mtDNA copy number . Collectively, the above studies illustrate that there are many instances where defects in mitochondrial metabolism and impairments in insulin action occur in conjunction with each other in skeletal muscle.
4. Mitochondrial dysfunction in tissues other than muscle
The liver plays a major role in regulating glucose homeostasis, producing glucose during the fasting state and storing glucose after the ingestion of a meal. Hepatic insulin resistance causes impaired glycogen synthesis and reduced suppression of endogenous glucose and is closely correlated with excess accumulation of lipid in liver. Chronic elevation of liver lipid content is referred to as non-alcoholic liver disease (NAFLD) and this condition progresses to non-alcoholic steatohepatitis (NASH) when inflammatory and fibrotic processes become involved.
A range of different parameters have been studied in rodents and humans with respect to liver mitochondrial metabolism. The collective findings indicate that the liver appears to be able to adapt to an excess of lipid by upregulating fatty acid oxidative capacity and TCA cycle activity, but this is not always coupled to a concomitant increase in electron transport chain activity, and as a consequence reactive oxygen species are produced (see  for an excellent review on the topic). There are also some
4.2. Adipose tissue
4.2.1. White adipose tissue
White adipose tissue (WAT) serves a principal role as the most important energy store in the body. However it has become increasingly clear over the last decade that WAT is also an active endocrine organ, releasing adipokines that influence whole-body energy homeostasis and insulin action. Mitochondrial content in WAT is low compared to other tissues, however the diversity of mitochondrial proteins in WAT has been shown to be greater than in muscle and heart . Intact mitochondrial metabolism is critical for maintaining normal WAT functions, such as the appropriate synthesis and secretion of adipokines and cycling reactions involved in lipid synthesis .
WAT mitochondrial content has been reported to be reduced in insulin-resistant humans and rodents. In women with T2D, electron transport chain genes were shown to be downregulated in visceral WAT independently of obesity and perhaps as a consequence of TNFalpha-induced inflammation . In obese humans, mtDNA copy number was reported to be lower than in control subjects and was directly correlated with basal and insulin-stimulated lipogenesis . In rodent models of genetic or dietary-induced obesity and insulin resistance, there are reductions in mtDNA copy number, mitochondrial density and mitochondrial OXPHOS activity [121-123]. Administration of thiazolidinediones promotes mitochondrial biogenesis in WAT in animals and humans, in conjunction with improved whole-body insulin sensitivity [46,123], suggesting that specific changes in WAT mitochondrial metabolism in obesity and T2D, may be imparting whole-body metabolic consequences. Indeed, recent work has shown adipose-restricted alterations in mitochondrial activity can have profound effects on global glucose and lipid homeostasis [124,125].
4.2.2. Brown adipose tissue
Unlike WAT, the principal function of brown adipose tissue (BAT) is energy dissipation, rather than energy storage. BAT has a high mitochondrial density per gram of tissue, and the unique presence of uncoupling protein 1 (UCP1) allows brown adipocytes to couple the oxidation of lipids, not to ATP synthesis, but to heat generation via proton leak across the mitochondrial inner membrane. Interest in brown adipose tissue has recently soared on the back of 3 important papers published in 2009 that unequivocally demonstrated the presence of functional BAT in humans [126-128]. There is an inverse correlation between BAT activity (as assessed by fluorodeoxyglucose PET) and obesity, suggesting that individuals with low BAT mitochondrial activity, may be prone to obesity and other metabolic diseases [126,127,129-131].
Like skeletal muscle, translocation of GLUT4 in response to insulin occurs in myocardium. This process is blunted in insulin-resistant humans and animals in association with other abnormalities in fuel metabolism ([132-134]. With respect to mitochondrial metabolism, genetic and diet-induced obesity and type 2 diabetes in rodents is associated impaired mitochondrial function [135-137]. MRS studies in individuals with T1DM, T2DM, obesity and/or NAFLD have also reported that that there is a decreased ratio of phosphocreatine:ATP in myocardium, potentially indicating derangements in mitochondrial substrate metabolism in these populations [138-142].
5. What factors lead to mitochondrial dysfunction in insulin resistance?
As noted in the previous section, mitochondrial dysfunction is frequently documented in insulin resistant states and there are many possible factors that may underlie this relationship.
5.1. Insulin resistance
Insulin is a potent anabolic hormone and it has been proposed that mitochondrial dysfunction may emerge secondary to insulin resistance. Insulin infusion in humans leads to increases in mitochondrial gene expression, higher oxidative enzyme activity and elevated ATP synthesis in muscle [143,144]. This response is attenuated in insulin-resistant T2D individuals, supporting a direct role for insulin resistance leading to mitochondrial dysfunction . Further evidence for this notion comes from a study by Karakelides et al, who showed that acute insulin removal from subjects with type 1 diabetes, caused reductions in mitochondrial ATP production and in mitochondrial gene expression in skeletal muscle . Additionally a recent study in patients with congenital defects in insulin signal transduction, reported that mitochondrial function (assessed by phosphocreatine recovery rates) in muscle was reduced in this population . Finally a recent study that induced insulin resistance by prolonged fasting, also reported defects in mitochondrial function . Overall these studies indicate that insulin can directly regulate mitochondrial biogenesis and metabolism, and therefore it is plausible that some of the mitochondrial defects observed in insulin-resistant subjects, could be a consequence of the insulin resistance itself.
5.2. Altered mitochondrial dynamics
Any perturbation in the dynamics of the mitochondrial network could potentially contribute to the pathogenesis of insulin resistance in skeletal muscle. The complex process of mitochondrial fission and fusion has been described above and alterations in key proteins mediating these dynamic events have been reported in insulin resistant and obese states. The expression of mitofusin 2 (MFN2), which appears to have additional pleitropic effects in cells beyond the maintenance of the mitochondrial network [148-152], is reduced in the skeletal muscle of obese insulin-resistant humans, type 2 diabetic humans and diabetic Zucker rats [149,153] and correlates with the capacity for glucose oxidation . Repression of MFN2 in L6E9 muscle cells and 10T/2 fibroblasts results in decreased glucose oxidation, cellular respiration, mitochondrial membrane potential, and causes fragmentation of the mitochondrial network  and liver-specific deletion of MFN2 results in glucose intolerance and impairments in insulin signaling . Recent work has also shown that mice deficient in the mitochondrial protease OMA1, display obesity and altered metabolic homeostasis, due to altered processing of the inner membrane fusion protein OPA1 and disruptions in mitochondrial morphology and fuel metabolism . It has also been reported that abnormalitieis in mitochondrial fission events may play a role lipid-induced insulin resistance. In C2C12 muscle cells, palmitic acid (but not other long-chain fatty acids) was shown to induce mitochondrial fragmentation in conjunction with insulin resistance and this effect could be blocked by genetic or pharmacological inhibition of Drp1 . Analysis of tissues from
5.3. Reduced physical activity
Physical inactivity has recently been reported to be as big a risk factor for non-communicable diseases as smoking, stressing the importance of exercise in metabolic health . Exercise is one of the major stimuli for mitochondrial biogenesis and chronic inactivity results in decreases in mitochondrial number in muscle . A number of studies have shown that obesity and other metabolic disorders are characterised by decreased physical activity levels and elevations in sedentary behaviour [160-162]. Interestingly the sedentary behaviours (e.g. sitting time) do not seem to be influenced by changes in weight and have been suggested to be biologically determined . Given these differences, it is likely that some of the mitochondrial defects reported in overweight or obese insulin-resistant subjects may be explained, in part, by low levels of physical activity.
5.4. Genetic and epigenetic factors
There is evidence in the literature that the metabolic phenotype of skeletal muscle may be strongly influenced by genetic programming. For example, despite being cultured under similar conditions for several weeks, studies have shown that primary human skeletal muscle cells in culture display a similar metabolic phenotype (e.g. gene expression and lipid partitioning) to that of the donor subject from which they originated [163,164]. Mutations in nuclear-encoded genes involved in mitochondrial function (e.g. PGC-1α, NDUFB6) have been linked with insulin action and T2D, as have mtDNA deletions [165,166]. An emerging area of research is also the regulation of mitochondrial function by epigenetic factors. Barres et al showed that the promoter of PGC-1a is methylated at non-CpG sites and exposure of primary human myotubes to hyperlipidemia or inflammatory stimuli, promoted PGC-1α hypermethylation. Intriguingly PGC-1α hypermethylation was observed in muscle of T2D patients in conjunction with reduced mitochondrial density . PGC-1α hypermethylation has also been linked with insulin resistance in non-alcoholic fatty liver disease . A number of other studies have also reported that methylation of other mitochondrial genes (e.g. NDUFB6 and ATP50) as well as TFAM, can be regulated by methylation and associated with insulin resistance. [166,168-170]. One recent study has also shown that methylation of mitochondrial DNA is also correlated with severity of NAFLD . In addition to methylation, acetylation can also influence gene transcription and the potential importance of this epigenetic factor is highlighted by a recent study showing that pharmacological inhibition of HDAC1 in cells and obese animals could promote mitochondrial biogenesis and improve metabolic phenotype . Overall the above studies indicate that specific chromatin modifications may influence mitochondrial function in insulin resistance and T2D.
5.5. Oxidative stress
Oxidative stress can be defined as a chronic imbalance between the production of reactive species and the protection against these species by antioxidant defenses, ultimately leading to macromolecular damage. Reactive oxygen species (ROS) are an unavoidable byproduct of metabolic reactions within cells and a major site for ROS production is the mitochondrion . Studies from a number of different groups have shown that in genetic or diet-induced obese rodents, there is increase ROS production [4,5,174,175]. Importantly, most [4,5,175,176], but not all studies  show that insulin action is improved by genetic or pharmacological attenuatation of mitochondrial ROS production, indicating an especially important role for generation of reactive species in this organelle. Since mitochondria are particularly susceptible to oxidative attack [178,179], it is possible that overactive ROS generation in response to obesity or high dietary lipid supply, may lead to defects in mitochondrial function. Indeed, Bonnard et al. fed mice a long-term diet rich in fat and sugar and concluded that under their specific experimental conditions, oxidative stress was involved in the induction of mitochondrial dysfunction .
5.6. Post-translational regulation of mitochondrial function
As noted above, there is an emerging appreciation for the fact that specific mitochondrial enzymes and pathways may be regulated by post-translational modifications. Several groups have shown that mitochondrial acetylation is increased in tissues of diet-induced obese mice [180,181]. SIRT3 is a key regulator of mitochondrial acetylation and the expression of this deacetylase enzyme is markedly reduced in a number of different experimental models of insulin resistance and diabetes [181-183]. SIRT3 KO mice display insulin resistance in muscle  and these mice also exhibit an accelerated development of the metabolic syndrome when challenged with long-term high fat diet, in association with pronounced hyperacetylation of liver mitochondria . Interestingly, in addition to showing that SIRT3 KO mice have a compromised phenotype, Hirschey et al have also shown that a point mutation in SIRT3 that results in reduced activity of this protein, is associated with the development of metabolic syndrome in humans . The above studies suggest that altered acetylation of mitochondrial proteins may associate with insulin resistance and impaired mitochondrial function, and while further study in this field is required, there is some evidence that other mitochondrial PTMs may also be altered in insulin resistance and T2D [184,185].
6. Mitochondrial dysfunction is not always linked with insulin resistance
Despite the frequent association of mitochondrial dysfunction and insulin resistance, evidence of a cause-and-effect relationship between the two is still lacking. In fact, a substantial literature now exists in both humans and rodents directly challenging the notion that deficiencies in mitochondrial oxidative capacity are an obligate part of the link between lipid accumulation (obesity) and insulin resistance.
6.1. Human studies
Trenell and colleagues used MRS to determine basal and maximal ATP turnover in muscle of well-controlled T2D patients compared with physical activity-, age- and weight-matched control subjects and observed no difference between the two groups . A similar finding was reported in a separate population where post-exercise phosphocreatine recovery indicated similar mitochondrial function between obese patients in either the early or advanced stages of T2D and normoglycemic controls matched for age, body composition and habitual physical activity levels . A further study from the same group also recently reported similar
The studies above provide evidence that at least in those populations, mitochondrial dysfunction does not seem to be present in a number of insulin resistant groups. In line with these examples of a discordant relationship between these two variables, several human intervention studies have also shown that changes in insulin sensitivity can occur without concurrent improvements in mitochondrial function. For instance, dietary restriction in overweight and obese subjects enhanced insulin sensitivity, without altering mtDNA, cardiolipin content or NADH-oxidase activity . Improved insulin sensitivity was reported in insulin-resistant subjects with a family history of T2D following 7 days of treatment with the anti-lipolytic agent acipimox, yet mitochondrial gene expression in muscle actually declined in these subjects . Treatment of diabetic patients with rosiglitazone improved insulin sensitivity, without altering
6.2. Rodent studies
To complement the studies in humans, a number of investigators have used gene-manipulated mice to more directly test whether specifically targeting mitochondrial metabolism, can induce changes in insulin sensitivity. Mitochondrial oxidative capacity was shown to be compromised in muscle-specific TFAM knockout mice, however these animals exhibited improved glucose clearance during a glucose tolerance test and maintained insulin-stimulated glucose uptake in muscle . TFAM knockout in adipose tissue was recently shown to protect against diet-induced obesity and insulin resistance, despite causing abnormalities in mitochondrial function . Similar findings were reported in mice with liver or muscle-specific deletion of apoptosis-inducing factor. These animals exhibited a gene expression pattern of mitochondrial oxidative phosphorylation deficiency similar to that observed in human insulin resistance [78,79], however they were lean and insulin-sensitive and did not manifest the usual deleterious effects of a high-fat diet . A number of groups have also targeted other regulators of mitochondrial function in mice. Due to their key role in mitochondrial biogenesis, muscle-specific knockout of PGC-1α or loss-of-function mutation of PGC-1β produced the expected decline in markers of mitochondrial function yet insulin sensitivity in muscle was preserved or in fact slightly enhanced in these animals compared to wild-type counterparts [26,199]. Two separately generated lines of muscle-specific PGC-1α transgenic mice have shown predictable increases in many mitochondrial parameters, but these animals are insulin resistant, potentially due to excessive fatty acid delivery into muscle  or decreased GLUT4 expression . In other examples of a dissociation between insulin resistance and mitochondrial dysfunction, it has been shown that Zucker diabetic fatty (ZDF) rats display normal
Dietary studies are another robust approach to test whether mitochondrial defects can be causally linked with changes in insulin action. Feeding rats an iron-deficient diet causes a deficiency in mitochondrial electron transport chain enzymes, however this is not associated with the development of insulin resistance . In 2007, our group showed that high-fat feeding in mice induced insulin resistance and was associated with increased expression of a PGC-1α and a number of mitochondrial proteins, elevated oxidative enzyme activity and higher fatty acid oxidation rates . Furthermore we also demonstrated that similar changes in markers of mitochondrial metabolism were present in muscle from high-fat fed rats,
One interesting observation to come out of the dietary studies in rodents is that from Muoio’s group. They reported that the increased mitochondrial fatty acid oxidation that occurs in response to high-fat feeding, results in the generation of acylcarnitines as incomplete oxidation products . This acylcarnitine signature, which is thought to disturb the mitochondrial CoA equilibrium and indicate a state of mitochondrial stress, has also been observed in humans with obesity and T2D [215,216]. Whether the acylcarnitines can directly contribute to the development of insulin resistance is currently unclear, but recent work suggests that these metabolites serve as a marker for the inability of mitochondria to efficiently transition between fuel substrates .
In summary, while there are a large number of studies that document an association between mitochondrial dysfunction and insulin resistance in lean and obese subjects, there is now a significant literature showing that alterations in mitochondrial function in muscle and insulin action are not always correlated. There are a myriad of differences between studies that may explain these discrepancies, including the particular population of individuals studied and their ethnic background and physical fitness level, the muscle group examined, the dietary regime employed in rodent and human studies (e.g. duration of feeding, fat content and composition) and the particular assay/technique used to assess mitochondrial function.
7. Are mitochondria good therapeutic targets for the treatment of insulin resistance and T2D?
While the precise role of mitochondria as pathogenic drivers of insulin resistance and T2D remains controversial, it should be recognized that even if mitochondrial defects arise secondary to the development of insulin resistance or as a consequence of obesity, reductions in oxidative fuel metabolism are likely to accelerate ectopic lipid deposition in non-adipose tissues and thereby exacerbate the insulin-resistant state. Therefore it would appear that stimulating mitochondrial biogenesis and fuel metabolism, could have beneficial effects for treating metabolic diseases. Evidence in this regard is summarised below
Exercise robustly stimulates mitochondrial biogenesis in muscle and has a multitude of health benefits, including improving insulin action. An 8 week cycling regime in sedentary elderly subjects increased muscle fatty acid oxidative capacity and in parallel improved insulin-mediated glucose disposal . A ten week exercise training program produced similar improvements in insulin sensitivity and muscle oxidative enzyme activity in both first degree relatives of T2D patients compared with age-, sex- and BMI-matched controls . Toledo reported that 4 months of exercise training in T2D patients increased skeletal muscle mitochondrial density, cardiolipin content and mitochondrial oxidation enzymes, in conjunction with a 60% improvement in insulin sensitivity . Aerobic cycling training for 10 weeks in obese, T2D male subjects induced an ~20% increase in insulin sensitivity and a 40% increase in mitochondrial content, with these changes of a similar magnitude to that observed in matched control subjects . Twelve weeks of combined progressive training was also sufficient to overcome the
7.2. Caloric restriction and modulation of sirtuin enzymes
Caloric restriction has been shown to improve insulin sensitivity in humans [222,223] and to improve markers of mitochondrial function in muscle of humans and rodents [224,225]. Many of the beneficial effects of calorie restriction have been proposed to be through stimulation of the SIRT1 pathway. Indeed studies have shown that activators of SIRT1 (e.g. resveratrol and more potent specific activators) can enhance mitochondrial metabolism and improve insulin action in rodent models of insulin resistance and T2D [226-229]. Recently it was also shown that 30 days of resveratrol supplementation improved some markers of mitochondrial function in muscle of obese subjects, in parallel with improved HOMA-IR and reduced hepatic lipid levels . These effects seem to be limited to metabolically compromised subjects, as resveratrol did not improve markers of glycaemic control in non-obese women with normal glucose tolerance . An alternative approach to using direct SIRT1 activators to mimic calorie restriction, is to increase the intracellular levels of NAD+, the obligate co-factor for the sirtuin reaction. Indeed two recent studies using different NAD+ precursors, nicotinamide mononucleotide or nicotinamide riboside, have both reported beneficial metabolic effects on metabolic homeostasis in animals models of insulin resistance and T2D [232,233].
7.3. Mitochondrial uncoupling
As noted above, energy dissipation or wasting can occur in a process known as mitochondrial uncoupling. While this occurs naturally through uncoupling proteins, there are also a pharmacological agents that can induce mitochondrial uncoupling, such as DNP (2,4-dintirophenol). These uncoupling agents are generally lipophilic weak acids, that cause mitochondrial uncoupling by transporting protons across the mitochondrial inner membrane into the matrix, deprotonating and then exiting as anions before repeating the catalytic cycle. Uncoupling agents have been successfully used in the past as obesity treatments. In the 1930’s DNP was shown to be an effective weight-loss drug in humans, providing an important proof-of-concept that the stimulation of energy expenditure by uncoupling is not necessarily compensated for by an increase in caloric intake . Despite its success as an anti-obesity therapy, DNP was withdrawn from the market in 1938 as it (like most uncouplers) has a narrow therapeutic window, with overdoses causing serious complications (even death) by compromising cellular energy homeostasis. Due to the current obesity epidemic, and illicit sales via the internet, it is alarming to see that DNP has made a comeback as a weight loss agent, with predictable lethal results [235,236]. Thus, while mitochondrial uncoupling with DNP does not appears to be a safe weight-loss therapy, other more recently described uncoupling agents may potentially have a safer profile for use in humans [237-239]. An additional approach may to be to upregulate physiological uncoupling in brown adipose tissue, via sympathomimetic agents or agonists for thyroid hormone or bile acid receptors.
7.4. Natural compounds
Natural compounds are another rich source of potential therapeutics for obesity and type diabetes, as there is often a long history of use of these compounds in the treatment of metabolic diseases states. One such compound is berberine, a natural plant alkaloid that has been used as a traditional medicine in many Asian countries. Berberine was shown to improve insulin sensitivity in a range of animal models  and there is also evidence of beneficial effects in humans . Although enhancing mitochondrial function appears to be an effective treatment for insulin resistance, we showed that berberine acted through inhibition of Complex 1 of the electron transport chain . This mild mitochondrial inhibition led to the activation of AMPK and subsequent metabolic benefits. Interestingly this pattern of mild inhibition of mitochondrial metabolism has been reported in other insulin-sensitising medicinal plants  and is also a characteristic of metformin and thiazolidinediones, which are frontline anti-diabetic therapies [242,244,245].
8. Concluding remarks
The mitochondrial dysfunction theory of insulin resistance, proposes that defects in mitochondrial metabolism are key events involved in the pathogenesis of insulin resistance. At present, the available literature does not provide strong evidence for this relationship, and there is mounting evidence that mitochondrial defects observed in insulin-resistant individuals are likely acquired (e.g. due to low physical activity or caloric excess), or develop secondary to the insulin resistance itself. Furthermore, with respect to muscle, another important issue that needs to be considered is whether the ~30% reduction in mitochondrial function that has been observed in some insulin-resistant humans would limit the oxidation of fatty acids, leading to lipid accumulation as proposed . At rest the rate of oxygen utilization in muscle is low; and the fact that muscle has enormous capacity to increase substrate oxidation over normal levels, means that there is substantial ‘spare’ capacity in this system to maintain fatty acid utilization under normal free-living conditions when energy requirements are low. Despite the unanswered questions about the precise role that mitochondria play in insulin resistance and T2D, therapies targeting this important organelle, should be explored for the the treatment of insulin resistance and its associated metabolic disorders.