|Home | About | Journals | Submit | Contact Us | Français|
The prevalence of type 2 diabetes mellitus (DM2) is increasing precipitously as a consequence of the aging population and obesity epidemic. Type 2 diabetes mellitus is characterized by both pancreatic beta cell dysfunction and insulin resistance in multiple tissues, abnormalities which precede and predict the development of DM2 [1, 2]. Because of its large mass, skeletal muscle accounts for the majority of insulin mediated glucose uptake in peripheral tissue [3, 4]. Thus, understanding the mechanism by which insulin resistance develops in skeletal muscle may offer insight into potential therapies for the treatment or prevention of DM2.
Abnormalities in skeletal muscle lipid metabolism and mitochondrial dysfunction have been proposed as possible mechanisms for skeletal muscle insulin resistance . In multiple studies of untrained subjects, elevation of intramyocellular lipid (IMCL) correlates with insulin resistance [6-8]. Lipid infusion induces insulin resistance several hours after FFA elevation [9, 10]. In humans, insulin resistance has been associated with diminished mitochondrial function [8, 11, 12], reduced expression of oxidative metabolism genes [13, 14], decreased mitochondrial size , and reduced mitochondrial density .
The purpose of this review is to explore the complex relationship between local lipid exposure, mitochondrial dysfunction, and insulin resistance at the level of human skeletal muscle. The focus will be on human studies although selected animal and cell-based studies will be presented for more detailed insights into pathophysiology. We begin with a brief overview of the normal physiology of insulin mediated glucose disposal and the associated abnormalities observed in insulin resistance. We will then discuss the evidence for local lipid excess contributing to skeletal muscle insulin resistance. Next, we will briefly review the relationship between skeletal muscle mitochondrial function and skeletal muscle insulin resistance with a focus on local lipid excess as a cause and/or consequence of mitochondrial dysfunction. After reviewing the role of lipid metabolism in skeletal muscle insulin resistance, we will present the concept that multiple abnormalities of non-muscle tissues and “cross talk” between tissues also contributes to the insulin insensitivity of skeletal muscle. Our major goal is to illustrate the conflicting nature of the evidence with regard to the biochemical pathogenesis of DM2 and, thereby, to highlight the difficulties in determining a causal relationships between lipid excess, insulin resistance and mitochondrial function in skeletal muscle.
Skeletal muscle is the largest site of insulin mediated glucose disposal in the human body [3, 4]. The physiology of skeletal muscle glucose uptake is shown in Figure 1. Normally, the binding of insulin to the insulin receptor stimulates autophosphorylation of tyrosine residues [16, 17] and subsequent activation of a receptor tyrosine kinase. This tyrosine kinase phosphorylates multiple intracellular substrates, including insulin receptor substrate (IRS) 1  and 2 , which play significant roles in the insulin response . The insulin response is mediated by IRS activation of phosphatidylinositol 3-kinase (PI 3-Kinase), a critical player in insulin signaling [18, 21-23] particularly with regards to glucose homeostasis [21, 24]. PI 3-Kinase facilitates the translocation of the insulin responsive glucose transporter (GLUT4) to the plasma membrane [25-27] through a mechanism likely mediated by phosphorylation of protein kinase B (AKT) , and/or an atypical protein kinase C (aPKC) [29, 30]. The translocation of GLUT 4 from the intracellular pool to the plasma membrane plays a crucial role in insulin mediated glucose entry into the skeletal muscle . Current evidence suggests a minimal role for insulin stimulating the ecotropic retroviral transforming sequence homolog (Cbl) and Cbl-associated pathway in mediating skeletal muscle glucose uptake [32, 33]
Contraction is another critical mediator of skeletal glucose uptake. Contraction stimulates glucose uptake through activation of p38 mitogen activated protein kinase (p38MAPK)  , calmodulin dependent kinase (CAMK) [36, 37], and AMP activated protein kinase (AMPK)  via a protein kinase (LKB1) . Enhanced glucose uptake from contraction appears to be due to increased GLUT-4 translocation due to several mediators: activation of myocyte enhancer factor-2 (MEF-2) [36, 37, 40], activation of a Rab GTPase-activating protein (AS160) [41-43] and improved insulin signaling , possibly through the tuberous sclerosis complex (TSC)  inhibitory effects on the mammalian target of rapamycin (mTOR)  and p70 S6 kinase [46, 47].
The convergence of contraction mediated glucose uptake and insulin mediated glucose uptake has been investigated. Insulin has been shown to stimulate p38MAPK activation . AMPK activation has been shown to improve insulin signaling . There is significant evidence that AS160 activation regulates both contraction mediated glucose uptake and insulin mediated glucose uptake [42, 43]
Skeletal muscle insulin resistance is characterized by impaired glucose uptake  resulting from impaired insulin receptor signaling [48-50]. Subjects with obesity  and subjects with DM2  exhibit reduced IRS-1 tyrosine phosphorylation and reduced PI 3-Kinase activity compared with their respective controls. The reduction in tyrosine phosphorylation of IRS-1 and IRS-2 has been related to increased serine/threonine phosphorylation of IRS-1 and IRS-2 . Proposed IRS serine/theronine kinases include inhibitor kappa B kinase (IKK) , c-Jun amino-terminal kinases (JNK) [53, 54], mTOR [55, 56], and protein kinase C isoforms [57, 58]. As a result of IRS serine/threonine phosphorylation, PI 3-Kinase levels are reduced which subsequently alters downstream effectors, i.e., decreased activity of aPKC [59-61] and possibly AKT [62-65] that decreases glucose uptake, presumably by reduced GLUT4 activity/translocation [5, 66, 67] A mouse model of insulin resistance (ob/ob mice) has demonstrated insulin resistance associated with impaired p38MAPK response to insulin but preserved p38MAPK response to contraction  (Figure 1).
Lipotoxicity, defined as the elevation of lipids and/or lipid metabolites within blood or tissues with subsequent metabolic derangement, is a postulated mechanism for skeletal muscle insulin resistance. In humans, the normal plasma fasting free fatty acid (FFA) concentration ranges from 350 to 550 umol/L [69, 70] and increases with prolonged fasting [69, 71], obesity [69, 72] insulin resistance  and DM2 . As the level of insulin resistance in untrained subjects correlates strongly with IMCL content as measured by muscle biopsy  or magnetic resonance imaging (MRI) , prolonged FFA exposure was initially hypothesized to increase skeletal muscle insulin resistance by increasing IMCL content.
This hypothesis was supported by findings from human studies involving lipid infusion. Lipid infusion has been shown to increase IMCL content [75, 76] and inhibit insulin stimulated glucose uptake in subjects who are healthy  and in subjects with DM2 . The inhibition of insulin stimulated glucose uptake inversely correlates with the increased FFA levels produced by the lipid infusion . An increase of plasma FFA levels to approximately 700 umol/l has been shown to disrupt insulin signaling by inhibiting insulin receptor tyrosine phosphorylation, IRS-1 tyrosine phosphorylation, PI 3-Kinase activity, and AKT serine phosphorylation . In healthy human subjects, the onset of lipid infusion induced insulin resistance (steady state FFA of approximately 1200 umol/l) was associated with a concurrent increase in activity of protein kinase C, an increase in the level of diacylglycerol (DAG) and a decrease in the level of IkappaB-α, an inhibitor of the nuclear factor—kappa B pathway . Animal studies have reported that lipid infusion induced insulin resistance is associated with activation of serine kinases such as protein kinase Θ [58, 80] and inhibitor kappa B kinase—beta (IKKB) , an activator of nuclear factor—kappa B pathway . Thus, short term exposure to a lipid substrate-rich environment (i.e. lipid infusion) stimulates skeletal muscle to shift substrate use towards fatty acids, with the consequence of inducing skeletal muscle insulin resistance and reduction of glucose uptake.
Although lipid infusion is a well documented method of producing skeletal muscle insulin resistance [10, 77], the evidence is less conclusive for high fat diets producing insulin resistance. In theory, a high fat diet should elevate blood triglyceride and FFA levels to increase skeletal muscle fatty acid accumulation and increase insulin resistance. Although animal studies have shown that chronic high fat feeding increases insulin resistance [82, 83], the data from human studies are contradictory [76 , 84, 85, 86 , 87, 88]. A portion of the uncertainty in human studies likely relates to the metabolic effects of dietary fat composition and quantity. When compared with subjects on a control diet, subjects given a Mediterranean-style diet (low in saturated fat and rich in monounsaturated and polyunsaturated fats) for 2 years reduced their weight (−5%), decreased insulin resistance (as measured by homeostasis model assessment), improved endothelial function (as measured by the L-arginine test) and lowered systemic inflammatory markers (as measured by high sensitivity-C reactive protein, Interleukin 6, Interleukin 7, Interleukin 18) . A prospective cohort study (4.4 years duration) of 13380 subjects found that participants who were the most adherent to a Mediterranean diet had a lower incidence of diabetes (0.41, 95% CI of 0.19 to 0.87) compared to participants who were least adherent to a Mediterranean diet . Unfortunately, the effects of this diet on IMCL levels and muscle lipid metabolites were not measured in these studies.
However, even when focusing on a high fat diet comprised of saturated fatty acids, the evidence for high fat intake increasing insulin resistance in humans remains unclear. In one longitudinal observational cohort study, high total and saturated fat intake was associated with higher fasting insulin levels, suggesting that a high saturated fat diet is associated with insulin resistance . However, several interventional studies in healthy humans have not documented a clear correlation between a high saturated fat diet and increased insulin resistance as measured by the hyperinsulinemic, euglycemic glucose clamp [76, 84, 86, 87] or by the frequently sampled intravenous glucose tolerance test [85, 88].
There is growing interest that increased levels of fatty acid metabolites, rather than IMCL alone, may be the mechanism responsible for lipid induced insulin resistance. The possibility that IMCL may not be harmful has been raised by the observations in endurance athletes who have levels of IMCL comparable to those of subjects with diabetes but do not have insulin resistance . In addition, elevated IMCL levels in transgenic mice do not also obligatorily correlate with insulin resistance. Transgenic mice overexpressing lipoprotein lipase in skeletal muscle have increased IMCL, fatty acyl CoA, diacylglycerol and ceramide levels in the setting of reduced muscle glucose uptake . In contrast, transgenic mice overexpressing diacylglycerol acyltransferase in skeletal muscle increased skeletal muscle IMCL in the setting of reduced diacyglycerol (DAG) and ceramide levels; these mice were protected from high fat diet induced insulin resistance . Proposed fatty acid metabolites responsible for lipid induced insulin resistance include 4-hydroxynonenal (4-HNE)  DAG [58, 79] ceramide,[94, 96] and long chain fatty acyl CoA (LC-CoA)  .
4-hydroxynonenal (4-HNE) is a by product of lipid peroxidation and an important mediator of free radical damage . In humans, 4-HNE elevation has been associated with obesity  but not with DM2  or reduced skeletal muscle mitochondrial respiration .
Diacylglycerol (DAG) is generated by lipid hydrolysis or by de novo synthesis. DAG is elevated in obese subjects compared to controls . Much interest has been on DAG activation of protein kinase C (PKC) as a mechanism of lipid induced insulin resistance and the activation of the novel PKCs (nPKC δ, Θ,ε,ŋ) in particular. In humans, lipid infusion induced insulin resistance has been associated with elevated skeletal muscle DAG level, elevated protein kinase C activity (elevated PKC δ and PKC ßII but not PKC Θ or ε) and decreased IkappaB-α protein, an inhibitor of NF-kappa beta . In rats, lipid infusion induced insulin resistance occurs in the context of increased skeletal muscle levels of DAG, PKC Θ activation, IRS-1 serine phosphorylation and decreased PI 3-Kinase activity, suggesting that lipid infusion alters insulin signaling in skeletal muscle . Due to the evidence linking DAG accumulation to abnormal insulin signaling, much effort has been made to modify DAG levels and examine the subsequent effects on insulin resistance. Reduction of DAG effects through increased fatty acid oxidation [101, 102] and increased IMCL synthesis [94, 101] have been associated with improvements in insulin sensitivity.
Ceramide is derived from long chain saturated fatty acids and serves as a precursor for complex sphingolipids. Ceramides may be produced by sphingomyelin hydrolysis or de novo synthesis from palmitoyl CoA. High levels of ceramides have been observed in obese insulin resistant humans [100, 104 ]. Exercise training is associated with improved insulin resistance and reduced ceramide levels [101, 102]. In healthy humans, total ceramide levels in skeletal muscle inversely correlate with insulin resistance, as measured by a hyperinsulinemic euglycemic clamp . Lipid induced insulin resistance (goal FFA at 1750 umol/l) has been associated with elevated skeletal muscle total ceramide levels . Lower rates of lipid infusion (goal FFA at ~ 800 umol/l) does not appear to affect intramuscular ceramide levels in humans; however, this particular study did not measure the effect of the lower rate of lipid infusion on insulin resistance or DAG levels. The mechanism of ceramide producing insulin resistance may be disruption of the insulin signaling cascade, since ceramide related inhibition of AKT [104, 107] and tyrosine phosphorylation of IRS-1 have been observed. An extensive discussion of ceramide effects on insulin resistance is presented in a review by Summers et al .
Evidence also suggests that fatty acid composition, specifically fatty acid chain length and saturation, may influence skeletal muscle insulin resistance. In humans, insulin resistance has been shown to correlate with increased saturated fatty acids in IMCL  and LC-CoA content in skeletal muscle . Animal studies have shown high fat feeding associated with increased LC-CoA and insulin resistance [82, 111]. Myotubes (C2C12) exposed to oleate or short saturated fatty acids (laurate and myristate) generate less DAG and ceramide and have less inhibition of AKT than myotubes exposed to palmitate (16:0) and other long chain saturated chain fatty acids (stearate, arachidate, and lignocerate) . In subjects with DM2, acipimox treatment for seven days reduced plasma FFA levels (−49%), improved insulin sensitivity (+33%), and decreased total intramuscular LC-CoA levels (−22%) .
Because the mitochondria are the primary cellular site for fatty acid oxidation and utilization, there is much interest in the role of reduced mitochondrial function contributing to “toxic” lipid metabolite accumulation and consequent insulin resistance. The mitochondria generate ATP through oxidative phosphorylation, which couples oxidation of reducing equivalents (ie NADH, FADH2) to the phosphorylation of ADP. Reducing equivalents generated from glycolysis (NADH), the tricarboxylic acid (TCA) cycle (NADH, FADH2) and beta-oxidation of fatty acids (NADH, FADH2) are oxidized by the electron transport chain (ETC) to produce the inner mitochondrial membrane proton gradient that drives ATP synthesis.
In humans and animals, numerous measures of mitochondrial capacity have been used, ranging from quantitation of mitochondrial density and enzyme content to functional evaluation of isolated mitochondria and in situ mitochondria. In vivo mitochondrial function has also been evaluated through the use of magnetic resonance spectrometry (MRS). Table 1 provides a selected listing of mitochondrial measurements, particularly those pertinent to this review.
Techniques for quantifying mitochondria range from electron microscopy based morphometric measures of size, morphology and tissue density , to measurements of mitochondrial DNA copy number (mtDNA), to quantification of inner mitochondrial membrane area . Measures of mitochondrial function include measures of fatty acid oxidation and oxidative phosphorylation. Indicators of fatty acid oxidation potential include measurements of enzymes associated with fatty acid transport protein into the cell [fatty acid transport protein (FAT/CD36) ], fatty acid transport into the mitochondria [carnitine palmitoyltransferase I (CPT1) ], and fatty acid beta oxidation [beta hydroxyacyl-Coenzyme A dehydrogenase (HADH)]  .Estimates of oxidative phosphorylation potential include measurement of enzymes associated with the TCA cycle (pyruvate dehydrogenase, citrate synthase, succinate dehydrogenase) and electron transport chain ( NADH dehydrogenase, succinate dehydrogenase, cytochrome c oxidase) . Active measures of mitochondria function include extraction of mitochondria to measure O2 consumption capacity with varying substrates [99, 120], CO2 production with lipid substrates [121, 122], and maximal ATP production with various substrates supporting oxidative phosphorylation . Recently, magnetic resonance spectroscopy has been employed to measure mitochondrial function in vivo by measuring the rate of ATP phosphorylation  , the rate of post ischemic phosphocreatinine recovery [124, 125] and TCA cycle flux [126, 127]. Markers of mitochondrial biogenesis include peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC-1a) [128, 129], mitochondrial transcription factor A (TFAM)  and AMPK [131, 132].
Uncoupling proteins (UCPs) play a key role in mitochondrial function. Uncoupling proteins ameliorate the generation of reactive oxygen radicals by allowing protons to leak across the inner mitochondrial membrane, reducing mitochondrial inner membrane potential . A side effect of UCP action is reduction of the efficiency of ATP synthesis in relation to the mitochondrial respiratory rate. UCP1 appears to be limited to the brown adipose tissue , UCP2 is expressed in a wide variety of tissues (including adipose tissue, skeletal muscle) , and UCP3 in humans appears to be restricted to skeletal muscle .
Evidence for reduced mitochondrial function has been seen both in the context of DM2 [11-15, 99, 124, 137] and in insulin resistant subjects without DM2 [8, 14, 127]. Specific measures of reduced mitochondrial function in such subjects include smaller mitochondrial size , decreased expression of oxidative phosphorylation genes [13, 14], lower levels of mitochondrial enzyme activity [11, 12] lower mitochondrial ATP production [11, 12], lower ATP synthetic flux rates [123, 137], slower TCA cycle flux rates , and slower post-ischemic phosphocreatinine recovery .
Reduced mitochondrial function can also take the form of “metabolic inflexibility,” a reduced ability to switch from predominant fat utilization during fasting to predominant glucose utilization during fed (insulin-stimulated) conditions . The severity of metabolic inflexibility (assessed by in vivo and in vitro measures) has been associated with the degree of insulin insensitivity in healthy subjects  and in subjects with a family history of DM2 .
The existence of multiple types of mitochondrial functional capacity measurements, however, has lead to discrepant and often contradictory observations between individual studies examining mitochondrial function and insulin resistance. When insulin resistant subjects were compared with control subjects, reductions in insulin signaling , mitochondrial density and cytochrome C oxidase activity were observed but the anticipated concordant changes in levels of succinate dehydrogenase (protein), pyruvate dehydrogenase (protein), mtDNA copy number, PGC-1alpha (mRNA and protein), and PGC-1beta (mRNA and protein) and TFAM (mRNA and protein) were not observed . In subjects with DM2, a reduction in mitochondrial respiration in the absence of expected changes in citrate synthase activity and fatty acid oxidation (HADH levels) were observed . In another study in subjects with DM2, reductions in skeletal muscle mitochondrial respiration were not observed when mitochondria respiration was normalized to mitochondrial DNA content or citrate synthase activity .
In vivo 31P MRS estimates of mitochondrial capacity using post exercise measures of phosphocreatinine and ADP recovery times have showed no differences between subjects with long standing insulin treated DM2 (>5 years), subjects with prediabetes, subjects with recently diagnosed DM2 (<1 month) and sedentary normal controls . In further support of a possible disconnect between mitochondrial ATP synthetic capacity and insulin resistance, comparison between Northern European subjects with DM2 and Asian Indian subjects with DM2 found that Asian Indians with DM2 had higher levels of insulin resistance despite the presence of higher skeletal muscle mitochondrial capacity, as measured by mitochondrial DNA copy number, mRNA of oxidative phosphorylation genes, citrate synthase enzyme activity, and maximal ATP production rate . Thus in humans, the relationship between mitochondrial function and insulin resistance remains indeterminate, with varying degrees of correlation observed depending on mitochondrial function measurement and patient selection. Given the recent findings in Asian Indians with DM2 , ethnicity may play a significant role in the etiology of insulin resistance and the mitochondrial abnormalities reported in non-Asian subjects [11-15, 99, 124, 137] are likely less relevant to the development of insulin resistance in Asian Indians.
Data from animal studies have also questioned the role of mitochondrial dysfunction in causing insulin resistance. A recent longitudinal study comparing Zucker diabetic fatty (ZDF) rats versus lean heterozygote littermates showed ZDF rats developing diabetes in association with increasing IMCL content, while skeletal muscle mitochondrial function (measured by 31P MRS, succinate dehydrogenase activity, citrate synthase activity) remained comparable to the lean littermates and mitochondrial fatty acid oxidation (measured by very long chain Acyl-CoA dehydrogenase activity) was increased compared with lean littermates . Several animal models of mitochondrial dysfunction including whole body LC-CoA dehydrogenase knockout mice , skeletal muscle specific PGC-1a knockout mice  and mice with targeted dysfunction of oxidative phosphorylation  have not demonstrated skeletal muscle insulin resistance.
Not surprisingly, the aforementioned effects of lipid excess on insulin sensitivity have generated interest in defining the role of excess lipid exposure on the development of skeletal muscle mitochondrial dysfunction. Although IMCL levels have been shown to be inversely correlate with MRS based measurement of in vivo mitochondrial function  this finding has not been consistently observed in other in vivo MRS studies [124, 137]. Exposure of isolated human and mice mitochondria to doses of free fatty acid metabolites (palmitoyl-L-carnitine, palmitoyl CoA, oleoyl CoA) similar to intramyocellular concentrations (0.5 to 2 umol/L) have been shown to stimulate ATP synthesis, whereas exposure to high levels of FFA metabolites (> 5 umol/L) showed a dose dependent inhibition of ATP synthesis . This evidence raises the possibility that exposure to high levels of FFA may cause mitochondrial dysfunction and lead to insulin resistance.
In 1963, Randle et al. proposed that excess lipid exposure inhibited glucose oxidation by increasing intramitochondrial CoA levels with subsequent inhibition of pyruvate dehydrogenase, and phosphofructokinase, two critical enzymes in the glycolytic pathway . As lipid infusion is a well established model for lipid insulin resistance in humans [10, 75-77], there is much interest in studying the effects of lipid infusion on skeletal muscle mitochondrial function. In humans, lipid infusion (6 hours, plateau FFA 2300±300 umol/L) has been shown to decrease activity of pyruvate dehydrogenase and increase mRNA levels of pyruvate dehydrogenase kinase isoform 4 ,an inhibitor of pyruvate dehydrogenase, suggesting that skeletal muscle preferentially oxidizes lipid in the setting of lipid infusion . In humans, lipid infusion (6 hour, plateau FFA 1037 ±29 umol/l) decreased mitochondrial function in vivo as measured by 31P MRS, with a decline in insulin sensitivity that correlated with declines in intramyocellular glucose-6-phosphate and insulin stimulated Pi →ATP flux . This decline in glucose-6-phosphate suggests that a major effect of lipid infusion is to decrease glucose entry into the cell, thus limiting glycolysis and glucose oxidation.
Lipid infusion effects on mitochondrial parameters have been measured. The effects of short term (6 hours or less) lipid infusion on mitochondrial gene expression have been mixed. A crossover study of lipid (6 hours, plateau FFA 1475±88 umol/l) versus glycerol (plateau FFA 129±14 umol/l) infusion in healthy humans demonstrated lipid infusion increasing IMCL, insulin resistance, and UCP3 mRNA levels while decreasing PGC-1a mRNA and PGC-1b mRNA levels . In contrast, another crossover study comparing lipid (5 hours, plateau FFA 1670 ±130 umol/l) versus saline (plateau FFA 490 ±87 umol/l ) versus saline+heparin (plateau FFA 670±90 umol/l) infusion showed no changes in PGC-1a or UCP3 mRNA levels . The discrepancy between these two studies, particularly with regards to PGC-1 and UCP3 measurements, may be related to several factors, including use of different controls (glycerol vs saline  ), shorter duration of infusion (6  vs 5 hours  ), use of heparin (no use  vs use ), concurrent hyperinsulinemic-euglycemic clamp (present  or absent ) and reference for reporting mRNA changes ( reference gene 36B4 , reference gene ß-actin , fold change from baseline levels  ) In particular, the effects of the hyperinsulinemic-euglycemic clamp on mitochondrial parameters must be considered, as the hyperinsulinemic euglycemic clamp has been shown to increase PGC-1a mRNA levels .This issue was clarified by a cross over study of prolonged lipid infusion (48 hours, plateau FFA 1730 ±430 umol/l) vs saline (48 hours, plateau FFA of 480 ±20 umol/l) in healthy subjects which demonstrated that lipid infusion increased insulin resistance (+24%) and decreased mRNA expression of numerous nuclear encoded mitochondrial genes including PGC-1 . In comparison with the studies described earlier, this study did not have a concurrent heparin infusion  and the muscle biopsy for evaluation of mitochondrial gene expression was performed after the lipid/saline infusion and prior to the hyperinsulinemic-euglycemic clamp . In contrast to the earlier studies [150, 151], microarray analysis of muscle gene expression was used, with specific confirmation of mRNA changes (in particular PGC-1) by PCR analysis (reference gene 18S ribosomal RNA) .
Although lipid infusion appears to reduce measures of mitochondrial function, the consequences of a high fat diet on mitochondrial function have been mixed. Based on the aforementioned lipid infusion data, a high fat diet would be predicted to increase FFA availability and to decrease glucose uptake by skeletal muscle. Therefore the accumulation of IMCL and lipid metabolites expected to occur with the ingestion of a high fat diet could promote mitochondrial damage. Conversely, intrinsic mitochondrial dysfunction might facilitate the development of insulin resistance as ingestion of a high fat diet in the setting of mitochondrial dysfunction may facilitate accumulation of IMCL and lipid metabolites. However, there are conflicting animal and human reports showing that a high fat diet may decrease [154, 155] not affect, [118, 156, 157], or increase [158-160] measures of skeletal muscle mitochondrial performance.
A short term high fat diet has been shown to reduce markers of skeletal muscle mitochondrial function. Compared to insulin sensitive subjects without a family history of DM2, insulin sensitive subjects with a family history of DM2 exhibited a decline in mRNA expression of PGC-1aand FAT/CD36 three hours after ingesting a single high fat meal (76% fat), suggesting that impaired ability to oxidize fat upon exposure to a high fat meal precedes the development of insulin resistance in subjects with a family history of DM2 . In another study, healthy young men who completed a three-day, high fat diet (50% fat) experienced reductions in the gene expression of PGC-1a(decline of 20%, p<0.01), PGC-1β (decline of 25%, p<0.05), and six genes involved in oxidative phosphorylation .
Compared to the short term (3 hours to 3 days) high fat diets in which skeletal muscle mitochondrial function was found to be reduced, longer duration of high fat feeding (3 to 7 weeks) in humans have had equivocal effects on measures of mitochondrial function, with alterations of beta oxidative capacity but not oxidative phosphorylation capacity. Comparing athletes feed a high fat (69% fat) versus usual diet (30% fat) for 15 days , skeletal muscle CPT1 activity increased without alterations of citrate synthase activity or HADH activity , suggesting that high fat diet in this study increased capacity for transfer of fatty acids into mitochondria but did not increase mitochondrial oxidative capacity . Similarly, athletes fed a high (53%) versus low fat (17%) fat diet for 5 weeks increased skeletal muscle IMCL levels but did not alter mitochondrial density . In untrained humans receiving a high fat (62%) diet for seven weeks, increased HADH activity but no change in citrate synthase enzyme activity was observed .
Short term high fat diets in animals have been shown to downregulate measures of mitochondrial function. A three day high fat diet (33% fat) in rats (age not stated: 320 grams average weight) downregulates gene expression of several oxidative phosphorylation enzymes (malate dehydrogenase, ATP synthase, NADH dehydrogenase, cytochrome c oxidase) in skeletal muscle . A high fat diet in mice (age 5 weeks, 45% high fat diet for 3 weeks) was associated with declines in skeletal muscle mRNA expression of numerous mitochondrial genes compared with control mice (10% fat diet for 3 weeks) as well as PGC-1a (decline of 90%, p<0.01), PGC-1β (decline of 90%, p<0.05), yet the activity of citrate synthase, cytochrome C oxidase, and HADH were not significantly changed . A high fat, high sucrose diet (36% fat, 35% carbohydrate , 50% sucrose) in mice (age 4 weeks) produced glucose intolerance at 4 weeks and diabetes at 16 weeks . Yet, these mice did not have evidence of mitochondrial dysfunction at 4 weeks as determined by measures of mitochondrial function such as mitochondrial density (mtDNA and electron microscopy), PGC-1a, and oxygen consumption of isolated muscle fibers; however, there was evidence of mitochondrial dysfunction by 16 weeks, as demonstrated by reduced mitochondrial density (mtDNA and electron microscopy), reduced mitochondrial gene expression (Cox 1, Cox 3, PGC-1a), and reduced oxygen consumption of isolated fibers. Due to the lack of normalization relative to mitochondrial content, mitochondrial dysfunction at the level of the individual mitochondria could not be assessed; nevertheless, the authors concluded that mitochondrial dysfunction at the level of skeletal muscle does not precede the development of insulin resistance .
Although prolonged high fat feeding in animals has been shown to reduce skeletal muscle mitochondrial function [155, 162], other studies have shown prolonged high fat feeding in animals to have no effect [163 , 164] or may even increase measures of mitochondrial function [158-160, 165, 166]. Rats (age not stated, weight 245 to 300 grams) fed a 5 week high fat diet (78% fat) as compared to a control diet (11% fat) had higher citrate synthase activity in skeletal muscle after 1 week of feeding, which was not maintained after five weeks of feeding, while HADH rose after 1 week of a high fat diet and continued to remain high after 5 weeks of a high fat diet . In a study of more prolonged feeding, rats (starting age 10 weeks) fed a high fat (60% fat) diet compared with rats fed a control diet (25% fat) for 36 weeks had no change in mitochondrial ATP production, citrate synthase activity, or cytochrome c oxidase mRNA levels . Insulin resistance in these rats was not specifically reported, however the high fat diet fed rats were heavier (~ 730 -770 grams) compared with the control diet fed rats (626 grams) .
Increases in mitochondrial function measures have also been observed in high fat feeding studies. Compared with a 15 day low fat diet (10.6%), recently weaned rats (age 25 days) subjected to a 15 day high fat diet (50%) had comparable succinate dehydrogenase and citrate synthase activity in skeletal muscle, yet the isolated skeletal muscle mitochondria from the high fat diet rats had higher rates of mitochondrial respiration and had higher rates of uncoupling with palmitate exposure . Mice (8 week old) fed a high fat diet (45% fat) versus a control diet (8% from fat) had higher rates of palmitate oxidation to CO2 production in muscle homogenates (at 5 and 20 weeks), increased oxygen consumption in isolated mitochondria (at 20 weeks) given sufficient substrate, and increased activity (at 5 and 20 weeks) of citrate synthase, HADH, medium chain acyl coA dehydrogenase, and CPT-1 . Similar to the previous study , a high fat diet may increase mitochondrial uncoupling  as evidenced by elevated protein expression of UCP3 (at 5 and 20 weeks) in the high fat diet mice compared with low fat diet mice. In another study, rats (age 6 weeks, personal communication with Dr. Holloszy) fed a high fat diet specifically designed to raise plasma FFAs (4 week diet, 60% fat+daily heparin injection for activation of lipoprotein lipase) demonstrated increased mtDNA copy number, increased palmitate oxidation to CO2 from muscle homogenate, increased enzyme levels of citrate synthase , increased oxidative phosphorylation enzyme levels (COX1, COX4, ATP synthase subunit α, and increased fatty acid oxidation enzyme levels (medium chain acyl CoA dehydrogenase, LC-CoA dehydrogenase, very long chain Acyl CoA dehydrogenase) . In contrast with previous studies, older rats (3 month old) subjected to a high fat diet (45% fat vs 13.8% fat for 40 days) initially increased in vitro muscle fiber mitochondrial ATP synthesis (day 14 to day 20) and mitochondrial respiration (day 14) which was not sustained and actually declined (compared with the low fat group) with continued feeding (day 40)  Mechanistically, this decline in in vitro muscle fiber muscle ATP synthesis and mitochondrial respiration observed at the conclusion of the high fat diet (day 40) was associated with increased fractional synthesis rate of mitochondrial and mixed muscle proteins, suggesting possible effects from protein turnover rates or disassociation of in vitro muscle fiber muscle ATP synthesis and mitochondrial respiration measurements from muscle/mitochondrial protein synthesis .
The inconsistency between high fat feeding and markers of mitochondrial function is likely related to inconsistency between the individual studies. A high fat diet appears to increase mitochondrial measures of beta oxidation. Whether it may affect mitochondrial oxygen consumption or ATP generation capacity may be related to contributing factors such as differences in active versus passive measures mitochondrial function, species, age at initiation of diet, diet composition, diet duration, diet effects on plasma fatty acid levels, and sampling timing during the dietary program. Most notably, the age of the animal at the time of exposure to the high fat diet may influence the effects of the high fat diet on mitochondrial function. Because enhanced mitochondrial function with high fat feeding has generally be observed in younger animals, this suggests an adaptive mechanism to high fat feeding that is more prominent in younger animals [158, 159, 166].
The available evidence demonstrates that lipid exposure may cause mitochondrial dysfunction; however, it has also been theorized that mitochondrial dysfunction may lead to accumulation of toxic lipid species with subsequent effects on insulin resistance . Mitochondria from untrained rats had higher rates of incomplete fatty acid oxidation compared with mitochondria from trained (4 weeks treadmill running) rats . In rats, prolonged blockage (4 weeks) of fatty acid beta oxidation by inhibition of CPT1 increased IMCL content and increased high fat diet induced insulin resistance . However, it appears that the effect of mitochondrial dysfunction on lipid metabolite accumulation and insulin resistance may be particularly dependent on the mechanism used to cause mitochondrial dysfunction and the extent of metabolite accumulation. Mice with blockage of fatty acid beta oxidation by a whole body knockout of LC-CoA dehydrogenase had increased skeletal muscle long chain Co A levels compared with control mice, yet maintained similar muscle glucose uptake and insulin signaling; this observation was attributed to comparable skeletal muscle diacylglycerol levels between the knockout and the control mice . Mice with partial CPT1 inhibition through a whole body knock out of malonyl-CoA decarboxylase, an enzyme which degrades malonyl-CoA — a natural inhibitor of CPT1, had lower rates of incomplete fatty acid catabolism and were protected from developing insulin resistance after a 12 week high fat diet . It has been hypothesized that increased beta oxidation in the setting of reduced TCA cycling or oxidative phosphorylation may lead to increased rates of incomplete fatty acid catabolism and reduced insulin signaling .
There is also interest in skeletal muscle insulin resistance arising from mitochondrial generation of ROS . In the process of generating ATP by oxidative phosphorylation, the mitochondrial electron transport chain produces ROS . Consequently, mitochondria have evolved several mechanisms to mitigate the effects of ROS including oxygen radical scavengers (ie mitochondrial superoxide dismutase - SOD2)  and UCPs .
Studies have examined the physiological effects of ROS by disrupting the protective mechanisms found in mitochondria. Compared with wild type mice, homozygous mutant mice lacking SOD2 have reduced succinate dehydrogenase activity and normal cytochrome c oxidase activity in the skeletal muscle and heart. These mice die approximately 10 days after birth from a dilated cardiomyopathy . Follow up preliminary studies of heterozygote mice with reduced levels of SOD2 (50%) demonstrated no difference in skeletal muscle insulin resistance (as measured by hyperinsulinemic clamp or in vitro glucose uptake at level of vastus lateralis muscle) compared with wild type mice .
In mitochondria, UCPs antagonize the generation of ROS by allowing protons to leak across the inner mitochondrial membrane and reduce the mitochondrial inner membrane potential . This process decreases the efficiency of oxidative phosphorylation by partially uncoupling respiration from ATP synthesis . Because generation of mitochondrial ROS depends on the mitochondrial inner membrane potential, reduction of the inner membrane potential through uncoupling reduces the rate of generation of ROS . Several types of uncoupling proteins (UCP 1 - 5) have been described  of which UCP3 has generated particular interest due to its expression in skeletal muscle [136, 179] and its possible role in exporting fatty acids out of mitochondria [180, 181] or transporting fatty acid peroxide anions across the inner mitochondrial membrane 
UCP3 was discovered in 1997 by screening a human skeletal muscle cDNA library . Increased function  and expression  of UCP3 has been observed with exposure to 4HNE  and ROS [133 , 162]. In humans, skeletal muscle UCP3 mRNA levels rise with fasting , lipid infusion , and saline infusion . Unlike adipose tissue UCP2 mRNA levels, skeletal muscle UCP2 and UCP3 mRNA levels do not correlate with BMI . In humans, the correlation between UCP3 expression and insulin resistance has been inconsistent. Increased skeletal muscle UCP3 mRNA levels has been observed in obesity  and DM2 [186 , 187]. Yet, reduced skeletal muscle UCP3 mRNA levels in DM2  and reduced UCP3 protein levels in DM2 have been observed as well. Although more studies are necessary to reconcile the differences, the discrepancy between UCP3 mRNA expression level in DM2 may be influenced by differences in withholding diabetes treatment prior to the biopsy (held overnight  vs more than 1 week [186, 187]) as well as use of a prebiopsy standardized diet (not used  vs 2 days [186, 187]). Thus, the human studies relating alterations in UCP3 to DM2 have remained inconclusive.
Of interest, several UCP3 knockout mouse models have been created. One mouse knockout model showed increased respiratory coupling in skeletal muscle mitochondria associated with increased generation of ROS; however these mice did not display any differences in weight, exercise tolerance, fatty acid oxidation, fasting glucose, or glucose oxidation compared with wild type mice . Another mouse UCP3 knockout model displayed fasting glucose levels comparable to wild type mice . Magnetic resonance spectroscopy has been used to measure skeletal muscle mitochondrial function in UCP3 knock out mice with contradictory results. Increased Pi → ATP flux observed with 31P NMR suggested elevated ATP synthesis, yet 13C NMR measurements from 13C acetate infusion demonstrated no change in TCA cycle flux rate. Thus, although UCP3 underexpression is associated with increased oxidative damage , the effects of the UCP3 underexpression on mitochondrial performance and insulin resistance remain uncertain.
Several mouse models of UCP3 overexpression have also been created. In one model, UCP3 was overexpressed using the rat myosin light chain 2 promoter . Compared with wild type mice, the transgenic mice had higher skeletal muscle oxygen consumption, lower fasting glucose levels, improved glucose tolerance and were resistant to diet induced obesity and diet induced insulin resistance . In another mouse model of UCP3 overexpression in skeletal muscle, the human α-skeletal actin promoter was used . These mice were hyperphagic, yet were leaner and more insulin sensitive than wild type controls . Further studies of these mice demonstrated maintenance of insulin signaling in the setting of a high fat diet. This appeared to result from lower DAG levels, reduced PKCΘ activity and preserved PI 3-Kinase activity . The effect of UCP3 overexpression on ROS remains unclear. Overexpression of UCP3 in L6 myotubes decreased the production of mitochondrial ROS and increased fatty acid oxidation . However, when L6 myotubes overexpressing UCP3 were deliberately exposed to palmitate, an increase in intracellular ROS was observed compared with control L6 myotubes . This discrepancy highlights the complex role of UCP3. Although UCP3 overexpression may reduce coupling and, thereby, reduce ROS production, perhaps the increased fatty acid oxidation arising from UCP3 overexpression may overwhelm the uncoupling effects. Thus, although UCP3 is the primary uncoupling protein in skeletal muscle, UCP3 overexpression appears to have a more significant physiological impact than its absence in diet induced obesity and insulin resistance. The mechanism of this protection remains unclear, but may be related to decreased generation of ROS or increased ability to oxidize lipids.
Although the skeletal muscle is the largest site of insulin mediated glucose disposal , skeletal muscle specific alterations may not necessarily improve overall insulin resistance. A mouse with skeletal muscle specific deletion of lipoprotein lipase exhibited improved insulin signaling and glucose uptake in skeletal muscle yet exhibited increased insulin resistance in the heart, white adipose tissue, and liver . As plasma FFA, TG and cytokine (Interleukin-1ß and Interleukin-6) were not significantly different between the knockout and wild type mice, the authors proposed that the observed effects may be due to the heart and liver shifting to lipid oxidation in compensation for the reduced skeletal muscle lipid oxidation . Mice with skeletal muscle specific knockout of PGC-1a demonstrated impaired glucose tolerance, in the surprising context of reduced beta cell function and normal skeletal muscle insulin sensitivity . Further investigation demonstrated increased expression of proinflammatory genes in the skeletal muscle (ie SOCS1, SOCS3, TNF-α, Interleukin-6, CD68) as well as increased serum Interleukin-6 levels . Exposure of beta cells to Interleukin-6 in vitro suppressed insulin secretion in response to hyperglycemia, providing evidence for “crosstalk” between skeletal muscle and beta cells .
Likewise, changes at the level of the liver and adipose tissue may influence skeletal muscle insulin resistance. In mice with adipose specific knock out of GLUT-4, muscle glucose uptake was impaired in vivo but preserved in vitro , suggesting the presence of circulatory factor (such as serum retinol binding protein 4, cytokines, etc) affecting skeletal muscle insulin sensitivity . Stress at the level of the endoplasmic reticulum in liver and adipose tissue (but not skeletal muscle) is associated with obesity and insulin resistance . Yet in mice, treatment of endoplasmic reticulum stress with orally administered ursodeoxycholic acid improved peripheral insulin sensitivity and muscle glucose uptake . These observations suggest skeletal muscle specific changes must be interpreted in the context of “cross talk” with other organs involved in glucose homeostasis. These observations suggest that the causes of alterations of skeletal muscle mitochondrial function and insulin sensitivity may be related to exogenous and endogenous factors.
This review explores the complex relationship between excess lipid exposure, mitochondrial dysfunction, and insulin resistance at the level of human skeletal muscle. Lipotoxicity — i.e., the elevation of lipids and/or associated lipid metabolites within blood and tissues with subsequent metabolic derangement — has been proposed as a possible mechanism of skeletal muscle insulin resistance. Intravenous lipid infusion is a well-documented method of inducing insulin resistance. Although IMCL content has been correlated with insulin resistance, there is increasing evidence that lipid metabolites such as 4-HNE, DAG, ceramide, and LC-CoA may play a more significant role than triglycerides in producing skeletal muscle insulin resistance.
The association between mitochondrial dysfunction and insulin resistance is unclear, particularly due to the varied options for measuring mitochondrial function. The effect of acute lipid exposure producing skeletal muscle insulin resistance in humans is well documented. The effects of chronic lipid exposure from dietary ingestion on skeletal muscle insulin resistance and skeletal muscle mitochondrial function remain disputed. The effects of skeletal muscle mitochondrial dysfunction on accumulation of lipotoxic species and skeletal muscle insulin resistance also remain uncertain. Certainly, pursuit of the role of lipid metabolites and their roles in the generation of skeletal muscle insulin resistance remain an exciting area for future research.
Moreover, alteration in skeletal muscle insulin resistance does not occur in isolation as functional perturbations of any component of the glucose homeostatic system may initiate development of insulin resistance in skeletal muscle through “cross talk” between tissues. Nevertheless, understanding pathophysiology of skeletal muscle insulin resistance remains critically important due to its role in preceding and facilitating the development of Type 2 diabetes.
Lisa Chow MD: Dr. Chow gratefully acknowledges support from NIH 5K12RR023247-02 grant.
Arthur From MD: Dr. From gratefully acknowledges support from NIH NCRR P41 RR008079 grant.
Elizabeth Seaquist MD:
Conflicts of interest:
Lisa Chow MD: none stated
Arthur From MD: none stated
Elizabeth Seaquist MD: none stated
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.