Oxidative stress and diabetes: what can we learn about insulin resistance from antioxidant mutant mouse models?

Abstract

The development of metabolic dysfunctions like diabetes and insulin resistance in mammals is regulated by a myriad of factors. Oxidative stress seems to play a central role in this process as recent evidence shows a general increase in oxidative damage and a decrease in oxidative defense associated with several metabolic diseases. These changes in oxidative stress can be directly correlated with increased fat accumulation, obesity and consumption of high calorie/high fat diets. Modulation of oxidant protection through either genetic mutation or treatment with antioxidants can significantly alter oxidative stress resistance and accumulation of oxidative damage in laboratory rodents. Antioxidant mutant mice have previously been utilized to examine the role of oxidative stress in other disease models, but have been relatively unexplored as models to study the regulation of glucose metabolism. In this review, we will discuss the evidence for oxidative stress as a primary mechanism linking obesity and metabolic disorders and whether alteration of antioxidant status in laboratory rodents can significantly alter the development of insulin resistance or diabetes.

Introduction

The prevalence of metabolic disorders is experiencing a rate of growth that is rapidly placing these diseases among the most significant that affect the world’s population. In particular, type 2 diabetes mellitus (T2DM), the most common form of metabolic dysfunction, currently affects more than 20 million people in the United States alone, with almost 60 million more diagnosed with pre-diabetes conditions [1]. Complications from diabetes include cardiovascular disease, blindness, nerve damage, and nephropathy. Thus, the increasing incidence of diabetes is a significant health concern beyond the disease itself [2]. The prevalence of T2DM is age-dependent, as more than 20% of people older than 60 years of age have diabetes compared to only about 8% of the population as a whole. Importantly, the strongest predictor of the development of T2DM is obesity or increasing fat accumulation. The global population is becoming older because of advances in medicine and at the same time becoming more obese due to an increasingly sedentary lifestyle and changes in diet towards excessive caloric intake. These two factors suggest that the growth of metabolic disorders worldwide is unlikely to slow. Understanding how factors like obesity lead to diabetes and insulin resistance has the potential to expand therapeutic options for future treatment of these diseases. Many inflammatory, endocrine, and intracellular pathways have been found to be dysregulated with obesity. However, the exact mechanism by which obesity causes metabolic diseases like T2DM is still unknown.

One of the earliest factors in the etiology of T2DM is the development of peripheral insulin resistance, or decrease in insulin sensitivity in the insulin-responsive tissues. Insulin resistance in vivo is caused either by inadequate insulin production (as defined by metabolic demands) or by abnormalities within the insulin signaling pathway [2]. This pathway is activated by binding of free insulin to the cell membrane-bound insulin receptor (IR). Subsequent activation of the intrinsic tyrosine kinase of IR then phosphorylates the β-subunit of IR which stimulates activation of the insulin signaling cascade through insulin receptor substrate protein-1 (IRS-1), phosphatidylinositol 3-kinase (PI-3K) and protein kinase B (PKB; Akt), ultimately leading to translocation of the glucose transporter (GLUT4) to the cell membrane of skeletal muscle and adipose cells and uptake of glucose [3]. Insulin resistance can be explicitly defined by abnormalities within this pathway. For example, muscle from subjects with T2DM, or from rodent models of diabetes, commonly show reduced insulin-stimulated insulin receptor and IRS-1 phosphorylation and PI-3K activity [4]. Obesity is known to drive the development of insulin resistance. However, not all obese subjects develop T2DM or insulin resistance, suggesting that the mechanism linking obesity with insulin resistance must be capable of being controlled under different circumstances.

It has been recently proposed that oxidative stress may be a primary factor in the etiology of obesity-induced insulin resistance and T2DM. It has long been known that the progression of several diseases and pathologies is greatly dependent on modulation of oxidative stress in mammalian systems [5,6]. As will be discussed in this review, it is becoming increasingly clear that consumption of adipogenic diets and the accumulation of adipose tissue can significantly increase levels of oxidative stress in mammalian systems. As a result, obesity is correlated with a drastic rise in oxidative damage to all cellular macromolecules ubiquitously among mammalian tissues [7,8,9,10,11]. Recent evidence also suggests that oxidative stress can have a significant inhibitory effect on insulin signaling in both in vitro and cell culture systems [12,13,14]. T2DM and insulin resistance are also strongly correlated with a pro-oxidant environment in both rodent and human studies, which suggests that oxidative stress may significantly alter both the pathogenesis and progression of these diseases in vivo [15,16].

A greater understanding of the role oxidative stress has in the development of T2DM and insulin resistance is important because oxidative stress can be modulated by both intrinsic and extrinsic factors, thus providing a plausible means for prevention of metabolic disorders. The focus of this review is to clarify the role of oxidative stress in the regulation of glucose metabolism and to explore whether modulation of the expression of antioxidant enzymes has a significant effect on this process. As a first step, we must first define what oxidative stress is, what effect oxidative stress has on the cell, and how eukaryotic cells defend themselves from oxidative stress.

Oxidative stress and the antioxidant defense system

Oxidative stress can be defined as a state of imbalance towards the factors that generate reactive oxygen radicals (e.g., superoxide or hydroxyl radicals) and away from the factors that protect cellular macromolecules from these reactants including antioxidants like superoxide dismutases, catalase, and glutathione peroxidases. The factors that generate reactive oxygen species (ROS) exist as products of normal cellular physiology as well as from various exogenous sources. Mitochondria are thought to be the source of most cellular ROS, specifically superoxide radicals. The reactions that generate ATP in the mitochondria require electrons from reduced substrates to be passed along the complexes of the electron transport chain. In the presence of molecular oxygen, electrons that leak from this process react and form the free radical superoxide. Superoxide anions are significant mediators in numerous oxidative chain reactions and are also a precursor to many other ROS [17]. Other significant intracellular sources of ROS include NADPH oxidases (which generate superoxide), nitric oxide synthases (nitric oxide) and lipoxygenases (fatty acid hydroperoxides) [17]. In addition, certain cell types within tissue systems may promote localized environments with elevated oxidative stress. For example, macrophages can produce localized oxidative stress as part of the inflammatory response [18]. Thus, low levels of ROS are typical within both the cell and higher order tissue and organ systems and some ROS (in particular superoxide and hydrogen peroxide) are required to support natural cellular function and regulate intracellular signaling [19]. However, excess ROS production (or reduced ROS regulation) can severely impair the cell and lead to macromolecular damage, dysfunction and death.

Under conditions of oxidative stress, free radicals that are not reduced or removed from the cellular environment can cause damage to all cellular macromolecules including nucleic acids, lipids and proteins [20]. DNA, both nuclear and mitochondrial, are susceptible to oxidation which results in mutations and single-strand breaks along with the formation of 8-hydroxyguanosine (8-OHdG) [21]. 8-OHdG is a relatively stable oxidation product and can be measured in both tissues and excreted urine which accurately represent the amount of DNA oxidation/repair rate as a measure of DNA damage within the body as a whole [22,23]. Oxidation of DNA has been strongly implicated in cellular senescence, apoptosis, and the development of cancerous cell phenotypes. On the other hand, oxidation of lipids can cause changes in structure and fluidity of cellular and organelle membranes that are detrimental to cellular processes and functions [24]. This ultimately affects cellular functions, further increasing cellular ROS concentrations [25]. In addition, oxidation of lipids may form lipid radical species that damage other cellular macromolecules. For example, lipid peroxides like malondialdehyde (MDA) and 4-hydroxynonenal (4HNE) can react with both DNA and proteins [26]. Proteins in particular are susceptible to attack by numerous forms of free radicals and ROS which can lead to many different forms of oxidative modification [5,27]. Increases in the accumulation of these forms of protein oxidative damage can lead to functional changes of proteins (generally detrimental) that can alter many cellular physiological processes [28,29]. Once oxidized, proteins must either be repaired or, if repair is not possible, degraded or cleared from the cell to minimize the potential negative effects of these damaged proteins. Almost all amino acids are susceptible to oxidative modification by one or more types of ROS. The sulfur-containing amino acids (cysteine and methionine) are unique in that there are specific enzymes to repair their oxidative damage (cysteine disulfides, methionine sulfoxides) [27,30,31]. However, oxidation to other amino acids, or unresolved damage to cysteine and methionine, can result in oxidation moieties that cannot be repaired. In cases where repair is not possible, oxidized proteins are generally labeled for degradation by the proteasome system or removed through autophagy processes. Despite the efficiency of clearance of oxidized protein, certain damaged proteins can remain and accumulate and promote cellular dysfunction [5,27].

Because of the potential detrimental effects of oxidative stress even under normal physiological conditions, aerobic organisms have evolved a complex antioxidant system consisting of both general antioxidants (that is, those that reduce oxidative stress by removal of ROS) and also specialized enzymes that can repair some forms of oxidation within cellular macromolecules. There are several enzymes involved in the antioxidant defense system, not all of which will be discussed herein. However, several enzymes merit the attention of this review because they contribute to a significantly large proportion of the antioxidant defense, because they play very specialized roles in defense and repair, and/or because they (and mutant models of these enzymes) are relatively well characterized (Table 1). For example, superoxide dismutases (Sod) reduce superoxide levels in the cell; these enzymes catalyze the conversion of the superoxide radical to molecular oxygen and hydrogen peroxide. Three main isoforms of Sod are found in mammals: CuZn superoxide dismutase (CuZnSod; Sod1), Mn superoxide dismutase (MnSod; Sod2), and extracellular superoxide dismutase (ECSod; Sod3). Each isoform is specifically localized to different cellular compartments with the primary location of Sod1 in the cytoplasm and in the mitochondrial intermembrane space [32,33], Sod2 in the mitochondrial matrix [32], and Sod3 in the extracellular fluids and interstitium [34]. Superoxide dismutases are then one of the first lines of defense against superoxide radicals produced by the mitochondria during cellular respiration and against superoxide produced by other cellular sources such as NADPH oxidases.

Table 1.

Major mammalian antioxidants discussed in this review

Antioxidant enzyme	Abbreviation used	Function	Family members
Superoxide dismutase	Sod	Conversion of superoxide to oxygen and H2O2	Sod1 (CuZnSod), Sod2 (MnSod), Sod3 (ECSod)
Catalase	Cat	Conversion of H2O2 to oxygen and water	Cat
Glutathione peroxidase	GPx	Conversion of hydroperoxides to water and alcohols	GPx1-8
Peroxiredoxin	Prdx	Reduction of hydroperoxides, peroxinitrite	Prdx1-6
Thioredoxin	Trx	Reduction of protein cysteine-thiol disulfide bonds	Trx1, Trx2
Methionine sulfoxide reductase	Msr	Reduction of methionine sulfoxide to methionine	MsrA, MsrB1, MsrB2, MsrB3

Peroxides, including those generated by Sod, are converted into water in the cell primarily by catalase, glutathione peroxidases, and peroxiredoxins. Catalase (Cat) is ubiquitously expressed among mammalian tissues and is primarily located in the peroxisomes. The primary catalytic function of catalase is the decomposition of hydrogen peroxide to oxygen and water [17]. In general, glutathione peroxidases (GPx) can reduce peroxides (including hydrogen peroxide and lipid hydroperoxides) to less toxic forms including water and alcohols. There are 8 putative GPxs that differ in tissue localization and substrate specificity, however, only GPx1 and GPx4 are nearly ubiquitously expressed. Glutathione peroxidase 1 (GPx1), the most abundant isoform of the mammalian GPxs, is ubiquitously expressed, and is responsible for much of the detoxification of H₂O₂ within the cytoplasm [17]. GPx4 is ubiquitiously expressed at low levels (compared to GPx1), and the specificity of GPx4 is for detoxification of lipid peroxides, including phospholipid hydroperoxides and hydroperoxides of cholesterol esters [35,36]. Other GPxs not discussed in this review include GPx2 (primarily located in the gastrointestine), GPx3 (plasma), GPx5 (epidydimal), GPx6 (olfactory) and GPx7 and GPx8 which actually have protein disulfide isomerase activity and participate in protein folding in the endoplasmic reticulum [37]. Peroxiredoxins (Prdx) are a relatively newly-discovered class of antioxidant with peroxidase activity that can reduce hydrogen peroxide, peroxynitrite and range of different organic hydroperoxides [38]. At least 6 different Prdx isoforms have been discovered in mammalian cells, each with specific cellular localization in most cellular compartments including cytosol, nucleus, membrane, mitochondria, or Golgi.

Detoxification of free radicals contributes to one side of the protection afforded by the antioxidant defense system. However, several components of this system can also reduce oxidative stress by removing oxidative damage from cellular macromolecules. As discussed above, Gpx4 can detoxify lipid peroxides including those phospholipids that make up cellular membranes [35]. In addition, thioredoxins (Trx) catalyze reduction of disulfide bonds in multiple substrate proteins. Through this reaction, Trxs act as antioxidants by detoxifying peroxides through peroxiredoxins and by reducing protein disulfides and methionine sulfoxides, either directly or through the actions of other oxidoreductases [39]. There are two forms of Trx in mammalian cells: Trx1 is located primarily in the cytosol while Trx2 is the mitochondrial form of thioredoxin. Methionine sulfoxide reductases (Msr) can also repair oxidation damage to proteins because they can catalytically reduce the oxidized form of methionine (methionine sulfoxide) back to unoxidized methionine [40,41]. There are two isoforms of Msr present nearly ubiquitously in mammalian cells: MsrA is located in both the cytosol and mitochondria, whereas the 3 isoforms of MsrB (MsrB1-3) in mammals seem to have discrete distribution among the cytosol and nucleus (MsrB1), mitochondria (MsrB2, MsrB3) and endoplasmic reticulum (MsrB3) [42].

Characterization of the enzymatic activity of nearly all of the components of the antioxidant defense systems described above was initially performed utilizing in vitro biochemical techniques and basic cell biology. However, it is difficult to determine whether these proteins have a physiologic role under such experimental paradigms. With the advent of gene-targeting molecular biology techniques, it has become possible to test whether genetic alterations of antioxidant genes in mice have significant effects on whole animal physiology by generating lines of knockout (or knockdown/heterozygote) or transgenic/overexpressing mice. Several of these lines of mice targeting specific antioxidants have been generated and characterized in terms of their oxidative stress resistance and the accumulation of oxidative damage to lipids, nucleic acids, and proteins. The results from these published reports are relatively equivocal in that antioxidant knockout mice tend to be sensitive to oxidative stress and show increased accumulation of oxidative damage whereas antioxidant overexpressing mice tend to be resistant to oxidative stress and show reduced oxidative damage (Table 2). For example, mice lacking Sod1 (Sod1^−/−) are particularly sensitive to agents or treatments known to elevate levels of oxidative stress such as MPTP (N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) or ischemia/reperfusion [43,44]. In addition, the levels of oxidized lipids, DNA, and proteins are higher in several tissues from Sod1^−/− mice suggesting a diminished capacity to reduce toxic ROS before they can damage macromolecules [45,46]. Conversely, overexpression of Sod1 (Sod1 TG) in mice increases resistance to toxic agents that generate oxidative stress (in both mice and cell lines derived from these mice) and reduces the accumulation of oxidized lipids [47,48,49]. In the case of some antioxidants, complete lack of these proteins (i.e., knockout) causes embryonic or post-natal lethality. Examples of these include mice lacking GPx4 (Gpx4^−/−) or Trx2 (Txn2^−/−) which do not completely develop in utero and mice lacking Sod2 (Sod2^−/−) which do not survive more than a few days post-natally [50,51,52]. However, mice carrying the wild-type allele on one chromosome and the knockout on the second (i.e, heterozygous; Gpx4^+/−, Txn2^+/−, Sod2^+/−) are viable and do show general trends of increased oxidative stress sensitivity and accumulation of oxidative damage (38–40). The stress resistance and damage accumulation phenotypes of other lines of antioxidant mutant mice are more thoroughly described in Table 2.

Table 2.

Oxidative stress in antioxidant mutant mouse models

Antioxidant	Model	Oxidative Stress	Oxidative Damage	Ref
Superoxide dismutases
CuZnSod	Sod1−/−	Sensitive	Increased (L, N, P)	[43,44,45,46]
	Sod1 TG	Resistant	Reduced (L)	[47,48,49]
MnSod	Sod2+/−	Sensitive	Increased (N, P)	[23,40,52,138,140, 141]
	Sod2 TG	Resistant	Reduced (L, P)	[142,143,144]
ECSod	Sod3−/−	Sensitive	Increased (L)	[145,146,147,148]
	Sod3 TG	Resistant	Reduced (L)	[149,150]
Catalase
	Cat−/−	No change/sensitive	Increased (L)	[151,152]
	Cat TG	Resistant	Reduced (N)	[47,48,49]
	MCAT (mito)	Resistant	Reduced (N)	[153,154]
Glutathione peroxidases
GPx1	Gpx1−/−	Sensitive	No change/increased (L)	[141,155,156,157, 158,159,160]
	GPx1 TG	Resistant	Reduced (L)	[161]
GPx4	Gpx4+/−	Sensitive	No change/increased (L, N)	[51,162,163]
	GPx4 TG	Resistant	Reduced (L)	[164,165]
Thioredoxins
Trx1	Trx1 TG	Resistant	Reduced (L)	[166,167,168]
Trx2	Txn2+/−	Sensitive	Increased (L, N, P)	[50]
Methionine sulfoxide reductases
MsrA	MsrA−/−	Sensitive	Increased (P)	[121,169,170]
	MsrA TG*	No change	No change (P)	[171]
MsrB	MsrB1−/−	ND	Increased (L, P)	[172]
Peroxiredoxins
Prdx1	Prdx1−/−	Sensitive	Increased (N, P)	[173]
Prdx2	Prdx2−/−	Sensitive	No change/increased (L, P)	[174,175]
Prdx3	Prdx3−/−	Sensitive	Increased (L)	[176,177]
	Prdx3 TG	Resistant	Reduced (L)	[125]
Prdx4	Prdx4−/−	ND	Increased (L)	[178]
	Prdx4 TG	ND	Reduced (N)	[124]
Prdx6	Prdx6−/−	Sensitive	Increased (L, P)	[179,180,181]
	Prdx6 TG	Resistant	Reduced (L, P)	[182,183]

ND = not determined

L = lipid oxidation

N = nucleic acid oxidation

P = protein oxidation

^*= only MEF from animals tested

Obesity and oxidative stress

Together, the components of the antioxidant defense system maintain the levels of ROS in the normal cellular homeostatic range. That is, under normal conditions this system reduces excess ROS to minimize damage, yet still allows for ROS required in intracellular redox signaling. However, certain pathological conditions, such as obesity, hyperglycemia and hyperlipidemia, have been shown to promote oxidative stress through elevated ROS production and/or reduced antioxidant defense. Several lines of evidence have suggested that fat accumulation (i.e., obesity) is correlated with high oxidative stress state in human populations. For example, plasma samples from obese human subjects show increased levels of 8-OHdG and in thiobarbituric acid reactive substances (TBARS), a measure of lipid peroxidation [7]. Another marker of lipid peroxidation, 8-Epi-prostaglandin F2α is also relatively high in plasma from obese human subjects [8]. High levels of oxidation associated with obesity are not limited to the plasma, but are also evident within tissues from obese individuals. Obesity is correlated with significantly higher levels of protein carbonyls in subcutaneous adipose tissue [53] and of 4HNE adducts in skeletal muscle [54]. Levels of oxidative stress and damage can also be significantly increased in laboratory models by experimental induction of obesity through either genetics or high nutrient feeding. Plasma lipid peroxidation and levels of H₂O₂ are significantly elevated in both genetically obese mice and mice fed a high fat diet [7]. These mice also show significant increase in lipid peroxidation and H₂O₂ in visceral adipose tissue [7]. Levels of protein carbonyls and protein-bound 4HNE are also elevated in the adipose tissue of high fat-fed mice [9]. In rats fed a high fat diet, MDA levels in the liver are significantly elevated relative to control mice [11]. Taken together, these data clearly show that obesity in both humans and laboratory rodents is associated with a pro-oxidant environment and increased oxidative damage.

The direct cause of this increase in oxidative stress with obesity is still not entirely clear. The sources of the pro-oxidant environment associated with obesity are multi-faceted, but alterations in mitochondria function appear to be a significant culprit. Under normal conditions, the oxidation of glucose or fatty acids to generate ATP through mitochondrial respiration results in the production of superoxide as a byproduct of the electron transport chain (ETC) [55]. Glycolysis and fatty acid oxidation produce NADH and FADH₂ which provide the electrons that fuel ETC and ATP production [56]. NADH and FADH₂ donate their electrons to complex I and II of the ETC respectively. These complexes then transfer electrons to ubiquinone which passes its electrons to complex III, to cytochrome C, to complex IV, and finally to molecular oxygen. Along the process, the electrons are used to shuttle protons across the mitochondrial membrane to generate the membrane voltage potential that drives mitochondrial respiration. Superoxide can be produced if electrons from coenzyme Q are received by molecular oxygen as the are shuttled from complexes I and II to complex III [55]. When caloric intake is excessive, as it is within a high fat or high carbohydrate diet, it is possible that the increased intake of glucose and fatty acids will generate more substrates entering into mitochondrial respiration. Consequently the number of electrons donated to the ETC may increase. Once the membrane potential reaches a threshold voltage, excess electrons might begin to back up at complex III and, with subsequent donation to molecular oxygen, generate higher levels of superoxide [55]. Interestingly, treatment of cell cultures with free fatty acids has been shown to increase ROS production indicating the elevation of fatty acids in obesity may provide an additional source of excess ETC substrates through increased fatty acid oxidation [7]. Several data are consistent with the idea that increased caloric intake or obesity are associated with increased mitochondrial superoxide production. High fat feeding in both mice and humans causes a significant increase in H₂O₂ emission from mitochondria isolated from skeletal muscle [57]. H₂O₂ emission was used as a surrogate of superoxide emission in these experiments as mitochondrial superoxide will be rapidly converted to H₂O₂ by Sod2. Increased ROS production has also been shown in mitochondria isolated from kidney [58], liver [59], and adipose tissue [60] from high fat fed or obese animals.

Another significant contributor to increased pro-oxidant environment in obesity is the generation of a pro-inflammatory environment. Inflammation can drive oxidative stress in vivo due to a large increase in free radical production by immune cells as part of the immune response. Obesity has been characterized as promoting chronic inflammation through several different mechanisms. For example, obesity is associated with changes in the T-cell subsets associated with adipose tissue [61,62] and with stimulation of infiltration (and subsequent activation) of macrophages into the growing adipose tissue [63,64,65,66]. Both situations result in the increased production of several different pro-inflammatory cytokines by immune cells, mature adipocytes, and pre-adipocytes [7,63,64,65,67,68]. The increased oxidative stress associated with chronic inflammation and other sources may also promote further production of pro-inflammatory cytokines in a “vicious cycle”. In cultured adipocytes, ROS increases production of the cytokine IL-6 and expression of the pro-inflammatory monocyte chemotactic protein–1 (MCP-1) [69]. In adipose tissue, this can promote macrophage infiltration further contributing to the pro-inflammatory environment (56, 64). ROS can also trigger signal transduction pathways (primarily through nuclear factor κ B; NFκB) that promote production of TNFα (tumor necrosis factor α) and IL-6 [70,71]. Finally, oxidative stress can also drive cells into cellular senescence, particularly adipocyte senescence, in part through cellular oxidation damage [72,73]. Adipocyte senescence also recruits macrophages and increases production of pro-inflammatory cytokines [73].

Other reported sources of the increased oxidative stress in obesity include increased superoxide production by NADPH oxidase and decreased expression/activity of components of the antioxidant defense system. NADPH oxidase is used primarily by neutrophils to generate superoxide as a mechanism to kill invading bacteria and fungi. NADPH oxidase is also activated by advanced glycation end-products (AGE) which are increased under conditions of high glucose [55,74]. With obesity and high fat feeding, there is also a general increase in NADPH oxidase activity, primarily in the adipose tissue [7]. Strikingly, treatment of the NADPH oxidase inhibitor apocynin reduces the levels of lipid peroxidation and H₂O₂ production in the adipose tissue of obese mice without reducing body weight or adiposity. Interestingly, apocynin treatment has no effect on oxidative stress in liver or skeletal muscle, suggesting that adipose tissue may be a primary source of NADPH oxidase-mediated oxidative stress with obesity [7]. Obesity has also been shown to be associated with reduced mRNA expression of Sod1, GPx1 and Cat in the adipose, but not in skeletal muscle or liver [7]. It is not entirely clear why the reduction of these genes was only in adipose tissue, however, one possibility may be the high oxidative stress in this tissue could suppress mRNA transcription in general [7]. These phenotypes of adipose tissue, increased superoxide production and reduced antioxidants, suggest that the general increase in oxidative stress associated with obesity might be due in large part to alterations in fat. However, it must be recognized that obesity can have significant effects on the oxidative environment in other tissues as well, including skeletal muscle [57].

Overall, these data strongly support the notion that obesity and high calorie/high fat/high carbohydrate diets promote oxidative stress and the accumulation of oxidative damage from multiple sources. These data also support one leg of the hypothesis that the mechanism by which obesity causes insulin resistance is through modulation of oxidative stress. The second leg of this hypothesis that must be addressed is the question of whether oxidative stress itself can promote insulin resistance.

Oxidative stress and insulin resistance

As addressed previously, there is a clear difference between ROS required for basic cellular mechanisms like cellular signaling and excessive ROS that contribute to oxidative stress. Free radicals and ROS have been shown to play an important part in the mammalian glucoregulatory system. For example, H₂O₂ production has been shown to regulate glucose-stimulated insulin release from β-cells and to modulate proximal and distal insulin signaling [75,76,77]. Insulin stimulation has been shown to promote H₂O₂ production that then enhances the insulin cascade by inhibiting protein tyrosine phosphatase (PTP) activity, leading to an increase in the basal phosphorylation level of both the IR and IRS proteins [78,79]. Similarly, oxidation of cysteine in phosphatase and tensin homolog (PTEN) has been shown to significantly significantly improve insulin response both in cell culture and in vivo [80]. Additionally, treatment of rats with L-cysteine, which can alter redox balance, can significantly improve hyperglycemia and increase insulin sensitivity [81]. However, these physiological effects are largely thought to be dependent on different forms of ROS or changes in redox state which are distinct from effects of oxidative damage. That is, these signaling pathways are controlled by careful maintenance of ROS within defined constraints modulated through production and antioxidant reduction rather than oxidative stress, i.e., a pathological imbalance towards increased ROS and/or reduced antioxidant defense. For example, while insulin stimulation promotes H₂O₂ production (ostensibly to increase insulin sensitivity), the relative concentration is small, production rapidly ceases when insulin is released from insulin receptor and there is a compensatory elevation in glutatione (GSH) [77,82]. However, increased oxidative stress under these definitions would be predicted to have a negative effect on glucoregulatory mechanisms through stimulation of stress-responsive pathways, by causing cellular dysfunction or by causing apoptosis/death of cells important for regulation of glucose/insulin such as pancreatic β-cells. Thus, this review will provide support for the prediction that excessive oxidative stress is detrimental to insulin signaling.

There are now several lines of evidence from both model organisms and clinical studies that identify a strong correlation between insulin resistance, T2DM and/or metabolic syndrome and oxidative stress. Several studies have shown diabetic patients show an increase in urinary 8-OHdG when compared to non-diabetic individuals. Importantly, even pre-diabetic individuals in these studies showed elevated 8-OHdG suggesting that oxidative damage to DNA is present even before the clinical development of diabetes [15,16,83]. Even among obese patients, the degree of oxidative stress has a strong correlation with metabolic dysfunction as shown by the finding that patients with the greatest degree of insulin resistance tend to have the highest plasma concentration of TBARS [84]. Hyperglycemia itself can cause an increase in oxidative stress as the level of lipid peroxidation in erythrocytes was found to be directly proportional to the glucose concentrations in vitro and to blood glucose concentrations in diabetic patients [85,86]. This increase in lipid peroxidation could be altered experimentally in diabetic rats by control of glycemia with insulin [87].

The pro-oxidant environment associated with diabetes, insulin resistance and metabolic syndrome may be largely due to reduced cellular capability to deal with oxidative stress. Diabetes is associated with reduced levels of components of the antioxidant defense system including GSH, vitamin E and vitamin C [88,89,90]. Diabetes and insulin resistance are also associated with reduction in total antioxidant activity, Sod activity, GPx activity and glutathione reductase activity [84,91,92]. Hyperglycemia may be a significant factor in the downregulation of antioxidant activity as it has been reported that glycation of Sod1 and Trx cause significant inhibition of each enzyme’s activity [93,94].

The most direct evidence that oxidative stress causes insulin resistance comes from a series of studies utilizing cell culture and ex vivo tissue models. Several cell culture models have shown that treatment of insulin-responsive cell lines, such as 3T3L1 and L6 myotubes, with H₂O₂ causes a significant decrease in insulin sensitivity [12,13,14,95,96]. Induction of insulin resistance in cell culture via TNFα or glucocortocoid treatment causes a significant induction of ROS [68]. Insulin resistance in these models can be prevented by treatment with different antioxidant compounds such as N-Acetyl Cysteine (NAC) and manganese (III) tetrakis (4-benzoic acid) porphyrin (MnTBAP) or by overexpression of Sod1, Sod2, or catalase [68]. In addition, treatment with palmitate causes insulin resistance in cell cultures which has been shown to be due in part to stimulation of ROS by increased lipid accumulation [97]. H₂O₂ also has a significantly detrimental effect on insulin signaling in ex vivo rat soleus muscle cultures [98]. In these cultures, H₂O₂ treatment caused a substantial loss of insulin stimulation both proximally at IR phosporylation and distally at Akt phosphorylation [99]. Furthermore, chronic exposure of soleus muscle culture to H₂O₂ caused a selective loss of IRS proteins which further inhibited insulin response [99]. Chronic exposure of soleus cultures to H₂O₂ also caused marked reduction in glucose transport activity both basally and with insulin stimulation [99].

It is still not clear exactly how oxidative stress causes insulin resistance or metabolic dysfunction, though several mechanisms have been proposed. The most common mechanism suggested is that oxidative stress stimulates stress signaling mitogen-activated protein kinases (MAPKs), such as c-jun N terminal kinase (JNK), that in turn cause inhibition of insulin signaling. ROS can activate JNK by oxidation and inactivation of JNK-inhibiting MAPK phosphatases [100] or of JNK-inhibiting glutathione-S-transferase [101] or by oxidizing Trx1 leading to Ask1 dissociation and subsequent Ask-mediated activation of JNK [102]. Activation of JNK has now become well accepted to play an important role in the etiology of T2DM and insulin resistance [67]. JNK family members are encoded by three genetic loci, Jnk1-3, of which the protein product JNK1 is known to be chronically activated in obesity and T2DM, at least in part due to lipotoxic stress [103]. Activation of JNK1 has been shown to directly phosporylate IRS-1 at inhibitory sites that prevent recruitment of this protein to the activated insulin receptor [104,105]. Thus, JNK-mediated IRS-1 phosphorylation disrupts downstream events in the insulin signaling pathway leading to insulin resistance. As might be predicted by the strong inhibitor effect of JNK1 on insulin signaling, mice lacking JNK1 show improved glucose tolerance and insulin sensitivity compared to control mice when fed a high fat diet [106].

Alternatively, oxidative stress could lead to insulin resistance by promoting the expression of several pro-inflammatory cytokines. As discussed above, obesity has been described as a chronic inflammatory disease in which fat accumulation can lead to increased production of several cytokines including TNFα and IL-6. TNFα and IL-6 have both been shown to lead to phosporylation of IRS-1 at inhibitory sites and consequently promote a significant decline in insulin sensitivity [107,108]. Insulin resistance caused by obesity in rodent models has been shown to be reversible by genetic ablation of either TNFα or its receptor, TNFαR [107]. While it has generally been thought that a pro-inflammatory state causes oxidative stress, there is also evidence the reverse also occurs, i.e., oxidative stress can also promote a pro-inflammatory state. For example, ROS trigger the expression of MCP-1, activate NfκB, increase TNFα and IL-6 production, and promote macrophage infiltration and adipocyte senescence [7,64,70,71,73,109].

Lastly, there may be a direct mechanistic link between oxidative stress and the etiology of insulin resistance through the accumulation of oxidative damage to critical macromolecules in insulin-sensitive tissues. Several groups have found an association between increased carbonylation and nitrosylation of proteins in insulin-sensitive tissues and obese or insulin resistant phenotypes [10,110,111,112]. Oxidation of specific proteins in vitro has a clear detrimental effect on their function [29,31] and there is a strong correlation between increasing oxidative stress and diminished protein folding and function in different animal models [28,113]. It might be predicted then that oxidation of particular proteins important for insulin signaling could potentially affect their function in propagation of insulin-stimulated signals. It has already been well established that transient oxidation of PTPs like PTEN can be used as a regulatory stimulus and directly affect insulin signaling [77,79,114,115]. However, evidence that oxidative stress causes oxidative damage that directly diminishes function of insulin signaling is limited. In older humans, it has been suggested that insulin resistance development is due, in part, to reduced function of these proteins [116]. In vitro studies have shown that oxidative stress can impair the ability of insulin receptor to correctly bind with insulin. In endothelial cell lines, this reduced binding of insulin diminishes the ability of insulin receptor to internalize insulin in cells, thereby diminishing insulin sensitivity [117]. Oxidative stress can also alter the ability of insulin signaling proteins to redistribute to correct subcellular locations that are necessary for efficient transduction of signals [118].

From a therapeutic standpoint, there is growing evidence that reduced oxidative stress may be beneficial in the prevention and treatment of insulin resistance, at least in some models. As discussed above, there is clearly a complicated relationship between the generation of oxidative stress (potentially by the obese state) and the effects of oxidative stress on the insulin signaling pathway (see Figure 1). However, simply decreasing oxidative stress by administration of antioxidants during, or prior to, obese conditions can alleviate some of the common diseases associated with obesity, e.g., glucose intolerance and insulin resistance (59). Administration of antioxidants (superoxide dismutase mimetics) can also alleviate some symptoms of experimentally-induced diabetes (60, 111, 112). As mentioned previously, mitochondria-derived ROS may alone be a significant factor in the development of insulin resistance in obesity. Supporting this, insulin resistance in high fat-fed mice can be abrogated by treatment with the mitochondria-targeted superoxide dismutase mimetic, MnTBAP [119] or the mitochondria-targeted small molecule antioxidant, SS31 [57].

Figure 1.

Modulation of oxidative stress is a central process in the development of obesity-induced insulin resistance. Increasing fat accumulation (Obesity) can generate oxidative stress in multiple ways, including increased mitochondria-derived ROS, increased activity of NADPH oxidases, decreased antioxidant expression, and promotion of a pro-inflammatory environment. By activating stress responsive pathways like JNK, increasing cytokine expression, or perhaps through direct oxidative damage of signaling proteins, this oxidative stress can lead to decreased insulin response of the insulin signaling pathway (right). These data suggest that reduction of oxidative stress may be a viable therapeutic or preventative treatment for insulin resistance.

However, other sources of ROS other than superoxide are also likely important in the development of insulin resistance as evidenced by data that show treatment of obese, diabetic mice with the NADPH oxidase inhibitor aponocynin reduces ROS production in adipose tissue and improves glucose metabolism [7]. There are some limitations to these types of studies including issues with the bio-availability of these antioxidant treatments in particular tissues. An alternate approach is the utilization of genetic mutant mice to determine the effect of particular antioxidants on glucose metabolism. By using these mice, it can be tested whether reduction or overexpression of individual antioxidants globally, or in a tissue-specific manner if the mutant exists, can have a significant affect on the development of metabolic diseases.

Antioxidant mutant mouse models and glucose homeostasis

The phenotypes of lines of antioxidant mutant mice described in Table 2 clearly show that expression/activity of a single antioxidant can have significant effects on the regulation of oxidative stress and the accumulation of oxidative damage in mice. However, what role these antioxidants, or particular forms of oxidative damage, have on physiological functions like glucose homeostasis has only begun to be addressed and much still remains to be elucidated about this relationship. If regulation of oxidative stress in general, or regulation of particular reactive oxygen species or oxidative damage moieties specifically, is important in the regulation of glucose homeostasis, it would be predicted that these effects should be evident in antioxidant mutant mice. The antioxidant mouse models in this review have been generated with global alteration in antioxidant expression. As discussed above, these models can address the effect of modulation of oxidative stress (either increased or reduced depending on the model) to a similar degree in all tissues including those important for regulating glucose metabolism, i.e., skeletal muscle, adipose tissue, liver and pancreas.

As discussed above, oxidative stress is largely thought to negatively affect glucose homeostasis in mammals (through stimulating stress-responsive pathways, causing cellular dysfunction or causing apoptosis/cell death). The prediction based on this assumption, would be that increased oxidative stress (such as that occurs in mice with reduced antioxidant expression) should result in an significant disability in the maintenance of glucose homeostasis. That is, these antioxidant knockout and heterozygote mice should display increased blood glucose, decreased blood insulin, glucose intolerance, insulin resistance, etc. This prediction seems relatively straightforward, however, the accumulating evidence collected from antioxidant mutant mice are equivocal in support of this idea at least when young mice are maintained under standard mouse husbandry conditions. For example, mice lacking (or having reduced expression of) different antioxidants show phenotypes suggesting that these mutants have reduced, improved, or no change in glucose homeostasis (Table 3). Because the lack of Sod1 has such a profound effect on oxidative stress in mice [45,46], severe defects in glucose homeostasis might be predicted in these mice. Wang et al found that Sod1^−/− mice are smaller in size (lower body weight), are hyperglycaemic and their blood insulin levels are low owing to some dysfunction in pancreatic β-cells [120]. However, these mice also show an improvement in glucose tolerance and insulin sensitivity which would not be predicted if increased oxidative stress was a causative factor in peripheral insulin resistance [120]. The improved insulin sensitivity may be a compensatory mechanism to the hypoinsulinemia as Sod1^−/− mice have significantly increased expression of IR in skeletal muscle [120]. These mice may therefore represent a phenotype more similar to Type 1 diabetes mellitus (generally described as pancreatic dysfunction causing hypoinsulinemia) rather than T2DM. In comparison, Hoehn et al. found that mice with a 50% reduction in Sod2 (Sod2^+/−) do not show any change in body weight or fasting glucose levels, but show a significant reduction in glucose tolerance relative to control mice [119]. However, these mice were relatively older when tested (18 months of age) and it may be these mice show no difference at a young age and aging promotes glucose intolerance to a greater degree in Sod2^+/− mice. Two groups have examined glucose metabolism in mice lacking GPx1 and have generated conflicting data. Both groups found Gpx1^−/− mice to be similar in size to control mice and hypoinsulinemic with the latter phenotype seemingly due to some dysfunction in the β-cells of the pancreas [80,120]. However, Wang et al. found that Gpx1^−/− mice had reduced fasting blood glucose and improved glucose tolerance [120], whereas Loh et al. found no difference in blood glucose and no change in insulin sensitivity [80]. The reasons for the discrepancies between the studies are not clear, but several possibilities are discussed below. Lastly, we have found that the lack of MsrA in mice has no effect on body weight through at least 22 months of age [121] and also has no effect on fasting blood glucose levels (unpublished data A.R. and A.B.S.).

Table 3.

Glucose homeostasis in antioxidant mutant mouse models

Model	Body weight	Adiposity	Blood glucose or insulin levels	Glucose tolerance or insulin sensitivity	Refs
Knockouts
Sod1−/−	↓	ND	Glucose ↑, Insulin ↓	Improved	[120]
Sod2+/−	–	ND	Glucose –, Insulin ND	Reduced*	[119]
Gpx1−/−	–	ND	Glucose –/↓, Insulin ↓	Improved/–	[80,120]
MsrA−/−	–	ND	Glucose –, Insulin ND	ND	[121]
Transgenics
Sod2 TG	–	–	Glucose –, Insulin –	–	[119]
MCAT (mito)	–	–	Glucose –, Insulin –	–**	[122]
GPx1 TG	↑	↑	Glucose ↑, Insulin ↑	Reduced	[123]
Prdx3 TG	–	ND	Glucose ↓, Insulin –	Improved	[125]
Prdx4 TG	ND	ND	Glucose –, Insulin ND	–	[124]

↓ = values compared to control mice are lower

– = values compared to control mice are unchanged

↑ = values compared to control mice are higher

ND = not determined

^*values from older mice

^**MCAT mice do show improved insulin sensitivity when older mice are tested

Contrary to the effect of increased oxidative stress, it would be predicted that mice with reduced oxidative stress (i.e., mice with overexpression of antioxidants) should show improved glucose homeostasis compared to control mice. As a whole, however, data from these mice suggest that overexpression of most antioxidants has minimal effects on glucose metabolism when mice are fed a normal diet. For example, Hoehn et al. found that overexpression of Sod2 has no significant effect on body weight, fat content, fasting glucose or insulin [119]. In addition, Sod2 TG mice are not different from control mice in glucose tolerance or insulin sensitivity [119]. Lee et al. found that mice with mitochondrial targeted catalase overexpression (MCAT) also show no differences in body weight, adiposity, or glucose homeostasis at least during young age [122]. However, there is evidence that MCAT mice are protected from declines in glucose regulation at old ages suggesting a protective role of catalase, or mitochondria, in age-associated dysregulation of glucose homeostasis [122]. Contrary to prediction, McClung et al. show that overexpression of GPx1 seems to have detrimental effects on the regulation of glucose homeostasis in mice. GPx1 TG mice show increased body weight and fat content, hyperglycemia and hyperinsulinemia and reduced insulin sensitivity [123]. Overexpression of peroxiredoxins either has no effect on blood glucose levels or insulin sensitivity (Prdx4 TG) or improves fasted blood glucose levels and improves glucose tolerance (Prdx3 TG) [124,125]. Because Prdx3 is localized to the mitochondria, these data might suggest the importance of peroxide detoxification in this organelle in regulating glucose sensitivity. Mice with overexpression of Trx1 have not yet been tested in terms of glucose homeostasis, however, Hui et al. found that mice lacking a functional Thioredoxin interacting protein (TrxIP) show concomitant increase in Trx1 activity that is associated with increased plasma insulin levels but improved glucose tolerance [126].

If oxidative stress is a significant contributor to the regulation of glucose metabolism, how then could this lack of a consistent effect in antioxidant mutant mice be interpreted? One interpretation may be that because the antioxidant defense is complex, the alteration of a single antioxidant might be expected to have little effect. For example, mice lacking both Sod1 and GPx1 show further exacerbation of hyperglycemia and hypoinsulinemia beyond that of either knockout alone [120]. On the other hand, the alteration of a single antioxidant does have significant effects on the levels of oxidative stress/damage in these mice (Table 2) so it may not be overall levels of stress/damage, but rather specific types of ROS or oxidative damage that are important in this process. It must also be mentioned that altering antioxidant levels in mice could potentially have a significant effect on regulation of ROS-modulated insulin signaling pathways. Certainly, more studies in mutant mice lacking different combinations of antioxidants (including those lacking >2 antioxidants) would need to be performed to completely address this.

A second interpretation may be that levels of oxidative stress may be relatively low under normal husbandry conditions and, thus, modulation of antioxidants has little effect overall. It may be that under the basal conditions of mice in the laboratory do not generate enough oxidative stress/damage to have a physiological effect on glucose homeostasis that is measurable. That is, under low levels of oxidative stress (normal, lean conditions) modulation of antioxidants may have less affect than when compared to situations with high levels of oxidative stress (obesity) and thus a greater necessity for the activity of antioxidants. In some ways, this might relate to the prevalence of metabolic dysfunction in human populations as well because insulin resistance, T2DM and other metabolic diseases are rare in patients who are lean and in relatively good health. However the prevalence of these metabolic diseases increases significantly when patients are obese or suffering from some other pathological conditions. It is feasible to predict that the phenotypes of antioxidant mutant mice may be different when maintained under different husbandry conditions (i.e., optimal conditions vs. chronic oxidative or metabolic stress). Clearly, modulation of antioxidant enzymes in mice has a strong effect when these mice are challenged with compounds that generate high levels of oxidative stress such as paraquat or diquat (Table 2). The effects of more mild oxidative challenges on these mice has yet to be fully elucidated; such challenges might include induction of inflammatory responses by treatment with agents like lipopolysaccharides or treatment with pathogens like influenza virus or Streptococcus pneumoniae. More directly related to topic of this review, differences in diet that lead to increased obesity or inflammation may also lead to a general oxidative stress condition under which modulation of antioxidant defense could have a clear and marked effect on glucose homeostasis in mice. By testing antioxidant mutant mice under multiple environmental conditions, such as maintenance on diets differing it fat content, it may be possible to more completely understand the role of oxidative stress on glucose homeostasis in broad terms.

Antioxidant mouse mutants in diabetes and obesity

If it is assumed that glucose regulation in young, healthy mice (and other rodents) is at its optimum, then results from the study of antioxidant mutant mice under such conditions may be limited. However, when under metabolic stress, such as obesity or consumption of high caloric diets, regulation of oxidative stress could be of great importance to the maintenance of glucose regulation and development of T2DM. Laboratory mice, particularly the C57Bl/6 strain in which the majority of studies are performed, do not typically develop T2DM or become obese without alterations to genes (such as ob/ob, db/db or NOD mice) or diet (such as high fat or western diets) [127]. In particular, diets high in fat content are a particularly useful experimental paradigm to test questions pertaining to obesity, insulin resistance, and T2DM. Because these diets can be fed to the same mice that have been previously fed normal diets, information about the development of adipose and glucose regulatory diseases can be directly assessed.

Under these obesity paradigms, it is relatively clear that antioxidants play a significant role in the regulation of glucose homeostasis changes associated with obesity. For example, Hoehn et al. showed that while Sod2 TG mice show no difference in glucose homeostasis on a normal diet, these mice are protected from glucose intolerance and insulin resistance that occurs in control mice fed a high fat (45%) diet [119]. Sod2 TG mice gained a similar amount of weight and consumed the same amount of food as control mice on this diet; however, Sod2 TG performed better in both glucose tolerance and insulin tolerance tests after high fat feeding and showed reduced blood insulin levels suggesting a protection from the development of insulin resistance. This protective phenotype of Sod2 overexpression could be recapitulated by treatment of high fat-fed control mice with a mitochondria-targeted superoxide dismutase mimetic, MnTBAP [119]. MnTBAP treatment has also been shown to improve glucose homeostasis in genetically obese mice [68]. In a set of elegant cell culture experiments, Hoehn et al. also clearly showed that mitochondrial superoxide production was a significant cause of insulin resistance. For example, induction of insulin resistance in 3T3L1 cell lines caused by treatment with insulin, TNFα, or dexamethasone was associated with increased superoxide from the mitochondria [119]. More directly, treatment with of L6 cell lines with antimycin A (an inhibitor of ETC Complex III) caused insulin resistance in these cells [119]. Treatment with mitochondria-targeted superoxide dismutase mimetics was sufficient to reverse the insulin resistant phenotype of these cells. Together, these data cleasrly implicate mitochondria-derived oxidative stress as a significant cause of insulin resistance assocaited with obesity and also suggest treatments designed to diminish mitochondrial superoxide production might can significantly inhibit the pathogenesis of T2DM-like metabolic dysfunction in mice.

Reduction in H₂O₂ in the mitochondria by either mitochondria-targeted overexpression of catalase or overexpression of peroxiredoxin 3 also appears to have a protective effect on glucose homeostasis in obese mice. Anderson et al. show that, when fed a 60% fat content diet, MCAT mice showed improved glucose tolerance and insulin sensitivity compared to high fat-fed control mice and reduced fasting blood glucose and insulin [57]. In addition, high fat-fed MCAT mice had improved insulin signaling in muscle relative to control mice. This high fat diet regiment caused a significant increase in H₂O₂ release from mitochondria isolated from skeletal muscle. H₂O₂ emission was drastically reduced in mitochondria isolated from high fat-fed MCAT mice providing a plausible mechanism for the protection of glucose homeostasis in this model. Treatment of high fat fed rats with a small molecule antioxidant, SS31, had a similar effect on mitochondria function, redox environment and glucose homeostasis [57]. Similarly, Chen et al. found that Prdx3 TG show a reduction in H₂O₂ emission from mitochondria isolated from skeletal muscle and reduced accumulation of lipid peroxidation in the form of F₂-isoprostanes and 4-HNE [125]. When fed a 60% fat diet, Prdx3 TG mice had reduced fasting and fed blood glucose and insulin relative to high fat fed control mice. Prdx3 TG mice were also more glucose tolerant than control mice indicating a preservation of glucose homeostasis in Prdx3 TG mice under metabolic challenge [125]. Together, the data from Sod2 TG, MCAT and Prdx3 TG mice show clearly that reduction of oxidative stress associated with high fat feeding can diminish obesity’s detrimental effects on glucose homeostasis. Perhaps more importantly, the result from high fat-fed Sod2 TG, MCAT, and Prdx3 TG mice highlight the potential importance of mitochondria in controlling both oxidative stress and glucose regulation under these conditions.

As briefly discussed earlier, antioxidants may have beneficial effects on insulin signaling through modulation of ROS-dependent signaling pathways rather than directly through modulating oxidative damage. Loh et al. found that Gpx1^−/− mice fed a high fat diet (36% fat) show a 3–4 fold increase in H₂O₂ levels in the skeletal muscle and a significant shift to an oxidized redox state in this tissue [80]. Despite this increase in oxidative stress, and contrary to what would be predicted, Gpx1^−/− mice maintained greater insulin sensitivity than their high fat fed controls both in measures of insulin tolerance tests and in insulin-induced phosporylation of Akt in muscle. This resistance to high-fat induced insulin resistance can partly be ascribed to an increased energy expenditure and reduced adipogenesis in Gpx1^−/− mice. However, the authors also found that treatment of these mice with a general antioxidant (N-acetylcysteine) caused a significant reduction of H₂O₂ levels and subsequent loss of insulin sensitivity. Therefore, regulation of insulin sensitivity in these animals seems to be modulated by H₂O₂ signaling, at least in part. In this context, it is interesting to note that GPx1 TG animals were found to be moderately obese and insulin resistant even when maintained on normal rodent chow [123]. Together, these data suggest that perhaps ROS-mediated signaling regulation of glucose metabolism may be regulated exclusively by glutathione peroxidases.

Concluding Remarks

Overall, there is substantial evidence pointing to oxidative stress playing a significant role in the progression of glucose metabolism dysfunction associated with obesity. Clearly, future studies examining more of antioxidant mutant mouse models in terms of obesity, glucose homeostasis and insulin resistance will lead to a greater understanding of how oxidative stress and ROS-mediated signaling work together to regulate metabolism. These models can also be further utilized to delineate the processes important for diabetes-associated pathologies. Currently, our understanding is limited, but we do know that antioxidants can significantly alter the pathogenesis of these debilitations. Retinal damage associated with diabetes, for example, has shown to be increased in Sod1^−/− mice [128] as has the rate of diabetes-induced cataract formation [129], whereas Sod1 TG mice show reduced diabetes-induced retinopathy [128,130]. Diabetes-induced nephropathy has been shown to be exacerbated in Sod1^−/− mice [131], whereas, mice overexpressing catalase in the kidney show reduced renal disease associated with diabetes [132]. Lastly, mice overexpressing Sod2 in the heart show reduced diabetes-induced left ventricular hypertrophy [133]. These studies support the hypothesis that oxidative stress promotes particular pathologies associated with diabetes and suggest further investigation utilizing antioxidant mutants could be extremely beneficial in the search for treatments for these pathologies associated with metabolic dysfunction.

While the data discussed in this review have primarily focused on oxidative stress in disease states, oxidative stress might regulate glucose metabolism more generally even with normal, healthy life. Metabolic dysfunctions like insulin resistance and diabetes are age-associated pathologies in people; that is, these conditions are much more prevalent in aged human populations than they are amongst young individuals [134,135]. Part of this can be explained by an overall increase in morbidity with age, including an elevated pro-inflammatory state, or with increasing adiposity with age. However, even after correcting for these pathological conditions there exists an age-associated decline in glucose homeostasis. Most accumulating data are consistent with the idea that oxidative stress may be a major regulator of the development of insulin resistance, diabetes, and complications from diabetes. The aging process itself has been proposed to be regulated by oxidative stress. The Oxidative Stress Theory of Aging suggests that progressive accumulation of oxidative damage to cellular macromolecules occurs over time and leads to physiological declines associated with aging [136,137]. The age-associated decline in glucose metabolism might result as a part of this accumulation of oxidative damage with age. A potential experimental paradigm to test this idea is by measurement of glucose regulation with age in the antioxidant mutant mouse models discussed herein. Some preliminary data does suggest that age-associated alterations in glucose metabolism may be altered in these models. For example, moderately old Sod2^+/− mice do show a decline in glucose tolerance relative to their littermate controls, even when maintained on standard rodent chow [119]. Sod2^+/− mice have previously been shown to have increased accumulation of oxidative damage with age relative to control mice and reduced mitochondrial function [52,138]. Conversely, MCAT mice are protected from age-associated declines in oxygen consumption, CO₂ production, energy expenditure, and insulin sensitivity [122]. In part, the maintenance of youthful metabolic parameters in MCAT mice seems to be due to an alleviation of mitochondrial H₂O₂ production and oxidative damage, both of which significantly increase with age. Taken together, these data suggest that age-associated declines in mitochondrial function and accumulation of oxidative damage can have a significant detrimental effect on glucose homeostasis.

In sum, life under constant metabolic challenge, whether by disease, diet or simply the aging process itself, can have severely detrimental effects on glucoregulatory processes. Oxidative stress may be a particularly important determinant of regulating these processes and, as such, might hold the key to treatment or prevention. To date, most clinical trial studies have suggested that dietary supplementation with antioxidants has little beneficial effect on prevention or treatment of diabetes in human populations [139]. However, the data from in vitro and rodent experiments suggest that intervention to reduce oxidative stress might hold some therapeutic potential for age-associated metabolic dysfunction. Why then has there been a disconnect between basic research and clinical studies? It may be that we have yet to address the importance of the timing of oxidative stress, or particular forms of ROS or damage, or particular environments (i.e., high fat diet or obesity) or tissue-specific effects of oxidative stress in relation to the whole of glucose homeostasis. In particular, adipose tissue specifically might be an important target of antioxidant therapy. Accumulation of visceral fat particularly contributes significantly to the pro-inflammatory, pro-oxidative stress associated with obesity. It will be interesting in the future to test whether adipose tissue-targeted overexpression of antioxidant, or treatment with antioxidants targeted to adipose tissue alone could have beneficial effects on obesity-induced insulin resistance. A better understanding of the systems biology of oxidative stress in terms of glucose regulation may be a critical step in addressing these questions. In the future, a more broad based understanding of how particular antioxidants respond under different environmental challenges could potentially greatly increase our understanding of treatment options for healthy life.

Abbreviations

4-HNE

4-hydroxynonenal

8-OHdG

8-hydroxyguanosine

AGE

advanced glucation end-product

Akt/PKB

protein kinase B

Cat

catalase

CuZnSod;Sod1

Copper/Zinc superoxide dismutase

ECSod;Sod3

extracellular superoxide dismutase

ETC

electron transport chain

GLUT4

glucose transporter

GPx

glutathione peroxidase

IL-6

interleukin-6

IR

insulin receptor

IRS-1

insulin receptor substrate 1

JNK

c-jun N terminal kinase

MAPK

mitogen-activated protein kinases

mCAT

mitochondria-targeted catalase overexpression

MCP-1

monocyte chemotactic protein 1

MDA

malondialdehyde

MnSod;Sod2

Manganese superoxide dismutase

Msr

methionine sulfoxide reductase

NADPH

nicotinamide adenine dinucleotide phosphate

NFκB

nuclear factor κ B

PI-3K

phosphatidylinositol 3-kinase

Prdx

peroxiredoxin

PTEN

phosphatase and tensin homolog

PTP

protein-tyrosine phosphatase

ROS

reactive oxygen species

T2DM

type 2 diabetes mellitus

TBAR

thiobarbituric acid reactive substances

Trx

thioredoxin

TG

transgenic

TNFα

tumor necrosis factor α