Targeting the NO/superoxide ratio in adipose tissue: relevance to obesity and diabetes management

Insulin sensitivity and metabolic homeostasis depend on the capacity of adipose tissue to take up and utilize excess glucose and fatty acids. The key aspects that determine the fuel‐buffering capacity of adipose tissue depend on the physiological levels of the small redox molecule, nitric oxide (NO). In addition to impairment of NO synthesis, excessive formation of the superoxide anion (О2•–) in adipose tissue may be an important interfering factor diverting the signalling of NO and other reactive oxygen and nitrogen species in obesity, resulting in metabolic dysfunction of adipose tissue over time. Besides its role in relief from superoxide burst, enhanced NO signalling may be responsible for the therapeutic benefits of different superoxide dismutase mimetics, in obesity and experimental diabetes models. This review summarizes the role of NO in adipose tissue and highlights the effects of NO/О2•– ratio ‘teetering’ as a promising pharmacological target in the metabolic syndrome.


Introduction
The majority of type 2 diabetes cases (up to 90%) are related to insulin resistance and obesity (Anderson et al., 2003), suggesting a causal connection between these conditions (Berger, 1992;Daly, 1994). According to the dynamic nature of energy balance maintenance (and long-term mechanisms of adiposity regulation), increase in energy uptake is paralleled by energy expenditure by peripheral tissues, including adipose tissue, and obesity develops when fuel intake chronically exceeds expenditure (Hill and Peters, 1998;Kashyap and Defronzo, 2007). The failure of adipose tissue expansion (not obesity per se) is the essential factor linking positive energy balance, diabetes and other cardiometabolic diseases (Lafontan, 2013). Thus, when the inflows of nutrients into adipose tissues exceed the capability of adipocytes to handle nutrient excess and the capacity of adipose tissue to expand by hyperplasia, more adipocytes become hypertrophic and defend themselves by developing insulin insensitivity (Gray and Vidal-Puig, 2007). Improper lipolytic and lipogenic response of adipose tissue during fasting/feeding conditions in either case results in overflow and ectopic deposition of lipids in non-adipose organs (Frayn, 2002). According to the lipid overflow hypothesis of systemic insulin resistance development (Unger, 2003), fat-specific insulin resistance is the earliest risk factor for systemic insulin resistance and type 2 diabetes pathogenesis. Thus, clarification of the molecular mechanisms underlying the lipid buffering capacity of adipose tissue, that is, oxidative capacity and insulin sensitivity, may aid in the improvement of therapeutic options for obesity and diabetes prevention.
Nitric oxide (NO) has emerged as a central regulator of energy metabolism and body composition that acts mainly by modulating the oxidative capacity and insulin sensitivity of adipose tissue (see Jobgen et al., 2006;Dai et al., 2013;Bernlohr, 2014;Sansbury and Hill, 2014). Supraphysiological production of NO by inducible NO synthase (iNOS) in adipocytes (Merial et al., 2000) and macrophages (Weisberg et al., 2003;Lumeng et al., 2007) and lack of endothelial NOS (eNOS)-mediated NO synthesis (Valerio et al., 2006) in adipose tissue are important risk factors leading to obesity. Moreover, uncontrolled flux of superoxide anion (О 2 •-) from the electron transport chain (ETC) in mitochondria and NADPH oxidase (NOX) under persistent nutrient inflow into adipocytes may divert the role of NO in hypertrophic adipocytes, leading to local and ultimately systemic insulin resistance and type 2 diabetes (see Maritim et al., 2003;Bashan et al., 2009;Afanas'ev, 2010). The current review discusses the underlying mechanisms and consequences of impairment of physiological signalling of reactive oxygen and nitrogen species (ROS/RNS) in adipose tissue, their effects on lipid buffer function, and implications in the development of obesity and obesity-related type 2 diabetes.
Adipose organ and roles in metabolic homeostasis: healthy and unhealthy expansion The function of adipose tissue in buffering daily lipid flux is traditionally understood as hydrolysis of triacylglycerol (TAG) and (re)esterification (Frayn, 2002). Compared with the other insulin-sensitive tissues, such as liver and muscle, adipose tissue is less important for postprandial glucose clearance, and the energy for cellular functions in adipocytes is primarily obtained via glycolytic ATP production. However, it appears that glucose uptake and metabolism as well as the oxidative capacity of adipocytes underlie metabolic flexibility and healthy expansion of adipose tissue during overnutrition and influence its role in protecting non-adipose tissues against lipotoxicity. The metabolic pathways facilitating adipocyte energy storage and release during fasting/fed transitions have been recently reviewed in detail by Rutkowski et al. (2015) and briefly presented in Figure 1. Fat cells (white adipocytes) compose the largest adipocyte population of human adipose tissue ( Figure 2). Adipocytes may additionally be brown or brite/beige according to their origin (endothelial or myogenic) and ultrastructure. The Insulin receptor eNOS Nuclear hormone receptors appearance and distinct origins of adipocytes reflect their specialization in energy partitioning. All adipocytes store excess calories as TAG, while brown and brite/beige cells are highly specialized for calorie combustion, specifically metabo-regulatory and/or thermo-regulatory thermogenesis. The metabolic inefficiency of these adipocytes is attributed to high mitochondrial content and the uncoupling protein 1 (UCP1), a protonophore that uncouples oxidative phosphorylation (OXPHOS) from respiration, generating heat instead of ATP (Cannon and Nedergaard, 2004). Generally, the number of brown and beige adipocytes found in white areas varies with age, strain and environmental conditions (Cinti, 2000). The number of brown adipocytes in humans is inversely correlated with body mass index and body fat mass (van Marken Lichtenbelt et al., 2009) as well as fasting glycaemia (Cypess et al., 2009) and positively correlated with resting metabolic rate (van Marken Lichtenbelt et al., 2009). The real significance of brown adipose tissue in overall human metabolism and energy expenditure is yet to be established (Kozak et al., 2010;Schlögl et al., 2013;Halpern et al., 2014), but there are indications that this tissue mediates cold as well as diet-induced thermogenesis in humans (as in mice and rats) (Vosselman et al., 2013a, b), and its lack or unresponsiveness under conditions of hypercaloric diet and overfeeding may explain a propensity towards easy weight gain (Stock, 1999;Vosselman et al., 2013b). Oxidative capacity and uncoupling also have significant implications for white, unilocular adipocytes, especially in obesity, in terms of regulation of metabolic function, and glucose and fatty acid partitioning during both fasting and feed

Figure 1
Adipocytes export fatty acids during times of energy deficit (involved pathways are marked with green arrows). The rise in cAMP, a sign of increased glucagon or adrenergic stimulation, and low insulin stimulate hydrolysis of triglycerides into glycerol and fatty acids. The increased cAMP pool activates PKA that, in turn, phosphorylates hormone-sensitive lipase (HSL) and perilipins to increase lipolysis. Glycerol and fatty acids are mostly exported into the circulation for systemic utilization. A proportion of fatty acids is re-esterified within the adipocytes, while another part may, after activation to form acetyl-CoA, enter the mitochondria for β-oxidation through carnitine palmitoyl transferase-1 (CPT-1). This rate-limiting enzyme is inhibited by malonyl-CoA, an intermediate of de novo lipogenesis regulated by acetyl carboxylase (ACC). ACC prevents the oxidation of fatty acids when adipocytes are in a lipogenic state. Inhibition of ACC by AMPK relieves this inhibition for β-oxidation. Under positive energy balance, insulin regulates glucose and fatty acid uptake in adipose tissue, and expression and activity of enzymes involved in their metabolism and deposition into TAG, that is, lipogenesis (marked by red arrows). In short, insulin through binding to its cell surface receptor stimulates tyrosine kinase activity, which phosphorylates key residues on several 'docking proteins', IRS proteins. Assembly of a stable complex leads to the regulation (in most cases, activation) of downstream signalling pathways. Recruited proteins include the p85 regulatory subunit of PI3-kinase, which stimulates signalling pathways ultimately leading to PI3-kinase-dependent serine/threonine PKAkt/PKB activation. Phosphorylation of Akt1 at two regulatory residues, Ser 473 and Thr 308 , is critical for complete Akt/PKB activation in adipose tissue. Akt stimulates the translocation of the glucose transporter, GLUT4, to the plasma membrane, thereby promoting uptake of glucose into the cell. A high level of circulating insulin also stimulates PDE3B, promoting cAMP hydrolysis, lowering PKA activity and PKA-dependent HSL phosphorylation, activation and lipolysis. Chronic insulin signalling enhances cAMP production through β-adrenoceptor activation in adipocytes but also disrupts the signalling pathway between β-adrenoceptors and PKA. Imported as well as de novo synthesized fatty acids from excess glucose combine with CoA, and after successive esterification, form TAG. The (re)-esterification process requires production of glycerol-3-phosphate as a substrate for fatty acid re-esterification into TAG. Glycerol-3-phosphate is mostly derived from glucose in the fed state (glycolytic intermediates). Because the glucose supply to the tissue is limited in the fasting state (lipolytic stimulation) and adipocytes have no significant glycerol kinase activity, glycerol-3-phosphate is acquired from lactate or pyruvate.

BJP
A Jankovic et al.
periods. During the fasting response, stimulation of β-oxidation in adipocytes may restrict high fatty acid export into the circulation and prevent or delay obesity development (Horowitz, 2001). In the fed state, adipocytes shift to glycolytic ATP production. Consequently, the levels of all metabolic products of glucose (CO 2 and pyruvate/lactate) increase up to 10-fold, along with adipocyte size (DiGirolamo et al., 1992). Intensified flux through the glycolysis and pentose phosphate pathway provides energy for adipocyte activity and directs excess metabolic substrates into TAG synthesis (lipogenesis) (DiGirolamo et al., 1992). Earlier, Rossmeisl et al. (2000) demonstrated that total uncoupling of OXPHOS in 3T3-L1 adipocytes induced by 2,4-dinitrophenol or ectopic UCP1 in white fat of transgenic aP2-UCP1 mice restricts in situ lipogenesis. Subsequent studies revealed that mitochondrial OXPHOS supports high energy-consuming processes, such as fatty acid storage, adipokine synthesis/secretion (Koh et al., 2007), insulin signalling and glucose uptake (Shi et al., 2008) and adipogenesis (Ryu et al., 2013), in differentiating 3T3-L1 adipocytes. Moreover, increased glucose and fatty acid flux/cycling through adipocytes occurring after treatment of adipocytes with insulin, corticosteroids, proinflammatory cytokines, lipid (Hoehn et al., 2009) or lactate (Carriere et al., 2014) or exposure of rats to the cold (Jankovic et al., 2015a), promote mitochondrial biogenesis and uncoupling. Besides increased energy dissipation and thermogenesis, these processes could be considered an adaptive stress response of adipocytes to the increased inflow of reducing equivalents to the mitochondrial ETC (Jeanson et al., 2015;Jankovic et al., 2015b). Thus, proper capacity for oxidation of glucose and fatty acids, OXPHOS, uncoupling and biogenesis of new mitochondria are indispensable in the daily metabolic regulation of adipocytes and long-term adipose tissue function and healthy expansion (Wilson-Fritch et al., 2004;De Pauw et al., 2009;Hao et al., 2010;Lu et al., 2010). In diabetes, the capacity of adipose tissue in energetic remodeling is impaired (Choo et al., 2006;Keller and Attie, 2010). In addition to metabolic and cellular plasticity, healthy adipose tissue expansion requires normal blood flow and vascularity (Frayn et al., 2003) as well as hyperplastic potential, because these factors account for greater lipid-buffering capacity (Rutkowski et al., 2015). Conversely, pathological (unhealthy) expansion of adipose tissue is characterized by changes in blood flow and the presence of enlarged, dysfunctional, that is, insulin-resistant, adipocytes (see Lafontan,

Figure 2
Adipose tissue is loose connective tissue composed of adipocytes and stromal-vascular cells, anatomically organized in distinct adipose tissue depots. In terms of energy balance, adipose tissue depots may appear as predominantly whiteenergy saving, brownenergy dissipating, or beige (brite or convertible), depending on the relative amount of white, beige or brown adipocytes. All adipose tissue depots in the body are functionally integrated in a highly dynamic multidepot adipose organ. The main brown (deep neck), beige (paravertebral) and white (gluteo-femoral) adipose tissue depots of the adult human adipose organ are presented. 2013). Adipose tissue-specific insulin resistance (impaired glucose and fatty acid uptake and utilization) appears to be an early and irreversible defect that explains the causal relationship between adipocyte dysfunction and systemic insulin resistance (Iozzo, 2009;Lafontan, 2013). Hitherto, studies on animal models (mutant mice, diet-induced obesity) and cultures of mouse and human adipocytes have provided strong support for the involvement of hypoxia, inflammatory signalling, endoplasmic reticulum stress and unfolded protein response, autophagy, dysfunction of mitochondria/impaired mitochondrial biogenesis, and oxidative stress in the development of fat-specific insulin resistance (see Wood et al., 2009;Blüher, 2009;Netzer et al., 2015).
Temporal progression of oxidative damage in the pathogenesis of obesity and associated metabolic disorders is poorly understood, because the classical 'markers' of oxidative damage (oxidation products of lipids, DNA and proteins), in contrast to levels in plasma, urine and various non-adipose tissues, are only minimally increased in mouse (Furukawa et al., 2004;Garcia-Diaz et al., 2007;Grimsrud et al., 2007) and human (Frohnert et al., 2011;Jankovic et al., 2014) adipose tissues in obesity. The data suggest that the increase in superoxide and ROS/RNS levels in the mitochondria of expanding adipose tissue precedes adipocyte dysfunction and progression of obesity-related metabolic disorders in metabolically healthy control mice (Houstis et al., 2006;Matsuzawa-Nagata et al., 2008) and individuals (Jankovic et al., 2014). Notably, increased О 2 •is a critical interfering factor in signalling of NO and other ROS/RNS essential for the lipid buffering function of adipose tissue, thus differentiating unhealthy from healthy expansion, that is, insulin-resistant from insulin-sensitive obesity (Sansbury and Hill, 2014).

Biological effects of NO in adipose tissue
NO, a gaseous signalling molecule similar to CO 2 , CO, H 2 S and O 2 , is toxic at high levels but essential in the regulation of biological processes when endogenously produced in nM concentrations. The multifaceted role of NO in adipose tissue corresponds to the extremely complex mechanisms of biological effects that depend on its (i) site and level of production, determined by enzymes involved in NO synthesis and availability of their substrates and cofactors, and (ii) interactions with different cellular biotargets, primarily ferrous iron (haem and non-haem), thiyl radicals, molecular oxygen and superoxide (Beckman and Koppenol, 1996;Wink and Mitchell, 1998).

Endogenous production of NO in adipose tissue
In virtually all cell types, NO is synthesized via NOS-catalysed oxidation of L-arginine (Moncada et al., 1989). Basal eNOS (NOS3) and iNOS (NOS2) expressions have been reported in rat and human adipose tissue and adipocytes (Ribiere et al., 1996;Elizalde et al., 2000;Gaudiot et al., 2000). Neuronal NOS (nNOS, NOS1) protein does not appear to be present in significant amounts (Engeli et al., 2004), although some studies have provided evidence for protein expression of nNOS in the cytoplasm (Fu et al., 2005;Jobgen et al., 2006) and mitochondria (Finocchietto et al., 2011) of adipocytes. eNOS is mostly membrane bound, while iNOS is localized in the cytoplasm of adipocytes and macrophages (Jobgen et al., 2006). However, following post-translational modification and protein/protein interactions, NOS proteins translocate into different cellular compartments, leading to localized biological effects of NO and regulation of different signalling pathways (Giordano et al., 2002;Villanueva and Giulivi, 2010). Under most physiological conditions, eNOS and nNOS synthesize low levels of NO (nM), whereas iNOS expression is up-regulated by LPS, TNF-α and interferon-γ, leading to the generation of high levels of NO (μM) lasting for several hours or days (Stamler and Meissner, 2001;Pilon et al., 2004). The reductive pathways, that is, synthesis of NO from nitrite (NO 2 À ) catalysed by several transition metal-containing proteins, such as xanthine oxidoreductase (XO) (Zhang et al., 1998), deoxymyoglobin/deoxyhaemoglobin (Nagababu et al., 2003), cytochrome c (Basu et al., 2008), complex III (Nohl et al., 2000), cytochrome c oxidase (CcOx) (Castello et al., 2006) or NOS (Vanin et al., 2007), represent a relevant NO source in hypoxic and acidic intracellular micro-environments (Shiva, 2013;Sparacino-Watkins et al., 2014) after exercise (Cosby et al., 2003) or ischaemia/reperfusion injury and myocardial ischaemic conditioning (Shiva et al., 2007; also see Andreadou et al., 2015). A recent study by Roberts et al. (2015 showed that exposure of rats in vivo and primary adipocytes ex vivo to hypoxia augments nitrate-mediated NO production with subsequent up-regulation of brown adipocyte-associated genes. These authors suggested that augmentation of the nitrate-stimulated browning response during hypoxia represents a physiological adaptation of adipocytes undergoing hypertrophy in obesity. This NO-producing pathway may be exploited therapeutically to maintain oxidative capacity of adipocytes to metabolize fatty acids and to counteract the obesity-related pathological metabolic state of adipose tissue (Roberts, 2015).
Interaction of NO with soluble guanylate cyclase (sGC) and cGMP-mediated signalling in adipocytes NO was initially identified as the first gaseous messenger molecule that acts through a completely novel mechanism, that is, binding to ferrous haem of sGC, leading to increased levels of the second messenger cGMP (Ignarro, 1990a, b). This NO-mediated mechanism covers a range of downstream signalling pathways in fundamental processes of virtually all cell types (Murad, 1988;Murad et al., 1990;Krumenacker and Murad, 2006), including adipocytes (Hemmrich et al., 2010). In adipose tissue, the reaction between NO and sGC, occurring at nM concentrations of NO (Bellamy and Garthwaite 2001;Rodríguez-Juárez et al., 2007), is mediated by constitutive eNOS. Recent data suggest that NO produced by the nitrite reductase activity of XO also acts through the sGC/cGMP signalling pathway . Physiological levels of NO produced in adipose tissue regulate blood flow and vascularization in a sGC/cGMPdependent manner, promoting substrate uptake and product removal via the circulation, thereby matching energy inflow/outflow with tissue perfusion (Jobgen et al., 2006). Additionally, endogenous NO directly mediates the metabolic response of adipocytes (see Jobgen et al., 2006, McKnight et al., 2010. First, physiological levels of NO stimulate the insulin-triggered (Roy et al., 1998) and (probably) insulin-independent uptake and oxidation of glucose (Tanaka
Upon activation by NO, AMPK phosphorylates and inactivates acetyl-CoA carboxylase, thereby reducing the conversion of acetyl-CoA to malonyl-CoA, which suppresses de novo fatty acid synthesis and activates carnitine palmitoyltransferase I, facilitating the transport and oxidation of fatty acids in mitochondria. NO also increases OXPHOS and mitochondriogenesis through an AMPK-mediated increase in PPARγ coactivator 1α (PGC-1α) expression (Tedesco et al., 2010). PGC-1α is the principal regulator of mitochondrial biogenesis and function. In combination with PPARγ, PGC-1α increases mtDNA replication, OXPHOS, mitochondrial fatty acid β-oxidation, UCP1 and mitochondrial antioxidant defence primarily in brown adipocytes (Bossy-Wetzel and Lipton, 2003;Kelly and Scarpulla, 2004) and possibly white adipocytes (Clementi and Nisoli, 2005;Tedesco et al., 2010). Through these pathways, NO supports healthy adipose tissue function, that is, long-term insulin sensitivity (Fu et al., 2005;Wu et al., 2007;Jobgen et al., 2009).
The groups of Fu et al., (2005) and Wu et al., (2007) showed that dietary supplementation with L-arginine, a substrate for NO synthesis, reduces body weight and fat mass in Zucker diabetic fatty rats. L-arginine induced a marked increase in expression of nNOS, haem oxygenase 3, AMPK and PGC-1α in white adipose tissue of Zucker diabetic fatty rats (Fu et al., 2005). In addition to stimulation of eNOS activity by high-dose L-arginine via direct administration, L-arginine can outcompete binding of the endogenous inhibitor of eNOS, asymmetric dimethyl-L-arginine, and force its removal from endothelial cells by driving the cationic amino acid transporter (Closs et al., 2012). Effects of NO on reducing fat mass may be attributed, in part, to the appearance of brown/brite adipocytes or brown-like phenotype (UCP1) in white adipocytes, that is, browning (Nisoli and Carruba 2006;Joffin et al., 2015;Roberts et al., 2015). We recently showed that L-arginine enhances the cold exposure-induced UCP1 level in mitochondria of unilocular white adipocytes in rats. Moreover, a similar increase in UCP1 protein expression was observed in rats maintained at room temperature (presented in Figure 3).
The essential role of eNOS-synthesized NO at nM levels in the metabolic plasticity of adipose tissue was confirmed in studies showing that eNOS knockout mice exhibit decreased UCP1 and PPARγ expression, reduced number of mitochondria, defective energy expenditure, increased body weight, insulin resistance and hypertension (Nisoli et al., 2007). eNOS À/À mice displayed exaggerated high-fat diet-induced weight gain (Shankar et al., 2000;Duplain et al., 2001;Cook et al., 2003). Furthermore, eNOS expression was remarkably diminished in fat tissue in obese rodents (Valerio et al., 2006) and humans (Perez-Matute et al., 2009;Georgescu et al., 2011), while its overexpression prevented diet-induced obesity, increased metabolic activity and promoted brown adipose tissue-like phenotype in white adipose tissue in mice (Sansbury et al., 2012). Similarly, constitutive activation of eNOS by knocking in a phosphomimetic point mutation at Ser 1176 was shown to promote resistance to diet-induced weight gain (Kashiwagi et al., 2013).
NO activates AMPK and its downstream pathways in a cGMP/PKG-dependent manner (Jobgen et al., 2006). Conversely, eNOS itself may be activated by AMPK. In endothelial cells, AMPK activates eNOS via phosphorylation at Ser 1177 and Ser 633 (Chen et al., 1999(Chen et al., , 2009). The latter phosphorylation event is slower and Ca independent and serves to maintain NO synthesis after the initial increase in NO induced by  Targeting NO/superoxide ratio in obesity BJP Ser 1177 phosphorylation, owing to a positive feedback loop (Schulz et al., 2009). On the other hand, AMPK-mediated post-translational phosphorylation of iNOS inhibits its activity and enhances insulin sensitivity in adipose tissue (Pilon et al., 2004). Thus, regulation of AMPK by NO may play an important role in controlling the relative eNOS and iNOS activities, NO levels and effects on adipose tissue.
Interaction of NO with CcOx: a potential key metabo-regulatory role in the postprandial response of adipocytes?
In addition to the ferrous iron in sGC, NO interacts directly with several ferrous haem proteins in near diffusion-limited reactions (k × 10 7 -10 8 M À1 ·s À1 ). The most important of these is the reaction with CcOx, the terminal complex of the mitochondrial ETC. The NO-CcOx system helps to fine-tune cellular respiration (Brown, 1995) and metabolism (Semenza, 1999).
Increased expression of mitochondrial NOS (potentially nNOS) in adipocytes of ob/ob mice led to reduced oxygen uptake via inhibition of CcOx (Finocchietto et al., 2011). The presence of mitochondrial NOS in adipocytes (as in other cells) is disputable, but other NOS stimulated by insulin may have similar mitochondrial effects (Jezek et al., 2010), because in adipocytes, insulin rapidly enhances NOS activity (Ribière et al., 2002;Engeli et al., 2004) as in endothelial cells (Dimmeler et al., 1999). Moreover, eNOS relocalizes upon post-transcriptional modification, that is, attaching to the outer mitochondrial membrane, at least in neurons and endothelial cells (Henrich et al., 2002;Gao et al., 2004). This finding indicates that eNOS regulates mitochondrial function, and conversely, mitochondria regulate eNOS activity (Nisoli and Carruba, 2006). Accordingly, NO synthesized by eNOS may transiently bind to CcOx and inhibit respiration under appropriate stimulation (e.g. by insulin) (Brown, 2001). This may induce a type of 'metabolic hypoxia', a phenomenon that restricts oxygen utilization (Moncada and Erusalimsky, 2002), that is, glucose oxidation. In postprandial adipocytes, 'metabolic hypoxia' may switch the cell to the glycolytic mode of ATP production and redirect glucose into lipids, similar to mtNOS-derived NO in muscle cells, in response to insulin (Finocchietto et al., 2008). Overexpression of nNOS or iNOS may contribute to ETC inhibition, oxygen uptake inhibition, mitochondrial dysregulation and insulin resistance progression in prediabetic states via the same mechanism (see Jezek et al., 2010).

Interaction of NO with oxygen and thiols
NO also binds to oxygen and thiyl radicals, especially at μM levels. Reaction of NO with O 2 , resulting in N 2 O 3 or NO 2 , may further cause S-oxidation and S-nitrosylation (S-nitrosation) of protein side chains (Stamler et al., 1992;Stamler, 1995;Beltrán et al., 2000;Hess et al., 2001). Moreover, these reactions may occur by direct covalent binding of NO with cysteinyl thiols of GSH and proteins upon transfer of NO + from S-nitrosothiol (SNO). S-nitrosoglutathione contributes to preservation of cellular NO activity as a delayed donor of NO. S-nitrosoglutathione and protein CysNO may subsequently react with GSH to generate a mixed disulphide and promote the release of HNO, another vasodilator. Such trans-S-nitrosation reactions within cells facilitate redistribution of NO among different -SNO pools (Liu et al., 1998;Yang and Loscalzo, 2005).
Protein Cys residues are capable of oscillating between reduced (ÀSH) and different oxidized states, including thiolate anion (ÀS À ), sulphenate (ÀSO À ), disulphide (ÀS-S-), sulphinate (ÀSO 2 À ) or sulphonate (ÀSO 3 À ). Among these, the latter two are considered irreversible oxidative modifications in mammalian cells (Jones, 2008). The redox state of the sulphur atom within Cys in the catalytic (regulatory or binding) domain of proteins may alter activity. In the biological context, oxidation and nitrosation reactions are compared with phosphorylation/dephosphorylation as the second prototypic post-translation mechanisms leading to redox regulation of protein activity/function (Lane et al., 2001).
In adipocytes, as in most cells, S-nitrosation of mitochondrial ETC thiol groups (especially complexes I and II) may slow down respiration (Moncada and Erusalimsky, 2002), acting synergistically in the transient inhibition of CcOx via NO binding. Both processes play an important role in the regulation of glycolytic (i.e. lipogenic) mode of postprandial adipocytes (Jezek et al., 2010).
Moreover, transient S-oxidation/S-nitrosation is indispensable in the maintenance of normal insulin signalling in adipocytes. Under normal (unstimulated or basal) circumstances, most protein Cys residues are maintained in the reduced state due to high levels and overall presence (cytoplasmic and mitochondrial) of GSH and thioredoxins (Trx), peroxiredoxins and the corresponding NADPHcoupled reductase systems (Fisher-Wellman and Neufer, 2012). This sets up a 'basal' phosphatase tone in unstimulated adipocytes, because protein phosphatases, particularly protein tyrosine phosphatases (PTPs) and the phosphoprotein phosphatase family of Ser/Thr phosphatases, are active under reduced conditions (Chiarugi, 2005;Wright et al., 2009;Fisher-Wellman and Neufer, 2012). However, a small shift in the redox state to prooxidative conditions, for example, due to insulin-mediated increase in ROS/RNS, promotes S-oxidation/S-nitrosylation of Cys in different proteins, such as protein kinases and phosphatases involved in insulin signalling. ROS/RNS play a superimposing role by controlling the phosphorylation state of proteins in the insulin signalling cascade via reversible S-oxidation/S-nitrosylation of PTPs, and their reactivity may therefore lead to changes in insulin sensitivity in adipose tissue. Moreover, NO-induced S-nitrosylation at the active site Cys residue of PTPs (SHP-1, SHP-2 and PTP1B), concomitant with eNOS-mediated NO burst in response to insulin action, supports NO-dependent regulation of tyrosine phosphorylation of the insulin receptor and its downstream effector kinases, insulin receptor substrate 1 (IRS-1) and PKB/Akt, at least in mouse endothelial MS-1 cells (Hsu and Meng, 2010). It is tempting to speculate that NO plays important permissive roles for transmission of the insulin signal in adipocytes also, via similar mechanisms.

Interaction of NO with superoxide
Within all the molecular targets of NO in the cell, superoxide is the most critical intervening factor of its biological effects, especially in the presence of high levels of NO. At equimolar BJP A Jankovic et al.
concentrations, these two reactive species combine at a diffusion-limited rate (k approximately 10 9 to 10 10 M À1 ·s À1 ) that exceeds the reactions of NO with all other biotargets (iron, thiols and oxygen) and the reaction velocity upon combination with superoxide dismutase (SOD) (k approximately 10 9 M À1 ·s À1 ), changing the biological action of NO and other ROS/RNS under oxidative stress conditions (Koppenol et al., 1992;Koppenol, 2001). The balance between local concentrations of NO, О 2 •and SOD is important in determining the effects of the NO/superoxide radical pair (Beckman and Koppenol, 1996;Rubbo et al., 1996). No significant competition occurs between NO and SOD for О 2 •under normal physiological conditions, because NO is in the nM range, compared with μM levels of SOD in cells. Under physiological conditions, reaction of О 2 •with NO constitutes only a minor part of the dismutation reaction of О 2 •by SOD. As a result, very little peroxynitrite (ONOO À ) is formed (Ferdinandy and Schulz, 2003). However, at increased rates of NO production (possibly via iNOS expression), when the levels of NO are ≥1 μM, ONOO À formation predominates over dismutation of О 2 •-(Depre and Hue, 1994;Csonka et al., 1999;Ferdinandy et al., 2000). Additionally, the increase in О 2 •levels due to increased production and/or decreased dismutation may lead to an effective increase in the probability of ONOO À formation. Both prerequisites, iNOSmediated NO production and increase in superoxide levels, occur in adipose tissue under conditions of obesity.
Adipose tissue in the obese state: the paradigm of NO/superoxide interactions As mentioned previously, decreased NO bioavailability is a sign of early risk of fat-specific insulin resistance development in obesity, and increased superoxide level in accumulating adipose tissue is the key factor in the initiation and progression of fat-specific insulin resistance. Inter alia, high О 2 •and NO levels are characteristic of adipose tissue in obesity. О 2 •antagonizes the direct effects of NO, and ONOO À formation and/or derived ROS/RNS can alter cell signalling, in part, by promoting oxidation, nitrosylation or nitration of a broad range of proteins, including enzymes of intermediary metabolism, mitochondrial complexes and insulin signalling.

Sources of high superoxide in obesity: antagonists of physiological NO functions
Superoxide in adipose tissue is primarily generated by overloading the mitochondrial OXPHOS system with metabolites from glucose, fatty acids and NOXs (Han et al., 2012). In recent years, eNOS uncoupling has been recognized as an important superoxide source in various tissues in obesity (Margaritis et al., 2013;Yu et al., 2014). XO, an enzyme that catalyses the conversion of hypoxanthine to uric acid and О 2 •-, has been proposed to play a critical role in superoxide production in mature adipocytes (Furukawa et al., 2004). Although hyperuricaemia has been linked to metabolic syndrome, the role of XO in adipose tissue as the source of superoxide remains poorly understood. The enzyme may be of importance, especially in impairment of adipose tissue blood flow, because pro-inflammatory cytokines irreversibly convert endothelial xanthine dehydrogenase to the oxidase form, XO (Vorbach et al., 2003).
Enhanced production of superoxide by mitochondria. During respiration, a significant proportion of oxygen molecules are incompletely converted to H 2 O and end up as О 2 •via a non-enzymatic pathway, owing to electron leak from complexes I and III of the ETC (Turrens and Boveris, 1980). Mitochondrial О 2 •is linked to hyperglycaemia-induced metabolic dysfunction in endothelial cell systems (Brownlee, 2001) and inflammation in adipocytes (Lin et al., 2005). A recent study showed that overfeeding (and/or high-calorie intake) increases the reduced state of electron carriers in the ETC and subsequently the probability of electron extraction by high reduction potential molecules, such as molecular oxygen (to create О 2 •-) (Matsuzawa-Nagata et al., 2008). This series of events is more evident in hyperglycaemia. When the elevated carbohydrate fuel supply exceeds the metabolic needs of the cell, reducing equivalents (NADH) are overproduced (Frizzell et al., 2012). A dramatic increase in NADH exceeds the electron acceptor capacity of ETC complexes, thereby perpetuating electron leak and subsequent О 2 •generation.
NOX. Superoxide/H 2 O 2 originating from NOX produced transiently in response to insulin stimulation acts as a second messenger for insulin signalling in adipocytes Kather, 1992, 1995), while excessive and long-term activation of NOX reduces insulin sensitivity (Furukawa et al., 2004). NOX is a membrane-associated multimeric oxidoreductase that transports electrons preferentially from cytosolic NADPH (although nonphagocytic NOX, especially the NOX1 isoform, may also use NADH as substrate) down to the electrochemical gradient through the membrane to oxygen, generating О 2 •that is rapidly converted to H 2 O 2 (Griendling et al., 2000). Among the seven isoforms of the catalytic subunit gp91 phox of NOX, phagocytic NOX2 and NOX4 are present in adipose tissue, especially in resident macrophages and adipocytes (Mouche et al., 2007). NOXs are significantly increased in adipose tissue in different genetic and dietinduced models of obesity in rats (Furukawa et al., 2004) and humans (Jankovic et al., 2014). Moreover, a higher gp91 phox protein level in visceral fat characterizes subjects with increased risk factors for metabolic syndrome, compared with metabolically healthy weight-matched subjects (Jankovic et al., 2014).
NOX4 generates H 2 O 2 even under basal conditions (via constitutive expression) (Nisimoto et al., 2010). Insulin and some anabolic hormones induce a several-fold increase in H 2 O 2 levels and expression of NOX4 (Goldstein et al., 2005).
Increased oxidative state can lead to immune cell activation (Schulz et al., 2014;Kröller-Schön et al., 2014). In particular, excess glucose and palmitate generate ROS via a mechanism that involves translocation of NOX4 into lipid rafts of adipocytes, leading to expression of monocyte chemotactic factors (Yeop Han et al., 2010;Han et al., 2012). Conversely, inflammatory cytokines and NOX2-mediated ROS originating from activated macrophages may be involved in augmentation of adipocyte NOX4 and consequent ROS production in obese adipose tissue (Furukawa et al., 2004). Possibly, Targeting NO/superoxide ratio in obesity BJP complex crosstalk of adipocyte and macrophage NOXs establishes a vicious cycle that augments superoxide production in adipose tissue in obesity and diabetes (recently reviewed in Karbach et al., 2014, andJankovic et al., 2015b).
Enhanced production of superoxide through eNOS uncoupling. In the absence of the coenzyme tetrahydrobiopterin (BH 4 ), NOS reduces molecular oxygen, rather than L-arginine, resulting in production of superoxide rather than NO, a phenomenon known as 'NOS uncoupling' . Several hypotheses have been proposed for intracellular BH 4 depletion, including ONOO À -mediated oxidation. Moreover, activation of the superoxide source (uncoupled eNOS) may stimulate the formation of superoxide and/or ONOO À in a positive feedback manner, in turn, oxidising BH 4 to the BH 3 radical, leading to further ROS/RNS formation (Kuzkaya et al., 2003). eNOS dysfunction was observed under conditions of oxidative stress and inflammation in perivascular adipose tissue (Filip et al., 2012;Margaritis et al., 2013). Increased superoxide production by uncoupling of eNOS is recognized as an important factor that decreases the vasodilatory role of perivascular adipose tissue in obesity. During early diet-induced obesity, adaptive overproduction of NO occurs in perivascular adipose tissue, while in established obesity, perivascular adipose tissue loses its vasodilatory properties via an increase in 'contractile' superoxide, leading to endothelial dysfunction and vascular disease (Fernandez-Alfonso et al., 2013). Increased nitro-oxidative pressure seen in perivascular adipose tissue in obesity may also occur in other adipose tissue areas. In fact, this may be critical for the impairment of blood flow, tissue oxygen and substrate supply in adipose tissues in obesity (see Alemany, 2012).
Lower antioxidant defence mechanisms. To prevent ROS/RNS excess, cells are equipped with antioxidant enzymes, such as manganese SOD (MnSOD), in the mitochondrial matrix and copper, zinc SOD in the cytosol and intermembrane space of mitochondria, which convert О 2 •into H 2 O 2 (Fridovich, 1995). H 2 O 2 is further reduced to H 2 O by catalase, GSH peroxidase and Trx/peroxiredoxin systems. GSH and Trx reduce peroxide concentrations and protein disulphide, subsequently producing oxidized GSH and Trx respectively. Oxidized GSH and Trx are converted back to reduced GSH, that is, Trx, via NADPH-dependent reductases, respectively, thus maintaining the reduced cellular redox state (Halliwell and Gutteridge, 2007). NADPH, the reducing power for GSH (Cys) and Trx recycling, is provided by the pentose phosphate pathway and malic enzyme. Accordingly, lower SOD levels may contribute to increased superoxide and thus higher probability of ONOO À formation in the metabolic syndrome. Likewise, the mitochondrial antioxidant enzymes, MnSOD and GSH peroxidase, show decreased activity in adipose tissue, with the most pronounced decline in obese type 2 diabetic subjects and, to a lesser degree, non-obese type 2 diabetic or non-diabetic obese subjects (Chattopadhyay et al., 2015). Consistent with these findings, our group showed that SOD levels were not significantly decreased in obese, insulin-sensitive subjects, while a significant decrease in SOD activity and MnSOD protein expression, in addition to lower levels of GSH, characterized visceral adipose tissue of subjects with increased cardiometabolic risk factors (Jankovic et al., 2014).
Besides conventional antioxidant mechanisms, uncoupling capacity plays a significant role in determining superoxide production from ETC. Uncoupling decreases the pressure on ETC complexes and increases electron transfer capability, thereby suppressing the leak of electrons and subsequent production of О 2 •-. UCP1 gene expression is undetectable in adipogenic precursor cells isolated from lean and obese individuals from subcutaneous abdominal white adipose tissue biopsies. However, after adipocyte differentiation, both gene expression and protein content of UCP1 are increased. UCP1 levels are significantly greater in cultures from lean, compared with obese individuals (Carey et al., 2014). Moreover, morbidly obese subjects express significantly lower levels of UCP mRNA than lean controls (Oberkofler et al., 1997). In parallel, fat mitochondria from obese type 2 diabetic subjects produce considerably more ROS, compared with those of controls and non-diabetic subjects. In the majority of cells, including adipocytes, the antioxidant role is mainly attributed to UCP2. However, recent data indicate that a short-term UCP1 increase may buffer the reductive pressure on mitochondrial ETC and consequent oxidative pressure in white adipocytes (Jankovic et al., 2015b). In addition, both uncoupling agents and MnSOD mimics that alleviate the О 2 •level in the ETC rapidly restore the insulin sensitivity of adipocytes (Hoehn et al., 2009). Thus, the antioxidant role of mitochondrial UCPs is important during adipocyte differentiation and their response to overfeeding, and its impairment may contribute to higher mitochondrial superoxide release. Moreover, reduction in mitochondrial number without concomitant reduction in nutrient uptake leads to an increase in net substrate flux through the remaining mitochondria and subsequent О 2 •production (Hoehn et al., 2009). Thus, lower or impaired mitochondriogenic potential may also contribute to increased superoxide levels in persistent overfeeding.
iNOS in obesity and insulin resistance iNOS À/À mice are protected from high-fat diet-induced insulin resistance. While wild-type and iNOS À/À mice on a high-fat diet develop obesity, obese iNOS À/À mice exhibit improved glucose tolerance, normal insulin sensitivity in vivo and normal insulin-stimulated glucose uptake in muscles. iNOS is increased in fat of genetic and dietary (high-fat feeding) models of obesity (Perreault and Marette, 2001). Chronic NO synthesis by iNOS represents an important contributory factor to nitrosative stress and development of fat-specific insulin resistance (Kaneki et al., 2007). In adipose tissue, the majority of iNOS is derived from phenotypic transformation of anti-inflammatory 'alternatively activated' macrophages to a more pro-inflammatory 'classically activated' form (Weisberg et al., 2003;Lumeng et al., 2007). In addition, high levels of TNF-α in adipocytes induce increased iNOS expression (Merial et al., 2000). Almost any factor that BJP A Jankovic et al.

Effects of interaction of NO with superoxide in adipose tissue in obesity
Interactions between NO and superoxide and the consequent effects in vivo are very complex, because they depend not only on the relative levels of reactants and kinetics but also on spatial (where) and temporal (at the same time) prerequisites. In contrast to NO that may act as either an intracellular or intercellular messenger, the half-life of superoxide is mainly restricted by SODs, and therefore, interactions most likely occur at the sites of superoxide production. In adipocytes, mitochondria and NOX4 enzyme localized in the cellular plasma membrane and intracellular membranous structures represent the major sites of NO/superoxide interactions (Bedard and Krause, 2007). This may also explain why the major burden of S-nitrosated/nitrated proteins is found in these compartments. Finally, ONOO À , the product of NO/superoxide interactions, exists as peroxynitrous acid (ONOOH) at physiological pH that is prone to proton-catalysed and carbon dioxidecatalysed homolysis, generating even more potent oxidants, such as hydroxyl ( • OH), nitrogen dioxide (NO 2 • ) and carbonate anion (CO 3 •-) radicals, which induce oxidation, nitrosation or nitration of protein side chains (see Toledo and Augusto, 2012). Protein modification by tyrosine nitration is mainly attributable to ONOO À . The reversibility of tyrosine nitration by denitrases (Kamisaki et al., 1998;Irie et al., 2003) and proteolysis (and resynthesis of nitrated proteins) (Souza et al., 2000) has altered the mainly negative perception of nitration and ONOO À . Accumulating data indicate that this protein modification plays an adaptive role in specific settings (Ferdinandy and Schulz, 2003), and the target proteins and extent of modification determine the resulting patho/physiological effects (Koeck et al., 2005(Koeck et al., , 2009. Extensive nitration of tyrosine residues in proteins is the fingerprint of different pathophysiological conditions, including the diabetic state (Koeck et al., 2009;Charbonneau and Marette, 2010;Pilon et al., 2010). All aspects of NO and superoxide interactions should be taken into account to explain why the higher production rate of NO and superoxide usually contributes to pathological conditions but also plays an adaptive role. This is very challenging when considering adipose tissue in obesity whose expansion is associated with pathologies but per se represents the adaptive response of tissue (schematically presented in Figure 4). The literature shows that the level of endogenously formed ONOO À increases in adipose tissue mainly as a result of hyperglycaemia, that is, established diabetes (Koeck et al., 2009). A similar increase in the ONOO À level in diabetic hearts has been reported (Pechánová et al., 2015;Varga et al., 2015). In these circumstances, increased signalling through advanced glycation end products or their corresponding receptors (RAGE) may additionally lead to excessive superoxide formation and ONOO À increase. Once formed, ONOO À triggers a vicious cycle, further decreasing NO bioavailability and increasing nitrooxidative stress. Using different mechanisms, ONOO À may oxidize and decrease eNOS-mediated NO production and divert direct sGC-mediated and CcOx-mediated bioeffects of NO in adipocytes, including insulin sensitivity and mitochondrial number and function (through low PGC-1α activation and/or expressionbiogenic, OXPHOS and antioxidant capacity of mitochondria). In addition, ONOO À may increase superoxide generation (again via eNOS and/or iNOS uncoupling, although the latter has not been confirmed for adipose tissue) and decrease superoxide neutralization (via nitration of MnSOD) and the pool of GSH. The latter parameters are initially increased in obesity as a compensatory response to increased levels of superoxide, SNO and ONOO À . The diminished antioxidant capacity of adipocytes (initially mitochondrial and subsequently cytosolic) renders more proteins susceptible to oxidation and nitration, resulting in inactivation or dysfunction. However, the important issue of which of these multiple effects represents an early event in ONOO À -mediated toxicity at a time point prior to the onset of irreversible functional changes and insulin resistance in adipocytes remains to be resolved.

Nitro-oxidative state and insulin signalling/sensitivity
A number of researchers are currently investigating the pathways with potential ONOO À involvement in relation to insulin signalling and resistance. Growing evidence suggests that tyrosine nitration can alter protein function by preventing functional phosphorylation (Mondoro et al., 1997;Rawlingson et al., 2003;Kaneki et al., 2007;Stadler, 2011). Hepatic insulin resistance in lipid-challenged mice results from ONOO À -mediated tyrosine nitration of insulin receptor (IR)β and IRS-1/-2 (which promotes inhibitory serine phosphorylation of IRS proteins) and Akt, directly inhibiting insulin signalling (Charbonneau and Marette, 2010). In addition, ONOO À mediates muscle insulin resistance via nitration of IRβ/IRS-1 and Akt , while targeted disruption of iNOS reverses high-fat diet-induced impairment of the insulin-stimulated tyrosine phosphorylation of IR, IRS-1 and IRS-1-associated PI3K activity, Akt and insulin resistance in muscle of mice (Perreault and Marette, 2001). Yasukawa et al. (2005) reported that a similar mechanism of ONOO À -mediated nitration may contribute to insulin resistance in adipocyte tissue. The group showed that pretreatment of cultured mouse 3T3-L1 adipocytes with the NO donor, S-nitroso-N-acetylpenicillamine, inhibits insulin-stimulated Akt/PKB activation. Activation of the renin-angiotensin-aldosterone system plays an important role in cardiovascular complications in metabolic syndrome, and angiotensin II is a strong trigger of tyrosine nitration and inactivation of kinases involved in insulin signalling (Csibi et al., 2010).
The data of Nomiyama et al. (2004) showed that constitutive production of ONOO À by the NO/О 2 •donor -3-(4-morpholinyl) sydnonimine hydrochloride (SIN-1)dose-dependently inhibited insulin-stimulated glucose uptake in rat-1 fibroblasts expressing human insulin receptors. SIN-1 reduced the IRS-1 protein level, and IRS-1 associated PI3K activity through tyrosine nitration of at least four tyrosine residues, including Tyr 939 , which is critical for the association of IRS-1 with PI3K. A continuous consumption of surplus of nutrients also increased tyrosine nitration in adipocytes (Koeck et al., 2009). Moreover, the extent of tyrosine nitration and the cellular nitroproteome profile in adipocytes reflects rise in the glucose Targeting NO/superoxide ratio in obesity BJP concentration during continuous exposure, as well as during the periodic fluctuation in the more physiological glucose concentration (Koeck et al., 2009). The nitrated proteins in glucose-overloaded adipocytes are primarily those involved in glucose metabolismglycolysis pathway and citric acid cycle (aldolase A, glyceraldehyde 3-phosphate dehydrogenase,

Figure 4
Interaction of NO and superoxide in the physiology and pathophysiology of adipose tissue. FA, fatty acid.

BJP
A Jankovic et al.
phosphoglycerate kinase, malate dehydrogenase, aconitase). Besides, increased nitration of some of the enzymes involved in lipid trafficking/oxidation contributes to insulin resistance in adipocytes. For example, both functionally important tyrosine residues (Tyr 19 and Tyr 128 ) in fatty acid binding protein 4 are found to nitrate in adipocytes exposed to high metabolic load (Koeck et al., 2009). The nitration of fatty acid binding protein 4 may affect its interaction with hormonesensitive lipase (Adida and Spener, 2006), translocation into the nucleus and interaction with PPARγ (Smith et al., 2007), which, in turn, may additionally down-regulate all PPARγmediated metabolic effects in adipose tissue, including insulin sensitivity (Koeck et al., 2009).
In contrast to these in vitro data, the evidence of high endogenous ONOO À levels and activity, primarily tyrosine nitration, in adipose tissue in obese or prediabetic animal models or humans is rather scarce. The reason may be that tyrosine nitration is a highly selective process limited to specific tyrosine residues on a surprisingly small number of proteins (Aulak et al., 2001;Gow et al., 2004;Kanski et al., 2005) and is thus detectable only at high ONOO À levels that mainly characterize already established diabetic conditions, as stated above.
Nevertheless, increased total protein S-nitrosylation, an alternative sign of ONOO À acting, has been detected in intra-abdominal adipose tissue of obese humans and high fat-fed or leptin-deficient ob/ob mice (Ovadia et al., 2011). Importantly, Yin et al. (2015) recently revealed the role of increased S-nitrosylation as the mechanism that links obesityassociated inflammation, lower PPARγ activity and insulin resistance in adipose tissue. They found that iNOS-mediated increase of nitro-oxidative stress translates from macrophages to adipocytes leading to S-nitrosylation of PPARγ on Cys 168 . This in turn potentiates proteasomal degradation of PPARγ, decreasing its transcriptional function. Moreover, expression level of the main PPARγ-regulated insulin-sensitizing adipokine, adiponectin, was also lower as a result of increased S-nitrosylation of PPARγ (Yin et al., 2015). Similar results were observed in vivo, in visceral adipose tissue of obese diabetic db/db mice (Yin et al., 2015).

Nitro-oxidative state and mitochondrial dysfunction
Disruption of mitochondrial function is implicated in fatspecific insulin resistance. Many mitochondrial proteins, including the enzymes involved in fat oxidation and energy supply, could be oxidized/nitrated under increased nitro-oxidative pressure, leading to depletion of ATP and impairment of lipid-buffering capacity in diabetic adipose tissue in obesity. The half-life of ONOO À under cellular conditions is short (~less than 1 s), but sufficient to cross biological membranes. Thus, ONOO À or peroxynitrous acid may enter mitochondria from the cytosol or be directly produced within mitochondria. Indeed, persistent hyperinsulinaemia and nNOS/mtNOS activation may contribute to mitochondrial dysfunction and insulin resistance of adipocytes (Jezek et al., 2010;Finocchietto et al., 2011). A remarkable increase in nNOS activity and NO level in mitochondria of ob/ob adipocytes inhibits CcOx and cause intense nitration at complex I impeding subsequent electron transfer from NADH (Finocchietto et al., 2011). Recurrent nitrosative stress of progressively higher intensity and duration would lead to parallel oxidative stress due to corresponding shifts in respiratory chain redox states that stimulate О 2 •formation. In fact, according to the 'Poderoso hypothesis', accumulating oxidative/nitrosative stress in mitochondria of fat cells could lead to retrograde modulation of the insulin signalling pathway and therefore represents the basis of type 2 diabetes aetiology (Jezek et al., 2010).
Nitro-oxidative state and lipolysis/re-esterification dysregulation As stated, nM levels of NO stimulate cAMP and PKA/hormone-sensitive lipase signalling, that is, lipolysis (Gaudiot et al., 2000;Jobgen et al., 2006), and decreased eNOS-mediated lipolysis could favour the development of obesity through retention of TAG within adipocytes. However, in obesity, diminished lipolysis can also be viewed as a mechanism limiting excessive fatty acid release and alleviating the development of insulin resistance and metabolic abnormalities (Girousse et al., 2013). Notably, impaired anti-lipolytic effects of insulin in the postprandial period could restrict fatty acid esterification in adipocytes, exposing non-adipose tissue to lipotoxicity. Recent studies have revealed that reduction in lipolysis improves glucose incorporation into adipocyte lipids without increasing fat mass (Girousse et al., 2013). Because high NO levels inhibit basal as well as catecholamine-stimulated lipolysis (Gaudiot et al., 1998;Andersson et al., 1999;Klatt et al., 2000), iNOSmediated increase in NO in adipose tissue under insulin stimulation is considered an important regulator of the lipogenic function of postprandial adipocytes (Engeli et al., 2004). However, recent data indicate that, in obesity, iNOS may, in fact, impair anti-lipolytic regulation in postprandial adipocytes, possibly via increased nitro-oxidative pressure. Specifically, iNOS increases S-nitrosylation of PKB/Akt and cAMP PDE (PDE3B). S-nitrosylation of PDE3B at Cys 768 and Cys 1040 (Ovadia et al., 2011) inhibits its cAMP hydrolytic activity (Zmuda-Trzebiatowska et al., 2007). The main metabolic consequence of increased Snitrosylation and inactivation of Akt/PDE3B axis in adipose tissue in obesity is impairment of insulin-induced antilipolysis (Ovadia et al., 2011). In addition to uncontrolled lipolysis, a high fatty acid output from adipose tissue leading to lipotoxicity and insulin resistance may result from impaired fatty acids reesterification. This metabolic pathway requires glycerol-3P synthesis, a pathway relaying on the activity of cytosolic phosphoenolpyruvate carboxykinase (PEPCK-C). Niang et al. (2011) andJaubert et al. (2012) showed that long-term leptin treatment inhibits PEPCK-C by increasing iNOSmediated PEPCK-C nitration, therefore limiting fatty acid -reesterification in rat adipocytes. The increased nitration of PEPCK-C by leptin may initially play beneficial role in obesity setting, by increasing energy expenditure (Jaubert et al., 2012). Nevertheless, in the long-standing hyperleptinaemic obesity increased nitration of PEPCK-C may restrict glyceroneogenesis, essential in the lipid buffering function of adipose tissue, ultimately leading to lipotoxicity.

Nitro-oxidative state and endoplasmic reticulum stress
The endoplasmic reticulum is essential for the folding and trafficking of proteins that enter the secretory pathway. The accumulation of misfolded or unfolded proteins in this organelle, known as endoplasmic reticulum stress, is a common pathogenic molecular mechanism underlying adipocyte-specific insulin resistance (Kawasaki et al., 2012;Boden et al., 2014). It has been recently shown in both genetic (ob/ob) and dietary (highfat diet-induced) models of obesity that, in the setting of obesity, inflammatory input through increased iNOS-mediated S-nitrosylation and impaired splicing activity of a key regulator of adaptive unfolded protein responseinositol-requiring protein 1leads to impaired endoplasmic reticulum function (Yang et al., 2015). These data were obtained with liver tissue, but it could be assumed that increased S-nitrosylation of the important proteins mediating unfolded protein response exists in adipose tissue in obesity. In support of this, injection of mice with a SOD mimetic (Mn (iii)tetrakis(4-benzoic acid) porphyrin) attenuated the induction of the unfolded protein response (Malhotra et al., 2008), while inhibition of iNOS reversed palmitateinduced endoplasmic reticulum stress response in 3T3-L1 adipocytes (Jeon et al., 2012).

Nitro-oxidative stress versus endogenous antioxidant defence
In addition to protein-bound thiols, ONOO À can oxidize GSH (Villa et al., 1994;Mayer et al., 1995;Prendergast et al., 1997). GSH is an efficient endogenous scavenger of ONOO À and plays a major role in cellular defence against this species. Thus, in the long term, exhaustion of GSH may Table 1 Potential therapeutic strategies to improve fat-specific insulin sensitivity by targeting NO/superoxide ratio in adipose tissue

Strategy (treatment) Molecular mechanisms Target cells and effects References
Exercise training and hypoglycemic agents render adipose tissue more susceptible to ONOO À -mediated nitro-oxidative damage (Pacher et al., 2007). In addition, lower reduced GSH/Trx levels may decrease de-nitrosylation of SNO moieties and further increase the S-nitrosylated protein level. Overall, changes in intracellular GSH level correspond to changes in lipogenesis and adipogenesis (Galinier et al., 2006). However, the GSH level decreases in adipose tissues in subjects with long-lasting obesity (Jankovic et al., 2014). This may represent an early sign of redox dysbalance that, if not compensated by other redox molecules like the Trx system, leads to increased vulnerability to oxidant insult and decreased fat storage buffering capacity of adipocytes in obese subjects (Jankovic et al., 2014). Interestingly, subcutaneous, compared with visceral, adipose tissue characterizes greater adaptive capacity of antioxidant defence on GSH depletion (Jankovic et al., 2014).
Perspectivesdoes targeting of the NO/superoxide ratio in adipose tissue hold promise?
Targeting of the NO/О 2 •ratio in adipose tissue may have relevance in the elucidation and management of obesity and diabetes pathogenesis. Although the use of antioxidant supplements may have beneficial effects on the overall health of diabetic patients, classical antioxidant therapy with vitamins failed to show benefits in curbing fat-specific insulin resistance, diabetes and related cardiovascular diseases (Münzel et al., 2010;Abdali et al., 2015). Adipose tissue in obesity is possibly a paradigm of parahormesis (Forman et al., 2014), whereas parallel increases in electrophiles, oxidants (superoxide, ONOO À ), nucleophiles and antioxidants (GSH, Trx and NADPH), in concert with increased nutrient metabolism in adipocytes, play important signalling roles, enabling adaptive expansion of adipose tissue in increasing nutrient intake (Jankovic et al., 2015b). Thus, antioxidant as well as the hormetic approach (to increase oxidants that would strengthen endogenous antioxidants) may interfere with the endogenous steady-state redox level in adipose tissue in obesity, thereby restricting its metabolic plasticity. The ideal approach is to increase NO bioavailability and/or simultaneously limit excessive superoxide formation (briefly summarized in Table 1). Currently, alternative antioxidant-promoting methods, such as L-arginine supplementation (Lucotti et al., 2006;McKnight et al., 2010), caloric restriction  or use of SOD mimetics (Hoehn et al., 2009), are considered promising therapeutic strategies.