H2O2 Signalling Pathway: A Possible Bridge between Insulin Receptor and Mitochondria

This review is focused on the mechanistic aspects of the insulin-induced H2O2 signalling pathway in neurons and the molecules affecting it, which act as risk factors for developing central insulin resistance. Insulin-induced H2O2 promotes insulin receptor activation and the mitochondria act as the insulin-sensitive H2O2 source, providing a direct molecular link between mitochondrial dysfunction and irregular insulin receptor activation. In this view, the accumulation of dysfunctional mitochondria during chronological ageing and Alzheimer’s disease (AD) is a risk factor that may contribute to the development of dysfunctional cerebral insulin receptor signalling and insulin resistance. Due to the high significance of insulin-induced H2O2 for insulin receptor activation, oxidative stress-induced upregulation of antioxidant enzymes, e.g., in AD brains, may represent another risk factor contributing to the development of insulin resistance. As insulin-induced H2O2 signalling requires fully functional mitochondria, pharmacological strategies based on activating mitochondria biogenesis in the brain are central to the treatment of diseases associated with dysfunctional insulin receptor signalling in this organ.


INTRODUCTION
Hydrogen peroxide (H 2 O 2 ) and superoxide anion radicals, collectively known as reactive oxygen species (ROS), are traditionally regarded as toxic byproducts of aerobic metabolism. Therefore, evidence that insulin action is facilitated by ROS was considered a redox paradox [1]. Most knowledge in this area comes from studies in fat cells. As early as the 1970s, insulin was discovered to stimulate the production of intracellular H 2 O 2 in adipose tissue [2,3]. Later, it was found that a membrane-bound NADPH oxidase, catalysing the reduction of molecular oxygen to superoxide, produced ROS in fat cells in response to insulin stimulation [4][5][6]. ROS are involved in upregulating very early insulin receptor signalling, since inhibition of the insulin-induced ROS generation attenuates tyrosine phosphorylation of the insulin receptor and its substrates [7,8]. Protein tyrosine phosphatases (PTP), which dephosphorylate and inactivate insulin receptors, are a target of insulin-induced ROS [7][8][9]. H 2 O 2 is a particularly important ROS for insulin receptor signalling, given that catalases almost completely inhibit the effects of insulin-induced ROS [8]. Thus, there is a signalling pathway enhancing very early insulin receptor signalling in fat cells, which involves NADPH oxidasemediated H 2 O 2 generation.
Emerging evidence suggests that mitochondria are the insulin-sensitive source of ROS essential for insulin receptor activation in neurons. These findings provide a direct molecular link between mitochondrial function and insulin receptor signalling, highlighting a role of mitochondrial dysfunction in the development of insulin resistance. This review is focused on the mechanistic aspects of the insulininduced H 2 O 2 signalling pathway in neurons and the factors affecting sit, which could promote the development of insulin resistance.

RECEPTOR IS REGULATED BY TYROSINE PHOSPHORYLATION
Cerebral insulin is produced by the pancreas and enters the cerebrospinal fluid by receptor-mediated saturable transport [10]. Little or no insulin is produced in the brain itself. Insulin binds to its cognate receptors and elicits a variety of biological responses [11]. Mice lacking the insulin receptor gene via targeted disruption die within the first week of birth due to severe diabetic ketoacidosis [12,13]. Most knowledge regarding insulin receptor signalling comes from studies in classical insulin target tissues such as fat, muscle and the liver, where insulin is essential for regulating energy functions such as glucose and lipid metabolism. Unlike its peripheral counterpart, the central insulin receptor is involved in anorexigenic responses, fertility and reproduction, memory formation, and neuronal survival, and is considered to have no direct effect on neuronal glucose metabolism [14][15][16][17][18][19][20].
The insulin receptor is a heterotetrameric protein composed of two α-subunits and two β-subunits linked by disulphide bonds [11]. The extracellular α-subunits have sites for insulin binding. The intracellular portion of the transmembrane β-subunits contains the insulin-regulated tyrosine kinase. The neuron-specific isoform of the insulin receptor (isoform A) differs from the peripheral insulin receptor (isoform B) and arises from the alternative splicing of exon 11, which removes 12 amino acids near the COOH terminus of the α-subunit [21,22]. There are some notable functional distinctions between the neuron-specific receptor and its peripheral counterpart. The neuronal insulin receptor binds insulin with a two-fold higher affinity, can also bind insulin-like growth factor-II with physiologically relevant affinity, shows no negative cooperativity, and has increased rates of receptor internalisation [23][24][25][26]. The neuronal isoform is the predominant insulin receptor isoform in the brain, with the highest density in the olfactory bulb, cerebral cortex, hypothalamus and hippocampus, where it is concentrated at synapses as components of postsynaptic densities [27][28][29].
The reversible tyrosine phosphorylation of the insulin receptor lies at the core of insulin signalling. Upon insulin binding, the insulin receptor undergoes autophosphorylation at three critical tyrosines within the activation loop of the tyrosine kinase domain, whose modification dramatically increases tyrosine kinase activity and triggers the downstream signalling cascades [30][31][32].
Dephosphorylation of the insulin receptor is catalysed by several members of the protein tyrosine phosphatase family, including the intracellular PTP1B, which has a major physiological role in the negative regulation of the insulin receptor itself and downstream effectors [9,[33][34][35]. There is evidence that a decrease in PTP1B activity improves insulin receptor signalling in neurons. A decrease in hypothalamic PTP1B lowers food intake, reducing body weight and improving insulin action and signalling in the hypothalamus [36]. A conditional deletion of PTP1B in the retina was shown to enhance insulin receptor signalling and cell survival [37]. Mice lacking PTP1B demonstrate increased tyrosine phosphorylation of the insulin receptor, improved systemic insulin sensitivity and obesity resistance [38,39].
It is generally accepted that PTP1B is a physiologically relevant target of insulin-induced H 2 O 2 [1]. A catalytic cysteine at amino acid position 215 of PTP1B is highly susceptible to direct oxidation by micromolar H 2 O 2 [40][41][42]. Depending on the reaction conditions, PTP1B undergoes a variety of inhibitory post-translational modifications, of which the S-glutathionylated form seems to be predominant in the intracellular milieu where glutathione is abundant [43][44][45]. The inhibitory S-glutathionylation is reversible and PTP1B can be reactivated by the glutaredoxin and thioredoxin systems [42]. As PTP1B dephosphorylates the already active phosphorylated form of the insulin receptor, H 2 O 2 -mediated PTP1B inhibition regulates insulin receptor activity in cells.

INSULIN-INDUCED ROS ARE REQUIRED FOR INSULIN RECEPTOR TYROSINE PHOSPHORYLA-TION IN NEURONS
Some evidence suggests that insulin-stimulated H 2 O 2 plays a critical role in early insulin receptor signalling in neuronal cells. Human neuroblastoma SK-N-BE(2) cells, which do not produce ROS in response to insulin, are insulin insensitive and demonstrate a lack of tyrosine phosphorylation of their insulin receptor and its substrates upon insulin stimulation [46][47][48]. Insulin stimulation of primary cerebellar granule neurons (CGN) was shown to trigger insulin receptor autophosphorylation, which was accompanied by the immediate release of H 2 O 2 [53]. Fig. (1) shows that the duration of the insulin-induced H 2 O 2 spike in CGN was less than 30 s. The average quantity of hydrogen peroxide per spike per µg of protein was 2.5 ± 0.2 nM, a one-order of magnitude greater than the baseline H 2 O 2 production in non-stimulated neurons (0.20 ± 0.01 nM; p < 0.05). This insulin-induced H 2 O 2 was shown to be important for the activation of insulin receptors, since the H 2 O 2 scavenger N-acetylcysteine (NAC) completely abolished insulin receptor autophosphorylation in CGN (Fig. 2).
Two sources of insulin-induced H 2 O 2 in cells have been identified to date, a membrane-bound NADPH oxidase in adipocytes [4][5][6] and mitochondrion in tissues rich in mitochondria (liver, heart and neurons). In neuronal tissue, insulin-stimulated H 2 O 2 production has been shown to be sensitive to selective inhibitors of mitochondrial ROS production (malonate and FCCP) [53] as well as to diphenyleneiodonium (DPI) [49], a non-selective inhibitor of mitochondrial H 2 O 2 production during reverse electron transport at mitochondrial complex II [50][51][52]. It should be noted that although DPI is frequently used as an inhibitor of stimulus-induced H 2 O 2 production, it does not discriminate between mitochondria and NADPH oxidases as sources of H 2 O 2 production, since DPI is a non-selective inhibitor of both. Fig. (1). Time-dependence of insulin-induced H 2 O 2 generation in CGN cultures. CGN were exposed to insulin 100 nM (•) or medium (○) at zero time. H 2 O 2 release was assayed with Amplex Red as described in [53].
Given that insulin-stimulated insulin receptor autophosphorylation is sensitive to inhibitors of mitochondrial ROS production, the process of mitochondrial superoxide/ H 2 O 2 generation is briefly summarised below.

MITOCHONDRIA AS H 2 O 2 SOURCE
Mitochondria are considered to be the main quantitative source of superoxide anion radical and H 2 O 2 in mammalian cells [54]. The superoxide is a primary mitochondrial ROS, which undergoes rapid stoichiometric dismutation to H 2 O 2 in the presence of cytoplasmic (SOD1) or mitochondrial (SOD2) superoxide dismutase. Since the rate of nonenzymatic dismutation of superoxide to H 2 O 2 is three orders of magnitude lower than the enzymatic one, SOD1 and SOD2 are critical components of signalling pathways where H 2 O 2 is a secondary messenger.

Fig. (2).
Dependence of insulin-stimulated insulin receptor autophosphorylation on NAC concentrations in CGN cultures. CGN were exposed to insulin 100 nM and increasing NAC concentrations.
Receptor autophosphorylation was assayed as described in [53].
It is generally accepted that much of the superoxide generated by these organelles results from leakage of electrons in two components of the mitochondrial electron transport chain: complex I (NADH-ubiquinone oxidoreductase) and complex III (ubiquinol-cytochrome c oxidoreductase) [55]. Complex I provides the highest rates of superoxide generation in isolated mitochondria from the brain, heart and muscle during the reverse electron flow from succinate-ubiquinone oxidoreductase (complex II). Rates of superoxide generation at complex III are considerably lower [52,[56][57][58][59]. Complex I releases superoxide into the mitochondrial matrix, where it is transformed stoichiometrically into H 2 O 2 by superoxide dismutase SOD2 (Mn-SOD) [54,[58][59][60][61]. As the reactivity of Mn-SOD is high (rate constant of 2 x 10 9 M -1 s -1 ) and its concentration in the mitochondria is 10 µM, the steady-state concentration of mitochondrial superoxide is very low [62,63]. Although superoxide anion radical itself cannot diffuse through the mitochondrial inner membrane, being a charged molecule at physiological pH, its neutral metabolite H 2 O 2 readily passes through the mitochondrial membranes into the cytoplasm. Therefore, the major signalling form of mitochondrial ROS is H 2 O 2 .
The majority of mitochondrial superoxide is produced by complex I during reverse electron transport (RET). RET occurs in the presence of a significant proton-motive force, when electron flow from complex II gives electrons to coenzyme Q, which, in turn, gives electrons to complex I to produce the superoxide. Mitochondria respiring on the complex II substrate succinate display the highest rates of RET-associated H 2 O 2 production [56,57,64]. This H 2 O 2 generation is extremely sensitive to changes in the protonmotive force. When respiration is coupled to ATP synthesis, mitochondria produce very low amounts of H 2 O 2 . However, even a small increase in the proton-motive force above a threshold value slightly exceeding the value in the active (phosphorylating) state gives rise to a very steep increase in H 2 O 2 production [56,65,66]. Depletion of the proton-motive force by the addition of uncouplers, inhibitors or ADP completely abolishes H 2 O 2 generation [56,57,67].
Several lines of evidence suggest that the RET-associated H 2 O 2 production in isolated mitochondria is regulated by the activity of complex II. During state 4 respiration, when complex II has the highest succinate dehydrogenase activity [68], the rate of H 2 O 2 generation is also the highest [56,65]. In the transition to state 3, when complex II is rapidly deactivated [59], H 2 O 2 generation is abolished [56,65]. Factors deactivating complex II (e.g., uncouplers, inhibitors and ADP) also inhibit RET-associated H 2 O 2 generation [68]. Malonate, the classic complex II inhibitor, dose-dependently slows the rates of RET and rotenone-sensitive superoxide production at complex I [69]. Succinate, the complex II substrate, promotes superoxide production at complex I [70]. The rate of H 2 O 2 production in isolated mitochondria depends on succinate concentrations and is in excellent compliance with Michaelis-Menten kinetics, meaning that the succinate dehydrogenase reaction is the rate-limiting step in RET-associated H 2 O 2 generation [71]. In view of all this evidence, complex II seems to be a key regulatory point, which controls rates of mitochondrial H 2 O 2 production through varying the rate of succinate oxidation.
Complex II activity is tightly regulated by a reversible oxidation-reduction reaction. Oxaloacetate (OAA), an intermediate of the citric acid cycle and an inhibitor of succinate dehydrogenase [72], binds to the oxidised form of complex II with an affinity at least one order of magnitude greater than to the reduced form [73]. Therefore, complex II is inhibited by oxaloacetate upon oxidation and becomes activated upon reduction, when it liberates oxaloacetate [73]. Consequently, the enzyme is obtained largely in the deactivated form, containing tightly bound oxaloacetate in a 1:1 ratio to the enzyme [74]. Ubiquinol, a reduced form of coenzyme Q, is a physiologically relevant reductant and activator of complex II [75,76]. The highest succinate dehydrogenase activity is observed in state 4 of respiration, when coenzyme Q is largely in the ubiquinol form and capable of reactivating complex II [58]. In the transition to state 3, when the ubiquinol pool is depleted via oxidation in the Q cycle, the rapid oxidative deactivation of complex II occurs [58,75]. Micromolar H 2 O 2 reversibly inhibits complex II activity by enhancing oxaloacetate-mediated inactivation of complex II [77,78]. This inhibition has been observed in intact mitochondria [77,78] and synaptosomes [79].
As a summary, the succinate dehydrogenase reaction plays a key role in mitochondrial superoxide/H 2 O 2 formation. There is a switch-like dependence of the rate of succinatesupported H 2 O 2 generation on the mitochondrial membrane potential and even a small decrease in the potential below the threshold value completely abolishes H 2 O 2 release by the mitochondria.

MITOCHONDRIA AS SOURCE OF INSULIN-INDUCED H 2 O 2
Several lines of evidence suggest that the mode by which insulin stimulates H 2 O 2 generation in mitochondrion-rich tissues is by varying the rate of succinate oxidation. Insulin stimulation triggers almost immediate H 2 O 2 release in the liver, heart and neurons, which is completely inhibited by malonate, a classic mitochondrial complex II inhibitor [53,80]. A dramatic transient increase in the rates of succinatesupported H 2 O 2 generation was observed in mitochondria isolated from tissues (liver and heart) pre-treated for 1 min with physiological insulin concentrations, as compared to mitochondria isolated from non-stimulated tissues (control) [71]. In terms of Michaelis-Menten kinetics, the insulin pretreatment resulted in a 3-to 4-fold increase in Vmax and a 2to 4-fold decrease in Km to a value of 2 to 9 µM succinate. For reference, succinate levels in human plasma vary from 1 to 9 µM at rest and increase to up to 125 µM in hypoxic conditions [81,82]. Therefore, a significant increase in insulin-induced H 2 O 2 generation can be achieved under steady-state physiological succinate levels through the acute increase in the rate of succinate oxidation at complex II. This is in full agreement with previous data related to insulinstimulated oxidation of the isotope-labelled [2,3-14 C] and [1,4-14 C] succinates in rat liver and muscle cells, where insulin stimulation acutely increased only the rates of mitochondrial [2,3-14 C]succinate oxidation, which was observed as early as within 30 s [83].
Mitochondria are an insulin-sensitive H 2 O 2 source involved in insulin-stimulated receptor autophosphorylation in neurons [53]. Malonate, the specific inhibitor of mitochondrial complex II, inhibits H 2 O 2 release and insulin receptor autophosphorylation in neurons stimulated with insulin. Succinate, a substrate of complex II, enhances insulin receptor autophosphorylation stimulated with suboptimal insulin concentrations. The uncoupler FCCP, a mitochondrion-depolarising agent and inhibitor of mitochondrial ROS production, inhibits insulin-induced H 2 O 2 release and insulin receptor autophosphorylation. Hence, insulin-induced H 2 O 2 generation in neurons is succinate-dependent and sensitive to uncoupler-induced mitochondrial depolarisation.
Altogether, these findings suggest that insulin stimulates the well-known process of succinate-dependent mitochondrial H 2 O 2 production, while components transducing the signal between the insulin receptor and mitochondria remain to be elucidated. Fig. (3) illustrates three modes of operation of complex II. In mode 1, when mitochondria actively produce ATP and the ubiquinol-to-ubiquinone ratio UQH 2 /UQ is low, complex II is inhibited by binding with oxaloacetate and the mitochondria does not produce H 2 O 2 or does but at low rates. In mode 2, when there is no ATP production and there is a high ubiquinol-to-ubiquinone ratio and high proton-motive force, complex II is de-inhibited and gives electrons to complex I to produce the superoxide. The superoxide undergoes a dismutation reaction with mitochondrial Mn-SOD to produce H 2 O 2 , which passes through the mitochondrial membranes into the cytoplasm. In this mode, mitochondria produce H 2 O 2 continuously at high rates. In mode 3, upon insulin stimulation, complex II becomes activated and provides succinate oxidation at rates of about one-order of magnitude higher than that in mode 2. In this mode, the mitochondria produce H 2 O 2 at the highest rates, but for a short time (seconds). It is possible to draw some tentative conclusions regarding a mode of cooperation between insulin receptor and mitochondria during insulin receptor activation (Fig. 4).
Insulin binding to the cognate receptor stimulates almost the immediate release of an H 2 O 2 spike from the mitochondria, which, in turn, upregulates autophosphorylation of the insulin receptor and activation of the receptor tyrosine kinase. This double positive regulatory feedback loop seems to involve the H 2 O 2 -mediated inhibition of protein tyrosine phosphatases, which otherwise dephosphorylate the insulin receptor and, thus, prevent full activation of the insulin receptor tyrosine kinase. Upon activation, the insulin receptor kinase triggers downstream signalling. Therefore, the insulin-induced H 2 O 2 signalling pathway plays a role in enhancing very early insulin receptor signalling. activation. In this view, the possible relationships between mitochondrial dysfunction and insulin receptor signalling and between insulin-induced H 2 O 2 and oxidative stress are discussed below. Fig. (4). Scheme of the cooperation between mitochondria and the insulin receptor during insulin receptor activation in neurons.

RELATIONSHIPS BETWEEN INSULIN-INDICED H 2 O 2 AND MITOCHONDRIAL DYSFUNCTION
Since mitochondrial H 2 O 2 is required for insulin receptor activation, it is obvious that some alterations in mitochondrial function could result in a malfunction in insulin receptor activation and insulin resistance. Requirements that must be met to trigger insulin-induced H 2 O 2 signalling include a high proton-motive force, a high ubiquinol-to-ubiquinone ratio and a complete conversion of ADP to ATP. Thus, it is clear that only fully functional mitochondria may serve as the insulin-sensitive source of H 2 O 2 and support insulin receptor activation. This provides a link between mitochondrial dysfunction and insulin resistance that is highly prevalent in ageing and Alzheimer's disease (AD). Chronological ageing is accompanied by a progressive decline in brain mitochondrial functions such as respiration with the complex I substrate NADH, enzymatic activity of complex I and complex IV, and ATP production [117]. Along with this, brain mitochondria are chronically depolarised in senescence [118,119]. Since age is the major risk factor for AD, the aging process itself plays an important role in promoting mitochondrial dysfunction in the brain. Indeed, mitochondrial dysfunction is one of the earliest and most prominent features in sporadic age-associated AD, as has been highlighted in excellent reviews [120][121][122]. Therefore, the age-related decline in mitochondrial functions seems to disrupt insulin receptor activation in neurons and lead to the development of cerebral insulin resistance in old age.

RELATIONSHIPS BETWEEN INSULIN-INDUCED H 2 O 2 AND OXIDATIVE STRESS
Oxidative stress is defined as an imbalance between oxidants and antioxidants in favour of the oxidants, potentially leading to damage [84]. Insulin-stimulated H 2 O 2 generation, which is low and short-term, does not contribute to significant oxidative damage in cells. As the threshold level for H 2 O 2 neurotoxicity in cerebellar granule neurons is 10 µM [85], the insulin-induced H 2 O 2 production of about 100 nM [53] is too low to induce apoptosis. The reaction of H 2 O 2 with ferrous iron (Fe 2+ ), which generates hydroxyl radicals suspected of being involved in protein carbonylation, lipid peroxidation and DNA oxidation [86], is too slow to occur, compared to the much higher rates of the competitive reactions of H 2 O 2 removal via the peroxiredoxin-thioredoxin system [87]. Indeed, a calculated half-time for the degradation of 100 nM H 2 O 2 with ferrous iron is hours, when the iron brain concentrations are as high as 300 µM [88] and the second-order rate constant is 4400 M -1 s -1 [89]. Therefore, insulin-stimulated H 2 O 2 production is far too low to contribute to oxidative damage in neurons.
On the contrary, there is evidence suggesting that oxidative stress may directly affect insulin receptor signalling. Micromolar H 2 O 2 inhibits insulin receptor autophosphorylation [90,91], insulin binding to the receptor, insulin-induced tyrosine phosphorylation of the receptor substrates and downstream signalling [90].

RELATIONSHIPS BETWEEN INSULIN-INDUCED H 2 O 2 AND ANTIOXIDANT SYSTEMS
Insulin-induced H 2 O 2 signalling is downregulated by intracellular antioxidants. There are several antioxidant systems that have been shown to be effective in scavenging intracellular H 2 O 2 . Among them, the peroxiredoxinthioredoxin and glutathione peroxidase-glutathione systems are the most important on the basis of competitive kinetic analysis. Peroxiredoxins (Prx) and glutathione peroxidase (Gpx1) were revealed to be the primary H 2 O 2 scavengers responsible for the majority of hydrogen peroxide metabolism, according to kinetic estimations based on published rate constants and abundance data [92,93]. Gpx1 is a selenoprotein that catalyses the reaction of glutathione and H 2 O 2 at a second-order rate constant of 6 x 10 7 M -1 s -1 [92]. The oxidised glutathione is then reduced by NADPH with glutathione reductase to complete the cycle. Peroxiredoxin family members (Prx1 through to Prx6) are thiol-dependent non-catalytic peroxidases that scavenge H 2 O 2 directly at a second-order rate constant of about 10 7 M -1 s -1 [94]. The oxidised peroxiredoxin is reduced by thioredoxin, which is in turn reduced by NADPHthioredoxin reductase. Therefore, the glutathione peroxidaseglutathione and peroxiredoxin-thioredoxin systems are channels that transfer reducing equivalents from NADPH to H 2 O 2 with the highest efficacy compared to other cellular antioxidant systems. The replenishment of the NADPH pool depends basically on the activity of glucose-6phosphate dehydrogenase, a rate-limiting enzyme in the pentose phosphate pathway, which supplies reducing equivalents to cells [95].

Insulin Insulin
Active Kinase P P P P P P Some data suggest strong ultrasensitivity of insulininduced H 2 O 2 signalling to the activity of antioxidant systems. Fig. (2) shows the dependence of the insulin-stimulated insulin receptor autophosphorylation on pre-treatment with N-acetylcysteine (NAC). NAC is a thiol compound, which is widely used for modelling antioxidant effects in in vitro experiments. Although the direct non-enzymatic reaction of NAC with H 2 O 2 is slow [96], NAC is at least three-orders of magnitude more effective in the enzymatic reaction with H 2 O 2 catalysed by glutathione peroxidase [97]. Thus, NAC is an artificial molecule that mimics glutathione in the glutathione peroxidase reaction. Even a small increase in NAC levels above a threshold value completely prevents insulin-stimulated insulin receptor activation. The curve fitting provides a Hill slope of 7.8, which indicates strong ultrasensitivity of insulin receptor autophosphorylation to intracellular antioxidant activity. Therefore, intracellular antioxidant systems represent an effective barrier for insulininduced H 2 O 2 signalling, which is operated in an "all-ornone" mode depending on the antioxidant activity. It can be hypothesised then that excessive activation of antioxidant systems could induce dysfunctional activation of insulin receptors and consequently, insulin resistance. Some evidence obtained from animal studies supports this idea. Mice overexpressing glutathione peroxidase, Gpx1, developed insulin resistance associated with hyperinsulinemia, hyperglycemia, obesity and a 70% reduction in insulinstimulated phosphorylation of insulin receptors, compared to wild type control mice [98]. On the contrary, mice lacking Gpx1 were protected from high-fat diet-induced insulin resistance, while administration of NAC rendered them more insulin-resistant and increased fasting glucose levels in the blood [99].
AD is a progressive neurodegenerative disease, which is commonly characterised by irregular cerebral insulin receptor signalling and insulin resistance [100][101][102][103] and elevated antioxidant defense, possible to compensate for oxidative stress. Oxidative stress is regarded as a primary progenitor of AD because oxidative markers appear prior to beta-amyloid senile plaques [104][105][106]. The compensatory upregulation of antioxidant systems has been found in AD. Activity of glucose-6-phosphate dehydrogenase is increased in AD brains as a response to oxidative stress [107][108][109]. A significant overexpression of glutathione peroxidase has been found in the AD hippocampus [110,111]. Moreover, several studies have demonstrated the upregulation of peroxiredoxins in AD brains. Protein levels of Prx1 and Prx2 were significantly increased in AD brains than in agematched control [112]. Prx1 was overexpressed in the human AD cortex [113]. Prx2 was significantly increased in the frontal cortex [114] and hippocampus of AD brains [115]. Prx6 was markedly elevated in astrocytes in many regions of AD brains [116]. Altogether, these findings indicate that oxidative stress provokes a significant increase in the capacity of antioxidant systems to metabolise H 2 O 2 in AD. In view of the ultrasensitivity of insulin receptor autophosphorylation to antioxidant activity in neurons, the elevated activity of the antioxidant systems in AD may contribute to dysfunctional insulin receptor activation and central insulin resistance.

MITOCHONDRIAL BIOGENESIS ACTIVATION AND TREATMENT OF INSULIN RESISTANCE
As insulin-induced H 2 O 2 signalling requires fully functional mitochondria, pharmacological strategies based on activating mitochondria biogenesis in the brain seems to be central to the treatment of diseases resulting from dysfunctional insulin receptor signalling such as cognitive impairments and AD. Some evidence supports this idea. It has been postulated that peroxisome proliferator-activated receptor-gamma agonists (PPARγ) may enhance cognition in AD patients by improving mitochondrial function [123]. Rosiglitazone, a PPARγ agonist and stimulator of mitochondrial biogenesis in the brain [124], restores insulin responsiveness and rescues behavioural deficits in Tg2576 transgenic mice [125]. In a small clinical study, 6 months of rosiglitazone treatment resulted in enhanced memory and cognitive function in AD patients compared to the placebotreated control [126]. In a phase II clinical trial, involving over five hundred patients with mild to moderate AD, rosiglitazone treatment enhanced attention and memory, while patients with apolipoprotein E4 allele did not respond to the therapy [127].

CONCLUSIONS
Insulin-induced H 2 O 2 signalling establishes a bridge between insulin receptors and mitochondrial function in neurons. Although the insulin-induced H 2 O 2 spike is short-term, it has unexpectedly high significance for the autophosphorylation of insulin receptors, the critical posttranslational modification leading to the activation of the receptor tyrosine kinase. The set of requirements that must be met to trigger insulin-induced H 2 O 2 generation indicates that only fully functional mitochondria may serve as the insulin-sensitive source of H 2 O 2 . It provides a direct molecular link between mitochondrial dysfunction and derangements in insulin receptor signalling and insulin resistance. Therefore, an accumulation of dysfunctional mitochondria during chronological ageing is a risk factor that contributes to the development of age-related diseases known to accompany dysfunctional cerebral insulin receptor signalling, e.g., age-associated cognitive deficits and Alzheimer's disease.
Compensatory upregulation of antioxidant enzymes under oxidative stress is another risk factor contributing to the development of insulin resistance in view of the high significance of insulin-induced H 2 O 2 for insulin signalling. As intracellular antioxidant defense under oxidative stress is calibrated against elevated oxidant levels, this higher level of defense becomes a barrier for insulin-induced H 2 O 2 signalling and prevents insulin receptor activation. Since insulin receptor autophosphorylation is ultrasensitive to changes in the activity of antioxidant systems, the compensatory upregulation of antioxidant enzymes under oxidative stress seems to be a key factor contributing to the development of insulin resistance in neurons.
As insulin-induced H 2 O 2 signalling requires fully functional mitochondria, pharmacological strategies based on activating mitochondria biogenesis and/or mitochondria turnover in the brain are central to the treatment of diseases resulting from dysfunctional insulin receptor signalling in this organ.

CONFLICT OF INTEREST
The author confirms that this article content has no conflict of interest.