Despite the wealth of pre-clinical support for a role for reactive oxygen and nitrogen species (ROS/RNS) in the aetiology of diabetic complications, enthusiasm for antioxidant therapeutic approaches has been dampened by less favourable outcomes in large clinical trials. This has necessitated a re-evaluation of pre-clinical evidence and a more rational approach to antioxidant therapy. The present review considers current evidence, from both pre-clinical and clinical studies, to address the benefits of antioxidant therapy. The main focus of the present review is on the effects of direct targeting of ROS-producing enzymes, the bolstering of antioxidant defences and mechanisms to improve nitric oxide availability. Current evidence suggests that a more nuanced approach to antioxidant therapy is more likely to yield positive reductions in end-organ injury, with considerations required for the types of ROS/RNS involved, the timing and dosage of antioxidant therapy, and the selective targeting of cell populations. This is likely to influence future strategies to lessen the burden of diabetic complications such as diabetes-associated atherosclerosis, diabetic nephropathy and diabetic retinopathy.
The pre-clinical evidence in support of a role for reactive oxygen and nitrogen species (ROS/RNS) in diabetic complications such as atherosclerosis remains steadfast . Investigations of numerous subcellular sources, as divergent as leakage from the mitochondrial electron transport chain , up-regulation of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase enzymes  or the generation of ROS/RNS by myeloperoxidases , have provided ample evidence for an altered ROS/RNS balance, tipped in favour of ROS/RNS accumulation. In addition, evidence supports a diabetes-associated decline in the endogenous mechanisms designed to lower oxidative stress. This includes antioxidant enzymes such as the superoxide dismutases, glutathione peroxidases, peroxiredoxins and catalase enzymes , the manifestation of which can be assessed as increased ROS-/RNS-mediated damage to DNA, RNA, lipids and proteins. These ROS-/RNS-led alterations at the cellular and molecular level lead to both structural and functional changes that define diabetes-associated conditions such as endothelial dysfunction, end-stage renal disease, diabetic retinopathy and cardiovascular complications including atherosclerosis [6,7].
The clinical evidence in support of ROS-/RNS-driven diabetic injury relies on easily accessible tissue such as urine or blood, where associations between increased ROS-/RNS-mediated damage and disease have been established, e.g. patients with diabetes exhibit increased urinary F2-isoprostanes , whereas Gradinaru et al.  report significant oxidative and glycoxidative damage in blood samples when assessing circulating proteins from patients with pre-diabetes and Type 2 diabetes.
Lowering ROS/RNS, via either their direct scavenging or reducing their production at the source, provides two diverse ways in which pre-clinical studies have approached the problem of excess oxidative burden. Antioxidants have shown promise in pre-clinical studies, spanning a range of animal models [10–12], with most showing attenuation of end-organ injury, although in certain instances the end-goal of lowering cardiovascular burden has not always been achieved [13,14]. On the other hand, clinical trials of non-selective antioxidants such as vitamins A, C and E have mostly been disappointing, e.g. three large clinical studies, HOPE (Heart Outcomes Prevention Evaluation) , GISSI (Italian group for the study of the survival of myocardial infarction)  and more recently the Physician's Health Study II , failed to show benefit after administration of vitamins E and C, with GISSI and HOPE reporting, contrary to expectation, that vitamin E increased the risk of developing heart failure and heart anomalies [15,16]. A recent meta-analysis across 26 studies similarly failed to show any benefit of antioxidant treatments such as vitamins and minerals on cardiovascular disease (CVD), cancer or mortality in healthy individuals with no known nutritional deficits . It is therefore not surprising that the great disconnect between the mounting supportive pre-clinical data and the negative clinical trial outcomes has cast doubt over the clinical relevance of antioxidant therapy. Indeed, questions have been raised about whether oxidative injury is causal or consequential, leaving little room for conventional antioxidant therapy at a time when diabetic complications are more prevalent than ever, significantly affecting the medical and economic burden of both developed and developing nations.
This is where a more rational approach to antioxidant therapy is needed. The use of vitamins and minerals in clinical trials now seems somewhat naïve given the recent pre-clinical advances being made in ROS/RNS involvement in CVD processes . It is now unanimously accepted that ROS/RNS differ in terms of their reaction kinetics, reaction preferences, diffusion parameters, sites of production and degradation kinetics. Similarly, not all antioxidants should be viewed as having the same properties. Indeed, vitamins C and E have opposing characteristics, with preferential partitioning into lipophilic and lipophobic environments, respectively. This is critical for substrate preference and may preclude the targeting of critical ROS/RNS species . Furthermore, an important consideration is the fact that the two-electron reactive ROS, hydrogen peroxide (H2O2), is now considered a significant ROS in terms of mediating both cellular signalling and injury. Lessening proatherogenic levels of this ROS, through the use of vitamins with their one-electron way of working for ROS removal, is therefore less likely to be effective  and alternative approaches are needed. Thus, based on recent advances in the understanding of radical biochemistry, this disconnect between pre-clinical and clinical data needs careful re-evaluation. The present review focuses on several key enzymatic systems that produce or eliminate ROS/RNS, highlighting a need for more targeted antioxidant approaches to reduce ROS/RNS, thereby drawing attention to novel antioxidant strategies for diabetic complications.
DIABETES AND ATHEROSCLEROSIS: PATHOLOGICAL MECHANISMS INVOLVING OXIDATIVE STRESS
Individuals with diabetes are at increased risk of developing macrovascular complications including atherosclerosis as well as cardiovascular events such as heart attacks and strokes . An inflammatory basis for atherosclerosis , centred around endothelial cell damage and immune cell infiltration , has been actively investigated and refined over the past 10–15 years. Intimately linked to these processes is a central role for ROS/RNS . Diabetic insult to the endothelium, whether as a consequence of the persistent hyperglycaemia or of hyperlipidaemia, triggers ROS/RNS production and reduces nitric oxide (NO) bioavailability, thereby up-regulating adhesion and chemotactic molecules to promote the recruitment and retention of inflammatory cells within the subintimal space [26,27]. In addition, smooth muscle cells (SMCs) from the medial layer undergo phenotypic modulation into highly migratory, proliferative and synthetic cells within the lesion (discussed in detail in Owens et al. ). A rich proinflammatory, pro-oxidative, profibrotic and prothrombotic tissue microenvironment ultimately stimulates growth and instability of the atherosclerotic plaque, predisposing to serious clinical outcomes such as myocardial infarction and stroke.
Much interest has focused on how diabetes promotes a proinflammatory state with particular attention on the heightened state of oxidative stress . One mechanism by which ROS/RNS triggers vascular inflammation is by increasing nuclear translocation of activated nuclear factor-κB (NF-κB) p65/p50 subunits in order to drive transcription of genes encoding adhesion, chemotactic and proinflammatory signalling molecules . Human aortic endothelial cells (HAECs) grown under hyperglycaemic conditions show increased expression of NF-κB-regulated genes, monocyte chemoattractant protein (MCP)-1 and vascular cell adhesion molecule (VCAM)-1, in an ROS-dependent manner . It is of interest that endothelial microparticles exposed to high glucose concentrations trigger ROS production and increase VCAM-1 as well as intercellular adhesion molecule (ICAM)-1 expression in target human coronary artery endothelial cells . In addition, advanced glycation end-products (AGEs) induce ROS production in HAECs via NF-κB-dependent and -independent mechanisms . Collectively, these studies highlight the pro-oxidant and proinflammatory effects of hyperglycaemia on the endothelium. In an effort to disrupt the feed-forward interaction between these atherogenic signalling pathways, current work is focused on identifying and inhibiting the primary source of ROS in the vascular wall.
Prominent role of Nox-derived ROS in atherogenic processes
Multiple enzymatic systems capable of generating ROS are up-regulated in the diseased vasculature including nicotinamide adenine dinucleotide phosphate [NAD(P)H] oxidase (Nox), nitric oxide synthase (NOS), myeloperoxidase (MPO) and xanthine oxidase (XO) . Studies of vascular endothelial cells have demonstrated the importance of Nox- and the mitochondrial electron transport chain-mediated generation of ROS in high-glucose- and AGE-mediated damage [30,34,35]. Across the vasculature, the Nox enzyme family is considered the major source of ROS  and a target for diabetes-associated vascular disease . In human atherosclerosis, intense superoxide production correlated with the increased expression of Nox-related proteins (gp91phox or Nox2, Nox1, Nox4 and p22phox), with Nox2 expressed mainly by macrophages and Nox4 also found in non-phagocytic vascular cells .
Recent evidence has identified a central role for Nox1 in accelerating atherosclerosis in the aorta of diabetic mice. In the study of Gray et al.  it was shown that high glucose levels selectively up-regulate the Nox1 isoform in cultured human endothelial cells. Importantly, diabetic apoprotein E (ApoE)/Nox1 double-knockout mice showed less plaque than diabetic ApoE knockout mice, which was accompanied by down-regulation of proinflammatory pathways. It is interesting that ApoE/Nox4 double-knockout mice failed to demonstrate reductions in lesion development, highlighting the central role played by Nox1 in proatherogenic processes . At present, from a therapeutic perspective, the targeting of Nox-specific isoforms in the vasculature is an area being actively pursued . Currently, there are no Nox isoform-specific inhibitors but use of the dual action Nox1/Nox4 inhibitor, GKT137831, reduced ROS formation and diabetes-associated atherosclerosis . Recently, an additional role has been ascribed to ebselen (2-phenyl-1,2-benzisoselenazol-3[2H]-one, discussed in more detail below) for its ability to disrupt assembly of some members of the Nox family of isoforms, namely Nox1 and Nox2. The use of ebselen has indeed been shown to reduce diabetes-associated atherosclerosis in diabetic ApoE knockout mice [40,41].
The role of Nox4 in vascular disease represents an area of active debate, with recent work by Schröder et al.  demonstrating protective functions of this isoform in models of angiotensin II (Ang II)-induced vascular hypertrophy and ischaemia-induced angiogenesis, potentially via hydrogen peroxide. Indeed, compared with aortas from wild-type mice, aortas from inducible Nox4-deficient animals showed increased inflammation and endothelial dysfunction. From a mechanistic perspective, loss of Nox4 appeared to cause endothelial dysfunction via pathways that included reductions in endothelial NOS (eNOS) and haem oxygenase-1 (HO-1), leading the authors to conclude that endogenous Nox4 protects the vasculature during ischaemic or inflammatory stress . Furthermore, Nox4 is known to promote the differentiation of many cell types, most probably through its function as a producer of hydrogen peroxide [43,44]. Therefore, clinical targeting of Nox4 may not be desirable in certain instances. Attention has also recently turned to the Nox5 isoform, present in humans and pigs but lacking in mice and rats, with initial evidence suggesting an important role for Nox5 in the propagation of aortic SMC proliferation . In human endothelial cells, vasoactive Ang II and endothelin (ET-1) were shown to regulate Nox5 through Ca2+/calmodulin-dependent processes, including selective activation of mitogen-activated protein kinases (MAPKs) which lead to differential Nox5-mediated functional responses by Ang II and ET-1 . However, the role of each Nox isoform in different vascular pathologies, particularly in the context of co-morbidities such as those often observed in patients with diabetes, still needs to be fully elucidated.
Diabetes preferentially activates inflammatory immune cells
The proinflammatory effects of diabetes include the aberrant activation of immune cells that are intimately involved in the development and progression of atherosclerosis. Evidence of T-cell and macrophage accumulation in atherosclerotic lesions has been well described in humans and mouse models of atherosclerosis [47,48]. T-cells isolated from patients with diabetes exhibit a more pronounced proinflammatory T-helper (Th)1 phenotype . More recently, the identification of pharmacological inhibition of Nox-derived ROS in T-cell-receptor activation [50,51] has highlighted the possibility of an immunomodulatory role for diabetes-induced oxidative stress. In support of this notion, we recently showed that inhibition of Nox-derived ROS/RNS attenuated T-cell activation and associated plaque development in diabetes .
Monocytes exposed to the diabetic milieu have also been shown to adopt an inflammatory phenotype . Plaque-derived macrophages from diabetic mice exhibit increased mitochondrial ROS production and proinflammatory gene expression, consistent with polarization to the M1 phenotypic state . Moreover, suppression of mitochondrial superoxide in bone marrow-derived macrophages, using transgenic mice that have a mitochondria-specific catalase, led to a significant reduction in inflammatory cell recruitment and overall reduction in lesion area . Similarly, deletion of the Nox subunit p47phox in monocytes/macrophages decreased atherosclerosis via a reduction in monocyte recruitment . Consistent with these findings, recent work by Padgett et al. , investigating macrophage redox status in the development of Type 1 diabetes, uncovered a novel mechanistic link between Nox-derived superoxide and macrophage M1/M2-like phenotypes, although CVD outcomes were not examined in that study. Increasing evidence supports a modulatory role for ROS in immune cell phenotypes, yet the exact mechanisms driving these proinflammatory changes in diabetes require further interrogation.
ROS are markers and mediators of the proatherogenic vascular SMC phenotype
Advanced atherosclerotic plaques are characterized by an accumulation of phenotypically modulated vascular SMCs (VSMCs). Diabetes markedly increases VSMC proliferation, migration and inflammatory responses, leading to accelerated plaque progression and instability [58–60]. In line with experimental data, a meta-analysis of studies conducted in patients with Type 2 diabetes concluded that significantly impaired vascular smooth muscle function is an important contributor to diabetes-associated vascular dysfunction . Indeed, high-glucose induction of ROS in VSMCs  can have adverse effects on VSMC phenotypes . Targeting Nox1 generation of ROS has been shown to abrogate agonist-induced migration [64,65], whereas inhibition of Nox enzymes with VAS2870 reversed hyperactive contractile responses in the diabetic aorta . These studies implicate Nox-derived ROS in VSMC pathophysiology relevant to CVD, warranting deeper investigation in the setting of diabetes.
Post-translational modifications of proteins in the diabetic milieu (including glycation and oxidation) trigger atherogenic responses in vascular cells. Emerging evidence from molecular studies has uncovered a complex regulatory role for protein modification by O-linked N-acetylglucosamine (O-GlcNAc; a process called O-GlcN-acylation) in cardiometabolic diseases, such as diabetes. Increased O-GlcN-acylation contributes to enhanced aortic calcification in diabetic mice together with up-regulated osteogenic gene transcription in primary mouse VSMCs . Accelerated medial calcification in diabetic rats was prevented by antioxidant treatment . Interactions between oxidative stress and O-GlcN-acylation of proteins have been demonstrated in cardiomyocytes after calcium overload  and in human aortic SMCs exposed to high glucose levels ; however, the functional implications of this interplay are not fully understood. Novel therapeutic approaches designed to decrease O-GlcNAc levels in diabetes are currently under investigation although the interplay between acute versus long-term O-GlcN-acylation, as seen during diabetes, needs further evaluation in light of the protective nature of acute modifications. Indeed, it may be the turnover of O-GlcN-acylation that plays a central role in the delicate regulation of the cardiovascular system .
Hyperglycaemia-induced epigenetic modifications of gene expression influence redox homoeostasis and inflammatory pathways in the vasculature
Chromatin modifications are increasingly explored as novel regulators of vascular cell pathology. It is now becoming even more appreciated that transient hyperglycaemia induces long-lasting activation of inflammatory and oxidative stress pathways via epigenetic regulation of the NF-κB subunit p65  and mitochondrial adaptor protein p66shc , respectively. Seminal studies by El-Osta and colleagues have shown persistent up-regulation and nuclear translocation of p65 in aortic endothelial cells after high glucose exposure mediated by the histone methyltransferase Set7 [74,75]. It is of interest that inhibition of mitochondrially derived superoxide prevented changes in Set7-mediated histone methylation and p65 gene expression . Recent examination of patients with diabetes validated Set7 as an adverse epigenetic signature of vascular cell inflammation, correlating with an increase in the oxidative stress marker 8-iso-prostaglandin 2α (8-isoPGF2α) . In addition, epigenetic regulation of p66shc in HAECs leads to persistent oxidative stress despite restoration of normoglycaemia . Together these studies highlight a critical role for ROS in epigenetic modulation of glucose-responsive pathways.
Novel strategies for targeting atherosclerosis in diabetes
Current therapeutic strategies have been successful in reducing but not totally removing the burden of CVD. Genome-wide profiling of chromatin modifications has identified adverse epigenetic signatures and their enzymatic writers as potential therapeutic targets, which may ultimately lead to more personalised therapeutic approaches. A better understanding of the altered protein biology in diabetes has also offered promising targets such as O-GlcN-acylated proteins for treating accelerated vascular disease. Additional therapeutic approaches include the attenuation of AGE formation , as well as the interruption of the production of inflammatory lipids [79,80]. Based on the vast amount of supportive pre-clinical data, oxidative stress is still an attractive target for the prevention and retardation of CVD . The central role played by Nox-derived ROS in multiple atherogenic pathways activated by diabetes confirms oxidative stress as a prime target for combating CVD. Indeed, novel Nox inhibitors, including GKT137831, have shown protective effects against diabetes-accelerated atherosclerosis [30,52]. Unlike previously untargeted antioxidant approaches, these Nox inhibitors represent a potentially promising area for anti-atherosclerosis pharmaceuticals. However, there are limitations, particularly with respect to the targeting of certain Nox isoforms such as Nox4. The current dual Nox1/Nox4 inhibitor, GKT137831, may not be suitable in situations where ROS such as hydrogen peroxide are required for functions that include stem cell differentiation into SMCs or cardiac cell differentiation [43,44].
DIABETIC NEPHROPATHY: PATHOLOGICAL MECHANISMS INVOLVING OXIDATIVE STRESS
Diabetic nephropathy (DN) is a major microvascular complication of diabetes, representing the most common cause of chronic kidney disease, leading to end-stage renal failure . The characteristic histopathological changes in DN include renal hypertrophy, defined as the development of thickened glomerular and tubular basement membranes, and the progressive accumulation of extracellular matrix (ECM) proteins in the glomerular mesangium and tubulointerstitium, resulting in glomerulosclerosis and tubulointerstitial fibrosis [83,84]. Functionally, DN is characterized by increasing albuminuria and an ensuing decline in glomerular filtration rate (GFR), leading to end-stage renal failure .
It is believed that the interactions between metabolic (hyperglycaemia, dyslipidaemia and impaired insulin signalling) and haemodynamic (systemic and intraglomerular hypertension, activation of the renin–angiotensin system) factors, as well as infiltration by inflammatory cells, elevated levels of growth factors and proinflammatory mediators, and activation of key intracellular signalling pathways and transcription factors, lead to the development and progression of diabetes-associated kidney disease [85,86]. A number of mechanisms have been proposed to explain the potential detrimental effects of hyperglycaemia-induced tissue damage, including flux through the polyol pathway, AGE formation, hexosamine pathway flux, activation of protein kinase C (PKC), elevations in Ang II and involvement of the Nox enzymes [83,86–88]. Importantly, oxidative stress is increasingly considered to be a major contributor to the development and progression of DN. A growing number of studies now support the view that the enhanced levels of ROS/RNS contribute to the inflammation and fibrosis of the kidney [87–89]. The elevated ROS/RNS levels most probably arise from diabetes-driven increase in ROS-producing enzymes of the kidney, but may also arise from defective responses  or down-regulation of endogenous antioxidant systems in the diabetic kidney .
Various renal sources of ROS have been suggested as relevant to the diabetic kidney, including enzymes of the mitochondrial respiratory chain, XO, uncoupled NOS and Nox enzymes [88,89]. Under physiological conditions, Nox enzymes have a very low constitutive activity in the kidney, but are highly up-regulated in disease states such as hypertension, hypercholesterolaemia and diabetes. Based on its high expression in diseased renal tissue, the Nox4 isoform of the Nox enzyme family was originally termed ‘Renox’, and was postulated to play a major role in the pathogenesis of DN [92–94].
Several studies have shown an up-regulation of Nox4 in the renal cortex of experimental models of DN. Furthermore, an increase in Nox4 expression was found in individual renal cells such as podocytes, mesangial cells, proximal and distal tubules in the presence of high glucose and after transforming growth factor β (TGF-β) or Ang II treatment [92,95–98]. It has been demonstrated by specifically targeting Nox4, through systemic administration of anti-sense oligonucleotides, that Nox4-derived ROS mediate renal hypertrophy and increase fibronectin expression in rats with streptozotocin-induced diabetes . Recently, the involvement of TGF-β and Ang II in the regulation of Nox4-derived ROS generation could be demonstrated in mouse podocytes and mesangial cells under high glucose conditions [95,98]. Furthermore, other studies have suggested that high glucose-mediated Nox4-derived ROS activate profibrotic processes via Nox4-sensitive, p38 MAPK-dependent and Akt phosphorylation pathways in mouse proximal and distal tubules [96,97,99]. Recent studies, in an experimental model of DN, have shown that genetic deletion and pharmacological inhibition of Nox4 resulted in renal protection from glomerular injury . This was evidenced by attenuated albuminuria, preserved renal structure, reduced glomerular accumulation of ECM proteins (collagen IV and fibronectin), attenuated glomerular macrophage infiltration, and reduced renal expression of inflammatory markers such as MCP-1 and NF-κB when Nox4 was reduced or absent. Despite these studies which strongly suggest a causal role for Nox4-derived ROS in DN, and therefore a potential target for the treatment of DN, one study in particular argues against an injurious role for Nox4. In the study of Babelova et al.  inducible Nox4 knockout animals exhibited worsened nephropathy in the STZ model of Type 1 diabetes, whereas in the unilateral ureteral ligation (UUO) model, renal expression of proinflammatory and profibrotic factors was elevated in these Nox4 knockout kidneys. Thus, the data of Babelova et al.  do not support the widely held view that Nox4 is deleterious in the diabetic kidney. It is clear that the renal role of Nox4 will need further studies to resolve these discrepancies.
Recent evidence also suggests a potential role for podocyte Nox5 in impaired renal function and hypertension. Indeed, when transgenic mice expressing human Nox5 in a podocyte-specific manner (Nox5pod+) were generated, Nox5pod+ mice exhibited early onset albuminuria, podocyte foot process effacement and elevated systolic blood pressure. Furthermore, subjecting Nox5pod+ mice to streptozotocin-induced diabetes further exacerbated these changes .
Despite overwhelming evidence in support of a role for ROS/RNS in diabetic nephropathy, a recent study by Dugan et al.  failed to show an elevation in the mitochondrial ROS, superoxide, in murine models of Type 1 diabetes. Reductions in mitochondrial ROS were linked to alterations in mitochondrial function via a pathway that includes 5′-AMP-activated protein kinase (AMPK). Mitochondrial biogenesis, pyruvate dehydrogenase activity and mitochondrial complex activity were rescued by treatment with an AMPK activator.
Agents targeting renal oxidative stress in DN
Several mechanism-based approaches have been examined to counteract oxidative stress-induced renal tissue damage, including the use of selective inhibitors of ROS enzymatic sources such as specific Nox inhibitors, as well as antioxidant supplements containing vitamins (e.g. vitamins E and C). A number of previous studies reported apocynin as a useful pharmacological inhibitor of NADPH oxidase. However, the specificity of apocynin towards the Nox isoforms is unclear. It has been shown that chronic inhibition of NADPH oxidase by apocynin in an experimental model of DN was associated with attenuation of plasma lipid peroxidation products, renal hydrogen peroxide production, mesangial matrix expansion and urinary protein excretion [104,105]. However, apocynin failed to inhibit superoxide generation in human embryonic kidney (HEK)-293 cells over-expressing Nox isoforms  and apocynin may have alternative or additional actions, including inhibiting rho kinase activity .
Recently, two groups of Nox-specific inhibitors, pyrazolopyridine compounds including GKT136901  and GKT137831  and trizolopyrimidine derivatives such as VAS2870 and VAS3947 [110,111], have been reported as potent compounds that inhibit the Nox isoforms. GKT136901 and GKT137831 are dual inhibitors for the Nox1 and Nox4 isoforms and are reported to inhibit Nox1-/Nox4-derived ROS production. GKT136901 has shown renoprotective effects in a mouse model of Type 2 diabetes . Furthermore, studies have demonstrated that administration of GKT137831 to an experimental model of DN replicated renoprotective effects to a similar degree to that of Nox4 deletion . Further validation for a protective role for GKT137831 comes from recent data in OVE26 mice, in which the Nox1/Nox4 inhibitor attenuated endpoints of DN. Importantly, this late-intervention study showed its potential clinical relevance in reversing established diabetic kidney disease because the drug was administered for 4 weeks to 4- to 5-month-old mice that had already developed renal injury . Indeed, this first-in-class, Nox1/Nox4, dual-specificity inhibitor has completed phase I safety trials and is now in phase II trials including clinical evaluation for DN (NCT02010242 on ClinicalTrials.gov). However, further refinement of the Nox inhibitors is probably warranted given the emerging evidence that the Nox isoforms are very tissue specific, with Nox4 predominantly expressed in the kidney. Hence, the development of isoform-specific Nox inhibitors is more likely to afford optimal protection in individual organs.
Despite satisfactory results from antioxidant therapy in experimental models of DN, vitamin supplementation has shown conflicting results to combat DN in patients with diabetes. A small cohort study has reported that high doses of vitamin E supplementation (1800 IU/day) showed a decrease in microalbuminuria and restoration of renal function in patients with Type 1 diabetes . Furthermore, supplementation of vitamin C resulted in some improvements in insulin action, glycaemic control and endothelial function in patients with diabetes . On the other hand, as discussed previously, a meta-analysis of clinical trials with a large number of participants failed to show any improvement in cardiovascular and cerebrovascular risk in patients with diabetes who were supplemented with high doses of vitamin E (16.5–2000 IU/day), but rather found an increase in all-cause mortality [15,116,117]. Therefore, the clinical relevance of vitamin E or C as antioxidant therapy for the treatment of DN remains controversial and is unlikely to be widely used.
A recent development has been the use of bardoxolone methyl, a nuclear factor erythroid 2-related factor 2 (Nrf2) transcription agonist and a regulator of the antioxidant response, to improve the GFR in patients with DN. Despite controversy, this drug class may offer a new approach to bolstering antioxidant defences in DN [118,119].
DIABETIC RETINOPATHY AND OXIDATIVE STRESS
Diabetic retinopathy (DR) is caused by progressive damage to the retina over decades of diabetes mellitus and represents a major health burden, with a staggering 93 million people estimated to have some form of the disease . DR is classified into an early stage of non-proliferative DR and a late stage of proliferative DR, according to the extent of microvascular damage in the retina. Vision loss in DR principally occurs from the breakdown of the blood–retinal barrier resulting in macular oedema, retinal detachment and intraocular haemorrhage. Neurons and glia make an important contribution to DR that is most probably due to the close anatomical arrangement of the vasculature, ganglion cells and glia [121,122]. Furthermore, DR is a state of subclinical chronic inflammation, whereby leukocytes migrate to the damaged retina and resident microglia/macrophages become activated to release injurious cytokines . Despite the progressive nature of the retinal injury that occurs in DR, there are no preventive treatments, with approaches such as laser surgery and anti-vascular endothelial growth factor (VEGF) agents applied in the end-stages of DR . Hence, understanding the early and underlying mechanisms involved in the development of DR is required to develop preventive medical treatments.
Although a number of factors contribute to DR, it is clear that excess ROS is an important underlying mechanism . In DR, hyperglycaemia induces excess ROS production via pathways such as glucose auto-oxidation, the polyol pathway, AGEs and activation of PKC, as well as the renin–angiotensin–aldosterone system [125,126]. Excess ROS damage retinal cells by stabilising the transcription factor, hypoxia-inducible factor-1α (HIF-1α), which leads to the increased expression of angiogenic and inflammatory mediators such as VEGF; these promote pathological angiogenesis and vascular leakage. VEGF may amplify the entire system with interactions between VEGF receptor 2 and Nox, leading to further production of ROS and activation of HIF-1α. Other amplifiers include the ROS-mediated induction of NF-κβ, which leads to the production of tumour necrosis factor α and subsequent generation of inflammatory factors such as interleukin-6, MCP-1 and ICAM-1 [126,127].
Agents targeting oxidative stress in DR
For many years, the development of ROS inhibitors for the treatment of DR has been of considerable interest. However, previous strategies such as antioxidant micronutrients have shown inconsistent results in human studies of DR, and may even act as pro-oxidants [128,129]. Attention then turned to the inhibition of ROS derived from Nox, because it was recognized that the primary function of Nox is the production of ROS. Numerous studies have evaluated the effects of purported global inhibitors of Nox such as apocynin in animal models of DR . However, these inhibitors are now identified as non-specific for Nox and not suitable for clinical translation . Increasing evidence indicates that individual isoforms of Nox influence organ pathology in a disease- and tissue-specific manner that may have relevance to DR [30,126,131]. Of the seven known isoforms of Nox, pre-clinical studies of DR have focused on Nox1, Nox2 and Nox4. In models of short-term diabetes, a genetic deficiency in Nox2 reduced retinal inflammation, which may be attributed to Nox2's essential role in phagocytic defence [132,133]. However, there is caution about targeting Nox2 as a medical treatment due to a deficiency in Nox2 being associated with severe infection and early death in humans . Furthermore, a recent study reported that long-term diabetes is lethal when induced in Nox2 knockout mice due to increased susceptibility to infection . Although the effects of Nox4 inhibition in DR have not been extensively studied, Nox4 siRNA attenuated retinal vascular leakage in diabetic animals .
Studies of the proliferative stage of DR are not possible in rodent models of diabetes because the retinal lesion does not progress to retinal neovascularization, and hence the oxygen-induced retinopathy model is often utilized. In a parallel study of Nox1, Nox2 and Nox4 knockout mice, the Nox1 isoform was reported to be the predominant Nox isoform involved in retinal neovascularization . Nox1 knockout mice exhibited reduced vasculopathy, as well as microglial density and leukocyte adherence to the vasculature . There is also evidence that Nox4 may be involved in oxygen-induced retinopathy, with increased Nox4 expression in the retina  and co-localization with retinal vascular endothelial cells . Another consideration is the role of Nox5 in DR. Nox1, Nox2, Nox4 and Nox5 are all expressed in the human retina as well as retinal endothelial cells and pericytes . This information, together with the emerging role of Nox5 in vascular injury, suggests that Nox5 may be relevant to microvascular disease in DR .
As mentioned earlier an exciting new direction in Nox research is the development of pharmacological inhibitors of Nox1/Nox4 isoforms that are now in clinical development. Recent studies indicate that the benefits of GKT137831 extend to ischaemic retinopathy with reductions in capillary degeneration and retinal neovascularization . Other approaches focusing on oxidative stress may also be relevant to DR and include boosting antioxidant defence mechanisms, e.g. increasing glutathione peroxidase 1 (GPx1) is likely to improve retinal damage, with a recent study showing that a deficiency in this key antioxidant enzyme exacerbates capillary degeneration and neovascularization in the retina of mice . Enhancement of Nrf2 has also emerged as a potential treatment target due to its protective effects in animal models of DR [141–143]. Overall, the imbalance between pro-oxidative and anti-oxidative mechanisms is likely to be critical in the pathogenesis of retinal damage in DR.
TARGETING ANTIOXIDANT DEFENCES: BOLSTERING KEY ENZYMES
To counterbalance the ROS-producing enzymes, an extensive antioxidant defence mechanism regulates redox flux and lessens oxidative burden. Several classes of enzymes target specific ROS/RNS, with preferences for distinct cell types and certain subcellular localizations . These highly specialized enzymes perform vital physiological functions, enabling certain ROS to drive cell signalling, e.g. a recent mechanism describes the formation of hydrogen peroxide gradients that are set up by the interplay of the antioxidant enzyme, peroxiredoxin 1, the Src tyrosine kinase and protein tyrosine phosphatases, which restrict hydrogen peroxide to the signalling site, thereby facilitating signal transduction from an extracellular ligand to intracellular compartments [144,145]. Yet it is in their capacity to lessen the ROS/RNS burden that antioxidants play a critical role in decreasing pathophysiological injury. The strongest evidence of a critical role for antioxidant defence in the protection against complications associated with diabetes comes from the analysis of both antioxidant-over-expressing and knockout mouse models. These approaches have not only enabled the identification of important enzymatic antioxidants but also shed light on the critical ROS/RNS levels that drive the pathogenesis of diabetic complications.
As detailed previously, ROS such as the superoxide radical (O2•−), which are generated by the hyperglycaemic/hyperlipidaemic milieu of diabetes, are counterbalanced in a first step of the antioxidant pathway by the superoxide dismutase (Sod) family of isoenzymes (Figure 1). The involvement of the cytosolic copper/zinc isoform, Sod1, in the protection against diabetic complications was demonstrated in diabetic Sod1 knockout mice which displayed marked structural damage (increased glomerular matrix proteins), oxidative damage (an increase in nitrotyrosine, and higher glomerular O2•− content) and functional injury (increased albuminuria) after 5 weeks of diabetes compared with diabetic wild-type mice . After 5 months, mesangial matrix expansion, renal cortical malondialdehyde content and severity of tubulointerstitial injury were all significantly increased . These data suggest that the increased build-up of O2•− interacts with NO to increase peroxynitrite-mediated injury, which is detected as increased nitrotyrosine in proteins. With respect to bolstering cytosolic Sod1 activity as a preventive strategy, surprisingly in ApoE knockout mice over-expressing Sod1, this strategy failed to prevent the development of atherosclerosis, in association with no reduction in markers of oxidative stress such as F2-isoprostanes . One explanation might be the second-step dismutation of O2•− into increased levels of hydrogen peroxide as a consequence of the diabetes-driven elevations in the substrate O2•− and over-expression of Sod1 (Figure 1). Indeed, as discussed by Yang et al. , it may therefore be more important to target the removal of hydrogen peroxide than O2•− in the prevention of atherosclerosis. This view is supported by experiments assessing the atherosclerosis burden in ApoE knockout mice over-expressing catalase, which catalyses the removal of hydrogen peroxide as water, where this strategy was found to lessen atherosclerotic plaque . However, it cannot be ruled out that elevation in mitochondrial O2•− levels may be more injurious than increases in cytosolic O2•− levels. These ROS would not have been removed in the ApoE/Sod1 transgenic mouse model. It may be particularly important to lessen mitochondrial O2•− because it is now recognized that mitochondrial ROS mediate crosstalk with other sources of ROS such as the cytosolic Nox enzymes, setting up a vicious cycle of ROS production [148,149]. In addition, this may lead to the uncoupling of eNOS [148,149]. Therefore therapeutic targeting to bolster the mitochondrial antioxidant capacity to remove mitochondrial O2•− through use of mitochondrially targeted antioxidants may be a better strategy to break this cycle, thereby inhibiting ROS production by the mitochondria and reducing Nox activity.
Diabetes-induced increases in ROS/RNS and their removal by enzymatic antioxidants of the antioxidant pathway
The antioxidant enzymes GPx1, catalase and the peroxiredoxins remove cellular hydrogen peroxide. This extensive array of specialised enzymes points to an important role for hydrogen peroxide in the maintenance of cellular homoeostasis. Although low levels of hydrogen peroxide are now considered significant in physiological processes , e.g. for angiogenesis , higher levels drive pathophysiological processes, e.g. in the propagation of proinflammatory and proatherogenic signalling [152,153]. This hormetic role for hydrogen peroxide means that fine-tuning its removal becomes critical in the protection against pathological injury while at the same time maintaining sufficient levels to drive cellular processes. Regulation of the endogenous hydrogen peroxide-neutralising enzymes appears to be a key mechanism in maintaining this balance.
Data now point to a pivotal role for GPx1 in the protection against diabetes-driven oxidative stress as well as proinflammatory and profibrotic mediators [41,154]. Indeed, comparisons between diabetic GPx1/ApoE double-knockout mice and diabetic ApoE knockout mice demonstrated that lack of GPx1 greatly accelerated end-points of atherosclerosis and kidney injury in diabetic GPx1/ApoE double-knockout mice [41,154]. The GPx1 selenium-dependent isoform is the most abundant of the GPx isoforms and is ubiquitously expressed in all cell types; it is mainly localized to the cytosol and mitochondria [155,156]. Similarly, Torzewski et al.  demonstrated an important role for GPx1 in lessening atherosclerosis in an environment of enhanced lipids. These pre-clinical data were shown against a back-drop of strong evidence for an important role for GPx1 in protection against the advancement of atherosclerosis in the clinic. Indeed, polymorphisms identified within the GPx1 gene (P198L and P197L), resulting in reduced GPx1 activity, have been linked with increased intima–media thickness of carotid arteries and an increased risk of CVD and peripheral vascular disease in patients with Type 2 diabetes . A recent meta-analysis of observational data confirms this strong association of GPx1 mutations and CVD across 2 cohort studies and 8 case–control studies involving 1430 cases and 3767 controls .
Additional studies suggest a protective role for GPx1 in the atherogenic process itself, e.g. reductions in red blood cell GPx1 activity were associated with an increased risk of cardiovascular events in several prospective cohort studies [160–162], whereas atherosclerotic plaques of patients with carotid artery disease have reduced GPx1 activity . Furthermore, it was suggested that the GPx-1 level was a valuable marker for monitoring cardiovascular events when considered in combination with plasma vitamin E levels . These data suggest that the targeting of GPx1 to bolster its antioxidant-like activity in diabetes is a potentially attractive and novel antioxidant strategy that should be pursued.
Similarly, the cytosolic thiol protein peroxiredoxin 2 (Prx2) reacts rapidly to remove hydrogen peroxide in a similar fashion to GPx1, albeit with higher affinity for hydrogen peroxide . In studies in hyperlipidaemic mice, it could be demonstrated that lack of Prx2 accelerated atherosclerosis via up-regulation of proinflammatory pathways which included the p65 subunit of NF-κB, c-Jun, c-Jun N-terminal kinases (JNKs) and p38 MAPK . The downstream effect of a lack of Prx2 was an increase in the expression of VCAM-1, ICAM-1 and MCP-1, which led to increased immune cell adhesion and infiltration into the aortic intima. Indeed, Prx2 deficiency showed a severe predisposition towards the development of atherosclerosis, suggesting an important role, as was observed in GPx1-deficient mice, for hydrogen peroxide-mediated injury in the atherogenic process.
Finally, further evidence of a role for hydrogen peroxide in the promotion of atherosclerosis comes from the analysis of mice over-expressing catalase. In these experiments, over-expression of catalase significantly reduced the severity of lesions in ApoE-deficient mice . This is supported by additional evidence where it could be shown that the onset and progression of diabetes are accompanied by reductions in various antioxidants, including catalase activity, in this setting [168–171]. However, a role for catalase in the protection against diabetes-mediated ROS injury is debatable because others report an increase in catalase activity [172,173]. More recently, mutations within the catalase gene have been suggested as contributing to the increased risk of diabetes . However, other studies report no association with catalase gene polymorphisms and the development of diabetic complications [175,176].
Novel strategies for bolstering antioxidant defences in diabetes
Pre-clinical evidence from GPx1 and Prx2 knockout mouse studies [154,166], as well as from catalase-over-expressing transgenic mice , point to an important role for hydrogen peroxide in driving the atherogenic process. Furthermore, several studies suggested that oxidative processes drive both the proatherogenic and the profibrotic processes underpinning diabetes-associated atherosclerosis and DN [30,154]. Evidence now suggests that treatment with an antioxidant such as ebselen has the potential to reverse both the atherosclerosis and the kidney injury associated with diabetes [41,154].
Ebselen and its analogues
Studies have been undertaken under both hyperlipidaemic and hyperglycaemic conditions to investigate the effectiveness of ebselen in attenuating diabetic complications. Ebselen is a synthetic, lipid-soluble, non-toxic, seleno-organic compound  with anti-inflammatory and antioxidant activities . Initially described as a GPx mimetic , it is now appreciated that ebselen may mediate its anti-inflammatory and antioxidant properties via multiple mechanisms, including disrupting the assembly of Nox1 and Nox2 isoforms at the cell membrane at concentrations well below those ascribed to its antioxidant function . In Prx2 knockout mice fed a high-fat diet, ebselen significantly attenuated plaque formation . In diabetic GPx1/ApoE double-knockout mice, treatment with ebselen by daily gavage for 20 weeks significantly attenuated atherosclerosis and endpoints of DN . The latter data clearly demonstrated that ebselen was able to lessen two often linked diabetic co-morbidities . Furthermore, these data strengthened the notion that ebselen reduces oxidative stress, because oxidative markers such as nitrotyrosine and urinary isoprostanes were significantly attenuated by this drug treatment . In addition, proinflammatory and profibrotic mediators such as VCAM-1, the F4/80 marker of macrophages, and renal VEGF and connective tissue growth factor (CTGF) expression, were significantly reduced by ebselen .
Current investigations are under way to improve the efficacy of synthetic GPx mimetics such as ebselen, with considerable effort to advance the biochemical functionality of the selenium moiety at the active site. Efforts are being directed towards either modifying the basic structure of ebselen or incorporating structural features of the native enzyme [181–183]. Recently, the use of a modified ebselen analogue has been investigated using in vivo studies in diabetic mice . Based on in vitro studies showing superior efficacy compared with ebselen, the modified selenenyl amide, m-hydroxy-ebselen, was selected and shown to reduce kidney injury and atherosclerosis to the same extent as seen with the parent compound ebselen. Importantly, these studies highlighted the significance of in vivo analyses to test the efficacies of novel ebselen analogues, as in vitro and cell-based assays only partly predict the in vivo situation . Other studies have highlighted the potential of diphenyl diselenide (DD), a simple organo-selenium compound with GPx-like activity, to reduce atherosclerosis. In these studies, low doses of DD reduced atherosclerotic lesions in hypercholesterolaemic low-density lipoprotein (LDL)-receptor knockout mice . Improvements were noted in endothelium-dependent vasorelaxation, and DD lowered oxidative stress markers (nitrotyrosine and malondialdehyde levels), prevented the up-regulation of MCP-1 and decreased inflammatory cell infiltration, suggesting an anti-atherogenic effect of DD by modulating pathways related to antioxidant and anti-inflammatory responses.
Antioxidant responses via Nrf2 activation
Nrf2 is an important redox-sensitive transcription factor that co-ordinates the up-regulation of a range of cytoprotective and detoxifying phase II enzymes, such as NAD(P)H:quinine oxidoreductase (NQO1) as well as antioxidant enzymes and proteins, e.g. catalase and HO-1, through an antioxidant response element-dependent pathway . Nrf2 was found to be up-regulated in response to hyperglycaemia in the kidney and specifically in certain renal [187–189] and coronary arterial endothelial  cells. Several agonists exist such as sulfuraphane , resveratrol  and oltipraz . Ebselen is also considered to be an Nrf2 activator , where it has been shown to modify Kelch-like ECH-associated protein 1 (Keap-1), thereby relieving the inhibitory effect of Keap-1 on Nrf2 . One specific Nrf2 agonist, bardoxolone methyl, has been investigated for its potential to reduce the key DN endpoint, estimated GFR, a marker of renal function, in DN patients . Indeed, in phase II clinical trials including the bardoxolone methyl treatment: renal function in CKD/Type 2 diabetes (BEAM) trial (NCT00811889) study, treatment of bardoxolone methyl in patients with DN appeared to be beneficial when compared with placebo [119,196]. However, the use of bardoxolone methyl was found to be associated with higher mortality in the treated group in a subsequent phase III trial, known as the BEACON (Bardoxolone Methyl Evaluation in Patients with Chronic Kidney Disease and Type 2 Diabetes) study. Thus, the trial was prematurely terminated because of safety concerns (NCT01351675 on ClinicalTrials.gov) due to an increased, albeit small, incidence of heart failure, possibly as a result of the drug's off-target effects [197,198].
Recent pre-clinical investigations into the use of a bardoxolone methyl analogue, dh404, may assist in identifying potential off-target effects of bardoxolone methyl and its analogues, and suggest that there may be a therapeutic window for optimal effect with this drug class . After administration of dh404 at increasing doses to diabetic, atherosclerosis-prone, ApoE knockout mice, it could be shown that dh404 retarded both atherosclerosis and diabetic kidney injury in a dose-dependent inverse manner, where the lowest dose of dh404 was most effective in retarding atherosclerosis and kidney injury. Mechanistically, it was shown that, at higher doses, where dh404 was less effective as a treatment modality despite increases in antioxidant defences, increased expression of the proinflammatory mediators, MCP-1 and the p65 subunit of NF-κB, was detected, signifying that higher doses up-regulated proinflammatory responses. Importantly, these studies suggested that this class of compound is worthy of further study to lessen diabetic complications but that careful evaluation of dosage is needed to identify maximal effect [199,200]. In addition, future studies would benefit from the identification of other Nrf2 activators with possibly even greater efficacy, as well as a deeper understanding of the mode of action of this novel class of compound. Indeed, the strategy of bolstering antioxidant defences through Nrf2 modulation may represent a new class of therapy , with potentially major advances over conventional therapy in the treatment of diabetic complications once the limitations of such an approach have been understood and appropriately addressed .
Early antioxidant treatment
Recent pre-clinical evidence in the diabetic Akita mouse model suggests that the timing of antioxidant therapy administration may be key to the protection against diabetic end-organ injury. In a late-intervention study, after evidence of significant structural and functional injury, administration of ebselen for 16 weeks failed to halt progression of the disease despite significant reductions in oxidative stress . This points to yet a further window of opportunity that needs to be considered in light of the failed clinical trials such as the heart outcomes prevention evaluation (HOPE) study in which patients were aged 55 years and older with established CVD or diabetes .
TARGETING NO BIOAVAILABILITY: DISRUPTING THE ENOS–CAV-1 INTERACTION
The development of both macro- and micro-vascular complications associated with diabetes is largely attributed to the adverse effects of hyperglycaemia on the vascular endothelium, and in particular hyperglycaemia-induced alterations in NO, the main protective mediator of endothelial health. Indeed, lowered NO bioavailability is a characteristic feature of endothelial dysfunction, which is now widely considered an independent prognostic tool for CVD associated with diabetes .
NO, produced by the constitutively expressed eNOS, is protective against CVDs through its potent vasodilatory action, and its ability to limit oxidative stress and inhibit proinflammatory leukocyte–endothelial interactions . Indeed, endothelial dysfunction is a consistent finding in clinical and pre-clinical models of diabetes. Patients with diabetes exhibit paradoxical vasoconstriction rather than vasodilatation when exposed to increasing amounts of acetylcholine in the coronary circulation , an indication of the extent of endothelial injury. Furthermore, in a recent study, it was demonstrated that eNOS phosphorylation at Ser1176, a known site for positive eNOS regulation, was diminished in the db/db Type 2 diabetic mouse model. This was rescued on ‘knock-in’ of the phospho-mimetic eNOS gene, resulting in improved vascular reactivity .
As endothelial dysfunction is the major contributor of diabetes-associated vascular diseases, targeting eNOS-derived NO is considered an unequivocal therapeutic target. However, simply activating the eNOS enzyme to increase NO is a more complex task. This is due to the fact that, in pathological states such as diabetes, the eNOS enzyme can be ‘uncoupled’ and switched to producing deleterious superoxide anion rather than protective NO. The resultant superoxide can inactivate NO, thereby forming peroxynitrite and further contributing to lower NO bioavailability . Indeed, in experimental diabetic conditions, eNOS-derived NO production is significantly inhibited whereas endothelial superoxide production is markedly increased. Moreover, the eNOS monomer:dimer ratio, an indicator of eNOS uncoupling, is augmented in cultured endothelial cells exposed to conditions of high glucose . Furthermore, tetrahydrobiopterin (BH4), an essential eNOS co-factor that stabilises the dimeric ‘coupled’ form of the enzyme, is oxidized in the diabetic state, thereby promoting eNOS uncoupling and peroxynitrite formation . Lastly, in the diabetic milieu, the heightened state of oxidative stress induced in particular by the Nox family of enzymes can directly induce eNOS ‘uncoupling’ .
Protein modifications induced by ROS such as peroxynitrite, are implicated in diabetes-associated vascular diseases , e.g. peroxynitrite formation causes, induced by high glucose levels, increased tyrosine nitration and inactivation of prostacyclin synthase, an enzyme involved in the synthesis of vasodilatory and anti-inflammatory prostacyclins . Recently, an elegant study by Cassuto et al.  demonstrated the novel contribution of peroxynitrite to microvascular dysfunction in diabetes. Peroxynitrite was found to alter the function of coronary endothelial caveolae, which are plasma membrane domains that localize signalling molecules such as eNOS. In both in vitro and in vivo diabetic settings, it was found that peroxynitrite levels were increased, which resulted in disassembly of caveolar organelles, decreased expression of endothelial caveolin-1 (Cav-1, the main coat protein of caveolae) and uncoupling of eNOS. Functionally, this resulted in significantly reduced flow-mediated dilatation of coronary arterioles and increased risk of coronary artery disease . The effect of disruption of caveolar assembly and eNOS uncoupling was further corroborated in isolated microvessels from Cav-1 knockout mice in which endothelial caveolae are absent . Sepiapterin, a BH4 precursor, restored flow-mediated vasodilatation in patients with diabetes, suggesting that eNOS uncoupling plays a direct role in microvascular dysfunction . These data further strengthen the concept that proper eNOS coupling and regulation are necessary to shift the balance against ROS-producing eNOS to protective NO-releasing eNOS, to protect the vasculature against diabetes-associated vascular diseases.
Despite the development of longer-acting nitrate therapy, the continued or frequent daily use of such agents soon results in tolerance. Once tolerance develops, patients lose the protective effects of long-acting nitrate therapy . Therefore, this approach has limitations for patients with diabetes who require long-term treatment to elevate NO. A recent novel approach to increase NO bioavailability has been to target proteins that directly influence eNOS function. One such example is the development of an antagonistic Cav-1 peptide, designed to mimic the scaffolding domain of Cav-1, an endogenous inhibitor of eNOS . By modulating the inhibitory domain of Cav-1, this peptide was able to release the inhibitory clamp on eNOS and increase NO bioavailability . In addition, this approach preserves caveolar function, enabling organelle assembly and trafficking to proceed unhindered. Indeed, in ex vivo studies using this peptide approach, it was shown that endothelial dysfunction is restored in aortic rings after treatment with CavNoxin, a mutant Cav-1-derived peptide, in an eNOS-dependent manner . To date, the Cav-1 antagonistic peptide is the only known small molecule peptide that is able to directly activate eNOS function. Although this strategy is still in pre-clinical development, the increased use of injectable therapies (small molecule peptides or antibodies) such as the various antibodies to PCSK9, albeit in other clinical CVDs, emphasizes the potential of these newer approaches in reducing CVD's burden. Indeed, with more than 100 peptide drug candidates in clinical development since 2001 and recent advances in peptide drug delivery, this is a rapidly advancing field. In addition to the Cav-1 antagonistic peptide, a primary screening of chemical libraries for compounds that increase eNOS gene expression revealed two small-molecular-mass compounds, known as AVE9488 and AVE3085, that have shown promise in their ability to increase endothelial NO production by simultaneous up-regulation of eNOS gene expression and reversal of eNOS uncoupling .
SUMMARY AND FUTURE PERSPECTIVES
Despite mostly negative clinical trials with non-specific antioxidants, overwhelming pre-clinical evidence supports a more targeted approach towards limiting ROS production, or the bolstering of appropriate antioxidant defences, in limiting diabetes-mediated micro- and macro-vascular complications. The present review has highlighted several newer approaches that are now being considered, as summarised in Figure 2. In particular, the exploration of specific Nox isoforms is shedding light on the specificity of these isoenzymes, with the targeting of Nox1 appropriate to limiting diabetes-associated atherosclerosis and the targeting of Nox4 to minimise diabetic kidney injury. With respect to antioxidant defences, it is becoming increasingly evident that the targeting of two-electron ROS such as hydrogen peroxide is more likely to protect against ROS-mediated damage and inappropriate ROS-driven signalling. In this respect, mimics of the peroxidases, such as ebselen and its analogues, are under current investigation. Improving the bioavailability of NO is yet another way to limit diabetes-associated vascular injury, with novel Cav-1 antagonistic peptides currently being investigated for their anti-atherogenic potential. However, results from clinical trials with newer compounds such as bardoxolone methyl have once again tempered enthusiasm for antioxidant-bolstering therapies. Nevertheless, with pre-clinical evidence suggesting that dosage and timing of treatment are issues that need further assessment, it is possible that such novel antioxidant therapies can ultimately be considered to reduce diabetes-related end-organ injury.
Specific targeting of diabetes-driven ROS/RNS
advanced glycation end-product
5′-AMP-activated protein kinase
- Ang II
glomerular filtration rate
glutathione peroxidase 1
human aortic endothelial cell
c-Jun N-terminal kinase
Kelch-like ECH-associated protein 1
mitogen-activated protein kinase
monocyte chemoattractant protein
nuclear factor erythroid 2-related factor 2
nitric oxide synthase
NF-E2-related factor 2
protein kinase C
reactive nitrogen species
reactive oxygen species
smooth muscle cell
transforming growth factor β
vascular cell adhesion molecule
vascular endothelial growth factor
vascular smooth muscle cell