The Role of NADPH Oxidases in the Etiology of Obesity and Metabolic Syndrome: Contribution of Individual Isoforms and Cell Biology

Abstract

Significance: Highly prevalent in Western cultures, obesity, metabolic syndrome, and diabetes increase the risk of cardiovascular morbidity and mortality and cost health care systems billions of dollars annually. At the cellular level, obesity, metabolic syndrome, and diabetes are associated with increased production of reactive oxygen species (ROS). Increased levels of ROS production in key organ systems such as adipose tissue, skeletal muscle, and the vasculature cause disruption of tissue homeostasis, leading to increased morbidity and risk of mortality. More specifically, growing evidence implicates the nicotinamide adenine dinucleotide phosphate oxidase (NOX) enzymes in these pathologies through impairment of insulin signaling, inflammation, and vascular dysfunction. The NOX family of enzymes is a major driver of redox signaling through its production of superoxide anion, hydrogen peroxide, and attendant downstream metabolites acting on redox-sensitive signaling molecules.

Recent Advances: The primary goal of this review is to highlight recent advances and survey our present understanding of cell-specific NOX enzyme contributions to metabolic diseases.

Critical Issues: However, due to the short half-lives of individual ROS and/or cellular defense systems, radii of ROS diffusion are commonly short, often restricting redox signaling and oxidant stress to localized events. Thus, special emphasis should be placed on cell type and subcellular location of NOX enzymes to better understand their role in the pathophysiology of metabolic diseases.

Future Directions: We discuss the targeting of NOX enzymes as potential therapy and bring to light potential emerging areas of NOX research, microparticles and epigenetics, in the context of metabolic disease.

Introduction

Obesity and diabetes have a profound effect on quality of life and place increasing financial and clinical burdens on our health care system (85, 194). In Western cultures where inactivity, high-fat intake, and nutrient excess are rampant, this poses a particular challenge. Considerable efforts have been channeled toward understanding metabolic disorders at the basic science and clinical levels.

Germane to this review, it is generally accepted that reactive oxygen species (ROS) play an important role in metabolic disorders and associated cardiovascular diseases (7, 86, 132, 136), including hypertension, hypercholesterolemia, coronary heart disease, and stroke (85). Nonetheless, as described recently by the COST Action BM1203 (EU-ROS) team (56), ROS and oxidative stress are all-too-commonly invoked on a casual basis for disease in an oversimplified way. Instead, complex redox biology should be viewed on the basis of the source of ROS, the amount and type of reactive species being produced, cellular location, and the redox-sensitive targets affected. Uncovering the source, localization, nature, and quantity of ROS, their redox modifiable targets, and downstream signaling during the development of obesity and metabolic disease could reveal important new therapeutic strategies to prevent or even halt disease progression.

The NADPH (nicotinamide adenine dinucleotide phosphate) oxidase (NOX) family of enzymes (Nox 1–5 and dual oxygenase [Duox] 1 and 2) (Fig. 1) is a professional superoxide anion (O₂^•−) or hydrogen peroxide (H₂O₂) generator (13), whose structure and complex formation have been the topic of many reviews (13, 73, 142, 143). In brief, activation of NOXs 1–3 requires coordinated complexation of membrane and cytosolic subunits the same as or related to the classical phagocytic subunits p22^phox, p47^phox, and p67^phox to produce O₂^•− (8, 13, 34, 51, 67, 146, 156, 169, 207). Nox4 associates with p22^phox in the membrane and does not require classical cytosolic subunits or their homologs, rather, activity can be mediated by Poldip2 in the cytoplasm, with reports showing both O₂^•− and H₂O₂ generation (4, 126). Nox5 and Duoxs 1 and 2 reportedly to date do not require cytosolic subunits and are activated by calcium binding to intracellular N-terminal EF hand motifs to produce O₂^•− and one or both of O₂^•− and H₂O_2, respectively (9, 14, 43, 55). NOXs are found in virtually every cell type and serve important roles in physiological and pathological signaling (13). Indeed, NOX enzymes have been implicated in the pathogenesis of metabolic disorders. Nevertheless, the investigation of the specific NOX isoform or isoforms involved and their functional roles has been limited by, among other factors: (i) the use of the inhibitors diphenylene iodonium (DPI) and apocynin, which are NOX isoform nonspecific and also act to inhibit other ROS-producing enzymes or as an antioxidant; (ii) data from global (also noninducible) knockout (KO) animals may be confounded by effects on other tissues or compensation during development; and (iii) limitations of present experimental tools, including a lack of effective antibodies for all NOX subunits, continued use of suboptimal ROS probes, and limited methods for detecting NOX redox targets, although recent advancements in oxidative proteomics could greatly enhance the ability to identify these targets. In addition, the field faces the challenge of addressing the exact mechanisms through which redox signaling is propagated in the presence of an extensive antioxidant system, which has a significantly lower Km value for H₂O₂ than modifiable cysteines in proteins. Whether NOX redox signaling directly oxidizes kinase cysteines, requires local inactivation of antioxidants such as peroxiredoxin, or relies on peroxiredoxin-mediated oxidative relay is poorly understood and remain critical questions in the field.

FIG. 1.

NOX isozymes. NOX complexes consist of their eponymous membrane-spanning catalytic subunit, which transfers electrons from NADPH through C-terminal NADPH- and FAD-binding sites and heme groups and reduces oxygen to superoxide anion. Nox1, 2, 3, and 4 complex with the stabilizing p22^phox membrane subunit. Canonical Nox2 requires assembly with cytosolic subunits (p47^phox_, p67^phox, p40^phox), while Nox1 and 3 require NoxA1 and NoxO1, the equivalent homologues for p47^phox, and p67^phox, respectively. Nox4 appears constitutively active and does not require cytosolic subunits but its activity can be modulated by Poldip2. In contrast, Nox5 and Duox 1/2 contain EF hand domains requiring calcium activation. NADPH, nicotinamide adenine dinucleotide phosphate; NOX, NADPH oxidase. Color images are available online.

Obesity and diabetes are complex conditions driven by a number of mechanisms, including inflammation, metabolic disruption, and ROS signaling. Numerous sources of ROS are implicated in the progression of metabolic diseases and resulting complications. The purpose of this review is to highlight recent advances and our present knowledge of specific NOX isoforms as to cell type and attendant signaling in the progression from normal metabolism to overweightness, obesity, metabolic syndrome, and diabetes. In this review, we present the gaps in knowledge and potential for future research endeavors. We focus on two key metabolic tissues (skeletal muscle and adipose) with additional emphasis placed on NOX in vascular pathologies, which are emerging as a leading cause of mortality in obese and diabetic patients. Notably, in this relatively nascent field, the literature is sparse in a number of areas.

Skeletal Muscle

In aggregate, skeletal muscle is a major metabolic organ in the human body. Reduced skeletal muscle energy consumption (i.e., physical inactivity) and consequential nutrient excess are major risk factors for the development of metabolic disorders (187). Skeletal muscle expresses Noxs1, 2, 4, and Duox1, and metabolic cues such as ATP, glucose, and fatty acids play a regulatory role in NOX expression and activity (49, 125, 160, 185). In response to exercise, NOX expression increases in a fiber-type-specific manner; specifically, Nox2 increases in type-2a fibers and Nox4 increases in type-1 fibers (125). In fact, Nox2 is activated in contracting muscle (demonstrated by Nox2-p47^phox proximity ligation assay) (89) by the ATP-protein kinase C (PKC) pathway (49) as well as by mechanical forces, potentially suggesting that NOX enzymes play a beneficial role in sensing muscle activity and mediating adaptations to those stimuli. Beneficial roles could include exercise adaptation (69, 94, 123), Ca²⁺ handling (94), glucose uptake (189), and insulin signaling (58). On the contrary, Souto Padron de Figueiredo et al. found that global Nox2 (KO) mice were protected from high-fat diet (HFD)-induced insulin resistance, suggesting that Nox2 may contribute to obesity-associated skeletal muscle insulin resistance (185). Considering the silencing of Nox2 was not cell specific, preserved insulin sensitivity might also be attributed to effects in other cell types (e.g., immune and endothelial cells). For instance, metabolically activated endothelial Nox2 contributes to endothelial insulin resistance, which has been suggested to impair skeletal muscle interstitial insulin levels and promote insulin resistance (111). In addition, innate immune cell populations, especially macrophages enriched in the Nox2 system, are elevated in skeletal muscle in both obese humans and mice (61, 102, 158). Interestingly, metabolically activated macrophage-derived conditioned media induce myocyte insulin resistance via PKC ϴ and Ɛ (101), known mediators of redox signaling. It is thus plausible that localized activation of NOX enzymes in these other cell types under obese or diabetic conditions could be damaging to skeletal muscle insulin sensitivity (101, 111) and could potentially impair other immune cell-mediated events, such as muscle recovery after ischemic injuries (84).

Homing in on the myocyte, short hairpin RNA (shRNA) knockdown of Nox2 in immortalized mouse myoblasts blocked high glucose- and high fat-induced suppression of insulin signaling (measured by pAkt) (185), suggesting that Nox2 or its oxidative metabolites play a deleterious autocrine role in skeletal muscle insulin resistance. Yet, it should be pointed out that, at basal levels (unstimulated conditions), knockdown of Nox2 surprisingly lowered insulin-induced phosphorylation of protein kinase B (Akt) without affecting glucose uptake, suggesting that Nox2-p-Akt signaling may play different roles on insulin resistance in the presence and absence of high glucose/fat challenge (185).

Taken together, those findings suggest that Nox2 signaling may mediate beneficial effects of exercise, and may also contribute to metabolic impairments associated with type II diabetes and obesity. Further investigation of these opposing effects is warranted and may depend on the quantity and location of ROS production (Fig. 2). In addition to its important metabolic functions, skeletal muscle interacts with cells in its milieu in a paracrine and perhaps even an endocrine manner. H₂O₂ produced during muscle contraction activates PGC-1α (69, 93), which regulates myokine production and muscle repair (52, 63, 139). These myokines play important endocrine roles throughout the body, such as in the regulation of adipose tissue phenotype and function as recently reviewed (21). Specifically, a number of myokines could potentially mediate vascular or adipose tissue NOXs; interleukin (IL)-6, BDNF, and IL-15 have been implicated to increase the expression or activity of NOX enzymes (98, 107, 200, 209), while irisin and IL-10 may act to downregulate NOX (97, 230). How the beneficial effects of exercise are transduced through myokine-mediated NOX expression and activity is relatively unknown. Therefore, future research aimed at understanding the role of NOX enzymes in skeletal muscle endocrine signaling and myokine impact on NOX expression and function in other tissues would be of great intrigue and interest.

FIG. 2.

Skeletal muscle Nox2 implicated in insulin resistance and exercise adaptation. (A) Skeletal muscle expresses Nox2, Nox4, and DUOX1/2. (B) In obesity and diabetes, Nox2 is upregulated. Increased Nox2 levels and ROS production induced by hyperglycemia and hyperlipidemia contribute to skeletal muscle insulin resistance, potentially through reduced insulin activation of Akt. (C) In response to exercise, an ATP-PKC pathway activates Nox2 leading to cellular adaptation, potentially mediated through PGC-1α. Akt, protein kinase B; PKC, protein kinase C; ROS, reactive oxygen species. Color images are available online.

Adipose Tissue (Visceral)

Adipose tissue is composed of a variety of cell types: adipocytes, preadipocytes, fibroblasts, vascular and lymphatic endothelial cells, and a number of resident immune cells, suggesting that a complex interaction between cell types sustains adipose tissue physiology and pathophysiology. Adipose depots act as the main storage sites for excess nutrients and, more specifically, visceral depots appear to represent an initial site of metabolic dysfunction on the road to obesity, metabolic syndrome, and type II diabetes. This is not mediated merely by global adipose depot expansion or increased adipocyte size (typified by the “healthy obese phenotype”), but requires visceral depot expansion and the disruption of autocrine, paracrine and endocrine adipose signaling (18, 135, 150). Further exemplifying this, lipodystrophy (loss of adipose tissue) induces metabolic derangements, such as insulin resistance, in the absence of obesity (106). This is potentially due to lipid accumulation in other tissues disrupting metabolic homeostasis and/or the loss of important adipokines such as adiponectin and leptin, which have been shown to modulate vascular NOX enzymes (96).

Adipocytes

A shift in adipose redox balance is a hallmark of metabolic disorders (86). Among the different NOXs, Nox4 is the most extensively studied in adipocytes (128, 175) while a few reports also examined changes in Nox 1 and 2 expression (174, 175). In healthy adipocytes, Nox4-mediated H₂O₂ production is increased following stimulation by insulin, amplifying its signaling through inhibition of the protein tyrosine phosphatase 1b (PTP1b), which promotes insulin receptor activation and glucose uptake (128). Along the same lines, in preadipocytes, Nox4-derived H₂O₂ is shown to augment insulin-mediated Akt activation and drive differentiation of preadipocytes into adipocytes (175). Nevertheless, the direct mechanisms and physiological implications of this Nox4-insulin cross talk in preadipocytes and mature adipocytes merit deeper investigation.

Likely an early occurrence in disease progression, nutrient excess leads to increased generation of intracellular H₂O₂ from Nox4 but not from mitochondria, causing expression of chemotactic factor genes and local inflammation that result in downstream insulin resistance in differentiated 3T3–L1 adipocytes (87). Importantly, in the same study, all of the NOX and Duox isoforms were probed and Nox4 was singularly and robustly induced at the RNA level, but less so at the protein level, perhaps suggesting a rapid turnover of Nox4 protein under glycolytic conditions. With little change in protein levels, the induction of Nox4-derived H₂O₂ is likely due to increased activity by posttranslational modification, potentially mediated by Poldip2 (126). Intriguingly, one recent study by Paredes et al. suggests that Poldip2 also plays a role in mitochondrial respiration; that is, downregulation of Poldip2 causes a shift toward glycolysis (155).

3T3–L1 adipocytes in culture exposed to high glucose and palmitate for 7 days exhibited increased glycolytic capabilities, NADPH content, and Nox4-derived H₂O₂ generation. Inhibition of the pentose phosphate pathway (PPP) or silencing of Nox4 both reduced H₂O₂ production and expression of monocyte chemoattractant protein 1 (MCP-1) (87), suggesting that PPP and Nox4 work in series to promote an inflammatory phenotype. Nox4 has been shown to activate NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells), a transcriptional regulator of MCP-1, and thus, relocalization and increased Nox4 activity may mediate this action of NF-κB intracellularly contributing to the inflammatory phenotype. Interestingly, these effects of Nox4 appear to be dependent on their localization to lipid rafts (87), although the signaling pathways underlying the translocation of existing Nox4 or directed localization of newly translated Nox4 in response to nutrient excess are unknown. Furthermore, whether Nox4-mediated redox signaling, via its intracellular molecular targets or cell-to-cell communication, is altered with the translocation is also unknown. It is, however, essential to point out that adipose tissue is heterogeneous (discussed in Resident and infiltrating immune cells section). This study suggests that redistribution of Nox4 to the lipid rafts increases adipocyte ROS production, intracellular NF-κB activation, and chemotactic signaling. Nonetheless, the direct consequences of these on intracellular or extracellular ROS and neighboring nonadipocytes remain undetermined. One plausible consequence could be that excessive extracellular H₂O₂ may mediate shifts in resident macrophage polarization in adipose tissue, as H₂O₂ has been shown to induce production of tumor necrosis factor (TNF)-α (141).

In vivo experiments support some of the findings revealed in cell culture. Global KO of Nox4 accelerated HFD-induced insulin resistance in mouse adipose tissue compared with wild type (WT) HFD controls (117). This is consistent with a fundamental salutary role of Nox4 on adipocyte insulin signaling even under pathological conditions. Furthermore, Nox4 also plays an essential role in adipocyte differentiation (175) potentially causing reduced adipocyte numbers in these mice, which could result in greater adipocyte hypertrophy under HFD. Indeed, adipocyte hypertrophy is associated with insulin resistance even in lean or healthy weight individuals (2, 82). Taken together, this may suggest that deficient Nox4 signaling during development or childhood resulting in diminished adipocyte number could predispose individuals to adipocyte hypertrophy and type II diabetes.

In addition, on account of the ubiquitous deletion of Nox4, contribution of this NOX to insulin resistance appears more complex and exhibits cell-type dependence. For instance, Nox4 in endothelial cells promotes insulin signaling and likely contributes to the observed accelerated loss of insulin sensitivity in global Nox4 KO (137). On the contrary, an adipocyte-specific Nox4 KO resulted in delayed insulin resistance and inflammation during the development of obesity (45). This effect might be attributed to the role of adipocyte Nox4, per se, in insulin resistance via recruiting immune cells (87), which could suppress insulin signaling through inflammatory cytokines, such as TNF-α. Consistent with this line of reasoning, when adipocyte ROS are elevated or glutathione defense is depleted, HFD-induced adipose dysfunction is accelerated (149). Dysfunction in these animals reportedly relates to restricted adipose expansion and an unfavorable profile of adipose inflammation, fibrosis, and lipogenesis (45). Furthermore, similar to cell culture experiments, KO of glucose-6-phosphate dehydrogenase, a rate-limiting enzyme of the PPP, prevents insulin resistance and adipose tissue inflammation (81). These experiments strongly suggest a link between the PPP and Nox4 activity in mediating immune infiltration and diabetic disease progression. Thus, it appears that adipocyte Nox4-derived H₂O₂ is essential for adipocyte function under physiological conditions, but redistribution and exacerbated Nox4 activity in metabolic diseases may initiate adipose inflammation and insulin resistance. Use of time-inducible adipocyte-specific Nox4 KO mice and coculture (immune cells and adipocytes) in future studies could help clarify the role of Nox4 in early versus late stages of metabolic disorders.

Knowledge of a role for Nox2 in adipocyte function and pathology is available yet scarce. Elevated expression of adipocyte Nox2 has not been shown under high-glucose or high-palmitate conditions in cultured 3T3–L1 adipocytes, but was reported in response to uric acid exposure (174), which is commonly elevated in metabolic disorders (31, 133, 196). Adipocytes challenged with uric acid showed NOX-derived ROS (likely O₂^•−)-mediated impairment of nitric oxide signaling (174), however, the specific NOX isoform responsible was not identified.

Resident and infiltrating immune cells

In healthy adipose tissue, metabolic homeostasis is, in part, dependent on resident immune cells. That is, resident eosinophils release IL-4 and maintain an alternatively activated macrophage (AAM) population (212). AAMs appear to promote adipocyte insulin signaling, nutrient uptake gene expression, and adiponectin production (147, 212). Indeed, depletion of eosinophils greatly reduced adipose AAM populations, which exacerbated HFD-induced insulin resistance (212). In this regard, AAMs appear to be protective even under HFD conditions. Intriguingly, transgenic overexpression of IL-5, which is known to increase the levels of resident eosinophils in adipose tissue, preserves adipose insulin sensitivity and systemic glucose tolerance (170), potentially through eosinophil-IL-4-mediated AAM retention.

With respect to the role of ROS on eosinophils in the adipose tissue, excessive H₂O₂ causes eosinophil apoptosis (168) and eosinophil populations decrease in adipose tissue with obesity (15). Combined with the observation that exposure to excess nutrients increases adipocyte ROS production (87), previous studies suggest that adipocyte-derived ROS may contribute to the observed loss of eosinophil and AAM in adipose tissue during the development of obesity. Therefore, experimental evidence is needed to definitively confirm this and identify the potential source of H₂O₂ leading to the loss of homeostatic immune cells.

Obesity is not only associated with a loss of beneficial or homeostatic immune populations but is likely also characterized by increasing populations of inflammatory immune cells, specifically macrophages. As discussed in the Adipocytes section, increased activity of adipocyte Nox4 in obesity, per se, is linked to adipocyte MCP-1 production (87). Consequently, potentiated infiltration of inflammatory phagocytes (145) would be expected to increase vicinal Nox2, the well-accepted most abundant NOX in phagocytes. In particular, infiltrating immune cell populations are predominantly comprised of macrophages (145), which express high levels of Nox2. As a consequence of exposure to excess nutrients and adipocyte inflammatory cytokines, macrophages are driven toward a proinflammatory polarization (109).

Shifting to NOX signaling in macrophages, we now focus on the roles NOXs play in the propagation of macrophage populations in adipose tissue. In macrophages, Nox2 can be activated by glucose and fatty acids (92), leading to NF-κB-mediated activation of inflammation (138) and inhibition of IL-10 (10), promoting an inflammatory phenotype. Release of inflammatory cytokines from proinflammatory macrophages, such as TNF-α, can further augment Nox2 activation, reinforcing the deleterious proinflammatory polarization of the macrophage population (220). The growing population of proinflammatory macrophages in obese adipose tissue gives rise to a low-grade inflammatory and oxidative environment, which further disrupts tissue homeostasis.

A potential disruption mediated by vicinal inflammatory macrophages is the reduction of the adipokine adiponectin in adipocytes, which is decreased in a dose-dependent manner by exogenous H₂O₂ (64). Furthermore, adiponectin is a key inducer of nitric oxide, IL-10, and AAM polarization. Therefore, diminished production of adiponectin in obesity likely contributes to a vicious cycle of higher NOX expression, ROS production, and attendant inflammation (5, 32, 148). Indeed, the reduction in adipose adiponectin production represents a key event in the progression of metabolic diseases; as such it has been targeted as a therapeutic for obesity and type II diabetes (1). In addition, loss of adiponectin, increased ROS, and resultant inflammation appear to be key factors in decreased sensitivity to insulin.

In obesity, the accumulation of proinflammatory macrophages in adipose tissue interferes with insulin sensitivity. In culture, macrophages impair adipocyte insulin signaling (215) through a TNF-α- and NF-κB-mediated increase in adipocyte PTP1b (184). PTP1b negatively regulates insulin receptor phosphorylation and diminishes insulin signaling. In healthy subjects, the autocrine role of Nox4-driven oxidation and inactivation of PTP1b leads to augmented insulin signaling. This is potentially disrupted by glucose and palmitate (128). It is thus likely that obesity-related upregulation of PTP1b expression and reduced inactivation contribute to the associated insulin resistance in adipocytes. Furthermore, because NF-κB is redox sensitive at higher levels of ROS (likely H₂O₂), paracrine H₂O₂ derived from phagocytic Nox2 may also contribute to upregulation of PTP1b expression. This could mean that autocrine H₂O₂ directly inactivates adipocyte PTP1b, while potential paracrine H₂O₂ from macrophages increases adipocyte PTP1b expression through NF-κB.

In vitro work establishes the potential for deleterious Nox2 upregulation of inflammation in macrophages, contributing to adipose tissue dysfunction. However, there are scant data on the direct in vivo actions of macrophage (or other myeloid cells) Nox2 on adipose function. While little is known about the direct actions of adipose tissue phagocyte Nox2 in vivo, we can glean some information perhaps from studies in global Nox2 KO mice. First, it appears that deletion of Nox2 reduces immune cell (especially macrophage and T cell) recruitment (24) and infiltration into adipose tissue in obesity (159). Second, KO of Nox2 preferentially shifts macrophages toward the anti-inflammatory AAM phenotype (159), likely explained by the role of Nox2 in the activation of inflammatory transcription factors (AP1 and NF-κB), inhibition of IL-10, and glucose metabolism discussed above (138).

Interestingly, Coats et al. (37) showed that phagocytic Nox2 may have a dual role in adipose tissue from obese mice. Nox2 KO mice fed a HFD for 8 weeks maintained normal insulin signaling and exhibited fewer proinflammatory cytokines compared with HFD controls as would be expected since Nox2 activates proinflammatory signaling. Conversely, after prolonged (16 weeks) exposure to HFD, the protective effect of Nox2 KO was lost and Nox2 KO mice presented with adipose tissue insulin resistance and inflammation similar to HFD-fed WT mice. Moreover, at 16 weeks, HFD-fed Nox2 KO mice exhibited increased lipid deposition in the liver, corroborating a similar finding from an earlier article by Costford et al. (38).

To test for a more definitive role of myeloid cells, studies were recapitulated in mice with a myeloid-specific ablation of Nox2 (37). While myeloid deletion affects numerous cell types (basophil, neutrophil, etc.), the authors show that these effects are likely due to the decreased ability of macrophages lacking Nox2 to breakdown lipids. Mechanistically, in macrophages, Nox2-dependent exocytosis of lipases was needed to break down apoptotic adipocytes into free fatty acids for uptake. This finding highlights a potentially beneficial role of macrophage Nox2 in apoptotic cell clearance, especially since Nox2 can be activated by apoptotic by-products, such as ATP.

However, this finding was not corroborated in a similar study by Pepping et al. In that study, 16 weeks of HFD-induced macrophage infiltration, adipose inflammation, and insulin resistance were all abolished by myeloid-specific deletion of Nox2 (159). A major difference in that study, as pointed out by the authors, was the time of HFD onset. Pepping et al. started HFD feeding after adulthood, while the aforementioned study started HFD feeding on weaning. This could mean that there is an age-dependent difference in the development of visceral adiposity, with early induction by HFD requiring macrophage Nox2 for tissue maintenance, while at later onset, only the deleterious role of Nox2 is revealed. Potential clinically significant distinctions between the two stages of HFD exposure augur pivotal studies that would hone in on the age-dependent function of adipose macrophages in obesity. This appears to be especially important based on the increased prevalence of childhood obesity over the past decades (80).

Based on this evidence, nutrient excess potentially leads to early alterations in adipocyte Nox4 signaling (Fig. 3). In response to nutrient excess, adipocytes augment chemoattractant release in an Nox4-dependent manner. This leads to increased proinflammatory phagocyte (predominantly macrophages) populations in visceral adipose tissue. Nox2-dependent signaling from macrophages promotes inflammation, insulin resistance, and altered adipocyte metabolism. However, some evidence supports that Nox2 is required for processing of apoptotic adipocytes and mitigating systemic lipid toxicity (38). Future work aimed at understanding the direct role of Nox2-derived O₂^•− or its downstream products on adipocyte function through coculture studies could uncover new pathways regulating metabolic disease progression.

FIG. 3.

Adipocyte Nox4 leads to a shift in adipose immune profile. (A) Nutrient excess leading to obesity causes a shift in immune cell populations. Adipose tissue comprised both adipocytes expressing mainly Nox4 and immune cells expressing Nox2. (B) Exposure to excess nutrients causes the localization of Nox4 to lipid rafts of adipocytes and increased H₂O₂. Elevated levels of adipose H₂O₂ increase MCP-1 and could potentially induce apoptosis of homeostatic immune cells, resulting in a shift toward insulin resistance and inflammation observed in obesity. H₂O₂, hydrogen peroxide; MCP-1, monocyte chemoattractant protein 1. Color images are available online.

Vasculature

While evidence clearly supports altered redox signaling early in adipose tissue contributing to metabolic disorders, disruptions in blood vessel NOX have also been reported early in obesity and may actually drive insulin resistance (105). This challenges, to some degree, the adipocentric view of metabolic disorders. Importantly, metabolic tissues and the vasculature are integrated in a number of ways, and more and more evidence illustrates that each depends on signaling from the other. Therefore, deciphering which organ is impaired by obesity first is a difficult task. Firstly, adipose tissue and skeletal muscle are perfused by capillaries that support the transport of (excess) nutrients, facilitate endocrine signals, and direct immune cell infiltration (105). Secondly, the vast majority of the vasculature is surrounded by specialized perivascular adipose tissue (PVAT) (17), which regulates vascular function and is, in turn, regulated by vascular signaling (5, 46). Finally, cytokines from adipose and skeletal muscle profoundly influence vascular function. In the context of NOX enzymes, adiponectin released from adipocytes was shown to suppress both Nox1 and 4 in vascular smooth muscle cells (VSMCs) and inhibit vascular NOX-derived O₂^•− (5, 96, 151), while the adipokine chemrin induces endothelial and VSMC NOX activity (144).

Understanding the physiological and pathophysiological roles of vascular NOX in the development and progression of metabolic disease is of high importance as cardiovascular mortality increasingly involves the risks associated with obesity and type II diabetes. Around 70% percent of diabetics die from some form of heart disease, with atherosclerosis likely subserving much of this disease risk (74). Furthermore, clarifying the role of specific isoforms in the context of specific vascular beds is crucial as their function and signaling targets may vary (118).

Perivascular adipose and adventitia

Similar to other adipose depots, PVAT immune cells contribute to normal function (211), but shifts in immune cell phenotypes and infiltration of proinflammatory immune cells (expressing Nox2) are a hallmark of dysfunctional adipose tissue arising in obesity, which was recently reviewed extensively (79). In rodents challenged with obesity, PVAT increases chemoattractant release (23, 46, 214). It is tempting to speculate that, similar to other adipose depots, Nox4 mediates MCP-1 expression in PVAT adipocytes as well. However, adipocytes in PVAT plausibly arise from different precursor cell lines compared with visceral depots (22). This warrants investigation into Nox4-mediated MCP-1 expression in PVAT adipocytes. Whether or not it is mediated by adipocyte Nox4, PVAT and adventitia show immune infiltration and increased expression of Nox2 in metabolic and vascular disorders (77, 223). Our understanding of specific NOX isoform functions in PVAT is significantly impeded not only by the diversity of PVAT along the vasculature (17) but also the use of apocynin or DPI, which are isoform nonspecific NOX inhibitors and have other NOX-independent side effects. Still, these studies in general highlight the role of PVAT immune infiltration (mainly macrophages), increased NOX expression (29), and exacerbated ROS production in obesity-induced impairment of vascular function (19, 100, 131).

Similar to visceral adipose tissue, increased ROS in PVAT promote the expression of proinflammatory cytokines, such as TNF-α, which can in turn activate vascular NOX enzymes (46, 204). Utilizing the Nox2-specific inhibitor Nox2ds-tat peptide in vitro on thoracic aorta PVAT from obese Zucker rats (a rodent model of metabolic syndrome), it was shown that Nox2 is responsible for a majority of oxidant production (46). In addition, vascular reactivity experiments revealed that PVAT Nox2 and TNF-α were essential for PVAT-mediated metabolic syndrome impairment of endothelial function, suggesting that PVAT Nox2 is dependent on inflammatory mediators to negatively influence vascular function.

While most evidence with respect to PVAT implies an “outside-in” mechanism of metabolic disease progression, recent evidence suggests that the PVAT both receives and sends signals to the vascular wall (5). Antonopoulos et al. showed that PVAT gene expression of adiponectin was inversely related to vascular O₂^•− in diabetic patients. In addition, using organ coculture experiments, activation of internal mammary artery with NADPH resulted in a similar upregulation of PVAT adiponectin gene expression potentially through a 4-hydroxynonenal-dependent pathway (5). This study provides evidence that vascular NOX signaling potentially activates compensatory gene expression in PVAT, but whether this change in gene expression corresponds to changes in protein level or changes in vascular function remains unexplored. While most studies support a detrimental role of PVAT ROS on vascular function in metabolic disease, the transmission of PVAT signals to the endothelium and VSMC is not well understood and may involve inflammatory mediators and/or intermediary cells such as adventitial fibroblasts.

Adventitial fibroblasts lie just below the PVAT and are recognized for their expression of Nox2 and 4 and contribution to vascular function (42, 48, 153). Fibroblasts are shown to be early responders in atherosclerosis progression (216). Signaling interactions between the PVAT and adventitial fibroblasts have been demonstrated (120, 171, 228) and may promote atherosclerosis-related proliferation and migration of adventitial fibroblasts (231). However, the role of NOX enzymes mediating this interaction in metabolic disease has not been studied and could represent an area ripe for exploration. Since H₂O₂ can induce NOX activity in a feed-forward manner in other cell types, it is plausible Nox2 activity in PVAT would start a chain reaction of increasing NOX activity throughout the vascular wall. Therefore, studies aimed at uncovering temporal and causal relationships between NOX isoform expression and functionality in the vascular cell types is expected to add to our understanding of the progression of vascular dysfunction in metabolic disease.

Endothelium and smooth muscle

The human endothelium expresses Nox1, Nox2, Nox4, and Nox5 at both membrane and intercellular locations, but a thorough understanding of the contribution of distinct cellular NOXs in metabolism and progression of metabolic dysfunction is lacking (115, 199). Delineating the roles of NOX isoforms in endothelial and smooth muscle cells is complicated by potential differences among vascular beds (118, 157). Metabolic disease is associated with endothelial dysfunction, VSMC remodeling, and loss of vascular compliance, all of which lead to the development of cardiovascular diseases such as hypertension and atherosclerosis. The NOX family of enzymes has been strongly implicated in the pathogenesis of metabolic disease-associated vascular dysfunction. Below, each vascular NOX isoform is discussed with regard to its cell type-specific role in the vascular complications of metabolic disease.

Nox4

In general, Nox4 is recognized for its role in physiological function. Nox4 promotes a differentiated contractile VSMC phenotype (36, 128), potentially dependent on subcellular localization of Nox4 to actin filaments. In the endothelium, similar to its function in adipocytes, Nox4 is essential to maintaining insulin signaling through Akt (137). In addition to regulating insulin signaling in the endothelium, Nox4 also appears to be an important player in the regulation of vascular tone. Importantly, Nox4-derived H₂O₂ does not react with nitric oxide. Endothelium-targeted overexpression of Nox4 lowers basal blood pressure and enhances aortic endothelial-dependent dilation ex vivo (165), potentially through NO-independent alterations in VSMC ion channels and membrane potential (165). Moreover, another study showed that laminar shear stress may promote endothelial nitric oxide synthase (eNOS) activation through Nox4-induced sulfenylation of SHP2 (tyrosine-protein phosphatase nonreceptor type 11) in the endothelium (173). These data support multiple roles for endothelial Nox4 in blood flow regulation. Further supporting the beneficial role of endothelial Nox4, Hu et al. reported that endothelial-specific, Nox4 dominant-negative mutant mice exhibited increased pulse wave velocity, a measure of aortic stiffness (an independent risk factor for cardiovascular events), in hyperglycemic apolipoprotein E (ApoE)^−/− mice, suggesting a key role of endothelial Nox4 in preventing the progression of arterial stiffness and atherosclerosis (91).

In addition to Nox4 regulation of vascular tone, some protective effects elicited by Nox4 may be mediated through repression of Nox2 and inflammation. In particular, global KO of Nox4 exacerbated aortic inflammation in ApoE^−/− mice, and had no effects on nitrotyrosine levels. Moreover, shRNA against Nox4 in human aortic endothelial cells increased Nox2 expression at both the basal level and after glucose insults, and exacerbated the expression of MCP-1, which was abrogated by Nox2 siRNA (71). These data suggest that there is a mechanism through which Nox4 downregulates Nox2 in the endothelium, resulting in blunted inflammation and vascular dysfunction. Conversely, isolated mouse coronary endothelial cells exposed to high glucose and insulin showed reduced Nox4 expression, which was prevented in cells isolated from Nox2 KO mice (59). From these studies, it appears Nox4-mediated downregulation of Nox2 (at least in the endothelium) is a key aspect of Nox4's vasoprotective effects. Based on this evidence, altered or diminished Nox4 in obesity may be an early event triggering endothelial dysfunction and vascular stiffness. Vascular stiffening before clinical manifestations of cardiovascular disease increases the risk of cardiovascular events and end-organ damage (205) and may represent a key pathological event for therapeutic intervention. Understanding Nox4-mediated pathways promoting vascular compliance may uncover novel targets to halt vascular complications of obesity before the development of deleterious cardiovascular pathologies, such as atherosclerosis.

Indeed, in the context of diabetic atherosclerosis, global Nox4 KO mice showed exaggerated disease progression associated with increased aortic macrophage infiltration and endothelial dysfunction (40, 47, 71, 113). These effects may, in part, be explained by increased expression of other NOXs, specifically Nox2 in the endothelium (71) and Nox1 in the VSMC (47). Moreover, disruption of endothelial Nox4 signaling in mice accelerated atherosclerotic progression (91) potentially through promotion of adhesion molecule expression, immune cell accumulation, and intima thickening (176). From global and endothelial-targeted Nox4 KO mice, it appears Nox4-mediated signaling is protective against diabetic atherosclerosis. However, global Nox4 KO may be dominated by its more pronounced effects on endothelial cells, while diverging influence of Nox4 on the VSMC may be masked. For instance, hyperglycemia has been reported to increase Nox4 expression and binding to p22^phox, essential for mediating IGF-1-stimulated signaling in VSMC (213), including oxidation/activation of sarcoma viral oncogene (Src) and SMC proliferation in vivo (213). These findings highlight a potential role for VSMC Nox4 in the development of atherosclerosis and vascular remodeling.

In aged ApoE^−/− mice, VSMC Nox4 expression is enhanced and associated with mitochondrial-ROS production as well as mitochondrial dysfunction. Furthermore, the same study also showed that aortic Nox4 expression correlates with age and atherosclerotic severity (201). Along the same line of reasoning, Tong et al. reported that targeting a nonfunctional Nox4 mutant to VSMC protected against atherosclerotic progression through downregulation of MCP-1, intercellular adhesion molecule 1 (ICAM-1), and vascular cell adhesion molecule (VCAM)-1 (192). In addition, Nox4 was found to promote epoxide hydrolase 2 expression, an enzyme that metabolizes epoxyeicosatrienoic acids. Activity of epoxide hydrolase 2 is strongly associated with VSMC proliferation as well as human coronary heart disease (114). However, the mechanism of Nox4-mediated upregulation of epoxide hydrolase 2 in rodents remains unclear as it appears to be independent of H₂O₂ production (192). In the study of VSMC nonfunctional Nox4 mice, left common carotid ligation-mediated induction of turbulent blood flow was used as the model for atherosclerosis, which likely disrupted endothelial function.

The contributions of endothelial signaling to VSMC Nox4 or vice versa have not been well defined. It is interesting that Nox4 has seemingly opposing effects in the two cell types (Fig. 4), where disturbed flow downregulates endothelial Nox4 leading to upregulation of VSMC Nox4 (91, 192). Uncovering endothelial and VSMC cross talk involving Nox4 expression or activity could elucidate mechanisms of metabolic disease-related atherosclerosis. While the studies above indicate that VSMC Nox4 is involved in lesion formation in mice and that Nox4 expression is correlated with atherosclerotic severity (201), human atherosclerotic lesions show decreased Nox4 expression (71), which may be instrumental for plaque instability and rupture (218). Temporal mechanisms underlying the expression and actions of Nox4 in atherosclerotic lesions are likely complex, but are expected to provide key insight into therapeutics aiming to reduce plaque rupture and thrombolytic events.

FIG. 4.

Diverging roles of NADPH oxidase 4 in the vascular smooth muscle and endothelium. (A) In vascular smooth muscle cells, elevated glucose levels induce increased expression of Nox4, which promotes proliferation and macrophage infiltration, both crucial components of atherosclerotic progression. (B) In the endothelium, exposure to elevated glucose diminishes Nox4 expression leading to the loss of Nox4-mediated salutary homeostatic signaling, enhancing insulin signaling activation of Akt, vasorelaxation (directly by H₂O₂ and indirectly through SHP2 and eNOS), and potentially repression of Nox2 and MCP-1 expression. eNOS, endothelial nitric oxide synthase; SHP2, tyrosine-protein phosphatase nonreceptor type 11. Color images are available online.

Nox4 has also been implicated in angiogenesis. Nox4 H₂O₂ production from vascular endothelial cells and cardiomyocytes has been shown to induce angiogenesis (30, 39, 227) and increase capillary density (39). In accordance with this, global Nox4 KO mice exhibited blunted exercise-induced angiogenesis (206). Mechanisms are expected to involve upregulation of vascular endothelial growth factor (VEGF) and repression of antiangiogenic factors. Peripheral capillary rarefaction has been previously described in early metabolic syndrome associated with elevated oxidant products (62), but the direct mechanisms remain unclear. It is possible that loss of Nox4 signaling in obesity contributes to vessel rarefaction through upregulation of other NOX isozymes. Determining the role of Nox4 in the maintenance of capillary density in adipose and skeletal muscle warrants investigation and may shed light on metabolic disease progression.

Nox2

In metabolic disease, endothelial Nox2 is generally considered to be deleterious as it impairs endothelial-dependent vasodilation. For instance, patients with hereditary deficiency of Nox2 exhibited enhanced flow-mediated vasodilation compared with healthy subjects (203). Supporting work in rodents also showed that Nox2 KO or treatment with Nox2ds-tat (aka gp91ds-tat) was protective against obesity and insulin resistance-associated vascular dysfunction (54, 186). During the development of metabolic syndrome and type II diabetes, Nox2 is highly active in the vessel wall and contributes extensively to vascular O₂^•− production (78) not only in endothelial cells but also in VSMCs (161).

Activation of endothelial Nox2 can be mediated by metabolic factors, including glucose, palmitate, and oxidized low-density lipoprotein (59, 112, 198). Evidence suggests that Nox2-mediated impairment of endothelial function is, in part, mediated by extracellular signal-regulated kinase (Erk)1/2 activation. Erk1/2 signaling may downregulate eNOS and insulin receptor signaling (54, 59). However, p-eNOS was only partially restored on Erk1/2 inhibition, suggesting that other mechanisms leading to endothelial dysfunction may also be at play. In addition, nitric oxide is directly inactivated by O₂^•− from Nox2, which may be mediated through colocalization of Nox2-eNOS in lipid rafts resulting in peroxynitrite formation (124, 226). Obese patients (regardless of sex) showed increased expression of endothelial p47^phox (needed for canonical Nox2 activation), suggesting Nox2 contributes to increased vascular O₂^•− and oxidative stress in these patients (178). Impaired nitric oxide and insulin signaling likely induces increased vascular tone as well as insulin resistance, leading to higher risk for cardiovascular pathologies and events.

Isolated primary coronary endothelial cells showed robust Nox2 activation, increased senescence, and apoptosis following 24 h of high glucose and insulin challenge (59), all of which potentially predispose the vessels to atherosclerosis and cardiac events (217). Even in isolated human adipose microvascular endothelial cells from healthy adults, Nox2 O₂^•− was induced by insulin, which appeared to inhibit insulin-induced nitric oxide production (129). Interestingly, Nox2 expression, as well as activity, was potentiated at what could be considered normal physiological levels of insulin (1 nM). This potentially suggests an important physiological role of Nox2 in mediating insulin signaling, which could provide negative feedback and fine tuning.

Although involvement of Nox2 under physiological circumstances requires further investigation, contribution of this isozyme to metabolic disease progression is validated by multiple studies. For instance, knocking out Nox2 abolished HFD-induced ROS production and Erk1/2 activation, and also reduced weight gain leading to improved insulin sensitivity and endothelial function in middle-aged mice (54). In addition, selective Nox2 inhibitor Nox2ds-tat effectively ameliorated insulin resistance-induced oxidative stress and vascular dysfunction independent of a change in body mass (186).

In an elegantly designed temporal study, endothelial function was assessed at 8, 12, 30, and 36 weeks of HFD in cerebral arterioles and carotid arteries. HFD for 12 weeks reduced cerebral arteriolar dilation to acetylcholine, which was prevented in the HFD-fed Nox2 KO mice despite the fact that no changes in fat mass, body mass, or basal glucose levels were observed (127). Nevertheless, since the study utilized global Nox2 KO mice, the direct contribution of endothelial Nox2 to the observed phenotype cannot be delineated as Nox2 in other cell types may also play a crucial role (e.g., macrophages and VSMC). It would be of great significance to conduct these experiments in cell-specific, inducible Nox2 KO mice, which were recently developed (172). Using these genetically constructed mice, it was shown that while Nox2 KO in myeloid cell reduced basal blood pressure, Nox2 KO in endothelial cells reduced blood pressure elevation in response to AngII (172). To what extent this observation can be recapitulated in obesity or diabetes studies is unknown. Future investigation using the cell-type-specific, inducible Nox2 KO mice would be essential to delineate the contribution of this isozyme in the progression of metabolic diseases.

Nox1

A number of studies in recent years have been conducted to investigate the specific role of vascular Nox1 in metabolic disorders. Qiu et al. reported that restoration of glycemic control via genetic deletion of myostatin in a rodent model of metabolic syndrome improved microvascular function and reduced Nox1 expression (162), implicated in eNOS uncoupling (180, 222) and may be regulated by Toll-like receptors (119). This contention is supported by a recent study from the same group showing a protective effect of global Nox1 KO on microvessel function (mesenteric arteries) in db/db mice (191). As Nox1 is expressed in both endothelial cells and VSMCs, global Nox1 KO-mediated improvements on both endothelium-dependent (acetylcholine-induced dilation) and smooth muscle-dependent responses (NO donor-induced dilation and myogenic tone) indicate that Nox1 present in both cell types likely contributes to metabolic vascular pathology. These findings are further supported in a Nox1 KO hyperglycemia mouse model. Alves-Lopes et al. showed that the contractile dysfunction of internal pudendal arteries was normalized in Nox1 KO hyperglycemic mice in comparison with the hyperglycemic WT counterparts, through reduced activation of Rho kinase and VSMC contractile state (3). Again, that study utilized global Nox1 KO and should be confirmed utilizing VSMC-specific KO of Nox1.

Studies investigating the contribution of Nox1 to atherosclerosis have shown conflicting results. Nox1 was originally shown to play a detrimental role in both HFD-induced and diabetic atherosclerosis (72, 177). Genetic deletion of Nox1 in the ApoE^−/− mice reduced carotid lesion severity, macrophage infiltration, and inflammatory cytokine levels, suggesting Nox1-derived O₂^•− production promotes atherosclerosis. However, a more recent study using the same mouse model (Nox1 and ApoE double KO) showed that Nox1 deficiency is associated with elevated levels of cholesterol, LDL, and plasma triacylglycerol compared with ApoE^−/− alone. Moreover, O₂^•− production in atherosclerotic aortas was surprisingly twofold higher in Nox1/ApoE double-KO versus ApoE^−/− mice, suggesting a protective role of Nox1-derived O₂^•− against atherosclerosis (183). Nevertheless, these results may potentially be revealed by the increased lipid burden in these animals and may in fact highlight an important role of Nox1 in nonvascular tissue in controlling metabolic processes in the ApoE^−/− model. Interestingly, in this study, atherosclerotic pathology only differed in severity at the aortic sinus between the Nox1 KO ApoE^−/− and the ApoE^−/− mice. This could be due to differences in cellular lineages along the aorta.

Possible explanations for opposing findings in the Sheehan et al. and Sobey et al. studies using similar animal models may include different diet fat content or induced vascular injury. In mice fed with higher fat content (42%), Nox1 KO seems protective (177), whereas in mice treated with lower fat content (22%), deleting Nox1 appears detrimental (183). The adverse vascular effects of Nox1 KO may be secondary to a role of Nox1 in controlling cholesterol and lipid circulation in other tissues. Finally, Nox1 KO was found to be protective when carotid ligation was implemented in concert with HFD to induce atherosclerosis (177). It is therefore possible that artery ligation-induced turbulent blood flow is a trigger for the detrimental activation of Nox1 in atherosclerosis development.

While recent advances have been made to understand the role of Nox1 in obesity- and diabetes-associated vasculopathy, many questions remain, especially cell type-specific involvement of Nox1. Development of endothelial- and VSMC-specific inducible KO of Nox1 would greatly assist in deciphering Nox1 mechanisms in various cell types and would also be well suited to answer the temporal contribution of Nox1 to obesity-associated metabolic disorders.

Nox5

Nox5 is not expressed in murine models and, therefore, robust in vivo data are lacking. Little is known about Nox5 in obesity and metabolic diseases, but a role in diabetes is implicated as high glucose increases endothelial calcium concentration (208), which is needed for Nox5 activation. In addition, high glucose-treated human pulmonary endothelial cells showed increased activity of PKCα. PKCα was subsequently shown to phosphorylate Nox5 and increase its activity in response to calcium stimulated by ionomycin in endothelial cells overexpressing Nox5 (26). In human VSMC, expression of Nox5 can be induced by factors elevated in obesity/type II diabetes, such as TNF-α, endothelin-1, and angiotensin II (154). Interestingly, overexpression of both catalytically active and inactive Nox5 variants in VSMC induces proliferation (154). This suggests a potential role for Nox5, independent of O₂^•− production, in the hyperproliferative VSMC phenotype seen in obesity/diabetic-induced atherosclerosis. In human subjects, Nox5 expression and activity are increased in coronary arteries of patients with coronary artery disease (76). Interestingly, Nox5 expression showed temporal and cell-specific differences with increased endothelial expression preceding upregulation in the vascular smooth muscle. The significance in timing of Nox5 expression is not known but may suggest that damaged endothelium and vascular inflammation promote VSMC Nox5 and atherosclerosis progression.

In summary, a dearth of data from vascular cell-specific KO mice limits present appreciation of cell-specific contributions of NOXs in metabolic disease-associated vascular pathologies. Based on present evidence, it is speculated that disruption of physiological endothelial Nox4 signaling is an early event in metabolic disease pathology. Downregulation of endothelial Nox4 leads to upregulation of other NOX isoforms, potentially in both endothelium and VSMCs. Eventually, loss of endothelial Nox4 signaling and potentiation of other vascular NOX enzymes lead to endothelial and VSMC dysfunction that ultimately result in the progression of atherosclerosis and increases in cardiovascular mortality. Future research utilizing temporal study design and cell-specific, inducible KO mice of vascular NOX enzymes should bring to light more exciting discoveries.

Emerging New Areas of Investigation

Microparticles

Microparticles are small (0.1–1 μm) diameter membrane vesicles, which display receptors, membrane components, and contain proteins along with nucleic acids from the cell of origin (134). Microparticles are released from virtually every cell type, including endothelium in response to stress, especially altered blood flow and apoptosis (134). Release of endothelial microparticles increases following the ingestion of a high-fat meal (197) and the response is exaggerated in diabetic subjects (197). A more recent study found that the increased levels of circulating microparticles in diabetic patients correlate with cardiovascular death (179). Together, the studies suggest a clinical link between microparticles, type II diabetes, and cardiovascular complications. Underlying mechanisms for these observations are not well understood, but may involve microparticle-induced endothelial inflammation and atherosclerosis, as injection of microparticles from injured endothelial cells impaired endothelial cell function and promoted macrophage infiltration of lesions in ApoE^−/− mice (95). In that study, microparticles released by cultured human coronary endothelial cells in response to high glucose showed elevated NOX-derived ROS, but the specific isoform was not identified (95). It would be intriguing to replicate the study with microparticles obtained from endothelial cells following NOX gene silencing and/or ex vivo treatment of microparticles with NOX inhibitors to probe a role for NOX-containing microparticles in metabolic disease-induced atherosclerosis. Recently, both Nox2 and Nox4 were shown to be expressed in circulating microparticles in lean and obese patients (50). It was observed that microparticles derived from obese patients showed higher expression of Nox4 and reduced expression of Nox2 compared with lean patients (50). Although limited, these intriguing findings suggest there is a potential for microparticles containing NOX to participate in diabetic-induced atherosclerosis.

Epigenetics

Epigenetics, in general, refers to post-translation modifications made to histones, various species of RNA, and DNA methylation that influence the structure and function of chromatin and gene transcription. Redox regulation of the epigenetic programming of gene transcription has been the topic of recent reviews (104, 110). Still, the direct role of NOX enzymes in epigenetic gene regulation, and in the context of metabolic disease and atherosclerosis, has been insufficiently explored, and represents a fertile field for future investigation. As discussed above, atherosclerosis is a common consequence of metabolic diseases and recent advancements have been made in the potential cross talk between redox signaling and epigenetics in the disease progression resulting in atherogenesis.

Original observations identified H₂O₂ being an instigator that promotes hypomethylation of DNA through increased expression of the ten-eleven translocation enzymes (TET) and depletion of upstream precursor S-adenosyl methionine, which correlates with reported diminished methylation in human tissues in metabolic disease (90). However, this notion has recently been challenged in VSMCs. In atherosclerotic lesions, Liu et al. (122) demonstrated that reduction in the transcriptional activating “mark” 5-hydroxymethylcytosine (5hmC), essential for VSMC contractile gene expression, was coupled with a decrease in TET2 expression. Therefore, loss of this marker and modulator may contribute to lesion progression. In agreement with a potential role for ROS in this process, Delatte et al. (44) found dedifferentiation and reduced 5hmC in VSMC in an animal model of oxidative stress. These observations may suggest that redox signaling and potentially NOX participate in the epigenetic reprogramming of VSMC in atherosclerosis.

In addition to modulation of DNA methylation, post-translational modifications to histone 3 may regulate VSMC proliferation. Redox modification or methylation of histone 3 was suggested to lead to enhanced proliferation. Garcia et al. (65a) showed enhanced glutathionylation of histone 3 in physiological proliferating cells and cancer cell lines, establishing a link between ROS production and an epigenetic marker of proliferation. However, the source of ROS and potential mechanistic relevance to atherosclerosis still remain unclear. In concert with histone 3 modifications and cell phenotype, DNA methylation lysine 4 appears to regulate the contractile VSMC phenotype. Indeed, multiple isoforms of histone demethylases have been shown to be upregulated in type II diabetes and contribute to atherosclerosis (28, 229). With that said, to our knowledge direct links between NOX and histone demethylase enzyme expression/activity have not been established. In addition, how redox modification of histone 3 promotes or interferes with histone 3 methylation is yet to be determined. Even though a direct role of NOX enzymes in these redox pathways has not been revealed, this is highly plausible based on the roles of NOX enzymes in atherosclerosis discussed in the Vasculature section.

Aside from NOX-derived ROS putatively driving epigenetic changes in metabolic disease, evidence exists linking epigenetic enzymes in the regulation of NOX expression. Two important families of enzymes regulating DNA accessibility through histone modification are the histone acetyl transferases (HATs) and histone deacetylases (HDACs). Canonical activity of HATs uses acetyl–CoA to add an acetyl group to lysine (K) residues at histone tails inducing chromatin structural remodeling allowing access to gene promoter regions. On the contrary, HDAC enzymes remove acetyl groups from histones and are generally associated with transcriptional repression. Paradoxically, some reports have shown that HDAC inhibitors reduce NOX expression and consequently NOX-derived ROS production in the vasculature (25, 130, 181).

It is proposed that vascular HDACs 1 and 2 increase in experimental diabetes (130) and may form a complex with transcription factors to regulate NOX transcription. This notion is supported by evidence of class 1 HDACs 1 and 2, at NOX promoter regions in VSMC (130) and HDAC inhibition reducing RNA polymerase 2 enrichment at NOX promoters in macrophages and lung fibroblasts (25). In opposition to this, the transcription of VSMC NOX was repressed by overexpression of class 2a HDAC4, suggesting an HDAC class-specific regulation of NOX expression. While little is known about HDAC-NOX pathways, indirect evidence suggests potential complex feedback and feed-forward pathways in hyperglycemia. Along these lines, it was demonstrated that H₂O₂ inhibits enzymatic activity of class 1 HDACs (53), setting up a potential feedback loop where NOX-derived ROS inhibit or activate (directionality remains to be determined) HDAC-mediated NOX transcription. In addition, class 2a HDAC4 has been shown to negatively regulate NOX expression in the VSMC (130), which is exported from the nucleus of myocytes under oxidative stress conditions (123). Thus, these new concepts suggest a potential role for NOX-derived H₂O₂ to facilitate HDAC4 export from the nucleus and remove its repression of NOX transcription (Fig. 5). While limited data exist to date, future investigation into the epigenetic regulations of NOX transcription and activity in the vasculature and, in turn, the impact of NOX-derived ROS on epigenetic regulation will enhance our knowledge of mechanisms governing vascular complications of obesity and type II diabetes.

FIG. 5.

Potential HDAC-mediated Nox expression in hyperglycemia. (A) Hyperglycemia may increase the expression of some class 1 HDACs (HDAC 1 and 2), which are suggested to complex with transcription factors and facilitate Nox expression. (B) The increased expression of Nox isozymes and increased oxidative stress induced by hyperglycemia have been implicated in the export of the class 2a HDAC4 from the nucleus. (C) HDAC4 is suggested to repress Nox expression suggesting a potential feed-forward signaling pathway where Nox-derived ROS remove HDAC4 repression of Nox transcription. HDAC, histone deacetylase. Color images are available online.

Therapies

A number of therapies (metformin, nitro fatty acids, and exercise) have been shown to alter NOX expression and activity. In some ways, these therapies may be more advantageous than targeted inhibition of a specific isoform as they might restore physiological NOX signaling across all isoforms instead of inhibiting one isozyme. Exercise benefits are multifaceted, although mechanisms of these beneficial effects are not completely understood. Evidence suggests exercise training suppresses p47^phox membrane localization (193) and decreased Nox2 expression (75) in rodent aortas, and inhibited Nox2 activity (20) in human subjects. In obese rats, exercise-triggered reduction in vascular NOX O₂^•− production, restoration of nitric oxide bioactivity, and suppression of vascular inflammation all contributed to increased vascular compliance (193). Exercise also exerts an anti-inflammatory effect, reducing cytokines known to enhance NOX expression and activity, such as TNF-α (195) and IL-6 (83). These exercise-mediated adaptations of vascular function in metabolic disease can conceivably reduce the risk of cardiovascular mortality in humans.

At present, metformin is a frontline drug for type II diabetes, in part, because it has widespread benefits dependent and independent of its glucose-lowering effects. Among its many effects, metformin has displayed inhibition of NOX activity through reducing p47^phox membrane localization in endothelial cells (11). Another therapeutic approach, presently in clinical trials, to treat metabolic disorders, is nitro-fatty acids. Nitro-fatty acids trigger post-translational modifications of reactive protein thiols resulting in activation or inactivation of the targeted protein. Biologically, nitro-fatty acids increase antioxidant defenses (12, 103) in human cells, and can even potentially inhibit NOX directly (70). Moreover, in rats, nitro-fatty acids accumulate in the adipose tissue (60), which may direct treatment to a key site of oxidative stress and inflammation. In addition, nitro-fatty acids have been shown to improve glucose tolerance and reverse pulmonary hypertension in a mouse model of obesity (99).

NOX inhibitors

Direct targeting of NOX enzymes has garnered much attention over recent decades. The oldest and most widely used inhibitors are DPI and apocynin, which are not isoform specific and have other non-NOX-related actions. Briefly, DPI inhibits flavoproteins, including eNOS and xanthine oxidase, while apocynin is an ROS scavenger and Rho kinase inhibitor. While apocynin has been shown to improve vascular function in obese mice (54), its broad profile of inhibition may interfere with salutary redox pathways and yield unwanted side effects. The more recent development of inhibitors, which show increased specificity within the NOX family may be more advantageous for understanding the roles of NOX enzymes in metabolic disease.

A number of compounds have been developed to target NOX enzymes, specifically GenKyoTex compounds (GKT-137831 and GKT-136901), and triazolo pyrimidine compounds (VAS2870 and VAS3947). In general, these compounds demonstrate inhibitory action on NOX enzymes, while having no or limited inhibitory action on xanthine oxidase or other flavoproteins.

The VAS compounds are postulated to interfere with complex formation and/or conformational changes of all the NOX enzymes with an IC₅₀ in the low μM range (66, 190). These attributes make VAS compounds potentially useful pan-NOX inhibitors, but limit their use in determining isoform-specific functions and therefore can only be used to draw a general conclusion about NOX involvement in a physiological or pathological process. Furthermore, experimental interpretation could be confounded by VAS compounds' off-target effects involving thiol alkylation that may replicate redox effects of ROS (188). Despite this, VAS compounds have been shown to reduce ex vivo constriction of aortas from mice with experimental diabetes (167) and improve endothelial-dependent dilation in fructose-fed rats (57).

The GKT compounds show considerable advantages in potency and isoform specificity over the VAS compounds. Pharmacokenetic studies demonstrate similar ability to inhibit Nox1 and Nox4 (<170 nM k_i), but less potency with respect to inhibiting Nox2 (>1530 nM k_i) and Nox5 (>410 nM k_i) (6, 65). Therefore, bioavailability should be monitored for in vivo application to ensure Nox1/Nox4 targeting. Although experimentally these compounds could be contrasted with gene-silencing techniques to determine isoform-specific effects, in vivo application in treating metabolic diseases may yield unwanted side effects especially in the vascular endothelial cells where Nox4 has been demonstrated to have vasodilatory and antiatherogenic functions as described in the Nox4 section. With that said, GKT-137831 has been used to inhibit VSMC inflammation and proliferation stimulated by adipokine chemerin in db/db mice (144) and to abrogate endothelin-1 signaling in cultured endothelial cells exposed to elevated glucose (152).

Another small molecule with NOX inhibitor actions is the synthetic organoselenium ebselen along with its congeners. Ebselen inhibits Nox1, Nox2, and Nox5 in the nM range (150, 500, and 700 nM, respectively) with Nox2 selectivity improved in subsequent analogs, such as that named Thr101, which has an IC₅₀ for Nox2 of 300 nM and IC₅₀ for Noxs1,4, and 5 above 3 μM (182). Other biological actions have been reported for low μM concentrations of ebselen such as inhibition of eNOS (225), suppression of amino acid metabolism (224), and action as a glutathione peroxidase mimetic (140). These effects, however, have not yet been assessed for ebselen congeners. Despite those widespread actions, which may complicate interpretation of ebselen-generated experimental results, the oral bioavailability, antioxidant actions, and ability to inhibit both Nox1 and Nox2 make ebselen a potential therapeutic option to treat metabolic disorders as long as in vivo concentrations are kept in a therapeutic window above the IC₅₀ for Nox2 (500 nM) and below the IC₅₀ for eNOS (8500 nM). In particular, ebselen was shown to reduce vascular dysfunction and improve insulin signaling in obese and diabetic rodents (16, 27, 33, 140, 219). In comparison, ebselen and its congeners appear to hold higher therapeutic potential than VAS or GKT compounds in the context of cardiometabolic diseases because they target Nox1, Nox2, and Nox5, which, as discussed previously, play key roles in the detrimental progression of metabolic disorders, but elicit little to no effects on the beneficial Nox4-mediated functions in vascular endothelial cells. Nevertheless, off-target inhibitory actions on eNOS still likely limit its clinical application. As such, a more rational approach strives to selectively target a single NOX isoform. Over the past decade, our laboratory has developed a number of isoform-specific NOX inhibitors, both for Nox2 and for Nox1.

Nox2-specific inhibitors

As discussed throughout the review, Nox2 is often linked to insulin resistance, inflammation, and oxidative damage; therefore, specific targeting of Nox2 in obesity and type II diabetes is expected to hold therapeutic value. The original peptide inhibitor of Nox2 (Nox2ds-tat, aka gp91ds-tat) interferes with p47^phox docking to Nox2 and is widely used in various disease models (41) (Fig. 6). Importantly, the peptide is specific to the Nox2-p47^phox interaction and shows no inhibitory effect on Nox4, conical Nox1, or hybrid Nox1 (Nox1-p47^phox) activity (41). This Nox2ds-tat peptide inhibitor has been shown to reverse atherosclerotic progression (163), block PVAT-mediated vascular impairment (46), and restore insulin signaling (186) in metabolic disease. Historically, peptidic drugs are viewed as not being advantageous due to concerns of bioavailability and membrane permeability. However, evolving technologies and strategies (166, 202) render the utility of a peptidic NOX inhibitor more likely. Recently, the laboratory has developed Nox2-selective tetrahydroisoquinoline small-molecule inhibitors CPP11g and CPP11h (35, 116). Similar to the Nox2ds-tat, these small molecules show no inhibition of other NOX isoforms and exhibit no known antioxidant properties (35). These compounds have been tested in vivo in an acute inflammatory mouse model, where they effectively ameliorated endothelial dysfunction in response to TNF-α (116). Thus, they may demonstrate promising therapeutic potential in obesity and type II diabetes similar to or even better than that observed with Nox2ds-tat.

FIG. 6.

Selective Nox inhibitors. Over the past two decades our laboratory has developed 4 Nox isoform-specific inhibitors. (A) The peptide NoxA1ds specifically interrupts the complexation between NoxA1 and Nox1, with potential benefits of selective Nox1 inhibition in metabolic diseases listed. (B) At present, one peptide and two small-molecule inhibitors exist for Nox2. Nox2ds-tat mimics the Nox2-p47^phox docking site inhibiting assembly and Nox2 superoxide production. The small-molecule inhibitors 11g and 11h are projected to fit into a p47^phox binding pocket disrupting the interaction with p22^phox and the assembly of the Nox2 isozyme specifically. Potential benefits of selectively inhibiting Nox2 in metabolic diseases are listed. Color images are available online.

Nox1-specific inhibitors

Similar to Nox2, Nox1, as discussed throughout the review, has been linked to a number of complications in metabolic disease, particularly atherosclerosis. Previously, the small molecule inhibitor ML171 was shown to inhibit Nox1 in a submicromolar range, while inhibition of other isoforms was only observed at significantly higher concentrations (68). This observation would render ML171 a relative selective inhibitor of Nox1. Even though this compound to date has been used sparingly, the therapeutic potential is promising due to the low IC₅₀ for Nox1. Our laboratory has recently developed an Nox1-selective peptide inhibitor, NoxA1ds (164), which blocks the interaction between NoxA1 and Nox1 (Fig. 6). Importantly, NoxA1ds showed no inhibitory effects on Nox2, Nox4, Nox5, or xanthine oxidase. In vitro administration of NoxA1ds inhibited hypoxia-induced superoxide production in human pulmonary endothelial cells (43a). Based on the literature discussed in this review, use of NoxA1ds may prove efficacious in reducing vascular resistance (3, 191) and preventing atherosclerotic development (72, 177) in metabolic disease.

In summary, these selective inhibitors are expected to be useful tools in determining isoform-specific signaling in cell culture and animal models of metabolic disease. Clinically, these selective inhibitors or their successors could have far-reaching indications and implications, including improving nitric oxide signaling, enhancing insulin sensitivity and mitigating inflammation. Nevertheless, it is plausible that global inhibition of even a specific isozyme could yield unintended side effects due to the diverse physiological and pathological roles of NOX in different systems and cell types. Therefore, the next frontier of NOX therapies (and many other therapies, for that matter) will be delivering specific NOX inhibitors to a specific tissue location or cell type. Emerging technologies such as microbubbles may be useful in achieving targeting specificity. Experiments have shown microbubbles carrying a drug payload can be targeted to release the drug in a specific area by ultrasound excitation (88, 108). Such therapies would likely reduce drug concentrations needed and avoid associated systemic side effects. In addition, manufacturing microbubbles to recognize specific cell surface marks may increase the delivery to sites of inflammation and injury (121, 210). In a similar manner, utilizing phagocytic cell biology to take up drugs and to preferentially deliver them in high inflammatory regions (221) may hold promise. Coupling NOX isoform-specific inhibitors with site- or tissue-specific drug delivery technologies is an exciting notion with a conceivably high therapeutic potential.

Conclusion

NOX enzymes play distinct isoform- and cell-type-specific functions, which are still not completely understood. The progression of metabolic dysfunction from nutrient excess to obesity and type II diabetes appears to be related to a loss of beneficial NOX signaling followed by increased expression of specific NOX isoforms and deleteriously exaggerated O₂^•− production. In metabolic disease, excessive ROS production from NOX enzymes contributes to inflammation, increased vascular tone, insulin resistance, and the progression of atherosclerosis. However, with NOX isozymes displaying important roles in both physiology and pathophysiology, it is likely that pan inhibition of NOX enzymes will not be a plausible therapeutic strategy. Targeting specific NOX isoforms potentially in a tissue- or cell-type-controlled manner may provide an advantageous therapeutic strategy in addressing metabolic disease pathologies.

Abbreviations Used

AAM

alternatively activated macrophages

Akt

protein kinase B

ApoE

apolipoprotein E

DPI

diphenylene iodonium

Duox

dual oxygenase

eNOS

endothelial nitric oxide synthase

Erk

extracellular signal-regulated kinase

H₂O₂

hydrogen peroxide

HATs

histone acetyl transferases

HDAC

histone deacetylase

HFD

high-fat diet

IL

interleukin

KO

knockout

MCP-1

monocyte chemoattractant protein 1

NADPH

nicotinamide adenine dinucleotide phosphate

NF-κB

nuclear factor kappa-light-chain-enhancer of activated B cells

NOX

NADPH oxidase

Nox

NADPH oxidase (specific isoform, denoted by subsequent number)

O₂^•−-

superoxide anion

PKC

protein kinase C

PPP

pentose phosphate pathway

PTP1b

protein tyrosine phosphatase 1b

PVAT

perivascular adipose tissue

ROS

reactive oxygen species

shRNA

short hairpin RNA

TNF

tumor necrosis factor

VSMC

vascular smooth muscle cells

WT

wild type

Author Disclosure Statement

P.J.P. is named on a filed patent (U.S. patent No. 9,187,528 B2) for NoxA1ds. All other authors declare that they have no conflicts of interest pertaining to the contents of this article.