NADPH (nicotinamide adenine dinucleotide phosphate) oxidases (Nox) are a family of transmembrane proteins of which there are seven isoforms: Nox 1–5 and Duox 1–2. Each Nox subunit serves as the catalytic core of a multi-subunit NADPH oxidase enzyme complex. Interestingly, the only known function of these enzymes is to generate reactive oxygen species (ROS), specifically superoxide (O2.−) or hydrogen peroxide (H2O2). They do so in response to signaling cascades initiated by osmotic or mechanical stress, hormones, vasoactive agents, and cytokines. Nox-generated ROS serve to modulate redox-sensitive targets in intracellular signaling pathways that control a wide variety of cell processes, including differentiation , fibrosis , growth , migration , proliferation , apoptosis , cytoskeletal rearrangement , and contraction . The importance of Nox proteins is evident not only from their roles in diverse cell processes, but also their vast tissue distribution. In mammals, Nox proteins are expressed in nearly all cell types, including several relevant to the cardiovascular system: cardiomyocytes , vascular smooth muscle cells (VSMCs) , endothelial cells (ECs) , adventitial fibroblasts , and leukocytes . Within the human vasculature, the crucial Nox isoforms are Nox1, Nox2, Nox4, and Nox5; these are the focus of this review.
Nox Structure and Function
The core structure of all Nox isoforms consists of six transmembrane domains and a cytosolic C-terminus (Figure 1). The third and fifth transmembrane domains contain conserved histidine residues that bind two prosthetic heme groups; and the C-terminus encodes two domains for the purpose of binding flavin adenine dinucleotide (FAD) and NADPH (reduced form). Nox use these prosthetic groups to transfer electrons from the cytosolic donor, NADPH, to FAD, then sequentially to each of the two heme groups and then to molecular oxygen on the opposite side of the membrane to generate ROS . Although all Nox use this mechanism to generate ROS, the modes of activation, subunit requirements, and subcellular localizations differ between isoforms. For instance, Nox1 and 2 are constitutively associated with another transmembrane protein, p22phox; and the full activity of Nox1/p22phox or Nox2/p22phox in response to agonist stimulation requires association with several cytosolic proteins: Rac, NoxO1 (or its homolog, p47phox), and NoxA1 (or its homolog, p67phox) are required for activation of Nox1/p22phox; and Rac, p40phox, p47phox, and p67phox are required for activation of Nox2/p22phox. Following complete enzyme complex assembly, Nox1 and Nox2 generate O2.−. Conversely, Nox4 and Nox5 do not require additional cytosolic proteins for their activation. Nox4 is constitutively associated with p22phox, is active in an agonist-independent manner, and produces H2O2 via dismutation of O2.− prior to its release from the enzyme ; whereas Nox5 is not associated with p22phox, is activated in a calcium-dependent manner due to an additional N-terminal calcium-binding domain, and produces O2.−[16, 17]. Once ROS are produced, they must interact with their target(s) in order to participate in cell signaling. However, O2.− does not readily cross membranes and is short-lived in its effect; and although it can be dismutated to longer lasting and membrane-diffusible H2O2, both of these ROS are relatively local in their effect. Therefore, the redox-dependent effects of Nox signaling are contingent on Nox localization such that ROS are produced in a specific location to affect a specific target . These differences in modes of activation, subunit requirements, type of ROS produced, and Nox localization are means by which NADPH oxidases achieve specificity in cell signaling.
Figure 1. Nox1, 2, 4, and 5 enzyme complex structure and function. The core structure of Nox1, 2, 4, and 5 consists of six transmembrane domains and a cytosolic C-terminus. The third and fifth transmembrane domains contain conserved histidine residues that bind two prosthetic heme groups; and the C-terminus encodes two domains for the purpose of binding FAD and NADPH. Nox use these prosthetic groups to transfer electrons from NADPH (in the cytosol) to FAD, sequentially to each heme group, and finally to molecular oxygen on the opposite side of the membrane to generate superoxide (O2.−). Nox1, 2, and 4 are constitutively associated with membrane-bound p22phox. Nox1 also associates with the cytosolic subunits NoxO1 (or its homolog p47phox), NoxA1 (or its homolog p67phox), and Rac. Nox2 associates with Rac, p47phox, p67phox, and p40phox. Nox4 does not require cytosolic subunits and performs two single-electron transfers to generate two superoxide anions that are dismutated to hydrogen peroxide (H2O2) prior to release from the enzyme. Nox5 does not associate with p22phox or cytosolic subunits, but requires calcium binding to its N-terminal domain to become activated.
Download figure to PowerPoint
Nox, Oxidative Stress, and Cardiovascular Disease (CVD)
In the context of physiological conditions, Nox produce ROS at levels that maintain physiologic cell processes such as cell signaling and the respiratory burst in neutrophils. However, under pathological conditions, Nox become overexpressed and overly stimulated, leading to the production of excess oxidants (molecules that readily accept electrons in chemical reactions), which can cause oxidative stress. Traditionally, “oxidative stress” refers to a disturbance in the normal redox state of a cell or tissue, meaning that there is either an increase in the production or a decrease in the removal of oxidants. This increase in oxidants results in an environment conducive to the creation of free radicals (compounds that contain unpaired electrons) that are highly reactive. Not only do Nox create oxidants and free radicals directly via the generation of H2O2 (an oxidant) and O2.− (an oxidant and free radical), they also create an environment in which additional oxidants and free radicals are likely to be produced. If cellular levels of oxidants exceed the antioxidant capacity of the cell, dysregulated cell signaling and oxidant-mediated injury to macromolecules can occur. Oxidative stress is implicated in a wide range of pathologies, but perhaps none more so than CVD. In humans, oxidative stress is involved in hypertension (HTN), atherosclerosis, myocardial infarction (MI), myocardial hypertrophy, restenosis, arrhythmias, and heart failure [19–23]. Although multiple cellular processes and enzymes can produce ROS and oxidative stress (including the mitochondrial electron transport chain, cyclooxygenase, lipoxygenase, heme oxygenase, cytochrome P450, xanthine oxidase, NO synthase, and peroxidase), Nox are the primary enzymes responsible for inducible ROS formation in the cardiovascular system [18, 24, 25].
Nox-derived ROS contribute to CVD by multiple mechanisms (Figure 2). First, ROS cause macromolecular damage of lipids, proteins, and DNA, which can lead to cell death . In the context of atherosclerosis, this cell death contributes to plaque erosion and thrombosis . Second, O2.− and H2O2 participate in cell signaling by oxidizing (and consequently activating or deactivating) redox-sensitive cysteine residues on target proteins that regulate cellular function, such as ion channels, phosphatases, kinases, cytoskeletal proteins, matrix metalloproteinases (MMPs), and transcription factors . For example, Nox-derived ROS modify the cysteine residue in the active site of protein tyrosine phosphatases (PTPs), thereby inactivating PTPs resulting in an increase in tyrosine phosphorylation of intermediates in signaling cascades, including the mitogen-activated protein kinases (MAPKs) . In VSMCs, MAPK phosphorylation causes migration and proliferation, thereby leading to neointima formation and vascular occlusion . Third, O2.− reacts with nitric oxide (NO), and as such reduces NO bioavailability, leading to: impaired vasodilation ; increased platelet adhesion and activation ; increased production of inflammatory molecules ; increased microvascular permeability ; vascular inflammation ; leukocyte infiltration of the endothelium [35, 36]; and plaque instability . Fourth, the reaction between O2.− and NO produces peroxynitrite (ONOO−), a powerful oxidant that causes lipid peroxidation, protein oxidation, protein nitration, and cell death . Lastly, ROS also oxidize low-density lipoprotein (ox-LDL), which contributes to atherogenesis by causing endothelial dysfunction, vascular inflammation, and foam cell formation . These findings clearly demonstrate the contribution of ROS to the pathogenesis of CVD. In addition, there have been numerous studies demonstrating the specific contribution of each vascular Nox to CVD. These studies described below not only reveal an important role for Nox1, 2, 4, and 5 in CVD, but also highlight the pressing need for targeted Nox therapeutics.
Figure 2. Role of NADPH oxidases in CVD. Risk factors for CVDs cause vascular injury and the release of cytokines and growth factors that increase the expression and activation of NADPH oxidases (Nox), which generate the ROS hydrogen peroxide (H2O2) and superoxide (O2.−) to cause oxidative stress. These ROS have the potential to cause oxidative macromolecular damage, oxidation of cell signaling intermediates, decreased nitric oxide (NO) bioavailability, increased peroxynitrite (ONOO−), and increased oxidized low-density lipoprotein (ox-LDL). Subsequent vascular inflammation and dysregulation lead to multiple CVD outcomes. ECM: extracellular matrix. CHF: congestive heart failure.
Download figure to PowerPoint
Nox1 is expressed in VSMCs and to a lesser extent in ECs, and contributes to the pathogenesis of atherosclerosis, restenosis, and HTN. Increased expression and activation of Nox1 in VSMCs causes increased degradation of extracellular matrix, cellular migration, and proliferation that contribute to neointimal hyperplasia, a precursor to restenosis and atherosclerosis . Furthermore, Nox1-deficient mice show a dramatic decrease in VSMC migration, proliferation, and the development of neointima in response to vascular injury . Moreover, in a mouse model of atherosclerosis (apolipoprotein E (ApoE) knockout and a high-fat diet), the deletion of Nox1 results in reduced levels of vascular ROS, aortic plaque area, and macrophage content of lesions . Additionally, in a murine model of HTN (infusion of angiotensin II, Ang II), the overexpression of Nox1 in VSMCs results in increased blood pressure and aortic hypertrophy as compared to control mice receiving Ang II [42, 43]. Moreover, mice that are deficient in Nox1 are protected from Ang II–induced HTN, and exhibit increased NO bioavailability, a marked reduction in aortic VSMC hypertrophy, and a decrease in extracellular matrix deposition [44, 45]. These studies indicate that Nox1 is a desirable therapeutic target for the treatment of CVD.
Nox2, also known as gp91phox, was the first Nox isoform to have been discovered. This protein is expressed at high levels in leukocytes, where its ROS production provides for the respiratory burst necessary for killing microbes within phagosomes. However, Nox2-dependent ROS production in leukocytes also contributes to the development of CVD. Furthermore, Nox2 is expressed in VSMCs, ECs, cardiomyocytes, and fibroblasts; and its activity in each of these cell types plays an important role in the development of CVD. Like Nox1, Nox2 also contributes to atherogenesis and HTN. In atherosclerotic human coronary arteries, Nox2 is most abundantly expressed in macrophages, and levels of the Nox2/p22phox mRNAs positively correlate with the severity of atherosclerosis . In ECs, Nox2 activity is necessary for the upregulation of inflammatory cytokines, such as TNF-α; the adhesion molecules necessary for leukocyte infiltration, including VCAM-1, E-selectin, and ICAM-1; and proinflammatory transcription factors, such as NFκB, all factors involved in vascular inflammation and atherogenesis [46, 47]. Furthermore, Nox2/ApoE double knockout mice have reduced O2.−, increased NO, decreased neointimal formation, and decreased atherosclerotic lesion of the abdominal aorta compared to ApoE knockout alone . A role for Nox2 in HTN has also been suggested by the finding that mice deficient in Nox2 are protected from Ang II–induced increases in blood pressure, vascular hypertrophy, and endothelial dysfunction . However, the contribution of Nox2-derived O2.− in HTN is controversial based on an alternate study which found that Ang II caused a similar increase in blood pressure in wild-type and Nox2 knockout mice, despite that mice deficient in Nox2 had a lower baseline blood pressure and were protected from Ang II–mediated aortic hypertrophy . Additional studies in Nox2-deficient mice have demonstrated that it plays a critical role in cardiac dysfunction. Nox2 knockout mice are less vulnerable to Ang II–induced cardiac hypertrophy and interstitial cardiac fibrosis . Although Nox2 is absent in all cell types in this model, these effects are presumably due to the absence of Nox2 in cardiomyocytes. This presumption is supported by further studies, demonstrating that cardiomyocyte-specific deletion of Rac1 (a required subunit for Nox2) also results in diminished cardiac hypertrophy in response to Ang II . Finally, Nox2 also appears to play a role in MI. Studies reveal that Nox2 expression is higher in infarcted than noninfarcted myocardium from patients who have died from MI  and that Nox2 deletion in mice protects against post-MI hypertrophy, fibrosis, and left ventricular dysfunction . Therefore, like Nox1, Nox2 also appears to be an attractive therapeutic target for the treatment of CVD. However, Nox2 inhibition in leukocytes could potentially lead to immune dysfunction, as is seen in X-linked chronic granulomatous disease (CGD), an immune disorder whereby defects in Nox2 lead to the absence of the respiratory burst in neutrophils and recurrent infections . So, although Nox2 appears to be a desirable therapeutic target, the potential side effect of immunodeficiency highlights the importance of therapeutics that selectively target pathological Nox2 activation in cardiovascular cell types while leaving physiologic Nox2 activation in neutrophils intact.
Nox4 is expressed in ECs, VSMCs, cardiomyocytes, and fibroblasts, and plays multiple and sometimes contradictory roles in the development of CVD. For instance, in ECs, Nox4 promotes the expression of endothelial surface adhesion molecules, adhesion of monocytes, and the transmigration of leukocytes into the vasculature . In addition, Nox4 has been implicated in promoting the migration, proliferation, and differentiation of both VSMCs and ECs [56, 57]. However, a transgenic model of Nox4 overexpression in ECs reveals Nox4-dependent vasodilation and lower blood pressure, under both unstimulated conditions and in response to Ang II compared to wild-type littermates . As indicated above, H2O2 is the primary product of Nox4, and H2O2 has been described as a vasodilator in multiple vascular beds, including human coronary arterioles . Nox4 overexpression in cardiomyocytes protects against contractile dysfunction and hypertrophy, whereas Nox4 deletion promotes these deleterious cardiac effects . And although renal Nox4 expression is known to be upregulated in rats with aldosterone/salt-induced HTN , this increased Nox4 expression may not be the cause of HTN in this model, but instead may be in response to HTN to counteract the prohypertensive effects mediated by Nox1 and 2. Although the function of Nox4 in CVD remains incompletely understood, available data suggest that Nox4 may have a complex role in CVD development. In contrast to the other Nox isoforms, therapeutic approaches to increase Nox4 activity may be protective in some situations of CVD.
Nox5 is the most recently characterized vascular Nox homolog, and its importance in CVD is not fully understood. The deficiency in our knowledge of Nox5 is due in part to the unusual finding that Nox5 is not expressed in rodents , and thus, studies have not been able to examine the effect of modulating Nox5 expression in murine models of CVD. In addition, the presence of multiple Nox5 splice variants also complicates its study . Nox5 is expressed in ECs and VSMCs in human vasculature and responds to agonists that contribute to CVD. In ECs, Nox5 is activated by Ang II and endothelin-1 , and its overexpression leads to increased proliferation and thus also to angiogenesis . In human aortic VSMCs, Nox5 is activated by platelet-derived growth factor (PDGF), and depletion of Nox5 by small interfering RNA (siRNA) leads to decreased PDGF-induced ROS production and proliferation of VSMCs . Furthermore, expression of Nox5 mRNA and protein is increased in human coronary arteries with atherosclerosis as compared to disease-free vessels, with expression of Nox5 predominantly in ECs in early lesions and in VSMCs in advanced lesions . These findings in human vascular cells suggest that Nox5 contributes to the pathogenesis of CVD; however, more work will be necessary to confirm its specific role in activating vascular cells to determine whether it will serve as an appropriate therapeutic target.
The remaining Nox homologs—Nox3, Duox1, and Duox2—are not known to be expressed in the human vasculature; and studies manipulating these isoforms have not revealed a CVD phenotype. Therefore, these Nox isoforms are unlikely to have a substantial role in CVD pathogenesis.
Other NADPH oxidase Subunits
Since p22phox is a required subunit for the Nox1, Nox2, and Nox4 enzyme complexes, and p22phox protein level is a major determinant of the expression level of each of these Nox isoforms, it is anticipated that the effects of altering p22phox expression represent the combined effects on all three of these Nox complexes. Furthermore, general observations from multiple studies suggest that Nox4 plays a contradictory role to Nox1 and Nox2 [58–60], which may explain the conflicting results of studies that manipulate p22phox expression. For instance, targeted depletion of p22phox by siRNA decreases the expression of Nox 1, 2, and 4 in kidney and also reduce Ang II–induced HTN . However, mice that overexpress p22phox in VSMCs do not develop HTN. Additionally, polymorphisms in the promoter of p22phox have seemingly opposing effects in CVD. For example, 930A>G causes increased phagocyte p22phox expression, increased ROS levels, decreased NO bioavailability, and is associated with HTN . However, the same polymorphism has a protective effect against premature coronary artery disease and the incidence of MI in individuals less than 40 years of age . These findings combined suggest that p22phox exerts a protective role in CVD via its association with Nox4-dependent signaling, whereas p22phox can exert detrimental effects in CVD via its association with Nox1 and/or Nox2-mediated signaling. These observations suggest that p22phox may be a poor therapeutic target in CVD.
As with p22phox, the remaining Nox enzyme complex subunits: p47phox, p67phox, and Rac, have also been evaluated to determine the role of Nox in CVD, and likewise, the outcomes of these studies are difficult to attribute to a specific Nox isoform due to the involvement of these subunits with both Nox1 and Nox2. However, these studies have been more consistent in revealing a detrimental role for p47phox , p67phox , and Rac  in CVD, and therefore support these subunits as potential therapeutic targets. An unresolved concern is that inhibition of p47phox, p67phox, or Rac for the treatment of CVD may also impair immune function by inhibiting Nox2 activation. However, it has been suggested by Drummond et al. that since phagocytic Nox2 activity must be two orders of magnitude lower than healthy controls to risk severe immunodeficiency; and since the activity of vascular Nox1 and Nox2 in animal studies is upregulated by less than one order of magnitude in CVD; it may be possible to titrate inhibitors targeting the shared cytosolic subunits such that vascular Nox is preferentially inhibited while immune function is preserved . If so, p47phox could be an ideal therapeutic target by inhibiting both Nox1 and Nox2 in the blood vessel.
In contrast, the cytosolic subunits NoxO1 and NoxA1 interact with Nox1, and not Nox2. NoxA1 has been shown to be necessary for VSMC migration and proliferation , and its overexpression increases O2.− production and neointimal hyperplasia following vessel injury. Additionally, NoxA1 is required for ROS production by EC in response to ox-LDL . And while the specific contribution of NoxO1 to CVD has yet to be demonstrated, it is known that NoxO1 is involved in Nox1-dependent ROS production in response to Ang II . Therefore, NoxA1 and NoxO1 may serve as appropriate targets for the treatment of CVD. However, NoxO1  and, to a lesser extent, NoxA1  are also involved in the activation of Nox3, which is expressed in fetal tissues and in adult inner ear, including the cochlear and vestibular sensory epithelia [77, 78]. Therefore, inhibition of NoxO1 or NoxA1 may impair fetal development or auditory and vestibular function. These potential side effects must be considered when determining the feasibility of NoxO1 and NoxA1 as potential therapeutic targets. For additional reviews of Nox in CVD, see [72, 79–81].
Therapeutic Barriers and Innovative Solutions
Nearly all currently available inhibitors of Nox are not isoform-specific; thus, treatment with these agents inhibits the function of multiple Nox isoforms. This could lead to potentially serious undesired effects, including immune dysfunction as a result of Nox2 suppression. Additionally, available Nox inhibitors do not specifically target pathologic signaling and leave physiologic signaling intact. Ideally, newly developed Nox inhibitors would be isoform-specific, cell-specific, and preferentially target pathological modes of Nox signaling. This level of specificity could potentially be achieved by directly interfering with the interface between a particular Nox isoform and the proteins with which it must interact to become active under pathological conditions, such as kinases, trafficking adapter proteins, or the cytosolic subunits of the NADPH oxidase. Specific sequence targeting could potentially be achieved by the following three approaches: mABs, small-molecule inhibitors, or aptamers.
mABs selectively bind to specific epitopes causing either activation or inactivation of the targeted protein. Multiple antibodies have been generated to p22phox, Nox2, and Nox4 [139, 140]. However, these antibodies have not been found to exert a therapeutic effect. An important limitation to this approach is that antibodies do not readily cross cell membranes, restricting their use to Nox isoforms that are expressed at the plasma membrane. Furthermore, since regulation of Nox activity occurs primarily on the cytoplasmic side of the membrane, the number of extracellular epitopes with regulatory effects are limited. This problem is highlighted by a recent study, showing that mABs generated against an intracellular component of Nox4 inhibit activity in a cell-free system, whereas an extracellular-targeted antibody is not inhibitory . Nonetheless, an antibody capable of interfering with electron transfer from the Nox heme group to molecular oxygen has the potential to inhibit ROS formation at the plasma membrane and within Nox-containing endosomes and may be a viable therapeutic option.
Small molecule inhibitors are organic compounds of less than 0.8 kD that bind to and alter the activity of biological molecules including DNA and protein. With the increased availability of large libraries of small molecules, investigators are developing protocols for high throughput screening to identify compounds that inhibit specific Nox isoforms in intact cells . Several inhibitors that are specific for Nox1 have been identified by this approach . It is important to note that the percent inhibition of Nox activity that is necessary to modify CVD development and preserve physiological activity is not known. For example, animal studies showing protection from CVD have relied primarily on mice with complete genetic deletion of an NADPH oxidase subunit [40, 48]. However, as little as 5–10% respiratory burst activity in neutrophils is sufficient to allow patients with CGD to remain healthy . It will be important to assess whether incomplete inhibition of Nox has similar protective effects and whether the small molecules effective in inhibiting Nox in vitro are capable of reducing CVD in vivo.
Aptamer technology is the newest approach and may be the most effective strategy in yielding Nox inhibitors that have all of the characteristics necessary to make them ideal therapeutics for CVD [144–146]. Aptamers are short (15–40 nucleotides) synthetic single-stranded DNA or RNA macromolecules that form intramolecular interactions, shaping these molecules into unique three-dimensional structures that can bind tightly to a particular target. Since there are a tremendous number of possible nucleotide sequences, allowing a vast diversity of possible three-dimensional structures, aptamers have the potential to bind to nearly any molecular target. SELEX (systematic evolution of ligands by exponential enrichment) is the process by which aptamers are subjected to repeated rounds of in vitro selection in order to select for aptamers that have interactions with specific targets such as proteins, nucleic acids, or cells. The advantages to this approach are: (1) These oligonucleotides recognize targets with specificities and affinities similar to those of antibody–antigen interactions, but unlike antibodies can be synthesized entirely in vitro. (2) Since these agents can be chemically synthesized, it is possible to substitute modified nucleotides for their native counterparts, to both promote nuclease resistance and improve their in vivo properties . (3) Aptamers can target specific protein isoforms, making it possible to target one Nox homolog over another . (4) Aptamers can specifically recognize post-translational modified sequences, which can be used to enhance selection for activated proteins. For example, some Nox and Nox subunit proteins are known to require phosphorylation for activation [122–124]. Therefore, peptides modified with constitutively phosphorylated residues to mimic activated Nox can be used as bait and nonphosphorylated peptides then used in a counter selection. (5) The potential of using aptamers as vascular therapeutics has already been demonstrated; several vascular proteins including VEGF (vascular endothelial growth factor) and factor IXa have been successfully targeted by aptamers and have undergone FDA approval for use as therapeutics [150, 151]. (6) Aptamers can be delivered in a cell type–specific manner by specifically selecting for aptamers capable of entering one cell or tissue type and using that pool to counter-select against aptamers that are capable of entering other cell types . Therefore, Nox-targeted aptamers could be selected for their ability to internalize into VSMCs but not ECs so as to inhibit VSMCs migration and proliferation while preserving EC function. This approach would provide a novel solution to the problem of late thrombosis associated with drug-eluting stents, which results in part from the inhibition of re-endothelialization caused by nonselective delivery of antiproliferative agents to both VSMCs and ECs. (7) Aptamers can be conjugated to both small-molecule inhibitors and to antibodies to improve the delivery of these agents to specific cell types . Therefore, aptamers can not only serve as Nox inhibitors, but also as delivery vectors for other types of Nox inhibitors. Aptamers have many unique properties that make them ideal as a novel strategy in the treatment of CVD. Aptamer technology is still in its infancy but has the potential to significantly alter the approach of treating multiple diseases, including CVD.