SOURCES OF VASCULAR NITRIC OXIDE AND REACTIVE OXYGEN SPECIES AND THEIR REGULATION

Abstract

Nitric oxide (NO) is a small free radical with critical signaling roles in physiology and pathophysiology. The generation of sufficient NO levels to regulate the resistance of the blood vessels and hence the maintenance of adequate blood flow is critical to the healthy performance of the vasculature. A novel paradigm indicates that classical NO synthesis by dedicated NO synthases is supplemented by nitrite reduction pathways under hypoxia. At the same time, reactive oxygen species (ROS), which include superoxide and hydrogen peroxide, are produced in the vascular system for signaling purposes, as effectors of the immune response, or as byproducts of cellular metabolism. NO and ROS can be generated by distinct enzymes or by the same enzyme through alternate reduction and oxidation processes. The latter oxidoreductase systems include NO synthases, molybdopterin enzymes, and hemoglobins, which can form superoxide by reduction of molecular oxygen or NO by reduction of inorganic nitrite. Enzymatic uncoupling, changes in oxygen tension, and the concentration of coenzymes and reductants can modulate the NO/ROS production from these oxidoreductases and determine the redox balance in health and disease. The dysregulation of the mechanisms involved in the generation of NO and ROS is an important cause of cardiovascular disease and target for therapy. In this review we will present the biology of NO and ROS in the cardiovascular system, with special emphasis on their routes of formation and regulation, as well as the therapeutic challenges and opportunities for the management of NO and ROS in cardiovascular disease.

I. INTRODUCTION

Nitric oxide (NO) is a small free radical molecule with critical signaling roles. The discovery of the function of NO in the vascular endothelium as endothelium-derived relaxing factor led to the awarding of the 1998 Nobel Prize to Drs. Furchgott, Ignarro and Murad (36, 324, 449, 491, 716). The functions of NO in mammalian systems extend beyond vascular signaling and are relevant in all organ systems, including but not limited to neuronal signaling, and host defense (448, 659, 738).

A number of oxygen-related species of high chemical reactivity are referred to as reactive oxygen species (ROS). These include oxygen radicals and peroxides, such as superoxide (O₂^·−) and hydrogen peroxide (H₂O₂), nitrogen radical species, such as NO and nitrogen dioxide (NO₂^·), and other species, such as peroxynitrite (ONOO⁻) and hypochlorite (ClO⁻). The species containing nitrogen are often treated separately as reactive nitrogen species (RNS). It is worth indicating that despite being long considered toxic species, most of these molecules have been shown to exert important signaling functions (249, 778, 937, 960). Therefore, the role of many of these molecules in health and disease is related to their production rates, steady-state concentrations, and the ability of the cellular antioxidant systems to modulate their activity.

In general, dysregulated production of ROS/RNS, as is the case for NO, leads to oxidative stress and deleterious consequences for living systems. However, as pointed out above, these molecules often have important signaling roles at low concentrations. For instance, the differences in response to NO at varying concentrations have attracted considerable attention. It has been shown that low levels (pM/nM) are physiological and related to the activation of high affinity primary binding targets such as soluble guanylyl cyclase (sGC) and cytochrome c oxidase (433, 863). An emerging paradigm proposes that intermediate levels (50–300 nM) can activate a range of positive and negative responses from wound healing to oncogenic pathways (938). Higher concentrations of NO (>1 μM) can lead not only to oxidative stress but also nitrative and nitrosative stress via the generation of peroxynitrite and nitrosating species (411, 412, 938, 939), and in combination with oxygen, can trigger posttranslational modification of proteins, lipids, and DNA (277, 433, 938).

The production of adequate levels of NO in the vascular endothelium is critical for the regulation of blood flow and vasodilation, as will be discussed at length in this review (299, 565, 573, 600, 786). In this context, it has become increasingly appreciated that oxygen levels can impact the oxidation/reduction properties of different proteins and regulate NO levels (FIGURE 1) (367, 578, 595, 931). For example, nitric oxide synthases (NOSs) produce NO using l-arginine and molecular oxygen (O₂) as substrates. Thus, under hypoxic or anoxic conditions, the generation of NO via NOS is compromised. However, a number of proteins that are involved in oxidative processes at basal oxygen levels can become de facto reductases as oxygen is depleted. The biological role of this transition is particularly prominent in the case of heme- and molybdopterin-containing proteins such as hemoglobin (Hb), myoglobin (Mb), and xanthine oxidase (XO) (185, 575, 578, 862, 880, 945, 990). Clinical intervention through these pathways continues attracting intense research.

FIGURE 1.

Oxygen and oxidoreductase enzymes regulate nitric oxide (NO) homeostasis. The gradient in the concentration of oxygen shifts the function of globins from oxidizing, NO-scavenging proteins to nitrite-reducing, NO-generating proteins.

The concept of oxygen-regulated oxidation and reduction processes in the metabolism of NO is not only relevant to NO generation but also to the scavenging of NO in the vasculature (FIGURE 1). In this regard, the role of globins like α-Hb and cytoglobin (Cygb) as catalytic NO dioxygenases that scavenge NO is a topic of current research (25, 593, 594, 898).

The translation of our knowledge about the biology of NO and ROS has encountered significant challenges. For instance, initial attempts to enhance NO levels using NO donors or supplementing with NOS substrates to reverse endothelial dysfunction have had limited success. The use of general antioxidants for the treatment of oxidative stress has also failed in most cases. Recent advances in the field have provided many clues on why these approaches have been unsuccessful. We will discuss these and other relevant physiological and pathophysiological issues and indicate how advances in basic biochemistry of the generation of NO/ROS have evolved our understanding and set new directions in the field.

In this review, we will present the biology of NO and ROS in the cardiovascular system with special emphasis on their routes of formation, chemistry, mode of action, and dysregulation in vascular disease. The formation pathways of NO and the mechanisms of NO signaling will be discussed in sect. II. The proteins and biological systems generating hydrogen peroxide and superoxide are treated in sect. III. Other ROS of particular relevance in the vascular system are discussed in sect. IV. Section V will study the cross-talk between NO- and ROS-generating systems. Finally, in sect. VI, we will discuss the current challenges and opportunities for the treatment of cardiovascular disease through the regulation of NO and ROS levels in pathological conditions.

II. NITRIC OXIDE GENERATION AND VASCULAR FUNCTION

The generation of sufficient NO levels to regulate the resistance of the blood vessels and hence the maintenance of an adequate blood flow is critical to the healthy performance of the vasculature (277, 299, 573, 600, 786). A number of mechanisms are involved in both the generation of NO and the response to NO signaling in the vasculature. In this section, we will overview the mammalian proteins involved in the generation and sensing of NO in the cardiovascular system. The production of NO in basal conditions is largely regulated by the activity of endothelial NOS (eNOS) in the vascular endothelium (324, 449, 716). Nevertheless, the contribution of other agents cannot be ignored. For instance, nitrite reduction by heme proteins can mediate hypoxic vasodilation and other physiological responses (367, 591, 968); neuronal NOS (nNOS) and inducible NOS (iNOS) can provide compensatory NO generation or exacerbated RNS synthesis (386, 444, 547, 704, 727). In this study, we will review NO synthases and other NO-generating biological systems.

A. Oxygen-Dependent Nitric Oxide Synthesis

The canonical pathways of NO formation rely on the specialized NOS enzymes. NOS are dimeric, multidomain enzymes that synthesize NO from molecular oxygen and l-arginine and use iron protoporphyrin-IX (heme), tetrahydrobiopterin (BH₄), FAD, flavin mononucleotide (FMN), and NADPH as cofactors (204, 904). The architecture of these proteins is complex, with a oxygenase/heme domain that binds BH₄ and heme, in which the oxidation of arginine to NO using molecular oxygen takes place, and a reductase domain evolutionarily related to cytochrome P450 reductase (CYPOR) that binds the cofactors FAD and FMN (204, 377). The reductase domain uses NADPH as electron source to reduce the FAD and FMN and finally transfer the electrons to the oxygenase/heme domain (FIGURE 2). Between both domains, there is a calmodulin (CaM) binding domain that regulates the electron shuttling between the reductase and oxygenase domains. To add to this complexity, the electron transfer between domains occurs between the reductase domain of one monomer and the oxygenase domain of the other monomer; thus, only the dimer is able to generate NO catalytically.

FIGURE 2.

Architecture of nitric oxide synthases (NOS). A: the arrangement of the domains in the NOS monomer. The oxygenase/heme domain (red) is connected to the reductase domain by a flexible linker, containing a calmodulin (CaM) binding sequence. The reductase domain includes a flavin mononucleotide (FMN)-binding domain (orange) that shuttles electrons from NADPH/FAD to the heme group and a FAD-containing domain (yellow) that uses NADPH as an electron source. B: the binding of CaM (blue) to NOS promotes electron transfer from the FMN domain of one monomer to the heme domain of the other monomer. C: three-dimensional structure model of NOS. The figure is assembled from the separated structures of the human endothelial NOS oxygenase domain (PDB:4D1O) (574), the CaM binding peptide bound to CaM (PDB:2N8J) (740), and the structure of the neuronal NOS reductase (PDB:1TLL) (338).

The role of NOS enzymes in vascular function and pathobiology has been extensively studied (277, 299, 569, 790). In this study, we will summarize the most relevant concepts about the function and regulation of the NOS isoforms in health and disease states.

NOS enzymes comprise three main isoforms: nNOS (NOS I), iNOS (NOS II), and eNOS (NOS III). Among these isoforms, eNOS is the enzyme more abundant in vascular endothelial cells and has the most significant impact on vascular function, with a variety of pathologies related with eNOS dysfunction. However, the specific roles of iNOS and nNOS in vascular disease are not to be ignored and will be also discussed below.

1. eNOS

eNOS is the constitutive NOS form in endothelial cells (ECs), and as such, it is the main contributor to vascular NO levels in physiological conditions (446). For example, the infusion of a competitive inhibitor of eNOS into the human systemic or coronary circulation decreases basal blood flow by ~25% (146, 722). Basal levels of NO produced by eNOS not only regulate blood flow but tonically inhibit platelet activation and the expression of inflammatory adhesion molecules on endothelium. The role of eNOS in the maintenance of optimal cardiovascular function is also highlighted by the experiments with eNOS-deficient mice. eNOS deletion has some limited effects on mouse blood pressure and vasodilation, as adaptive processes increase the production of nNOS and prostaglandins (444, 913). However, eNOS^−/− mice show impairments in angiogenesis and wound healing (563). In the apolipoprotein E (ApoE) knockout (KO) atherosclerosis model, addition of the eNOS deletion accelerates the atherosclerotic process (533). Notably, overexpression of eNOS is also detrimental (708), highlighting the delicate balance of eNOS function.

Although eNOS is predominantly expressed in the endothelium (Table 1), and the ECs are proposed to regulate NO signaling in the vasculature, a number of recent studies using chimeric cross bone marrow transplant mouse models have shown that the red blood cell also expresses a functional eNOS that contributes in part to systemic blood pressure responses (181, 536, 1025). This erythrocyte eNOS can contribute to endocrine NO signaling to modulate both blood pressure and myocardial injury and appears to be tonically regulated by arginase through the availability of the eNOS substrate l-arginine (1044).

Table 1.

Tissue distribution of NOS isoforms in the cardiovascular system

	Tissue	Cell Types	Cellular Location	References
nNOS (NOS1)	Heart	ECs	Plasma membrane	114, 126, 599, 841, 977, 1037
	Lung	VSMCs	Caveolae
	Blood vessels	Adventitial fibroblasts	Sarcoplasmic reticulum
		Cardiomyocytes
iNOS (NOS2)	Heart	ECs	Plasma membrane	126–128, 521, 525, 604, 688, 698, 867, 903, 984, 1064
	Lung	VSMCs	Phagosomes
	Blood vessels	Adventitial fibroblasts	Golgi
		Cardiomyocytes	Mitochondria
		Macrophages and other leukocytes
eNOS (NOS3)	Heart	ECs	Plasma membrane	126, 181, 283, 358, 518, 819, 878, 1025
	Lung	VSMCs	Lipid rafts
	Blood vessels	Cardiomyocytes	Caveolae
		Erythrocytes
		Platelets

EC, endothelial cell; eNOS, endothelial NOS; iNOS, inducible NOS; NOS, nitric oxide synthase; nNOS, neuronal NOS; VSMC, vascular smooth muscle cell.

eNOS is constitutively expressed; a number of posttranslational modifications impact the ability of eNOS to produce NO, from changes in the NO production rates to, in some cases, the complete blockade of NO synthesis (296). The relative presence of these modifications is critical to enzyme activity. General measurements of eNOS protein such as the determination of monomer/dimer ratio by Western blot or a single phosphorylation site status show a necessarily simplistic view of the activity of eNOS in cells. A deeper analysis of the main regulatory factors of eNOS activity, including individual phosphorylation sites, other posttranslational modifications, and the levels of the cofactors and substrates, with special interest on tetrahydropterin/dihydropterin (BH₄/BH₂) ratios and l-arginine/asymmetric dimethylarginine (ADMA) concentrations are necessary for a better assessment of the eNOS function in vivo and the assessment of endothelial dysfunction conditions. In this section, we will describe the regulation of eNOS activity by posttranslational modifications; the role of the cofactors BH₄ and l-arginine and their counterparts BH₂ and ADMA, and the role of arginases and l-arginine transport metabolons will be treated in sect. IIIB. A detailed study on other regulatory mechanisms, including protein/protein interactions, shear stress, and other mechanisms, has been provided by other reviews (59, 296, 299, 870).

a) phosphorylation.

A number of amino acids have been shown to be phosphorylated in human eNOS (296, 669). Because of their impact in eNOS activity, the more relevant phosphorylation sites are Thr495, Ser1177 (Ser1179 in bovine eNOS) and Tyr657. Other phosphorylation sites include Tyr81, Ser114, Ser615, and Ser633 (Table 2).

Table 2.

Posttranslational modification sites in human eNOS

Site	Kinase	Effect of Modification	References
Phosphorylation
Thr495	PKC	Impairs CaM binding	297, 407, 646
	AMP kinase	Decreases NO synthesis rates
	Constitutive	Can cause uncoupling through the reductase domain
Ser1177	Akt	Activates electron transfer through the reductase domain	74, 236, 321
	PKA	Increases NO synthesis rates
	AMP kinase
	CaMKII
Tyr81	Src	Slight increase in activity	320, 322
		May be involved in Ca2+/CaM sensitivity
Ser114	Constitutive		74, 327
Ser615	PKA	No change in NO synthesis	74, 647
	Akt	May modulate protein/protein binding interactions
Ser633	PKA	No change in NO synthesis	74, 647
	PKG
Tyr657	PYK2	Decrease in NO synthesis	292
Glutathionylation
Cys382		Not determined	160
Cys689		Decrease flow through the reductase domain	162
		Decreased NO synthesis
		Can cause uncoupling through the reductase domain
Cys908		Decrease flow through the reductase domain	162
		Decreased NO synthesis
		Can cause uncoupling through the reductase domain

CaM, calmodulin; NO, nitric oxide

The activation of eNOS requires Ca²⁺ and CaM; however, changes in resting Ca²⁺ concentrations are not strictly necessary to regulate NO synthesis by eNOS. In turn, CaM affinity to eNOS is regulated via phosphorylation of Thr495. Thr495 is located in the CaM-binding site of eNOS, and the phosphorylation of Thr495 impairs CaM binding, thus blocking Ca²⁺/CaM-dependent activation of the enzyme (741). In resting endothelial cells, Thr495 is generally phosphorylated (297). The phosphorylation has been attributed to protein kinase C (PKC) (297, 646), and the residue is dephosphorylated by protein phosphatase 1 (PP1) (646). The equilibrium between phosphorylated and dephosphorylated Thr495 is mainly modulated by the changes in intracellular Ca²⁺ levels. Thus, bradykinin and Ca²⁺ ionophores promote Thr495 dephosphorylation and eNOS activation. Ser1177 phosphorylation activates electron flow through the reductase domain and, hence, increases NO synthesis in functional eNOS. Ser1177 is located in the C-terminal portion of eNOS in a helical C-terminal tail element that blocks flavin reduction and regulates the conformational equilibrium of the reductase domain (404, 631). The mechanism is similar to that described for nNOS Ser1412 (7, 947). In resting cells, Ser1177 is usually not phosphorylated. Phosphorylation is induced by a variety of signals, and the kinases involved are dependent on the inducing factor (Table 2). Some agonists like bradykinin and Ca²⁺ ionophores can induce Thr495 dephosphorylation via PP1 and Ser1177 phosphorylation via activation of CaMKII at the same time. Shear stress activates Ser1177 via PKA-dependent phosphorylation. VEGF, insulin, and estrogens promote Ser1177 phosphorylation via Akt kinase.

Tyr657 is located in the FMN domain of eNOS in close proximity to the FMN cofactor. The mutation of Tyr657 (292) causes complete loss of NO synthesis and l-citrulline formation, suggesting a blockade of the intramolecular electron transfer. Thus, Tyr657 phosphorylation effectively inactivates eNOS. The phosphorylation of Tyr657 is catalyzed by proline-rich tyrosine kinase 2 (PYK2). This kinase is activated in endothelial cells by several stimuli including angiotensin II, oxidative stress, and insulin (296). The detrimental role of this phosphorylation on NO synthesis has spurred interest in the pharmacological inhibition of PYK2 for the treatment of cardiovascular disease (94, 628, 869, 983).

b) glutathionylation.

Recent reports indicate that eNOS can also be modified posttranslationally by S-glutathionylation (162). At least three residues have been shown to be susceptible: Cys689, Cys908 (162), and Cys382 (160) (Table 2). The process can be reversed by glutaredoxin-1 and thioredoxin (160, 907). The reaction increases eNOS uncoupling and the formation of superoxide with diminished NO synthesis activity. Glutathionylation appears to be a detrimental modification caused by oxidative stress conditions (162, 522) and is very sensitive to the ratio of oxidized and reduced glutathione (160).

c) s-nitrosation.

S-nitrosation of eNOS has also been reported (777). This reaction is also dependent on eNOS myristoylation and Ser1177 phosphorylation and involves the nitrosation of the Zn-binding cysteines (Cys96 and Cys101 in the bovine eNOS) (271, 272). As a consequence of S-nitrosation, the formation of the eNOS dimer is blocked, resulting in the loss of NO synthesis. Denitrosation of Cys96-NO and Cys101-NO restores eNOS activity. Additional studies using mutant eNOS Cys96Ser and Cys101Ser, which cannot be S-nitrosated, indicate that the enzymatic activity is not altered by treatment with NO donors, confirming a role for the modification of these thiols in the regulation of eNOS activity (271).

d) other posttranslational modifications.

eNOS can be acylated at different residues. The myristoylation of the N-terminal glycine targets eNOS to the membrane (284), whereas the palmitoylation of Cys15 and Cys26 targets eNOS specifically to caveolae (82, 336). These modifications are not expected to change the intrinsic NOS activity but are important to modulate eNOS localization and function.

Hyperglycemia can result in N-acetyl glycosylation of Ser1177. This modification results in a nonphosphorylable Ser1177 and limits eNOS response to agonists and vasorelaxation (257, 280, 1035).

In summary, we want to remark that the generation of NO by eNOS in optimal conditions, at least as studied in vitro with saturating concentrations of coenzymes and substrates and absence of deleterious modifications, is a well-balanced process with a very limited production of superoxide as a side reaction. However, a number of circumstances can lead to changes in the NO generation cycle and trigger the production of superoxide instead of NO. These processes are generally termed “eNOS uncoupling” as the consumption of reducing equivalents by eNOS is no longer coupled to the formation of NO and is instead driven to the generation of superoxide from molecular oxygen. This process is largely deleterious and has been linked to endothelial dysfunction and other vascular pathologies. eNOS uncoupling will be discussed in sect. IIIB along with other mechanisms of superoxide generation.

2. iNOS

Unlike eNOS and nNOS, which are constitutively expressed enzymes regulated via CaM binding and posttranslational modifications, iNOS has a much higher CaM affinity, so it binds CaM at very low Ca²⁺ concentrations, and thus, its activity is not regulated by Ca²⁺/CaM but mainly at the level of gene transcription (169). The kinetic parameters of iNOS lead to a catalytic activity that generates higher NO levels than eNOS and nNOS, is less sensitive to NO-dependent autoinhibition, and generates higher levels of other nitrogen species, such as nitrate (816, 904).

The expression of iNOS is usually limited to airway epithelium and neuronal cells (Table 1), where it is highly expressed and active under basal conditions (604, 984, 1064), and activated macrophages and hepatocytes, where it is expressed in the setting of inflammatory stimuli (688).

Although healthy cells in the vasculature do not present significant levels of iNOS, several pathologic processes show increased iNOS activity in blood vessels (386, 704, 727). As iNOS can generate higher NO levels than eNOS, this iNOS activation leads to excess NO and severe impairment of vascular function. This effect is mediated by several pathways including continuous activation of sGC and competition for BH₄ with eNOS (386). Overall, the excess NO limits the response of blood vessels to vasodilators and decreases NO sensitivity (263).

As pointed out, iNOS expression is prevalent in macrophages. In these cells, iNOS induction is necessary for the generation of high levels of NO in the phagosome and bacterial lysis and, therefore, is a basic tool for the innate immune response (519, 688). However, in pathological situations, increased NO synthesis can be deleterious (e.g., in macrophages recruited to the atherosclerotic plaque). The high rate of NO production by iNOS, together with superoxide formation from iNOS or other sources, can lead to the production of peroxynitrite (see sect. IVA) (222, 624, 981, 1034), a very toxic oxidant and nitrating agent (78, 709). Studies in which iNOS deletion was added to the ApoE^−/− atherosclerosis model indicate improved cardiovascular function in the ApoE^−/−iNOS^−/− mice compared with ApoE^−/− controls, indicating a detrimental role of iNOS in atherosclerosis progression (532).

Finally, it should be noted that iNOS-derived NO can also have positive effects in certain conditions. iNOS activation has been identified as a mechanism of protection against ischemia/reperfusion damage. This observation appears to be related to a preconditioning effect caused by increased levels of NO (109, 483, 579, 1009).

3. nNOS

Similar to eNOS, nNOS is a constitutively expressed form of NOS activated by CaM binding. NO synthesis rates for nNOS are generally higher than eNOS, but also tightly regulated by Ca²⁺/CaM (unlike iNOS), and nNOS is also very susceptible to feedback inhibition by NO (4, 902). This enables nNOS to produce NO in a pulsatile manner instead of generating sustained low levels. These features appear to be more related to its role in synaptic transmission (811, 812). Although nNOS is commonly found in neurons (120), it is also found in other tissues including vascular smooth muscle cells (VSMCs), adventitial fibroblasts, ECs, and cardiomyocytes (114, 126, 599, 841, 1037) (Table 1). Expression levels of nNOS in the vasculature are lower than for eNOS; however, there is increasing evidence of important functions for nNOS-derived NO in vascular physiology (188).

Early experiments in rodents indicated an important role of nNOS in the regulation of cerebral blood flow (731). Selective inhibition of nNOS causes increases in blood pressure and decreased response to acetylcholine in normotensive rats (139). In eNOS^−/− mice, nNOS can be activated by shear stress, partially compensating the loss of eNOS in the vasculature (444, 547). Experiments with nNOS KO mice indicate increased neointima formation in carotid artery ligation and balloon injury models (666), indicating that nNOS appears to limit vascular injury independently of eNOS. To investigate the role of nNOS in atherosclerosis, the nNOS^−/− deletion was incorporated in mice carrying the ApoE^−/− deletion. The double deletion indicates a protective role for nNOS in the development of atherosclerosis, as the mice carrying both deletions showed accelerated progression of the disease (534).

The effects of nNOS in the human vascular system have been demonstrated by studies with nNOS-specific inhibitors. These works indicate that NO synthesis from nNOS has definite roles in the vasodilatory response, including, but not limited to, microvascular tone and coronary flow (498, 843, 844).

Another relevant pathway for nNOS-dependent NO signaling not mediated by sGC is the formation of S-nitrosothiols (462). This function has special relevance in brain function in health and neurodegenerative diseases (686). A more specific description of S-nitrosation and its signaling potential is presented in sect. IID.

4. Pharmacological regulation of NOS enzymes

There have been significant efforts to develop specific NOS inhibitors for research and pharmacological purposes (17, 752).

a) nonisozyme-specific nos inhibitors.

Early research indicated that analogs of the substrate l-arginine had inhibitory properties on the three NOS isoforms. Among these, N^G-monomethyl-l-arginine (l-NMMA) and N^G-nitro-l-arginine (l-NNA) and its precursor N^G-nitro-l-arginine methyl ester (l-NAME) were characterized as general NOS inhibitors and continue to be broadly used in research, especially l-NAME (FIGURE 3).

FIGURE 3.

Chemical structure of the nitric oxide synthase (NOS) substrate l-arginine, tetrahydrobiopterin, and selected NOS inhibitors.

b) 7-nitroindazole.

7-Nitroindazole (7-NI) (FIGURE 3) is one of the earliest inhibitors showing isoform specificity. Although it was shown that all three isoforms can bind 7-NI with very similar affinity (17), in vivo studies showed inhibition of nNOS without significant effects in blood pressure, a surrogate of eNOS inhibition (664). It appears that 7-NI may have differential properties in cell permeability, with limited uptake in endothelial cells (402). Thus, although 7-NI cannot be accurately described as a specific nNOS inhibitor, it can behave as such in vivo.

c) 1400w.

N-[3-(aminomethyl)benzyl] acetamidine (FIGURE 3) is a specific inhibitor of iNOS (344). It remains one of the most widely used NOS inhibitors in research because of its cell and tissue permeability. Its selectivity seems related to the irreversible effect on the faster reacting iNOS, whereas the inhibition on nNOS and eNOS is reversible. A similar effect is observed in N(5)-(1-iminoethyl)-l-ornithine (l-NIO) (FIGURE 3) (279).

d) GW273629 and GW274150.

These sulfur-substituted acetamidine amino acids (FIGURE 3) are specific inhibitors for iNOS (16). With safer toxicity profiles than 1400W, GW273629 and GW274150 have been used in clinical trials for migraine (437, 965) (NCT00242866; NCT00319137). Although both compounds were ineffective, it is not clear if this is due to pharmacokinetic issues (965).

e) vaS203.

The pterin 4-amino-tetrahydrobiopterin (VAS203) (FIGURE 3) is a BH₄ analog that can inhibit all NOS proteins by replacing the BH₄ cofactor (1008). VAS203 has shown efficacy in the treatment of traumatic brain injury and has been used in phase II clinical trials (897) (NCT02012582).

The structural similarities between isoforms (and particularly between eNOS and nNOS) have made the development of specific eNOS and nNOS inhibitors particularly challenging. Structure-based inhibitors have allowed the development of new inhibitors with improved isoform specificity (337, 752). Cell permeability and other pharmacokinetic considerations have precluded the clinical use of novel inhibitors; however, newer compounds have been developed; for instance, novel-specific nNOS inhibitors have been tested in animal studies (29, 254, 255, 1051). The further development of new clinically available specific inhibitors for NOS, in particular for the iNOS and nNOS isoforms, is to be expected.

B. Nitrite-Dependent Nitric Oxide Synthesis

Despite early indications to the contrary (124, 329), nitrate and nitrite have been often overlooked as inert NO oxidation metabolites circulating in plasma and in cells. However, during the last two decades, a novel paradigm of nitrite as a source of NO, especially in hypoxia, has emerged (184, 185, 216, 364, 366, 369, 600, 603, 861, 968). Indeed, it has become increasingly accepted that the routes for the production of NO in vivo are not only oxidative (as in NOS) but also reductive (particularly by nitrite-reducing proteins such as globins and molybdopterin enzymes). These pathways can work synergistically to maintain NO levels in response to changes in oxygen tension. We foresee this theme of synergistic oxidative and reductive pathways to be potentially relevant to the generation of other reactive species in health and disease.

A few years after the discovery of the role of NO as the endothelial-derived relaxing factor, increasing evidence supported the presence of nonenzymatic NO generation (89, 605, 1075). These studies and observations of arterial-to-venous gradients of nitrite in the human circulation indicated a potential role for nitrite as a NO source in vivo (369). Subsequent studies have extended this notion to a more global paradigm that encompasses the role of nitrate and nitrite as part of a cycle of NO generation that includes both enzymatic and nonenzymatic processes.

Nitrate and nitrite can enter the circulation through the dietary intake of nitrate-rich foods, particularly leafy green vegetables (1002). In addition, NO generated from NOS enzymes is eventually oxidized to nitrate and nitrite as well. Mammals do not possess efficient systems for nitrate reduction, but oral commensal bacteria have enzymatic nitrate reduction pathways able to generate significant amounts of nitrite from nitrate reduction in the saliva (467, 602). Nitrate is not only acquired from dietary sources but is also concentrated from the blood into the saliva via the salivary glands, pumped through the sialin transporter (761). After consumption of nitrate-rich foods (beet root, spinach, kale, etc.), the concentration of nitrate and nitrite in saliva can reach levels as high as 10 mM and 2 mM, respectively (601). A portion of this nitrite is eventually absorbed into the circulation. Studies using antiseptic killing of the mouth microbiome, or systemic NOS inhibition, suggest that about half of basal plasma nitrite comes from the salivary reduction of dietary nitrate and the other half from the oxidation of NO produced mainly by eNOS (485, 557, 763, 865). NO oxidation to nitrite is proposed to occur via the oxidase activity of plasma ceruloplasmin (865).

Once nitrite accumulates in the plasma, the reduction of nitrite to NO in the vasculature is mediated by several proteins, including deoxygenated Hb (deoxyHb) (185). Increased physiological levels of nitrite in plasma from a nitrate-rich diet can lower blood pressure (554). The effect is dependent on the generation of nitrite from oral bacteria, as antiseptic mouthwash can eliminate both the increase in circulating nitrite and the blood pressure decrease (485). Subsequent studies have further validated the link between nitrite levels and improved cardiovascular function (486, 502).

Numerous metal-containing proteins have been shown to catalyze the reduction of nitrite to NO. These proteins generally contain a heme or molybdopterin cofactor. The presence of many of these proteins in diverse components of the vasculature, including blood cells, blood vessels, and heart tissue, makes them relevant for the generation of NO in vascular biology. In the subsequent sections, we will discuss the known mechanisms for nitrite reduction that are relevant for cardiovascular function. We also note that, to date, existing evidence indicates a role for Hb, Mb, and XO on biological NO generation, whereas the role of the other possible mechanisms in vivo are still a matter of discussion.

1. Inorganic nitrite reduction

Along with the role of different proteins in the reduction of nitrite, a number of nonenzymatic processes can reduce nitrite to NO in vivo. In general, these systems require concerted electron and proton donation, which is optimal at lower pH and hypoxic conditions, thus making them particularly relevant in ischemic events (1074, 1075). The simplest mechanism involves the disproportionation of nitrite, a process accelerated at acidic pH as it requires the protonation of nitrite to nitrous acid (HNO₂). Two molecules of nitrous acid, then disproportionate, to form N₂O₃ and water (Eqs. 1–4)

$${\text{2NO}}_2^{-}{\text{+\hspace{0.17em}2H}}^{\text{+}}\to 2{\text{HNO}}_2$$ (1)

$${\text{HNO}}_{\text{2}}+{\text{HNO}}_{\text{2}}\to {\text{N}}_{\text{2}}{\text{O}}_{\text{3}}+{\text{H}}_{\text{2}}\text{O}$$ (2)

$${\text{N}}_{\text{2}}{\text{O}}_{\text{3}}\to {\text{NO}}^{\text{·}}+{\text{NO}}_{\text{2}}{}^{\text{·}}$$ (3)

$${\text{2NO}}_2^{-}+{\text{2H}}^{\text{+}}\to {\text{NO}}^{\text{·}}+{\text{NO}}_{\text{2}}{}^{\text{·}}+{\text{H}}_{\text{2}}\text{O}$$ (4)

This process is prevalent in very low pH conditions, as in the stomach (89, 605). In fact, this mechanism has been shown to be responsible for nitrite-dependent increases in gastric mucosal blood flow (98) and may relate to the role of nitrite in host defense as acidified nitrite is a potent antimicrobial agent (99, 261).

Several molecules present in cells have been shown to reduce nitrite to NO. Ascorbic acid (vitamin C) is a widely present cellular antioxidant that can catalyze the reduction of nitrite to NO in vitro (205). This effect is also observed with in vivo infusions of ascorbic acid and nitrite (217). Polyphenols can be assimilated through the diet and, like vitamin C, can reduce nitrite to NO (325, 733, 834).

Dietary interventions have indicated improved vascular function linked to the consumption of polyphenols from different sources, including cocoa and dealcoholized red wine (60, 168, 225, 417, 834). Combined use of polyphenols and nitrite appears to have additive effects in vascular protection (791). The effects of these dietary components can be, in part, due to the interaction with different enzymatic systems (30, 558, 824), but the link between these compounds and NO formation deserves further investigation (600, 789).

2. Heme proteins

Heme proteins have emerged as critical parts in the generation of NO under hypoxia. The specific role of globins in nitrite reduction and NO homeostasis has been studied in recent reviews (39, 770, 931). It should be noted that competing reactions, namely the scavenging of NO by the ferrous heme [rate constants in the order of 1.7 × 10⁷ to 1.5 × 10⁸ M⁻¹s⁻¹ (180, 967), Eq. 6] and the rate of NO dioxygenation [rate constants in the order of 3.4 × 10⁷ to 8.9 × 10⁷ M⁻¹s⁻¹ for mammalian Hb/Mb (264, 341), Eq. 8] greatly decrease the amount of bioavailable NO generated by the reactions of nitrite with heme proteins (931). However, nitrite-dependent NO generation in vivo catalyzed by Hb and Mb is well documented (183, 185, 365, 862, 975); it is very likely that the measurement of this NO generation is due to the fact that the high concentrations of Hb and Mb (in the mM range) can overcome the low efficiency of the nitrite reduction reaction.

a) hb.

The ability of deoxyHb to reduce nitrite has been extensively documented (124, 248, 366, 447). Initial reports by Brooks demonstrated that deoxyHb reacts with nitrite, generating nitrosyl Hb and methemoglobin, consistent with electron transfer from a ferrous Hb to nitrite (124). The studies by Huang et al. (445) confirmed the overall stoichiometry of two molecules of deoxyHb generating one molecule of nitrosyl Hb and one molecule of methemoglobin (Eq. 7), consistent with electron transfer from a ferrous Hb to nitrite and indicated the critical influence of pH in the process. By studying the same reaction using alkyl nitrites, which do not show a pH dependence in their reaction rates with deoxyHb, it was established that the reaction of nitrite requires a proton and thus formally requires nitrous acid (247, 248). This property also makes the reaction faster in low pH conditions. The overall process can be written as (Eqs. 5–7)

$$\text{DeoxyHb\hspace{0.17em}}\left({\text{Fe}}^{\text{2+}}\right)+{\text{H}}^{+}+{\text{NO}}_2{}^{-}\to \text{MetHb}\left({\text{Fe}}^{3+}\right)+\text{NO}+{\text{OH}}^{-}$$ (5)

$$\text{DeoxyHb}\left({\text{Fe}}^{2+}\right)+\text{NO}\to \text{Hb}-\text{NO}$$ (6)

$$2\text{DeoxyHb}\left({\text{Fe}}^{2+}\right)+{\text{H}}^{+}+{\text{NO}}_2{}^{-}\to \text{MetHb}\left({\text{Fe}}^{3+}\right)+\text{HbNO}+{\text{OH}}^{-}$$ (7)

This scheme has been found to apply to virtually all the heme protein–mediated nitrite reduction reactions studied to date, including Hb (248, 447), Mb (247, 862), neuroglobin (Ngb) (932, 945), Cygb (182, 571, 780), cystathionine β-synthase (CBS) (149, 350), globin X (184), plant and cyanobacterial Hb (phytohemoglobins) (905, 946), flavohemoglobin (340), heme-albumin (43), cytochrome c (41), protoglobin (40), and microperoxidase-11 (42).

The reaction of deoxyHb with nitrite shows an increase in the instantaneous reaction rate as the reaction progresses (447). Studies with T-state and R-state Hb–stabilizing compounds have unambiguously identified the reason for this change is the shift from T-state to R-state during the course of the reaction, as the NO formed in the reaction binds to the unreacted deoxyHb to stabilize the R-state. This process is referred to as R-state autocatalysis. The rate constants for the reaction of Hb with nitrite change from 0.12 M⁻¹s⁻¹ for T-state Hb to 6.0 M⁻¹s⁻¹ for R-state Hb (Table 3) (447). As the distribution of R-state and T-state Hb in vivo are regulated by oxygen concentration, this phenomenon has important relevance for the reaction in vivo. In fact, the reaction of Hb with nitrite has been modeled mathematically at different oxygen tensions. The combination of a faster rate at higher oxygen, but increased availability of deoxyHb at low oxygen, leads to bell-shaped dependence of the observed rates versus the oxygen concentration with the maximal rates of NO generation around 50% Hb oxygen saturation, consistent with experimental data (367, 368, 794).

Table 3.

Rate constants for the reaction of nitrite with selected deoxygenated heme proteins

Protein	K, M−1s−1	Reference
Hemoglobin
Human (T-state)a	0.12	447
Human (R-state)a	6	447
Myoglobin
Horse hearta	2.9	945
Sperm whalea	5.6	945
Neuroglobin
Human (S-S)a	0.12	945
Human (SH)a	0.062	945
Cytoglobin
Humanb	0.14	571
Humanc	1.14	182
Cystathionine-β-synthase
Humand	0.6	350

^a100 mM sodium phosphate, pH 7.4, 25°C; ^b100 mM sodium phosphate, pH 7.0, 25°C; ^c100 mM phosphate, pH 7.4, 37°C; ^d100 mM HEPES, pH 7.4, 37°C.

b) physiological implications of hb-dependent nitrite reduction.

The functional implications of the Hb-dependent nitrite reductase reaction remains a subject of intense research, particularly in the context of viable physiological models of erythrocytic NO transport (366, 424, 591). There is consensus in the field that NO-erythrocytic interactions regulate vascular function, particularly that oxygen desaturation of red blood cell Hb stimulates increased vascular NO bioavailability, leading to NO-dependent hypoxic vasodilation (266). However, the molecular mechanism of this regulation is more contentious. Three models by which this erythrocytic regulation of NO production occurs have been proposed: 1) hypoxic release of ATP from the red blood cell (882) that stimulates endothelial NO production by binding to endothelial purinergic receptors, 2) the S-nitrosation of cysteine 93 of the β-chain of Hb (SNO-Hb) (866), and 3) the reduction of nitrite by deoxyHb (as described above) (425).

Understanding the role of each of these potential mechanisms is an ongoing area of interest. The SNO-Hb hypothesis proposes that SNO-Hb is formed when Hb is oxygenated (R-state) in the lungs and remains stable. Once Hb becomes deoxygenated (T-state), the SNO-Hb reacts with existing thiols to release a vasodilatory signal (471, 728, 886). In the last 25 years since this proposal, numerous groups have debated key elements of the hypothesis including the mechanism of SNO-Hb formation and release and physiological levels of SNO-Hb. Additionally, a genetic murine model in which the cysteine 93 of β-Hb was replaced with alanine (and in which SNO-Hb formation was abolished) demonstrated no impairment in hypoxic vasodilation, questioning the role of SNO-Hb in physiologic hypoxic vasodilation (454). Importantly, more recent studies have shown that the β-93 cysteine KO mouse shows more cardiac injury in ischemic conditions, suggesting that SNO-Hb may play a more prominent role in cardiac rather than vascular function (1060).

With regards to the nitrite reduction hypothesis, human studies show a significant gradient of nitrite from the arterial to venous circulation, concomitant with the production of NO (detected as iron-nitrosyl Hb) in the venous circulation (185, 369). In vitro experiments in isolated aortic rings demonstrate that nitrite mediates vasodilation in conditions of Hb deoxygenation (185, 196). Consistent with these experiments, infusions of physiological nitrite concentrations mediate vasodilation in humans, which is enhanced during exercise and not affected by NOS inhibition. This effect is accompanied by iron-nitrosyl Hb formation, consistent with nitrite reductase chemistry (185).

A central controversy relevant to all theories of erythrocytic NO transport is the question of how NO generated can escape the red cell without reacting with Hb. Although theoretical calculations suggest that this type of NO escape should be limited, accumulating studies by many research groups now demonstrate the production of bioavailable NO generated from the incubation of erythrocytes with nitrite (185, 196, 883, 975). The production of NO gas has been measured by chemiluminescence in vitro from the reaction of deoxygenated Hb, Mb, or erythrocytes with nitrite. More physiological biosensor experiments demonstrate that incubation of nitrite with erythrocytes can induce cGMP production and inhibit activation in platelets coincubated with this reaction (591, 883, 975). The mechanism for NO release remains unclear but has been postulated to involve the formation of NO⁺ (684) or N₂O₃ as intermediates (69). Additionally, NO escape may relate to the localization of its formation, particularly at the surface of the erythrocyte (813) or via concerted reactions of NO and autoxidation of oxygen to form NO and NO₂ (N₂O₃) at partial oxygen saturations (i.e., the reactions of nitrite with both oxygenated Hb and deoxyHb at around 50% Hb oxygen saturation) (384). More work is needed to understand the biophysics of NO signaling from the erythrocyte during nitrite reactions with deoxygenated red blood cells and Hb.

c) mb.

The physiological relevance of the reaction of Hb with nitrite suggested that similar processes could involve the closely related globin, Mb. The ability of Mb to catalyze nitrite reduction to NO was first confirmed in vitro. Consistent with the monomeric structure of Mb and its lack of allosteric regulation, the reaction rate is constant through the reaction, with a bimolecular rate constant similar to that of R-state Hb (Table 3) (447). In vitro studies of isolated mitochondria in the presence of Mb and nitrite have shown that Mb-dependent NO is bioavailable and can inhibit mitochondrial respiration in a similar manner to a conventional NO donor (862). In the cardiovascular system, the heart is the most relevant organ for the function of Mb as a nitrite reductase. In a murine myocardial infarction (MI) model, nitrite has been shown to decrease infarct size, and this effect was lost in Mb KO mice, which cannot reduce nitrite to NO (774). Through this mechanism, cardiac Mb is thought to be critical for the reduction of circulating nitrite accumulated through diet or through the therapeutic process of remote ischemic preconditioning, which mediates cardioprotection (267, 775).

d) ngb.

Ngb is a recently discovered six-coordinate globin that is evolutionarily related to Hb and Mb (134). The physiological function of Ngb is unknown, although because of its generally low concentration in cells (with the exception of the retina in the eye), it is probably not related to oxygen transport and storage as Hb and Mb are (38, 39, 133). It was recently demonstrated that, as observed with other globins, deoxygenated Ngb can also reduce nitrite to NO (932, 945). Furthermore, this process is redox regulated by the redox state of two surface thiols. Thus, oxidation of two cysteine residues to form an intramolecular disulfide bond doubles the rate of nitrite reduction (Table 3). This redox transition occurs in a physiologically relevant range and can be controlled by the cellular GSSG/GSH ratios (945). It appears that Ngb is not present in vascular cells, except for sympathetic nerves (898). The reaction of Ngb with nitrite may have implications for microcirculation in the brain, where several studies have found a vasoprotective effect of Ngb (140, 503, 917).

e) cygb.

Like Ngb, Cygb is a six-coordinate globin discovered in the early 2000s (132, 493, 951). Cygb is ubiquitous in human tissues, mainly found in fibroblasts and cells of related lineage such as osteoblasts and chondroblasts, but also in other cell types, including neurons (403, 687). Interestingly, it has been identified in VSMCs (398, 571, 594). Cygb can reduce nitrite to NO at rate constants similar or slightly higher than Ngb (182, 571) (Table 3). In addition, the oxygenated form of Cygb reacts with NO to form nitrate in a NO dioxygenation reaction that scavenges NO (339) (Eq. 8)

$${\text{Fe}}^{2+}-{\text{O}}_2+\text{NO}\to {\text{Fe}}^{3+}+{\text{NO}}_3{}^{-}$$ (8)

This reaction is common to most heme proteins (264, 341, 342, 931). In practice, this indicates that Cygb expression can regulate the diffusion of NO and possibly prevent excessive vasodilation at high levels of NO, not unlike the proposed role of α-Hb in vascular endothelial cells (898). This role has been proposed for Cygb in the vascular endothelium (398, 593, 594). It should be noted that the scavenging of NO is more efficient at high O₂ tensions, which favor the formation of the ferrous/oxygenated heme species, whereas nitrite reduction would be prevalent when the oxygen concentration is low, and the ferrous/deoxygenated species is more abundant. Thus, both processes can synergize to regulate the concentration of NO (593, 594). The presence of an efficient reduction system for Cygb in smooth muscle cells via cytochrome b₅ and cytochrome b₅ reductase further supports the possible relevance of these processes in vivo (25, 593).

f) cbs.

CBS is a pivotal enzyme for the metabolism of homocysteine (649). Mutations in the CBS gene are associated with hereditary homocystinuria (531). CBS binds heme and 5′-pyridoxal phosphate. The 5′-pyridoxal phosphate group is critical for the reaction of homocysteine with serine or cysteine, whereas the role of the heme domain in the catalysis of CBS in unknown, but this heme appears to be functional and redox active (868). The CBS-mediated reduction of nitrite to NO by the CBS heme has been recently reported (149, 350). The reaction proceeds with rate constants similar to these of T-state Hb, Ngb, or Cygb (Table 3) (149, 350). The presence of CBS in endothelial cells (809) suggests that CBS could play a role in vascular nitrite reduction in vivo.

g) nos.

As NOS proteins contain a heme group, it is conceivable that they can catalyze the reduction of nitrite as observed in other heme enzymes. Indeed, in anoxic conditions, eNOS has been shown to catalyze nitrite reduction to NO (970). The contribution of eNOS to nitrite reduction could be significant in specific tissues, including red blood cells and kidney (654, 998).

3. Molybdoproteins

The relevance of molybdopterins in the generation of NO in the vasculature has been increasingly appreciated since this reaction was first described for XO (650, 1061). As the molybdopterin group does not react directly with oxygen or NO, these reactions that decrease NO release from heme proteins do not limit Mo-dependent nitrite reduction reactions. Thus, the reduction rates by molybdoproteins can be relevant in vivo at lower rate constant values as compared for heme protein reduction rates. The reduction of nitrite by all four mammalian molybdenum-containing enzymes has been characterized. Recent reviews on the nitrite metabolism by molybdenum enzymes are available (614, 616).

a) xo.

Mammalian molybdopterin-containing enzymes catalyze different oxidation-reduction reactions including, but not limited to, xanthine oxidation, sulfite oxidation, and drug metabolism processes (839, 840). To our knowledge, the ability of mammalian Mo-containing enzymes to reduce nitrite to NO was first described for XO (370, 578, 650, 1061). Earlier work had shown the ability of XO to also reduce nitrate to nitrite (50, 308). Subsequently, several reports have shown that all mammalian Mo-containing enzymes [XO, aldehyde oxidase (AO), sulfite oxidase, and the mitochondrial amidoxime-reducing component (mARC)] are able to reduce nitrite at different intrinsic rates (Table 4) (575, 880, 990).

Table 4.

Kinetic parameters for the reaction of nitrite with selected molybdopterin proteins

Protein	kcat, s−1	KM, mM	K, M−1s−1	Reference
Xanthine oxidase
Human, pH 7.4a	0.41	2.2	186	616
Human, pH 6.3b	1.2	0.67	1800	616
Rat, pH 7.4a	0.55	1.9	289	616
Rat, pH 6.3a	0.58	0.25	2300	616
Aldehyde oxidase
Human, pH 7.4a	0.47	4.1	115	616
Rat, pH 7.4a	0.67	3.6	186	616
Rat, pH 6.3a	0.66	0.43	1500	616
Sulfite oxidase
Human, pH 7.4c	0.002	1.6	1.3	990
Human, pH 6.5c	0.004	1.7	2.4	990
mARC-1
Human, pH 7.4d	0.1	9.5	11	880

^aReaction contains 50 μM of reductant (aldehyde); ^breaction contains 750 μM of reductant (aldehyde); ^creaction contains 5 μM of reductant (sulfite); ^dreaction contains 1 mM NADH, 0.2 μM cytochrome b₅ reductase and 2 μM cytochrome b₅.

Although the architecture of these enzymes is variable, existing evidence indicates that the reduction of nitrite takes place in the molybdopterin site, where Mo reduces nitrite to NO in a single electron transfer reaction similar to that of the heme proteins (615, 616) (Eqs. 9 and 10)

$${\text{Mo}}^{4+}+{\text{H}}^{+}+{\text{NO}}_2{}^{-}\to {\text{Mo}}^{5+}+\text{NO}+{\text{OH}}^{-}$$ (9)

$${\text{Mo}}^{5+}+{\text{H}}^{+}+{\text{NO}}_2{}^{-}\to {\text{Mo}}^{6+}+\text{NO}+{\text{OH}}^{-}$$ (10)

Depending on the redox potential of the enzyme, nitrite reduction may occur by either of the two reactions or only by Mo⁴⁺ (Eq. 9) pathways or one specific pathway (614, 616). The pH dependence of the reaction is not linear as in the case of the heme-dependent reactions, probably because of the protonation of important catalytic residues at low pH that offsets the effect of the increased proton concentration. These circumstances lead to a maximal rate for XO-dependent nitrite reduction around pH 6.3 (Table 4) (617). As noted above, unlike nitrite reduction by heme, the Mo cofactor does not bind NO. Thus, this reaction potentially has a higher yield of free NO.

In the vasculature, XO has been identified in the surface of epithelial cells, endothelial cells (5, 798, 979, 1010), and erythrocytes (357, 998). To date, XO is the most relevant nonheme nitrite reductase in vivo. Many studies have shown the relevance of XO in mammalian physiology (357, 568, 577, 578, 998). The use of specific XO inhibitors such as allopurinol and oxypurinol has allowed the determination of specific XO effects on nitrite reduction and NO metabolism (357, 568, 575, 998).

b) ao.

AO is a molybdenum-containing protein with high sequence and structural similarity to XO. Its function is yet unknown, although it has been related to the metabolism of retinoic acid, neurotransmitters, and a variety of xenobiotics (513, 756, 920). AO has been shown to reduce nitrite to NO, albeit a slower rate than XO (Table 4) (568, 575, 617, 1001). As observed for XO, the pH dependence of nitrite reduction shows a bell-shaped pattern with a maximal rate around pH 6.5 (617). The relevance of AO-dependent nitrite reduction has not been studied in so much detail as for XO, but depending on their tissue abundance and the availability of their reducing substrates, some studies suggest that the magnitude of AO activity in the vasculature can be similar to XO (568, 745, 1073).

c) sulfite oxidase.

The molydopterin-containg sulfite oxidase catalyzes the oxidation of sulfite to sulfate, preventing the accumulation of toxic levels of sulfite. Impaired sulfite oxidase function leads to severe neurological damage and death (510). Like XO and AO, the molybdopterin cofactor of sulfite oxidase is able to catalyze nitrite reduction to produce NO (Table 4) (990). Unlike other molydopterin proteins, in sulfite reductase only the Mo⁴⁺ center, and not the Mo⁵⁺ species, is able to catalyze the reaction (990). The reaction has a marked O₂ dependence with very low NO formation in the presence of O₂. This is most probably related to the fast oxidation of the Mo⁴⁺ species to Mo⁶⁺ by molecular oxygen as observed in other molybdopterin proteins (880, 990).

d) marc 1 and 2.

Two isoforms of the mARC are expressed in mammals (705). Sequence homology suggests that mARC proteins belong to the sulfite oxidase family. The structure of the mARC proteins has not been elucidated, but electron paramagnetic resonance data seem to support these similarities (1043). mARC proteins have a not yet understood physiological function. Existing evidence suggests that they may be involved in the detoxification of N-hydroxylated substrates; mARC enzymes are required for the detoxification of some hydroxylamines (706). mARC can also catalyze the reduction of the NOS reaction-intermediate N^G-hydroxy-l-arginine to l-arginine in a reaction that may be involved in NO biosynthesis (528).

Both mARC enzymes can reduce nitrite to NO in a process very sensitive to oxygen-mediated inhibition (Table 4) (880). As mARC enzymes can use the cytochrome b₅ reductase/cytochrome b₅ system as a source of electrons, the three proteins can form a mitochondrial metabolon for the reduction of nitrite to NO under hypoxic conditions (880).

4. Other proteins

a) carbonic anhydrase.

The reduction of nitrite by carbonic anhydrase has been described (1). Unlike heme- or molybdenum-containing proteins, the Zn²⁺ in the active site of the carbonic anhydrase is not redox active, thus invoking a nitrite anhydrase mechanism as described in Eqs. 1–4. Such reaction would be of notable interest in vascular physiology because of the ubiquitous presence of carbonic anhydrase in erythrocytes; however, this activity remains highly controversial (615).

b) cytochrome c.

The electron carrier cytochrome c has also been noted as a situational nitrite reductase. As cytochrome c is a six-coordinate heme protein with no distal site available for nitrite ligation, the reaction is only observable in conditions where the protein is partly unfolded, a situation that can be elicited by cardiolipin or other anionic phospholipids (477). In the presence of cardiolipin-containing liposomes, cytochrome c has been shown to catalyze nitrite reduction to NO in a reaction that may be prevalent in apoptotic processes (68).

c) cyp and cypor.

The heme P450 cytochromes (CYPs) are involved in a wide range of detoxification processes. CYP 2B4 and microsomal rat liver CYP fractions have been characterized as nitrite reductases (576). Notably, the associated CYP-reducing protein, CYPOR, can reduce nitrate to nitrite, forming a metabolon that can tentatively catalyze the complete process from nitrate to NO (576).

C. The NO Receptor sGC and the Modulation of Vascular Tone

1. Structure and function of sGC

The effects of NO in the vasculature at concentrations in the nM/pM range (397) are mainly mediated through the activation of the canonical NO receptor sGC (223, 662, 711, 751). sGC is a cytosolic, heterodimeric protein comprising two homologous subunits, α and β. Although at least two isoforms for either α and β subunit exist (termed α1, α2 and β1, β2) with different tissue distributions (129), the α1β1 heterodimer is the most common form. Each monomer contains four domains: an N-terminal heme-binding domain, a Per-Arnt-Sim (PAS) domain, a helical/coiled coil motif, and a C-terminal, guanylyl cyclase domain that catalyzes the formation of cGMP from GTP (FIGURE 4). It should be noted that despite the homology between the N-terminal domains, only the N-terminal domain of the β subunit binds heme (489). The heme domain in the β subunit is the NO-sensing center of sGC. Remarkably, unlike most other hemoproteins that bind oxygen, CO and NO, this heme group shows a high selectivity toward NO and can also bind CO weakly but shows no affinity toward molecular oxygen (625). When NO binds to the heme, changes in the heme coordination cause a conformational change in the N-terminal heme domain that relieves an inhibitory interaction between the heme domain and the catalytic C-terminal domain, activating cGMP synthesis (1017). The exact details of the conformational changes that mediate enzyme activation are not yet known and are the focus of ongoing structural studies (145, 313, 958).

FIGURE 4.

Architecture of soluble guanylyl cyclase (sGC). Top: arrangement of sGC domains in the sequence of α and β subunits. Bottom: model for the interaction of the N-termini domains of the α and β subunits of human sGC. Domains are colored according to the top panel. Catalytic GTP cyclase domains are not shown. The model for the human sGC protein is based on the model for Manduca sexta sGC derived from chemical cross-linking, small-angle X-ray scattering, and homology modeling (312, 313, 662).

The activation of sGC by NO leads to an increase in cGMP synthesis of at least two orders of magnitude (895). cGMP exerts a number of downstream signaling effects on phosphodiesterases such as PDE5 and other targets including cGMP-dependent kinases and cGMP-gated ion channels, eliciting a vasodilatory response (673, 674, 993).

Apart from changes in NO levels or downstream signaling, the activity of sGC itself is compromised in a number of pathological conditions (355, 363, 600, 853, 888, 1022). Situations directly related to sGC that can account for a decrease in sGC activity include changes in the expression of sGC, defective incorporation or loss of sGC heme, a deficient reduction of the sGC heme iron, leading to an NO unresponsive ferric enzyme, posttranslational modifications, ubiquitination and proteosomal degradation, and defects in cGMP synthesis.

Changes in sGC related to the mRNA transcript have been described. For example, expression of a variant α2 subunit (α2i) containing a 31 amino acid insertion has been shown in different tissues. This subunit can form a dimer with β1 but produces an inactive sGC form (80). Alternatively, several splice variants of the α1 and β1 monomers have been identified in vivo (626, 787, 852). These splice forms can be related to decreased sGC activity and appear to be more abundant in disease conditions such as aortic aneurysm (626).

The incorporation of the heme group in the sGC β1 monomer is regulated by the chaperone heat shock protein 90 (Hsp90) (95, 354, 356, 818). This process is also regulated by NO, with low NO doses improving Hsp90/sGC interaction and heme insertion, whereas high NO blocks heme insertion (354, 985). The relevance of heme-free sGC in the context of cardiovascular disease is not well characterized. Several lines of evidence suggest the presence of pools of heme-free sGC in vivo (354, 356, 799, 943), and it is reasonable to expect that the exacerbated ROS conditions that exist in the origin and development of many cardiovascular pathologies can promote heme oxidation and subsequent heme loss (312, 363, 888). Thus, a plausible therapeutic strategy can target the inactive, heme-depleted sGC by use of agonists (activators) that can bind to the empty heme site activating the enzyme. Some examples of these drugs are discussed in the next section.

In addition to the presence of heme-free protein, it is increasingly appreciated that a pool of ferric sGC can also exist in smooth muscle cells (641, 769, 891). As noted, the oxidation of the heme iron is probably the intermediate step in the generation of free-heme sGC (312, 363, 888). The generation of ferric sGC is one of the mechanisms of sGC desensitization, and this form of the protein, whereas nonresponsive to stimulator drugs, can be rescued by sGC activators (302, 891). Recent work indicates that CYB5R3 (methemoglobin reductase) may mediate the reduction of ferric sGC to NO-responsive ferrous sGC in cells (769). These findings open the possibility of novel pathways to regulate sGC responses in blood flow regulation and for targeted therapy in diseases associated with oxidative stress and endothelial dysfunction.

Posttranslational modifications of sGC have been also described. Prolonged exposure to NO has been long known to cause a decrease in sGC activity, usually characterized as sGC desensitization (85, 671, 831). This process is related to some of the tolerance profiles observed for NO donor treatments (459). sGC desensitization is probably due to different concurrent mechanisms, including S-nitrosation and heme iron–nitrosylation. Recent reports indicate that several cysteine residues in sGC can undergo S-nitrosation, leading to decreased activity (93, 822, 823). The formation of a heme iron–nitrosylated form of sGC, which has decreased guanylyl cyclase activity, has been reported (953). Ubiquitination of sGC targets the protein for proteosomal degradation, a process that can be inhibited by sGC activators (643). Further research on these sGC modifications, and interventions to prevent their occurrence, is ongoing (92).

2. Pharmacological regulation of sGC

Because of the pivotal role of sGC in the regulation of the NO response, substantial efforts have been made to develop pharmacological regulators of sGC function. In theory, direct targeting of sGC offers notable advantages, bypassing the complex mechanisms that may limit NO generation and availability under pathological conditions (FIGURE 5). Two main groups of compounds, activators and stimulators, have been developed with different mechanisms of action. In both cases, the target of the drugs appears to be the NO-sensing, heme domain of the β subunit (662). A large number of sGC agonists have been developed over the last three decades; we discuss below a few of the more widely studied compounds. For a more complete view of the development of these and other sGC-related drugs, the reader is referred to more specific reviews (274, 302, 888).

FIGURE 5.

Soluble guanylyl cyclase (sGC) function in healthy and endothelial dysfunction states. Oxidative stress conditions cause oxidation of BH₄ to BH₂, superoxide production in the endothelial cells, and promote oxidation and heme loss in smooth muscle cell sGC.

a) stimulators.

Stimulators are compounds that enhance the activity of functional sGC and require the presence of ferrous (Fe²⁺) heme-bound sGC. These compounds can cooperate with endogenous NO-mediated activation.

The benzyl indazole YC-1 (FIGURE 6) was the first developed sGC stimulator (524). The effects of YC-1 on sGC activation were found to be synergistic with NO donors and required heme-bound sGC (310, 436, 896). The compound presented some limitations and a lack of specificity with cGMP-independent effects and inhibition of PDE5 (309, 326, 991). Based on YC-1 structure, a number of optimized compounds have been developed such as BAY 41–2272 and BAY 41–8543 (889). These compounds are more specific than YC-1 and despite some poor pharmacokinetic properties, have been extensively used in research (105, 353, 670, 760, 764, 852, 889).

FIGURE 6.

Chemical structure of selected soluble guanylyl cyclase (sGC) stimulators and activators.

Further studies led to the development of another YC-1–related compound, BAY 63–2521 (riociguat) (83, 655) (FIGURE 6). Riociguat (commercialized as Adempas) was approved in 2013 by the Food and Drug Administration for the treatment of pulmonary arterial hypertension and chronic thromboembolic pulmonary hypertension (351, 352).

Another second-generation stimulator compound is BAY 1021189 (vericiguat) (301) (FIGURE 6), which has completed Phase IIB trials (287, 349, 744) and is currently in Phase III clinical trials for the treatment of heart failure with reduced ejection fraction (301) (NCT02861534).

B) activators.

sGC activators are compounds that can trigger the catalytic activity of sGC independently of the redox state of the sGC heme or even in the absence of bound heme. In general, these compounds work as heme analogs, replacing the heme group and triggering a similar conformational change to that of NO-bound heme in the native enzyme, thus activating cGMP formation (302).

High-throughput screening assays led to the discovery of a novel series of sGC regulators that do not require the sGC heme. The first developed drug in this category was the compound BAY 582667 (cinaciguat) (890) (FIGURE 6). Apart from the recovery of sGC synthesis, it was noted that BAY 582667 could limit the degradation of heme-oxidized and heme-free sGC (643). Cinaciguat showed promising results in animal studies and acute decompensated heart failure (553), but the trials were stopped because of the development of low blood pressure (269).

Another sGC stimulator that has reached clinical use is HMR 1766 (ataciguat) (827) (FIGURE 6). Phase II clinical trials are currently evaluating the effect of ataciguat on aortic valve calcification (NCT02481258).

D. S-Nitrosothiols and the Modulation of Vascular Function

1. S-nitrosation and vascular function

Although activation of sGC is considered the predominant mechanism of NO-dependent signaling, it is now recognized that posttranslational modification of cysteine residues by S-nitrosation comprises a significant alternative mechanism of NO signaling. The first protein shown to be S-nitrosated was albumin (885), and its existence was first proposed to represent a reservoir of NO bioactivity. However, S-nitrosated albumin is now considered to be a product of more detrimental NO “scavenging” by the protein (332, 956). It is now well recognized that a plethora of proteins of various function and localization are S-nitrosated physiologically or in pathological conditions. For example, as described above in sect. IIB2, S-nitrosation of Hb is an active area of research and SNO-Hb has been considered a reservoir of NO activity. In this regard, decreased levels of SNO-Hb have been reported in conditions such as pulmonary hypertension and hypothesized to contribute to disease pathogenesis (586, 729).

Accumulating studies demonstrate that beyond simply representing a reservoir of NO activity, S-nitrosation is a mechanism of enzymatic regulation. S-nitrosation of specific cysteine residues can modulate enzymatic activity and function. One prototypical example of this paradigm is the type 2 ryanodine receptor (RyR2). The RyR2 is a Ca²⁺ release channel that releases Ca²⁺ from the sarcoplasmic reticulum to mediate cardiac excitation/contraction coupling. The magnitude and duration of Ca²⁺ release from through the RyR2 determines contractility of the myocyte. Although the RyR2 contains many surface-exposed thiols, physiological S-nitrosation of specific cysteine residues activate the protein and contribute to maintainance of healthy excitation contraction coupling (1038). A lack of this S-nitrosation has been associated with cardiac arrhythmias as well as heart failure (131, 372, 373). Similar to the RyR2, a multitude of enzymes have now been shown to be activated or inhibited by S-nitrosation. Although review of all these proteins is beyond the scope of this manuscript, the S-nitrosation of mitochondrial complex I and its cardioprotective signaling is discussed in sect. IIIC3, and the S-nitrosation of eNOS as a regulator of its activity is outlined in sect. IIA1. The role of S-nitrosation in regulating cardiovascular function has been reviewed elsewhere (586, 677, 914).

In the following section, we review the chemical mechanisms by which S-nitrosothiols are formed. It is important to note that physiological levels of S-nitrosothiols are determined not only by the rate of S-nitrosation but also by protein denitrosation. Although S-nitrosothiols degrade through transnitrosation reactions (outlined below) and interaction with other low molecular–weight thiols, the existence of denitrosating enzymes has been reported. The best characterized enzyme in this class is S-nitrosoglutathione reductase (GSNOR), a member of the aldehyde dehydrogenase enzyme group, which catalyzes the denitrosation of glutathione and thus regulates the equilibrium of S-nitrosation (592).

2. Formation of S-nitrosothiols

The formation of S-nitrosothiols is, on occasion, described as the reaction of NO with a thiol (R-SH) group to generate a protein-bound nitrosothiol (R-SNO). However, it should be noted that NO is not a nitrosating species and does not react directly with thiols. Formation of S-nitrosothiols involves the reaction of the deprotonated thiol (R-S⁻) and a formal nitrosonium ion (NO⁺) (Eq. 11)

$${\text{R-S}}^{-}+{\text{NO}}^{+}\to \text{R-SNO}$$ (11)

In physiological conditions, the thiolate is generally provided by glutathione or cysteine, yielding GSNO or Cys-NO, respectively. The nature of the nitrosating species can vary, generating different possible mechanisms for the reaction. Most of these routes require the formation of the nitrogen dioxide radical intermediate (NO₂^·) (Eqs. 12a, 12b)

$${\text{NO}}^{·}+{\text{O}}_2\leftrightarrow {\text{O}}_2{\text{NO}}^{·}$$ (12a)

$${\text{O}}_2{\text{NO}}^{·}+{\text{NO}}^{·}\to 2{\text{NO}}_2{}^{·}$$ (12b)

The reaction of NO₂^· with a thiolate in the presence of NO can generate the S-nitrosothiol via a thiyl (RS^·) radical (Eqs. 13a, 13b)

$${\text{NO}}_2{}^{·}+{\text{R-S}}^{-}\to {\text{R-S}}^{·}+{\text{NO}}_2{}^{-}$$ (13a)

$${\text{R-S}}^{·}+{\text{NO}}^{·}\to \text{R-SNO}$$ (13b)

The reaction of NO₂^· with another molecule of NO₂^· or NO generates dinitrogen tetroxide (N₂O₄) or dinitrogen trioxide (N₂O₃), respectively, which are strong nitrosating species and can react directly with the thiolate (691, 692, 1018) (Eqs. 14a–d)

$${\text{NO}}_2{}^{·}+\text{NO}\leftrightarrow {\text{N}}_2{\text{O}}_3$$ (14a)

$${\text{NO}}_2{}^{·}+{\text{NO}}_2{}^{·}\leftrightarrow {\text{N}}_{\text{2}}{\text{O}}_{\text{4}}$$ (14b)

$${\text{N}}_{\text{2}}{\text{O}}_{\text{3}}+{\text{R-S}}^{-}\to \text{R-SNO}+{\text{NO}}_2{}^{-}$$ (14c)

$${\text{N}}_{\text{2}}{\text{O}}_{\text{4}}+{\text{R-S}}^{-}\to \text{R-SNO}+{\text{NO}}_3{}^{-}$$ (14d)

Peroxynitrite (sect. IVA), formed by the reaction of NO and superoxide, is another source of NO₂^·, either through homolytic decomposition, via peroxynitrous acid, into NO₂^· and OH^· radicals (Eqs. 15a–b) or via reactions with CO₂ (Eq. 15c) or metal centers (Eq. 15d)

$${\text{ONOO}}^{-}+{\text{H}}^{+}\to \text{ONOOH}$$ (15a)

$$\text{ONOOH}\to {\text{NO}}_2{}^{·}+{\text{OH}}^{·}$$ (15b)

$${\text{ONOO}}^{-}+{\text{CO}}_2\to {\text{NO}}_2{}^{·}+{\text{CO}}_3{}^{-·}$$ (15c)

$${\text{ONOO}}^{-}+{\text{\hspace{0.17em}M}}^{n+}->{\text{\hspace{0.17em}NO}}_2{}^{·}+{\text{\hspace{0.17em}M}}^{(n+1)}=\text{O}$$ (15d)

These reactions leading to the formation of NO₂^· are well characterized (123, 548, 585, 1018). Global kinetic analysis of the reactions involved suggests that the main source of NO₂^· in vivo may be the decomposition of peroxynitrite (585). This result is consistent with the unfavorable rates for the direct formation of NO₂^· from oxygen and NO in vivo, (1018) although the exact mechanisms of formation remain a topic of current research (123, 872).

An additional pathway involves the formation of a ferric nitrosyl heme which can be in transient equilibrium with a ferrous-nitrosonium species (Eqs. 16a–c)

$${\text{Fe}}^{3+}+\text{NO}\leftrightarrow {\text{Fe}}^{3+}-\text{NO}$$ (16a)

$${\text{Fe}}^{3+}-\text{NO}\leftrightarrow {\text{Fe}}^{2+}-{\text{NO}}^{+}$$ (16b)

$${\text{Fe}}^{2+}-{\text{NO}}^{+}+{\text{R-S}}^{-}\to {\text{Fe}}^{2+}+\text{R-SNO}$$ (16c)

Finally, the reaction of an existing S-nitrosothiol with a thiolate group can result in transnitrosation (Eq. 17)

$$\text{R-SNO}+\text{R}’\text{-}{\text{S}}^{-}\leftrightarrow \text{R}\text{-}{\text{S}}^{\text{-}}+\text{R}’\text{-}\text{SNO}$$ (17)

Transnitrosation is a fundamental reaction of S-nitrosothiols and allows for the formation of GSNO or S-nitrosated proteins to mediate the formation of other S-nitrosothiols and probably serve as a reservoir for S-nitrosothiols in the cellular environment. Altogether, we want to note that the formation of the S-nitrosothiols is a complex process, and existing evidence indicates the requirement of significant concentrations of NO and usually the presence of superoxide and/or peroxynitrite.

The formation of S-nitrosothiols and their functional roles in vivo is a growing field, with protein S-nitrosation identified in most physiological pathways. It should be considered that the stability and precision of this modification is under intense study to determine its parallels with other protein modifications such as phosphorylation (1023, 1024). Recent studies of the proteome of eNOS^−/− mice compared with wild type mice identify a more limited set of S-nitrosated proteins (125, 378). Altogether, it appears that the detection of S-nitrosated proteins in the presence of exogenous NO or R-SNO species can lead to nonphysiological S-nitrosation and more sensitive methods, monitoring protein S-nitrosation levels in vivo are necessary for accurate detection of these modifications.

III. SUPEROXIDE AND HYDROGEN PEROXIDE GENERATION AND VASCULAR FUNCTION

The presence of ROS in the vasculature has historically been considered detrimental and a consequence of abnormal generation and/or an inability of the endogenous reductant and antioxidant systems to scavenge these reactive species. The excess of superoxide, or other ROS, in the cell causes the modification of different parts of the cellular machinery and is a known player in the genesis and progression of cardiovascular disease (227, 408, 409, 570, 672, 700, 893). Conversely, the signaling properties of many of these molecules are increasingly recognized. Indeed, the generation of superoxide and hydrogen peroxide, among others, has been shown to mediate critical processes such as angiogenesis, hypoxic adaptation, and energy homeostasis (96, 249, 291, 720, 778, 825, 960). In this section, we will discuss the main sources of superoxide and hydrogen peroxide generation in the vasculature and their endogenous and pharmacological regulation. Among these, NADPH oxidases (NOXs) appear to be the most important sources quantitatively. In healthy situations, the generation of ROS by eNOS, mitochondria, or XO is limited, and in the case of mitochondria, it is probably more related to signaling pathways (289, 290). However, there is significant evidence that not only NOXs but also the generation of ROS by uncoupled eNOS, mitochondria, or XO has particular relevance in the origin and development of cardiovascular disease.

A. NOXs

NOXs are multiprotein enzyme complexes that catalyze the electron transfer to molecular oxygen to form ROS, more commonly superoxide, but also hydrogen peroxide (79, 252, 383, 546, 912). To date, seven isoforms have been described [NOX 1–5 and dual oxidases (DUOX) 1 and 2]. Each isoform consists of a core, catalytic subunit with a transmembrane domain and a variable number of regulatory subunits. The core subunit defines the name of the NOX complex. Thus, NOXs complexes show differences in their molecular architecture, activator proteins, localization, and function. NADPH is the preferred substrate for all NOX isoforms, with the NOX2 isoform being particularly NADPH specific (48, 63, 177, 1047).

The putative mechanism of all NOX proteins can be described as an electron transfer chain in which electrons are transferred from NADPH to the FAD cofactor in the flavoprotein/dehydrogenase domain and from FAD to the heme moieties (Eqs. 18, 19), where molecular oxygen is reduced to superoxide. The specific mechanism of superoxide generation is still unknown. According to one hypothesis, the reduced (Fe²⁺) heme moiety binds oxygen to form a ferrous/oxy species that decays to ferric heme and superoxide (Eq. 20) (245, 980). Routes that do not involve the formation of a stable heme/oxy complex are also possible (Eq. 21) (456). Notably, the reaction of NOX1–3 and 5 generates superoxide, whereas NOX4 and DUOX1/2 produce hydrogen peroxide (26, 936). Existing evidence suggests that the primary species generated is superoxide in all isoforms, and the formation of hydrogen peroxide is mediated by the peroxidase-like domain in DUOX or by the dismutation of superoxide molecules in NOX4 (Eq. 22). The mechanism of NOX4 is still controversial, but elegant work by Takac et al. indicates that the E-loop (sequence between helices V and VI) of NOX4 shows notable differences with the other NOX isoforms (925). This loop in NOX4 seems to provide steric hindrance to slow down superoxide release, forcing two superoxide molecules to react with each other in a reaction that appears to be catalyzed, in part, by a histidine residue (925).

$$\text{NADPH}+\text{FAD}+{\text{H}}^{+}\to {\text{NADP}}^{+}+{\text{FADH}}_2$$ (18)

$${\text{FADH}}_2+2{\text{Fe}}^{3+}\to \text{FAD}+2{\text{Fe}}^{2+}$$ (19)

$${\text{Fe}}^{2+}+{\text{O}}_2\to {\text{Fe}}^{2+}-{\text{O}}_2$$ (20a)

$${\text{Fe}}^{2+}-{\text{O}}_2\to {\text{Fe}}^{3+}+{\text{O}}_2{}^{-·}$$ (20b)

$${\text{Fe}}^{2+}+{\text{O}}_2\to {\text{Fe}}^{3+}+{\text{O}}_2{}^{-·}$$ (21)

$$2{\text{O}}_2{}^{-·}+2{\text{H}}^{+}\to {\text{O}}_2+{\text{H}}_2{\text{O}}_2$$ (22)

The architecture of the NOX/DUOX core subunits has been described in detail elsewhere (13, 79, 383, 546) (FIGURE 7). NOX core subunits 1–4 are structurally similar with a N-terminal, membrane domain and a C-terminal flavoprotein, dehydrogenase domain with high sequence homology to the ferredoxin/NADP reductase family (490). NOX5 has a similar architecture but has an additional N-terminal domain before the transmembrane domain with four EF hand motifs that are responsive to Ca²⁺. The reductase domain of NOX5 also includes a CaM-binding motif not found in the NOX1–4 proteins. When Ca²⁺ is bound, the EF hands fold over the reductase CaM element to relieve autoinhibition (63, 944). DUOX1 and 2 have a similar architecture to NOX5 and include an additional transmembrane domain connected to a putative extracellular peroxidase domain (262). The ability of this domain to act as a peroxidase is still under debate (638, 639).

FIGURE 7.

Structure of the NADPH oxidase (NOX) core protein membrane and cytosolic domains. The sequence comprises six transmembrane helices and a cytosolic flavoprotein/dehydrogenase domain. The transmembrane helices are indicated by colors: I, red; II, orange; III, yellow; IV, green; V, cyan; and VI, dark blue. The location of loops B and C is indicated. The dehydrogenase domain is shown in purple. The cofactors heme and FAD are shown as red and yellow sticks, respectively. Figure was drawn with PyMOL based on the structures for the transmembrane (PDB: 5O0T) and dehydrogenase (PDB: 5O0X) domains of NOX5 (613).

Recently, the crystal structure of a NOX core subunit (NOX5) has been determined (613). This structure is consistent with previous evidence suggesting the presence of six transmembrane α helices with five connecting loops (FIGURE 7) (613). The transmembrane domain includes four conserved histidines that serve as binding sites for two heme moieties bound between helices III and V (288, 613), and the flavoprotein domain binds the cofactor FAD and includes a binding site for NADPH (49, 480, 613, 980).

NOX isoforms are regulated by a large number of protein/protein interactions, and most isoforms are not functional without the assembly of the NOX protein with several additional proteins. NOX1–4 are associated to another membrane component, p22^phox (24). Although there is no structural data available for p22^phox, sequence analysis and computational models have been used to elaborate a consensus prediction of a three-helix transmembrane domain and a cytoplasmic C-terminal domain (637). Other models are consistent with the presence of two transmembrane helices instead of three (135, 451, 930).

In the human vasculature, the most relevant isoforms appear to be NOX1, NOX2, NOX4, and NOX5 (252, 253, 555). The relative architecture of these three proteins and their activator proteins are shown in FIGURE 8. NOX1 activation requires p22^phox, NOX-organizer 1 (NOXO1), NOX-activator 1 (NOXA1), and the GTPase Rac1 (24, 108, 348). The activator protein p47^phox can replace NOXO1 in a noncanonical mode (61). Recent studies indicate a possible role of EBP50 in NOX1 assembly (14). NOX2 activation requires p22^phox, p47^phox, p67^phox, p40^phox, and Rac1 (276). As described for NOX1, a noncanonical mode of activation can use NOXO1 instead of p47^phox (927). The activation mechanisms of NOX4 are not completely understood. NOX4 appears to be constitutively active in the presence of p22^phox without the participation of cytosolic proteins (24, 627). This, in part, can be explained by the constitutive activity of its dehydrogenase domain (694). However, a number of NOX4-interacting proteins have been recently identified. The protein polymerase delta–interacting protein 2 (poldip2) seems to exert regulatory effects on NOX4 (212, 607), although poldip2 is involved on many other pathways (316). The membrane protein Toll-like receptor 4 (TLR4) has been shown to increase NOX4 activity (86, 724), and Tks4/5, a homolog of NOXO1, also regulates NOX4 activity (231). Additional interactions with protein disulfide isomerase have also been reported (465). In addition, gene splicing also appears to modulate NOX4 activation and play important roles in cardiovascular disease (971).

FIGURE 8.

Putative assembly architecture of NADPH oxidase (NOX)1, NOX2, NOX4, and NOX5 and their regulatory proteins.

The regulation of NOX expression is modulated by a variety of agonist species and transcription factors. We will only highlight some of the main agonists for each isoform; a more detailed view is covered in recent reviews (79, 555, 737, 912).

1. NOX2

NOX2 is the main NOX isoform in neutrophils and macrophages where it mediates the oxidative burst (198, 845). Genetic defects in the main constituents of NOX2 can trigger an inability of the immune system cells to kill invading bacteria and fungi. This condition, characterized by an increased susceptibility to infections, is known as chronic granulomatous disease. Most mutations causing the disease (around 90%) are found in the membrane component NOX2 (also known historically as gp91^phox) and the interacting protein p47^phox. Mutations in p22^phox, p67^phox, and Rac2 have been also identified (428, 438).

NOX2 expression has been also reported in other cell types including endothelial cells and adventitial fibroblasts (376, 382, 657, 714, 715, 736, 876) (FIGURE 9, Table 5). Thus, the effects of NOX2 activity depend on different cell types and have important links to the immune response (392, 488). The current knowledge about NOX enzymes in the cardiovascular system indicates that NOX2 is usually the main player in the development of pathological conditions and increased NOX2 activity is linked to vascular disease (142, 252, 478).

FIGURE 9.

Cellular distribution of NADPH oxidase (NOX) species in the blood vessels. Predominant forms in each cell type are shown in bold.

Table 5.

Tissue distribution of NOX isoforms in the cardiovascular system

	Tissue	Cell Types	Cellular Location	References
NOX1	Heart	ECs	Plasma membrane	9, 423, 431, 652, 876, 879, 909
	Blood vessels	VSMCs	Endosomes
		Adventitial fibroblasts	Caveolae
		Cardiomyocytes
NOX2	Heart	ECs	Plasma membrane	9, 166, 376, 382, 427, 468, 473, 580, 657, 714, 715, 736, 742, 876, 950, 964
	Lung	VSMCs	Phagosomes
	Blood vessels	Adventitial fibroblasts	Endoplasmic reticulum
		Cardiomyocytes
		Hematopoietic stem cells
		Macrophages neutrophils, monocytes, dendritic cells
NOX4	Heart	ECs	Plasma membrane	9, 10, 100, 118, 166, 179, 201, 228, 265, 347, 423, 431, 540, 541, 556, 736, 742, 876, 906, 964
	Lung	VSMCs	Focal adhesions
	Blood vessels	Adventitial fibroblasts	Endoplasmic reticulum
		Cardiomyocytes	Nucleus
		Cardiac fibroblasts Hematopoietic stem cells	Mitochondria
NOX5	Heart	ECs	Plasma membrane	81, 166, 201, 396, 469, 494, 848
	Blood vessels	VSMCs	Endoplasmic reticulum
		Cardiomyocytes
		Cardiac fibroblasts

EC, endothelial cell; NOX, NADPH oxidase; VSMC, vascular smooth muscle cell.

In the vasculature, different effectors can increase NOX2 expression and activity (555, 737). A large number of cytokines and hormones have been shown to induce NOX2 expression, including angiotensin II (658, 714, 802), thrombin (725), and TNF-α (275, 964). Circulating lipids also increase NOX2 production in ECs in vitro (159, 208, 856, 1063). The role of endothelial NOX2 in the development of cardiovascular disease is controversial. Some studies have described that endothelial overexpression of NOX2 increases endothelial dysfunction (88, 676). However, in other cases, an increased expression of endothelial NOX2 has not been linked to pathological conditions even though an increased production of superoxide and macrophage recruitment was observed (243, 443).

Global KO studies in mice indicate a negative role of NOX2 in disease conditions. The deletion of the organizer subunit p47^phox ameliorates the hypertensive response to angiotensin II, decreases endothelial superoxide formation and eNOS uncoupling, and improves survival after MI (238, 549, 550). The mice combining the mutation of p47^phox−/− with ApoE^−/− appear to be protected from atherosclerotic lesions (253) compared with ApoE^−/− mice (64, 976), although some reports do not find significant differences in atherosclerosis progression or lesion size (443). It should be noted that the use of p47^phox−/− as a surrogate for the NOX2 deletion is partly limited by the involvement of p47^phox in other protein complexes, including not only the noncanonical activation of NOX1, as already mentioned above but also its interaction with protein disulfide isomerase (214, 559).

Other studies have directly targeted the NOX2 protein. Mouse models of NOX2 deletion have generally shown similar benefits as those using p47^phox-deficient mice. The NOX2^−/− mice are also protected against angiotensin II–dependent hypertension (151, 988). The combination of NOX2^−/− with ApoE^−/− can decrease the atherosclerotic lesions (475), but this is not always the case (508). Other works have found a positive effect of the NOX2 deletion in mice treated with high fat diet. In these studies, mice carrying the deletion gain weight but show no increase in ROS production or endothelial dysfunction as compared with the control animals (256, 609).

Finally, it is worth noting that the role of NOX2 in the immune system and in vascular endothelial cells will have, as noted, different impacts in cardiovascular disease (573). The relative contribution of these sources for the initiation and development of atherosclerotic lesions and other pathologies remains an issue of active research. The current paradigm indicates that macrophage NOX2 is more detrimental in the development of cardiovascular disease (395, 876), with endothelial NOX2 probably involved in the initiation, but not so much in the development of the disease, and a subsequent progress of the atherosclerotic lesion dependent on immune cells (243, 573, 976).

2. NOX1

The isoform NOX1 is expressed in VSMCs and in lower levels is also found in ECs (556, 909) (FIGURE 9, Table 5). NOX1 is found in caveolae and signalosomes of vascular cells (423, 431, 652). NOX1 expression is activated by several factors including angiotensin II, PDGF, and thrombin (382, 492, 630, 695, 999). The nature of the agonist can vary in different cellular compartments (651, 652, 1055).

Increased NOX1 activity is associated with vascular disease, and studies have shown links between high NOX1 levels and hypertension (1015), diabetes (163, 380, 1003), ischemia/reperfusion injury (479), and coronary artery disease (393). NOX1-dependent signaling has been linked to angiogenesis. For instance, knockdown of NOX1 can decrease angiogenesis (343), whereas increased NOX1 promotes tumor angiogenesis and growth (582). These findings indicate a critical role of NOX1-derived superoxide in angiogenesis signaling via PPARα-, AKT-, and ERK1/2-mediated pathways.

In agreement with these observations, the overexpression of NOX1 in VSMCs increases blood pressure and hypertrophy in mouse hypertension models (235). siRNA against NOX1 or NOXO1 decreased superoxide production and eNOS uncoupling in diabetic mice (1049). NOXA1 has been also associated with atherosclerotic injury, with overexpression of NOXA1 increasing ROS production and increased NOXA1 levels detected in atherosclerotic lesions (695).

Studies of NOX1 function have also taken advantage of the generation of several KO lines. Mice with deletions of NOX1 (345, 630), NOXO1 (511), and NOXA1 (293) have been developed. NOX1 deletion limits the blood pressure increase elicited by angiotensin II, although it is unclear if basal blood pressure levels are reduced (345, 346, 630).

The effect of NOX1 in atherosclerosis has been investigated similarly to NOX2, using a Nox1^−/−ApoE^−/− mouse model. The results of different studies are contradictory, with examples of improved outcomes for NOX1 deficient mice subjected to high fat diet (854) and diabetic mouse models (380), whereas in other cases, NOX1 deletion did not seem to have an effect (875).

The effects of the p47^phox deletions, often used as a surrogate for the NOX2 deletion, can be partly due to changes in NOX1 activity. This notion is supported by a diminished effect of angiotensin II in the p47^phox KO (549, 550) as compared with NOX1 (345, 630) or NOX2 deficient models (988).

Altogether, the role of NOX1 in cardiovascular disease seems secondary to NOX2, but still significant, as shown in several models (345, 380, 630, 854). It is important to consider that therapies focused on NOX2 can overlook compensatory effects of NOX1 in part by the sharing of a number of regulatory elements that can activate the production of superoxide in both proteins.

3. NOX4

NOX4 is the most abundant isoform in ECs, and it is also present in VSMCs and adventitial fibroblasts (10, 265, 736, 832, 876, 1036). In the heart, NOX4 has been identified in ECs, cardiomyocytes, and fibroblasts (682, 1058). The intracellular location of NOX4 suggests a role in signaling and may be involved in endoplasmic reticulum stress (817) (FIGURE 9, Table 5). The expression of NOX4 is upregulated by TGF-β (201, 906). Other regulators include TNF-α (70, 681, 884, 1048), hypoxia (455), and PDGF (487).

The role of NOX4 in vascular biology appears to be generally protective. Endothelial-targeted overexpression of NOX4 has beneficial effects and decreases blood pressure (779). Existing evidence supports a role of NOX4 in maintaining blood pressure levels (552, 723, 779, 832).

Endothelial-specific overexpression of NOX4 in mice promotes angiogenesis and eNOS function (193) and decreases blood pressure (779). Comparable results were achieved with NOX4 knockdown studies in cells (642). A beneficial effect of NOX4 in ischemia/reperfusion has also been observed (989).

Studies with NOX4^−/− mice indicate protection from endothelial dysfunction in wild-type mice compared with the NOX4 deletion (832). In general, NOX4 shows protective effects in ApoE^−/− mice (194, 229, 230, 379, 552, 838). One study has related NOX4 to hyperglycemia and insulin resistance in mice (1012). However, NOX4 deletion in the ApoE^−/− mouse does not ameliorate diabetes-associated vasculopathy (380). Other studies have found no change in NOX4 levels in diabetic rat models (1003). Heart-specific NOX4 overexpression has protective effects in the response to chronic load stress in part due to increased HIF1α response and increased angiogenesis (1058). It should be noted that some aspects of NOX4 function in the cardiovascular system remain controversial. For instance, splicing of NOX4 appears to influence cardiovascular disease, with a shorter protein (NOX4D) exerting beneficial signaling functions in the nucleus, whereas full-length NOX4 is overexpressed in ischemia with detrimental effects (971).

NOX4 protective role may be due to its generation of hydrogen peroxide, and not superoxide (253). As hydrogen peroxide does not scavenge NO, changes in NOX4 activity do not directly impact NO levels. The signaling mechanisms of hydrogen peroxide in ECs (941) lead to the increased activity of eNOS and other enzymes (121, 141, 193, 251, 815, 940). Consistent with this role of NOX4-derived ROS in angiogenesis (642, 989), it has been shown that NOX4 can modulate the role of VEGF-induced dimerization of CD146 necessary for cell adhesion (1065), and NOX4-derived ROS can induce the reversible glutathionylation of SERCA and increase Ca²⁺ levels, thus regulating VEGF-signaling cell migration (273).

In a final note, it should be also considered that although NOX4 activity has been shown to reduce fibrosis in atherosclerosis (230), it can also promote fibrosis, particularly in the lung, but also in kidney and other tissues (110, 201, 416, 596, 820). The basis of these differences in NOX4 roles are not completely understood, but could be based on the specific roles of NOX4 in different cell types and/or detrimental effect of eNOS activation in conditions in which eNOS is uncoupled (323).

4. NOX5

NOX5 is the last discovered NOX isoform (62). In human blood vessels, NOX5 is expressed in ECs and VSMCs (81, 469) (FIGURE 9, Table 5).

NOX5 expression can be activated by most of the factors that activate other vascular NOX enzymes such as angiotensin II, TNF-α, PDGF, thrombin, and endothelin-1 (81, 660, 681). As indicated above, the particular architecture of NOX5 indicates a regulation of its activity through Ca²⁺ levels (62, 469). Phosphorylation of Ser475, Ser490, Thr494, and Ser498 within the flavoprotein domain increases Ca²⁺ sensitivity (463, 717, 718); this phosphorylation is mediated by different kinases, including PKC isoforms (164).

Unlike the rest of the NOX proteins, NOX5 does not have an homolog in rodents, which has hampered the research of this isoform (871). New approaches, such as mice strains incorporating the human NOX5 gene, may provide new information about NOX5 function (439, 517).

Despite these difficulties, increasing data suggest a role of NOX5 in cardiovascular disease. Higher NOX5 expression levels have been observed in patients with coronary artery disease (391), but causation has not been proven. NOX5 expression also increases in the human heart after MI (396). Expression of NOX5 in VSMCs and in macrophages and monocytes can contribute to the development of atherosclerotic lesions (622, 623). NOX5 also appears to modulate proliferation and cell migration processes (371, 719). Altogether, increased NOX5 activity seems to be involved in a pathological response (661). Further work is required to delineate the role of NOX5 in human cardiovascular disease.

5. Pharmacological regulation of NOX enzymes

A plethora of NOX inhibitors, including small molecules and peptides, have been developed (22). Most of the inhibitors have shown poor specificity between isoforms, hampering the progress in the field (174). In this section, we will present some of the most relevant inhibitors because of their wide use or high isoform specificity. A more comprehensive analysis of the available inhibitors for NOX enzymes can be found in recent reviews (22, 174, 175, 252).

a) diphenylene iodonium.

Diphenylene iodonium (DPI) (FIGURE 10) is a nonspecific inhibitor of all NOX isoforms (18, 197, 246). Its mechanism of action is based on the reaction with the FAD moiety of the NOX dehydrogenase domain (699). Unfortunately, this reaction is general for any flavin-containing protein, which means that DPI is also an inhibitor of many other ROS-producing enzymes, including, but not limited to, NOS, mitochondrial complex I and XO (18). Some reports indicate that the use of low levels of DPI may selectively target NOX in the presence of other flavoproteins (241); however, the lack of specificity makes unlikely the general use of DPI as a NOX inhibitor in vivo.

FIGURE 10.

Chemical structures of selected NADPH oxidase (NOX) inhibitors.

b) apocynin.

Apocynin (FIGURE 10) is a methoxy-substituted catechol generally considered to be a specific NOX2 isoform inhibitor. Apocynin inhibits NOX2 by reacting with cysteine residues in p47^phox that appear to be necessary for enzyme assembly (894). It should be noted that this mechanism of action converts apocynin into a NOX1 inhibitor when NOX1 uses p47^phox for activation in the noncanonical mode. Moreover, there are several factors that question the inhibitory effects of apocynin; notably, apocynin is an antioxidant and scavenger of ROS like HOCl and hydrogen peroxide (426, 735). Thus, although apocynin remains widely used in research and has led to important insights in the function of NOX enzymes (237, 413, 581), the results of experiments with apocynin need be considered with caution. In particular, effects in isoforms other than NOX1 and NOX2 should be validated with other approaches.

c) nox2 docking sequence-tat.

NOX2 docking sequence-tat (NOX2ds-tat), the first isoform-specific NOX inhibitor, is a peptide based on the sequence of the B-loop of NOX2 (199, 218, 782). NOX2ds blocks specifically the interaction between NOX2 and p47^phox (199). The addition of the tat sequence (nine amino acids from HIV-tat) confers membrane penetration ability to the peptide (782). The inhibitor shows high specificity toward NOX2 (199). The chemical nature of the compound makes oral delivery unsuitable and has limited its pharmacological application, however, because of its isoform specificity and lack of off-target effects remains widely used in research (244, 461, 476, 762, 782, 910). Notably, a similar approach with NOX4 did not result in significant inhibition of NOX4 (200), which suggests a strong interaction between dehydrogenase and transmembrane domains in NOX4 and points out a critical function of the B-loop in this interaction (460), very probably linked to the constitutive activity of this NOX isoform.

d) noxa1 docking sequence.

NOXA1 docking sequence (NOXA1ds) is a peptide developed to block the interaction between NOX1 and NOXA1. The sequence contains 11 amino acids from the putative binding region of NOXA1 to NOX1. The selected sequence shows low similarity with p67^phox to prevent cross reactivity with NOX2. The peptide prevents NOX1 assembly and does not inhibit NOX2, NOX4, or NOX5 as tested in in vitro assays (772, 792).

e) ml171.

ML171 (2-acetyl phenothiazine) (FIGURE 10) is one of the few small-molecule, NOX isoform–specific inhibitors reported to date. ML171 inhibits NOX1 at nanomolar concentrations, with IC₅₀ values at least 20-fold higher for NOX2, NOX3, NOX4, and XO (359). The mechanism of action of ML171 seems to involve only interaction with the NOX1 protein and not with its activating subunits. ML171 showed off-target effects with the inhibition of serotonin and adrenergic receptors in the micromolar range (359). The study of differently substituted phenothiazines indicates some isoform selectivity that can represent a promising pathway to novel small-molecule, selective NOX inhibitors (847).

f) vas2870 and vas3947.

VAS2870 and VAS3947 (FIGURE 10) are recently described NOX inhibitors based on a triazolopyrimidine scaffold (892, 934, 1016). VAS2870 appears to inhibit all isoforms, with proven inhibitory effects at least for NOX1, NOX2, NOX4, and NOX5 isoforms (21, 520). The mode of action of these compounds is unknown, although for NOX2, they appear to block complex assembly (21). These inhibitors do not have a noticeable effect on XO- or eNOS-dependent ROS production, so they can be considered specific pan-NOX inhibitors. VAS3947, a more soluble derivative of VAS2870, shows similar inhibitory properties (1016). Off-target effects for VAS2870, namely nonspecific alkylation of thiols, have been reported (915).

g) gkt-136901 and gkt-137831.

GKT-136901 and GKT-137831 are pyrazolopyridines structurally related to VAS2870 and VAS3947. GKT-136901 and GKT-137831 (FIGURE 10) are the first inhibitors suitable for oral administration, and currently GKT-137831 is in phase 2 clinical trials (472). GKT-136901 was first characterized as a selective NOX1 and NOX4 inhibitor, although it can also inhibit NOX 2 at higher concentrations (545). Inhibitory effects against NOX5 have also been documented (680). No off-target effects were detected against a panel of 135 enzymes, including XO and eNOS (545). GKT-137831 shows similar inhibitory properties with potent inhibition of NOX1 and NOX4 and inhibitory effects on NOX2 and NOX5 at higher concentrations (35). The basis of the inhibition by these compounds is not known, although the activity of GKT-136901 as peroxynitrite scavengers has been recently reported (826).

h) gsk2795039.

The discovery of a novel, specific NOX2 inhibitor was recently reported (435). GSK2795039 (FIGURE 10) appears to be specific toward NOX2 and does not show off-target effects toward eNOS or XO. Existing evidence suggests that GSK2795039 inhibits NOX2 through competition with NADPH for the binding site of the NOX2 dehydrogenase domain (435).

B. Uncoupling of Endothelial NOS

NO production is critical for vascular function as described in previous sections (299, 573, 600, 786). In conditions in which the substrates and cofactors are present in saturating concentrations, NOS enzymes catalyze the synthesis of NO from molecular oxygen and l-arginine with high efficiency. The electrons transferred from NADPH are almost quantitatively used to catalyze l-arginine oxidation, and the ratio of NADPH to l-citrulline is close to the optimal value of 1.5 molecules of NADPH per l-citrulline (and NO) generated (3, 590). However, a number of factors can alter the electron transfer in NOS so the amount of NADPH required to generate NO becomes much larger. This is largely due to the reduction of other substrates, usually molecular oxygen, to generate superoxide. This situation when the consumption of NADPH is not correlated with a stoichiometric production of NO is generally termed eNOS uncoupling (488, 570, 870). Although the superoxide generation is the common manifestation of this NOS dysfunctional state, it is important to recognize and understand the causes of eNOS uncoupling to develop adequate interventions. In addition, a number of situations impinge NOS function and NO generation, but not necessarily cause increased superoxide production, despite being usually characterized as uncoupling. Factors that can trigger eNOS uncoupling are discussed below.

Initial reports indicated that nNOS had the ability to produce superoxide in a reaction dependent on NADPH and Ca²⁺/CaM (750). The effect of NOS inhibitors on this process was diverse, with some inhibitors efficiently suppressing superoxide production (e.g., l-NAME), whereas others had little effect (e.g., N^G-monomethyl-l-arginine) (749). The reaction is observed in the three NOS isoforms (750, 973, 1034). The effect of cofactors and NOS inhibitors was thoroughly investigated (150, 250, 590, 749, 973, 1033). Superoxide can be generated by the oxygenase/heme domain, in a Ca²⁺/CaM-dependent way, or independently by the reductase domain from Ca²⁺/CaM (653, 1032). In conditions that promote uncoupling, the heme group is usually the main generator of superoxide species through the formation of a ferrous/oxy complex that decays to ferric heme and superoxide (Eq. 20) (FIGURE 11) (250, 901, 973, 1033).

FIGURE 11.

Superoxide generation by nitric oxide synthase (NOS) monomer and dimer in the presence or absence of calmodulin (CaM).

The loss of the dimer structure renders eNOS unable to produce NO, as the heme group of one monomer must receive electrons from the FMN domain of the other monomer. Thus, conditions that increase the monomer/dimer ratio decrease NO output and can increase superoxide production through the reductase domain but also decrease superoxide formation in the oxygenase domain (FIGURE 11). Therefore, it should be noted that measurement of dimer/monomer ratios as an indication of eNOS uncoupling is not a reliable and reproducible method, and more specific methods monitoring posttranslational modifications and substrate levels are preferable.

The concomitant formation of NO and superoxide can be especially deleterious as both radicals can combine to form peroxynitrite at diffusion-limited rates. Peroxynitrite is a strong oxidant and nitrosating species (78). Therefore, the superoxide generated by the uncoupled eNOS not only induces toxic effects on its own but 1) acts as a NO scavenger, decreasing NO availability and 2) generates peroxynitrite, a more harmful species (78). Peroxynitrite can in turn oxidize BH₄ to BH₂, further promoting eNOS uncoupling (543). Notably, the deleterious effect of superoxide on the endothelium relaxing effect of NO was discovered even before the chemical nature of NO as endothelium-derived relaxing factor was elucidated (385). In the following sections, we dissect the different causes of NOS uncoupling and overview the current status of pharmacological approaches to prevent or reverse this dysfunctional state. Special emphasis is dedicated to the two most relevant aspects in pathophysiology, namely the dysregulation of BH₄ and l-arginine levels.

1. BH₄/BH₂

The binding of the cofactor tetrahydrobiopterin (BH₄) is a requisite for NO synthesis by NOS enzymes. BH₄ binding synergistically enhances the affinity toward the substrate l-arginine. In addition, BH₄ plays a critical role in the catalytic cycle of NOS (87, 204, 377, 904, 933, 1000, 1007). The ability of BH₄ to support electron delivery to the heme group at critical steps, generating a transient radical species, is specific to BH₄. Other pterins are, in general, unable to support NO synthesis in mammalian NOS enzymes, or can only support a NO synthesis rate much slower than BH₄ (753). BH₄ can be oxidized in vivo by molecular oxygen or ROS to generate dihydrobiopterin (BH₂), a pterin very structurally similar to BH₄ but unable to support NO synthesis (544, 753, 933).

Biological de novo synthesis of BH₄ is mediated by three enzymes that catalyze the stepwise process from GTP to BH₄, namely GTP cyclohydrolase I (GCH1), pyruvoyl tetrahydropterin synthase (PTPS), and sepiapterin reductase. In this route, the reaction of GTP with GCH1 is the limiting step. An additional salvage pathway can produce BH₄ from sepiapterin and/or BH₂. In this pathway, sepiapterin reductase reduces sepiapterin to BH₂, whereas dihydrofolate reductase can reduce BH₂ to BH₄ (87, 1007).

The conversion of BH₄ to BH₂ in the setting of endothelial dysfunction has been widely reported (192, 258, 550). The accumulation of BH₂ becomes deleterious as the binding site of BH₄ in NOS can also bind this pterin. This is particularly problematic in the case of eNOS (190). iNOS and nNOS show higher affinity toward BH₄ than for BH₂ (around 10-fold in the absence of l-arginine) (6, 514, 753); however, eNOS shows almost identical binding affinities for BH₄ and BH₂, and it is thus very sensitive to imbalances in the concentrations of BH₄ and BH₂ (190). Thus, rather than the relative amount of BH₂, the BH₄/BH₂ ratio gives a more precise estimation of the endothelial function (190, 192, 974).

Mechanistically, when NOS binds BH₂ instead of BH₄, the ferrous/oxy heme cannot hydroxylate l-arginine to N^G-hydroxy-l-arginine, NO synthesis is blocked, and the oxygen is eventually oxidized to superoxide. Thus, not only is the NO synthesis completely abolished but the protein also becomes an oxidase, generating superoxide instead of NO (191, 972, 1007) (Table 6).

Table 6.

Relative effect of different factors on eNOS superoxide and NO production

Superoxide Production	Pterin	Substrate	Other	NOSynthesis
+++	None	l-arginine		None
+++	None	ADMA		None
+++	None	None		None
++	BH2	l-arginine		None
++	BH2	None		None
+	BH4	l-arginine	S1177p	+
=	BH4	l-arginine		=
=	BH4	l-arginine	T495p	–

Qualitative assessment based on data from Refs. 150, 161, 250, 334, 753, 973, 974, 1033. ADMA, asymmetric dimethylarginine; eNOS, endothelial nitric oxide synthase; NO, nitric oxide.

2. l-Arginine/ADMA

In saturating conditions of the NOS cofactors, the synthesis of NO only requires l-Arginine and O₂ as substrates (904). l-Arginine is also important for the dimerization of NOS (51, 515) and increases BH₄ binding affinity (514, 753). The absence of l-arginine leads to the production of superoxide in an otherwise active form of NOS; thus, l-arginine availability can have an important effect on eNOS uncoupling (Table 6). Given the concentration of l-arginine in ECs [0.1–0.8 mM (1030)] and the high affinity of NOS enzymes toward l-arginine [eNOS K_M = 2.9 μM (746)], it was long thought that l-arginine availability would not be a functional issue for NOS in vivo. However, in recent years, it has become clear that extracellular l-arginine and ADMA concentrations, compartmentalization, and membrane transporters are important for NOS function, as will be discussed below in sect. IIIB4 (167, 668, 1030).

ADMA (FIGURE 3) is mainly produced from the breakdown of proteins containing methylated arginine residues. The same processes also yield the methyl arginines N^G-methyl monoarginine (FIGURE 3) and symmetric dimethylarginine (FIGURE 3) (668). Notably, ADMA and N^G-monomethyl-l-arginine (but not symmetric dimethylarginine) are NOS inhibitors and can be important regulators of NOS activity in vivo (566, 963). The metabolism of ADMA and its effects on cardiovascular disease have been covered by recent reviews (668, 790, 962). Soon after its discovery in cells and tissue, ADMA was recognized as an important marker for cardiovascular disease progression (656, 829, 961, 1011, 1053, 1068). More recent observations indicate that the ADMA/l-arginine ratios in plasma may be a better marker than either metabolite assessed independently (75, 589, 606, 790, 966).

3. Other causes of eNOS uncoupling

a) disruption of the zn/s cluster.

The dimeric structure of NOS is held together by interactions between the oxygenase domains, and among these interactions, the formation of a Zn/4S cluster with the participation of two cysteines from each NOS monomer (195). The oxidation of the cysteine residues involved in the formation of the Zn/S cluster disrupts the dimer structure. Work by Zou et al. indicated that this cluster is particularly sensitive to peroxynitrite (1070). This observation is not unexpected, as metal-thiol bonds are some of the main targets of peroxynitrite (78, 526).

The breakdown of the Zn/S cluster is, in most cases, secondary to eNOS uncoupling. As discussed above, the monomeric eNOS cannot catalyze the NO synthesis reactions, and the monomerization renders the heme group inactive. The superoxide formation from the monomer is low as compared with the uncoupled dimer, and thus, this mechanism does not increase the rate of superoxide formation. However, the breakdown of the Zn/S cluster is a marker that indicates the generation of peroxynitrite in the cellular environment. Although this is not a 100% specific marker of eNOS uncoupling (the superoxide source may be different from NOS, for example, a NOX enzyme), the correlation of the Zn/S cluster breakdown and eNOS uncoupling is probably elevated (Table 6, FIGURE 12).

FIGURE 12.

Progression of nitric oxide (NO) synthesis, superoxide, and peroxynitrite during the development of endothelial nitric oxide synthase (eNOS) uncoupling.

Another modification that disrupts the Zn/S cluster is S-nitrosation of the cysteine residues Cys96 and Cys101, as discussed in sect. IIA1.

b) glutathionylation of cysteine residues.

Recent reports indicate that eNOS can be modified posttranslationally by S-glutathionylation (162). At least three residues have been shown to be susceptible of modification: Cys 689, Cys908 (162), and Cys382 (160) (Table 2). The process can be reversed by glutaredoxin-1 (160). The glutathionylation of Cys689 and Cys908 causes eNOS uncoupling. These residues, located in the reductase domain, appear to disturb the electron transfer between FAD and FMN cofactors, and indirectly limit the rate of reduction of the oxygenase domain. Consistent with this hypothesis, the addition of the inhibitor l-NAME decreases superoxide production only in part, indicating that most of the superoxide production occurs in the reductase domain via electron transfer from the reduced FAD and FMN to molecular oxygen (162) (Table 6, Figs. 11, 12).

S-glutathionylation has been involved in the development of angiotensin II–mediated endothelial dysfunction (328) and nitroglycerin resistance (522). Both processes appear to be initiated by mitochondrial ROS. Consistent with this view, the increase in ROS levels, BH₄ depletion, and glutathione oxidation, with subsequent eNOS S-glutathionylation, appears to be intimately related (189, 215).

Along with glutaredoxin-1, thioredoxin has also been proposed to deglutathionylate eNOS, and this process is independent of glutathione oxidation (907). This offers therapeutical opportunities in oxidative stress situations in which glutathione levels are low (432).

c) phosphorylation.

Phosphorylation of Thr495 is a negative regulator of CaM binding to eNOS (297, 407, 646). The Thr495-phosphorylated NOS shows decreased electron transfer from the reductase domain to the heme domain and decreased NO synthesis (588). The accumulation of electrons in the reductase domain causes an increased uncoupling from the reductase domain, which, as discussed previously, is less pronounced than when this occurs in the heme but can cause significant production of superoxide (Table 6).

4. Pharmacological interventions in eNOS uncoupling

The sensitivity of the eNOS system to oxidative stress converts eNOS into a main target in conditions of ROS imbalance. According to the kindling hypothesis of endothelial dysfunction, the initial formation of ROS by other sources can promote uncoupling of NOS, amplifying the initial excess of ROS (488). The uncoupling causes eNOS activity to change from healthy (producing NO) to toxic (generating superoxide) (303). As such, the uncoupling of eNOS has been noted as a target for therapeutic potential (488, 569, 570, 600, 784). Different interventions have been developed to address this misdirected reactivity. As the status of eNOS uncoupling changes with the severity of the cause, different treatments may prove more efficacious at different stages (FIGURE 12). Current strategies are discussed below.

a) bh₄.

Many studies have shown that BH₄ levels correlate with eNOS activity (796, 1006). Conversely, low levels of BH₄ are found in endothelial dysfunction and cardiovascular disease (421, 429, 440, 550, 560, 858, 899). Spearheaded by these observations, interventions to reverse eNOS uncoupling via BH₄ supplementation have been numerous, and in general, the supplementation of BH₄ or positive modifications in the BH₄ synthesis pathways have been positive. A large number of studies on mice and other animals have been described; for the sake of conciseness, we will only discuss some of the recent human studies. For a more complete overview of the BH₄ supplementation experiments including studies in other organisms, the reader is referred to recent reviews (87, 828).

Acute supplementation of BH₄, either orally or injected, generally improves vasodilatation. This effect has been observed in a number of human studies, including healthy individuals (450, 743, 987) and patients with hypertension (429), hypercholesterolemia (318, 899, 1031), chronic heart failure (849), coronary artery disease (618, 850), heart failure with reduced ejection fraction (178), or rheumatoid arthritis (619) or smokers (418, 929, 957). Studies, including the assessment of coronary artery function in patients with atherosclerosis, did not find an improvement after BH₄ treatment (1028).

Several studies monitoring endothelial function after chronic BH₄ administration have been conducted. In hypercholesterolemia patients, oral BH₄ for 4 wk (400 mg twice daily) improved endothelial function (186). A treatment with oral BH₄ for 2–6 wk in patients with coronary artery disease did not show improvement; and actually increased BH₂ levels (202). In patients with hypertension, a 4–8-wk, 400 mg/day treatment was able to decrease blood pressure (747). Studies on pulmonary hypertension also show improvement in the 6-min walking distance (788). A recent study using 400 mg daily for 1 wk improved endothelial function in rheumatoid arthritis patients (619).

In most examples, BH₄ is able to improve endothelial function, but the study by Cunnington et al. (202) clearly indicated that in some conditions, BH₄ supplementation is not a practical therapeutic approach. In severe oxidative stress conditions, the BH₄ will be rapidly oxidized to BH₂, further increasing the BH₂/BH₄ ratio (FIGURE 12). As the treatment with antioxidants has been shown to improve endothelial function in certain conditions (56, 419, 420, 422, 572), it is possible that the coadministration of BH₄ with ascorbic acid or other antioxidants can circumvent these issues. Of note, BH₄ is also an antioxidant, and it is not clear if some of the observed effects at the supraphysiological concentrations used are just due to short-term antioxidant effects; related pterins, such as folate, can also produce similar effects (32, 33, 860). Techniques that can assess the oxidative stress conditions in tissues can certainly improve the use of BH₄ therapy with a more personalized approach. Pleiotropic effects of some drugs can improve BH₄ availability; this has been observed for statins such as atorvastatin (31).

As an alternative to BH₄, supplementation using other related compounds such as sepiapterin has been also used. Sepiapterin can be converted into BH₄ via the salvage pathway by sepiapterin reductase and dihydrofolate reductase as described in the previous sections. Sepiapterin supplementation improves endothelial function in several mouse models, including the atherosclerosis model ApoE^−/− mice (560), diabetes models (721), or obesity models (527). Mouse GCH1 KO mice show abolished endothelial NO synthesis and increased superoxide production that can be reversed by sepiapterin supplementation (173). It should be noted that this pathway has BH₂ as an intermediate, which could lead to detrimental effects. In conditions in which dihydrofolate reductase is inactive or absent, the recovery of eNOS coupling by sepiapterin supplementation is ineffective (452).

Folic acid has wide effects on the metabolism, impacting cysteine/homocysteine metabolism, but also pathways involving tetrahydrofolate and dihydrofolate and hence BH₄ biosynthesis (598). The effects on eNOS uncoupling seem related to the antioxidant properties of the folic acid metabolites (900). Supplementation of folic acid (5-methyltetrahydrofolate) shows positive effects on eNOS function (900, 978) and has been shown to prevent nitroglycerin tolerance in short-term human studies (375). Folic acid supplementation studies in human blood vessel samples have shown increases in BH₄ levels and BH₄/BH₂ ratios (33).

To improve endogenous BH₄ synthesis, efforts have focused in GCH1, the rate limiting enzyme during de novo BH₄ synthesis (143). Overexpression of GCH1 in cells and mouse models improves eNOS function (19, 143). Human variants of GCH1 show measurable effects in blood pressure, indicating the direct relationship between GCH1 and eNOS function (1057).

b) l-arginine.

Several lines of evidence indicate that the availability of l-arginine may be jeopardized under endothelial dysfunction, and the concentrations of inhibitory l-arginine metabolites, such as ADMA, can also increase in pathological conditions (107, 1053). In this regard, different studies have investigated alternatives to improve l-arginine supply either via direct l-arginine supplementation, the use of l-arginine precursors, or through the inactivation of l-arginine consuming arginases.

The direct supplementation of l-arginine was shown to increase NO production in cells. This finding was rather surprising, as most cells have intracellular l-arginine concentrations manyfold higher than the eNOS K_M value, and thus, this observation has been termed the “arginine paradox” (12, 797). Further studies showed that l-arginine import via the CAT1 transporter is critical for eNOS activity (167, 634). A membrane metabolon including arginine succinate lyase, argininosuccinate synthetase, hsp90, and eNOS associates in the caveolae and is necessary for regular eNOS function (270, 294, 634). Another factor contributing to the arginine paradox, as will be discussed below, is the concentration of endogenous eNOS inhibitors, mainly ADMA. At high ADMA concentrations, l-arginine supplementation may not overcome eNOS inhibition, indicating the need to determine l-arginine/ADMA ratios to better assess l-arginine availability (102, 106, 954).

In animal models of MI, l-arginine supplementation shows mixed results. Some studies show improvement in functional recovery (712), whereas studies in l-arginine–treated dogs show detrimental outcomes (665). Further studies have explored l-arginine supplementation in human patients. Again, the results have been heterogeneous. The causes of this variability are still unclear. Many studies have reported a lack of effect on l-arginine in endothelial function markers, and on occasion, the supplementation of l-arginine has been associated with higher mortality effects, such as tests of l-arginine in patients after MI (835). Other studies show impaired recovery in peripheral artery disease patients (1014). In contrast, most studies detect decreases in blood pressure with supplementation. A recent meta-analysis indicates that doses between 4 and 24 g/day decrease blood pressure (systolic, 5 mmHg; diastolic, 3 mmHg) (240). Other recent studies show improved endothelial function and insulin sensitivity in patients with impaired glucose uptake and metabolic syndrome (6.6 g/day l-arginine for 2 wk) (663) and improved endothelial function and decreased homocysteine levels in patients with elevated blood pressure and hyperhomocysteinuria (2.4 g/day for 4 wk) (781).

Altogether, it is not clear if the levels of l-arginine are the best target for intervention. In some cases high, l-arginine levels have even been correlated with higher incidence of ischemic heart disease (1045). It seems that the involvement of l-arginine in different metabolic pathways makes the use of l-arginine levels as a biomarker a poor indicator of outcomes (668, 857, 1029). Studies of a panel of l-arginine metabolites indicate some correlations, particularly for ADMA, but also the need to consider other factors such as age, sex, and body mass index (75, 771). The interaction of l-arginine levels with ADMA and other metabolites is the focus of current studies. As pointed out above, the ratio of l-arginine to ADMA appears to be a better indicator of cardiovascular risk (28, 103, 104, 606, 1050). It has been noted that when ADMA levels are low, the effects of exogenous l-arginine are limited; however, the therapy seems to be more beneficial in patients with higher ADMA levels, in which the treatment increases the l-arginine/ADMA ratios and improves endothelial function (103).

Other l-arginine–related species may allow for interventions of l-arginine levels. For example, l-citrulline has also been used to increase l-arginine levels (842, 1029). This supplementation can avoid the formation of detrimental high ADMA levels (632). The studies of l-citrulline in humans show mixed results. In some cases, l-citrulline increased NO synthesis, but no changes in blood pressure were observed (506). Other studies have observed positive effects of l-citrulline including reduced blood pressure (20, 285, 286).

Finally, it must be noted that arginases play an important role in the regulation of l-arginine levels (144, 259, 734, 804). Exacerbated arginase activity can decrease l-arginine levels and cause eNOS uncoupling (91, 144, 667, 804, 1039). Thus, the use of arginase inhibitors is gaining increasing attention. Several human studies have used the arginase inhibitor nor-N^G-hydroxy-l-arginine (FIGURE 3). In these cases, arginase inhibition showed protection against ischemia-reperfusion injury (530) and improved endothelial function in patients with coronary artery disease and diabetes (855) and in patients with hypercholesterolemia (529). Further studies in this field and the search for novel natural and synthetic arginase inhibitors are warranted (757).

C. Mitochondria

The mitochondrial electron transport chain (ETC) represents a physiologically significant generator of ROS within vascular cells. Mitochondria generate ATP through the transfer of electrons from ETC complexes I and II to complex IV, where oxygen binds and is reduced to water. Through electron transfer steps, protons are pumped from the matrix to the inner membrane space, establishing a proton-motive force that couples electron transfer to the phosphorylation of ADP to ATP at complex V (FIGURE 13). Although the majority of electrons entering the ETC ultimately reduce oxygen to water at complex IV, a small proportion (~2–11%) of electrons escape from the chain at complex I or III to generate superoxide (34, 115, 678). Although this superoxide, being a charged molecule, likely does not escape the mitochondrion, manganese superoxide dismutase (MnSOD) located within the mitochondria catalyzes the dismutation of superoxide to hydrogen peroxide, which can freely diffuse out of the organelle to mediate cellular signaling or contribute to oxidative stress.

FIGURE 13.

Mitochondrial electron transfer chain. Main sites of superoxide production are indicated.

1. Mechanisms of ROS production

Mitochondrial ROS production is regulated by a number of factors including oxygen levels, substrate availability, and mitochondrial morphology/dynamics, all of which ultimately modulate the redox state of the ETC. Complex I, a 1,000-kDa protein comprised of ~45 subunits, is the most significant site of superoxide generation within the ETC and produces superoxide by two distinct mechanisms depending on the reduction status of the ETC and strength of the proton-motive force across the inner mitochondrial membrane (FIGURE 13). A FMN center within the complex, which accepts electrons from NADH, serves as the entry point to the ETC. In conditions in which electron transport through the ETC is slow, this FMN can become reduced and react with oxygen, catalyzing superoxide production. This occurs in conditions of high NADH/NAD⁺ ratios, such as when cellular ATP demand is low or the ETC has sustained damage. In contrast, when electron supply is high through the ETC, coenzyme Q downstream of complex I becomes reduced. A high proton-motive force present in this condition can then push electrons in the reverse direction from coenzyme Q to the complex I FMN, generating superoxide by reverse electron transport. This type of ROS generation occurs when succinate, the substrate for complex II, is present in high concentration and is responsible in vivo for the ROS generated after ischemia upon reperfusion (171, 172).

Complex III also represents a site of superoxide production within the ETC, although the rate of generation at this site is significantly lower than by reverse electron transport through complex I. Complex III catalyzes the Q cycle in which electrons are accepted from ubiquinone to be used to reduce cytochrome c for subsequent transfer to complex IV (FIGURE 13). In this process, ubiquinol (QH₂) binds to the Q₀ site of complex III and undergoes a one-electron oxidation that produces an unstable semiquinone radical. Although in most cases this semiquinone is rapidly oxidized, its reaction with oxygen leads to superoxide production at this site (567, 678). Importantly, specific complex III inhibitors, such as Antimycin A, can stabilize this semiquinone radical, significantly enhancing superoxide production by complex III (678). Physiologically, it is unclear which factors stabilize this species to regulate superoxide production. However, a number of studies demonstrate that complex III–dependent ROS production is important in physiological signaling pathways, including hypoxic sensing (997).

Although complexes I and III are considered the predominant sources of ROS within the mitochondrion, it is important to note that mitochondrial enzymes beyond the ETC have been reported to generate ROS. Pyruvate dehydrogenase (887), α-ketoglutarate dehydrogenase (130, 887, 952), and monoamine oxygenase A and B (290) have all been demonstrated to serve as ROS sources. These sources, although lower in their ROS-generating capacity, potentially play a role in specific mitochondrial-dependent redox signaling pathways. Notably, other reactive species–producing enzymes have also been demonstrated to translocate into the mitochondrion in specific conditions. For example, eNOS has been shown to enter the mitochondrion of pulmonary smooth muscle cells upon stimulation with asymmetric dimethyl arginine or endothelin-1 (768, 916). Additionally, although controversial, NOX4 has been reported to be localized within the mitochondrion either constitutively or in specific pathological conditions (152, 306, 540).

2. Physiological role of mitochondrial ROS

Although it has been recognized since the 1960s that mitochondria produce ROS, the physiological role of this phenomenon remained unclear until decades later. Accumulating data now demonstrate that mitochondrial ROS both mediate essential vascular signaling pathways and can also contribute to the pathogenesis of cardiovascular disease. This section will provide an overview of what is known about the role of mitochondrial ROS in cardiovascular health and disease.

Early studies viewed mitochondrial ROS production as pathological and focused on the deleterious effects of mitochondrial superoxide production. Mice completely lacking MnSOD, leading to greater mitochondrial superoxide concentrations, died within weeks of birth (584), confirming the essentiality of mitochondrial superoxide scavenging. Further, mice partially or conditionally lacking MnSOD showed dilated cardiomyopathy characterized by left ventricle dilation, decreased wall thickness, and hypertrophy (584, 696). Notably, tissue from the left ventricle of heart failure patients showed an increase in superoxide production and a decrease in MnSOD protein levels and activity, consistent with a role for increased mitochondrial superoxide in the pathogenesis of heart failure (814). It is now evident that mitochondrial ROS production also plays a major role in the pathogenesis of MI. During myocardial ischemia, the ETC as well as the NADH and quinone pools become maximally reduced. Additionally, there is a significant buildup of succinate, the substrate for complex II, which further supports the reduction of the quinone pool. Together, these conditions promote reverse electron transfer upon reperfusion, which results in significant superoxide production from complex I (171). Pharmacological inhibitors of complex I, including classical inhibitors as well as NO donors and nitrite, decrease this reverse electron transfer–mediated ROS production, leading to attenuated infarct size and protection of myocardial function after ischemia/reperfusion in a number of animal models (discussed in the next section) (136, 165, 170, 864).

Beyond the heart, changes in mitochondrial redox status can also have detrimental effects in the vascular wall and circulating cells. Decreased MnSOD activity or increased mitochondrial superoxide generation causes endothelial dysfunction and impairs acetylcholine-dependent vasodilation, particularly with age (1005). This dysfunction occurs not only through the direct reaction of superoxide with NO but is also associated with the oxidative damage of mitochondrial DNA and proteins. Notably, specific polymorphisms in the gene for MnSOD have been established as an independent risk factor for coronary artery disease (317).

Importantly, although mitochondrial ROS can propagate cellular damage, low physiological concentrations of mitochondrial oxidants have been shown to mediate a number of essential cellular signaling pathways (290). For example, mitochondria have been proposed to serve as a cellular oxygen sensor, such that in hypoxic conditions, mitochondrial ROS production by complex III leads to the stabilization of HIF1α and downstream hypoxic adaptation (84, 997). A similar pathway exists in the pulmonary vasculature in which mitochondria-generated oxidants signal the conserved phenomenon of hypoxic vasoconstriction (995). On a cellular level, mitochondrial ROS have also been shown to be required for the differentiation of some hematopoietic stem cells as well as human mesenchymal stem cells (707, 948). Additionally, mitochondrial ROS have been implicated in pathways leading to cell proliferation and senescence (233, 1046) as well as in the activation of other major cell signaling molecules such as the MAP kinase JNK and NF-κB (330, 689).

Notably, mitochondrial ROS modulate mitochondrial function, number, and cellular metabolic balance through a number of mechanisms. Mitochondrial hydrogen peroxide can oxidize the catalytic subunit of AMP kinase, leading to its autophosphorylation and subsequent activation to modulate metabolism within the cell (481, 1067). Although activation of AMP kinase can lead to mitochondrial biogenesis and increased mitochondrial number, this pathway can also directly be stimulated by mitochondrial hydrogen peroxide-dependent activation of the MAP kinase Akt (911, 919). As a counterbalance, mitochondrial ROS have also been shown to induce organellar degradation (e.g., mitophagy) under specific conditions (810). Through these mechanisms, mitochondrial ROS potentially play a role as a mediator in elaborate feedback pathways to regulate mitochondrial energetics and redox signaling.

The role of mitochondrial oxidants in the pathogenesis of some disease processes remains controversial. For example, two competing hypotheses have been proposed for the role of mitochondrial ROS in the pathogenesis of pulmonary arterial hypertension. Studies by Schumacker and colleagues suggest that increases in pulmonary mitochondrial superoxide production leads to the release of intracellular Ca²⁺ stores and subsequent vasoconstriction (996). To the contrary, Archer and colleagues demonstrate in animal models of pulmonary arterial hypertension that mitochondrial ROS production in the pulmonary artery smooth muscle cells is decreased, leading to inhibition of the K_V1.5 channel, which ultimately results in membrane depolarization, intracellular Ca²⁺ increases, and vasoconstriction (645). Although the role of mitochondrial ROS in pathogenesis remains controversial and a hot area of research, there is no doubt that changes in mitochondrial redox status play a role in disease progression.

3. Pharmacological interventions at the level of mitochondrial function

With the recognition that mitochondrial ROS play a role in disease pathogenesis, the development of antioxidants targeted to the mitochondrion is an active area of research. Several compounds have been developed with encouraging results in vascular disease and other pathologies.

a) mitoq.

The most extensively tested mitochondrial antioxidant is probably MitoQ (FIGURE 14). This small molecule consists of the antioxidant ubiquinol linked to the lipophilic cation triphenylphosphonium (500). The positive charge of the triphenylphosphonium cation is delocalized over its large hydrophobic area, which allows it to freely be taken into the mitochondria based on the mitochondrial membrane potential (873). MitoQ has been used extensively in animal models of disease and been shown to be protective against ischemia/reperfusion injury, cardiac hypertrophy, and hypertension (210, 211) among other pathologies.

FIGURE 14.

Structures of selected mitochondria-targeted compounds.

MitoQ has also been tested in phase II clinical trials for both Parkinson’s disease and hepatitis C (NCT00329056; NCT00433108). Although Parkinson’s patients taking MitoQ showed no difference in motor function compared with those on placebo treatment, this trial did establish that administration of MitoQ to humans for 1 yr is safe (874). In contrast, chronic hepatitis C patients treated with MitoQ for 28 days showed significant lowering of liver injury markers such as circulating alanine transaminase (333). These positive results have led to further phase IIB trials in hepatitis as well as the initiation of clinical trials to test MitoQ in other vascular pathologies.

b) szeto–schiller-31.

Szeto–Schiller peptides are another class of molecules that target antioxidant activity to the mitochondrion. Szeto–Schiller-31 (SS-31) (Bendavia/elamipretide; Stealth Peptides) is a small water-soluble tetrapeptide that accumulates within the mitochondrion and binds to cardiolipin in the inner mitochondrial membrane where it scavenges ROS and prevents lipid peroxidation (921) (FIGURE 14). These peptides have been shown to mediate protection in animal models of ischemia/reperfusion (922) and insulin resistance (27). Phase I clinical trials established that the molecule was safe and well tolerated (360). Multiple phase II trials are underway, examining the effects of the drug in conditions such as skeletal muscle disorders (NCT02245620), mitochondrial disorders (NCT02976038, NCT02805790, NCT02367014), and heart failure (NCT02814097, NCT02788747, NCT02914665).

c) mitosno and sodium nitrite.

In addition to pure mitochondrial-targeted oxidant scavengers, potential therapeutics that modulate ETC activity to decrease mitochondrial ROS production are being investigated. For example, MitoSNO (FIGURE 14), a mitochondrially targeted S-nitrosothiol, was shown in rodent models to significantly attenuate infarct size after MI (170). Mechanistically, this protection is dependent on MitoSNO transnitrosating a critical cysteine in the active site of complex I. This S-nitrosation of Cys39 in the ND3 subunit of complex I inhibits its catalytic activity and, hence, attenuates reverse electron transport and ROS production (170). Notably, other nitrogen-centered molecules, such as nitrite, have been shown to promote S-nitrosation of complex I and attenuate ischemia/reperfusion injury (864). Nitrite administered before or during cardiac ischemia has been shown to protect cardiac function and attenuate infarct size in a number of small and large animal models (226, 374, 864). These preclinical studies have led to ongoing clinical testing of nitrite as a therapeutic for MI as well as cardiac arrest.

D. XO

Xanthine oxidoreductase (XOR) is expressed as a xanthine dehydrogenase (XDH) but can become XO by posttranslational modifications (FIGURE 15). XOR is a dimeric protein containing one molybdopterin cofactor, two iron/sulfur clusters (2Fe/2S), and one molecule of FAD per 145-kDa subunit. The N-terminal domain contains the two Fe/S clusters, a central domain contains the FAD cofactor, and a NAD⁺/H binding site and the C-terminal domain contains the molybdopterin cofactor and a purine binding site (268). The enzyme has an important role in the catabolism of purines, catalyzing the stepwise oxidations of hypoxanthine to xanthine and xanthine to uric acid. Purine oxidation takes place in the molybdopterin site where the reaction leads to the reduction of the Mo⁶⁺ to Mo⁴⁺. The electrons are then transported within the enzyme, first to the Fe/S clusters and then to the FAD. The fully reduced FADH₂ can reduce NAD⁺ to form NADH.

FIGURE 15.

Architecture of xanthine dehydrogenase (XDH)/zanthine oxidase. A and B indicate the arrangement of the domains in the XDH/xanthine oxidoreductase (XOR) monomer. The middle, FAD-containing domain (yellow), is connected to the N-terminal domain containing two iron/sulfur clusters (Fe₂S₂) (green) and the molybdopterin domain (pink) by two flexible linkers. Posttranslational changes modify the protein activity from a XDH (A), NADH-producing enzyme, to the xanthine oxidase (XO) (B), hydrogen peroxide/superoxide-generating enzyme. C: three-dimensional structure of the XDH monomer (PDB:1FO4) (268).

1. ROS generation by XOR

During normal operation, XDH shows a limited production of superoxide/hydrogen peroxide via electron leak through the reduced FAD cofactor to O₂. This reaction only produces significant levels of ROS if NAD⁺ levels are low; however, these conditions can happen in the setting of vascular hypoxia (406, 562).

Oxidation of cysteine residues 535 and 992 in rat or bovine XDH or limited proteolysis turns XDH into XO (23, 72, 693, 773, 803, 994). XO shows a decreased affinity for NAD⁺ and increased O₂ affinity. The comparison of the crystal structures for XDH and XO indicates the movement of a loop that partially blocks the NAD⁺ binding site and alters the electrostatics of the FAD environment (268, 542). Altogether, these changes in substrate affinity lead to an increased electron flow from the molybdopterin domain to oxygen instead of NAD⁺. Thus, XO catalyzes the oxidation of hypoxanthine to xanthine and subsequently xanthine to uric acid, generating hydrogen peroxide and superoxide in a reaction amplified under ischemia and hypoxia (147, 307, 496, 497, 499).

XO is expressed in ECs (393), and its activity is increased by angiotensin II (551), TNF-α, LPS (117, 311), IL-6, and hypoxia (410, 748). There are low-circulating levels of plasma XO in baseline conditions that can increase in disease conditions by the release from damaged cells in liver or other tissues (1010). The binding of plasma XO to ECs can lead to vascular dysfunction (45, 71, 562, 612, 1010).

2. Pharmacological interventions targeting XO

Extensive evidence supports a role of XO in the development of cardiovascular disease (137, 148, 157, 282, 430, 466, 561, 633). As XO has become an interesting pharmacological target in the treatment of heart and cardiovascular diseases, numerous studies have used different strategies for the inhibition of XO.

a) tungsten.

Tungsten can be used as a global inhibitor of molybdenum-containing enzymes as it replaces the Mo atom in the molybdopterin, rendering the enzymes that use this cofactor inactive. Because of this mechanism of action, it should be noted that effects independent of XO will also be observed because of inhibition of AO, sulfite oxidase, and mARC. Tungsten is a powerful tool for the study of molybdenum enzymes in cell and animal studies and has been found to alleviate atherosclerosis in ApoE^−/− mice (833). However, because of its broad effects, the clinical use of tungsten is not feasible.

b) allopurinol.

Allopurinol is structurally similar to the XO substrate hypoxanthine and is metabolized to the active compound oxypurinol (FIGURE 16). Both allopurinol and oxypurinol have been used for the treatment of gout for decades (305, 710). Beyond its use for hyperuricemia and gout, more recent studies have used allopurinol and oxypurinol as XO inhibitors for the treatment of cardiovascular disease in human studies. In small trials, the pharmacological inhibition of XO with allopurinol/oxypurinol improved cardiovascular conditions in several pathologies, including heart failure (58, 278), dilated cardiomyopathy (148), and diabetes (137, 224). Amelioration of endothelial function in smokers was also observed (387). Despite these initial results, subsequent trials with larger number of patients have failed to report substantial positive effects as compared with the placebo groups (361, 405), although the meta-analysis of available studies indicates some improvement of cardiovascular function (430). Altogether, the effect of allopurinol in cardiovascular disease is still controversial. It has also been pointed out that allopurinol and oxypurinol have limited capacity to inhibit XO bound to glycosaminoglycans (620). It is very possible that the use of XO inhibitors needs to be complemented with other therapies targeting other sources of endothelial dysfunction (928).

FIGURE 16.

Structures of xanthine oxidase substrates and inhibitors.

c) febuxostat.

Febuxostat (2-(3-cyano-4-isobutoxyphenyl)-4-methyl-5-thiazole carboxylic acid) (TEI-6720, TMX-67) (FIGURE 16) is a XO inhibitor developed by Teijin, with higher selectivity and potency than allopurinol and oxypurinol (702). The selectivity is partly due to a lack of structural resemblance with purines; this also reduces side effects of purine and pyrimidine metabolism (1041). The higher affinity and specificity of febuxostat can have advantages over allopurinol and oxypurinol. Notably, the activity of febuxostat is not impaired by XO immobilization on the surface of ECs (620). Febuxostat was approved in 2008 in Europe and by the Food and Drug Administration in 2009 for the treatment of gout (commercial name Uloric in United States, Adenuric in Europe) (76, 77, 837). Studies in rodents have shown improvement in atherosclerosis and hypertension after treatment with febuxostat (697, 859). The safety of febuxostat for the treatment of cardiovascular disease in humans is still under investigation (331, 610).

d) topiroxostat.

Topiroxostat (FYX-051, 4-[5-pyridin-4-yl-1H-[1, 2, 4] triazol-3-yl]pyridine-2-carbonitrile) (FIGURE 16) is a novel XO inhibitor developed by Fuji Yahuhin Co. Ltd. Topiroxostat is a potent inhibitor of XO and induces a covalent modification in the enzyme that inactivates XO after complex dissociation (629, 703, 821). Topiroxostat is currently under study for the treatment of hyperuricemia in clinical trials (NCT02327754; NCT02837198) (441, 442).

e) nitrite.

Nitrite has also been used to inhibit XO-catalyzed ROS generation and to take advantage of its nitrite reductase activity to generate NO (reviewed in sect. IIB3) while decreasing superoxide production (1071). Recent work indicates that this treatment can be effective in the reduction of blood pressure in hypertensive patients (357).

E. Other Sources of ROS

Other proteins have been related to the direct or indirect production of ROS in the cardiovascular system, including, but not limited to, peroxidases, cyclooxygenases (COX), lipoxygenases, monoamine oxidase, heme oxygenase (HO), glucose oxidase, cytochrome b₅ reductase, and CYPOR (209, 505, 790, 908). In this section, we will briefly describe some of these ROS sources.

1. Peroxidases and COX

Mammalian peroxidases are heme proteins that catalyze the oxidation of a wide range of substrates. These proteins are activated by their reaction with hydrogen peroxide to generate an oxidizing heme species that can hydroxylate or oxygenate a substrate. Mammalian peroxidases include thyroid peroxidase, myeloperoxidase (MPO), lactoperoxidase, eosinophil peroxidase, and COX. MPO, lactoperoxidase, and eosinophil peroxidase are integral components of the immune defense (516).

a) mpo.

MPO is found predominantly in neutrophils and monocytes and catalyzes the reaction of hydrogen peroxide to produce several reactive species, mainly hypochlorous acid (HOCl) (discussed in sect. IVB) (516, 1021). Macrophage localization in the atherosclerotic plaque leads to detectable MPO activity and products of HOCl oxidation (213, 414). Several studies have correlated low circulating MPO levels with better cardiovascular prognosis (239, 260, 644, 690, 1059).

b) cox.

COX are proteins from the peroxidase family, evolutionarily related to the peroxidase enzymes of the immune system (1052). They utilize lipid substrates, generally arachidonic acid, and produce a variety of prostaglandins, lipid oxidation products with important physiological effects (156). Two isoforms are expressed in mammals: COX-1 and COX-2, and both of them have been related to endothelial dysfunction (981, 982). Interestingly, some interplay between COX-2 and other ROS/RNS-generating proteins such as iNOS (52) and XO (701) have been described.

2. Lipoxygenases

Lipoxygenases catalyze the insertion of oxygen in polyunsaturated fatty acids (119, 535, 1040) and are also biological sources of superoxide (537). Several lipoxygenases have been related to inflammation and cardiovascular disease, in particular, 5-lipoxygenase and 12/15-lipoxygenase (203, 597, 636).

3. AO

The Mo-containing protein AO has also been described as a potential source of superoxide in vivo (538, 539). Although the function of AO in mammals is yet to be established (434, 935), AO appears to play an important role in drug metabolism and other metabolic pathways (335, 512, 513, 756, 920). Studies with rat AO have shown a significant production of superoxide when the enzyme uses NADH as a substrate (539). Recent work on the human enzyme shows that the production of superoxide by the human AOX1 isoform is lower than that of the rat enzymes, and the enzyme is not reactive toward NADH (304). However, a known single nucleotide polymorphism in the AOX1 gene was shown to produce a mutant enzyme with increased superoxide formation rates in vitro (304). The possible impact of these single nucleotide polymorphisms in vivo remains to be determined.

4. CYPOR and CYP

CYPOR is a diflavin protein, structurally and evolutionarily related to the reductase domain of the NOS enzymes (377) that functions as a general reductase of the many CYP proteins. The P450 system is conspicuous for its role in detoxification and hormone synthesis (675, 1013). CYPOR, either directly or in combination with the CYP proteins, has been described as a source of superoxide (758, 851). The presence of these proteins in endothelial cells makes them a significant source of superoxide in the vasculature (295, 298). Because of the functional similarities with the dehydrogenase domain of NOX, some NADPH-dependent superoxide measurements can be unable to differentiate CYPOR from NOX as the source of superoxide (783).

5. HO

HO mediates the degradation of heme in mammals. This enzyme catalyzes the degradation of the protoporphyrin IX (heme b) to biliverdin (805, 806). This process generates ferrous iron (Fe²⁺) and CO, which are well-known toxic species (122, 739, 795, 959), but CO can also be, in low concentrations, a signaling molecule (713, 739, 805, 807). Three HO isoforms exist; the inducible isoform HO-1 is, to date, the most relevant to vascular physiology. HO-1 expression in basal conditions is low or undetectable in most tissues. High levels of HO-1 are present in the spleen, consistent with an important role in the processing of heme from red blood cell turnover (713, 805). The interplay between HO-1, CO production, and NOX inactivation is a focus of current interest in cardiovascular research (685, 793, 805, 924).

IV. OTHER ROS

A. Peroxynitrite

Peroxynitrite (ONOO⁻) is a strong oxidizing and nitrating agent formed by the reaction of NO with superoxide (Eq. 23). This reaction occurs at a diffusion-limited rate and is governed by the flux of NO and superoxide production (78). Another physiological route by which peroxynitrite can be formed is through the reaction of hydrogen peroxide with nitrite in acidic conditions (Eq. 24) (808).

$${\text{NO}}^{·}+{\text{O}}_2{}^{-·}\to {\text{ONOO}}^{-}$$ (23)

$${\text{H}}_{\text{2}}{\text{O}}_{\text{2}}+{\text{HNO}}_2\to \text{ONOOH}+{\text{H}}_{\text{2}}\text{O}$$ (24a)

$$\text{ONOOH}+{\text{OH}}^{-}\to {\text{ONOO}}^{-}+{\text{H}}_{\text{2}}\text{O}$$ (24b)

Despite its rapid rate of formation, peroxynitrite reacts relatively slowly in comparison with other RNS, with many biomolecules, including metal centers, proteins, DNA, and lipids (78, 526, 709). Additionally, studies show peroxynitrite can traverse cell membranes by both diffusion and through ion channels, suggesting that peroxynitrite can mediate oxidation/nitration distant from its site of formation (221, 611).

The reaction of peroxynitrite with biomolecules is generally considered as a detrimental event. Peroxynitrite can directly react with a variety of metal centers, resulting in their oxidation. In the cardiovascular system, peroxynitrite can oxidize heme proteins, such as Hb and Mb, resulting in the oxidation of their heme oxygen–carrying site from ferrous iron to ferric heme (101, 942). In the mitochondrion, a similar reaction oxidizes the electron transport protein cytochrome c (942). Reaction of peroxynitrite with Zn/S clusters such as that present in eNOS, causes dysfunction of the enzyme as outlined above (sect. IIIB3) (78, 526). Additionally, peroxynitrite reacts rapidly with Fe/S clusters leading to the inactivation of critical enzymes such as mitochondrial aconitase (155, 401).

Peroxynitrite is perhaps most well recognized for its reaction with cysteine and tyrosine residues in proteins. Peroxynitrite-dependent cysteine oxidation results in the formation of a sulfenic acid intermediate that can further be oxidized to a disulfide. This type of oxidation modulates the activity of a number of enzymes. For example, peroxynitrite-mediated oxidation of ETC components such as complex I and complex V results in inhibition of mitochondrial function (767). Oxidation of tyrosine phosphatases results in the inhibition of their enzymatic activity and the aberrant propagation of phosphorylative signaling (926). In contrast, oxidation of cysteine residues on metalloproteinases results in the activation of these enzymes, which may also be detrimental to tissue integrity and repair (992). In addition to proteins, peroxynitrite can directly oxidize reduced glutathione, depleting the antioxidant capacity of the cell (37).

Beyond cysteine oxidation, peroxynitrite can mediate the covalent addition of a nitro group (−NO₂) to the aromatic ring of tyrosine, resulting in tyrosine nitration. This modification is measured as the hallmark of peroxynitrite presence and is traditionally viewed as a sign of nitrative or oxidative stress (78, 765). Importantly, protein nitration can also occur through the reaction of nitrite with heme proteins in the presence of hydrogen peroxide (153, 154). Thus, methodology beyond nitrotyrosine measurement is required to confirm the presence of peroxynitrite in biological systems.

Although a growing number of proteins have now been shown to be nitrated, multiple studies have shown that nitration is a selective process (47, 484). Tyrosines for which the aromatic ring is closer to the surface of a protein, in which there is a neighboring negative charge, and those in hydrophobic regions are more likely to be nitrated (66, 877). Several studies provide comprehensive lists of proteins identified to be nitrated in physiology and disease (44, 73, 381, 709). These proteins include, but are not limited to, enzymes involved in antioxidant defenses such as MnSOD (219, 220), detoxification enzymes such as CYP (587), nucleic acid regulatory proteins such as histone deacetylases (457) and histones (453, 504), and cell structural proteins such as tubulin (730, 732) and myosin heavy chain (648). Importantly, the nitration of these proteins decreases enzyme activity, disrupts metabolism and cellular detoxification, perturbs cytoskeletal organization, and ultimately contributes to the cytotoxic effects of peroxynitrite.

Although peroxynitrite is traditionally regarded as a harmful species, it should be noted that it may have some beneficial signaling properties. For example, lipids are also a significant biological target for peroxynitrite, and although the species can abstract a hydrogen atom to initiate lipid peroxidation, a detrimental process that can lead to cytotoxicity in many cases (65, 766), peroxynitrite can also mediate lipid nitration (54, 1027). Nitrated fatty acids are a class of electrophilic signaling molecules which have been shown to react with a number of cellular targets to mediate protective inflammatory and metabolic signaling (830). Primary mechanisms of nitrated fatty acid signaling include activation of PPARγ (55, 583) and Michael addition to thiol-containing proteins in a number of signaling pathways (53, 955). Through these signaling mechanisms, nitrated lipids mediate a number of physiological effects in animal models including attenuation of inflammatory bowel disease (113), protection after ischemia/reperfusion (683, 801), decreases in atherosclerosis (800), and reversal of obesity-induced pulmonary hypertension (495).

B. HOCl

Hypochlorous acid (HOCl) is generated by the enzyme MPO, localized within leukocytes by the following reaction (300) (Eq. 25)

$${\text{H}}_{\text{2}}{\text{O}}_{\text{2}}+{\text{Cl}}^{-}\to \text{HOCl}+{\text{OH}}^{-}$$ (25)

This highly reactive oxidizing and chlorinating agent plays a major role in the microbicidal actions of neutrophils (15, 1021). Although HOCl reacts with a variety of biomolecule targets (755), in proteins, it predominantly oxidizes cysteine and methionine residues to disulfide and methionine sulfoxide, respectively (726). With regards to metal centers, HOCl can react with heme and Fe/S centers to mediate oxidative bleaching. Additionally, HOCl reacts more slowly with tyrosine residues to mediate the formation of 3-chlorotyrosine (1019). Notably, 3-chlorotyrosine is utilized as a marker of the presence of both MPO (as this is the only enzyme known to generate HOCl) and HOCl. Oxidation of amino acids and metal centers of enzymes results in the HOCl-dependent impairment of a number of critical enzymes.

The role of HOCl in the modification of proteins and lipids is particularly well studied and pertinent to the pathogenesis of vascular disease. In the context of atherosclerosis, HOCl has been shown to oxidize both high and low density lipoproteins (90, 470, 881, 1042). Furthermore, chlorinated proteins and lipids are present in atherosclerotic plaques (90, 415, 621). Hypochlorous acid decreases eNOS activity and NO bioavailability by a number of direct and indirect mechanisms, disrupting protective NO signaling (2, 464, 1054). The oxidation and chlorination of proteins, such as matrix metalloproteinases, leads to detrimental vascular remodeling (314).

C. Hydroxyl Radical

The hydroxyl radical (OH^·) is one of the most reactive species formed in biological systems. In this context, OH^· is generally formed as a product of Fenton reactions involving free iron or copper ions (Fe²⁺; Cu⁺) and hydrogen peroxide (Eq. 26) (281, 986) or from the reaction of hydrogen peroxide with superoxide (Haber–Weiss reaction) (Eq. 27) (394).

$${\text{Fe}}^{2+}+{\text{H}}_{\text{2}}{\text{O}}_{\text{2}}\to {\text{Fe}}^{3+}+{\text{OH}}^{·}+{\text{OH}}^{-}$$ (26a)

$${\text{Cu}}^{+}+{\text{H}}_{\text{2}}{\text{O}}_{\text{2}}\to {\text{Cu}}^{2+}+{\text{OH}}^{·}+{\text{OH}}^{-}$$ (26b)

$${\text{O}}_2{}^{-·}+{\text{H}}_{\text{2}}{\text{O}}_{\text{2}}\to {\text{OH}}^{·}+{\text{O}}_2+{\text{OH}}^{-}$$ (27)

The process is generally more complex than indicated in Eq. 26 and includes the formation of a transient metal/peroxide complex that can react in a different fashion depending on the metal ligands and possible substrates. Therefore, in some cases, the reaction may not produce free hydroxyl radical but other metal-based oxidizing species (759, 1020).

Although the endogenous production of OH^· in healthy conditions is apparently negligible (474), the excess of hydrogen peroxide and superoxide in oxidative stress conditions can yield an increased amount of this toxic species (1072).

In physiology, perhaps the most common source of Fenton chemistry is the presence of increased concentrations of free iron and/or copper (122, 458, 959). In the cardiovascular system, this can be related to aging (122) or hemolytic disease (1026). The source of this iron is most likely plasma-released Hb; whereas Hb or heme are not catalysts for the Fenton reaction, the reaction of plasma Hb with peroxides can cause the release of free iron (389, 759).

The hydroxyl radical reacts with multiple organic substrates including, but not limited to, DNA, lipids, and proteins at rates near the diffusion limits. This lack of specificity makes the use of antioxidants to prevent hydroxyl radical toxicity impractical. Ideally, the best strategies will be based on the chelation of the active metal ions in an environment that blocks their redox reactivity. Proteins like ceruloplasmin, transferrin, or lactoferrin limit the reaction (388, 1020). The binding of NO to heme iron can also limit heme reactivity and prevent iron release from heme (482).

D. Other Species

The reactions of NO can generate other reactive species that have a poorly characterized role in vivo. Among these species, we will mention the nitrogen dioxide radical (NO₂^·), dinitrogen trioxide (N₂O₃), and carbonate radical (CO₃^·−) (46, 692).

1. Nitrogen dioxide

Nitrogen dioxide (NO₂^·) is formed in the reaction of NO with oxygen (Eq. 28) (509, 1018). A more reactive species than NO, it is often used as a surrogate for the detection of NO. Other formation routes include the reactions of nitrite with oxygenated Hb (501) and peroxynitrite decomposition (46).

$${\text{NO}}^{·}+{\text{O}}_2\leftrightarrow {\text{O}}_{\text{2}}{\text{NO}}^{·}$$ (28a)

$${\text{O}}_{\text{2}}{\text{NO}}^{·}+{\text{NO}}^{·}\to 2{\text{NO}}_2{}^{·}$$ (28b)

2. Dinitrogen trioxide

Dinitrogen trioxide (N₂O₃) is a strong nitrosating species also formed in the oxidation of NO (Eq. 29) (691, 692, 1018).

$${\text{NO}}_2{}^{·}+{\text{NO}}^{·}\to {\text{N}}_{\text{2}}{\text{O}}_{\text{3}}$$ (29)

As discussed previously in section IIB2, the formation of N₂O₃ in the red blood cell via an anhydrase reaction involving Hb, NO, and nitrite (69) constitutes a possible pathway to preserve NO bioactivity (367).

3. Carbonate radical

The carbonate radical can be formed in vivo by the reaction of the hydroxyl radical with carbonate or bicarbonate (138) or via the reaction of peroxynitrite with carbon dioxide (Eq. 30) (112, 608, 640).

$${\text{OH}}^{·}+{\text{CO}}_3{}^{2-}\to {\text{OH}}^{-}+{\text{CO}}_3{}^{·-}$$ (30a)

$${\text{OH}}^{·}+{\text{HCO}}_3{}^{-}\to {\text{H}}_{\text{2}}{\text{O}}^{-}+{\text{CO}}_3{}^{·-}$$ (30b)

Several examples of biological oxidations mediated by the bicarbonate anion, in particular through interactions with peroxynitrite, have been reported in the literature [e.g., (46, 635, 918, 1056)]. In relation to cardiovascular biology, it is important to note that XO has also been characterized as a target and source of carbonate radical (111).

V. CROSS-TALK BETWEEN ROS AND RNS SOURCES

Our description of sources of NO and ROS may give the impression of a rigorous separation of roles for each protein or enzymatic system. This is certainly not the situation in physiological systems. We have discussed how some enzymes, in particular eNOS, can produce different species in response to posttranslational modifications or lack of cofactors or substrates (sects. IIA1 and IIIB). Current research has greatly advanced our knowledge on how different systems can influence the NO/ROS generation from other sources in which it has been termed ROS-induced ROS release (1066, 1069).

Mitochondria and NOX are the main effectors of downstream signaling to other sources and to each other. Mitochondrial uncoupling and increased ROS generation activates the mitochondrial permeability transition pore (mPTP), allowing the release of ROS to the cytosol (1069). In response to the increased levels of ROS, the PKC is activated, which in turn, can activate NOXs (319, 776). Other routes of mitochondrial triggered activation of NOX involve the activation of NOX1 via activation of PI3K and Rac1. The blockade of mitochondrial ROS by rotenone was shown to prevent NOX1 activation (FIGURE 17) (564).

FIGURE 17.

Cross-talk between mitochondria, NADPH oxidase (NOX), and other reactive oxygen species (ROS) sources. Mitochondrial ROS are released to the cytosol via the mitochondrial permeability transition pore (mPTP). These can cause activation of PKC and phosphatidylinositol 3-kinase (PI3K) kinases and Rac1 GTPase with subsequent activation of NOX enzymes and further ROS production. Alternatively, NOX can release ROS that activate protein kinase C, isoform ε (PKCε) and in turn activate the mitochondrial ATP-dependent potassium channels (K_ATP), increasing mitochondrial ROS. Cytosolic ROS can increase ROS-generating activity of endothelial nitric oxide synthase (eNOS) and xanthine oxidase (XO).

Indeed, activated NOX can in turn, increase ROS production by the mitochondria (11, 116). This process appears to require the function of mitochondrial ATP-dependent potassium channels (K_ATP), as shown by the ability of K_ATP inhibitors to block the increase of mitochondrial ROS production after angiotensin II–induced activation of NOX (242, 507). Of note, the activation of NOX via angiotensin II and the angiotensin II type 1 receptor is mediated by PKC; however, the isoform PKCε is also an activator of the K_ATP channel, which, as indicated, also stimulates mitochondrial ROS production (FIGURE 17) (187, 242).

Another case of significant cross-talk between NOX and mitochondria has been provided by studies of nitroglycerin resistance. In this case, the development of tolerance was mainly due to mitochondrial ROS release and was independent of NOX, whereas the activation of NOX, triggered by mitochondrial ROS, was the effector of the endothelial dysfunction associated to nitroglycerin tolerance (1004).

Secondary to the main role of mitochondria and NOX, but not to be ignored, XO and eNOS can also amplify the formation of ROS; for instance, XO is indirectly activated by angiotensin II, increasing cellular ROS production (551). A direct role of mitochondrial ROS on XO activation has also been described (362). As discussed in previous section, increased ROS eventually causes eNOS uncoupling via BH₄ depletion, Zn/S cluster breakdown, and other routes, further perpetuating the generation of superoxide and other ROS (FIGURE 17) (488, 570).

The existence of many different routes interconnecting ROS/RNS sources may suggest that the normalization of dysfunctional conditions is a herculean task; however, it also presents the promise of normalizing a disease state by the direct inhibition of a single source of oxidative stress. Recent work indicates that this can be a feasible approach (206, 207, 234, 836). The development of novel, specific mitochondrial inhibitors (discussed in sect. IIIC3) is a promising step in this direction (8, 679, 923, 1062).

VI. CONCLUDING REMARKS: ADVANCES, CHALLENGES, AND THERAPEUTIC OPPORTUNITIES

Over the last decade, our knowledge about NO and ROS generation has increased substantially. From the initial ideas considering that oxidative species need to be suppressed and that oxidative stress could be combated with adequate antioxidant treatment, we now can appreciate the critical signaling properties of NO, hydrogen peroxide, or superoxide.

The initial approach to the treatment of oxidative stress was based on the supplementation with different antioxidants, in particular vitamins C and E and β-carotene. The use of antioxidants has been markedly ineffective when not damaging (97, 390, 400, 785). In fact, the situation has been termed the “antioxidant paradox” (399, 400). The failure of these studies, along with their limitations, has been discussed at length in this review, and a number of explanations for this phenomenon have been proposed (67, 158, 252, 400, 672, 949).

At the same time, we are increasingly appreciative of the cross-talk between different radical species, generation pathways, and cellular compartments. Different examples of species interaction have been discussed throughout this review; for instance, the oxidation of BH₄ by superoxide will trigger eNOS uncoupling with decreased NO production and increased superoxide production. In another example, NO, through its high affinity for heme groups or via S-nitrosation, can act as an important modulator of NOX (176, 315, 846) or can also modulate the insertion of heme in the receptor sGC. The generation of NO through NOS enzymes and oxidative pathways in normoxic conditions is supplemented by the actions of reductive, nitrite-dependent pathways in hypoxia (603, 968). We also note how some proteins are able to switch, in an oxygen-dependent manner, from NO scavengers into nitrite-reducing, NO-producing systems. The ability to interact with different oxidation states of sGC and use pharmacological tools to recover sGC-dependent signaling provides a new tool for the treatment of conditions in which NO signaling is compromised. Further research aimed to the understanding of these dynamics will be critical for the treatment of endothelial dysfunction and other redox imbalance-mediated pathologies.

The interactions between mitochondria and ROS-producing proteins, especially NOXs, are of particular current interest (116, 206, 234, 836). Mitochondria are a physiologically significant source of ROS with important implications in physiology, not only in cardiovascular disease but in many other conditions like aging (57, 84, 136, 165, 170, 290, 864, 997).

It is also significant that many of our advances have evolved not only from the clinical field but also from basic science. Chemical knowledge of NO and ROS reactivity has been available well before its biological relevance was established. This knowledge is critical for the understanding of the biology of ROS and can provide new tools and strategies for the study of ROS biology (232). The studies on purified protein systems, although necessarily simplistic, have provided insights on the reactivity of most proteins involved in NO and ROS biology that have been widely validated in vivo. Conversely, the study of genetic pathologies such as chronic granulomatous disease has helped to establish the relevance of NOX proteins even before these were thoroughly characterized.

In summary, we have progressed from a simpler approach leveraging antioxidant vitamins and ROS scavengers, which failed, to a more fundamental discovery of the enzymatic sources and reaction pathways of ROS. The future treatments need to address these issues. It is clearer now that specific inhibitors directed to the sources of ROS can yield more effective therapies. A better understanding of NOS function has allowed the development of new strategies to couple eNOS and target downstream effectors like sGC. Advances on NOX research have led to the development of novel inhibitors on their way to widespread use for the treatment of cardiovascular disease and other pathologies. New highly specific inhibitors of other ROS sources like the XO-targeted febuxostat and mitochondrial-targeted drugs provide additional pathways to modulate cellular ROS production. We foresee that the availability of these specific drugs and a better biomarker-based diagnosis of the dysregulated systems will allow for more personalized treatments.

Finally, antioxidants targeted to specific cells or organelles can still become relevant in the management of oxidative stress pathologies, in particular, but not limited to, mitochondria-targeted antioxidants (873). The development of these treatments directed toward specific organs, tissue, and cell types can be one of the next big advances in the field.

GRANTS

This work was supported by National Institutes of Health Grants R01 HL098032, R01 HL125886, P01 HL103455, T32 HL110849, and T32 HL007563 (to M.T. Gladwin), R21 ES027390 (to J. Tejero), and funding from the Institute for Transfusion Medicine and the Hemophilia Center of Western Pennsylvania (to M.T. Gladwin and S. Shiva).

DISCLOSURES

M.T. Gladwin is a coinventor on a National Institutes of Health government patent application for the use of nitrite salts in the treatment of cardiovascular diseases. J. Tejero and S. Shiva do not declare any conflicts of interest, financial or otherwise.