Role of Nitric Oxide Carried by Hemoglobin in Cardiovascular Physiology: Developments on a Three-Gas Respiratory Cycle

Abstract

A continuous supply of oxygen is essential for the survival of multi-cellular organisms. The understanding of how this supply is regulated in the microvasculature has evolved from viewing erythrocytes (red blood cells, RBCs) as passive carriers of oxygen to recognizing the complex interplay between hemoglobin and oxygen, carbon dioxide, and nitric oxide – the three-gas respiratory cycle – that insures adequate oxygen and nutrient delivery to meet local metabolic demand. In this context, it is blood flow not blood oxygen content that is the main driver of tissue oxygenation by RBCs. Herein we review the lines of experimentation that led to this understanding of RBC function; from the foundational understanding of allosteric regulation of oxygen binding in hemoglobin in the stereochemical model of Perutz, to blood flow autoregulation (hypoxic vasodilation governing oxygen delivery) observed by Guyton, to current understanding that centers on S-nitrosylation of hemoglobin (aka S-nitrosohemoglobin; SNO-Hb) as a purveyor of oxygen-dependent vasodilatory activity. Notably, hypoxic vasodilation is recapitulated by native SNO-replete RBCs and by SNO-Hb itself, whereby SNO is released from hemoglobin and RBCs during deoxygenation, in proportion to the degree of hemoglobin deoxygenation, to regulate vessels directly. In addition, we discuss how dysregulation of this system through genetic mutation in hemoglobin or through disease is a common factor in oxygenation pathologies resulting from microcirculatory impairment, including sickle cell disease, ischemic heart disease and heart failure. We then conclude by identifying potential therapeutic interventions to correct deficits in RBC-mediated vasodilation to improve oxygen delivery—steps towards effective microvasculature-targeted therapies. To the extent that diseases of the heart, lungs, and blood are associated with impaired tissue oxygenation, the development of new therapies based on the three-gas respiratory system have the potential to improve the well-being of millions of patients.

OVERVIEW

Disruptions in blood flow and oxygen delivery are hallmarks of the major ailments that afflict modern society, with diseases of the heart, lung, and vasculature directly accounting for 36% of the 2.5 million American deaths annually – and indirectly accounting for another 6% of deaths of Americans whose primary cause of death was identified as diabetes, kidney disease, or liver disease.^1,2 Diseases of blood flow and oxygen delivery have become a global concern as ischemic heart disease, stroke, and chronic obstructive pulmonary disease kill close to 18 million annually.³

On the surface, diseases of the heart, lungs, and vasculature can be viewed as heterogeneous conditions with broad-based contributions from genetics, environment, diet, and activity level. They involve numerous molecular pathways and offer many therapeutic targets. Nevertheless, these diseases all share dysfunction at the level of the tissue microvasculature where oxygen exchange occurs and for which targeted interventions are lacking. Given that the primary function of the heart and lungs is to oxygenate tissues, surprisingly little is known about the control of oxygen delivery, and therapies that effectively target the microvascular origins of hypoxemia remain a holy grail of cardiovascular medicine.

In this review, we take a focused look at the underpinnings of oxygen delivery to tissues. Originally presented by Max Perutz in terms of an allosteric model for cooperative oxygen binding and release from hemoglobin (Hb),^4,5 this is now understood to be only half the story. The other half entails regulation by Hb of microcirculatory blood flow supplying that oxygen, i.e., understanding how red blood cell (RBC)/Hb access to tissue microcirculation is matched to local tissue oxygen demand. Our studies show that nitric oxide (NO) carried by Hb dilates the microvasculature to increase local blood flow and thus oxygen delivery (autoregulation of flow). Specifically, although free nitric oxide (NO) cannot exist in meaningful amounts in the blood, as it is rapidly sequestered by hemes and eliminated in reactions with OxyHb^6,7, NO bioactivity is preserved in RBCs through its conjugation to Cys thiols in Hb to form an S-nitrosothiol (SNO), S-nitrosohemoglobin (SNO-Hb).^8,9 SNO-Hb is especially well-suited to regulate local blood flow because the release of SNO from Hb is linked thermodynamically to the release of oxygen,^10,11 providing the basis for the expanded three-gas respiratory cycle.

Here we detail the essential roles of NO processing by Hb in respiratory cycle physiology – from RBC function to microvascular blood flow – that provide the foundation for mammalian life. We illustrate how this interaction of NO and Hb serves as the basis for novel therapeutics to enhance RBC function.

Part 1. The Traditional View of Blood Flow and Tissue Oxygenation

Microvascular Blood Flow

i. Hypoxic control of the vascular circuit

The centrality of blood and blood flow to mammalian physiology has made it a focus of study by physicians and scientists for millennia. Only relatively recently have imaging techniques allowed visualization of capillary blood flow, represented by RBC passage through capillaries in single file and in intimate contact with the endothelium (Figure 1A). The RBC and endothelial lining within tissues may be viewed as an integrated vascular unit (IVU), highlighting the opportunity to target the microvasculature through RBCs.

Figure 1. Tissue blood flow is controlled by hypoxic autoregulation during arterial-venous transit.

A. RBCs traversing microvascular capillaries in single-file requires RBC deformation due to intimate contact of RBC and endothelial cell membranes. RBCs adjoining the endothelium constitute an integrated vascular unit. B. Blood flow through the canine hind limb increases linearly as arterial blood oxygenation decreases, demonstrating that hypoxic vasodilation to maintain oxygen delivery is directly and effectively coupled to HbO₂ saturation, not to P_O2 or P₅₀. C. Reactive hyperemia of the gastrocnemius muscle (shown as % over basal flow) in mice bearing human wildtype hemoglobin (C93), or where β-globin is mutated (C93A) and cannot bear SNO at this site. Modified panel A from Ref²⁶⁸, panel B from Ref¹³, panel C from Ref⁹, with permission.

One of the earliest investigations in this regard was reported by Roy and Brown in the their 1880 paper, The Blood-Pressure And Its Variations in the Arterioles, Capillaries and Smaller Veins.¹² After devising ingenious methods to quantify pressure and flow in small blood vessels, they conducted multiple assessments of vascular beds at baseline, under conditions of anemia and following application of agents that could constrict or relax blood vessels. With these efforts they noted “…certain restrictions must be made in applying the term “contractility” to express the nature of these vital changes in the calibre of the capillaries. It is possible that the term “change of elasticity” may in reality be more appropriate…with smooth muscles, and with the capillary walls…the expansions and contractions of which they are capable, may be the result of a change in the statical relations of the ultimate molecules”.¹² Roy and Brown’s interpretation was exceedingly prescient in setting the stage for the discovery of hypoxic vasodilatation some 80 years later.

In the 1960s, Arthur Guyton set out to investigate the regulation of local blood flow, to expand on Roy and Brown’s work to “elucidate further the basic mechanism by which local blood flow is linked to each tissue’s need for nutrients”.¹³ He designed experiments where canines were instrumented with catheters and pressure monitors that allowed the rate of blood flow through the hind limbs to be measured as a function of Hb oxygen saturation. Specifically, one pulmonary vein was connected to one femoral artery, so that limb muscle blood flow could be measured while blood oxygen saturation (within the isolated limb and lung) was varied by regulating the fraction of inspired oxygen (FiO₂). The resultant data provided definitive proof for hypoxic auto-regulation: as Hb oxygen saturation was reduced in stages, flow through the limb increased proportionately (Figure 1B). Direct infusion of RBCs with reduced oxygen saturation precisely recapitulated the flow responses. Thus, tissues actively maintain an adequate supply of oxygen by controlling their own blood flow.¹³ This effect is independent of altered metabolic activity of the muscle (and metabolites).

Guyton initially proposed that auto-regulation was driven by oxygen availability; he also considered the idea that the lungs removed a vasoconstrictor substance from RBCs. Ultimately, he expressed frustration at the lack of progress on solving this problem.¹⁴ As it turns out, the answer would entail oxygen-dependent release of a vasodilator, and the mechanistic clues were to be found in two key observations:

1. Vasodilation was produced upon infusion of RBCs into the hindlimb.

2. The degree of vasodilation by RBCs was inversely proportional to oxygen saturation of Hb in preparations (but independent of P_O2). Vasodilation by RBCs is thus a linear function of Hb saturation, ranging from 100–0% saturation, which implicates Hb conformational state (allostery),^4,5 as described below.

ii. Hemoglobin allostery in control of microvascular blood flow

Hemoglobin oxygen saturation is known to be the primary determinant of Hb structure through the allosteric transition between oxygenated R-state vs deoxygenated T-state (see below).^4,5 Thus, the Guyton data implicated the release of a vasodilator from the deoxygenated structure of Hb. That such a mechanism is operative in vivo is clearly supported by data showing that normal mice exhibit elevated muscle blood flow after brief ischemia (reactive hyperemia), but this hypoxia-induced flow increase is markedly impaired in mice bearing hemoglobin with a single point mutation, cysteineβ93 to alanine, described in detail in Part 2 (Figure 1C); it should be noted that Cysβ93 reactivity is linked allosterically to oxygen binding.^10,11

iii. Role of RBCs versus endothelium

In current understanding¹⁵, the flow of blood is controlled most obviously by the pumping action of the heart, but the vascular system itself is an integral component of blood flow regulation at multiple levels. In major arteries, blood ejection from the heart under pressure induces stretching of vessel walls, whose resistance maintains mean arterial pressure well over mean venous pressure even when the heart is not ejecting blood. The compliance of arterial resistance vessels is actively regulated through the tone of surrounding smooth muscles. This level of regulation is often mediated or modulated by vasoactive substances released from local nerves or into general circulation (angiotensin II, bradykinin, acetylcholine, noradrenaline, etc) acting on G protein-coupled receptors often used as targets for drugs to control blood pressure, but many vasodilators share a common endothelial cell-derived relaxation factor (e.g., nitric oxide (NO)). As arteries divide into arterioles within tissues, blood flow through individual arterioles into specific capillary beds is also regulated by smooth muscles acting as sphincters that respond to circulating vasoactive signaling mediators, local signals and nerve input to increase or decrease the flow through the arteriole into a specific capillary bed versus through shunt vessels.¹⁵ Notably, the G protein-coupled receptor and canonical NO/PKG pathways utilized by circulating, neuronal or locally-derived vasoregulatory substances are too slow to control moment-to-moment blood flow (i.e., RBCs experiencing local hypoxia are gone by the time the vasculature can respond via these signals) and are utilized to regulate longer-term basal blood flow changes.^16,17 In fact, eNOS expression is lost in small vessels controlling tissue oxygenation. One prominent model posits that regulation of flow through such feeder arterioles is controlled by metabolic demand and hypoxia in tissues downstream of the arteriole, where upstream arteriole-associated muscles that see relatively constant flow of highly oxygenated blood detect distant downstream metabolic demand.¹⁸ One suggestion is that endothelial cells coupled by gap junctions conduct calcium-based and other electrical signals back out to arterioles.^19–21

Here we focus on the role of RBCs as mediators of flow subserving oxygenation. RBCs traverse capillaries in plug-flow mode; that is, in single file and in intimate contact with the endothelium (see Figure 1A). This ensures effective gas exchange, but may require that both the RBC and capillary wall/endothelial cell deform during passage. Capillaries devoid of smooth muscle still are able to regulate their diameter and compliance through the engagement of surrounding pericytes (which have contractile apparatus) as well as by their own intrinsic cytoskeletal regulation.^17,22–24 It is in this intimate tissue microcirculation, including small arterioles and capillaries (i.e., within the IVU, see above), that RBCs themselves are most able to regulate tissue blood flow and tissue oxygen delivery. Once arteries and arterioles are maximally dilated, remaining resistance (roughly half of normal) is primarily due to capillaries, indicating that tissue microcirculation is among the main drivers of systemic vascular resistance. Since resistance to flow in a tube is proportional to the inverse-fourth power of the radius, even small changes in arteriole and capillary diameter can dramatically affect blood flow through a tissue (and thus its oxygenation).²⁴

Red Blood Cells and Hemoglobin in Tissue Oxygenation

i. RBC flow in the microcirculation

Each human RBC is 6–8 μm in diameter and circumnavigates the vasculature every 40–60 seconds. Their characteristic biconcave disk shape maximizes surface relative to volume, facilitating rapid gas exchange to and from the Hb within. Furthermore, the RBC cytoskeleton and shape are adaptive, and RBC rheology (the ability to deform) is essential for effective tissue perfusion, since loss of deformability (as in sickle cell disease) reduces oxygen delivery. RBCs traffic through miles of microvasculature in series; arterial-venous transit times are on the order of a second (Figure 1A). The small micro-vessels do not express eNOS, avoiding the flow-limiting vasoconstriction that would result from NO sequestration by RBCs (Text Box 1). RBCs in fact serve as their own source of SNO to regulate blood flow (see below).

Text Box 1. Strict conservation of amino acids in Hb that are required for oxygen delivery through carriage and blood flow. Tissue oxygenation is a function of O₂ content and flow of blood. His/Phe in Hb determines O₂ content; βCys93 regulates blood flow.

1. O₂ Content × Blood Flow ≈ O₂ delivery

2. O₂ Content ≈ Hb−(His/Phe)−Fe

3. Blood Flow ≈ Hb−(Cys93)

ii. Conservation of residues in Hb that regulate oxygen delivery

Hemoglobin is a tetrameric protein with a molecular weight of 65 kDa. Hb was a focus of extensive structural and biochemical analyses throughout the 20^th century, making it the most studied protein. Hb would also provide a foundation for the concept that allostery governs the posttranslational regulation of proteins – “the second secret of life”²⁵ – and garner Hb the moniker “honorary enzyme”. A more recent understanding of Hb/NO interactions reveals an essential (no longer honorary) enzymatic activity of Hb in converting NO to SNO, to regulate oxygen delivery in the respiratory cycle.^11,26,27

Adult Hb protein consists of two α and two β globin chains. The mature α-chains have 141 amino acids while the β-chains have 146.^4,28 Each chain contains 8 helical segments (A through G) with one heme molecule positioned between helixes E and F. There are over 1,000 human Hb variants^29,30 whose effects on oxygen affinity and respiratory physiology range from negligible to severe life-shortening decrements (notably HbS in sickle cell disease, the result of an amino acid switch from glutamate to valine at position 6 of the β-chain).^31–33 Despite this degree of variation, there is significant sequence homology across species, although only 3 residues are strictly conserved within humans and most higher-order species. The first are the proximal histidines in the E and F helixes (His58 and His97 in α-globin; His 63 and His 92 in β-globin) that bind the heme iron.^34–36 The second residue is the phenylalanine at position 43 in α-globin and 42 on β-globin (Phe43/Phe42),^37,38 which locks the heme group in place. All these residues are required for Hb to bind O₂. The third strictly conserved residue is the Cys at position 93 of the β-chain (βCys93)^20,39, which binds NO.⁸ Notably, the reactivity and function of this Cys is linked to binding of oxygen (i.e., so called ‘thermodynamic linkage’) (Figure 2A) (although this residue has only trivial effects on oxygen affinity per se).¹¹ Multiple lines of evidence (detailed in the following sections) point to βCys93 as having a key role in the regulation of blood flow governing oxygen delivery, through release of NO in an O₂ dependent manner (Text Box 1). Alternatively stated, allosteric regulation of Hb by release of oxygen promotes release of NO bioactivity to regulate blood flow.

Figure 2. Oxygen-driven conformational changes in SNO-βCys93-Hb allow SNO release.

A. Crystal structure of nitrosylated R-state hemoglobin. The β-globin hemes (with iron-coordinating His63 and His87 residues and heme-stabilizing Phe42 residue) and the reactive βCys93 residues, are NO bound. Cys sulfur atom, yellow; NO nitrogen atom, blue; NO oxygen atom, red; hemes shown as ball-and-stick, and β-globin backbone as ribbon. Derived from 1buw structure in Ref²⁶⁹ using PyMol. Note: SNO is released from Hb in the T-state, preventing crystallization. B. Model comparing SNO-Hb βCys93 in the oxygenated R-state (red backbone) with SNO buried (left), with deoxygenated T-state (blue backbone) with SNO exposed for NO bioactivity release (right). Cys sulfur atom, green; NO nitrogen atom, blue; NO oxygen atom, red. C. Isolated SNO-Hb (2SNO-Hb[FeO₂]₄) in the oxygenated state (21% O₂; R-state) (left) induces contraction of blood vessels (aortic rings) through scavenging of NO derived from endothelium, but SNO-Hb in the deoxygenated state (<1% O₂; T-state) (center) induces relaxation through release of SNO. Hb Cys-SNO content and O₂ saturation are linked and linearly related to vessel relaxation activity (right), grey points extrapolated according to Refs.^11,101 D. Blood flow through the substantia nigra of rat brain is dependent on SNO-Hb and blood oxygenation (at the indicated measured tissue P_O2). SNO-repleted (circles) or native (squares) Hb was infused in femoral vein over 3 min starting at time 0. SNO-Hb mediated vasodilation is allosterically regulated by O₂ release from Hb and thus proportional to tissue hypoxia. Modified panels B and D from Ref⁸⁴, panel C from Refs^104,106, with permission.

In this regard, classic work by Bohr and by Haldane⁴⁰ on the influence of pH and carbon dioxide on Hb oxygen affinity and on the reciprocal influence of Hb oxygen saturation on hydrogen ion and carbon dioxide binding, respectively, helped characterize the allosteric cooperative binding and release of oxygen occurring within the RBCs to delineate the respiratory function of blood. Perutz provided a mechanistic explanation for this cooperative allostery through two distinct structures of Hb: an oxygenated/relaxed (“R”) structure (after binding oxygen in the lungs) and a deoxygenated/tense (“T”) structure (after release of oxygen in tissues).^4,28 Oxygen binding to and release from Hb drives the highly-cooperative allosteric transition between T and R states that allows the all-or-none action of Hb to tightly bind 4 diatomic oxygens under high P_O2 but also release all 4 oxygens in low P_O2.^4,28 A remaining problem, however, would be to integrate this function of Hb with blood flow through tissues, which represents the primary determinant of oxygen delivery under all but the most extreme conditions. While Guyton would coin the term “autoregulation of blood flow” and ascribe RBCs to its control, he also noted that the actual control mechanism remained unknown.¹³

Part 2. THE HUMAN RESPIRATORY CYCLE IS A THREE-GAS SYSTEM: OXYGEN, NITRIC OXIDE, AND CARBON DIOXIDE

The Third Gas

i. Historical context: Nitric oxide, S-nitrosothiols and S-nitrosylation

Until the late 1980s, NO gas was considered primarily an air pollutant and irritant, before becoming appreciated as a signaling ‘gasotransmitter’. Palmer and colleagues,^41,42 and Ignarro and colleagues,⁴³ building on work from Furchgott⁴⁴ and Hibbs,⁴⁵ characterized the enzymatic generation of NO by vascular endothelial cells. Three forms of nitric oxide synthase (NOS) are known, deriving their names from the originating tissues (n, neuronal; e, endothelial; and i, inducible). nNOS and eNOS (responsible for generating endothelium derived relaxing factor (EDRF), whose identification as NO is credited to Furchgott, Ignarro and Moncada) are constitutively expressed highly in neurons and endothelium, respectively, while iNOS expression in inflammatory cells is rapidly induced in response to infection.⁴⁶

It should be appreciated that ~50% of NO bioactivity in blood, as measured by NO/SNO in RBCs and plasma, derives from eNOS that resides in vascular endothelium and to a lesser extent in RBCs.^8,47 Similarly, eNOS contributes ~50% of the NO metabolites (primarily nitrite) circulating in plasma. It follows that nNOS and iNOS within tissues also contribute substantially. There is also growing appreciation of food-derived and microbiota sources of NO⁴⁸ that enter the circulation in the form of SNOs and nitrite/nitrate.⁴⁹ Ultimately, all NO metabolites find their way into the circulation. The final metabolic end-product nitrate is excreted, but other NO-derived species, including NO, SNO, and nitrite, can be processed by Hb into SNO-Hb and used for regulation of tissue oxygenation (see below). Non-specific inhibition of NOSs will deplete SNO-Hb by 50 to 80% and is tolerated⁸, whereas nearly complete elimination of SNO-Hb compromises tissue oxygenation and decreases survival⁹ (see below).

Mechanistic insight into how NO dilates blood vessels can be attributed to Murad,⁵⁰ who proposed that NO is the vasoactive product of nitroglycerin that activates guanylyl cyclase (see below). However, only in 2002 was it determined how nitroglycerin actually produces NO bioactivity (i.e., enzymatically via Aldehyde Dehydrogenase 2).^51,52 The research on nitroglycerin that presaged identification of endothelium-derived relaxing factor as NO by Furchgott,⁴⁴ Ignarro,^53,54 and Moncada⁷ et al is a thoroughly reviewed story. Yet even as NO received the title “Molecule of the Year” in 1992,⁵⁵ there was growing recognition that NO gas itself did not account for many of the in vivo activities being attributed to it.^56,57 Rather, burgeoning roles for “NO bioactivity” in multiple cellular functions were identified with a class of molecules called S-nitrosothiols (SNOs), reflecting redox activation of NO^. to NO⁺ that reacts with free thiols, including cysteines in proteins but also low molecular weight thiols (i.e., glutathione, cysteine, acetyl-CoA), to form SNOs.^57–60

Proposed by Ignarro as intermediates in nitroglycerin bioactivation,⁶¹ SNOs were discovered as endogenous biomolecules by Stamler and colleagues, and shown to endow NO with manifold activities.^{8,57–59,62,63} A remaining puzzle at that time had been to explain how vascular NO avoided sequestration by hemes in Hb, which should prevent vasodilation (Text Box 2); SNOs solved this mystery as well, since they are impervious to heme. But most importantly, as Stamler conceived, SNOs also serve as the basis of a widespread posttranslational modification with the potential for allosteric regulation of all classes of proteins,^57,58,83 but particularly of Hb,⁸ with ramifications essential for mammalian life.^9,64 Thus, SNO modification (S-nitrosylation) of proteins provided an explanation for many aspects of the cellular influence of “NO”, extending far beyond vasodilation via one single mechanism. Hemoglobin would thus provide a paradigm for NO function in biology, as it had done earlier for the posttranslational modification phosphorylation.

Text Box 2. Classic reactions of NO with Hb that are close to diffusion-limited and effectively irreversible, resulting in NO sequestration and elimination to produce vasoconstriction in vitro.

1. NO + Hb(Fe²⁺) → Hb−Fe²⁺−NO

2. NO + Hb(Fe²⁺)O₂ → Hb(Fe³⁺) + NO₃⁻

ii. Nitric oxide bioactivity in blood and tissues – Central Role for S-nitrosylation

In conjunction with the recognition of NO gas as EDRF was the identification of its receptor: the heme of soluble guanylyl cyclase (sGC).⁵⁰ Binding of NO to the heme-iron in sGC leads to increased conversion of GTP to cyclic GMP (cGMP), which in turn activates protein kinase G (PKG) in vascular smooth muscle. These steps explained how NO produced by the endothelium resulted in vascular relaxation to reduce blood pressure, and provided a foundational paradigm for the field: NO transduction requires heme-iron in sGC.

However, the ability of NO to signal through cGMP to cause vasodilation in fact varies considerably among vessels of different sizes, from location to location in the vasculature, and also among species: for example, while NO-driven vasodilation of murine thoracic aorta is cGMP dependent, rabbit thoracic aorta is only about half mediated through cGMP (and thus half independent of canonical cGMP/PKG) and that in abdominal aorta is entirely cGMP-independent.^65–67 Additionally, this sGC pathway acts in vascular smooth muscle and most prominently in larger vessels; capillaries lack muscle or sGC. Within blood vessels, NO gas also is ephemeral. High concentrations of heme iron in RBC Hb will rapidly sequester vast quantities of NO, lowering the concentration of NO below that necessary to activate sGC; blood flow to tissues would be impaired by systemic vasoconstriction in the absence of additional mechanisms. The solution to this problem was to understand that NO is redox-activated in vivo, entailing formal oxidation of NO to NO⁺, to allow its reaction with thiols.^56,58 S-nitrosothiols formed in this manner are impervious to heme yet also remain endowed with vasodilatory activity; they also retain activity in the microcirculation where NO itself loses effectiveness.⁶⁸ Another conundrum solved by SNOs was to explain the seemingly endless functions of NO in physiology, which are mostly independent of sGC/PKG. Protein sulfhydryl groups, rather than heme iron, provide the substrate for ubiquitous regulation of protein function by SNO. In retrospect, the foundational paradigm of the NO field – transduction via heme – may represent an exception to a more general and widespread rule of S-nitrosylation-mediated cellular signaling through conjugation of NO with sulfhydryls to form SNOs in both proteins and small molecules to transduce the cellular actions of NO. Notably, Cys thiol-based modifications in proteins also provide a basis for redox signaling, and S-nitrosylation is the most developed example in the class: the prototypic redox-based signal.

It is now recognized that protein S-nitrosylation is a ubiquitous post-translational modification and a major distributor of NO bioactivity, since a vast array of the functions of NO in cellular regulation are carried out independently of sGC/PKG^69–71 (and nitrosothiols can also activate sGC/PKG⁶⁹). Thousands of proteins have been identified as targets of S-nitrosylation;^71–73 and activity, subcellular localization, stability, trafficking and interactions of these proteins change in response to addition or removal of SNO.⁷⁴ The breadth of cellular activities regulated by S-nitrosylation may rival those controlled by phosphorylation.^75,76 It is evident that disruption of S-nitrosylation is an important component of many pathologic conditions associated with disorders in oxygenation.⁷⁷

The field of NO biology has traditionally operated on the assumption that, in contrast to other prominent post-translational modifications, protein SNOs are formed non-enzymatically and stochastically. Multiple chemical routes to SNO formation have been proposed, largely on the basis of theoretical analyses. These include the oxygen-dependent oxidative formation of NO-donating intermediates such as N₂O₃. However, this proposed chemistry is subject to kinetic and other constraints in the cellular milieu and is therefore poorly-suited for the ubiquitous regulatory role that has been established for protein S-nitrosylation. Recent evidence shows that protein SNOs are frequently formed by transnitrosylation, the transfer of the NO group from Cys on one protein to free Cys on another,^76,78 by analogy to transfer of ubiquitin by E3 ligases or palmitate by acyl transferases. It has been shown that de novo SNO formation also is enzymatic, entailing SNO synthases that conjugate NO with Cys residues.⁷⁹ Of significance in this regard is the growing appreciation that the redox requirement for SNO formation by NO can be met by the participation of transition metals, in particular by iron in Hb and other proteins. Hemoglobin is thus a paradigmatic SNO synthase, as discussed below, and in fact exhibits a transnitrosylase activity as well, i.e., the ability to transfer its SNO to partner proteins;⁸⁰ these enzymatic functions of Hb form the basis of the mechanism for SNO export that governs autoregulation of blood flow (see below).

SNO Homeostasis in RBCs

i. The Hb paradigm for enzymatic S-nitrosylation

In vivo SNO homeostasis is a balance between S-nitrosylation and denitrosylation. Many thousands of SNO proteins and sites are known, and surveys using algorithms for S-nitrosylation sites predict that the majority of proteins are subject to S-nitrosylation and the majority of these involve conserved sites, including βCys93 in Hb.^73,81–83 Knowledge of the factors and conditions that control site-specific S-nitrosylation is accumulating, including roles for allosteric effectors, SNO motifs, protein-protein interactions and low molecular weight SNO-thiol carrier proteins. These principles of specificity find their origins in significant part in studies of SNO-Hb, which has served as a paradigm for protein regulation by S-nitrosylation.^8,84–87 As one example, the classic acid-base motif for S-nitrosylation derives from residues surrounding the conserved Cys93 in Hb β-chain.⁸⁶

S-Nitrosylation reactions entail a formal one-electron oxidation of NO to NO⁺ that is often catalyzed by transition metals (Text Box 3). The role of iron in catalysis of protein S-nitrosylation was discovered in Hb, which can use either NO or nitrite as substrates for its SNO synthase activity to form SNO-Hb.^11,27,87,88 SNO-Hb also exemplifies transnitrosylase activity: SNO-Hb binds to and transnitrosylates AE1/Band 3 protein at an N-terminal Cys,⁸⁷ as well as protein disulfide isomerase, both of which help shuttle NO activity out of RBCs.^87,89 Thus, Hb exhibits two enzymatic activities: the SNO synthase activity that catalyzes auto-S-nitrosylation using NO or nitrite²⁷ as substrates, and transnitrosylase activity that transfers this SNO to specific partner proteins.^85,90

Text Box 3. NO activated by iron in the Fe(III) state reacts with cysteine thiol to generate a SNO-protein.

$$\text{NO }+{\text{ Fe}}^{3+}\to {\text{NO}}^{+}+{\text{ Fe}}^{2+}$$ Eq 1.

$${\text{NO}}^{+}+\text{ Cys}\to \text{SNO}-\text{Cys}$$ Eq 2.

These enzymatic activities in Hb have provided a conceptual foundation for understanding enzymatic S-nitrosylation in other cells and by other enzymes. Thus, NOSs are found in super-complexes containing enzymes responsible for generating and transmitting SNOs:⁷⁹ SNO synthases to generate de novo SNOs, and transnitrosylases to transfer of SNO to additional targets.^79,90,91 In sum, while NOSs serve as the principal source of NO, protein nitrosylases mediate the iron-to-Cys and Cys-to-Cys transfer of functional NO to target proteins to disseminate the signal. Further, as with other modes of post-translational modification (e.g., kinases), specificity is aided by amino acids neighboring the modified Cys that constitute a SNO motif^78,84 and by compartmentalization of enzymes together with substrates. Indeed, Hb is present within a super-complex that includes GAPDH, a well-characterized transnitrosylase.⁹²

ii. Termination of SNO signaling in RBCs

By analogy to kinase/phosphatase cycles, S-nitrosylation is terminated by denitrosylases that remove SNO from proteins and low molecular-weight thiols. Denitrosylases may also have a protective role in reducing cellular injury caused by nitrosative stress through non-directed S-nitrosylation of inappropriate targets. The first enzyme to be identified as a denitrosylase in a physiologic context was S-nitrosoglutathione reductase (GSNOR),^93,94 which specifically breaks down S-nitrosoglutathione (GSNO), a major small molecule conveyor of NO signaling. GSNOR activity regulates SNO-Hb levels through equilibria between GSNO and SNO-Hb that has been implicated in vasodilation⁹⁵ and tissue oxygen homeostasis.⁹⁶ Additional candidate denitrosylases present in RBCs include thioredoxins and protein disulfide isomerases, but their roles remain to be understood.

In sum, S-nitrosylation of proteins is in regulated equilibria, maintained through both enzymatic S-nitrosylation and denitrosylation pathways. Hemoglobin governs these equilibria in the blood and microcirculation. Notably, in pathologic states, denitrosylase activity may increase even as there is a need for more (not less) SNOs. For example, pulmonary GSNOR activity elevates in response to simulated high altitude exposure that has been intimately linked to SNO-Hb biology.^97,98

Physiological Reactions of NO with Hemoglobin: Making SNO-Hb

i. Reactions of NO/Hb in vitro

Hemoglobin would appear to be an ideal sensor governing oxygen delivery in the vasculature since it is the oxygen saturation of blood Hb, not P_O2, that determines blood flow (see Figure 1B). Similarly, NO would appear to be an ideal transducer of this vasodilatory activity. However, this role for NO initially presented a conundrum because the identification of NO in signaling was based on the ability of Hb to inactivate NO (see Text Box 1).^7,44,53,54 The challenge was thus to understand how Hb could deploy NO on the one hand and inactivate it on the other, and to meaningfully link these activities to physiology as the RBCs transit the circulatory system.¹¹ The answer would lie in the critical distinction between NO and SNO (the latter is bioactive in RBCs, the former is not⁸) and in the allosteric mechanism in Hb that facilitates redox conversion of NO into SNO (Text Box 2). The resolution of this problem was in the discoveries that Hb itself acts as an enzyme to promote its own S-nitrosylation and that generation and release of SNO from RBCs is linked to Hb oxygen saturation-driven allostery.^8,84

Reactions of NO with Hb are critically dependent on NO concentration and NO/Hb ratio; physiological reactions (NO<1μM; NO/Hb<1/100) will be summarized here briefly. A more detailed description can be found in Singel and Stamler.¹¹ As noted previously, Hb has 2 α and 2 β subunits, and it exists in two classic structures, R (relaxed, oxygenated) and T (tense, deoxygenated), that reflect the number of heme-bound oxygen ligands. Hemoglobin transitions from T to R as the third oxygen ligand binds, providing the basis for cooperativity. The β-subunits of Hb each possess one highly-reactive Cys93 residue, and its reactivity is linked to binding of oxygen (see Figure 2B). Physiological concentrations of NO bind cooperatively to βCys93 in oxygenated Hb (avoiding NO oxidation to nitrate; Text Box 2).⁹⁹ Further, physiological concentrations of NO will transfer from heme to Cys upon oxygenation.^100,101 NO can thus react with both heme and βCys93 in Hb, but reaction with βCys93 is favored in oxygenated Hb. Conversely, reactions of NO with heme-iron are favored over thiol in the deoxygenated state. Notably, reactions of NO with heme-iron in Hb also take place preferentially within the β-subunit under physiological conditions, since oxygen is released more rapidly from β- than α-chains, providing more β-chain vacancies; in addition, NO generated from nitrite or SNO-Hb preferentially forms βFeNO.^11,27 Thus, reactions of Hb with NO or nitrite show β-chain predominance in physiological situations, presaging formation of SNO.

Hemes in Hb may be vacant (deoxygenated), ligated (by O₂ or NO) or oxidized (Fe³⁺ or met). Most Hb proteins will have either 4 vacant (deoxygenated) or 4 fully-ligated (oxygenated) hemes, due to cooperativity. But a few will contain a mix of NO/O₂/Fe³⁺. These multivalent Hbs enable reaction of NO with βCys93, as Fe³⁺ supports the redox requirement to form NO⁺. Thus, multivalent Hbs may be viewed as specialized micro-populations of Hb that provide the main route to SNO-Hb.^11,27,86,87 Interestingly, these micro-populations decorate the RBC membrane, where SNO-Hb is found in high concentrations, in closest proximity to the blood vessel wall.⁸⁷

ii. Reactions of NO/Hb in vivo

The behavior of NO within multivalent Hb micro-populations has to be placed in the context of the respiratory cycle. Thus NO (itself or derived from nitrite or other SNOs) will bind to the vacant heme iron (Fe²⁺ or Fe³⁺) of deoxygenated Hb, mainly in the venous circulation, to generate HbFeNO.²⁷ In effect, RBCs transiting from tissues to the lungs will accumulate HbFeNO. This FeNO, which resides primarily in the β-chains, shows behavior of Fe³⁺NO – a redox-activated species in equilibrium with SNO-Hb.^11,101,102 Upon oxygenation in the lungs and Hb transition from T-state to R-state, the NO group is transferred from β-heme to βCys93 to generate SNO-Hb.^27,100,103 Conversely, the transition from high to low oxygen tension in the peripheral arterioles and capillaries (R to T transition) rotates this SNO from a protected pocket to a solvent-exposed orientation (see Figure 2B) that promotes its release as SNO-based vasodilatory activity (Figure 2C) along with the oxygen.^{8, 84, 103, 104} This can be observed as deoxygenation-induced SNO-Hb mediated vasodilation in vitro in direct proportion to Hb deoxygenation (Figure 2C) and SNO-Hb-dependent blood flow changes in brain regions, regulated by tissue oxygen level in situ (Figure 2D).⁸⁴ Remarkably, purified SNO-Hb transitioning from R- to T-state essentially recapitulates the hypoxic vasodilation observed by Guyton (see Figure 1B). Thus, circulating levels of SNO-Hb are dependent on Hb oxygen saturation (higher in artery and lower in vein).¹⁰⁵ Notably, only a small fraction of the NO carried by Hb is released from RBCs (i.e., transferred to acceptor thiols) during artery-to-vein transit; most is auto-captured by hemes within Hb β-subunits.^104,106 Thus, Hb functions as an enzyme that generates SNO (via auto-S-nitrosylation), as well as a propagator of SNO-based signals (through its transnitrosylase activity that exports SNO), thereby mediating a form of hypoxic vasodilation in the periphery to govern oxygenation-driven microvascular blood flow.

Coupling of RBC Vasodilation to Oxygen Concentration: Releasing SNO

i. Role of oxygen gradients

Blood flow in the microcirculation is regulated by physiological oxygen gradients, with RBCs causing blood vessels to dilate or constrict depending on the state of Hb oxygenation in response to local metabolic demand.^11,13,84,107 For example, in vitro tests of aortic rings show vasoconstriction by RBCs under high P_O2 but vasorelaxation in low P_O2 (Figure 3A). Just as for physiological hypoxic autoregulation (Figure 1B), RBC-mediated hypoxic vasorelaxation also demonstrates graded responses to oxygen levels from 100 to 0% (Figure 3B). SNO-Hb serves as a bioactive intermediate in this system,^{8,84,103,108,109} actuating graded vasodilation in proportion to degree of hypoxia (see Figure 2C,D).^84,105,110 Thus, the Guyton physiology is recapitulated by both purified preparations of SNO-Hb and native human RBCs (Figure 1B, Figure 2C, Figure 3B). Further, whereas in systemic arterioles, SNO-Hb will elicit increases in blood flow, SNO-Hb entering the lung (in T-state) may influence ventilation-perfusion matching.¹¹¹ RBCs thereby facilitate both the uptake and delivery of oxygen¹⁰⁵ as they manipulate both basal microvascular tone and the immediate responses to hypoxic demand.^112–114 The specific targets of RBC-derived SNO in vasculature and arteriolar walls that cause vasorelaxation remain to be identified, but include sGC.^115,116 Inhibition of sGC prevents human RBC-mediated hypoxic aortic (and pulmonary; see Figure 6 below) artery vasorelaxation, while inhibition of eNOS or denuding of the endothelium does not (Figure 3C). In a mouse model replacing mouse Hb with human Hb chains, mutating β-globin Cys93 so it can no longer carry or release SNO leads to reduced ability of isolated RBCs to effect hypoxic vasodilation (Figure 1C, Figure 3D).

Figure 3. RBC vasodilation through SNO-Hb.

A. Regulation of aortic ring tension after addition of native human RBCs (A-C) in high or low oxygen (Hb in R- vs T-state). RBCs pretreated with physiological amounts of NO (1μM) to produce ~500 nM SNO-Hb; representative traces (left) and quantification (right). RBCs recapitulate effects of native SNO-Hb (see Figure 2B). B. Graded relaxation by native untreated RBCs as a function of P_O2 ranging from ~10% O₂ (63 mmHg; HbO₂ ~90% saturated) to ~0.5% O₂ (3 mmHg; HbO₂ <10%) across the R-T transition. C. Hypoxic (~1% O₂; Hb T-state) vasodilation of aortic rings by native (untreated) human RBCs, after denuding (“D”) endothelial cells, after NOS inhibition by L-NAME, after cGC inhibition by ODQ, or in aorta from eNOS-knockout mice. Representative traces (left) and quantification (right). D. Hypoxic (~1% O₂) vasodilation of aortic rings by mouse RBCs bearing human wildtype α,β,γ globins (C93), or where human β-globin is mutated (C93A) and cannot bear SNO at this site. Note that mouse RBCs have a predominant ATP-dependent vasorelaxation pathway (compared to human RBCs) that contributes to the residual activity in endothelium-intact vessels, and these mice carry residual C93 in fetal Hb. E. Hypoxic (~1% O₂) vasodilation of aorta by native human RBCs is not altered by addition of micromolar nitrite, before, during or after addition of RBCs. Representative traces (left) and quantification (right). F. Model of P_O2-driven SNO flux in and out of RBCs. SNO-Hb is formed de novo under high P_O2. Under low P_O2 in tissues, Hb conformation change allows SNO-Cys93 to transfer SNO to the transmembrane transporter AE1 (Cys 317),⁸⁷ which shuttles SNO between coupled Cys residues to expose SNO on the outer membrane of the RBC, where it will equilibrate with SNO carriers in the blood and with SNO proteins in the vessel endothelial wall to elicit vessel dilation. Under high P_O2, the AE1–Hb SNO equilibrium is proposed to run in reverse to return SNO into the RBC and onto Hb. Note; whereas SNO-Cys93 and SNOCys317 have been identified in NO export, involvement of Cys843 is proposed based on exchange of selenium between Cys317/Cys843 in studies on selenium import from plasma-to-Cys317.¹⁵⁰ Modified panel A from Ref⁸⁷, panel B from Ref¹⁰³, panels C and E from Ref¹¹⁶, and panel D from Ref⁹, with permission.

Figure 6. Physiological and pathophysiological alterations in SNO-Hb.

A. Hb oxygen saturation (upper) and SNO-Hb levels (lower) measured in humans during staggered ascent to 5,000 meters and return to base altitude over 19 days. B. Levels of SNO-Hb and iron-nitrosyl Hb (HbFeNO), and thus total HbNO, are significantly reduced in patients with severe Pulmonary Arterial Hypertension (PAH), compared to age-matched controls. C. In PAH patients, NO/SNO levels inversely correlated with disease severity; specifically, higher pulmonary arterial pressures (PAP) were associated with lower RBC HbNO levels. D. Representative data showing that human RBCs dilate pulmonary artery in vitro in an NO-dependent manner (blocked by inhibition of sGC with ODQ) and that relaxations of aorta are impaired in RBCs from PAH patients. E. Group data demonstrating impaired ability of RBCs from PAH patients to produce hypoxic vasodilation in an in vitro aortic ring bioassay. F. SNO-Hb level is reduced in human sickle RBCs, and correlates with disease severity. G. Sickle RBCs are impaired by thiol oxidation of membrane transporter protein AE1 (Band 3), which is required for SNO export. H. Sickle RBCs from severe SCD patients are ineffective in hypoxic vasodilation in the in vitro aortic ring bioassay. I. The amount of NO bioactivity (ratio of SNO to total HbNO) is significantly less in blood from patients with Peripheral Artery Disease (PAD) compared to age-matched healthy controls. J. Representative tissue oxygenation (StO₂) tracings depicting the delayed hyperemic response in a PAD patient after a brief period of arterial occlusion. K. Mean (± SD) foot reperfusion times following thigh occlusion are significantly longer in the PAD cohort. (* p< 0.05). Panel A modified from Ref⁹⁸, panels B-E from Ref¹⁰⁹, panels F,G,H from Ref¹⁵², with permission; panels I,J,K, unpublished data.

ii. Other oxygen-dependent mediators

RBCs may dilate blood vessels by additional mechanisms, but physiological roles for these mediators in microvascular oxygen delivery are less clear.

a. ATP.

RBCs can liberate ATP at low P_O2 to increase endothelial/eNOS-derived NO through an ATP receptor-eNOS pathway (note: this phenomenon is much more prominent for mouse RBCs than human,^116,117 and likely accounts for the residual relaxation effect shown in Figure 3D). But studies have demonstrated the half-life of circulating ATP^114,118 far exceeds the requirements for autoregulation, and hypoxic vasodilation is preserved following pharmacologic and genetic inhibition of eNOS.^119–121 A reasonable interpretation of these data is that ATP most likely influences the set-point (basal tone) of vessels, whereas RBC-derived SNOs primarily effect flow autoregulation (moment-to-moment changes in local blood flow).^{11,113,114,122} In this regard, it is important to emphasize that human RBCs can also dilate vessels denuded of endothelium.¹¹⁶

b. Nitrite.

Nitrite has been touted as a mediator of tissue-derived vasoactivity in ischemic (and acidic) tissues,¹²³ and nitrite may also facilitate vasodilation under hypoxic conditions. Also, as described below, RBCs possess several proteins endowed with nitrite reductase activity, the ability to reduce nitrite to NO. These ideas are not in dispute, but attempts to connect nitrite to vascular physiology (i.e., physiological hypoxic vasodilation by blood, and autoregulation in particular) have been fraught with problems, both methodological and conceptual. Early studies in fact measured pharmacologic vessel relaxation by nitrite, not hypoxic vasodilation.¹²⁴ Nitrite has not been shown to emulate known physiology: it has not been shown to regulate blood flow governing tissue oxygenation in the microcirculation (we are unaware of any study) or to enhance the vasodilatory activity of Hb or of RBCs (see Figure 3E) (we are unaware of even one study) or even to dilate blood vessels in the presence of excess Hb (NO is sequestered and eliminated; Text Box 2).^105,116 A current model for nitrite’s potential role in hypoxic vasodilation was based on a misunderstanding of the putative importance of the “P₅₀” of Hb (the P_O2 at which Hb is 50% oxygen saturated, and at which nitrite is most effectively converted to iron-nitrosyl Hb).¹²⁵ However, hypoxic vasodilation is not linked to P₅₀¹²⁶ but rather increases as Hb oxygen saturation declines to zero.^11,13 Further, it is still not adequately appreciated that heme iron-nitrosyl Hb levels (generated by nitrite/Hb) reflect NO sequestration, not bioactivity (especially α-globin heme-bound NO). Nonetheless, nitrite can serve as an effective source of βFeNO and thus of SNO in Hb.^27,88,103 Also, nitrite has been shown to mediate an antiplatelet effect of RBCs in vitro¹²⁷ (questionable relevance to hypoxic vasodilation notwithstanding). Finally, effects of nitrite in the blood should not be conflated with vasodilatory effects of oral nitrite, which are mediated by SNOs formed by acidification in the gut, whereas circulating nitrite at physiological concentrations is essentially inactive.^{49, 128} Taken together, we suggest that nitrite may contribute to the reservoir of SNOs in blood, including SNO-Hb, and may serve as a general source of NO bioactivity under hypoxia—a ‘back up’ system. But currently, in the absence of genetic data or other stringent criteria to implicate nitrite in known physiology, the field of study is heavily reliant on pharmacology.

c. Xanthine oxidase (XO) and carbonic anhydrase (CA).

An accumulating literature has identified nitrite reductase activities with both XO and CA. These data are intriguing, but their physiological relevance remain to be determined. For example, pharmacological inhibition of XO in fact preserves blood flow (e.g.¹²⁹), improves vasodilator capacity¹³⁰ and lowers blood pressure,¹³¹ whereas one might anticipate the opposite from a nitrite reductase. Likewise, a role for CA in RBC mediated vasodilation in vivo has not been observed.¹³²

d. eNOS.

eNOS in RBCs may contribute to regulation of blood pressure¹³³ and RBC rheology.¹³⁴ Pharmacological activation of RBC eNOS has also been associated with cardioprotection in ex-vivo models,¹³⁵ but the mechanism is unknown. eNOS activity declines at low P_O2. Our bias is that the primary role of eNOS is in signaling within RBCs, as in other cells. It is not involved directly in RBC vasodilation under hypoxia.^11,116

iii. Unifying hypothesis

Hemoglobin, eNOS, and CA are located in close proximity at the RBC membrane; Hb and CA are both associated with band 3/AE1, which is required for SNO export (as described below). CA has been implicated in enzymatic activity (nitrous anhydride activity) that generates SNO precursors from nitrite¹³⁶ and may also be involved in nitrite import/export. eNOS may serve as regulated source of NO/nitrite at the membrane. Thus, the machinery for generating SNOs in general and SNO-Hb in particular is located within a signaling complex at the RBC membrane. While this machinery may have multiple roles (and not all elements may be required for all roles), it is tantalizing to speculate that these components may work in concert to enable RBCs to dilate blood vessels and generate NO bioactivity.

Red Blood Cell SNO Bioactivity

i. SNO-Hb as mediator of hypoxic vasodilation

Isolated Hb can be S-nitrosylated selectively at βCys93 using the donor Cys-NO (pH 7.8, borate, etc).⁸ SNO-Hb constricts blood vessels at high P_O2 but dilates under hypoxia, illustrating allosteric regulation by O₂ (Figure 2C). In contrast to earlier expectations about heme destruction of NO activity, NO can evade heme in the form of SNO on Hb.^8,84 However, excessive SNO-loading of Hb leads to modification of sites other than βCys93 and thus loss of allosteric regulation: vasodilation becomes independent of P_O2. Similarly, RBC-mediated vasodilation was first described with artificially SNO-loaded RBCs.⁸ But SNO-loading of RBCs with CysNO is not selective for βCys93, and the higher the loading (millimolar in some studies) the more CysNO modifies alternative targets (other Hb thiols, cellular glutathione and membranes), leading to P_O2-independent vasodilation by aberrant RBC-donors. To overcome this problem, we developed physiological loading techniques: at physiological levels of NO (~1 μM), Hb in RBCs is preferentially SNO-modified at βCys93 (Figure 3A).⁸⁴ Notably, RBCs or native Hb modified selectively with SNO at βCys93 elicit vasodilation in proportion to the level of hypoxia (Figure 3B),⁸⁷ and blood flow in multiple brain regions is increased by hypoxia, but reduced by hyperoxia (see Figure 2D).⁸⁴ Native RBCs with endogenous SNO can elicit vasodilation.⁸⁷ Further, half of the vasodilatory activity was in RBC membranes associated with the Band 3 protein at the RBC surface, and Band 3 inhibitors block RBC vasodilation in vitro and in vivo (Figure 3F).⁸⁷ SNO-replete native RBCs produce a graded vasodilatory response to hypoxia (Figure 3A,B), replicating the pattern of hypoxic autoregulation observed by Guyton (see Figure 1B).¹⁰³ RBCs depleted of SNO show impaired blood flow,^116,137 most clearly demonstrated in the basal blood flow deficits in the human βCys93A mutant knock-in mouse.^9,138, as well as the failure of the βCys93A mouse to increase flow in response to systemic hypoxia^9,138, or following vessel occlusion (see Figure 1C).

ii. Physiological context of SNO-Hb vs alternative mediators.

There has been remarkable controversy about the role of RBCs as mediators of vasodilation. In part this stems from conflation of multiple mechanisms for blood flow regulation acting at multiple levels of the vasculature and operating on distinct time-scales. In our work (and in this review), we have focused on the mechanism of physiological hypoxic autoregulation governing blood flow as a direct function of O₂ saturation, as initially demonstrated by Guyton. Auto-regulation of blood flow is the only vascular physiology we are aware of that is directly mediated by RBCs. Other vascular effectors do not appear to explain this physiological effect. One example is the observation that RBCs release ATP under hypoxia to elicit vasodilation.^{117,139–141} Endothelial cells do as well; thus, the mechanism is not dependent on Hb O₂ saturation (as per autoregulation; see Figure 1B). ATP-mediated vasodilation requires an intact endothelium as well as eNOS activity, since ATP acts through endothelial GPCR-stimulated NO generation (whereas RBC SNO-Hb does not).^139–141 However, hypoxic regulation of blood flow governing tissue oxygenation is unaffected by eNOS inhibition in humans.^11,27,86,87 ATP release is more relevant for rodent than human RBCs^9,116 and cannot operate in microvessels, which are lacking eNOS.

Similarly, it has been claimed that nitrite in plasma or in RBCs mediates hypoxic vasodilation.^124,142 There is no doubt that nitrite can alter vascular tone and blood flow in general. The questions are, under what circumstances does this occur physiologically, and through what mechanisms? Nitrite can utilize Hb to generate Fe-NO, and although this occurs under hypoxic conditions, this activity is maximal at oxygen P₅₀ (P_O2 ~ 26mmHg) rather than increasing further as O₂ declines.^143,144 Furthermore, it is not clear that the NO actually escapes the heme iron (Fe-NO) to elicit vasodilation, as isolated vessels were reported to exhibit a shift in P_O2 threshold for initiating hypoxic vasodilation by RBCs with or without nitrite, but did not exhibit enhanced vasodilation.¹²⁴ A 5-fold exercise-induced blood flow increase in human limb (in the presence of eNOS inhibitor) does not reduce plasma nitrite in either venous or arterial blood.¹⁴⁵ Also, eNOS deletion lowers nitrite but does not affect autoregulation of blood flow^11,27,86,87 (by contrast with SNO-Hb depletion⁹). It is therefore unclear how nitrite levels could regulate hypoxic vasodilation in the graded manner observed physiologically.

Published examples of vasodilation by nitrite in vitro and in animal models always utilize supra-physiological levels; no studies report endogenous levels (~100nM) (e.g., Refs^124,146). The pharmacological effects of nitrite are not enhanced by RBCs or by Hb (Figure 3E),¹¹⁶ and RBCs block nitrite mediated relaxations unless nitrite>>Hb at very non-physiological levels. Nitrite has been reported to be processed into SNO by RBCs/Hb, but the time-frame for this reaction is too slow to account for hypoxic vasodilation during arterial-venous transit.²⁷ Overall, it is seems unlikely that nitrite mediates directly a significant part of physiological vasodilation under hypoxia by RBCs. Moreover, in the absence of defined pathways targetable with inhibitors or genetic modifications, nitrite-based models generally have not been amenable to direct and stringent testing against defined criteria. (note: xanthine oxidase was proposed as one mediator of nitrite bioactivation to NO,^147,148 but testing using inhibitors did not support a physiological role;¹⁴⁹ see also above). Nitrite is more likely a back-up system (and contributor to SNO bioactivity) than mediator of any known physiological response.

iii. Role of βCys93 in hypoxic vasodilation

The oxygen-dependent structure-activity relationship of SNO-Hb βCys93 first was reported twenty years ago.⁸⁴ When Hb is deoxygenated, this SNO is exposed to solvent and reactivity is high, a result of Cys93 being forced to the protein surface beyond the His146-Asp94 salt bridge. When Hb is oxygenated, this salt bridge is broken and Cys93 points inward away from solvent, protecting the SNO group. Thus, SNO-βCys93 assumes positions in deoxygenated (T) and oxygenated (R) Hb that dictate its reactivity: Cys93-SNO can act as an SNO donor in the T conformation to mediate hypoxic vasodilation, while being protected from release in the R conformation. The degree of deoxygenation of Hb directly correlates with the degree of vasodilation driven by SNO released from RBCs (see Figure 1B and Figure 3B).^11,13 This represents the molecular mechanism of autoregulation of microcirculatory blood flow, as determined by strict genetic criteria in Cys93 mutant animals, providing for second-by-second regulation of oxygen delivery.⁹ No alternative function for the essential βCys93 in Hb, whose reactivity is linked allosterically to oxygen binding, has been shown, and no alternative mechanism to recapitulate the physiology of autoregulation, whose role is essential for oxygen delivery in tissues, has been proposed.

Hemoglobin is both cytosolic and associated with membrane proteins; this latter population appears responsible for the export of NO bioactivity from RBCs.⁸⁷ Under hypoxia, membrane SNO-βCys93-Hb transfers NO to Cys residues within the N-terminal cytosolic domain of band 3/AE1;⁸⁷ the N-terminal Cys in AE1 is in communication with a Cys within the intra-membrane domain-spanning region of AE1, which in turn communicates with plasma.¹⁵⁰ Protein disulfide isomerase at the plasma membrane⁸⁹ and glutathione/thiols in plasma⁹⁵ also appear to participate in SNO shuttling. Thus, RBC SNO distribution is governed by a set of P_O2-coupled equilibria (Figure 3F), with delivery of NO bioactivity via sequential transnitrosylation reactions that represent a shift across these coupled equilibria – in effect, a “bucket brigade”. In this model there is bidirectional movement of NO/SNO bioactivity as NO shuttles through exofacial, membrane cytoskeletal, and cytosolic SNO species. Furthermore, there is evidence to suggest that some conditions (e.g., sickle cell, blood storage) can negatively impact components of NO export, viz. cleavage of the N-terminal domain of AE1¹⁵¹ and RBC membrane oxidation,¹⁵² which provide mechanistic contributions for defects in RBC-mediated vasodilation.

There is increasing evidence that the “bucket brigade” model of SNO flux may well be a general mechanism of transport and utilization of stored SNO. Such findings challenge the accepted wisdom that NO/SNO reactions are stochastic but in accord with SNO placement and transfer being specifically targeted through enzymatic means. The enzymatic machinery for S-nitrosylation comprising three enzymes – NO synthases, SNO synthases, and transnitrosylases – can be found within the RBC linked with Hb.^79,153 In particular, eNOS is located at the membrane of RBCs^47,133,154 in immediate proximity to membrane Hb, which converts NO into SNO. SNO-Hb can then transfer NO to AE1 with which it directly interacts. Thus, Hb can function both as a SNO synthase and transnitrosylase. To be clear, eNOS may provide a source of NO/nitrite that yields SNO-Hb, but SNO-Hb is plentiful so that RBC vasodilation is not acutely dependent on eNOS (Figure 3C).

iv. Arterial-venous SNO gradients: blood gases

Deoxygenation-dependent SNO release from Hb and RBCs during transit of tissue microcirculation generates a gradient of SNO-Hb, with higher levels in arteries and lower levels in veins (Figure 4A).¹⁰³ While some have contested this finding, samples in non-supportive studies were exposed to room air and/or high oxygen pressures in freeze/thaw that would eliminate the O₂/NO gradients,^145,155 and the finding has significant support both in vitro and in vivo. In isolated Hb, SNO-Hb is quite stable in the oxygenated state but decays rapidly in the deoxygenated state (and decay is accelerated by an external SNO acceptor such as glutathione) (Figure 4B).⁸ Within native RBCs, SNO decay exhibits a characteristic linear relationship to Hb O₂-saturation (also accelerated by glutathione) (Figure 4C)¹⁰⁵ that appears to underlie the linear increase in blood flow observed by Guyton in hypoxia (see Figure 1B). That is, in human blood gases in vitro, Hb-SNO is directly correlated with Hb-O₂, consistent with arterial-venous gradients in vivo that subserve autoregulation of blood flow. Moreover, venous but not arterial blood in situ extrudes bona fide SNO, captured on extracellular glutathione and identified by mass spectrometry (Figure 4D).¹⁵⁶ Mice bearing RBCs with human Hb unable to bind SNO at Cys93 (discussed in detail below) demonstrate a loss in glutathione-augmented RBC-mediated vasorelaxation of blood vessels in low oxygen (Figure 4E),⁹ consistent with export of vasorelaxant SNO activity from RBCs as they encounter tissue hypoxia.

Figure 4. Human arterial-venous gradient in RBC SNO-Hb results from deoxyHb release of SNO that can be acquired by glutathione (extracellular thiol acceptor).

A. Arterial and venous blood O₂ and SNO-Hb values from healthy humans. B. SNO-Hb (purified preparation; 2SNO-Hb[FeO₂]; metHb <5%) levels decline upon deoxygenation, and SNO loss is accelerated markedly by addition of glutathione (GSH) as SNO acceptor. C. Linear relationship between Hb oxygen saturation and SNO-Hb (expressed as ln(SNO-Hb/Hb) in native human RBCs: SNO-Hb declines in RBCs with oxygen desaturation, effectively creating the gradient between artery and vein (see Figure 4A and 4D). Addition of GSH accelerates loss of Hb-SNO from RBCs reflecting equilibrium between SNO inside and outside RBCs. D. Export of SNO by native RBCs in vivo. GSH incubated with whole venous blood (blue, deoxy) acquires offloaded SNO (GSNO detected by LC/MS), while oxygenated (arterial) blood does not (red). Ultraviolet light-treatment to remove SNO from venous blood, control (purple). E. Aortic ring relaxation by native RBCs isolated from wildtype C93 mice is augmented by extracellular GSH, but augmentation is lost using RBCs from βC93A mutant mice. Modified panel A from Ref¹⁰³, panel B from Ref⁸, panel C from Ref¹⁰⁵, panel D from Ref¹⁵⁶, panel E from Ref⁹, with permission.

v. Rheological effects of SNO

Rheology is another function impacted by RBC NO bioactivity,¹⁵⁷ where either excess or deficiency in SNO reduces RBC deformability to impair microvessel passage. Incubation of human RBCs with NOS inhibitors results in a rapid decline in deformability required for effective RBC passage in tissues (see Figure 1A). This defect was reversed by administration of exogenous NO donors or by the addition of excess L-Arg to the incubation mixture.¹⁵⁸ (Note that these data point to an important function of eNOS within RBCs under normoxia.) Similar results have been observed in vivo: either NO scavenger or NOS inhibitor decreased RBC velocity by decreasing deformability even as vessel diameters remained constant.¹⁵⁹ Reduced RBC deformability has been identified in human pathologies including hypertension,¹⁵⁷ type II diabetes,¹⁶⁰ and sepsis.¹⁶¹ S-Nitrosylation of RBC cytoskeletal proteins, notably α- and β-spectrin, is viewed as a key component in maintaining deformability,¹⁶² so losses in membrane SNOs may be additive to losses in SNO-Hb with respect to impairing export of NO bioactivity and thus oxygen delivery.

Essential Physiology of RBC SNO-Hb (βCys93).

i. Integrated response to hypoxia

Strict conservation of βCys93 in mammalian Hb and (thermodynamic) linkage of its function to Hb deoxygenation (allosteric regulation by P_O2) play out in a remarkable set of physiological actions revealing the critical role of βCys93 SNO-based vasodilatory activity in the integrated response to hypoxia (Figure 5). In essence, the defining features of autoregulation as described by Guyton,¹³ as well as pulmonary and central (nucleus tractus solitarius) responses to hypoxia are disrupted in genetically-modified mice that express human Hb (α, β, and fetal γ chains) in which Hb βCys93 was replaced with an Ala residue that cannot accept SNO. Moreover, the most fundamental aspects of cardiovascular and respiratory function in oxygen delivery (Text Box 1) are dependent on βCys93 and its role hypoxic autoregulation of blood flow.

Figure 5. Impact of βCys93 on cardiovascular and respiratory function.

A. Representative blood flow tracings during graded hypoxia in a γβC93 (wildtype human Hb) and a γβC93A (Cys mutant Hb refractory to S-nitrosylation) mouse. B. Muscle blood flow and C. P_O2 responses to hypoxia are significantly compromised, with baseline blood flow and P_O2 lower in the γβC93A and βC93A mutant mice. D. Representative ECG recordings in a C93 (control) and a C93A mouse at FiO₂ of 0.21 and 0.05 with T- and ST-waves marked, indicating myocardial ischemia at baseline in the C93A mouse and myocardial infarction/injury with brief (5 min) hypoxic challenge. E. During transient progressive hypoxia, ST-wave elevation (and hyperacute T-waves) indicative of acute ischemic injury are significantly greater and far more frequent in βC93A and γβC93A mice vs. γβC93 control mice. F. At FiO₂ = 0.21 (room air), T-wave amplitude is significantly reduced in βC93A and γβC93A mice vs. γβC93 mice, indicative of ischemia under basal conditions. G. Post-ischemia necrotic area (white) in heart after induced myocardial infarction is greater in βC93A mice (right) compared to wildtype βC93 mice (left). H. Cardiac output after transaortic constriction pressure overload (TAC) is depressed in βC93A and γβC93A mice, indicative of heart failure due to microvascular impairment. I. Survival during chronic TAC pressure overload heart failure is reduced in βC93A and γβC93A mice. J. Breathing frequency fails to increase in βC93A mutant mice after return to normoxia following hypoxic challenge indicating central respiratory impairment. K. Lung tidal volume fails to increase in βC93A mutant mice after return to normoxia following hypoxic challenge. L. Minute ventilation fails to increase in βC93A mutant mice after return to normoxia following hypoxic challenge. Modified panels A-F from Ref⁹, panels G-I from Ref¹³⁸, panels J-L from Ref¹⁶³, with permission.

ii. Essential role in cardiovascular and respiratory function

Most strikingly, reactive hyperemia after transient interruption of blood flow through muscle is significantly decreased in βCys93A mice (Figure 1C), demonstrating an in vivo role for Hb-SNO in this classic vasodilatory process that restores tissue oxygenation following hypoxia. These data are supported by mechanistic studies. Using an in vitro aortic ring relaxation assay, RBC-induced vascular relaxation at low P_O2 was impaired in RBCs from βCys93A mice (Figure 3D, Figure 4E), and export of NO bioactivity was significantly reduced.⁹ Consistent with these findings, basal skeletal muscle blood flow and P_O2 (Figure 5A–C) were diminished in the absence of βCys93. Further, mutant mice were unable to increase blood flow or normalize tissue oxygenation in response to progressive hypoxia, despite normal cardiac function, which represents the signature feature of aberrant hypoxic vasodilation in vivo.⁹ Strikingly, βCys93A mutant mice (Figure 5D–F) showed ECG signatures of ischemia under normoxia (i.e., decreases in T wave amplitude that reflect diminished coronary blood flow),⁹ indicating chronically inadequate perfusion in the absence of SNO-Hb. Furthermore, ST elevations (indicative of acute myocardial infarction) were recorded following brief hypoxic challenge. Mortality after transient exposure to hypoxia was much greater in βC93A mice than in control mice. Litter sizes were markedly reduced in βC93A mutants,^9,146 but these mutant mice retain Cys93-SNO in fetal γ-globin chains expressed in adults and also exhibit compensatory blood vessel growth due to chronic hypoxia,⁹ likely explaining why any survive at all. Additionally, βC93A substitution was shown to increase infarct size and to predispose mice to heart failure and death (Figure 5G–I).¹³⁸ Mice bearing βC93A were tested for hypoxia-induced facilitation of breathing, where resumption of normal P_O2 after brief hypoxia leads to increased ventilation rate.¹⁶³ Mice lacking hemoglobin S-nitrosylation were unable to increase breathing frequency or duration upon return to normoxia (Figure 5J–L). Altogether, these findings establish that mice lacking SNO-βCys93-Hb are fundamentally impaired in hypoxic responses despite carrying normal amounts of oxygen (βCys93 mutation has minimal effects on Hb P₅₀) (Text Box 1), and that SNO-βCys93 is essential for tissue oxygenation by RBCs, for regulation of microvascular blood flow, and for normal cardiovascular and respiratory function. Taken together with work by Palmer showing that low-molecular-weight SNOs derived from RBC SNO-Hb transduce hypoxia-mimetic responses in the lung,⁹⁷ it is clear that βCys93 serves a critical role in the integration of cardiovascular and respiratory systems to hypoxia, including the central ventilatory drive in the brain and pulmonary arterial pressure in the lung, as well as peripheral blood flow and tissue oxygenation.

Methodological Controversies: SNO Quantification in Hb Micropopulations

Prior to development of the biotin-switch technique in 2001¹⁶⁴, only a few labs were measuring endogenous SNO-proteins, using cumbersome tools. The biotin-switch greatly simplified the measurement of protein S-nitrosylation. Various improvements to the original assay (such as by collecting SNO proteins directly on beads in SNO-Resin Assisted Capture, SNO-RAC¹⁶⁵) have ultimately allowed for the identification of thousands of proteins whose activity is regulated by the addition or removal of SNO from specific Cys residues. For example, we have used SNO-RAC to show that SNO-Cys can be found endogenously not only in the β-chains of various Hbs, but also in the α- (note: Hb’s have 6 Cys)⁹ suggesting P_O2-independent roles; however, switch methodology does not easily quantify absolute amounts of SNO and cannot detect iron-nitrosyl.

Several quantitative methods of measuring SNOs in blood have been developed, but all have limitations. In particular, it is critical to maintain P_O2 during all manipulations to minimize loss or movement of NO within the hemoglobin tetramer prior to measurement.¹⁰³ To measure SNO levels in blood Hb, RBCs typically are isolated and lysed, and Hb is purified using column chromatography. Subsequent liberation of the covalently-bound NO is accomplished by chemical or photolytic cleavage; the free NO is then usually quantified by chemiluminescence.¹⁶⁶ Photolytic cleavage of NO does not depend on reagents and conditions.^8,167 By contrast, chemical methods used to selectively release specific NO forms from Hb produce different and varying results. Unfortunately, these chemical methods were not validated against physiological Hb standards (Hb micro-populations),¹⁶⁸ creating confusion and misunderstanding of how much NO is bound and the nature of species being quantified.

The tri-iodide method¹⁶⁹ developed to release NO from nitrite and SNOs in isolated systems, is case in point. Because of its simplicity, the assay was adopted for in vivo study by many groups, with a succession of modifications purported to allow for selective release or elimination of specific NO species, including the addition of multiple reagents to “stabilize” or “release” SNO (e.g., potassium ferricyanide, sulfanilamide, N-ethylmaleimide, potassium cyanide), solution alkalinization, and endothermic reaction conditions.^170–174 However, this methodology does not preserve the oxygen tension in blood gases – essential for accurate measurement of SNO-Hb – and artifactually alters the molecular distribution of NO/SNO/FeNO prior to chemiluminescence quantification.¹⁶⁸ It was further shown that the method could not detect key SNO and iron-nitrosyl species, including Hb micro-populations that form SNO, nor could it make accurate SNO/FeNO assignments in Hb.¹⁶⁸ Unsurprisingly, RBC SNO-Hb levels were found using this assay to be widely divergent from results using other types of assays.¹⁷⁰ This has provided a basis for some to reject a role for SNO-Hb in hypoxic vasodilation.^175,176 In contrast, methods designed to maintain a physiologic environment and avoid addition of multiple chemicals have revealed the oxygen-dependence of SNO-Hb levels and identified altered RBC SNO-Hb levels in human diseases (see below).

The 3C method (so-named because it employs carbon monoxide and cuprous chloride-saturated Cys) requires minimal sample preparation, and is noted for its ability to measure SNO-Hb in intact RBCs. SNOs are selectively converted to free NO upon addition of RBCs to the cuprous chloride solution while carbon monoxide serves to prevent autocapture of the released NO by heme.¹⁰⁵ A strength of the assay is that sample preparation simply involves washing isolated RBCs prior to assay, so physiological P_O2 in samples can be readily preserved. The weakness is that the output measure is total RBC SNO level, not specific measures of SNO-Hb, HbFeNO, or total HbNO. Also, carbon monoxide can prevent auto-capture of NO by ferrous (Fe²⁺) but not ferric (Fe³⁺) heme, and a major population of NO in Hb is found within SNO/Fe³⁺-NO^11,102,106 Nonetheless the 3C method has effectively and reproducibly identified oxygen-sensitive arterial-venous gradients in RBC-SNO (Figure 4C),¹⁰⁵ as well as altered RBC SNO levels in septic patients, in patients with acute respiratory distress syndrome and in test subjects at high altitude.⁹⁸

Our preferred method for SNO-Hb quantification is photolysis-chemiluminescence.^{103,115,177,178} Its limitation is the requirement to isolate Hb while maintaining the relevant physiological P_O2 of the source blood throughout the preparation. Concentrations of SNO-Hb and HbFeNO are readily calculated.^87,109 This method has been well validated for selective determination of RBC HbNO micropopulations, without artifactual contamination from any other species, including nitrite and nitrate.^168,179 Unfortunately, this technique requires sophisticated equipment, experienced operators, and extensive sample preparation. Arterial and venous blood gas values along with NO levels determined using photochemiluminescence from a cohort of healthy individuals are presented in Figure 4A, documenting the typical oxygen-dependent arterial-venous gradient in NO levels determined with this technique.

Pharmacological Controversies: SNO-loading of RBCs

Study of isolated RBCs treated with NO is complicated by the need to maintain the proper disposition and redox state of the NO-Hb within. Replacing SNO lost from Hb-βCys93 is a major source of variability among research teams. Under physiological conditions (<1 μM NO; NO/Hb <1:100), Hb will load SNO primarily on βCys93,⁸⁷ and thus maintain hypoxic regulation, but Hb has several additional cysteine residues.^4,28 Despite a well worked-out protocol using limiting NO and a cycle of Hb deoxygenation/reoxygenation for placing SNO on βCys93, allowing assay under near-physiological conditions,⁸⁷ many groups continue to use a more-is-better approach that massively overloads RBCs with SNO prior to activity assays. While physiologically-loaded SNO-RBCs (Figure 3A) exhibit graded release of SNO with oxygen decline (Figure 3B), overloading results in SNO binding at multiple inappropriate sites in Hb and other RBC proteins and on glutathione, and in the prominent production of metHb. Studies using SNO-overloaded RBCs are unfortunately common (one recent example¹⁸⁰), and it appears unlikely that such “donor” assays relate to RBC physiology. Most importantly, RBCs pharmacologically overloaded with SNO lose the oxygen-dependence of SNO release, since only βCys93 is linked to oxygen-dependent allosteric transitions in Hb.¹¹ Instead, they export the donors used to load the RBCs as well as GSNO using high K_m (non-physiological) transporters.

Physiological Controversies: Physiology vs “physiological response”, and genetic criteria

The complexity of Hb/NO reactions and of microvascular physiology has led to controversy. NO/SNO reactions with Hb are dependent on NO concentration and NO/Hb ratio.¹¹ Major disagreements have resulted between groups that use physiological (micromolar or less) versus high pharmacological (typically millimolar) amounts of NO/NO donors to test the role of RBCs and native SNO-Hb in hypoxic vasodilation. Misunderstanding has also arisen over what constitutes “physiology” in order to differentiate the role of SNO-Hb versus other vasodilators acting at multiple levels within the vasculature. Some used platelets as a surrogate model for blood vessels. Many have not appreciated that vasodilation by Hb had a critical context in the physiology of autoregulation, which is often conflated in the literature with vasodilation under systemic hypoxia (“hypoxic vasodilation”). Autoregulation is one form of hypoxic vasodilation operating in the microcirculation to optimize tissue oxygen delivery. Its key features are: linear-inverse relationship to O₂ saturation (not P_O2); coupling to arterial-venous gradients; optimization of tissue oxygen extraction. Autoregulation is a physiological flow response linked to tissue oxygenation, whereas many alternative forms of vasodilation may be enhanced under hypoxia, but will not optimize tissue oxygenation due to shunting. Postulated mediators of hypoxic vasodilation are compared in Table I relative to these criteria for fulfilling hypoxic autoregulation.

Table I.

Proposed Mediators of Microvascular Autoregulation

Criteria for Autoregulation	RBCs	SNO-Hb	Nitrite	ATP
Increase microvessel blood flow	Yes	Yes	Unknown	Yes
Δ flow proportional to Hb O2 saturation	Yes	Yes	No	Unknown
Erythrocyte replicated	Yes	Yes	No	Yes
Endothelium-independent action (denude)	Yes	Yes	Yes	No
NOS-independent action (acute inhibition)	Yes	Yes	Yes	No
Localized action coupled to O2 saturation	Yes	Yes	No	No
Rapidly altered by local tissue O2 saturation	Yes	Yes	No	Unknown
Replicated in vitro	Yes	Yes	No	No
Replicated by in vivo infusion	Yes	Yes	No *	No
Impaired by inhibition/depletion	Yes	Yes	No	No

^*While infused nitrite increases blood flow, it does not increase tissue O₂ extraction (i.e., venous O₂ increases not decreases) indicative of increased shunted flow not increased hypoxia-governed flow.⁸⁴

Controversy has arisen about whether Hb and RBCs are able to release SNO to augment microcirculatory blood flow. Much data from our group and others demonstrate that isolated RBCs with native SNO-Hb or loaded to contain physiological SNO-Hb maintain SNO in 20% oxygen but lose SNO when removed from oxygen, and this loss is greatly augmented by addition of glutathione to the incubation buffer (Figure 4B; Figure 4E). These data strongly support a model where binding of NO to βCys93 is thermodynamically linked to binding of O₂ to heme.¹⁰ A similar experiment using blood directly demonstrated that this loss of native SNO-Hb occurs proportionate to the oxygen level, and this loss is also increased by addition of glutathione (Figure 4C, compare to Figure 1B). Rapid isolation of the plasma from such a reaction and immediate LC/MS identifies the production of authentic SNO-glutathione (Figure 4D). Our interpretation of these and other experiments is that Hb acts as a SNO synthase, linking oxygenation/deoxygenation transitions to SNO formation from NO and to SNO release from the βCys93 residue (with transfer to acceptor glutathione); it has been further shown that thiols in RBC membrane participate in the transfer of SNO from βCys93 to extracellular glutathione.⁸⁷

On the other hand, others have contended at various times that NO cannot react with thiols under physiological conditions,¹⁷⁰ that Hb does not become S-nitrosylated under physiological conditions,¹⁷⁰ that Hb does become S-nitrosylated under physiological conditions in rodents but not in primates,¹⁸¹ that SNO-Hb does form but does not dispense SNO from RBCs,¹⁸² that SNO-Hb does form and does dispense vasodilatory activity, but this is not increased under hypoxia,¹⁸³ that SNO-Hb does dispense vasodilatory activity from RBCs but only in sepsis,¹⁸⁴ that vasodilatory activity can be released from native RBCs but not via NO/SNO,^139–141 that non-NO/SNO vasodilatory activity can be released from native RBCs in response to NO/SNO,¹⁸⁵ that NO/SNO can be in fact released from RBCs but not from SNO-Hb,¹⁸⁰ and that RBC vasodilatory activity is instead due to nitrite.^124,186 Most recently, nitrite is reported to act on RBCs to dilate blood vessels through SNO-based mechanisms (but specifically excluding SNO-Hb).^127,187,188 Given these disagreements, based primarily on distinct methods and the questionable relevance of certain models (e.g. platelet aggregation, blood pressure, nitrite/NO pharmacology) to microvascular physiology^146,189,190 it was hoped that physiological investigations of the relationship of hypoxia to proposed functions of βCys93 would lead to consensus. Unfortunately, this optimism has not yet proven true.

A particularly informative example stems from the development of genetically-modified mice that express human Hb (α, β, and fetal γ chains) in which Hb βCys93 was replaced with an Ala residue that cannot accept SNO. Isbell et al¹⁴⁶ reported that these animals were normal. While their assessment is correct for the functions they measured, they assessed parameters with no bearing on microvascular blood flow or potential RBC effects on it (e.g., blood pressure, lung histology) and, remarkably, they did not measure tissue P_O2 or blood flow, or induce hypoxemia. They concluded “…SNO-Hb is not essential for the physiologic coupling of erythrocyte deoxygenation with increased NO bioactivity in vivo”.¹⁴⁶ The paper received significant attention and citations. It also presents a cautionary tale on how hypoxic vasodilation governing tissue oxygenation should be studied: the delivery of oxygen at the tissue level is a micro-hemodynamic flow parameter, the extent of which cannot be quantified by measuring macro-hemodynamic parameters such as systemic and pulmonary arterial blood pressures.^189–191 A recent analysis of RBCs from these mice also concluded that NO-mediated effects of RBCs are not mediated by nitrosylation of βCys93¹⁸⁰, but the measures assessed (platelet aggregation, recovery in cardiac pressure following ischemia/reperfusion in an isolated heart model) are not microvascular or even related to blood vessels, and the methods used for studying isolated RBCs and Hb are entirely pharmacological (massive NO loading) rather than physiological. When appropriate indices have been studied in these mutant mice, including microvascular blood flow and tissue oxygenation, dramatic impairments were identified:^9,138,163 hypoxic vasodilation resulting after transient interruption of blood flow through limbs is significantly decreased in βCys93A mice (Figure 1C),⁹ RBC-induced hypoxic vascular relaxation is impaired (Figure 3D, Figure 4E),⁹ export of NO bioactivity is significantly reduced,⁹ both skeletal muscle blood flow and P_O2 (Figure 5A–C) is diminished at baseline and in response to progressive hypoxia (Figure 5D–F),⁹ myocardial ischemia is evident at baseline¹³⁸ and βC93A mice show markedly increased mortality (Figure 5G–I).¹³⁸ Additionally, mice bearing βC93A are unable to increase breathing frequency or duration upon return to normoxia (Figure 5J–L).¹⁶³ Altogether, these findings establish that mice lacking SNO-βCys93-Hb are far from normal, and that SNO-βCys93 is essential for tissue oxygenation by RBCs, for regulation of microvascular blood flow, and for normal cardiovascular and respiratory function. Further evidence for the physiological role of SNO-Hb in regulating tissue oxygenation comes from studies in patients in various conditions and disease states (see Part 3, below).

Part 3. S-NITROSOTHIOLS AND PATHOLOGIES OF OXYGENATION

SNOs in Hypoxia

Studies at high altitude with reduced oxygen availability have been particularly useful in understanding the human SNO-based response and its key role in adaptation to hypoxia. Two distinct evolutionary paths have allowed Tibetans and Andeans to adapt to the 40–50% reduction in oxygen partial pressure at 4000–5000 meters above sea level.¹⁹² High altitude Andeans have compensated by increasing blood oxygen content through polycythemia: RBC Hb content is ~3–4 g/dl higher than at sea level, driven through elevated erythropoietin and erythropoiesis. This comes at a significant physiologic cost of pulmonary hypertension, increased cardiovascular risk, and poor pregnancy outcomes. In contrast, Tibetans accommodate by increasing the efficiency of oxygen delivery and do not typically succumb to the pathophysiologic events common in their Andean counterparts. These accommodations start in the lungs with higher resting ventilation rate (but normal pulmonary pressures) and end in the microvasculature with higher capillary density and blood flow. Tibetans have only modest elevations in Hb levels, and thus low blood oxygen content, but marked elevations in RBC-SNO to enhance RBC-mediated microvascular blood flow and end-organ oxygen delivery.¹⁹³ Interestingly, on the 4000-meter Ethiopian plateau, distinct groups utilize each of these two adaptation pathways.¹⁹⁴

Acute adaptation to altitude occurs over a period of days-weeks through a combination of Tibetan and Andean pathways. In field studies of acclimating mountain climbers, staggered ascent produces stepwise increases in Hb/hematocrit and SNO-Hb even as arterial oxygen saturation declines (Figure 6A). The rise in total Hb and SNO-Hb directly correlates with exercise performance at altitude, but these adaptations fade rapidly upon return to sea level.⁹⁸ In a hypobaric chamber, rapid “ascent” (over minutes) of healthy subjects to 5,000 meters raises pulmonary pressure, increases cardiac output, and produces a significant decline in systemic vascular resistance as, presumably, RBCs effect SNO-driven hypoxic vasodilation in the periphery in an attempt to maintain tissue oxygenation. SNO utilization for microvascular vasodilation results in a rapid depletion in SNO-Hb since deoxy-Hb favors SNO off-loading.¹⁰³ An inability to adapt to high altitude, notably by not increasing NO bioavailability, can lead to serious altitude sickness, and associated pulmonary hypertension. Notably, a study of pulmonary arterial hypertension patients found that SNO-Hb levels were markedly decreased and their RBCs exhibited deficits in the ability to effect hypoxic vasodilation (Figure 6B–E).¹⁰⁹

SNOs in Sepsis

Sepsis produces dramatic induction of iNOS expression¹⁹⁵ leading to increases in NO production that result in dysregulated protein S-nitrosylation, including of Hb.¹⁸⁴ In RBCs from septic patients, SNO-Hb levels are ~4-fold higher than in healthy controls;⁹⁵ as a consequence, SNO-based vasodilation becomes independent of P_O2,¹⁸⁴ potentially leading to shunts in blood flow. The pathologic importance of excessive NO production is reinforced by the finding that GSNOR^−/− mice (with increased SNO levels) have higher mortality in sepsis models compared to WT littermates.⁹⁵ Additional reductions in tissue oxygenation can be attributed to sepsis-induced changes in RBC rheology, specifically, decreased deformability and increased rigidity that hinder movement (flow) of RBCs through the microvasculature.¹⁹⁶ RBC rheology is further impaired by reactive oxygen species and redistribution of membrane phospholipids, the degree of which appears to be dependent on the infectious agent.¹⁹⁷ Recent experiments in a rabbit model of lipopolysaccharide-induced shock suggest that hyper SNO-Hb is impaired in SNO release, perhaps reflecting small arterial-venous O₂ gradients (since O₂ and SNO release are coupled), contributing to microvascular dysfunction in oxygen delivery.¹⁹⁸

SNOs in Diseases with Impaired Perfusion

Here we discuss three distinct oxygenation pathologies as examples of diseases defined (at least in part) by deficiencies in RBC-SNO.

Pulmonary Arterial Hypertension

(PAH) is defined by hemodynamic changes, but on a microscopic level it can be viewed as a disorder of microvascular blood flow with impairments in tissue oxygenation, suggesting a SNO-based component. In a cohort of PAH patients with mild hypoxemia, SNO-Hb levels were significantly less than those observed in healthy humans (Figure 6B).¹⁰⁹ The magnitude of SNO-Hb reduction was inversely correlated with pulmonary artery pressure (Figure 6C). Further, using an in vitro bioassay system, RBCs from these patients exhibited a marked reduction in ability to effect hypoxic vasodilation (Figure 6D–E). Palmer and colleagues have demonstrated elegantly in mice that high flux of SNO released chronically from RBCs is hypoxia-mimetic, activating transcriptional responses that raise pulmonary pressure.⁹⁷ In their model, imbalanced RBC-SNO signaling results from feeding mice N-acetyl cysteine, which acts as a SNO acceptor to reduce SNO-Hb (shifting equilibrium in favor of the extracellular SNO pool) and increases vasodilatory activity systemwide to result in dysfunction that resembles PAH,⁹⁷ indicating that SNO-Hb is an element of integrated system-level transduction of hypoxia.

Sickle Cell Disease

(SCD) results from a point mutation substituting Val for Hb β-chain Glu6 to create the variant HbS. Whereas normal RBCs dilate microvasculature in response to decreasing oxygen tension, sickle RBCs deform, become rigid, and at times polymerize to induce vaso-occlusion, pain, and hypoxic tissue injury.¹⁹⁹ HbS has reduced oxygen affinity that disfavors SNO-Hb formation, resulting in low SNO-Hb content in SCD (Figure 6F).¹⁵² In addition, the redox state of sickle RBCs is oxidizing, which leads to an increase in membrane thiol oxidation, including oxidation of the AE1 thiols that would be used to export NO bioactivity (Figure 6G). Hypoxic vasodilation by sickle RBCs is thereby impaired. Thus, SCD can be viewed as chronic condition of disrupted RBC-SNO homeostasis. In support of this view, the magnitude of the HbS SNO deficiency is commensurate with disease severity. In bioassays, impairment in hypoxic vasodilation by RBCs correlated with lower SNO-Hb levels and with disease severity (Figure 6H). These observations present SCD in terms of defects in SNO trafficking, with respect to: first, movement of NO from heme to Cys in HbS (impaired SNO synthase activity); second, movement of NO from HbS to AE1 (impaired transnitrosylase activity); and third, movement of NO bioactivity across the RBC membrane into the vascular space (oxidative stress, impaired AE1 transnitrosylase activity). In vitro re-nitrosylation restored HbS SNO levels and hypoxic vasodilatory activity. SNO-based therapies have yet to be tested in this patient population but other pre-clinical data suggest potential benefit. Specifically, hydroxyurea administered to RBCs from Townes mice enhanced S-nitrosylation of βCys93²⁰⁰ – a result that might well be augmented in the presence of a direct S-nitrosylating agent.²⁰¹

Peripheral Artery Disease

(PAD) is viewed as the consequence of systemic atherosclerosis reducing blood flow in arteries, and can culminate in critical limb ischemia (CLI).^202–205 It is well-appreciated that blood flow remains impaired in a significant fraction of patients after procedures to restore bulk flow, indicating microvascular disease. Dysregulation of NO bioactivity is an important contributor to PAD/CLI morbidity and mortality.^206–210 However, clinical trials have failed to show a strong therapeutic benefit from non-specific modulation of NO.²¹⁰ We considered that the defect might be RBC-based, and assessed the relationship between SNO-Hb and limb blood flow in PAD patients compared to age-matched healthy controls. SNO-Hb levels in PAD patients were less than half those of healthy subjects (Figure 6I). As a consequence, we postulated that PAD patients with altered SNO-Hb levels would exhibit an aberrant (i.e., delayed) hyperemic response. Representative foot tissue oxygenation recovery tracings from a healthy subject and a PAD patient after release of an occluding cuff are depicted in Figure 6J–K. The PAD patient exhibited a protracted re-oxygenation response compared to the healthy subject, consistent with findings in βCys93A mice subjected to occlusion (Figure 1C). Further work will be required to define SNO interventions to improve PAD and prevent CLI.

RBC Transfusion

Blood transfusion to correct anemia can be lifesaving. However, evidence continues to accumulate that administration of stored RBCs may not always be beneficial and may actually cause harm^211–215 – findings of particular concern because even mild anemia is prognostic of adverse patient outcomes.^216,217 Transfusion is premised on the assumption that additional RBCs will improve tissue oxygenation. However, there has been no rigorous demonstration that transfusion improves oxygen delivery in the general case, and accumulating data suggest that this goal is not being met. Banked RBCs do not replicate the function of native RBCs.^218–221 Instead, administration of stored RBCs is an independent risk factor for morbidity and mortality across a diverse set of clinical situations.^{211–214,220,222} Deleterious outcomes linked to transfusion (acute kidney injury, MI, and death)²²⁰ suggest that banked blood may exacerbate rather than correct anemia-induced tissue hypoxia.^223,224 In animal studies, this is in fact the case.²²⁵

To the extent that increasing RBC number by transfusion does not prevent adverse outcomes or reverse risks of anemia, banked blood is evidently doing some harm.^212,220,226 Thus, raising hematocrit into the normal range is not advocated currently. Instead, the focus is on identifying anemic transfusion thresholds that avoid adverse outcomes.^227,228 Donated blood undergoes time-dependent changes, including decreased RBC flexibility,^229,230 hemolysis,²³¹ increased adhesiveness²³² and decreases in 2,3-DPG.²³³ However, none of these measures is correlated with altered oxygen delivery in vivo. Moreover, even freshly processed blood impairs tissue oxygenation, prior to storage-related biochemical changes;²³⁴ and transfusion of blood that is only a few days old has been associated with marked increases in mortality.²¹⁵ This is not to say that other components of the storage lesion do not have a role in the adverse effects of transfusion²³⁵, but these do not account for defective oxygen delivery capability in animals²²⁴ or humans.²³⁶

Based on the role of SNOs in tissue oxygenation, we proposed that the inability of banked blood to improve oxygen delivery relects a deficiency in SNO-Hb.¹³⁷ Indeed, Hb-βCys93Ala mutant mice deficient in SNO-Hb, but with otherwise normal RBCs, exhibit basal tissue hypoxia (see Figure 5C). Administration of RBCs deficient in SNO-Hb creates tissue hypoxia.¹³⁷ Thus SNO-Hb depletion, whether genetically or experimentally induced, impairs tissue oxygenation, despite normal oxygen content and function of RBCs. Storage of blood leads to rapid losses in SNO-Hb (Figure 7A) that precisely parallel losses in the ability of banked RBCs to effect hypoxic vasodilation.^115,178 This defect in vasodilation can be corrected by repleting SNO-Hb (Figure 7B). These findings were consistent with data demonstrating declines in tissue oxygenation upon receipt of stored blood (Figure 7C) and the reduced ability of stored SNO-Hb-depleted blood to improve hypoxic coronary blood flow (Figure 7D). Clinically-relevant small and large animal transfusion paradigms across three different species uniformly demonstrated that banked blood deficient in SNO-Hb failed to correct anemia-induced reductions in blood flow and oxygen utilization.¹³⁷ In a sheep model, mimicking patient transfusion by removal of 2 units of blood and replacing with 2 units of stored sheep RBCs resulted in reduced tissue oxygen extraction, while SNO-repleted RBCs increased tissue oxygen extraction and raised muscle P_O2 (Figure 7E,F).

Figure 7. Blood storage diminishes SNO-Hb, hypoxic vasodilation and O₂ delivery, and RBC SNO repletion improves tissue O₂ delivery.

A. Storage of human-donor blood results in rapid loss of SNO-Hb. B. Loss of SNO-Hb correlates with diminished hypoxic vasodilation by RBCs. Aortic vasorelaxation by RBCs is restored by repleting SNO-Hb to physiological level (~1–5μM; see Figure 3A). C. Peripheral tissue oxygen saturation in mice declines following transfusion with stored blood (1–2 units equivalent), consistent with diminished hypoxic vasodilation and microvascular plugging (see Figure 1A, 1C). D. Canine coronary artery blood flow in vivo elicited by renitrosylated RBCs is significantly greater than that produced by SNO-depleted (stored) RBCs, with the degree of change greater under hypoxia. E. Time course of change in arterial-venous O₂ content (mean ± SD) in sheep after replacement of 2 units of RBCs stored for 14 days (untreated, squares) or repleted with physiological amounts of SNO (1–5 μM) immediately prior to transfusion (circles, renitrosylated). *, higher than baseline, X, lower than baseline, p<0.05. F. Skeletal muscle P_O2 in sheep transfused as in (E). *, higher than baseline, p<0.05. Modified panels A,B,D from Ref¹¹⁵, Panel C from Ref²⁷⁰ and Panels E,F from Ref¹³⁷, with permission.

We assessed SNO-Hb levels and patient status during neonatal cardio-pulmonary bypass.²³⁷ Intra-operative transfusion in this setting (including circuit priming) is associated with adverse outcomes.^238–240 At our institution, packed RBCs no older than 5 days are used. Yet even with this very fresh blood, we documented adverse effects of transfusion on oxygenation and kidney function, consistent with loss of SNO-Hb within hours of donation.^115,178 Specifically, the percent of blood volume replaced by transfusion directly correlated with ventilator time, and inversely correlated with kidney function. In addition, an inverse association was found between SNO-Hb and post-operative increase in Hb (ΔHb), reflecting the amount of SNO-Hb-deficient RBCs circulating in the patient. Decline in SNO-Hb and increase in ΔHb correlated with probability of kidney dysfunction and hypoxia-related complications, and SNO-Hb was an inverse predictor of outcome. These findings suggest that SNO-Hb and ΔHb are prognostic biomarkers following pediatric cardio-pulmonary bypass, and that maintenance of RBC-derived NO bioactivity might confer therapeutic benefit.

Part 4. THERAPEUTIC ATTEMPTS TO IMPROVE OXYGENATION BY RBCs

Interventions to increase SNO-Hb have the potential to improve oxygen delivery, as well as reverse vasculopathy of NO deficiency. Here we review the effectiveness of current NO-based therapy in hypoxic conditions and in restoring SNO-Hb levels in particular.

Nitric Oxide

In 1999, the originally-approved clinical indication for inhaled NO gas (iNO) issued by the FDA was limited to “treatment of term and near-term (>34 weeks) neonates with hypoxic respiratory failure associated with clinical or echocardiographic evidence of pulmonary hypertension”. At the time, anticipation was high that iNO would be a universal remedy for a wide range of oxygenation pathologies. However, after 20 years, the approved indication for iNO has not changed. In this regard, it is important to appreciate that NO gas is fundamentally different from the products of in vivo NOS activity, which derive their activity in large part from SNOs. NO gas is unreactive toward thiols and relatively ineffective in S-nitrosylation reactions.⁵⁶

Nitrates and Inorganic Nitrite

While nitroglycerin is effective in the treatment of angina and heart failure by dilation of coronary arteries and capacitance vessels, it only weakly improves microvascular flow through a RBC-mediated mechanism,²⁴¹ explaining its limited use in microvascular and RBC pathologies.

Inorganic nitrite has been suggested as an alternative to tolerance-inducing NO donors.²⁴² The pharmacological effects of nitrite depend on route of administration. Through interactions with salivary enzymes and gastric acidification, oral nitrite forms SNOs, which likely mediate its circulatory effect.⁴⁹ In contrast, i.v. infusion of nitrite has a bi-phasic effect, likely reflecting nitrite conversion to SNO-Hb at very low concentrations^103,27 but direct vasodilatory effects at very high concentrations. Pharmacological concentrations actually inhibit SNO-Hb formation,^27,100 perhaps explaining equivocal or negative results in some clinical studies (e.g.^243–246) versus subtle cardiovascular improvements in others.²⁴²

Regulation of Endogenous NO and SNO

Supplementation of L-Arg, the substrate for enzymatic generation of NO/SNO, would seem to be a simple treatment to raise SNO levels. However, >98% of L-Arg is used for non-NO-related metabolism, and supplemental L-Arg has failed to provide sustained benefit in clinical trials, including exercise performance,²⁴⁷ post MI therapy,²⁴⁸ and peripheral artery disease.²¹⁰ Arginase inhibitors should allow L-Arg to be utilized preferentially by NOS, and arginase inhibition has been reported to improve RBC-dependent recovery in an ex-vivo cardiac ischemia-reperfusion animal model,^249–251 independently of Hb βCys93¹⁸⁰ (in contrast to physiological reactive hyperemia, which is critically dependent on Hb βCys93 (Figure 1C), and suggesting that the cardiac effect is independent of RBC vasodilation). Another approach to increasing NO is to target endogenous NOS inhibitors, such as asymmetric dimethylarginine (ADMA).^252–254 ADMA increases with RBC storage to impair NO generation and RBC-mediated vasodilation.²⁵⁵ In animal models, restoring NO generation in stored RBCs confers protection in models of cardiac injury.²⁵¹

Under physiologic conditions, SNO-protein levels are determined by the balance between nitrosylation and denitrosylation, with steady-state levels determined by denitrosylases. In RBCs, inhibition of GSNOR raises SNO-Hb levels and increases vasodilatory activity of RBCs.⁹⁵ Small molecule inhibitors of GSNOR have been advanced to human testing,^256,257 but their effects on oxygenation and RBC pathologies remain untested.

S-Nitrosylating Agents

Ideal S-nitrosylating agents would react with specific thiols to form stable adducts with designed activity. Advantages of SNOs include their resistance to inactivation by heme in Hb, as well as ability to dilate microcirculatory vessels that are refractory to NO itself.⁶⁸ Since sGC/cGMP vasodilation by NO/SNOs varies with vessel size, NO as such is reliant on sGC to induce vasodilation, while SNOs including SNO-Hb^258,259 vasodilate by multiple mechanisms.⁶⁵ Moreover, auto-regulation of blood flow governing tissue oxygenation is evidently sGC-independent, since sGC deficiency or inhibition has no reported effect on tissue oxygenation,²⁶⁰ in stark contrast to deficiencies in SNO-Hb.^9,138

A lead compound suitable for clinical use to enhance RBC-based vasodilation, inhaled ethyl nitrite (ENO), has advanced into clinical trials. ENO effectively generates SNO-Hb in physiological range,²⁶¹ and has potent vasodilatory effects mediated by RBCs that correct tissue hypoxia in animal models and humans.^{8,137,258,262–265} A Phase I safety/dose-ranging study in young healthy adult volunteers with right-heart catheterization for hemodynamic monitoring showed ENO raises SNO-Hb and recapitulates hypoxic vasodilation.²⁶⁶ Thus, hemodynamic changes induced by ENO were likely mediated by a dose-dependent decline in systemic vascular resistance, consistent with a direct role of hypoxic vasodilation by RBCs. Most notably, skeletal muscle hypoxia was reversed by ENO: calf muscle oxygenation (StO₂) in subjects on FiO₂ of 0.12 remained at normoxic baseline, with an increase in RBC SNO-Hb levels.²⁶⁶ Taken together, these data support the interpretation that ENO improved skeletal muscle oxygen delivery and utilization through a SNO-based mechanism to offset reduced oxygen availability.

S-Nitrosylation in Transfusion

SNO loss during RBC storage is massive relative to the amount of NO produced in vivo (~1 μmol/day/70kg). Thus, during the immediate post-transfusion period, SNO-deficient RBCs cannot be restored to normal SNO levels, and instead act as sinks for NO and impair SNO homeostasis of native RBCs. Transfusion recipients are thus predisposed to vasoconstriction and ischemia just when microcirculatory vasodilation is needed to improve end-organ oxygen delivery. Correcting RBC SNO-Hb deficit produces post-transfusion improvements in oxygen utilization and organ function in several preclinical paradigms (see Figure 7).¹³⁷ SNO-repletion also reverses storage-related declines in RBC deformability.²⁶⁷ Collectively, these data offer compelling support for the proposition that SNO-Hb deficiency contributes to the clinical finding that standard transfusion regimens do little to improve end-organ oxygen delivery, whereas repletion of SNO-Hb at the time of transfusion results in sustained improvements in P_O2 and related parameters of oxygen sufficiency.

Part 5. SUMMARY

RBCs should no longer be viewed as passive conductors of oxygen. Instead they are the integral regulators of a three-gas respiratory cycle that employs NO bioactivity to ensure that oxygen delivery and carbon dioxide removal are matched to local metabolic demand within tissue microvessels. The physiologic hypoxic vasodilation observed by Guyton in the 1960s is recapitulated by SNO-replete RBCs and by SNO-Hb itself, whereby SNO is released from Hb and RBCs during deoxygenation in tissues to regulate vessels directly. Appreciating these inter-relationships and their roles in acute and chronic pathologies of oxygenation is imperative to our understanding of normal cardiovascular physiology, and to developing better therapies for major causes of human morbidity and mortality.

Nonstandard Abbreviations and Acronyms:

ADMA

Asymmetric dimethylarginine

cGMP

cyclic guanosine monophosphate

CA

Carbonic anhydrase

Cys

cysteine

EDRF

endothelium-derived relaxing factor

ENO

ethyl nitrite

eNOS

endothelial nitric oxide synthase

FiO₂

fractional inspired oxygen

GSNO

S-nitrosoglutathione

GSNOR

S-nitrosoglutathione reductase

HAPE

high-altitude pulmonary edema

Hb

hemoglobin

HbFeNO

hemoglobin iron nitrosy

His

histidine

iNO

inhaled nitric oxide

iNOS

inducible nitric oxide synthase

IVU

Integrated vascular unit

L-Arg

L-Arginine

L-Cit

L-Citrulline

metHb

met-hemoglobin

nNOS

neuronal nitric oxide synthase

NO

nitric oxide

PAD

peripheral artery disease

PAH

pulmonary arterial hypertension

Phe

phenylalanine

PKG

protein kinase G

P_O2

partial pressure of oxygen

RBC

red blood cel

SaO₂

systemic arterial oxygen saturation

sCG

soluble guanylyl cyclase

SNO

S-nitrosothiol

SNO-Hb

S-nitrosylated hemoglobin

SNO-RAC

SNO-resin assisted capture

StO₂

tissue oxygen saturation

SvO₂

systemic venous oxygen saturation

XO

Xanthine oxidase