The Role of Carbon Monoxide and Heme Oxygenase in the Prevention of Sickle Cell Disease Vaso-Occlusive Crises

1.0. Overview of Sickle Cell Disease

Sickle cell disease (SCD) is a painful, lifelong hemoglobinopathy with substantial morbidities and premature mortality. It is inherited as a point mutation in the hemoglobin (Hb) beta-globin gene where glutamic acid at position 6 is substituted by valine. This mutation provides resistance to malaria, but, in the homozygous form, results in SCD, which is characterized by anemia and painful vaso-occlusive crises (VOCs). The change in molecular structure allows Hb in the deoxygenated state to form polymers that promote Hb autoxidation, red blood cell (RBC) membrane damage, decreased RBC deformability, intravascular and extravascular hemolysis, inflammation, vaso-occlusion, and ultimately organ injury¹.

Intravascular hemolysis results in the release of free Hb into plasma. In normal individuals, free Hb is rapidly bound to circulating haptoglobin with high affinity (K_d = ~10⁻¹² M) and carried to CD163 receptors on macrophages for endocytosis and degradation^2,3,4. However in SCD patients, the continuous release of Hb from hemolyzed sickle RBCs depletes the plasma haptoglobin resulting in free Hb circulating in plasma⁵. Free Hb in the vasculature rapidly quenches the vasodilator nitric oxide (NO) resulting in vasospasm^6,7. In the pro-oxidative SCD vasculature, the reaction of NO and other oxidants with ferrous Hb oxidizes the heme iron from ferrous Hb to ferric Hb (metHb). MetHb is unstable and releases hemin into the vascular space and it is hemin release that initiates the pathophysiological cascade ultimately resulting in the vascular damage underpinning the widespread tissue and organ damage that is the hallmark of this devastating disease^8,9,10.

In normal individuals, the released hemin is rapidly bound to circulating hemopexin with high affinity (K_d< 10⁻¹³ M) and carried to CD91 (LRP1) receptors on hepatocytes for endocytosis and degradation^11,12,13. However, in SCD patients the continuous formation of metHb in plasma and release of hemin depletes plasma hemopexin levels, in a similar manner to hemoglobin depleting haptoglobin, resulting in free hemin bound to lower affinity RBC-derived microparticles, albumin and lipoproteins^14,15,16,17. Hemin is released continuously from these carriers to cell membranes of circulating immune cells, platelets and endothelial cells in the vessel wall^18,14. Hemin in the cell membrane is cytotoxic, promotes oxidative stress, and activates innate immune toll-like receptor 4 (TLR4) signaling initiating activation of monocytes and endothelium^{8,19,20,21,22,23}. The resulting inflammatory cascades include degranulation of endothelial Weibel-Palade bodies, enzymatically-derived oxidant production, NF-κB-driven inflammasome activation, pro-inflammatory cytokine production, and adhesion molecule expression. Hemin, cytokines, lipopolysaccharide (LPS), infections, and the microbiome can all promote pro-inflammatory responses leading to adhesion of sickle RBCs, leukocytes, and platelets to the endothelium and the formation of multicellular aggregates leading to vaso-occlusion^{8,24,25,26,27,28}. The subsequent re-opening of occluded vessels promotes ischemia-reperfusion pathophysiology, inflammatory pain, and tissue injury that amplify the innate immune inflammatory responses^29,30,31.

SCD manifests clinically through a variety of severe morbidities, including stroke, auto-splenectomy with risk of resultant overwhelming infection, pulmonary hypertension, acute chest syndrome, infarction of organs, priapism, leg ulcers, and renal injury³². According to the Centers for Disease Control (CDC), the median survival among adults in North America is approximately 42 years although a higher median survival rate was reported in an adult-only study in patients with SCD in the United Kingdom³³. In addition to the impact on survival, the ravages of this microvascular occlusive disease severely and progressively impact the quality of life from infancy through adulthood³⁴.

Although sickle cell trait has been shown to have evolved in equatorial Africa, individuals with sickle trait have migrated to most major geographic regions of the world³⁵. It is estimated that SCD affects approximately 100,000 Americans³⁶, as well as approximately 37,000 patients in the EU³⁷, and millions of patients globally. The prevalence of the disease is increasing as infant survival rises worldwide³⁵. The number of children born with SCD is expected to exceed 14 million worldwide in the next 40 years, with about 79% of these children being born in sub-Saharan Africa in 2010 where the mortality is highest under 5 years of age ³⁸.

A substantial medical need exists to reduce VOCs, also termed sickle cell crises and the consequent morbidities and mortality. Hydroxyurea remains the only therapeutic approved for the prevention of VOCs in SCD. Hydroxyurea reduces, but does not eliminate the incidence of VOC’s or the morbidities and mortality of SCD. In response to the unmet medical need, there are new therapeutics in development that target a variety of mechanisms, each with its own unique potential risks and benefits³⁹. This review is focused on carbon monoxide (CO), as this small molecule gasotransmitter can impart potent salutary medical benefits based upon specific mechanisms of action at non-toxic low doses. We discuss the promising data that have emerged showing the benefits of CO and the efforts underway to develop it as a potent therapeutic.

2.0. Overview of the Heme Oxygenase/Carbon Monoxide Pathway

The discovery of nitric oxide (NO) gas in 1987 revealed the novel concept that endogenous production of a gas as a signal transmitter can impart diverse and critical systemic physiologic effects across a wide spectrum of biological and pathological processes. Unlike NO which is highly reactive, CO is stable. CO binds to cellular heme moieties and modulates the activity of heme-containing enzymes including mitochondrial oxidases, guanylate cyclase, nitric oxide synthase and NADPH oxidase^40,41. Ultimately, the target for CO will depend on the cell type and the expression of hemoproteins that are present in the cell. For example, guanylate cyclase is highly present in vascular smooth muscle cells, but very low in endothelial cells.

CO is generated endogenously when heme is degraded by the heme oxygenase (HO) enzymes. Most, if not all, endogenous CO is generated by HO degradation of heme. These enzymes generate CO in the cytosol. In recent years, interest in HO has grown beyond its role as a metabolic enzyme that degrades heme with research exploding across multiple scientific disciplines. HO is now classified as an anti-inflammatory cytoprotective enzyme that generates bioactive CO that mimics many of the effects of HO and has been described as an anti-inflammatory gasotransmitter. In addition, heme degradation by HO produces iron that is stored in ferritin and the bile pigment biliverdin that is metabolized to bilirubin by biliverdin reductase (Figure 1).

Figure 1.

Heme degradation releases bioactive products

Intravascular hemolysis of RBCs with the resultant free Hb and heme within the circulation is a normally occurring event, albeit occurring at a low rate. Normally, haptoglobin and hemopexin scavenge circulating Hb and heme, respectively, and deliver them to macrophages and hepatocytes for processing by HO. It is when these systems are overwhelmed that pathology such as those of SCD result. Free heme in the presence of reactive oxygen species produced by activated granulocytes is cytotoxic to endothelium¹⁸. Heme uptake by endothelium synergizes with polymorphonuclear granulocyte-mediated damage⁴². In addition, heme activates innate immune TLR4 signaling on endothelium, leukocytes and platelets leading to vascular cell activation that includes oxidant production, NF-κB activation, adhesion molecule expression, pro-inflammatory cytokine production, and necroptosis/apoptosis. In many instances these reactions amplify tissue injury caused by initiating events such as trauma, infection, and/or hemolysis. Detoxification of free heme by HO and the mitigation of heme-induced innate immune responses are critical for survival.

There are two HO enzymes, HO-1 and HO-2, that degrade heme into equimolar quantities of iron, biliverdin and CO (Figure 1,^43,44). The isoforms differ in their structure, location and biological role. HO-1 is highly inducible and found in all cells and is upregulated by a variety of agents including heme, pathogens, ischemia/reperfusion injury (IRI), oxidants, and, in certain instances, CO itself^{45,46,47,48,49,50,51}. HO-2 is constitutively expressed primarily in the brain, vasculature, and testes where it is involved in neurotransmission, vasomotor tone, and spermatogenesis^52,53. HO-1 and CO are recognized as critically protective due to their ability to modulate inflammation and apoptosis^40,54,55. At low doses, equivalent to that generated by HO-1, CO has been shown in a substantial number of research observations to induce HO-1 itself, as well as modulate a wide spectrum of genes linked to inhibition of oxidative stress, inflammation, and apoptosis, including Nrf2, NF-κB, STAT3, CREBH and various MAP kinases^{51,54,56,57,58,59,60,61}. Importantly, these effects of CO have been documented across a variety of CO administration approaches, including increasing endogenous production of CO⁴³, inhalation of CO gas, and intravenous administration of CO in the form of CO attached to a transition metal, such as molybdenum and ruthenium (CO Releasing Molecules [CORMs]), or even iron as a CO-saturated Hb^40,51.

CO is a tasteless, odorless, non-corrosive gas that exists as a stable diatomic molecule consisting of one carbon and one oxygen atom. It is physiologically present in all species that utilize heme protein. In healthy humans basal levels of CO in blood range between 0.5 and 1.5% carbon-monoxy hemoglobin (COHb)⁶². CO binds competitively with oxygen for Hb, with an affinity for Hb 200–250 times that of oxygen. In addition to decreasing the oxygen carrying capacity of Hb, the binding of CO to Hb locks Hb in the R state, increasing the affinity of Hb for oxygen and thereby shifting the oxygen dissociation curve to the left. In addition to Hb, CO also binds to myoglobin and a variety of intracellular hemeproteins including cytochromes and other metallo-enzymes. Exogenous CO is generated during the combustion of fossil fuels, or emitted from the oceans or volcanoes where it is found in large quantities. CO is rapidly taken into the body by inhalation from the atmosphere. It is readily absorbed and excreted through the lungs and rapidly distributed by the blood stream essentially unchanged except for a very low percentage that is converted to CO₂⁶². In the tissues a portion of CO offloads from Hb, much like O_2, where it can influence cellular function.

The toxicological effects of CO have been well studied. The half-life of CO in sedentary adults at sea level is 4 to 5 hours⁶². Blood COHb levels are considered a reliable biological marker of the dose or exposure of CO an individual has received. Levels of COHb above 20–30% are associated with adverse symptoms such as headaches and dizziness. Mortality is documented with levels above 50%^62,63,64. Levels of COHb lower than 15% are well tolerated for both short and prolonged time periods, as indicated in the epidemiological and clinical literature^62,63,64. It is reported that exertional capacity is reduced when COHb is above 5% and individuals with advanced cardiovascular disease are at risk of symptoms with exertion when levels are above 3%⁶². Headache and nausea have been reported when COHb is raised above 5%, yet rigorous Phase I safety trials have documented that COHb of up to 13.9% showed no adverse effects (see Safety Studies of CO below) and ongoing Phase 2 trials have documented few related adverse events. It should also be noted that there is controversy and debate as to reproducible physiologic effects of CO on the body and the accompanying symptoms that result⁶⁵.

2.1. Carbon Monoxide as a Gasotransmitter

In contrast to the established CO toxicity demonstrated at high COHb levels, low-dose CO is now accepted as a potent gasotransmitter, capable of regulating a host of physiologic and therapeutic processes^40,66. In recent years, research has focused extensively on the mechanisms underlying how CO imparts its salutary effects. The large body of generated data clearly shows that CO provides cytoprotection in models of sterile inflammation including organ transplantation⁶⁷, acute lung injury⁶⁸, SCD, stroke⁶⁹, and other conditions^43,70,71. In each instance the protection observed was associated with modulation of the innate and adaptive immune response and restoration of tissue and organ function. One underlying theme is that CO at low concentrations abrogates IRI providing cytoprotection as demonstrated in an IRI lung injury model⁷². Research has shown the paradoxical effect that CO increased tissue oxygen availability by displacing O₂ within intracellular compartments⁷². In the lung, CO-mediated protection in wild-type mice was abrogated in mice null for the gene encoding plasminogen activator inhibitor (Pai1^−/−) mice. These data established a fundamental link between CO, hemostasis and the prevention of IRI based on the ability of CO to derepress the fibrinolytic axis. CO has been shown to be effective both prophylactically and therapeutically even when started at the peak of pathology. In short, CO seems to befit the need of the tissue to restore normal function. This has been defined as homeo-’dynamic’⁴⁰.

3.0. Overview of Potential Carbon Monoxide Actions in Sickle Cell Disease

How could CO modulate SCD? Exogenous CO may potentially affect SCD through six distinct mechanisms of action (Table 1). First, CO could affect HbS polymerization. CO can bind to HbS polymers and enhance their melting which minimizes their persistence in red cells within the circulation, and their ability to rapidly seed additional polymers during transit in hypoxic tissues. These polymers may contribute to abnormal rheological flow and ultimately vaso-occlusion⁷³. However, the equilibrium solubility of HbS with CO is linearly proportional to the amount of bound CO⁷⁴. Thus, very high concentrations of CO may be required to have a clinical benefit in SCD through this mechanism. Yet, Sirs showed a reduction of sickle RBCs from 10.2% to 3.9% with a COHb level of 4% in blood from a single patient⁷⁵, and there have been reports on studies of CO effects on HbS polymerization, with one showing in vitro binding of CO to sickle polymers⁷³ and a second study utilizing a microfluidic device showing concentrations of CO as low as 0.1% preventing occlusion when perfused with blood from SCD individuals⁷⁶. The melting of polymers may be a key event minimizing the persistence of these polymers in the circulation and as a consequence, limiting these acting as a seed for rapid polymerization in hypoxic tissue⁷³. Further, given the high affinity of CO for Hb, it is a logical supposition that the impact of CO on polymerization could limit sickle-associated clinical effects, but the extent and duration of these effect in vivo have received little focus. Second, CO inhibits the Ca²⁺-permeable cation conductance channel or P-sickle, thereby inhibiting sickle RBC dehydration⁷⁷. It has been postulated that an increase in RBC hydration could decrease the rate of HbS polymer formation.

Table 1.

Potential CO targets and mechanisms of action in SCD.

Potential Mechanism	Potential Effect	Potential Pathways/Targets/Markers
1. Anti-polymerization73–76	CO binds to HbS, modulating HbS polymerization	Ferrous HbS
2. Sickle red blood cell hydration77	Inhibition of Ca2+-permeable cation conductance	Psickle
3. Anti-oxidative78–84	Decreased oxidative stress	Decreased HbS oxidation, inhibition of NADPH oxidase, and activation Nrf-2-responsive genes including HO-1 and anti-oxidant enzymes.
4. Anti-inflammatory51,54,56–61,104–115	Modulation of pro- and anti-inflammatory mediators	Activation of Nrf-2-responsive genes including HO-1 and anti-inflammatory proteins; decreased NF-κB activation leading to decreased VCAM-1, ICAM-1, P-selectin, and pro-inflammatory cytokines; increased nuclear phospho-p38 MAPK and phospho-Akt; decreased innate immune TLR4 signaling and cell membrane expression; and decreased inflammasome activation
5. Vasodilation93,94	Increased cGMP	Guanylate cyclase and indirectly by nitric oxide synthase (NOS)
6. Anti-apoptotic97–100,151–153	Decreased mitochondrial damage and cell death	Cytochrome C release, caspase activation, and mitochondrial membrane potential

Third, CO may have anti-oxidative effects by inhibiting the oxidation of ferrous HbS to ferric HbS through its ligation to heme and thereby preventing heme release from ferric HbS^78,79. CO may also inhibit superoxide production by NADPH oxidase 2 (NOX2) through CO binding to heme groups on the gp91^phox subunit⁸⁰. In addition, CO activates Nrf2-responsive anti-oxidant genes including enzymes involved in glutathione production and regeneration, NADPH production, and iron sequestration such as HO-1 and ferritin.

Fourth, CO has remarkable anti-inflammatory properties through the activation of Nrf2-responsive genes leading to HO-1 upregulation and NF-κB downregulation⁵¹. A wide spectrum of genes are modulated by CO including Nrf2, NF-κB, STAT3, CREBH and various MAP kinases^{51,54,56,57,58,59,60,61}. CO has been shown to modulate inflammation by increasing HIF1α and PPARγ signaling, largely dependent on the ability of CO to generate reactive oxygen species, generated in large part by the mitochondria^81,82,83,84. Clues to the protective effect of CO are delineated in the work of Ferreira et al showing that mice with HbS are resistant to malaria infection, due in part to the induction of HO-1 and its products including CO⁸⁵. HO-1 is highly inducible by CO and both HO-1 and CO modulate inflammatory responses^{51,54,56,57,58,59,60,61}. HO-1 and CO-mediated inhibition of NF-κB activation reduces adhesion molecule and pro-inflammatory cytokine expression, which inhibits vaso-occlusion in SCD mouse models. Other anti-inflammatory effects include increases in phospho-p38 MAPK and phospho-Akt as well as decreases in innate immune TLR4 signaling and TLR4 trafficking to the cell membrane surface^86,87,88. CO also inhibits NLRP3 inflammasome activation^89,90,91,92. Fifth, NO bioavailability is decreased in SCD. However, CO can substitute for NO binding to the heme protein guanylate cyclase and increase cGMP production leading to vasodilation of blood vessels^93,94. CO has been shown to have a relationship with NO in the vasculature where CO increases endothelial cell repair after injury and can induce apoptosis of hyperproliferative vascular smooth muscle cells that cause pulmonary hypertension. In each instance, CO increases NO bioavailability that is involved in the observed benefits^95,96. Sixth, CO has anti-apoptotic activity by preventing mitochondrial depolarization, inhibiting cytochrome C release, suppressing p53 expression, and inhibiting caspase activation^97,98,99,100.

Although the potential efficacy of CO in SCD may rely upon all of these mechanisms of action, the exact dosing regimen of CO and resulting COHb concentration to maximize efficacy and ensure safety, will depend on the interplay of these mechanisms in the clinical setting. Currently available data indicates that given the relatively short half-life of CO, the prevention of polymerization through CO binding to HbS (Mechanism 1 above) and the vasodilatory effects of CO (Mechanism 5) are anticipated to require more frequent and higher doses, while the anti-inflammatory and anti-apoptotic actions (Mechanisms 4 and 6, respectively) will likely require less frequent and lower doses. Preclinical data suggests that, in transgenic SCD mice, smaller, daily doses with peak COHb levels at 2% to 6.5% are effective in down-regulating the pathophysiological mechanisms that are associated with SCD^51,101.

3.1. Reported Studies of CO in Sickle Cell Disease

A substantial number of studies, including cell based models and in vivo in animals and humans have been conducted supporting the potential efficacy, safety, and bioavailability of CO in SCD using a variety of delivery mechanisms, including inhaled and parenteral administration.

3.1.1. Preclinical In Vitro Sickle Cell Disease Studies

A number of in vitro studies have demonstrated the potential for CO to directly limit the sickling phenomenon through binding to HbS. As early as 1963, it was shown that the presence of CO reduces the proportion of sickled cells in vitro⁷⁵. A more recent study in a microfluidic device confirmed that a small CO concentration prevented occlusion, even without the presence of oxygen⁷⁶. It has also been demonstrated that CO binds to and melts HbS polymers⁷³. In addition, Bains and colleagues demonstrated that HO-1 induction, exogenous biliverdin or CO administration by CORMs markedly decreased adhesion of RBC’s from sickle cell patients to cultured endothelial cells¹⁰².

3.1.2. Preclinical In Vivo Sickle Cell Disease Efficacy Studies

Data from four studies in four different transgenic SCD mouse models demonstrate that modulation of the HO-1/CO pathway is extremely effective in thwarting SCD pathophysiology and suggests low dose CO as a novel treatment approach. A seminal study by Belcher and coworkers first showed the beneficial effects of both the up-regulation of HO-1 and administration of CO in two SCD mouse models¹⁰³. Their study demonstrated that the up-regulation of HO-1 modulates the pathobiology of SCD. Overexpression of HO-1 in both S+S-Antilles and Berkeley SCD transgenic mice led to significantly reduced hypoxia/reoxygenation-induced vascular stasis. This protective effect of HO-1 was associated with the mitigation of inflammation, including the reduction in leukocyte-endothelium interactions and the inhibition of NF-κB, VCAM-1, and ICAM-1 expression. In contrast, the inhibition of HO-1 activity was shown to exacerbate vascular stasis. Most importantly, these anti-inflammatory and anti-vaso-occlusive effects of HO-1 were nearly identical in sickle mice inhaling low concentrations (250 parts per million [ppm]) of CO gas for 1 hour over 3 consecutive days).

A parallel study confirmed these effects of HO-1 in limiting vaso-occlusion and inflammation in SCD mice. In this study, livers of sickle mice overexpressing wild type HO-1, but not an enzymatically inactive HO-1, showed marked activation of the phospho-p38 MAPK and phospho-Akt cell signaling pathways, reduced levels of NF-κB p65 in liver nuclear extracts, and decreased sVCAM-1 in serum¹⁰⁴. Similarly, hypoxia-induced vascular stasis was only inhibited in sickle mice overexpressing enzymatically active HO-1. Vascular stasis was inhibited in the venules of dorsal skin despite the absence of the HO-1 transgene in the skin, indicating that overexpression of the wild type HO-1 transgene in the livers of sickle mice had distal effects on the vasculature suggesting a mediator that could be transported and exert effects remotely.

This work was extended further investigating the anti-inflammatory effects of CO in S+S-Antilles SCD mice¹⁰¹. In this study, treatment of SCD mice with 25 or 250 ppm of inhaled CO for 1 hour per day, 3 days per week for 8 to 10 weeks showed decreased vascular inflammation and end-organ pathology. Inhaled CO treatment significantly reduced total white blood cell, neutrophil and lymphocyte counts. Moreover, reduced staining for myeloid and lymphoid markers in the bone marrow was observed, and bone marrow exhibited a significant decrease in colony-forming units by granulocyte-macrophage colony-forming cell assays. Anti-inflammatory signaling pathways including phospho-Akt and phospho-p38 MAPK were also markedly increased in the livers of CO-treated mice. Importantly, treated sickle mice had a significant reduction in liver parenchymal necrosis, reflecting the anti-inflammatory and anti-vaso-occlusive benefits of CO administration, and these effects were seen at a peak COHb level of 2% and at higher levels.

A follow-on study demonstrated that the protective effects of CO extended to two additional SCD transgenic mouse models using a PEGylated COHb molecule (MP4CO) for intravenous CO administration⁵¹. This effect was not seen with the oxygenated molecule MP4OX. Similar to the studies described above, the administration of CO in the NY1DD transgenic SCD mouse model markedly induced HO-1 activity and inhibited NF-κB activation and hypoxia/reoxygenation-induced microvascular stasis. Importantly, CO administration also upregulated nuclear Nrf2, an important transcriptional regulator of HO-1 and other cytoprotective antioxidant genes. Additionally, in a heterozygous HbAS-Townes sickle trait mouse model, administration of CO significantly reduced mortality in a model of acute lung injury and cardiac dysfunction. CO administration decreased NF-κB activation, P-selectin and von Willebrand Factor expression on blood vessels, indicating anti-inflammatory and anti-adhesive effects. Importantly, these effects were seen at peak COHb levels of 6.5%.

3.1.3. Sickle Cell Disease Clinical Efficacy and Epidemiologic Studies

Data from three independent studies provide preliminary clinical evidence for the COmediated prevention of VOC in SCD patients. An early clinical study by Sirs first demonstrated that a 30 minute administration of inhaled CO to a SCD patient producing an estimated COHb concentration of 4%, resulted in a 60% reduction of sickled RBCs in the peripheral blood⁷⁵. Beutler and colleagues extended this work, demonstrating that the administration of inhaled CO to two SCD adult patients increased Cr⁵¹-tagged RBC survival, suggesting that the rheologic properties of sickle cells were favorably influenced¹⁰⁵. Importantly, although these two studies comprise a very limited number of patients, the Sirs study suggests the possibility of a relatively low minimum effective dose, which clearly will need to be studied in formal clinical trials. In addition, an epidemiologic study carried out in London showed a significantly lower hospital admission rate for VOC on days when higher ambient atmospheric levels of CO were recorded¹⁰⁶. Importantly, this significantly lower hospital admission rate was not related to the other five atmospheric pollutants that were investigated.

3.1.4. Safety Studies of CO

The clinical safety of CO in SCD depends on the target dose and COHb levels. As discussed above, a target dose resulting in COHb levels as low as 2% has been reported to be efficacious in SCD in preclinical studies, and a COHb of approximately 4% has been reported to significantly reduce sickled RBCs in a clinical experience, suggesting a target range of <10% COHb as appropriate to evaluate for efficacy.

At the time of writing, a review of clinicaltrials.gov show there are 8 completed and 10 on-going clinical studies testing a variety of CO-delivery modalities (but not oral) (Table 2). Thus 18 studies with a variety of sponsors have Phase 1 and Phase 2 clinical investigations under regulatory guidance, providing strong support for the perception of safety at targeted CO-Hb levels for Phase 1 and Phase 2 clinical investigation.

Table 2.

Completed Phase 1 or Phase 2 clinical studies^* supported by a number of sponsors and employing a variety of CO-administration mechanisms

Clinical Trial	Description/Indication	CO Product
Phase 1	Single Dose Safety (Normal Volunteers)	Inhaled gas107
Phase 1	Repeat Dose (10 Days) Safety (Normal Volunteers)	Inhaled gas107
Phase 1	Single Dose Safety (Normal Volunteers)	CORM (Sanguinate™)110
Phase 1b	Repeat Dose Safety (Sickle Cell Disease)	CORM (Sanguinate™)109
Phase 1b	Repeat Dose Safety (Sickle Cell Disease)	CORM (MP4CO)108
Phase 2	Chronic Obstructive Pulmonary Disorder	Inhaled gas111
Phase 2	Experimental Endotoxemia	Inhaled gas112
Phase 2	Mitochondrial Biogenesis (skeletal muscle)	Inhaled gas113

^*https://clinicaltrials.gov/

Two successful Phase 1 clinical trials in normal volunteers have been conducted with inhaled CO gas, demonstrating the safety and tolerability of CO up to a peak COHb level of 13.9%¹⁰⁷. In addition, two Phase 1b clinical safety studies conducted with CO using either intravenous MP4CO or Sanguinate^™, demonstrated safety and tolerability in SCD patients in all dose groups, with a COHb increase over baseline of 2% for MP4CO^108,109. The potential for therapeutic development of CO is demonstrated by four additional Phase 1 and 2 clinical studies with inhaled and Hb-delivered CO conducted in non-SCD indications^{110,111,112,113}.

Despite these supportive studies, the safety of CO at concentrations at or below COHb of 10% in SCD patients is based on limited data. Individuals with SCD may have underlying SCD-associated end-organ disease that may be negatively impacted by CO administration, which may result from less oxygen to tissues. It is also possible, but unlikely given the current data, that CO may actually increase HbS polymerization. Given the current limited information on the impact of CO in individuals with SCD, careful study in SCD patients is warranted.

4.0. CO and Neuronal Injury

CO has been shown to be of benefit in a number of animal models of organ ischemia, including brain, kidney, heart, liver, and gut ischemia. In SCD there are major concerns around cerebral ischemia, as cerebrovascular occlusive events (CVOE) in patients with SCD are common and are major clinical events that uniformly determine adverse outcomes¹¹⁴. These events can manifest across the age spectrum as overt strokes, but cerebral infarcts can also be silent, with the cumulative events over time resulting in chronic cerebral ischemia, cognitive impairment, and poor quality of life^115,116. Silent cerebral infarcts (SCI) documented by radiologic changes have been reported in 10% to 30% of patients with SCD¹¹⁶ and a prevalence of SCI of 53% has been documented in adults with SCD¹¹⁷. The prevalence of clinical stroke in SCD is reported to be as high as 24% and a crippling CVOE can occur as early as two years of age, with the highest incidence being within the first decade of life^118,114. The incidence of stroke has been demonstrated to be 10% per 36 months period in children with SCD who have abnormal transcranial Doppler ultrasound (TCD)¹¹⁹.

The current standard of care aimed at limiting stroke episodes includes annual TCD evaluations for all children with SCD. Those children showing elevated blood flow velocities indicative of cerebrovascular narrowing and children who have experienced a stroke are treated by regular (approximately monthly) transfusion of RBCs^114,116. RBC transfusions have been shown to substantially reduce the risk of stroke in SCD patients, but are associated with significant complications such as iron overload and alloimmunization. No other therapy, including hydroxyurea, has been demonstrated to be effective in the primary prevention of stroke in SCD patients, although in SCD patients previously impacted by a stroke, hydroxyurea combined with phlebotomy, when compared under clinical trial with transfusions and chelation therapy, showed non-inferiority¹²⁰.

4.1. The potential for HO-1 and CO to limit Cerebro-vascular Occlusive Episodes in Sickle Cell Disease

Neuroprotective strategies that aim to reduce neuronal damage are an important approach to improving CVOE outcomes in SCD. These strategies have been explored and primarily carried out in non-sickle cell animal models including stroke, autoimmune inflammation, and traumatic and non-traumatic brain injury. Traditional neuroprotective strategies target the acute ischemic event in hopes of lessening the severity of brain injury. More recent data suggest that post-ischemic inflammation¹²¹ also contributes to brain injury following stroke, providing a possible target for neuroprotective agents, even when delivered in a delayed time window. The potential involvement of the HO/CO pathway in CVOE may include both a prevention and treatment modality. First, in preventing VOCs across the vasculature in all organs, as is discussed above, CVOEs will also likely be prevented. Second, HO and CO have been shown to mitigate the sequelae of CVOEs that occur in a number of non-SCD animal models of neural injury, that to date have included models of cerebral infarction and cerebral inflammation, among others.

4.2. HO and CO in Cerebral Ischemia

Data from six studies from four independent laboratories^{69,122,123,,124,125,126} implicate HO-1 and CO in limiting non-SCD cerebral ischemia and suggest that targeting this pathway will offer a novel approach to the treatment of CVOE in SCD. Additional research provides support for these observations.

Panahian and colleagues first showed that HO-1 provides neuroprotection in cerebral ischemia¹²³. Transgenic mice overexpressing HO-1 in neurons had smaller infarcts and less cerebral edema compared to wild type mice; HO-1 was shown to impart anti-apoptotic and anti-oxidant actions in the brain mediated by inhibition of p53 protein nuclear localization and accompanied by decreases in iron staining and tissue lipid peroxidation. HO-1 was also necessary to protect the brain following N-methyl-D-aspartate (NMDA) injection, as Hmox1^−/− mice had significantly greater lesion volume than wild-type littermate controls. In vitro data corroborated these results: survival after NMDA exposure was reduced in Hmox1^−/− mouse neurons compared to wild-type neurons. This research substantiated a study demonstrating that overexpression of HO-1 using an adenoviral vector, attenuated brain damage after focal cerebral ischemia in rats¹²⁵.

A number of studies have extended this initial work on HO-1 by investigating the effects of exogenous CO administration. In a model of transient focal cerebral ischemia, inhaled CO at the time of reperfusion reduced infarct size at 48 hours to a similar degree as that observed in transgenic mice overexpressing HO-1⁶⁹. Moreover, CO treatment was associated with improved neurologic deficit scores at 24 and 48 hours. This observation was supported using an intravenous, PEGylated COHb in a transient focal cerebral ischemia model. COHb levels of 11% were associated with reduced infarct volume and better neurologic deficit scores at 24 hours compared to animals that received either no transfusion or transfusion with PEG-albumin control¹²². Wang and colleagues¹²⁴ extended the work by Zeynalov using a cerebral vascular occlusion model and inhaled CO in wild-type and Nrf2^−/− mice. Administration of 125 and 250 ppm inhaled CO to wild-type mice at the onset of reperfusion and extended over 20 hours significantly reduced cerebral infarct size and resulted in better behavioral scores at 7 days post-ischemia compared to untreated mice. CO administration resulted in Nrf2 dissociation from Keap1, Nrf2 nuclear translocation, increased binding of Nrf2 to Hmox1 DNA antioxidant response elements, and increased HO-1 expression. The neuroprotective effects of CO were completely abolished in Nrf2^−/− mice. Zhang evaluated the effects of CO in a 2-hour middle cerebral artery occlusion (MCAO) model in rats using CO delivered via two different types of cell free Hb¹²⁶. The study demonstrated decreased infarct size after COHb administration (with approximately 5% peak COHb levels) 3 days after MCAO as well as increased blood flow to the infarct border zone. Importantly, these studies demonstrate that administration of CO, using various delivery strategies, has similar and significant neuroprotective effects after acute ischemic stroke.

4.3. HO and CO in Cerebral Inflammation

Abundant data indicate that low CO levels modulate the immune response to limit detrimental autoimmune neuroinflammation. Following ischemic stroke there is a breakdown in the blood-brain barrier that allows cells of the systemic immune system to gain access to the ischemic brain^125,127,128. Antigens that are normally sequestered from the immune system may thus be encountered in the brain as well as in peripheral lymphoid organs. Indeed, by 24 hours after the induction of experimental cerebral ischemia, lymphocytes are found within the central nervous system (CNS) and CNS antigens (for example proteolipid protein) are being presented to the immune system in cervical lymph nodes¹²⁹. These observations suggest that an autoimmune response to brain antigens could develop after stroke. In fact, studies in the 1970’s showed that stroke survivors had more robust cellular immune responses towards myelin basic protein (MBP) and “brain antigens” than patients with multiple sclerosis and acute inflammatory demyelinating polyneuropathy^130,131,132. More recent data demonstrate lymphocyte responses to MBP and antibody responses to neurofilament and portions of the N-methyl-D-aspartate receptor among stroke survivors^133,134,135. These immune responses have been largely considered an epiphenomenon of cerebral tissue injury.

In experimental and clinical studies both animals and patients who develop Th1 immune responses in the brain, suffer worse stroke outcomes^{121,136,137,138}. It has also been shown that Th17 responses negatively impact stroke outcome^{139,140,141,142,143,144}. These data suggest that the post-ischemic immune responses affect outcome. It has been shown that induction of MBP-specific Treg responses prior to ischemia prevents Th1 responses to MBP and improves outcome^137,145,146. Therapies that can be initiated in a delayed time frame after stroke onset, especially when tissue plasminogen activator is not indicated, including CVOE in SCD, are sorely needed and the modulation of the immune response is an intervention that has potential for improving outcome.

In addition to the potential neuroprotective properties of CO, there are abundant data suggesting that CO modulates immune responses and limits detrimental autoimmune inflammation. CO has multiple effects on the immune system, which in sum, are associated with decreases in inflammation and detrimental Th1 and Th17 responses. Importantly, as pertains to the systemic encounter of CNS antigens after stroke, CO inhibits the capacity of dendritic cells to present soluble antigens to T cells¹⁴⁷, which may, in part, be related to a block in the up-regulation of TLR4¹⁴⁸. CORMs increased expression of HO-1, reducing the secretion of IFN-γ and IL-17 from T cells¹⁴⁹. CO also blocks the production of IL-2 in naïve T cells¹⁵⁰, thereby inhibiting T cell proliferation. These effects are well demonstrated in models of experimental allergic encephalomyelitis (EAE). In the EAE model, these effects were associated with a decrease in CD8 T cell accumulation in the CNS, a decrease in MHCII expression in the CNS, inhibition of Th cell proliferation and effector cell function, and a decrease in myelin-reactive T cell proliferation¹⁴⁶. Of note, the effects of CO appear to be directed towards inhibition of Th1 and Th17 responses as opposed to up-regulation of Treg responses as the secretion of Treg cytokines was not affected¹⁴⁹. Given the inhibitory effects of CO on the development of the Th1 and Th17 immune responses, and the fact that it can halt ongoing EAE, CO appears poised to be an effective immune modulator in post-ischemic settings including SCD-associated CVOE.

4.4. HO and CO in Brain Injury

A number of studies in various in vitro and in vivo models of cerebral injury demonstrate that CO provides neuroprotection, providing support for the use of CO in the treatment of SCD-associated CVOE’s. Four studies focused on in vitro cellular injury models in neurons, astrocytes and microglia demonstrated that CO prevents apoptosis^{151,152,153,154}. Vieira and colleagues demonstrated that preconditioning with CO gas inhibits apoptosis in murine cerebellar granule cells treated with either glutamate or tert-butylhydroperoxide (t-BHP) to mimic excitotoxicity and oxidative stress, respectively. CO treatment induced de novo protein synthesis in these cells. This study also demonstrated that CO is a critical component of the anti-apoptotic role of HO-1, as it was found that exogenous CO administration could reverse the lack of HO-1 in an oxidative stress model. Queiroga et al, found that the anti-apoptotic properties of CO extended to astrocytes where it prevented apoptosis induced by either t-BHP or diamide, a thiol crosslinking agent. CO inhibited increases in mitochondrial membrane permeability, a key event in the intrinsic apoptotic pathway. Almeida and colleagues confirmed these effects showing that CO stimulates oxidative phosphorylation and improves metabolism, in part by preventing caspase-3 activation, mitochondrial depolarization, and changes in plasma membrane permeability ¹⁵³. Additionally, Bani-Hani and colleagues showed that CO treatment using a CORM attenuated thrombin-induced activation of BV-2 microglia and reduced lactate dehydrogenase, a marker for cell breakdown¹⁵⁴. Moreover, inhibition of HO-1 did not change the activity of CO, suggesting that CO is directly responsible for the observed neuroprotective effects. Finally, Yabluchansky et al utilized CORM-3 in a rat model of hemorrhagic stroke to demonstrate that CO reduces brain damage via a modification of the inflammatory response¹⁵⁵.

Three in vivo studies in models of cerebral injury have been described^156,157,158. The first report demonstrated the cerebro-protective effects of CO, delivered using a CORM, in a seizure-induced neonatal vascular injury model in piglets¹⁵⁶. In this study, CO pretreatment, in contrast to air-treated control animals, prevented the loss of postictal cerebrovascular reactivity (indicative of cerebrovascular injury) to selective physiologically relevant vasodilators, both endothelium-dependent and endothelium-independent. Moreover, CO was shown to elicit vasodilator properties on the cerebral circulation, protecting the brain from cerebrovascular injury caused by seizures. The second study extended this finding, demonstrating that inhaled CO provided cerebral cytoprotection in a model of deep hypothermic circulatory arrest in piglets. CO pre-treatment completely abrogated cell death in the neocortex and hippocampus compared to air-treated control animals. CO-treated animals, in contrast to air-treated controls, maintained normal oxygen/glucose and lactate/glucose indices and corresponding cerebral sinus pressures with no change in systemic hemodynamics or metabolic intermediates suggesting that CO was acting, in part, to preserve cellular bioenergetics¹⁵⁷. The third study demonstrated that CO, whether inhaled or delivered as a CORM, reduced cell death and ameliorated the progression of neurological deficits in a mouse model of traumatic brain injury (TBI). Treatment with CO after TBI significantly reduced pericyte death and improved functional recovery as compared to vehicle-treated control animals. In addition, CO-treated animals showed higher levels of phosphorylated neural nitric oxide synthase within neural stem cells. This study suggests that CO treatment prevents pericyte death and promotes neurogenesis¹⁵⁸. Collectively, these data suggest that chronic administration of CO, in addition to preventing VOCs, could also mitigate the morbidities caused by specific organ ischemia, including the brain in SCD patients.

Despite the accumulating and abundant encouraging preclinical data that support the potential impact of HO and CO on improving SCD-CVOE outcomes, studies in human subjects with SCD need to proceed with caution given the host of SCD-associated organ morbidities. For example, cerebral blood flow reserve in SCD patients is often reduced and, given the possible reduction in cerebral oxygen reserve that may occur under raised CO-Hb levels, carefully constructed adequately controlled studies will be required in order to determine the risk/benefit ratio with CO in these patients.

5.0. Methods of CO Administration

Although there is substantial support for the potential therapeutic use of CO for the treatment and prevention of VOCs in SCD, going back as far as 1963⁷⁵, this molecule has not yet advanced to licensure as a therapeutic agent for SCD. One reason for this is the difficulty in administering CO in a chronic, environmentally safe, appropriately dosed and controlled manner. Three approaches to CO administration have been tested, including inhaled CO, CORMs, and an oral formulation of CO. Both CORMs and an oral formulation of CO are currently in development for the treatment and prevention of VOCs, respectively.

5.1. Inhaled CO

Although inhaled CO is the most obvious form of CO administration, this presents substantial challenges for further development, especially for chronic use that would be required for the prevention of VOCs. CO gas for inhalation is readily and commercially available and this route of administration has been used to generate the vast majority of preclinical and clinical data regarding CO and SCD. Inhaled CO, however, is difficult to accurately dose given the variability of patient ventilation, mask and cannula compliance, and pulmonary absorption associated with the variable capacity of the lung. Inhaled CO also has environmental safety concerns, as it requires the presence of large amounts of CO gas in cylinders. In addition, the chronic use of CO gas in an outpatient or home environment, as would be necessary for the use in the prevention of SCD VOCs, has significant logistical and compliance hurdles for patients, and potential exposure risks to others in close proximity. Despite the above challenges, the administration of inhaled CO is being utilized in a number of ongoing non-SCD targeted Phase 1 and Phase 2 clinical studies, primarily in acute indications (www.clinicaltrials.gov).

5.2. CO Releasing Molecules (CORMs)

CORMs, also termed CO prodrugs, where CO is bound to carrier molecules, including both small carrier molecules and large carrier molecules, have been explored as an alternative to inhaled CO. Small molecule CORMs have typically used transition metals such as molybdenum and ruthenium as well as boric acid/sodium borate as carriers for CO, first described in the pioneering work led by Motterlini and Foresti¹⁵⁹. Although still under development, these metal-based CORMs present challenges for further development, including potential dose-limiting toxicity of the carrier molecule^160,161. Based on these concerns, further small molecule CORM development has focused on encapsulated metal-based CORMs¹⁶², where the CORM is bound to a macromolecular carrier. Organic CORMs, where a small organic molecule releases CO under physiologic conditions¹⁶⁰, and hybrid CO-RMs that are tethered to molecules that are known to induce HO-1 are also under evaluation¹⁶³. These approaches as well as other, similarly innovative approaches are being explored in the development of CORMs^155,163, with particular focus on minimizing toxicity and permitting dosing at levels likely to permit the evaluation of efficacy. Given the potential dose-limiting toxicity of the currently reported carrier molecules, if the CO release and stability prove to be appropriate, these small molecule CORMs would likely be targeted for the treatment of VOCs rather than for the chronic use necessary for the prevention of VOCs.

Large molecule CORMs, typically with PEGylated Hb carriers intended for intravenous administration, are being developed, and two such products (MP4CO and Sanguinate^™) have advanced into clinical evaluation. The Phase 1b clinical trial results of both MP4CO (no longer in development) and Sanguinate have been reported^108,109 and a Phase 2 clinical trial with Sanguinate in the treatment of VOCs is ongoing (NCT02411708). However, in the case of these Hb-based compounds, the announced intent of the companies as well as the documented toxicity of cell-free Hb suggest that these molecules, should they succeed in clinical development, would be limited to acute use for treatment of VOC rather than chronic use for the prevention of VOC¹⁶⁴. As an added issue, when heme is released from the exogenously administered hemoglobin, substantial acute changes that could result in toxicity as reported with hemoglobin solutions may occur^164,165, and the potential for iron overload also exists if used for chronic treatment. Regardless, as new small and large molecule CORMs with a spectrum of chemical properties, reactivity and toxicology are being developed, these may result in novel innovative agents for this highly challenging disease.

5.3. Oral Formulation of CO

The potential to orally administer CO in a liquid formulation has been pursued to avoid the barriers to development of inhaled CO and CORMs detailed above^166,167. CO is absorbed as well across gastrointestinal surfaces as across the pulmonary epithelium¹⁶⁸. The development of a CO-containing liquid comprised of Generally Recognized as Safe (GRAS) excipients (HBI-002) was recently documented in abstract form and presented at research meetings^166,169,170. The bioavailability of CO through oral administration of HBI-002 has been demonstrated in rodents, where an increase in COHb levels of up to 10% was achieved. Oral administration of CO avoids many of the issues associated with gas administration. As compared with inhaled CO, it allows more precise dosing and avoids environmental safety issues associated with compressed gas cylinders. In addition, it avoids the challenges associated with the reported CORMs, as no releasing molecule or reaction to release CO is required. Oral delivery of the liquid formulation provides a platform for feasibility and compliance given the preference for oral dosing by patients, especially for chronic administration and for the administration of a therapeutic outside of hospital settings.

6.0. Conclusions

CO is physiologically generated during the metabolism of heme by the HO enzymes and is measurable in blood, typically correlating with the stress response. There is much therapeutic potential for using CO in higher than physiological doses, but at levels below known toxicity to treat SCD patients. This has been known for over fifty years. Substantial preclinical and clinical data with CO has been generated, providing compelling support for its use as a potential therapeutic. More recent data underlying the therapeutic mechanisms of CO, including in SCD, have become available and these effects have been demonstrated in a plethora of in vitro preclinical and animal studies including multiple SCD mouse models. The data indicate that CO has key signaling impact on a host of metallo-enzymes as well as key genes that, in sum, result in modulation of inflammatory, oxidant and apoptotic effects as well as vasomotor tone and adhesive properties resulting in tissue preservation of vascular flow. However the role of CO as an anti-polymerization of HbS agent, while well-demonstrated, is not fully apparent at the low COHb levels targeted by these therapeutic approaches. Further work is needed to explore the known potential of CO to melt HbS tactoids and prevent polymerization, particularly within the pharmacokinetic dosing spectrum envisaged for this therapy and elucidating how this interacts with the signaling actions of CO. Further research is also indicated and hopefully forthcoming, to carefully elucidate the safety and benefits of this potential gasotransmitter therapy across the age spectrum of patients impacted by the host of patho-physiological complications of this devastating disease.

In summary, CO has important anti-vaso-occlusive and immunomodulatory effects that, together, provide cells and tissues with powerful protective and pro-survival systems in the setting of SCD. In particular, considerable scientific data provide evidence for a beneficial impact of CO on cerebrovascular complications, suggesting that in SCD, CO could potentially limit these highly problematic neurologic outcomes.