Nitrite decreases sickle hemoglobin polymerization in vitro independently of methemoglobin formation

Abstract

The root cause of sickle cell disease (SCD) is the polymerization of sickle hemoglobin (HbS) leading to sickling of red blood cells (RBC). Earlier studies showed that in patients with SCD, high-dose nitrite inhibited sickling, an effect originally attributed to HbS oxidation to methemoglobin-S even though the anti-sickling effect did not correlate with methemoglobin-S levels. Here, we examined the effects of nitrite on HbS polymerization and on methemoglobin formation in a SCD mouse model. In vitro, at concentrations higher than physiologic (>1μM), nitrite increased the delay time for polymerization of deoxygenated HbS independently of methemoglobin-S formation, which only occurred at much higher concentrations (>300μM). In vitro, higher nitrite concentrations oxidized 100% of normal hemoglobin A (HbA), but only 70% of HbS. Dimethyl adipimidate, an anti-polymerization agent, increased the fraction of HbS oxidized by nitrite to 82%, suggesting that polymerized HbS partially contributed to the oxidation-resistant fraction of HbS. At low concentrations (10μM-1mM), nitrite did not increase the formation of reactive oxygen species but at high concentrations (10mM) it decreased sickle RBC viability. In SCD mice, 4-week administration of nitrite yielded no significant changes in methemoglobin or nitrite levels in plasma and RBC, however, it further increased leukocytosis. Overall, these data suggest that nitrite at supra-physiologic concentrations has anti-polymerization properties in vitro and that leukocytosis is a potential nitrite toxicity in vivo. Therefore, to determine whether the anti-polymerization effect of nitrite observed in vitro underlies the decreases in sickling observed in patients with SCD, administration of higher nitrite doses is required.

Introduction

Sickle cell disease (SCD), one of the most common hemoglobinopathies, affects approximately 100,000 Americans and millions of individuals of predominantly African heritage worldwide (Hassell, 2010; Piel et al. 2017). The disease results from a mutation in the β-globin gene leading to the formation of an abnormal hemoglobin variant, sickle hemoglobin (HbS). When deoxygenated, HbS polymerizes forming stiff fibers, which distort red blood cells (RBCs) into a ‘sickle’ shape. Consequently, patients with SCD develop hemolytic anemia, recurrent episodes of vaso-occlusion and pain crises, chronic vascular endothelial injury and inflammation, multisystem organ damage, and have significantly reduced life expectancy (Piel et al., 2017). Polymerization of HbS is the central event in the pathobiology of SCD and therapies aimed at decreasing the formation of HbS fibers could decrease sickling and benefit SCD patients (Eaton and Bunn, 2017).

Over 70 years ago, Linus Pauling’s group showed that when RBCs from patients with SCD were exposed to nitrite in vitro, methemoglobin [oxidized hemoglobin with a heme iron configuration changed from ferrous (Fe²⁺) to ferric (Fe³⁺)] was formed and sickling was prevented (Itano, 1950). A decade later, Beutler and colleagues showed that ingestion of high-dose nitrite decreased sickling, increased RBC survival, and increased methemoglobin levels in patients with SCD (Beutler, 1961). At the time, the investigators hypothesized that by forming methemoglobin S, nitrite reduces the concentration of deoxygenated HbS in RBCs thereby decreasing sickling (Beutler, 1961). However, the anti-sickling effects of nitrite did not correlate with methemoglobin levels and despite its desirable effects, nitrite was not pursued as a therapy for SCD (Beutler, 1961).

Recent preclinical studies have led to renewed interest in inorganic nitrite as a potential therapy for SCD. Nitrite has been shown to inhibit platelet activation (Akrawinthawong et al., 2014) , aggregation (Srihirun et al., 2012; Wajih et al., 2017) and thrombus formation (Park et al., 2014), improve RBC deformability, inhibit RBC adhesion to endothelial cells, and decrease hemolysis in SCD mice (Wajih et al., 2017). These salutary effects of nitrite in cardiovascular pathologies and SCD have been attributed to its reduction into nitric oxide (NO). Relevant to SCD is the fact that nitrite can be reduced into NO in hypoxic environments where deoxygenated hemoglobin predominates and where NO release could promote vasodilation, reduce platelet activation and inflammatory response, all effects that could prove beneficial in SCD (Cosby et al., 2003; Shiva et al., 2007; Amdahl et al., 2019). However, nitrite can also have potentially harmful effects including generating methemoglobin, increasing ferric heme (hemin) release, and altering the redox state of RBCs (Zavodnik et al., 1999; May et al., 2000; Belcher et al., 2014; Ansari et al., 2015; Ansari and Mahmood, 2016). Therefore, if nitrite is to be pursued as a therapy for SCD, a better understanding of its mechanisms of action and of its toxicity is needed. We investigated the effects of nitrite on the kinetics of HbS polymerization in vitro. In addition, we determined the dose-response relationship between nitrite concentration and methemoglobin formation, hemin release, and reactive oxygen species formation in vitro using blood from SCD mice. Further, in vivo, we examined the effects of long-term nitrite administration on the hematologic profile of SCD mice.

Materials and Methods

The Supplemental File provides additional details of methods used in this investigation.

SCD mouse model

The NIH Clinical Center Animal Care and Use Committee approved the experimental protocol. We studied the humanized Townes SCD mouse model (Wu et al., 2006; Hanna et al., 2007; Khaibullina et al., 2015; Wang et al., 2016), which expresses human and no mouse hemoglobin. Homozygous Townes mice (HbSS) express human hemoglobin S and control Townes (HbAA) express human hemoglobin A (Wu et al., 2006; Hanna et al., 2007).

Hemoglobin S polymerization assay:

We measured the delay time for HbS polymerization using a high phosphate buffer model system (Adachi and Asakura, 1979). Briefly, RBCs were washed three times with saline and lysed with five volumes of lysis buffer (5mM potassium phosphate buffer, 0.5mM EDTA, pH = 7.34). Lysis buffer was removed by filtering 500μl of hemoglobin solution (14,000g, 15min) through pre-washed centrifugal filters (30KDa cutoff, Millipore-Sigma). The recovered hemoglobin was diluted with 1.8M phosphate buffer (pH = 7.3) to [hemoglobin] = 0.2243 – 0.2315 g/dL (median: 0.2261) calculated using published extinction coefficients (60 liters / (mM⁻¹ x cm⁻¹), 577 nm absorbance, tetrameric form, 64,500g molecular weight) (Adachi and Asakura, 1979). Diluted hemoglobin samples were aliquoted, nitrite was added (from 10x stock solutions, also prepared in 1.8M buffer) and aliquots were incubated at 37°C for 2hrs. We used dithionite as the deoxygenation agent for the polymerization assay as it has been reported before (Adachi and Asakura, 1979; Adachi and Asakura, 1982; He and Russell, 2002). Polymerization assays (in duplicates) were carried out in pre-chilled 96 well plate with 20μl sodium dithionite solution (100mM, freshly prepared in deoxygenated 1.8M buffer with careful tube inversions) added to wells followed by 200μl of nitrite-treated hemoglobin solutions. Final hemoglobin levels during the polymerization phase were 20% diluted from above values. Wells were immediately isolated from room air with 50μl TW oil (Inland Vacuum Industries). Sealed plates were incubated at 4°C for 30min and transferred into a plate reader pre-heated to 30°C. Absorbance (700nm) was acquired every 10 seconds for 50 minutes. The polymerization delay time was annotated as the time point in which its absorbance derivative remained equal or above zero for at least 5 successive points.

Blood collection and preparation

Blood was collected using a protocol preventing nitrite contamination (Almeida et al., 2017) and a complete blood count was obtained (Hemavet, Drew Corporation). Blood was diluted with deoxygenated phosphate buffered saline (PBS, pH = 7.4) prepared by bubbling nitrogen gas for 20 min; these procedures maintain the integrity of RBC membranes preserving intracellular milieu. All experiments were carried out with 10% final blood content (hemoglobin=0.6-0.8 g/dL) on room air (21% oxygen). The anti-polymerization agent dimethyl adipimidate dihydrochloride (DMA) (cat #285625-5G, Millipore Sigma) was used according to a protocol shown to block 95% of deoxygenation-induced RBC sickling (Lubin et al., 1975; Pennathur-Das et al., 1984). Blood samples were aliquoted and treated with nitrite and vehicle aliquots received equal volumes without added drugs. Aliquots were incubated for 2hrs (37°C) for complete nitrite metabolism (Almeida et al., 2020).

Methemoglobin assay:

Methemoglobin was assayed using the method described by Cruz-Landeira and colleagues (Cruz-Landeira et al., 2002). For each experiment, the effect of nitrite on methemoglobin content was calculated by fitting the nitrite dose-response curve with the Hill equation (4 parameters, SigmaPlot v14), which considers that methemoglobin content is not zero (undetectable) at baseline. We determine the Emax (relative maximum response), nH (Hill coefficient) and EC₅₀ (the concentration that produces 50% of the respective maximum response). Values for these parameters were averaged across samples and a Hill equation was also fitted to the averaged result for each treatment condition. This method cannot distinguish between methemoglobin and nitrosyl-methemoglobin. However, the formation of nitrosyl-methemoglobin is small relative to formation of methemoglobin when hemoglobin is incubated with nitrite (Keszler et al., 2008).

Hemin assay:

Because heme and hemin cannot be distinguished chemically, hemin (free ferric heme) also refers to “free heme” (free ferrous heme) (Ghosh et al., 2013; Belcher et al., 2014; Vallelian et al., 2018). Heme refers to total heme (protein-bound heme plus its free form). A commercial kit selective for hemin over total heme was used to assay its intra-RBC levels (Abcam, cat# ab65332, Abcam customer service, personal communication) removing the need for sample deproteination. Plasma was separated by centrifugation and washed RBCs were diluted in deoxygenated PBS to hemoglobin=0.8 g/dL and incubated with nitrite for 2h at 37°C. Aliquots were further diluted with double-distilled water to fit the standard curve suggested by the manufacturer.

Fast RBC purification method:

We isolated RBC adapting the method described by Beutler and colleagues (Beutler et al., 1976; Beutler and Gelbart, 1986) which is based on the mechanical ability of RBCs to deform and pass through a cellulose filter while white blood cells (WBCs) are retained by their relative rigidity. Briefly, a 1:1 mixture (0.5g each) of cellulose type 50 (Sigma cat# S5504) and type 101 (Sigma cat# S6790) was mixed with 5ml of ice-cold PBS and was poured into an upside-down 10ml syringe and its plunger was used to gently compact this mixture into a cake reaching ≈ 2.5ml syringe mark. Whole blood was layered onto the top of the cake followed by 7ml of ice-cold PBS. The plunger was reinserted and used to press the solution through the cellulose cake (≈1 drop/2 sec). Cells were precipitated from the filtrate by centrifugation (1000g, 10min, 4°C) and washed 2x with Hank’s balanced salt solution with 1.2 mM calcium and 0.8mM magnesium added.

Reactive oxygen species (ROS) assay:

Purified RBCs were resuspended in Hank’s balanced salt solution to a hemoglobin concentration of 0.8g/dL and distributed into 450μl aliquots. Nitrite was added (from 10x stocks; 50μl/aliquot) and aliquots were incubated at 37°C for 15min to initiate nitrite metabolism. The ROS dye chloromethyl 2′,7′-dichlorodihydrofluorescein diacetate (1μl from 500μM stock, final Dimethylsulfoxide = 0.2%) (Thermo Fisher) was added to each aliquot and 200μl were distributed into duplicate wells in a black 96 well plate. Plates were sealed and incubated in a plate reader pre-heated to 37°C for 3 hours. Fluorescence was acquired every minute (Ex=495 and Em=520nm). Data were analyzed at 150min to document that the dye response did not saturate. Given that there was a significant loss of RBCs during the purification process, each sample examined consisted of purified RBCs pooled from 2 animals. Twelve animals were then needed for the six samples in each genotype group, HbAA or HbSS.

Adenosine triphosphate (ATP) assay:

Purified RBCs resuspended in Hank’s balanced salt solution hemoglobin = 0.8g/dL) were incubated for 2hrs (37°C) with indicated drugs (from 10x stocks prepared in Hank’s balanced salt solution). ATP was assayed with a commercial kit (Thermo Fisher, cat#: A22066).

Nitrite and nitrate supplementation:

Animals weight was measured before and after a 4-week administration of distilled water (vehicle) or sodium nitrite (1g/L), or sodium nitrate (4 g/L). Solution bottles were replaced twice a week and drinking volumes recorded. After 4 weeks of treatment, blood was collected as described. 3-isobutyl-1-methylxanthine, a phosphodiesterase inhibitor, (0.5mM, final Dimethylsulfoxide = 0.5%) and protease inhibitors cocktail (Complete EDTA-free, 25x stock, Roche) were added (final blood dilution = 4.5%). Blood was kept on ice at all times to preserve nitrite levels (Almeida et al., 2017) and a complete blood count was obtained. Plasma was separated from RBCs by centrifugation (1000g, 10min, 4°C), transferred into new microcentrifuge tube and re-spun in order to remove any remaining RBC. RBCs were washed 2x with 700μl deoxygenated PBS (with added 3-isobutyl-1-methylxanthine and protease inhibitors) to remove any contaminating plasma. All samples were frozen at −80°C until processing.

Nitrite and nitrate assays:

We followed procedures for sample processing and selective nitrite assay using the ascorbic and acetic acids chemiluminescence method as described (Almeida et al., 2015; Almeida et al., 2017). Samples (plasma and RBCs) were deproteinated with chloroform to avoid methanol-induced nitrite decomposition (Almeida et al., 2015; Almeida et al., 2017). For RBC nitrate assays, washed RBCs were diluted in distilled water to 5-6 g/dL hemoglobin. Samples for nitrate assays (RBC and plasma) were deproteinated with filtration through a pre-rinsed 10KDa cutoff filter (0.5ml); preliminary experiments indicated that pre-rinsed filters were nitrate-free. Nitrate was assayed (in triplicates) with a Greiss-based fluorometric method following manufacturer’s protocol (Cayman Chemicals). Intra-RBC nitrite and nitrate levels are reported either adjusted to hemoglobin content (nmols/g of hemoglobin) or as concentration (μM) by considering the amount of hemoglobin inside a single RBC (mean corpuscular hemoglobin) and the single RBC volume (mean corpuscular volume).

Statistical analysis

Data recorded at one time point, which were not normally distributed, were analyzed using Kruskal-Wallis or Chi-square tests. Repeated measures data (nitrite dose-responses in vitro) were analyzed using a repeated-measures ANOVA with interaction terms (genotype by treatment). Model fit diagnostics were examined to ensure that model assumptions were satisfied. Dependent variables measurements were transformed to the natural logarithmic scale to meet model assumptions when appropriate. All p-values are reported unadjusted for multiple comparisons (Rothman, 2014; Streiner, 2015), thus the reader should take this under consideration when interpreting reported p-values, particularly those close to 0.05. All analyses were conducted using SigmaStat 14.0 (Systat Software).

Results

Nitrite alters polymerization kinetics in sickle cell disease mice

We evaluated the effect of nitrite on polymerization kinetics measuring the delay time (latency period) for the onset of HbS polymerization. After a 2-hour incubation, increasing doses of nitrite yielded prolongation in the delay times for HbS polymerization (p<0.001, for overall effect of nitrite, Fig. 1A and 1B). Compared with vehicle, nitrite prolonged delay times at all concentrations examined (1μM, p=0.046, 3μM, p=0.002, 10μM, p=0.001, 30μM, p<0.001, and 100μM, p<0.001, Fig. 1A and 1B), indicating that at concentrations equal to or above 1μM, nitrite had significant anti-polymerization effects. At nitrite concentrations equal to or greater than 300μM, delay times were greater than 50 min (Fig. 1A).

Figure 1. Nitrite alters polymerization kinetics in sickle cell mice.

A. Curves show mean absorbance traces (n=7) for each nitrite concentration as indicated (vehicle, 1, 10, 100 and 300μM). The vertical dotted lines indicate the latency period for the onset of polymerization (delay time) on each concentration curve. In the presence of 300μM nitrite, the delay time for polymerization was greater than 50 min indicating lack of HbS polymerization. We note that polymerization recordings in the presence of nitrite showed a progressive decrease in initial absorbance levels (t=0). We did not attempt to quantify such effect (if any), but it may suggest that nitrite also displays a depolymerization effect. B. Mean ± standard error (N=7) delay times in the presence of increasing nitrite concentrations (1μM – 100μM). For nitrite concentrations greater than 300μM, the delay time was greater than 50min and is not shown. The right y-axis shows the delay times at each nitrite concentration relative to that recorded with vehicle. Nitrite increased delay times for polymerization (p<0.001, for overall treatment effect). Post hoc tests revealed that compared with vehicle, at all concentrations studied, nitrite prolonged the delay time for hemoglobin S polymerization (1μM, p=0.046, 3μM, p=0.002, 10μM, p=0.001, 30μM, p<0.001, and 100μM, p<0.001. Veh represents vehicle.

The efficacy of nitrite in forming methemoglobin is altered in sickle cell blood

We next examined a possible mechanism underlying the anti-polymerization effect of nitrite testing the hypothesis that, by forming methemoglobin, nitrite could decrease intraerythrocyte concentrations of deoxygenated HbS thereby decreasing polymerization. At baseline conditions [vehicle, room air (21% oxygen)], methemoglobin levels in HbSS and HbAA mouse blood were similar (9.80±0.63% and 8.65±0.42, mean ± SEM, HbSS and HbAA respectively, p=0.139, Fig. 2A). At the same conditions, nitrite concentrations of up to 100μM, yielded no significant increases in methemoglobin content in HbAA or HbSS blood (Fig. 2A), even though at these lower concentrations, nitrite increased HbS polymerization delay times (Fig. 1). Conversely at 21% oxygen, increases in methemoglobin formation were observed only at nitrite concentrations of 300μM or greater (Fig. 2A).

Figure 2. Hemoglobin S polymerization affects the pharmacodynamics of nitrite-induced methemoglobin formation in HbSS blood.

Data are shown as mean ± standard error and the curves represent the best fit for the data. A. On room air, nitrite differentially induces methemoglobin formation in HbAA and HbSS blood. In a dose-dependent manner, nitrite oxidized 100% of the hemoglobin in HbAA blood but only a fraction of hemoglobin in HbSS blood [p<0.001, comparing HbAA and HbSS relative maximum response, Emax)]. However, the nitrite concentration that produces 50% of the maximum response (EC₅₀) of methemoglobin formation in HbSS and HbAA were similar (p=0.433). These data indicate that nitrite is equally potent (EC₅₀) to induce methemoglobin in both genotypes but is more efficacious (Emax) in HbAA compared with HbSS. R² = 0.991 and 0.996 for HbAA and HbSS curve fits, respectively. HbAA, N=5 and HbSS, N=6. B. Dimethyl adipimidate dihydrochloride (DMA) treatment increased nitrite-induced methemoglobin formation in HbSS blood from 70% to 82%, while a trend, this increase did not reach statistical significance (p=0.077). With DMA treatment, there was a small change in EC₅₀ (p=0.039, HbSS compared with HbSS+DMA) indicating that with reduction in polymerization, nitrite is more potent in inducing methemoglobin-S formation. R² = 0.9923 (HbAA+DMA), 0.9988 (HbSS+DMA) and 0.9957 (HbSS+Veh) for curve fitting. N=6 per genotype/treatment.

We then examined the pharmacodynamics of nitrite by measuring its efficacy [relative maximum effect (Emax)] and potency [concentration producing 50% of the respective maximum effect (EC₅₀)] in forming methemoglobin in HbAA and HbSS blood. Surprisingly, at the highest concentration tested (10mM), the Emax for nitrite-induced methemoglobin formation was lower in HbSS than in HbAA blood (p<0.001, Fig. 2A, supplemental Table 1, supplemental Fig. 1), indicating that nitrite was less efficacious in oxidizing hemoglobin in HbSS than in HbAA blood. In HbAA, higher nitrite concentrations (3-10 mM) formed 100% methemoglobin, whereas in HbSS blood the highest concentration (10mM) only formed approximately 70% of methemoglobin (Fig. 2A, supplemental Table 1, supplemental Fig. 1). The relative EC₅₀ for nitrite-induced methemoglobin formation in HbSS and HbAA were similar (p=0.433, Fig. 2A, supplemental Table 1) indicating that nitrite was equally potent in forming methemoglobin in both genotypes. Overall, in these experimental conditions (room air), nitrite had high efficacy, acting as a full agonist in HbAA, whereas it had lower efficacy, acting as a partial agonist for methemoglobin formation, in HbSS blood.

We next tested the hypothesis that polymerized HbS, contributes to the fraction of HbS resistant to nitrite-induced oxidation to methemoglobin. We treated HbAA and HbSS blood with DMA, an inhibitor of HbS polymerization. Addition of DMA did not alter Emax or EC₅₀ (Supplemental Fig. 2 and Supplemental Table 1) of nitrite-induced methemoglobin formation in HbAA blood, indicating that DMA was not toxic to RBCs. In contrast, in HbSS blood, DMA increased nitrite-induced methemoglobin formation from 70% to 82%, an effect that did not reach statistical significance [(p=0.077, Emax HbSS vs HbSS(+DMA), Fig. 2B and supplemental Table 1)]. DMA also decreased the EC₅₀ of nitrite (p=0.039, EC₅₀ HbSS vs HbSS(+DMA), Fig. 2B and supplemental Table 1) indicating that inhibition of polymerization made nitrite more potent in forming methemoglobin-S in HbSS blood. Together, these DMA-induced changes suggest that polymerized hemoglobin partially contributes to the fraction of HbS that is resistant to nitrite-induced methemoglobin formation in HbSS blood.

We also examined whether differences in antioxidant capacity could explain the discrepancy in nitrite-induced methemoglobin levels. In HbAA and HbSS blood, NADH, which can activate methemoglobin reductases or ascorbic acid, which increases RBC anti-oxidant defenses, did not significantly alter EC₅₀ or Emax for nitrite-induced methemoglobin formation (supplemental Fig. 3, supplemental table 1) in HbAA or HbSS blood.

Nitrite-induced changes in hemin release, redox state, and RBC viability

As methemoglobin formation is associated with hemin release, we measured the effect of nitrite on intra-RBC free hemin levels (Fig. 3A, B). At baseline conditions (vehicle, 21% oxygen) hemin levels in HbSS were approximately 30-fold higher than in HbAA RBCs (p=0.002, Fig. 3A). Given this difference in baseline levels, we show the data as fold-change from vehicle. Nitrite yielded concentration-dependent increases in intra-RBC hemin levels in both HbAA and HbSS RBCs compared with respective vehicle (Fig. 3B). In HbAA and HbSS RBCs, the relative maximum change in nitrite-induced hemin compared with respective baseline levels were similar (Emax approximately 2.2-fold-change from vehicle, p=0.877, Fig. 3B and supplemental Table 2). However, nitrite was more potent in increasing intra-RBC hemin levels in HbSS (EC₅₀, 0.41 ± 0.05 mM) than in HbAA (EC₅₀, 0.8±0.13 mM) RBCs (p=0.004, Fig. 3B and supplemental Table 2). Significant increases in hemin release in HbSS samples only occurred at nitrite concentrations >300μM. These nitrite concentration >300μM were equivalent to those that generated significant increases in methemoglobin (Fig. 2) but were significantly higher than concentrations that reduced HbS polymerization without forming methemoglobin (Fig. 1).

Figure 3. Pro-oxidant profile of nitrite.

In panels A, C, and E, box plots show median and first and third quartiles and whiskers the 10th and 90^th percentiles. In panel B data are shown as mean ± standard error and the curve represents the best fit for the data. In panels D and F, data are shown as mean ± standard error and dotted curves were inserted for clarity purposes only and do not represent fitted curves. A. Hemin content in HbSS RBCs was ≈30-fold higher than in HbAA (p=0.002, N=6 per genotype). B. Nitrite was equally efficacious (relative maximum effect, Emax) at producing hemin in HbAA and HbSS (p=0.877). The nitrite concentration producing 50% of the respective maximum effect (EC₅₀) was lower in HbSS compared with HbAA blood indicating that nitrite was more potent at producing hemin in HbSS (p=0.004). R² = 0.9940 and 0.9948 for curve fitting of HbAA and HbSS data points, respectively. N=6 for each genotype. C. At baseline, HbSS RBCs have higher levels of reactive oxygen species (ROS) compared with HbAA (p=0.007). D. Nitrite (10μM – 1mM) did not change ROS output in HbAA or HbSS RBCs. However, exposure of HbAA RBCs to 3-10mM nitrite increased ROS levels (p<0.001 for each concentration compared with vehicle-treated). In HbSS RBCs, 10mM nitrite treatment decreased ROS levels (p<0.001, 10mM nitrite compared with vehicle-treated). In panels C and D, N=6 per genotype (each of the 6 samples included blood pooled from 2 mice). E. At baseline, ATP levels were higher in purified HbSS RBCs compared with HbAA (p=0.002, N=4 per genotype). F. Exposure to nitrite (10μM – 10mM) did not change ATP levels in HbAA RBCs, whereas, 10mM nitrite significantly reduced ATP levels in HbSS RBCs (p=0.01, 10mM treatment compared with vehicle-treated, N=7 for each genotype).

We also investigated the effects of nitrite on the redox state of RBCs. For these experiments, we isolated RBCs and removed white blood cells from the blood ensuring that ROS measurements reflect solely those produced by RBCs (supplemental file and supplemental Fig. 4). At baseline conditions, HbSS RBCs had higher ROS levels compared with HbAA (p=0.007, Fig. 3C). Overall, the effects of nitrite on RBC ROS levels varied according to genotype (p<0.001 for genotype by treatment interaction, Fig. 3D). Compared with vehicle, nitrite doses of 3-10 mM yielded increases in ROS in HbAA RBCs (p<0.001, Fig. 3D). In contrast, compared with vehicle, nitrite at 10mM yielded reductions in ROS levels in HbSS RBCs (p<0.001, Fig. 3D). We hypothesized that this reduction in ROS levels in HbSS RBCs at high nitrite concentration reflected RBC toxicity. We then measured RBC ATP levels as a surrogate measurement of cell viability. At baseline, HbSS RBCs had higher ATP levels compared with HbAA (p=0.002; Fig. 3E). In HbAA RBCs, exposure to nitrite concentration of up to 10mM, yielded no significant changes in ATP levels compared with vehicle (Fig. 3F). Conversely, in HbSS RBCs, exposure to 10mM of nitrite, decreased ATP levels compared with vehicle (p=0.01, Fig. 3F), thus suggesting a dose-dependent RBC toxicity.

Nitrate exposure did not form methemoglobin

In vitro, exposure to nitrate (10μM – 10mM) yielded no methemoglobin formation in HbAA or HbSS blood (data not shown). We did not investigate the effects of nitrate on hemin, ROS, or ATP levels.

Long-term nitrite and nitrate administration differentially changed nitrite and nitrate levels in blood compartments of HbAA and HbSS mice

Over the 4-week treatment, HbAA and HbSS mice ingested similar amounts of nitrite and of nitrate (p= 0.751, HbAA compared with HbSS for each drug, Supplemental Fig. 5A) and at the doses administered, no significant changes in weight or water intake were noted (Supplemental Fig. 5B–D), indicating that animals tolerated nitrite and nitrate well.

In vehicle-treated animals, HbSS had higher plasma nitrite compared with HbAA mice (p<0.001, Fig. 4A). The effect of nitrite administration on plasma nitrite levels varied according to genotype (p=0.013 for treatment by genotype interaction, Fig. 4A). At week four, nitrite-treated HbAA (p<0.001), but not HbSS mice (p=0.244) had higher plasma nitrite levels compared with respective vehicle-treated animals (Fig. 4A).

Figure 4. Effect of long-term supplementation with nitrite (+NI) and nitrate (+NA) differentially alter their levels in blood compartments.

Box plots show the median and first and third quartiles and the whiskers the 10^th and 90^th percentiles. A. Among vehicle-treated (Veh) animals, HbSS had higher plasma nitrite compared to HbAA mice (p<0.001). The effect of nitrite supplementation on plasma nitrite varied according to genotype (p=0.013 for treatment by genotype interaction). At week four, nitrite-treated HbAA (p<0.001), but not HbSS mice (p=0.244) had higher plasma nitrite levels compared with respective vehicle-treated animals. B. The effect of nitrite and nitrate supplementation on RBC nitrite levels also varied according to genotype (p<0.001, for treatment by genotype interaction). In vehicle-treated animals HbSS had higher RBC nitrite levels compared with HbAA mice (p<0.001). Compared with vehicle, nitrite and nitrate administration yielded higher RBC nitrite levels in HbAA (p<0.001 for both) but no significant changes in HbSS animals (p=0.612 and p=0.539 for nitrite and nitrate respectively). C. Compared with respective vehicle-treated animals, nitrate-treated HbAA (p<0.001) and HbSS (p=0.001) mice had higher plasma nitrate levels. D. Across treatments, nitrate levels were lower in HbSS compared with HbAA RBCs (p<0.001 for overall genotype effect). Across all genotypes, nitrate-treated mice had higher RBC nitrate levels compared with nitrite-treated animals (p=0.002). E. Across treatments, HbSS had higher methemoglobin levels compared with HbAA mice (p<0.001, for overall genotype effect). Compared with vehicle, nitrite or nitrate administration yielded no significant changes in methemoglobin levels in HbAA or HbSS mice. N=20 for each experimental group (equal number of age-matched male and females).

Regarding RBC nitrite, in vehicle-treated animals, HbSS had higher RBC nitrite levels compared to HbAA mice (p<0.001, Fig. 4B) and the effects of nitrite and nitrate supplementation varied according to genotype (p<0.001, for treatment by genotype interaction Fig. 4B). At week four, compared with vehicle-treated, nitrite- and nitrate-treated HbAA had higher RBC nitrite levels (p<0.001 for both). In contrast, neither nitrite or nitrate treatment yielded significant changes in HbSS RBCs nitrite levels (p=0.612 and p=0.539 for nitrite and nitrate respectively, Fig. 4B).

In vehicle-treated animals, HbSS and HbAA mice had similar plasma nitrate levels (p=0.885, Fig. 4C). Compared with respective vehicle-treated animals, nitrate-treated HbAA (p<0.001) and HbSS (p=0.001) mice had higher plasma nitrate levels at week four (Fig. 4C). Across genotypes, nitrite supplementation did not alter plasma nitrate levels (p=0.488, Fig. 4C). Regarding RBC nitrate levels, across all treatments, HbSS mice had lower RBC nitrate levels compared to HbAA animals (p<0.001 for genotype effect, Fig. 4D). Across genotypes, nitrate-treated mice had higher RBC nitrate levels compared with nitrite-treated animals (p=0.002, Fig. 4D). The data shown in Fig. 4B and 4D presents RBC nitrite and nitrate levels as intracellular concentrations (μM). The same data adjusted to hemoglobin content (as surrogate for total RBC protein), which considers the anemia present in HbSS, is shown in Supplemental Fig. 6.

We investigated if nitrite or nitrate administration modified blood methemoglobin levels in vivo. Across treatments, HbSS had higher methemoglobin levels compared with HbAA mice (p<0.001, for overall genotype effect, Fig. 4E). Nitrite and nitrate administration yielded no significant changes in methemoglobin levels in HbAA or HbSS mice (p=0.841, for overall effect of treament, Fig. 4E). Additionally, across genotype and treatments, males (8.37 ± 0.31%, mean ± SEM) had higher methemoglobin levels compared with female (7.45 ± 0.31) mice (p=0.035, for overall sex effect).

Nitrite and nitrate supplementation differentially altered hematologic parameters in SCD mice

The effect of nitrite and nitrate administrations on WBC varied according to genotype (p=0.018 for treatment by genotype interaction, Fig. 5A). In keeping with previous reports, in vehicle-treated animals, HbSS had higher WBC compared to HbAA animals (p<0.001, Fig. 5A) and this leukocytosis worsened in nitrite- (p=0.002) and nitrate-treated (p=0.002) HbSS mice. In contrast, in HbAA animals, nitrite and nitrate supplementation yielded no significant change in WBC, Fig. 5A.

Figure 5. Hematological effects of nitrite (+NI) and nitrate (+NA) supplementation.

Box plots show the median and first and third quartiles and the whiskers the 10^th and 90^th percentiles. A. Among vehicle-treated mice, white blood cell (WBC) count was elevated in HbSS compared with HbAA samples (p<0.001) and nitrite and nitrate supplementation worsened the leukocytosis observed in vehicle-treated HbSS animals (p=0.002 for both nitrite of nitrate). Nitrite or nitrate supplementation did not affect WBC count in HbAA animals. B., C., and D. Across treatments, RBC count, hemoglobin, and hematocrit were lower in HbSS compared with HbAA mice (all p<0.001, for genotype effect). Compared with respective vehicle-treated animals, nitrite-treated HbAA and HbSS mice had higher RBC, hemoglobin, and hematocrit (all p<0.001 for treatment effect. N=20 animals per genotype/treatment; equal number of males and females.

Across treatments, RBC count, hemoglobin, and hematocrit were lower in HbSS compared with HbAA mice (all p<0.001, for genotype effect, Fig. 5B–D). Compared with respective vehicle-treated, nitrite-treated HbAA and HbSS mice had higher RBC, hemoglobin, and hematocrit (all p<0.001 for treatment effect, Fig. 5B–D). Across treatments, HbSS mice had higher mean corpuscular volume (p<0.001), mean corpuscular hemoglobin (p<0.001), platelet counts (p=0.036) and red cell distribution width, and lower mean corpuscular hemoglobin concentration (p<0.001, supplemental Fig. 7). Nitrite or nitrate supplementation did not alter any of those hematologic parameters (all p>0.35, Supplemental Fig. 7).

Discussion

In the experimental conditions used in this investigation (in vitro, 21% oxygen, high phosphate buffer, high nitrite/RBC ratio), in RBCs from sickle mice, nitrite increased the delay time for HbS polymerization (Fig. 1) at concentrations that were not associated with increases in methemoglobin formation (Fig. 2), hemin release (Figs. 3A and 3B), or increases in ROS (Figs. 3C and 3D). In keeping with previous reports (Almeida et al., 2020), vehicle-treated HbSS mice had higher nitrite levels in plasma and RBCs compared with HbAA animals (Figs. 4A and 4B). Compared with vehicle, administration of nitrite or nitrate for four weeks increased nitrite levels in blood compartments of HbAA. Conversely, in HbSS mice, the same protocol failed to further increase plasma or intra-RBC nitrite levels (Figs. 4A and 4B).

Over 60 years ago, Beutler and Mikus reported that nitrite reduced sickling and increased RBC survival in patients with SCD (Beutler, 1961). While the authors hypothesized that those effects could be related to nitrite-induced formation of methemoglobin, decreases in sickling did not correlate with methemoglobin levels (Beutler, 1961). Here we found a possible alternative explanation as nitrite at concentrations observed physiologically (0.1 – 1μM) or slightly above it (1 – 10μM) significantly increased the delay time for HbS polymerization (Fig. 1), suggesting that nitrite has anti-polymerization effects. As HbS polymerization leads to sickling of RBCs, these findings reveal a possible mechanism, which could have led to those decreases in sickling observed in nitrite-treated SCD patients (Beutler, 1961). Moreover, our results suggest that these nitrite anti-polymerization properties, are unrelated to methemoglobin formation, which was only observed at much higher concentrations (Srihirun et al., 2012; Akrawinthawong et al., 2014; Wajih et al., 2017)

It is surprising that nitrite had anti-polymerization effects as delay times for polymerization (Fig. 1) were measured after RBCs were exposed to nitrite at 21% oxygen, a condition in which oxygenated HbS predominates and nitrite metabolism is expected to produce nitrate and methemoglobin. Instead, in those conditions, at concentrations slightly above physiologic, nitrite increased delay times for polymerization without increasing methemoglobin (Figs. 1 and 2). One possibility is that nitrite increased levels of S-nitrosated HbS (SNO-HbS), which has higher affinity for oxygen (Bonaventura et al., 1999; Hrinczenko et al., 2000; Bonaventura et al., 2002), thereby decreasing the concentration of deoxygenated HbS and increasing the delay time for HbS polymerization. Other studies support this hypothesis. For example, in vitro, SNO-Hb forms rapidly when RBCs are exposed to low nitrite concentrations (>200nM) (Bryan et al., 2004) and in humans, NO inhalation increases levels of SNO-Hb and of nitrosyl(heme)hemoglobin (nitrosyl-HbS) (Gladwin et al., 2000). It is then reasonable to hypothesize that at slightly supra-physiologic concentrations (1-10μM), nitrite can produce enough SNO-HbS that can inhibit HbS polymerization. While the mechanisms underlying this anti-polymerization effect of nitrite were not examined further, akin to observations in patients with SCD (Beutler, 1961), our data suggest that methemoglobin accumulation is not necessary to decrease sickling as methemoglobin formation only occurred at nitrite concentrations much higher than those associated with increases in polymerization delay times (>300μM, Fig. 2) .

Our dose-response studies indicated that 100% of HbA but only approximately 70% of HbS could be oxidized to methemoglobin at higher nitrite concentrations (Fig. 2). This differential effect of nitrite in forming methemoglobin-A and methemoglobin-S is surprising given that the main recognized functional difference between HbA and HbS is their solubility (Adachi and Asakura, 1979). Our findings that the anti-polymerization agent DMA increased nitrite-induced methemoglobin-S formation suggest that polymerized HbS partially contributes to the fraction of HbS that is resistant to nitrite-induced oxidation to methemoglobin. Another possible contributing factor for this discrepancy in nitrite-induced methemoglobin comparing HbAA and HbSS is that sickle RBCs have an increased capacity to reduce methemoglobin. In fact, others have shown that sickle RBCs have elevated methemoglobin reductase activity (Zerez et al., 1990; Chaves et al., 2019), normal NADH levels (Zerez et al., 1990) and manipulations that raised intra-RBC NADH levels produced corresponding decreases in methemoglobin HbS half-life (Zerez et al., 1990). Our findings that addition of NADH, a substrate of methemoglobin reductase type I, or ascorbic acid, an antioxidant, did not modify the levels of methemoglobin (Supplemental Fig. 3), suggests that processes that reduce methemoglobin (Zerez et al., 1990), do not play a role in the differences of nitrite efficacy in forming methemoglobin comparing HbAA and HbSS blood. It is possible that differences in experimental conditions could explain these contradictory results. We measured methemoglobin levels 2 hours after the incubation of RBCs with nitrite, a time that is sufficient for complete nitrite metabolism (Almeida et al., 2020) and gradual accumulation of methemoglobin, which is known to occur over 60 to 90min of nitrite exposure (Zavodnik et al., 1999; May et al., 2000; Kohn et al., 2002). Therefore, it is possible that added NADH delayed methemoglobin accumulation but its effect went undetected in our experimental conditions. Nevertheless, in this investigation, the fraction of HbS resistant to oxidation by nitrite appears to be unrelated to enhanced methemoglobin reductase type I activity.

Previous studies have shown that nitrite increases ROS production and protein and lipid oxidation and decreases levels of reduced glutathione and the activity of anti-oxidation and metabolic enzymes in a dose-dependent manner (0.1 – 10mM) (Zavodnik et al., 1999; May et al., 2000; Ansari et al., 2015; Ansari and Mahmood, 2016). Our findings that nitrite toxicity mediated by methemoglobin formation and hemin release was observed only at higher concentrations (>300μM) are in concert with those studies (Zavodnik et al., 1999; May et al., 2000; Ansari et al., 2015; Ansari and Mahmood, 2016). However, we observed increases in ROS formation in HbSS RBCs only at the highest nitrite concentration studied (>3mM, Fig. 3D). We measured ROS and ATP avoiding signal from contaminating WBCs, which produce ROS in response to activation of toll-like receptor 4 by hemin (Figueiredo et al., 2007; Belcher et al., 2014) by using purified RBCs following guidelines suggested by the European Red Cell Society (ERCS)(Minetti et al., 2013). Our discrepant results might then be explained by the fact that some investigations, unlike ours, did not isolate RBCs and the samples might have contained WBCs (Minetti et al., 2013). Therefore, it is reasonable to attribute our findings of lack of increased ROS signal at lower nitrite concentrations to the use of a RBC isolation procedure that completely eliminates WBCs from the samples. Nevertheless, our findings suggest that in vitro, at concentrations lower than 100μM nitrite does not alter ROS formation in RBCs.

We had previously shown that HbSS mice and patients with SCD in steady state already have elevated nitrite levels in blood compartments compared to controls(Almeida et al., 2020). Here, our in vitro and in vivo findings suggest that in order to achieve RBC nitrite levels that are associated with decreased HbS polymerization in vitro and with other potentially beneficial effects (Srihirun et al., 2012; Akrawinthawong et al., 2014), higher nitrite doses are needed. However, the nitrite dose administered here (1g/L nitrite corresponding to ingestion 448.5 ± 19.5 mg/Kg/day) significantly increased plasma and RBC nitrite in HbAA, but not in HbSS mice.

One possibility is that, in HbSS mice, the mechanism responsible for nitrite accumulation in sickle RBCs may be saturated and that additional nitrite administration will not yield further increases in RBC nitrite. Studies administering nitrite at higher doses will test this hypothesis. It is worrisome that nitrite yielded significant further increases in WBCs in SCD mice (Fig. 5A) given that leukocytosis is a predictor of poor outcome in SCD (Platt et al., 1994). However, others have shown that nitrite reduces leucocyte and platelet adhesion to endothelial cells in culture and to vascular cell wall (Wajih et al., 2017). As the mechanisms for leukocytes differentiation (Sottoriva and Pajcini, 2021) and vascular wall adhesion (Morikis et al., 2021) are different at the molecular and vascular levels, it is possible that, despite increasing leukocytosis, nitrite may have a net positive impact on leukocyte adhesion to endothelial cells and vasculature.

We acknowledge that our experimental design and conditions need to be considered when interpreting some of our results. First, we examined the delay time of HbS polymerization in vitro (Fig. 1) using a commonly used model system with high concentration phosphate buffer and lower HbS concentrations. While these conditions have several technical advantages (see supplemental material and methods for details), phosphate and hemoglobin concentrations are different in vivo (Chen et al., 2004). Further, as we did not examine the potential changes in polymerization kinetics and RBC sickling in HbSS mice after nitrite treatment, it is unknown whether our findings in vitro can be replicated in vivo. Another consideration is that our present findings of increased RBC, hematocrit, and hemoglobin levels in nitrite-treated HbAA and HbSS mice (Fig. 5) are contrary to our previous studies in this mouse model (Wang et al., 2018; Almeida et al., 2020). Additionally, while nitrite administration increased nitrite levels in blood compartments of HbAA, but not in HbSS mice (Figs. 4A and 4B), it increased RBC and hemoglobin in both genotypes (Figs. 4B, 4C, and 4D). While it is possible that differing husbandry (single housing, different mouse chow) conditions and experimental procedures explain these discrepant results, future studies examining different nitrite doses will determine whether our in vitro findings will translate to beneficial effects in vivo.

Overall, our findings indicate that nitrite decreases HbS polymerization and may have a positive impact in SCD at concentrations slightly above physiological levels (1 - 10μM). Further, nitrite toxicity mediated by methemoglobin formation and hemin release only occurs at much higher concentrations (>300μM). Therefore, our findings suggest that if the in vitro anti-polymerization effects are to be explored in vivo, nitrite should be administered at doses much higher than those used here. The present investigation can inform the design of future in vivo studies to determine whether further elevating intra-RBC nitrite is possible and whether it favorably impacts sickling kinetics, decreases SCD complications and organ damage, without significant toxicity in vivo.

Highlights.

• The root cause of sickle cell disease is polymerization of sickle hemoglobin

• There is renewed interest in nitrite as a therapy for sickle cell disease

• Supra-physiologic nitrite concentrations decrease sickle hemoglobin polymerization

• This effect was independent of methemoglobin formation or hemin release

• Nitrite may play a therapeutic role in sickle cell disease

Abbreviations

SCD

sickle cell disease

RBC

red blood cells

HbA

hemoglobin A

HbS

hemoglobin S

NO

nitric oxide

HbSS

homozygous for the human hemoglobin S mutation

HbAA

homozygous for the human hemoglobin A mutation

EDTA

ethylenediaminetetraacetic acid

PBS

phosphate buffered saline

DMA

dimethyl adipimidate dihydrochloride

Emax

relative maximum effect of the drug (Emax)

EC₅₀

concentration producing 50% of the respective maximum effect

WBC

white blood cells

ROS

reactive oxygen species (ROS)

ATP

adenosine triphosphate

SNO-HbS

nitrosated HbS (SNO-HbS)

Nitrosyl-HbS

nitrosyl(heme)hemoglobin S

NADH

nicotinamide adenine dinucleotide reduced form NADH

NI

nitrite

NA

nitrate

Data availability statement

The data that support the findings of this study are available upon request to the corresponding author.