Comodulation of NO-Dependent Vasodilation by Erythroid Band 3 and Hemoglobin: A GP.Mur Athlete Study

GP.Mur, a red blood cell (RBC) hybrid protein encoded by glycophorin B-A-B, increases expression of erythroid band 3 (Anion Exchanger-1, SLC4A1). GP.Mur is extremely rare but has a prevalence of 1–10% in regions of Southeast Asia. We unexpectedly found slightly higher blood pressure (BP) among healthy Taiwanese adults with GP.Mur. Since band 3 has been suggested to interact with hemoglobin (Hb) to modulate nitric oxide (NO)-dependent hypoxic vasodilation during the respiratory cycle, we hypothesized that GP.Mur red cells could exert differentiable effects on vascular tone. Here we recruited GP.Mur-positive and GP.Mur-negative elite male college athletes, as well as age-matched, GP.Mur-negative non-athletes, for NO-dependent flow-mediated dilation (FMD) and NO-independent dilation (NID). The subjects were also tested for plasma nitrite and nitrate before and after arterial occlusion in FMD. GP.Mur+ and non-GP.Mur athletes exhibited similar heart rates and blood pressure, but GP.Mur+ athletes showed significantly lower FMD (4.8 ± 2.4%) than non-GP.Mur athletes (6.5 ± 2.1%). NO-independent vasodilation was not affected by GP.Mur. As Hb controls intravascular NO bioavailability, we examined the effect of Hb on limiting FMD and found it to be significantly stronger in GP.Mur+ subjects. Biochemically, plasma nitrite levels were directly proportional to individual band 3 expression on the red cell membrane. The increase of plasma nitrite triggered by arterial occlusion also showed small dependency on band 3 levels in non-GP.Mur subjects. By the GP.Mur comparative study, we unveiled comodulation of NO-dependent vasodilation by band 3 and Hb, and verified the long-pending role of erythroid band 3 in this process.


INTRODUCTION
Hemoglobin (Hb) modulates vascular tone by controlling nitric oxide bioavailability in the blood stream. The chemical reaction between Hb and NO metabolites is coupled with the dynamics of O 2 saturation and desaturation of Hb (1)(2)(3)(4)(5). While intraerythrocytic Hb is considered the most powerful NO scavenger in human body, its effect on NO-associated vasodilation is limited by the RBC membrane (6). Hb-mediated NO bioavailability has been proposed to require band 3, the main anion transporter on RBCs. The interaction between band 3 and deoxygenated Hb in the submembranous domain has been suggested to assist the transport of NO metabolites (e.g., nitrite and SNO) across the red cell membrane by unclear means (7)(8)(9)(10). Currently, the role of band 3 in RBC-dependent NOx processing and hypoxic vasodilation is enigmatic.
GP.Mur (also known as Miltenberger subtype III or Mi.III), is a Southeast Asian special blood type with 1-10% prevalence among Filipinos, Vietnamese, Thai, and Taiwanese, but is very rare among Caucasian (0.0098%), northern Asian like Han Chinese and Japanese (0.006%) (11)(12)(13)(14). Biochemically, GP.Mur is a hybrid protein originated from homologous gene recombination of the highly homologous GPYB and GYPA genes (15). As GP.Mur contains a segment of glycophorin A (GPA) and GPA acts as a chaperone to band 3 (16,17), the consequence of the expression of GP.Mur is to support increased expression of band 3 on the erythrocyte surface (18). Band 3, a Cl − /HCO − 3 antiporter, allows HCO − 3 permeation across the erythrocyte membrane, which is physiologically essential for intraerythrocytic CO 2 /HCO − 3 conversion and postnatal CO 2 respiration (19)(20)(21). People with GP.Mur blood type express more band 3 on RBCs and show faster CO 2 respiration after a mild physical challenge (18,22). Unexpectedly, from our previous human exercise study on Taiwanese adults aged up to 50 years old, GP.Mur+ subjects exhibit slightly higher blood pressure than the non-GP.Mur counterparts (22,23).
The structural-functional hallmarks of GP.Mur+ red cells are enhanced band 3 expression (18) and altered band 3 complex structure with other membrane proteins on the red cell surface (21,(24)(25)(26). GP.Mur+ red cells present stronger band 3-AQP1 interaction (18,26), contain reduced amount of RhAG and consequently altered band 3-Rh/RhAG macrocomplex organization (25,27), and are more resilient physically in the osmolarity-fragility test (18). These presumably could affect band 3 interaction with submembranous deoxy Hb and metHb (28) and/or result in differential NOx processing, vasodilation, and blood pressure setting. To test this hypothesis, we examined NO-dependent FMD and NO-independent vasodilation primarily in a uniform cohort-Taiwanese elite college athletes who have been trained professionally for national and/or international sports competitions for approximately half of their lifetime. Together with blood measurements for band 3 and NOx changes during FMD, we identified co-modulation of FMD by erythroid band 3 and Hb.

Subject Recruitment, GP.Mur Phenotyping, and Study Design
This human study protocol had been approved by MMH-Institutional Review Board (IRB registration: 19MMHIS081e). For uniform cohorts, we primarily recruited male elite athletes from top sports universities in Taiwan, where they enrolled with their competition scores from national/international games. Before college, all of them had professional sports training and competition experiences for at least 6 years. All subjects were free from major illness. To maintain eligibility for sports competitions, they were prohibited from taking most medication. We also recruited age-and BMI-matched, healthy male nonathletes as controls. All subjects had given written informed consents before participating in this study.
GP.Mur blood type was first determined by red cell serology using anti-Mur and anti-Mi a antisera, and then confirmed by PCR-sequencing, as described (29). It has long been observed in Taiwan that the prevalence of GP.Mur among elite Taiwanese athletes is much higher. We pre-screened 129 elite male college athletes and found 32 of them with GP.Mur blood type (24.8%). Among the elite athletes, 51 of them agreed to participate in this study, and 19 of them bear GP.Mur phenotype (37.3%). As for the non-athlete subjects, only one out of 20 bear GP.Mur (5%), which is similar to the occurrence frequency of GP.Mur in general Taiwanese population.
The study design was illustrated in the flow chart in Figure 1. Briefly, each subject was tested on two separate days: 1 day for ultrasound measurements of FMD and then NID; the other day for assessments of the changes of blood plasma NO metabolites during FMD (blood withdrawal mostly completed within 1 min after cuff deflation). The subjects were not allowed intake of food or caffeinated drinks at least 2 h prior to FMD tests on both days. They were asked not to practice sports at least 6 h before testing.

Ultrasound Measurements for FMD and NID
We followed the FMD and NID protocols provided by the Hitachi-Aloka eTRACKING (echo-tracking) system (30). The Aloka eTRACKING program was installed in the Hitachi Arietta 60 (Ar60) ultrasound system equipped with a 12-2 MHz transducer (L441).
Before ultrasound examination, each subject was allowed to calm down in bed in a dark, quiet examination room, and his left wrist was wrapped with a pneumatic cuff. The L441 transducer locked in a probe holder was placed about 5-10 cm above his left elbow to image his left brachial artery. Under the FMD mode of eTRACKING, a selected region of his brachial artery was first focused and diameter changes in the focused brachial artery were recorded over time. The FMD recording began with 1min baseline (stabilization), 5-min inflation of the wrist pressure to 50 mmHg above his systolic blood pressure (SBP), followed by deflation and 5 more minutes of monitoring the changes of the brachial arterial diameter. The peak of reactive hyperemia generally appeared within 30-90 s post-arterial occlusion, and the FIGURE 1 | The study flow chart combined ultrasound measurements of FMD and NID, and blood sampling before cuffing and right upon pressure deflation which allowed assessment of plasma NOx changes by arterial occlusion during FMD. The study was arranged in two separate days: Day 1 for ultrasound-based tests; Day 2 for blood sampling before and after 5-min brachial arterial constriction. In Day 1, the subject was started with the ultrasound measurement for FMD, followed by a minimal 20-min break to allow recovery of the tested brachial artery, and then the second ultrasound measurement for NID. FMD took ∼11 min (including 1 minute of stabilization, 5 min of cuff pressuring to 50 mmHg above the subject's SBP, and another 5 min of observation after wrist cuff deflation). The NID measurement also began with 1-min stabilization, followed by oral administration of NTG and then 10-min monitoring of the changes of the brachial arterial diameter. In Day 2, the setting was very similar to FMD in Day 1 (5 min of pressuring to 50 mmHg above the subject's SBP), but ultrasound was replaced by blood sampling before and after cuffing.
brachial arterial diameter generally returned to its baseline level within 2 min (see Figure 2A).
After FMD, the subject was given a 30-min break and then stabilized again in bed for ultrasound examination of NID on a similar region of the brachial artery as for FMD. NOindependent dilation recording by eTRACKING also began with 1-min baseline. The subject was then given 0.6 mg sublingual nitroglycerin (Nitrostat R , Pfizer), and his changes of the brachial arterial diameter were recorded for 10 min. These subjects generally reached maximal vasodilation by the 6th minute post-nitroglycerin. Nitroglycerin-induced hyperemia generally persisted till the end of the total 11-min recording (see Figure 2A).

Determination of Blood Plasma Nitrate and Nitrite Concentrations
To find out blood plasma NOx changes during FMD, on a separate visit the subject underwent a similar mechanical stimulus-5-min cuff inflation to 50 mmHg above his SBP on his right upper arm. Venous blood was sampled from his left median cubital vein at two timepoints: (1) before cuffing; (2) after deflation of the arm cuff. This allowed us to evaluate the consequence of regional ischemia and onset of NO-dependent vasodilation reflected by plasma NOx. Our blood sampling protocol for NOx during reactive hyperemia followed a report by Allen et al. (31), with slight modification.
Right after blood withdrawal into an ACD tube, the sample was immediately centrifuged to separate plasma from RBCs. Blood plasma was further filtered using an Amicon Ultra-4 Centrifugal unit with 10-kDa cutoff at 3,220 × g for 30 min at 4 • C, and then aliquoted and stored at a −80 • C freezer. Due to the short lifetime of nitrite, the handling process from blood withdrawal to sample storage in −80 • C was limited to 45 min or less. Plasma NOx and NO − 3 were quantitated using the Nitrate/Nitrite colorimetric assay kit (Cayman Chemical, MI, USA); plasma NO − 2 was quantitated with the Nitrate/Nitrite fluorometric assay kit (Cayman).

Band 3 Quantitation
RBCs were hypotonically ruptured to obtain the membrane fraction (ghost). Total proteins in freshly lysed ghosts were quantified by Lowry protein assay (Bio-RAD, CA, USA), and band 3 levels measured by band 3 sandwich ELISA. The band 3 sandwich ELISA was developed in house. Briefly, anti-band 3 rat monoclonal BRAC 18 (International Blood Group Reference Laboratory [IBGRL], NHS, UK) was applied as the coating antibody (1:500 dilution) onto an ELISA plate overnight at 4 • C. RBC ghost lysates were then reacted with the coated BRAC 18 for 30 min at 37 • C, followed by PBST (0.05% Tween-20-supplemented PBS) washes and incubation with another anti-band 3 mouse monoclonal BRIC 170 (IBGRL; 1:500 dilution) for 2 h at 37 • C. The signal of BRIC 170sample binding was amplified by incubation with 1:5,000 HRP-conjugated, anti-mouse-specific secondary IgG (Jackson ImmunoResearch Laboratories, PA, USA); this secondary antimouse antibody had been immunoaffinity-purified to yield minimal cross reactivity with rat. The standard curve for absolute quantitation of band 3 in the sandwich ELISA used a custommade band 3 peptide that contains the known epitope sequence of BRIC 170 (AA 368-382); the custom-made peptide sequence is NGGPDDPLQQTGQLFGGLVR (AA 365-384 of human band 3 ([NP_000333.1]). One's band 3 level relative to all RBC membrane proteins in his sample (w/w) was obtained by dividing the band 3 concentration by the ghost protein concentration and expressed in percentile (%).

RESULTS
To probe into the potential impacts of GP.Mur/band 3 on NOdependent reactive hyperemia in a uniform cohort, this study recruited primarily male elite college athletes from top-ranking Taiwan sports universities where the prevalence of GP.Mur is much higher than the general Taiwanese population. The athlete subjects all have similar age, lifestyle, and enrolled in rigorous sports training programs aiming for national/international competitions for approximately half of their lifetimes ( Table 1). They were divided into GP.Mur-positive (19/70) and GP.Murnegative athlete groups (32/70). For comparison, twenty age-and BMI-matched non-athlete subjects were also recruited (only 1/20 GP.Mur+). Compared to the non-GP.Mur non-athletes (19/70), both groups of athletes showed similarly lower heart rates and slightly higher systolic blood pressuer ( Table 1).

FMD Limited More by Hb in GP.Mur+ Than in Non-GP.Mur Subjects
In a study on hypertensive patients with high-altitude erythropoiesis, FMD in these patients was inversely correlated with their abnormally high Hb (>21 g/dL) and excessive number of circulating RBC cells (33). Here, our subjects were healthy young students with Hb and RBC count in the normal ranges. Their Hb and RBC count were not different because of GP.Mur ( Table 1). But only in GP.Mur+ subjects was FMD significantly inversely correlated with Hb ( Figure 3A: thick solid line). This trend would not be obvious in most non-GP.Mur people with normal Hb, as shown with the poor fitting in Figure 3A (Figure 3A). From the data, Hb in GP.Mur+ RBCs was more effective in scavenging NO, and this likely involved distinct components in the GP.Mur red cell membrane. In contrast, the correlation between Hb and NID was poor for all three groups (p > 0.05), indicating that NID was not associated with Hb (Figure 3A, right). These results confirm the central role of Hb in NO-dependent vasodilation.
On the other hand, FMD correlation with RBC count was not much different between GP.Mur and non-GP.Mur (Figure 3B: r or the slopes for GP.Mur+ athletes, −0.47; for non-GP.Mur athletes, −0.31). From their similar dependency of FMD on RBC count, their drastically different FMD sensitivity to Hb had to be primarily triggered by their differences in the RBC membrane composition (i.e., GP.Mur/band 3 protein complexes).

Blood Plasma Nitrite Levels Dynamically Correlated With Band 3 Expression
Similar to the FMD/NOx study by Allen et al. (31), we observed that occlusion of the brachial artery followed by cuff deflation decreased systemic plasma nitrate and increased systemic plasma nitrite contents (Table 2, Figure 4A). But the concentrations of plasma nitrite, nitrate, or NOx before pressure cuff or right after cuff deflation were not different between GP.Mur+ and non-GP.Mur athletes (  and nitrite consumption. Under physiologically oxygenated conditions, most nitrite is constitutively generated in tissue from oxygenation of NO produced by eNOS (34), and is consumed by reduction into NO primarily via deoxy Hb (35). Individual eNOS levels were expected to vary, and hence individuals with high [NO − 2 ] pre−FMD generally showed high [NO − 2 ] upon deflation (Figure 4B).
Band 3 complexes have been suggested to assist Hb in NOdependent hypoxic vasodilation, either as a nitrite transporter (7) or/and as a mediator for permeation of NOx/SNO metabolites across the red cell membrane (8,9). Band 3 expression was slightly higher in people with GP.Mur ( Table 2) (18,22,36). We thus explored the possible involvement of band 3 in NOx changes during FMD. The concentrations of systemic plasma nitrite before cuff pressing and after deflation were both directly correlated to band 3 expression levels on the RBC membrane ( Figure 4B). The correlation between band 3 and systemic nitrite was better before than after regional arterial occlusion (Figure 4B: larger Pearson's correlation r before cuffing [left] than after cuff deflation [right]). Notably, for GP.Mur+ athletes, their band 3-nitrite correlation was much smaller and diminished after cuff deflation. We calculated the extent of systemic nitrite changes induced by this regional hypoxia ([NO − 2 ] post -[NO − 2 ] pre , or [NO − 2 ] cuffing ), which was significantly correlated with erythroid band 3 levels but only in non-GP.Mur subjects (Figure 4C). This discrepancy suggests different impacts of GP.Mur+ and non-GP.Mur membrane structure on NOx metabolism, as seen in Figure 3A.

DISCUSSION
Intraerythrocytic Hb is the major NO scavenger inside human body, but deoxy Hb as a nitrite reductase also generates NO in low O 2 during the respiratory cycle. The sources of NO for maintaining the vascular tone mainly come from constitutive conversion of L-Arginine by endothelial NO synthase (eNOS) and from reduction of NOx by heme reductase like deoxy Hb. NO production by deoxy Hb complements NO production by eNOS, which takes place preferentially in high O 2 . This may explain, at least in part, why eNOS expression alone cannot fully reflect NO bioavailability or FMD (37)(38)(39).
Upon deflation of the cuff pressure in FMD, we observed decrease of plasma nitrate and increase of plasma nitrite for both GP.Mur and non-GP.Mur groups (Table 2, Figure 4A). The reaction of NO in oxygenated RBCs mainly forms nitrate; the reaction of NO in oxygenated tissues mainly forms nitrite. Thus, decreasing nitrate suggests lower NO and a need to increase NO/nitrite bioavailability from both tissues and RBCs. Nitrite synthesis was accelerated to increase NO bioavailability.
This study probed into the effects of GP.Mur/band 3 on NOdependent vasodilation. Our age-matched, young elite athletes all had similar blood pressure, but GP.Mur+ athletes had already exhibited lower FMD than non-GP.Mur counterparts (Figure 2). Their Hb levels were all within the normal range, but Hb in GP.Mur+ RBCs was more effective in limiting NO-dependent vasodilation (Figures 2, 3). This suggests that GP.Mur/band 3associated structural features could further facilitate NO scavenge by intraerythrocytic Hb (Figure 3A). Band 3 protein complexes differ between GP.Mur and non-GP.Mur in two aspects: (1) there are more band 3 molecules on the GP.Mur+ RBCs; (2) GP.Mur+ band 3 complexes are structurally distinguishable from non-GP.Mur (18,(25)(26)(27). In particular, AQP1 and band 3 interact more strongly in GP.Mur+ than in non-GP.Mur RBCs (26). The water channel AQP1 is also a NO gas channel (40)(41)(42). Thus, we hypothesize that Hb in GP.Mur+ RBCs was more effective in scavenging NO for the two reasons: (1) more band 3 on the GP.Mur+ membrane could attract more Hb to the inner leaflet of the cell membrane; (2) their stronger AQP1-band 3 interaction could channel NO influx via AQP1 to band 3-bound Hb (Figure 5A). Though Hb was not different between GP.Mur+ and non-GP.Mur groups ( Table 1), more Hb distributed on the inner leaflet would result in more efficient NO scavenge. Since NO gas diffuses into erythrocytes extremely rapidly (at a rate near its diffusion limit), the rate of NO scavenge by intracellular Hb is critical to NO bioavailability and NO-dependent vasodilation (as in Figure 3A).
We also found that erythroid band 3 levels were directly correlated with [NO − 2 ] FMD in non-GP.Mur but not in GP.Mur+ subjects (Figure 4). Band 3 likely could support permeation of NO − 2 across the red cell membrane, as proposed by other groups (7,8). Band 3 transports HCO − 3 at the rate of ∼10 4 /s. If band 3 transports NO − 2 at a similar rate as HCO − 3 , this is still much slower than the rate of NO diffusion through the plasma membrane. Thus, the different correlations between band 3 levels and [NO − 2 ] FMD more likely involved different degrees of NO scavenge between GP.Mur and non-GP.Mur.
We hypothesize that for non-GP.Mur RBCs, a higher band 3 protein level shall decrease membrane fluidity and hence reduce NO diffusion into the red cells (Figure 5B, bottom). This is because the inner leaflet of the RBC membrane is mainly covered by metHb. Despite that metHb in RBCs is only 0.5-3%, it reacts with NO at a much slower rate than oxyHb or deoxyHb, and thus forms a barrier for NO diffusion into the red cells. Increase of band 3 expression in non-GP.Mur RBCs thus could reduce NO entry and scavenge; more plasma nitrite could then be observed ( Figure 5B). On the other hand, the lack of the correlation between erythroid band 3 and plasma nitrite levels in GP.Mur+ RBCs suggest that factors other than membrane barriers to NO diffusion (i.e., metHb and fluidity) were involved. These other factors include more extensive lining of Hb with cytoplasmic GP.Mur band 3 in the inner leaflet, and unique channeling of NO to submembranous Hb via AQP1-band 3 complexes (Figure 5A). Because of these additional factors, higher band 3 levels in GP.Mur+ RBCs conceivably also promote NO scavenging. These effects were opposite to the decrease of NO entry in less fluid red cells ( Figure 5B). These may explain why the direct correlation between plasma nitrite and band 3 levels seen in non-GP.Mur subjects was absent in GP.Mur (Figures 4, 5).
A final interesting observation from this study is the lower plasma nitrate (but not nitrite) among the elite athletes, regardless of GP.Mur ( Table 2). Skeletal muscles are the dominant nitrate reservoir over other organs or blood (43). It has been reported that the nitrate stores, particularly that in skeletal muscles, are essentially reduced to nitrite and NO for vasodilation during high-intensity exercise (43). Our athlete subjects were routinely trained at high intensity. Indeed, their nitrate store was more depleted than that in the non-athlete controls ( Table 2). Thus, from our results, even young athletes will need nitrate supplementation more than non-athletes of similar age, not just for sports performance but also for their vascular health (43).

Limitation
This work nonetheless is limited by the small sample size for a FMD study (44). The main reason is the low prevalence of GP.Mur blood type in northern Taiwan (< 2%) where this study was conducted. To minimize variations among subjects, we recruited only male elite athletes from top sports universities in the region. But this type of study may be even harder to conduct in the future, as we have noticed gradual dilution of the GYP.Mur gene in Taiwan in the past half century.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by MMH-Institutional Review Board (IRB registration: 19MMHIS081e). The patients/participants provided their written informed consent to participate in this study.