Loss of alpha-globin genes in human subjects is associated with improved nitric oxide-mediated vascular perfusion

Abstract

Alpha thalassemia is a hemoglobinopathy due to decreased production of the α-globin protein from loss of up to four α-globin genes, with one or two missing in the trait phenotype. Individuals with sickle cell disease who co-inherit the loss of one or two α-globin genes have been known to have reduced risk of morbid outcomes, but the underlying mechanism is unknown. While α-globin gene deletions affect sickle red cell deformability, the α-globin genes and protein are also present in the endothelial wall of human arterioles and participate in nitric oxide scavenging during vasoconstriction. Decreased production of α-globin due to α-thalassemia trait may thereby limit nitric oxide scavenging and promote vasodilation. To evaluate this potential mechanism, we performed flow-mediated dilation and microvascular post-occlusive reactive hyperemia in 27 human subjects (15 missing one or two α-globin genes and 12 healthy controls). Flow-mediated dilation was significantly higher in subjects with α-trait after controlling for age (P = .0357), but microvascular perfusion was not different between groups. As none of the subjects had anemia or hemolysis, the improvement in vascular function could be attributed to the difference in α-globin gene status. This may explain the beneficial effect of α-globin gene loss in sickle cell disease and suggests that α-globin gene status may play a role in other vascular diseases.

1 |. INTRODUCTION

Humans normally have four α-globin genes encoding the α-globin protein. Individuals who are missing one α-globin gene have no clinical phenotype and are called silent carriers. Those missing two α-globin genes are commonly referred to as α-thalassemia trait and have microcytosis, while individuals missing three or four α-globin genes have moderate (hemoglobin H) or profound anemia (α-thalassemia major), respectively.¹ Individuals with sickle cell disease (SCD) can co-inherit loss of one or two α-globin genes, which has long been known to influence SCD physiology.² The SCD patients missing one or two α-globin genes have higher hemoglobin, and lower bilirubin and reticulocyte counts.³ Interestingly, absence of α-globin genes in SCD is associated with decreased rates of stroke,⁴ cerebral vasculopathy,^5,6 acute chest syndrome, and leg ulcers,⁷ with risk reductions ranging from 34% to 71%. These effects are thought to be related to higher nitric oxide (NO) bioavailability⁷ due to the associated lower hemolytic rate³ and therefore lower NO consumption. The presence of α-thalassemia trait in SCD also leads to improved red blood cell (RBC) hydration and deformability,^8,9 although the exact mechanism is not known.

The effects of α-thalassemia trait in SCD are often attributed to effects on the RBC, but recent studies showed that the α-globin gene and protein are present in the endothelial wall of the myoendothelial junctions of human arterioles, where α-globin interacts with endothelial NO synthase to modulate NO scavenging, thereby affecting the degree of vasoconstriction.^10,11 In a mouse model, disruption of this interaction using an α-globin mimetic peptide increases endothelial NO activity, independently of NO production, and reduces systemic hypertension.¹² Furthermore, treatment with the α-globin mimetic also increases blood perfusion in mice, and similarly ameliorates vasoconstriction in biopsied human arterioles ex vivo.¹³ Thus, there appears to be a relationship between absence of α-globin in the vessel wall and increased arterial vasodilation, but studies of vasodilation in humans have not been reported.

These experiments suggest that the association between α-globin and NO-mediated vascular signaling in arteries and arterioles^10,11 could explain the clinical effect of missing α-globin genes in SCD that is independent of the effects of α-globin on RBC deformability, and is consistent with the idea that vaso-occlusive crises are related to both the rate of blood perfusion and RBC deformability.¹⁴ Chronic hemolysis in SCD impairs NO bioavailability through release of plasma hemoglobin and arginase, which scavenge NO and eliminate NO precursors, respectively.¹⁵ However, given that α-globin is present in the arteriolar endothelium, a reduced number of α-globin genes, and consequently reduced α-globin protein, might decrease local NO scavenging in the arteriole and improve microvascular perfusion, a potential mechanism for the effect of missing α-globin genes on SCD complications. Therefore, to isolate the effects of α-globin on perfusion from those of hemolysis and hemoglobin S polymerization in humans, we investigated whether loss of α-globin genes improves vascular perfusion in non-SCD subjects with α-thalassemia trait, a naturally occurring human model of α-globin gene knockout.

2 |. METHODS

Subjects missing one or two α-globin genes and healthy controls were recruited from the Hematology Clinic at the Childrenʼs Hospital Los Angeles (CHLA). Informed consent was obtained from all patients as approved by the CHLA Institutional Review Board. Blood samples were obtained from all subjects for determination of hemoglobin, mean corpuscular volume (MCV), reticulocyte count, plasma hemoglobin, lactate dehydrogenase (LDH), and α-globin gene status, which was determined using the α-Globin StripAssay (ViennaLab Diagnostics GmbH, Austria).¹⁶ Patients with hemoglobin H and α-thalassemia major were excluded due to potential confounding effects of anemia and transfusion therapy on blood perfusion. We also excluded any subjects with diabetes, systemic hypertension, history of smoking, or other vascular disease that could impact vascular reactivity.

We performed flow-mediated dilation (FMD) using Doppler ultra-sound (Vivid Gi, GE Healthcare, Chicago, Illinois) and the Brachial Analyzer software (Medical Imaging Applications, Coralville, Iowa) to measure brachial artery diameter before and 7 minutes after a three-minute cuff occlusion of the forearm just above the wrist to 30 Torr above baseline systolic blood pressure.¹⁷ Microvascular post-occlusive reactive hyperemia was measured simultaneously using laser Doppler flowmetry (LDF; Periflux 5000, Perimed AB, Järfälla, Sweden) and photoplethysmography (PPG; Nonin Medical Inc., Plymouth, Minnesota) in the fingertip.¹⁸ The LDF uses light scatter to determine the concentration and velocity of moving red cells, assessing blood perfusion in a 1-mm³ region just below the skin surface.¹⁹ Also, PPG similarly measures volumetric microvascular perfusion.²⁰

Blood pressure was also measured in the finger at a sampling rate of 200 Hz (Nexfin, Amsterdam, Netherlands). All signals were recorded continuously using the Acqknowledge data acquisition system (Biopac Systems Inc., Goleta, California) and exported to a specially designed browser in MATLAB for processing and analysis.^21,22 Vascular response was quantified by maximal percent change from baseline in brachial artery diameter, LDF, or PPG after occlusion. We also calculated vascular compliance by the instantaneous ratio of the change in fingertip blood volume to the change in pressure, using PPG and blood pressure, respectively.

Statistical analysis was performed with JMP version 14 (SAS Institute Inc., Cary, North Carolina). Univariate analyses used Student t test or Wilcoxon non-parametric test for continuous variables, and Pearsonʼs χ² test or Fisherʼs exact test for dichotomous variables. Skewed continuous variables were transformed to most closely approximate normality for regression analyses. Stepwise multivariate linear regression was used to assess vascular response in each experiment, including analysis of variable interaction and confounders. A P < .05 was considered significant for the final model.

3 |. RESULTS

The 27 subjects enrolled in the study included 12 controls (four α-globin genes) and 15 α-trait subjects, with 10 missing one α-globin gene and five missing two. The α-trait subjects had a lower mean MCV than controls (P = .010), but hemoglobin levels and markers of hemolysis were normal in both groups (Table 1). There was no detectable difference in FMD between individuals missing one and two α-globin genes, including after correction for MCV; thus, these subjects were combined into one group referred herein as α-trait. Baseline systolic and diastolic blood pressure (SBP and DBP) were higher in α-trait subjects (P = .027 and .049, respectively), but the association with α-trait was lost after controlling for age (P = .195 and .205, respectively); thus, SBP and DBP were not included in the FMD analysis.

TABLE 1.

Subject characteristics

	Control (n = 12)	α-trait (n = 15)	P
Age, y	28.7 [±13.4]	40.3 [±16.3]	.060*
Sex			.795†
Male	5 [42%]	7 [47%]
Female	7 [58%]	8 [53%]
Race/ethnicity			.014‡
Asian	1 [8%]	9 [60%]
Black	11 [92%]	5 [33%]
White		1 [7%]
Vitals
Height, cm	167 [±9.8]	172 [±8.3]	.206§
Weight, kg	71.3 [±20]	78.2 [±19]	.366§
BMI, kg/m2	25.6 [±7.5]	26.4 [±5.4]	.783§
SBP, mm Hg	118 [±13]	128 [±11]	.195§∥
DBP, mm Hg	73 [±11]	80 [±9]	.205§∥
HR, bpm	76.2 [±14]	75.4 [±8.6]	.859§
RR, bpm	15.1 [±2.7]	13.9 [±2.0]	.287*
O2 saturation, %	98 [±1.2]	98 [±0.8]	.203*
Laboratory values
Hemoglobin, g/dL	13.1 [±1.4]	12.7 [±1.4]	.511§
MCV, fL	85 [±7.3]	76.6 [±8.4]	.010§
Absolute reticulocyte, K/μL	62.6 [±24.7]	67.6 [±18.9]	.510*
LDH, U/L	478 [±106]	570 [±91]	.027§¶
Plasma hemoglobin, mg/dL	9.8 [±9.1]	7.9 [±6.3]	.401§

Note: Values expressed as mean [±SD] for continuous variables, and as n [%] for categorical variables. The P value was determined by

^*Wilcoxon rank sum test,

^†Pearsonʼs χ² test,

^‡Fisherʼs exact test or

^§Student t test. Bold P values indicate significant differences (P < .05).

^∥Corrected for age.

^¶Both values are in the normal range.

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; HR, heart rate; LDH, lactate dehydrogenase; MCV, mean corpuscular volume; RR, respiratory rate; SBP, systolic blood pressure.

Thus, FMD was significantly higher in α-trait subjects (least squares mean ± SE; 5.5 ± 0.5) than in controls (3.6 ± 0.5) after adjusting for age (Figure 1A; P = .036). The known decrease in FMD with age²³ (Figure 1B) was the same in control and α-trait (P = .963). Although the α-trait and control groups were not ethnicity-matched (P = .014), ethnicity was not a statistically significant determinant of FMD on univariate or multivariate analysis. So, FMD was not different between the nine Asian and five black subjects with α-trait (P = .505; data not shown).

FIGURE 1.

Multivariate analysis of flow-mediated dilation (FMD) vs α-trait and age. A, Least squares means of FMD in control and α-trait. FMD is significantly higher in α-trait (P = .036) after controlling for age. B, Multivariate regression plot. Regression lines are shown with control indicated by red points and α-trait indicated in blue. Model R² = 0.22 (F_2,24 = 3.407, P = .050)

We did not detect any effect of α-trait on microvascular perfusion by LDF or PPG after cuff occlusion, or by calculated compliance (data not shown).

4 |. DISCUSSION

Note, FMD is thought to reflect NO-mediated vasodilatory effects due to shear in conduit vessels,^24–26 and the extent of arterial dilatation after cuff release largely depends on the amount and duration of shear stress.²⁷ When the blood pressure cuff at the wrist is released, blood flows into the hand. Conduit arterial flow in the brachial artery, and hence shear stress, is also clearly increased by vasodilation of the arteriolar networks in the hand that are downstream to the occluding blood pressure cuff at the wrist.²⁸ This explains why a decrease in endothelial α-globin residing in myoendothelial junctions, which are in greater number in small resistance arteries and arterioles but are also present to a lesser extent in larger vessels,²⁹ resulted in an approximately 50% higher FMD in α-trait subjects. We had expected to see an effect of α-trait at the microvascular level through a direct effect on NO bioavailability but were unable to detect changes in microvascular perfusion using LDF, PPG or compliance measurements in α-trait individuals. Both LDF and PPG measurements are highly variable and reflect blood perfusion in small volumes of microvascular tissue, whereas FMD measures perfusion in a conduit artery that reflects many downstream microvascular networks. Thus, the effect of α-globin gene loss at the microvascular level is effectively multiplied in FMD, resulting in the detection of a greater vascular perfusion response in α-trait vs control subjects.

We had also expected a gene dosage effect on vascular perfusion, yet we did not find any difference in FMD between subjects missing one vs two α-globin genes. There are two α-globin genes, HBA1 and HBA2, located close to each other within the α-globin gene cluster on chromosome 16. In our study, most subjects missing one α-globin gene carry the common 3.7-kilobase deletion on chromosome 16 that yields a fusion gene between parts of HBA1 and HBA2.³⁰ This fusion gene acts as a single α-globin gene in the RBC, but it is unclear to what extent the fusion gene is expressed in the endothelium given that HBA1 and HBA2 are expressed in different proportions in the RBC vs the vascular endothelium.³¹ Thus, the difference in vascular α-globin protein expression between one gene and twogene loss could be similar, which would explain the lack of gene dosage effect.

Our study was not ethnicity-matched, with one subject identifying as Asian in the control group and the rest as black, while the α-trait group had a 2:1 mix of Asian to black subjects and one white subject (Table 1). Subgroup analysis showed no difference in FMD between Asian and non-Asian α-trait subjects, consistent with the lack of effect of ethnicity on FMD by multivariate analysis, suggesting that the difference in FMD is due to α-trait and not to non-specific effects of ethnicity.

As none of the subjects had evidence of anemia or hemolysis (Table 1), the difference in FMD between the control and α-trait groups could be best explained by the increased vasodilation due to the loss of α-globin genes resulting in lower α-globin protein levels.³² These findings provide a plausible explanation for the beneficial non-RBC related effect of α-thalassemia trait in SCD and suggest that exploring the α-globin gene status in other vascular diseases may point to a role for NO in the pathophysiology of these disorders.