Hypoxic vasodilatory defect and pulmonary hypertension in mice lacking hemoglobin β-cysteine93 S-nitrosylation (original) (raw)
Hb βCys93 mediates hypoxic vasodilation in vitro. We have previously shown that native βCys93 mutant RBCs induce vasodilation less effectively under hypoxia than control RBCs (7) and that human RBCs loaded physiologically with NO gas recapitulate hypoxic vasodilation by native RBCs (15). However, it has been reported recently that mouse RBCs preloaded with SNO (via treatment with CysNO) do not show hypoxic vasodilation or differences in vasodilation between mutant and control RBCs (22). In working with mouse RBCs, we noted that Hb was not modified by exogenous CysNO as readily as was Hb within human RBCs and that met-Hb (which eliminates the allosteric transition in Hb) formed in very high amounts (data not shown). We therefore developed a protocol (see Methods) optimized for SNO loading of Hb within humanized mouse RBCs. With this procedure, we are able to load SNO predominantly onto βCys93, although mouse RBCs load SNO less well and produce higher met-Hb than do normal human RBCs. Under these optimized loading conditions, the C93 (control) mouse RBC preparations contained approximately 10 SNO per 1000 Hb tetramers (~2 SNO/1000 heme) and were oxidized to approximately 10% met-Hb, while C93A (mutant) RBCs had significantly less SNO (~6 SNO per 1000 Hb tetramers or ~1 SNO/1000 heme) (Figure 1A) and indistinguishable met-Hb levels. CysNO increased SNO-Hb 5-fold over the level observed in fresh, untreated C93 RBCs and in C93A mutant RBCs, which had significantly less SNO-Hb also at baseline (Figure 1B). Adding these SNO-loaded C93 RBCs to aortic ring bioassays from wild-type mice resulted in vasorelaxation under hypoxia, but not under normoxia (Figure 1, C and D) (as seen with physiological amounts of SNO-Hb and native RBCs; refs. 25, 29), fulfilling the sine qua non requirement of hypoxic vasodilation. In contrast, SNO-loaded C93A RBCs produced significantly less vasorelaxation under hypoxic conditions than C93 RBCs but were equally vasoconstrictive under normoxia. These results demonstrate that Hb βCys93 is the primary and preferred site of _S_-nitrosylation in RBCs, that carefully loaded mouse RBCs recapitulate the hypoxic vasorelaxation found using human RBCs under basal conditions (14, 15, 25), and that this hypoxic vasorelaxation effect is significantly diminished when SNO can no longer bind to or be released from βCys93. Thus, the allosterically regulated βCys93 mediates hypoxic vasodilation by RBCs.
Reduction of SNO-Hb in Cys93A RBCs and impairment of hypoxic vasodilation. (A) SNO levels after treatment of RBCs from C93 control and C93A mice with CysNO (n = 10 for C93; n = 9 for C93A). (B) Baseline SNO in untreated RBCs from C93 and C93A mice (n = 14 each). (C) RBC-mediated hypoxic vasodilation of isolated aortic rings in vitro. SNO-loaded RBCs in A were added to bioassays under hypoxia (1% O2) or normoxia (20% O2) (n = 6 each). Data shown as mean ± SD. *P < 0.05, **P < 0.01 vs. C93, 2-tailed Student’s t test. (D) Representative aortic ring bioassay response to adding SNO-loaded C93 versus C93A RBCs (arrow) over time, in 1% O2 or 20% O2, normalized to 100% tension with phenylephrine. (E) Vasodilation in abdominal aorta in vivo at baseline and after aortic ligature release. Data shown as mean ± SEM. n = 11 C93 (4.0 ± 0.3 months); n = 10 C93A (3.7 ± 0.4 months). *P < 0.05 vs. C93, 2-way ANOVA. (F) Peak vasodilation of aorta, calculated from peak response from each mouse. Data shown as mean ± SD. *P < 0.05 vs. C93, 2-tailed Student’s t test. (G) Representative short axis M-mode images of abdominal aorta depicting aortic dilation (and impairment in C93A). Dashed lines represent vessel wall positions at diastole, for calculating diameter. Vertical scale bar: 2 mm; horizontal: 50 ms. (H) Mean blood flow increase after aortic ligature release. Data shown as mean ± SEM. n = 16 C93 (3.8 ± 0.9 months); n = 17 C93A (3.8 ± 0.7 months). *P < 0.05 vs. C93, 2-way ANOVA. (I) Peak flow increase, calculated from each mouse. Data shown as mean ± SD. *P < 0.05 vs. C93, 2-tailed Student’s t test. (J) Representative abdominal aortic blood flow curves.
Peripheral vasodilation by Hb βCys93 in vivo. In vitro bioassays with isolated aortic rings and static, dilute RBCs have limitations. To assess the role of Hb βCys93 in regulating hypoxic vasodilation in vivo, we performed 2 types of experiments based on classic reactive hyperemia paradigms (30, 31) but using abdominal aorta in situ. First, flow through the abdominal aorta was blocked temporarily by ligating the abdominal aorta for 5 minutes to create tissue hypoxia, the ligature was released, and the diameter of the abdominal aorta upstream of the ligation site was measured at diastole in real time using ultrasonography. In this model of reactive hyperemia, the diameter of the aorta underwent a transient increase after hyperemic flow had normalized, and the extent of this increase, shown as dynamic vasodilation, was significantly reduced in the C93A mice (Figure 1E). The overall dilation of the abdominal aorta increased by approximately 8.8% over the basal diameter in C93 (control) mice bearing wild-type human Hb (Figure 1F) compared with the diameter at the same location prior to occlusion. By contrast, the diameter increase in C93A mouse aorta was significantly blunted, at only approximately 4.4% over its basal diameter (Figure 1F); representative M-mode images for individual mice are shown in Figure 1G. This indicates that Hb βCys93-derived SNO within RBCs contributes about 50% of the vasodilation effect following temporary occlusion. The remaining half is attributed to local, shear force–induced endothelial NO production upon restored flow (flow-mediated dilation, FMD; refs. 32, 33), which has often been assumed to be responsible for the full effect but without empirical evidence. Moreover, it had not been previously possible to discern the role of endothelium versus RBCs because endothelial NOS (eNOS) inhibition reduces levels of SNO-Hb and RBC SNO (13, 34). We conclude that vasodilation following occlusion is evidently mediated by both RBCs and endothelium, the former stimulated by hypoxia (35) and the latter by shear.
In a second experiment, we directly measured blood flow through the abdominal aorta downstream of the ligation site using an ultrasound flow probe. Increases in flow following ligature release result from microcirculatory vasodilation downstream, which provides a surrogate measure of NO vasodilatory activity. Generally, very small diameter increases distributed across the microcirculation result in marked increases in flow, as flow is a function of (radius)4. As with abdominal aorta diameter, we measured blood flow continually through the cardiac cycle and calculated systolic peak and mean blood flow for the first second of each 15-second window to determine dynamic flow increases. Mean flow increase is shown in Figure 1H. In C93 control mice, the mean aortic blood flow increased nearly 1.2-fold over flow measured prior to the blockade, but in the C93A mice, this flow was significantly blunted, reaching only approximately 0.8-fold (Figure 1I); representative traces for individual mice are shown in Figure 1J. Similarly, peak flow at systole was increased 54% over basal in control mice, but this was significantly reduced to 31% in C93A mice (Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.155234DS1). We conclude that about 50% of the hyperemic flow increase, reflecting microcirculatory vasodilation following hypoxia, is mediated by Hb βCys93 within RBCs.
Right ventricular hypertrophy in young and aged animals in the absence of βCys93. Loss of ability to carry and release SNO from Hb βCys93 leads to tissue hypoxia that is compensated in part by increased cardiac workload (7). Additionally, hypoxia may result in pulmonary hypertension, evidenced by right ventricular hypertrophy. We found total heart weight to body weight ratio to be significantly elevated in both young adult and old C93A mice, compared with C93 control mice (Figure 2A). Further, both right ventricle weight (Figure 2B) and left ventricle weight (Figure 2C) as a ratio to body weight were elevated significantly in C93A mice. However, C93A mice were lighter than C93 controls, sufficiently so to potentially confound interpretation when aged (C93 26.86 ± 2.76 g and C93A 25.04 ± 2.00 g for young mice, P < 0.01; and C93 42.82 ± 7.91 g and C93A 34.23 ± 4.43 g for aged mice, P < 0.01). More importantly, the ratio of right ventricle weight to the weight of the left ventricle plus septum (Figure 2D) was significantly elevated in both young and aged C93A mice (independent of body weight differences). Thus, both young and aged C93A mice showed signs of right heart hypertrophy, suggesting increased pulmonary vascular tone.
Right ventricular and pulmonary artery signs of pulmonary arterial hypertension in C93A mice with age. (A) Total heart weight (HW) to body weight (BW) ratio in young and in aged C93A versus C93 mice. (B) Right ventricle (RV) to BW ratio in young and in aged C93A versus C93 mice. (C) Left ventricle (LV) to BW ratio in young and in aged C93A versus C93 mice. (D) RV to LV + septum weight (LV+S) ratio in young and in aged C93A versus C93 mice. (E) Pulmonary artery (PA) blood flow velocity-time integral (VTI) in young and in aged C93A versus C93 mice. (F) PA diameter in young and in aged C93A versus C93 mice. (G) Mean velocity of PA blood flow in young and in aged C93A versus C93 mice. (H) Peak velocity of PA blood flow in young and in aged C93A versus C93 mice. For panels A–D, young mice (n = 31 C93, 3.4 ± 0.6 months of age, and n = 36 C93A, 3.4 ± 0.4 months of age) and aged mice (n = 28 C93, 19.5 ± 1.8 months of age, and n = 21 C93A, 19.7 ± 3.1 months of age) were assessed. For panels E–H, young mice (n = 16 C93, 3.8 ± 1.3 months of age, and n = 19 C93A, 3.0 ± 0.8 months of age) and aged mice (n = 15 C93, 20.9 ± 1.6 months of age, and n = 23 C93A, 21.8 ± 1.2 months of age) were assessed. Differences were assessed using Student’s t test (2 tailed). *P < 0.05, **P < 0.01 C93A vs. C93, for young or aged animals compared separately.
Age-related signs of pulmonary hypertension in the absence of βCys93. While young adult C93A mice had normal pulmonary artery blood flow velocity-time integral (VTI), pulmonary VTI was significantly reduced in aged C93A mice (Figure 2E). Likewise, pulmonary artery diameter was normal in young C93A mice but was significantly increased in aged C93A mice (Figure 2F). Mean and peak blood velocity in the pulmonary artery were both significantly reduced in both young and aged C93A mice (Figure 2, G and H).
Pulmonary hemodynamics in young and aged animals. In contrast to the prior (7) and above data in conscious mice, in young anesthetized mice at baseline (Supplemental Table 1), no significant changes were apparent, while in aged anesthetized C93A mice, systolic, diastolic, and mean pulmonary arterial pressures and right ventricular systolic pressure trended toward increases versus control mice (Supplemental Figure 2, A–D). Measures of right ventricular function in aged versus young mice, including dP/dt max, dP/dt min, contractility index, and average dP/dt over the isovolumic relaxation period (IRP average dP/dt) showed a similar trend (Supplemental Figure 2, E–H). Taken together with right ventricular hypertrophy and pulmonary artery dilation in aged C93A animals at baseline, and age-related reductions in pulmonary blood flow velocity in conscious C93A mice (Figure 2), our results suggest that with activity, stress, and aging, abnormal pulmonary vascular reactivity and/or right ventriculo-arterial coupling results in right-sided dysfunction at baseline, while in unstressed animals under anesthesia, right-sided changes are attenuated.
Pulmonary hypertension with chronic hypoxia in the absence of βCys93. Patients with chronic hypoxia-induced pulmonary hypertension have low SNO-Hb, and their RBCs show impaired vasodilatory responses in vitro (36). We therefore exposed C93A mice to chronic hypoxia. Based on our previous study demonstrating that mutant mice survived acute exposure to 10% hypoxia but succumbed quickly at 5% O2 (7), we housed young mice at 10% O2 for 4 weeks (all mice survived) and examined cardiac function and pulmonary artery pressure using echocardiography and invasive catheterization, respectively. Pulmonary artery diameter was significantly larger in C93A mice after chronic hypoxia (Figure 3A), and pulmonary VTI was diminished (Figure 3B). The mean and peak velocity of blood ejection from the right heart were also reduced in chronically hypoxic C93A mice, consistent with right heart dysfunction subsequent to pulmonary arterial hypertension (Figure 3, C and D). Indeed, systolic, diastolic, and mean pulmonary artery pressures were all significantly elevated in C93A mice versus control mice (Figure 3E). Furthermore, right ventricular systolic pressure was significantly higher in C93A mice than in C93 (Figure 3F). Analysis of pressure-time curves revealed that right ventricle dP/dt max, dP/dt min, contractility index, and average dP/dt over IRP (IRP average dP/dt) were all significantly elevated in C93A mice (Figure 3, G–J). However, the time constant of relaxation (τ) did not differ (Figure 3K). The right ventricles of C93A mice showed evidence for increased fibrosis compared with C93 controls (Figure 3, L–N). Overall, these changes are indicative of pulmonary arterial hypertension with right ventricular dysfunction.
Pulmonary hypertension and right ventricular dysfunction in hypoxic C93A mice. All comparisons are between hypoxic C93A (red bar) versus C93 (green bar) mice. (A) Pulmonary artery diameter. (B) Pulmonary artery blood flow VTI. (C) Mean velocity of pulmonary artery blood flow. (D) Peak velocity of pulmonary artery blood flow. (E) Systolic pulmonary arterial pressure (sPAP), diastolic pulmonary arterial pressure (dPAP), and mean pulmonary arterial pressure (mPAP). (F) Right ventricular systolic pressure (RVSP). (G) Maximal rate of change in right ventricular (RV) pressure (dP/dt max). (H) Minimal rate of change in RV pressure (dP/dt min). (I) RV contractility index. (J) RV average dP/dt over isovolumic relaxation period (IRP average dP/dt). (K) Time constant of relaxation (τ). (L–N) RV fibrosis in young mice housed in 10% O2 for 4 weeks, visualized by Picrosirius red staining. C93 lacking fibrosis (L, representative of 3 tested), C93A displaying developing fibrotic areas (M, representative of 4 of 5), and C93A with fibrosis (N, observed in 1 of 5). * indicates fibrotic areas; scale bar: 50 μm. For all quantitative panels, data are presented as mean ± SD. Young mice exposed to 10% O2 for 4 weeks: for panels A–D, n = 19 C93, 4.9 ± 1.3 months of age, and n = 23 C93A, 4.4 ± 1.1 months of age); for panels E–K, n = 19 C93 mice (4.9 ± 1.3 months of age) and n = 20 C93A mice (4.3 ± 1.1 months of age). Differences were assessed using Student’s t test (2 tailed). *P < 0.05, **P < 0.01 C93A vs. C93.
As a confirmatory measure of the effects of chronic hypoxia, we compared young mutant C93A mice under normoxia with young C93A mice housed under 10% O2 for 4 weeks (using data shown in Figures 2 and 3 and Supplemental Figure 2). Pulmonary artery diameter increased in young C93A mice after chronic hypoxia (Supplemental Figure 3A). Pulmonary artery blood flow VTI, mean blood velocity, and peak blood velocity all were significantly reduced in young C93A mice (Supplemental Figure 3, B–D). Further, systolic, diastolic, and mean pulmonary artery pressures were increased in young C93A mice versus controls (Supplemental Figure 3, E–G), and right ventricular pressures, dP/dt max, dP/dt min, and average dP/dt over the IRP were also increased (Supplemental Figure 3, H–L). Finally, there was significant hypertrophy of the right ventricle compared with the left ventricle and of the total heart compared with body weight (Supplemental Figure 4, A–D). Thus, pulmonary hypertension with cor pulmonale is induced by hypoxia independently of age in C93A mice.
Left heart function with aging and chronic hypoxia. While left ventricular ejection fraction and fractional shortening were normal at baseline in young adult C93A animals (Supplemental Table 1), these functional measures were reduced in aged animals (Supplemental Figure 5, A and B). Further, chronic hypoxia led to modestly reduced left ventricular ejection fraction and fractional shortening in young C93A mice versus control C93 mice (Supplemental Figure 5, C and D; and Supplemental Figure 6, A and B), accompanied by increases in left ventricular end-systolic and -diastolic diameters and volumes (Supplemental Figure 5, E–H; and Supplemental Figure 6, C–F). Parameters that did not show differences in young C93A mice between normoxia and chronic 10% O2 included left ventricular end-systolic volume, inner diameter at systole, ventricular end-diastolic volume, and inner diameter at diastole (Supplemental Figure 6, C–F).
Increased mortality in the absence of βCys93 under chronic hypoxia. We assessed survival of mice housed chronically under 10% O2. The C93 wild-type mice all survived through 50 days but were all dead by 154 days, with a mean survival time of 87 days (Figure 4, A and B). In contrast, C93A mice died much sooner, with the first mouse dying at 25 days and the last at 78 days, with a mean survival time of 51 days (Figure 4, A and B). This is consistent with mice lacking ability to mediate hypoxic vasodilation of peripheral and pulmonary vasculature.
Increased mortality during chronic hypoxia in C93A mutant mice. (A) C93 and C93A mice were housed under 10% O2, and survival was assessed daily. The percentage of mice surviving each day is shown in a Kaplan-Meier plot. Curves were compared using log-rank (Mantel-Cox) and Gehan-Breslow-Wilcoxon test. **P < 0.01 C93A (n = 15, 7.6 ± 0.5 months of age) vs. C93 (n = 21, 7.8 ± 0.8 months of age). (B) Average days of survival at 10% O2, plotted as mean ± SD, with individual data points shown. **P < 0.01 C93A vs. C93 by Student’s t test (2 tailed).