|Home | About | Journals | Submit | Contact Us | Français|
Vascular tissues produce carbon monoxide (CO) via HO-dependent and HO-independent mechanisms; the former in tandem with biliverdin and iron and the latter as a lone product. CO has been shown to function as both a vasoconstrictor and vasodilator, however, factors that dictate the vasoregulatory phenotype of this gas are unknown.
Herein, we demonstrate that CO-mediated vasoconstriction is mechanistically linked to enhanced reactive oxygen species (ROS) production that masks vasodilatory pathways.
Sprague Dawley rat interlobar and interlobular arteries were examined in terms of superoxide (O2-) generation and vascular reactivity in the absence and presence of antioxidants. Both authentic CO and the CO-releasing molecule (CORM)-3 constricted renal arteries and increased O2- production in a dose-dependent manner. Antioxidants tempol, ebselen and deferoxamine inhibited CO-induced O2- production and converted CO from constrictor to dilator. CO-induced O2- generation was found to involve the activity of multiple oxidases including nitric oxide synthase, NADPH-oxidase, xanthine oxidase and complex IV of the mitochondrial electron chain. Furthermore, inhibition of these enzymes converted CO from constrictor to dilator. Similarly, biliverdin and bilirubin inhibited CO-induced O2- production and vasoconstriction, allowing for a vasodilatory response to CO to be expressed. CO-induced vasoconstriction was dependent on a non-thromboxane agonist of the thromboxane receptor, while vasodilatory mechanisms of CO relied on the activation of soluble guanylate cyclase and calcium-gated potassium channels.
CO-induced vasoconstriction involves the generation of ROS which, when negated, allows for the expression of vasodilatory pathways which are masked by the primary oxidative stress response to this gas.
Vascular tissues generate carbon monoxide (CO) via heme oxygenase (HO)-dependent and HO-independent pathways. CO is liberated during heme metabolism by HO, generating equimolar quantities of biliverdin and iron, while HO-independent sources of CO include oxidation of organic molecules and peroxidation of membrane lipids 1, 2. Following seminal reports that described a vasorelaxant effect of CO in the hepatic microcirculation, evidence supporting a vasodilatory role of CO has accumulated 3, 4. For example, pharmacological inhibition of HO product generation, presumably leading to a reduction in CO production, increases renal vascular resistance and constricts pressurized gracilis muscle arterioles 5, 6. The functional role of HO in the vasculature, however, is not synonymous with the biological effects of CO; as CO exerts pleiotropic actions in terms of vasoregulation, functioning as both a vasodilator and vasoconstrictor.
Exogenous CO was first shown to dilate rat coronary arteries and proposed to act in a manner similar to nitric oxide (NO); that is, via activation of soluble guanylate cyclase (sGC) 7, 8. The role of CO in maintaining vascular tone has been proposed to be enhanced under conditions in which NO bioavailability was reduced, a setting in which CO production was enhanced 5, 9. CO has also been shown to hyperpolarize vascular smooth muscle via activation of calcium-activated potassium channels (KCa) 10, 11. Finally, CO was demonstrated to reduce constrictor responses to phenylephrine and 20-HETE, while reducing the synthesis of the vasoconstrictor agent endothelin 12, 13.
While the majority of data supports a pro-dilatory role for CO, evidence for the existence of vasoconstrictor effects of CO has also accumulated. CO binds sGC with 30-100 fold less potency than NO, thus the ability of this gas to efficiently activate sGC in physiological settings has been in question 14. Moreover, CO has been proposed to inhibit NO synthesis and/or actions, interfering with the expression of vasodilatory mechanisms mediated by NO 15, 16. Furthermore, both endogenous and exogenous CO were found to cause constriction of isolated rat gracilis muscle arterioles 15. To date, the mechanism of CO-induced vasoconstriction has not been elucidated.
Relevant to the apparent discrepancy as to whether CO functions as a vasodilator or vasoconstrictor are findings that vascular endothelial cells exposed to CO suffer oxidative stress 17. As reactive oxygen species (ROS) have been implicated in pathways associated with direct vasoconstriction as well as impairment of vasodilation, we hypothesized that CO of vascular origin may promote vasoconstriction in a ROS-dependent manner 18-20. The present study was designed to investigate the vasoregulatory mechanisms associated with CO by examining the relationship between CO and ROS generation in rat renal arterial vessels.
Male Sprague Dawley rats (250-300 g, Charles River, Wilmington, DE) were used in compliance with protocols approved by the Institutional Animal Care and Use Committee.
Lucigenin chemiluminescence was used to measure O2- production by freshly isolated renal interlobar arteries as per a protocol modified from Omar et al.21. In an effort to further confirm levels of O2- production by vessels, an intravital fluorescence microscopy detection technique was utilized. O2- production by renal interlobar arteries was assessed by quantifying a fluorescent signal arising from oxidized dihydroethidium (DHE) using a Nikon epifluorescence inverted microscope (Diaphot) outfitted with a SIT camera (Hamamatsu, Japan) as previously described 22.
Freshly isolated renal interlobar arteries were exposed to authentic CO (1-μmol/l) for 1 h at 37°C in gas-sealed vials. Supernatant was collected for measurement of H2O2 using Amplex red as previously described 23. As an index of ONOO- formation, vessels were homogenized in K+ phosphate buffer, separated by SDS–PAGE and probed with an anti-nitrotyrosine antibody (Millipore, 1:5000).
Isoprostane production by renal interlobar arteries was assessed using the Cayman Chemical STAT-8-isoprostane EIA kit (CN-500431).
Renal interlobular arteries were dissected into segments 1-2 mm in length and mounted on a pressure myograph (model CH/200/Q, Living System Instrumentation; Burlington, VT) as previously described 12. CO or the CO-releasing molecule (CORM)-3 was added to the superfusion buffer and changes in internal diameter were recorded 24. The response to CO was studied in vascular preparations pretreated and not pretreated with tempol, ebselen, deferoxamine, N (G)-nitro-L- arginine methyl ester (L-NAME), apocynin, allopurinol, rotenone, carbonyl cyanide m-chlorophenylhydrazone (CCCP), pegylated superoxide dismutase (SOD), pegylated catalase, polyethylene glycol monomethyl ether, biliverdin, bilirubin, uric acid, indomethacin or a thromboxane receptor (TP) antagonist (SQ29548). Finally, the effect of CO in vessels pretreated with tempol was examined in the absence and presence of an inhibitor of sGC (ODQ) or the KCa channel blockers tetraethylammonium (TEA) and iberiotoxin (IBTX). Data are reported as internal diameter (ID) or change in ID (μm).
SOD and catalase activity was quantified in vascular homogenates from vessels treated with and without CO (1-μmol/l) at 37°C for 1 h using commercial kits (Cayman Chemical 706002 and 707002, respectively).
Data are expressed as mean ± SEM for the given number (n) of experiments. Results were analyzed by student's t-test or by two-way ANOVA and the Newman-Keuls post hoc test was performed. The null hypothesis was rejected at p< 0.05.
Initial experiments were conducted to establish the vasoregulatory phenotype of CO in rat renal arteries under basal conditions. Shown in Figure 1A, exposure to authentic CO or the CO-releasing molecule CORM-3 reduced (p<0.05) internal diameter of interlobular arteries in a dose-dependent manner. Comparatively, vasoconstrictor response to 0.1 and 1.0-μmol/l authentic CO correlated to 10 and 100-μmol/l CORM-3, respectively. These findings are consistent with gas chromatography/mass spectrometry detection of CO release by CORM-3 in physiological buffers, in which about 1% of detectable CO recovery was observed (data not shown). Importantly, inactivated CORM-3 (iCORM-3), which does not actively release CO, had no effect on internal diameter of vessels.
As shown in Figure 1B, freshly dissected vessels generate O2-ex vivo. Exposure of vessels to CORM-3 led to a dose-dependent increase in lucigenin-detectable O2- production that was about 2-fold higher than baseline at 100-μmol/l CORM-3 (p<0.05). DHE fluorescence, an indicator of O2− production, was found to be stable in isolated renal arteries under basal conditions. Consistent with other detection methods, administration of CORM-3 led to a gradual increase in DHE-detectable O2- production over a period of 1-3 m before stabilizing and remaining elevated during a plateau phase (Figure 1C). Importantly, iCORM-3 was found to have no effect on vascular O2- generation by either method (Figures 1B and 1C). Complementary studies confirmed CO was the bioactive molecule as authentic CO (1.0-μmol/l) increased O2- generation by vessels from 62±3 to 99±9 cpm/ug protein (p<0.05) as measured by lucigenin chemiluminescence.
While O2- anion may be the initial ROS formed following exposure of renal arteries to CO, it is unclear whether other reactive intermediates are formed and/or propagate the vasoregulatory effects of CO. To this end, we examined the production of H2O2 and nitrotyrosine, the latter an indicator of ONOO- formation, in vessels exposed to CO. We observed a slight increase in H2O2 production that did not reach statistical significance, by renal interlobar arteries exposed to CO for 1 h (3.54±1.1 vs. 4.56±1.0 nmol/mg protein; n=6 and 7, respectively). Nitrotyrosine levels were undetectable under basal conditions and were unaltered following exposure to CO (data not shown).
To further investigate the impact of oxidative stress on the vasoregulatory actions of CO, antioxidants known to preferentially target various ROS were applied. Tempol (1-mmol/l), an SOD mimetic, prevented CORM-3-induced increase in O2- by renal arteries (Figure 2A). Ebselen (1-μmol/l) and deferoxamine (500-μmol/l), a glutathione peroxidase mimetic and an iron chelator, respectively, similarly prevented (p<0.05) CORM-3 from promoting an increase in O2-. A link between CO-induced ROS generation and vasoregulation was established by the observations that tempol, ebselen and deferoxamine, prevented CO-induced vasoconstriction and uncovered a vasodilatory response to this gas (Figure 2B). A pegylated form of SOD (500U/ml) was also effective at converting the response to CO from constriction to dilation (Figure 2B). The capacity of ebselen and deferoxamine to convert CO from constrictor to dilator, however, may implicate downstream ROS in these processes. Pretreatment of vessels with pegylated catalase (1000U/ml) similarly converted the response of CO from constriction to dilation (Figure 2B). Polyethylene glycol monomethyl ether (125 ug /ml), used as a control for pegylated SOD and catalase, had no effect on CO-mediated vasoconstriction (Figure 2B). Collectively, these findings implicate O2- and downstream intermediary reactive species in the implementation of the vasoregulatory actions of CO.
To investigate the contribution of various oxidases to the implementation of CO-induced oxidative stress and vasoactivity, the effects of this gas were studied in the absence and presence of inhibitors of the major sources of O2- in the vasculature; nitric oxide synthase (NOS), (NADPH)-oxidase, xanthine oxidase (XO), and the mitochondria. As shown in Figure 3, inhibition of NOS, NADPH-oxidase and XO with L-NAME (1-mmol/l), apocynin (100-μmol/l) and allopurinol (100-μmol/l), respectively, inhibited CORM-3-induced elevation in O2- levels. Importantly, inhibition of NOS, NADPH-oxidase and XO also converted the response to CO from vasoconstriction to vasodilation (Figure 4).
The mitochondrial respiration chain has been reported to differentially generate O2- depending on the specific activation or inhibition of particular oxidases. Rotenone (10-μmol/l), an inhibitor of complex I in the mitochondrial respiration chain did not alter O2- production or internal diameter in response to CORM-3 and CO (Figure 3 and and4D),4D), respectively. On the other hand, CCCP (5-μmol/l), an inhibitor of complex IV minimized (p<0.05) CORM-3-induced O2- production by over 70% (from 16±6.4 to 76.0±18.6 in controls and from 30±2.4 to 46±5.7 cpm/μg protein in vessels pretreated with CCCP) and converted CO from constrictor to dilator at 1000-nmol/l CO (Figure 4D).
Consistent with previous reports demonstrating the antioxidant properties of biliverdin and bilirubin, 100-nmol/l of either bile pigment inhibited the CORM-3-induced increase in O2- production (Figure 5A). Consistent with our initial findings using other antioxidants, biliverdin and bilirubin converted the response of CO from vasoconstriction to vasodilation in a concentration-dependent manner (Figures 5B and 5C, respectively). Another endogenous antioxidant, uric acid (200-μmol/l) also converted the response of renal arteries to CO from vasoconstriction to vasodilation (-3.0±0.0 vs. 4.8±0.9 and -9.5±0.7 vs. 12.3±0.3 ID (μm) at 100- and 1000-nmol/l CO, respectively).
Pretreatment of renal interlobular arteries with indomethacin (1-μmol/l) did not alter the vasoconstrictor response to CO, suggesting cyclooxygenase (COX) metabolites do not contribute to vasoconstrictor mechanisms associated with this gas (Figure 6A). On the other hand, blockade of the TP receptor (SQ29548; 1-μmol/l) prevented (p<0.05) CO-induced vasoconstriction, implicating a non-thromboxane agonist of the TP receptor in mediating this response. Importantly, TP receptor blockade with SQ29548 had no effect on phenylephrine (1-μmol/l)-induced reduction in internal diameter of renal vessels (-25.1±15 vs. -27.3±1.0).
Oxidative stress promotes the formation of vasoconstrictor, non-enzymatic oxidation products of arachidonic acid known as isoprostanes 25. We found that incubation of freshly isolated renal interlobar arteries with CO (10-μmol/l) increased isoprostane formation from 201±47 to 393±60 pg/mg protein (p<0.05). Collectively, these observations suggest the contribution of an isoprostane to the vasoconstrictor actions of CO.
Pretreatment of renal interlobular arteries with tempol allows for the expression of a CO-induced vasodilatory response (Figure 6B). In this experimental setting, the dilatory action of CO was partially reduced (p<0.05) by ODQ (10-μmol/l) and completely prevented (p<0.05) by both TEA (1-mmol/l) and iberiotoxin (1-μmol/l). The combination of TEA and ODQ also prevented the vasodilatory actions of CO. In the absence of tempol, both ODQ and TEA sensitized (p<0.05) vessels to CO-induced vasoconstriction. ODQ enhanced CO-induced vasoconstriction from -5.0±0.6 to -7.3±0.6 μm and from -12.8±0.9 to -18.0±1.9 μm, at 100- and 1000-nmol/l CO, respectively. TEA enhanced CO-induced vasoconstriction from -5.0±0.5 to -13.0±1.2 μm and from -12.8±0.9 to -21.5±0.5 μm, at 100- and 1000-nmol/l CO, respectively.
We then investigated the capacity of CO to alter the enzymatic activities of two critical antioxidative enzymes. The activities of SOD and catalase were measured in homogenates from vessels exposed to CO (1-μmol/l) for 1 h. CO did not significantly alter the activities of SOD (2.25±0.57 vs. 1.39±0.04 U/mg protein in untreated and CO-treated, respectively) or catalase (22.4±3.3 vs. 27.5±1.4 nmol H2O2 conversion/mg protein/min in untreated and CO-treated, respectively).
Vascular tissues generate CO which, depending on experimental conditions, has been implicated in mediating vasoconstriction as well as vasodilation 7, 8, 11, 15, 16. We report here for the first time that both the vasoconstrictor and vasodilatory responses to CO are critically conditioned by redox mechanisms. The vasoconstrictor action is linked to increased oxidant activity which promotes formation of isoprostanes. The vasodilatory action is linked to mechanisms involving sGC and KCa channels, and requires conditions that offset the pro-oxidant activity of CO to be expressed.
That CO and CORM-3 elicit constriction of isolated, pressurized, renal interlobular arteries is consistent with earlier reports that CO constricts pressurized gracilis muscle arterioles, an action attributed to inhibition of NO synthesis 15. The current study offers an alternative explanation to the vasoconstrictor action of CO that involves oxidative stress as a determinant for the generation of isoprostanes, which promote contraction of vascular smooth muscle and thus mediate the constrictor action of the gas. Consistent with early reports of pro-oxidant actions of CO in endothelial cells and brain, we found that both authentic CO and CORM-3 cause increase of O2- levels in renal interlobar arteries 2, 17, 26 This action of CO may entail activation of multiple oxidases, since the CO-induced elevation of vascular O2- levels was blunted or minimized in arterial vessels pretreated with L-NAME (NOS inhibitor), apocynin (NADPH oxidase assembly inhibitor), allopurinol (XO inhibitor), or CCCP (mitochondrial oxidase complex IV inhibitor). Relevant to this point, CO is capable of binding and inhibiting NOS 27, 28. It is unclear, however, whether NOS inhibition by CO is accompanied by uncoupling of the enzyme with resultant generation of O2- as occurs in the presence of tetrahydrobiopterin deficiency. Cytochrome c oxidase, a constituent of mitochondrial oxidase complex IV, is also amenable to inhibition by CO with attendant generation of ROS production 29, 30. On the other hand, in respiratory epithelial cells, CO was reported to inhibit rather than to stimulate NADPH-oxidase dependent generation of O2-30.
Given our finding that multiple oxidases appear to contribute to CO-induced elevation of vascular O2-, one should consider the possibility that a feed-forward cycle links the initial surge in O2- production via NOS uncoupling and/or cytochrome c oxidase inhibition, with the secondary activation of multiple oxidases by downstream intermediate pro-oxidant molecules such as H2O2, OH- radical and ONOO-. These are volatile molecules which can rapidly cycle between species via pathways such as the dismutation of O2- to H2O2, the conversion of H2O2 and nitrite to ONOO-, the Fenton reaction-mediated generation of OH- radical, or the combination of O2- and NO to form ONOO-31. Previous studies offered evidence of feed-forward propagation of oxidative stress in the vasculature via H2O2-dependent activation of NADPH oxidases, XO, eNOS uncoupling, and augmentation of intracellular iron 32. Despite a previous report that CO increased intracellular H2O2 production in the brain, the present observations that incubation of arterial vessels with CO failed to result in a significant augmentation of H2O2 or nitrotyrosine levels (an index of ONOO-), argues against the notion that H2O2 and ONOO- are implicated in the propagation of the oxidative stress serving to sustain CO-induced augmentation of vascular O2- levels 33. However, the possibility that CO-induced oxidative stress is sustained by intermediate reacting molecules via activation of multiple oxidases fits well with our finding that pretreatment of arterial vessels with ebselen (a glutathione peroxidase mimetic which also scavenges ONOO-), or deferoxamine (a chelator of iron and other transition metals which limits OH- radical generation), prevents CO from increasing O2-. That deferoxamine blocked CO-induced increase in vascular O2-suggests that iron or other transition metals play a role in ROS propagation initiated by CO. Free iron can be deleterious to cells due to its participation in the Fenton reaction which involves H2O2 and yields OH- radical, a highly reactive oxidant toxic to biological molecules 34. That deferoxamine did not alter basal vascular levels of O2- may be taken to indicate that under resting conditions metal-driven reactions promoting oxidative stress are nominal.
We have also given consideration to the possibility that CO-induced elevation of vascular O2- levels results from an inhibitory action of the gas on antioxidant enzymes such as catalase and SOD. Catalase is a heme-containing enzyme which has been suggested to be a target for CO, leading to inhibition of its catalytic activity 35. This is not the case in our study, as treatment with CO did not alter catalase activity measured in freshly isolated arterial vessels acutely exposed to the gas. Treatment with CO was also without effect on the activity of SOD measured in isolated arterial vessels. Recently, CO was reported to inhibit cystathionine beta-synthase 36. Inhibition of this enzyme may overwhelm endogenous anti-oxidative defense mechanisms via excessive homocysteine accumulation and/or a reduction in intracellular glutathione.
Linking the increase in O2- production to the vasoconstrictor actions of CO in renal arteries, we demonstrate that CO-induced vasoconstriction is converted to dilation by exogenous antioxidants and inhibition of intracellular sources of O2-. That a reduction in O2- levels prevents CO-mediated constriction, confirms a role for ROS in the constrictor response. However, the ability of antioxidants to convert the actions of exogenous CO from constrictor to dilator, suggest that ROS may be simultaneously preventing the expression of vasodilatory pathways. In the present study, dilation to CO in the presence of antioxidants was found to be mediated by activation of sGC and KCa channels, consistent with reports in other resistance vessels 11, 37. Interestingly, sGC and K channels have been shown to be negatively regulated by ROS. BKCa in rat cerebral arterial smooth muscle cells is reversibly inhibited by ONOO- while ROS-mediated heme oxidation impairs sGC activation in blood vessels 38, 39. Thus, antioxidant intervention may provide a dual impetus to both antagonize pro-constrictor mechanisms, as well as to relieve inhibitory influences on vasodilator pathways (e.g. KCa, sGC) associated with oxidative stress.
The mechanism associated with CO-induced vasoconstriction, which appears to involve the generation of O2- and potentially downstream ROS, has not been elucidated to date. ROS are known to lead to the generation of non-enzymatic metabolites of arachidonic acid known as isoprostanes that are capable of constricting vessels via activation of the TP receptor 25. As CO was found to enhance vascular isoprostane formation, we hypothesized that isoprostanes may be downstream mediators of CO-induced vasoconstriction. That a TP receptor antagonist, but not indomethacin, inhibited vasoconstriction to CO, provides seminal evidence that isoprostane-mediated activation of the TP receptor mediates CO-induced vasoconstriction.
Paradoxically, we observe vasoconstriction in response to exogenous CO, yet previous work has demonstrated that a reduction in endogenous CO formation via the inhibition of HO activity similarly promotes vasoconstriction 5. These findings suggest that endogenously produced CO functions as a vasodilator, while exogenous CO functions as a vasoconstrictor. As HO-mediated heme metabolism concurrently generates CO and endogenous antioxidants biliverdin/bilirubin, we hypothesized that co-generation of biliverdin/bilirubin functions to neutralize pro-oxidant/-constrictor effects of endogenously formed CO. In fact, the pro-oxidant and pressor effects associated with angiotensin II- and DOCA salt-induced hypertension were reduced by elevated bilirubin levels 40, 41.
Consistent with previous reports of bile pigments functioning as antioxidants, exogenous biliverdin and bilirubin inhibited O2- production and vasoconstriction in response to CO. Concentrations of biliverdin and bilirubin used in the present study (10-1000-nmol/l) were consistent with previous work and believed to be in a physiological range, well below plasma concentration (5-17μmol/l) 42. Ultimately, intracellular concentrations of biliverdin and bilirubin are contingent on lipid/water solubility, binding proteins, uptake/diffusion and intracellular heme metabolism. Numerous mechanisms have been proposed regarding the antioxidant capacity of biliverdin and bilirubin. Perhaps the most impressive effects of bilirubin in terms of cellular protection, is its ability to safeguard against lipid peroxidation 42. Plasma bilirubin may function as a chain-breaking antioxidant, acting on secondary oxidants (e.g. ONOO-) involved in the propagation of ROS-mediated damage 43, 44. Bilirubin was additionally shown to inhibit the activation process of NADPH oxidase, a major source of vascular O2-, and inhibit protein kinase C activity-dependent ROS production 45, 46. Furthermore, bilirubin may undergo a “recycling” process whereby biliverdin is converted to bilirubin via the enzyme biliverdin reductase, followed by bilirubin oxidation by ROS to biliverdin 47. It should also be noted that other endogenous antioxidants may provide an alternative to bile pigments, such as in a setting in which CO is formed independently of HO. For example, uric acid has been previously shown to suppress ROS production in response to angiotensin II and we demonstrate herein the ability of this compound to convert the response of CO from constrictor to dilator 48. Collectively, these findings support the notion that the antioxidant capacity of endogenously formed compounds (biliverdin, bilirubin and uric acid) function to unmask dilatory mechanisms associated with CO.
It may be anticipated that acute administration of CO would elicit vasodilation, as biliverdin/bilirubin would be present when redox balance is in equilibrium, however, the current studies were conducted in an isolated, non-blood perfused system. Plasma bilirubin has been shown to have a large capacity to combat oxidative stress, therefore, lack of this pigment may reduced the antioxidant capacity of the vessel wall and allow for CO to elicit vasoconstriction 42, 44. In an intact system, we have previously demonstrated that infusion of CO into the renal artery of rats does not alter blood pressure; therefore the effects of CO may be largely dependent on environmental redox balance or in this case, experimental conditions49.
The debate as to whether CO of vascular origin functions as a vasodilator or vasoconstrictor has been fueled by conflicting reports in the literature. The role of HO in vasoregulation has oft focused on CO as the quintessential bioactive product of heme metabolism; however it may be prudent to consider that concurrent generation of biliverdin/bilirubin is critical in dictating the vasoregulatory phenotype of CO. In as much, these considerations may be applicable to systems outside of the vasculature, a concept that is consistent with the work of other investigators demonstrating synergistic actions of CO and biliverdin 50. HO-independent sources of CO may also be physiologically relevant as CO formed as an isolated product likely increases ROS production, potentially leading to vasoconstriction. In conclusion, this study demonstrates for the first time that CO constricts renal arteries in a ROS-dependent manner which when antagonized, allows for vasodilatory pathways associated with CO to become unmasked.
We thank Dr. Kim D. Lamon for his critical review of the manuscript.
Sources of Funding: This work was supported by NIH grants HL034300 and HL018579.