Perivascular Superoxide Anion Contributes to Impairment of Endothelium-Dependent Relaxation
Role of gp91phox
Background— Like endothelial and smooth muscle cells, vascular adventitial fibroblasts contain a substantial NAD(P)H oxidase superoxide anion (O2−)-generating system activated by angiotensin II (Ang II). Based on the ability of nitric oxide (NO·) to diffuse rapidly through tissue and the fast reaction rate of NO· and O2−, we postulated that the interaction between NO· and adventitial NAD(P)H oxidase–derived O2− contributes to impairment of endothelium-dependent relaxation (EDR).
Methods and Results— C57Bl/6 mouse abdominal aortas were simultaneously perfused intraluminally and suffused adventitially with physiological buffer at 37°C. After constricting the vessels with phenylephrine, an acetylcholine dose-response curve was obtained while monitoring changes in diameter by videomicroscopy. Endogenous O2− was increased by treating the adventitial side of the aortas with Ang II (10 pmol/L), leading to impairment of EDR. EDR impairment was reversed by adventitial suffusion of superoxide dismutase (SOD) of aortas from wild-type mice. Ang II–treated aortas from gp91phox−/− mice, which lack significant adventitial O2−, exhibited greater EDR and were not affected by SOD. Adventitially suffused SOD failed to penetrate the media, indicating that the effects of SOD were localized to the adventitia. Adventitial application of the O2−-generating system xanthine/xanthine oxidase or the potent NO scavenger oxyhemoglobin impaired EDR.
Conclusions— O2− derived from adventitial gp91phox-based NAD(P)H oxidase contributes to impairment of the action of endothelium-derived NO·.
Received May 15, 2002; revision received August 13, 2002; accepted August 16, 2002.
In many vascular studies, the adventitia is damaged or removed. The simple removal of the blood vessel from the animal is likely to injure the adventitia, and during dissection the adventitia is often reduced or excluded by attempts to remove adherent adipose tissue. Moreover, endothelial removal commonly involves rolling a vessel ring using forceps pressed on its luminal surface, which is likely to cause adventitial or medial damage. These and other manipulations are likely to skew assessment of changes in vascular tone under both normal and pathological conditions, and significant variation in vascular responsiveness is likely to result depending on the degree of damage.
Numerous studies have implicated superoxide anion (O2−) in hypertension.1,2⇓ O2− interferes with nitric oxide (NO·)- dependent vasodilatation3 and participates in endothelium-dependent constriction.4 It also inhibits basal cGMP formation in vascular smooth muscle cells as well as the increase in cGMP produced by S-nitroso-N-acetylpenicillamine.5 Although NO· has been considered highly reactive and ephemeral in nature, several theoretical and empirical reports suggest that it diffuses relatively unimpeded through tissue unless opposed by its two major scavengers, O2− and oxyhemoglobin.6,7⇓
An inducible phagocyte-like NAD(P)H oxidase has been reported in the endothelium8–11⇓⇓⇓ and smooth muscle12,13⇓ and has been cited in the regulation of NO· bioactivity. Several key studies described a functional oxidase in these cells and implicated their contribution to impaired endothelial function.9,14,15⇓⇓ Because these O2− sources are near the site of eNOS-derived NO·, it is broadly accepted that they interfere with the actions of endothelium-derived NO·. We reported finding significant O2− generation by an angiotensin II (Ang II)–inducible NAD(P)H oxidase in the adventitia,16,17⇓ and Wang et al18 suggested involvement of this source in the development of passive tone. However, the ability of adventitial O2− to reduce endothelium-dependent relaxation (EDR) remains in question. We examined the possibility that the adventitia contributes significantly to impairment of NO·-dependent EDR through the generation of O2−.
Ang II, phenylephrine, acetylcholine, SOD, and H2O2 were purchased from Sigma, and PBS was obtained from Mediatech.
Perfusion of Abdominal Aortas
Abdominal aortas from heparinized 9- to 13-week-old wild-type C57Bl/6Tac mice (Taconic) and gp91phox−/− (Jackson Labs) and eNOS−/− mice on a C57Bl/6J background were placed in bicarbonate-buffered physiological salt solution (PSS) containing (in mmol/L) NaCl 119, KCl 4.7, MgSO4 1.17, CaCl2 1.6, NaH2PO4 1.18, NaHCO3 24, EDTA 0.026, and glucose 5.5 and bubbled with 95% O2 and 5% CO2 (pH 7.4). Adipose tissue was removed in cold PSS, taking care not to damage the adventitia, and samples were placed in a heated (37°C) suffusion/perfusion chamber that allowed compartmental additions of agents (luminal perfusate, adventitial suffusate). The aorta was cannulated with glass micropipettes and secured with 10-0 nylon suture (Ethicon), also used to tie off collaterals19; and the lumen was perfused with PSS to remove blood. The inflow end (perfusate) was attached to a reservoir with an atmosphere of 95% O2 and 5% CO2 that allowed adjustment of intraluminal pressure. Aortas were equilibrated to ≈80% of mean arterial pressure (80 mm Hg), continuously perfused and suffused with warm PSS, and then contracted with 1 μmol/L phenylephrine to produce 30% to 40% of maximal contraction. Maximal contraction was achieved at 10−4 mol/L phenylephrine and represented an ≈20% decrease in aortic diameter.
Acetylcholine was continually luminally infused at increasing concentrations, and a dose-response curve was generated. Diameter was measured at 10-minute intervals by videomicroscopy (Nikon, Optiphot-2) using the 30-minute value for comparison. There were 2 reference points for determination of percent relaxation: 0% relaxation as no change from maximum contraction before dose-response and 100% relaxation as return to the diameter before phenylephrine contraction. In separate experiments, H2O2 was diluted into PSS buffer and suffused after preconstriction with phenylephrine. Ang II, SOD, and oxyhemoglobin were applied as described in the figure legends.
Detection of O2−
To examine whether oxygenation of the buffer enhanced O2− production by isolated mouse aortas, total aortas (from just below the descending arch to just above the iliac arteries) were cleaned of adipose and loose connective tissue, cut into rings, and incubated for 30 minutes in modified Krebs-HEPES buffer (pH 7.4) at 37°C with diethyldithiocarbamate (10 mmol/L) to inhibit endogenous Cu/Zn superoxide dismutase then either gently bubbled with 95% oxygen and 5% CO2 or not bubbled. Rings were transferred to tubes containing lucigenin (5 μmol/L) and incubated for 10 minutes at 37°C in the dark.20 After incubation, tubes were placed in a luminometer and luminescence integrated over 5 minutes (10 cycles). Tiron (10 mmol/L) was added, and 20 more cycles were read, averaging the final 3 values. Differences between the average values of the first 10 readings and the last 3 readings (with Tiron) were expressed as Δ chemiluminescence/min per mg tissue weight.
Biotinylation of Superoxide Dismutase
SOD was biotinylated using an EZ-Link kit (Pierce). Sulfo-NHS-LC-biotin was added to Cu/Zn-SOD in 12-fold molar excess and incubated for 2 hours at 4°C, then separated from free biotin using a dextran desalting column.
Frozen aortas suffused with biotinylated or unmodified SOD were cut into 4- to 5-μm sections and dried for 1 hour at room temperature. Sections were fixed in acetic acid (10 minutes), dried at room temperature, washed with PBS, blocked with goat serum (Sigma) (1:100) for 30 minutes, and incubated with ABC reagent (Vector) for 30 minutes. Sections were incubated with peroxidase substrate (AEC) for 5 minutes to obtain the desired stain intensity. After washing with PBS, sections were photographed at ×10 and ×40 magnification.
Data are expressed as mean±SEM. Comparisons were made using ANOVA. If differences were noted between treatment groups, a t test was performed for point analyses and the results adjusted using Hochberg’s method. P<0.05 was considered significant.
EDR Response to Adventitial Application of Oxyhemoglobin
To examine whether selective addition of an NO· scavenger to the perivascular space attenuates EDR, 10 μmol/L oxyhemoglobin was applied to the adventitial suffusate, with the knowledge that proteins with a lower molecular weight are unable to penetrate beyond the vascular adventitia. Oxyhemoglobin impaired EDR compared with vehicle (Figure 1).
Effect of Adventitial Xanthine/Xanthine Oxidase Suffusion on EDR
To test whether exogenously applied adventitial O2− attenuates EDR, preconstricted aortas were adventitially suffused with vehicle, xanthine (X, 500 μmol/L) plus xanthine oxidase (XO, 0.01 U/mL), or X/XO plus SOD. Acetylcholine-induced relaxation was impaired in the presence of X/XO and was restored by adventitial SOD suffusion (Figure 2).
Effect of Adventitial Application of SOD on Ang II–Induced EDR Impairment
Endogenous adventitial O2− was increased by suffusing the adventitia for 3 hours with 10 pmol/L Ang II, which increases O2− in the mouse aorta20 without increasing diameter (5.4±0.2 versus 5.5±0.2 arbitrary units). Phenylephrine-induced contraction was the same in vehicle and Ang II groups; however, Ang II shifted acetylcholine-induced relaxation to the right. EDR impairment was reversed by adventitial suffusion of SOD (300 U/mL), but SOD suffusion did not improve relaxation in the vehicle group (Figure 3).
To test whether adventitially applied SOD could penetrate other vascular layers, we suffused the adventitia with biotinylated SOD. SOD was detected in the adventitial space surrounding the external elastic lamina (Figure 4); aortas not suffused with biotinylated SOD did not stain.
Comparison of EDR in Ang II–Treated Aortas From gp91phox−/− Versus Wild-Type Mice
Our previous data suggested that adventitial fibroblasts contain a gp91phox-based NAD(P)H oxidase that is primarily responsible for O2− generation in the adventitia.16 The present study revealed that in the absence of SOD, Ang II–treated aortas from gp91phox−/− mice showed greater relaxation than those from Ang II–treated wild-type mice. Moreover, SOD had no effect on EDR of Ang II–treated aortas from gp91phox−/− mice (Figure 5).
Effect of H2O2 on Phenylephrine-Constricted Mouse Aortas
To test whether the relaxant effect of SOD is partly attributable to generation of H2O2, we examined the ability of adventitially applied H2O2 to relax phenylephrine-preconstricted aortas; however, contrary to our hypothesis, we observed vasoconstriction with H2O2 (Figure 6).
Lack of Effect of Oxygenation on O2− Production by Mouse Aortas
To examine whether in vitro buffer oxygenation with 95% O2 in this preparation enhances aortic O2− production, we compared nonoxygenated aortas with those oxygenated with 95% O2/5% CO2 but observed no difference in O2− detection (8.7±4.2 versus 9.5±4.7 mU/min per mg tissue, oxygenated versus nonoxygenated). This is likely related to a low Km for O2 of NAD(P)H oxidase.21
Our findings support the hypothesis that adventitial NAD(P)H oxidase–derived O2− impairs endothelium-derived NO-dependent relaxation. Adventitial delivery of the O2−-generating system X/XO caused significant impairment of EDR that was reversed by coadministration of SOD. Ang II–induced impairment of EDR could be ameliorated by adventitial delivery of SOD. Moreover, Ang II–treated aortas from gp91phox-deficient mice exhibited enhanced EDR (compared with wild-type controls), whereas adventitial SOD delivery had no effect. These data suggest that O2− derived from adventitial gp91phox-NAD(P)H oxidase can impair EDR.
Undoubtedly, endothelial and smooth muscle NAD(P)H oxidase–derived O2− can interfere with NO·. Numerous studies have implicated both oxidases in impairment of endothelium-dependent relaxation, and both are widely accepted as important regulators of vascular tone.2,9,10⇓⇓ Görlach et al9 clearly demonstrated the functional role of endothelial oxidase in O2− production, showing that phorbol ester–induced aortic O2− was blocked by removal of the endothelium or gp91phox deletion. They also showed that gp91phox deletion improves acetylcholine-induced relaxation of mouse aortas under basal conditions. Mollnau et al15 associated Ang II–induced NAD(P)H oxidase expression with uncoupling of NOS and decreased NO·-induced signaling via guanylate cyclase and cGMP-dependent protein kinase, suggesting that endothelial or smooth muscle NAD(P)H oxidase impairs endothelium-dependent relaxation. In the present study, we focused on the role of adventitial NAD(P)H oxidase in endothelium-dependent relaxation impairment, independent of the effect of endothelial and smooth muscle oxidase.
Although NO· scavengers are known to impair vascular relaxation, we questioned whether adding a scavenger to the perivascular space could attenuate EDR. Applying oxyhemoglobin to the adventitia impaired EDR, consistent with the notion that adventitial scavenging of NO· reduces overall vascular NO· derived from the endothelium and thus its relaxant effect. These findings support remote NO· scavenging reducing EDR.
Our results suggest that in addition to sources adjacent to the endothelium, stimulation of adventitia-derived O2− can contribute to impairment of EDR. Previous studies showed that the adventitia is a significant O2− source and suggest that adventitial O2− acts as a barrier to exogenously applied NO· and affects passive tone22; however, its ability to interfere with EDR was not explored. Adventitial application of X/XO impaired EDR, which was reversed by adventitial SOD suffusion, consistent with adventitial O2− impairing endothelium-derived NO· bioactivity. The apparent convergence of dose-response curves at high acetylcholine concentrations may be explained by the ability of endogenous NO· to overcome ambient O2− levels or the possibility that peroxynitrite (formed by the reaction of these two free radicals) is contributing to vasodilatation.23 The fact that in the presence of X/XO, SOD enhanced acetylcholine-induced relaxation more than in vehicle controls may partly reflect SOD conversion of O2− to very high concentrations of H2O2, which cause relaxation.24 Thus, H2O2 accumulation in this scenario could enhance the relaxation observed with X/XO plus SOD. Previous studies showed that endothelial NOS plays a significant role in relaxation of the mouse thoracic aorta, but this was not always true for smaller blood vessels.25 The markedly reduced ability of acetylcholine to relax preconstricted abdominal aortas from eNOS−/− versus wild-type mice in our study (not shown) confirmed that eNOS-derived NO· plays a highly significant role in acetylcholine-induced relaxation of the abdominal aorta.
We previously described an Ang II–inducible NAD(P)H oxidase in the rabbit aortic adventitia,16 which is also present in rats22 and mice.17 In the present study, Ang II–induced EDR impairment was reversed by adventitial SOD suffusion, suggesting that elevated adventitial O2− lowers vascular steady-state NO· levels in the same way as oxyhemoglobin (the sink hypothesis).7 In contrast, adventitial SOD suffusion did not improve relaxation of vehicle-treated aortas, consistent with low basal O2− under normal conditions. Indeed, the highly impenetrable elastic lamina blocks proteins and adenoviral vectors26 and would sequester large molecules such as SOD, XO, and oxyhemoglobin. Adventitial suffusion of biotinylated SOD (the smallest of the 3 proteins) resulted in SOD being detected only in the adventitial space. This confirmed that SOD was not penetrating the media and is consistent with adventitial O2− impeding EDR.
Interestingly, relaxations of aortas (treated adventitially with Ang II) from gp91phox−/− mice were greater than those of wild-type mice, suggesting that adventitial gp91phox-based oxidase could interfere with EDR. Moreover, adventitial SOD suffusion had no effect on EDR of Ang II–treated aortas from gp91phox−/− mice, consistent with the reduced ability of the aortic adventitia in gp91phox−/− mice to produce Ang II–stimulated O2−.27 This suggests that the O2− source impairing EDR is adventitial gp91phox-based NAD(P)H oxidase. The data also corroborate that the effects of SOD we observed were specific to O2− scavenging. Although previous studies addressed the role of ubiquitous gp91phox NAD(P)H oxidase in normal acetylcholine-induced EDR, our present study tested whether adventitial gp91phox-containing NAD(P)H oxidase might contribute to impaired EDR. We are presently developing methods of assessing the relative contribution of the various cellular sources of NAD(P)H oxidases to impaired EDR.
We postulated that adventitial O2− scavenges endothelium-derived NO· and thus impairs EDR; yet O2− could be constricting the outer vascular medial layers or stimulate vasoconstrictor release from the adventitia. For instance, reactive oxygen species (ROS) activate cyclooxygenase and enhance the vasoconstrictor action of prostaglandin H2.28 They also stimulate vascular smooth muscle cells to release heat shock protein-90 and cyclophilin A,29 which can activate ERK1/2 in an autocrine fashion and mediate smooth muscle cell contraction.30 These studies strongly support a paracrine effect of oxidative stress in the vasculature; and although we are not aware of studies examining such a role, it is tempting to speculate that such mechanisms also exist in adventitial fibroblasts.
We initially postulated that the ability of SOD to improve relaxation was partly attributable to generation of H2O2. In fact, under physiological conditions, Ang II can activate endogenous extracellular SOD.31 Thus, when adventitial O2− is elevated and endogenous SOD is enhanced, the resulting product, H2O2, could become a paracrine signaling agent, diffusing from the adventitia into the media, because it is stable under biological conditions and can penetrate cell membranes.32 Attempts to determine the direct contribution of H2O2 to the effect of SOD on EDR were impeded by the ability of the peroxide scavenger catalase to decrease NO· as well.33 For this reason, we examined the ability of exogenous H2O2 to relax mouse aortas. Contrary to our hypothesis, we observed a constrictor response to adventitially applied H2O2 between 10−6 and 10−4 mol/L. A previous report by Fujimoto et al24 demonstrated only a relaxant effect of H2O2 in mouse aortas at concentrations of ≥10−4 mol/L; however, they applied H2O2 to an organ chamber bath with access to the entire vessel ring, whereas we applied it directly to the adventitia. Thus, the difference we observed might be related to a constrictor effect of H2O2 on the outer media or secondary release of a vasoconstrictor from the adventitia, whereas the effect they observed could have been mediated by direct smooth muscle relaxation or release of a vasodilator from the endothelium. Nevertheless, because vascular O2− is reported to be in the nanomolar range,34 application of exogenous SOD would at most be expected to cause acute production of nanomolar concentrations of H2O2. Such concentrations did not relax the aorta and are therefore unlikely to have contributed to the ameliorative relaxant effect of SOD.
Hypertension, atherosclerosis, and vascular injury all activate perivascular cells and increase macrophage levels in the perivascular space,35–37⇓⇓ which, combined with characteristic fibroblast proliferation, will likely result in increased perivascular O2− production and hence EDR impairment. It is conceivable that generation of Ang II or cytokines by perivascular adipose tissue38 and interstitium39 as well as macrophages and mast cells present in the adventitia37 could potentiate endogenous O2− production. Moreover, adventitial NAD(P)H oxidase activity may affect nitrergic neurotransmission; eg, adventitial fibroblasts are juxtaposed to adventitial nerve endings that produce NO·. Therefore, in vivo adventitial O2− may play an even more important role in the control of vascular tone.
In summary, we believe our data support the hypothesis that adventitial O2− plays a significant role in the impairment of EDR and perhaps a modulatory role in endothelial function. Thus, previous preparations of blood vessels that damaged the adventitia may have overlooked an important modulator of tone. These data also highlight the potential role of an activated adventitia in vascular disease and support the concept of cross-talk between vascular segments via ROS. Additional studies targeting adventitial ROS are needed to clarify their role in vascular physiology and pathophysiology.
This work was supported by National Institutes of Health grants HL55425 and HL28982 and by American Heart Association grants 95011900 and 9808086W. We thank Dr William C. Sessa for his assistance and insightful discussion.
- ↵Nakazono K, Watanabe N, Matsuno K, et al. Does superoxide underlie the pathogenesis of hypertension? Proc Natl Acad Sci U S A. 1991; 88: 10045–10048.
- ↵Wu L, de Champlain J. Effects of superoxide on signaling pathways in smooth muscle cells from rats. Hypertension. 1999; 34: 1247–1253.
- ↵Görlach A, Brandes RP, Nguyen K, et al. A gp91phox containing NADPH oxidase selectively expressed in endothelial cells is a major source of oxygen radical generation in the arterial wall. Circ Res. 2000; 87: 26–32.
- ↵Bayraktutan U, Blayney L, Shah AM. Molecular characterization and localization of the NAD(P)H oxidase components gp91-phox and p22-phox in endothelial cells. Arterioscler Thromb Vasc Biol. 2000; 20: 1903–1911.
- ↵Touyz RM, Chen X, Tabet F, et al. Expression of a functionally active gp91phox-containing neutrophil-type NAD(P)H oxidase in smooth muscle cells from human resistance arteries: regulation by angiotensin II. Circ Res. 2002: 90: 1205–1213.
- ↵Griendling KK, Minieri CA, Ollerenshaw JD, et al. Angiotensin II stimulates NADH and NADPH oxidase activity in cultured vascular smooth muscle cells. Circ Res. 1994; 74: 1141–1148.
- ↵Laursen JB, Rajagopalan S, Galis Z, et al. Role of superoxide in angiotensin II-induced but not catecholamine-induced hypertension. Circulation. 1997; 95: 588–593.
- ↵Mollnau H, Wendt M, Szöcs K, et al. Effects of angiotensin II infusion on the expression and function of NAD(P)H oxidase and components of nitric oxide/cGMP signaling. Circ Res. 2002; 90: e58–e65.
- ↵Pagano PJ, Clark JK, Cifuentes-Pagano ME, et al. Localization of a constitutively active, phagocyte-like NADPH oxidase in rabbit aortic adventitia: enhancement by angiotensin II. Proc Natl Acad Sci U S A. 1997; 94: 14483–14488.
- ↵Cifuentes ME, Rey FE, Carretero OA, et al. Upregulation of p67phox and gp91phox in aortas from angiotensin II-infused mice. Am J Physiol Heart Circ Physiol. 2000; 279: H2234–H2240.
- ↵Wang HD, Hope S, Du Y, et al. Paracrine role of adventitial superoxide anion in mediating spontaneous tone of the isolated rat aorta in angiotensin II-induced hypertension. Hypertension. 1999; 33: 1225–1232.
- ↵Rey FE, Cifuentes ME, Kiarash A, et al. Novel competitive inhibitor of NAD(P)H oxidase assembly attenuates vascular O2− and systolic blood pressure in mice. Circ Res. 2001; 89: 408–414.
- ↵Wang HD, Pagano PJ, Du Y, et al. Superoxide anion from the adventitia of the rat thoracic aorta inactivates nitric oxide. Circ Res. 1998; 82: 810–818.
- ↵Ríos CD, Ooboshi H, Piegors D, et al. Adenovirus-mediated gene transfer to normal and atherosclerotic arteries: a novel approach. Arterioscler Thromb Vasc Biol. 1995; 15: 2241–2245.
- ↵Wang HD, Xu S, Johns DG, et al. Role of NADPH oxidase in the vascular hypertrophic and oxidative stress response to angiotensin II in mice. Circ Res. 2001; 88: 947–953.
- ↵Liao D-F, Jin Z-G, Baas AS, et al. Purification and identification of secreted oxidative stress-induced factors from vascular smooth muscle cells. J Biol Chem. 2000; 275: 189–196.
- ↵Touyz RM, El Mabrouk M, He G, et al. Mitogen-activated protein/extracellular signal-regulated kinase inhibition attenuates angiotensin II-mediated signaling and contraction in spontaneously hypertensive rat vascular smooth muscle cells. Circ Res. 1999; 84: 505–515.
- ↵Fukai T, Siegfried MR, Ushio-Fukai M, et al. Modulation of extracellular superoxide dismutase expression by angiotensin II and hypertension. Circ Res. 1999; 85: 23–28.
- ↵Shi Y, Niculescu R, Wang D, et al. Increased NAD(P)H oxidase and reactive oxygen species in coronary arteries after balloon injury. Arterioscler Thromb Vasc Biol. 2001; 21: 739–745.
- ↵Capers Q,IV, Alexander RW, Lou P, et al. Monocyte chemoattractant protein-1 expression in aortic tissues of hypertensive rats. Hypertension. 1997; 30: 1397–1402.
- ↵Cassis LA, Lynch KR, Peach MJ. Localization of angiotensinogen messenger RNA in rat aorta. Circ Res. 1988; 62: 1259–1262.
- ↵Guan S, Fox J, Mitchell KD, et al. Angiotensin and angiotensin converting enzyme tissue levels in two-kidney, one clip hypertensive rats. Hypertension. 1992; 20: 763–767.