INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE PHAGOCYTIC NADPH oxidase or respiratory burst oxidase is present in neutrophils and macrophages and has an important role in immune defense. It is a multicomponent enzyme complex, which includes two membrane-spanning polypeptide subunits p22^phox and gp91^phox, which comprise flavocytochrome b₅₅₈, and at least three cytosolic subunits p47^phox, p67^phox, and Rac2 (3). Exposure of the cell to a variety of agonists induces the association of the cytosolic with the membrane-associated components, causing activation of the normally dormant oxidase, which then catalyzes the one electron reduction of oxygen to superoxide, using NADPH as the electron donor (8).

A vascular isoform of the oxidase has been recently described as the most important source of superoxide in the vascular wall (5, 13). The characterization of this vascular enzyme has been a matter of intensive research. Similarity with the leukocyte NADPH oxidase was suggested on the basis of pharmacological, spectral, or immunolocalization studies. In pulmonary smooth muscle, the gp91^phox subunit of the oxidase was reported to be present, and a functioning cytochrome b₅₅₈ was identified (20). The cytosolic subunits p47^phox and p67^phox have been described in endothelial cells (15) and adventitial fibroblasts (16), and the membrane-bound subunit p22^phox was shown to be expressed in endothelial cells (4) and vascular smooth muscle cells (40). In addition, antisense antagonism of p22^phox inhibited the superoxide production after treatment of vascular smooth muscle cells with angiotensin II.

However, major questions regarding the structural identity of the vascular enzyme still remain. Some past studies reported the inability to detect the gp91^phox subunit in vascular smooth muscle cells from the rat aorta, although a functioning cytochrome b₅₅₈ was identified (4). On the other hand, in human endothelial cells, it was reported that gp91^phox was detected, but the cytochrome was not demonstrated (20).

The functioning of the vascular enzyme is also a matter of controversy. Whereas the phagocytic oxidase generates superoxide early after exposure to an activating stimulus and in large amounts, the vascular oxidase produces superoxide at a lower and constant rate (12). In addition, the enzyme specificity regarding its source of reducing equivalents remains unclear. Whereas the phagocytic oxidase accepts electrons from NADPH with a Michaelis constant (K_m) 10 times lower than the K_m for NADH (3), the vascular enzyme has been reported to accept electrons both from NADH (22) or NADPH (26).

In parallel with increasing evidence that reactive oxygen species can act as ubiquitous second messengers (44), it is becoming apparent that NAD(P)H oxidase-derived oxidants can also exert a similar role in the vascular system (13, 14, 21, 38). However, whereas the phagocytic oxidase has an established function in immune defense, questions remain regarding the physiological role of the vascular oxidase. Superoxide, which is the primary product of the enzyme, has been implicated in a series of pathological conditions, such as hypertension and restenosis after angioplasty. It has been shown that the administration of SOD normalizes the blood pressure in rats made hypertensive by elevated levels of angiotensin II (19) and abolishes the vasospasm observed after coronary angioplasty in dogs (18). It is unclear, however, whether or not the vascular oxidase is able to affect vasomotility.

A mouse model of the X-linked chronic granulomatous disease (X-CGD) has been generated with a nonfunctional allele for the gp91^phox subunit of the phagocytic NADPH oxidase by using targeted homologous recombination in murine cells (29). These X-CGD mice are viable when housed in a protective environment with normal growth and development and have the same number of circulating neutrophils that are able to migrate to the peritoneal cavity after injection of thioglycollate. However, the neutrophils of these mice are unable to release superoxide after stimulation with phorbol esther, as detected by cytochrome c reduction and the nitro blue tetrazolium assay (29). These mice were previously used to elucidate the role of the superoxide-generating NADPH oxidase in focal cerebral ischemia and reperfusion (42) or oxygen tension sensing in the pulmonary vasculature (2); however, vascular superoxide generation in these transgenic mice has not been previously studied.

In this study, we further characterize the vascular oxidase to determine whether this enzyme is identical to the phagocytic oxidase. We specifically determined whether the vascular oxidase contains the major cytochrome b₅₅₈ containing the gp91^phox subunit that is critical for the function of the phagocytic oxidase. Superoxide production of vessels and homogenate fractions from gp91^phox-deficient (/) mice and wild-type control is characterized, and the effect of knockout of gp91^phox on the function of the vascular oxidase in the control of vascular tone is assessed. It was observed that superoxide generation and vasoconstriction from the oxidase are not altered by knockout of gp91^phox.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Chemicals. The cell membrane permeant SOD mimetic (SOD_m) M40403 was obtained from Monsanto/MetaPhore Pharmaceuticals (St. Louis, MO) (33). The spin-trap 5,5'-dimethyl-pyrroline-N-oxide (DMPO) was obtained from Dojindo Laboratories (Kumamoto, Japan). SOD from bovine erythrocytes, diethyldithiocarbamate (DETC), oxypurinol, 17-octadecynoic acid, indomethacin, rotenone, diphenyleneiodonium (DPI), antimycin, thenoyltrifluoroacetone (TTFA), N-nitro-L-arginine methyl ester (L-NAME) hydrochloride, phenylephrine hydrochloride, NADH, and NADPH were purchased from Sigma Chemicals (St. Louis, MO) and then stored and diluted according to the manufacturer's directions.

Animals. gp91^phox knockout mice were obtained from The Jackson Laboratory (Bar Harbor, ME). These mice were generated with a null allele of the gene that encodes the 91-kDa subunit of the oxidase cytochrome b₅₅₈. Affected hemizygous male mice lack phagocyte superoxide production (29). C57/BL6 6- to 8-wk-old mice were used as wild-type controls. This mouse strain has been well characterized previously. Leukocytes obtained from these mice have no detectable gp91^phox on Western blotting (29).

Superoxide detection by EPR spectroscopy in intact vessels. Mice were killed by intraperitoneal injection of pentobarbital sodium (100 mg/kg). Thoracic aortas were removed, freed from periadventital and loose connective tissue, and immediately incubated with the spin-trap DMPO (50 mM) diluted in buffer (PBS + EDTA 0.01 mM, pH = 7.40). NADPH or NADH (0.5 mM each) was added to the tube with a final volume of 0.6 ml. After 10 min of incubation at room temperature, the vessel segment was transferred to the flat cell, with surrounding buffer added in sufficient amounts to fill the top compartment containing the vessel. In some experiments the SOD_m M40403 (20 µM), DETC (0.1 mM), or DPI (20 µM) was added to the solution from the beginning of the incubation period.

Vessel homogenate and particulate fraction preparation. The thoracic aortas from 10 mice were excised, freed from periadventitial and loose connective tissue, and placed in cold Tris · HCl buffer (50 mM), pH = 7.40, containing 0.1% mercaptoethanol, 1.0 mM phenylmethanesulfonyl fluoride (PMSF), and protease inhibitors (Complete Mini, Boehringer-Mannheim). The vessels were minced under liquid nitrogen, collected with 1 ml of the buffer, sonicated 10 times (3 s each) on ice at medium power, and centrifuged at 1,000 g for 10 min at 4°C. The supernatant was collected and transferred to a 4.0-ml ultracentrifuge tube. It was then centrifuged at 20,000 g for 15 min at 4°C. The supernatant was then centrifuged at 100,000 g for 45 min at 4°C. A small pellet was harvested from the bottom of the centrifuge tube and resuspended in buffer. Protein was quantified by the Bradford method. This preparation yields the particulate fraction (PF), which contains mainly membranes and microsomes (31).

Enzymatic source of superoxide in the vessel wall. To determine the enzymatic source of superoxide in the vessel wall, experiments were performed with PF, which was previously incubated with compounds that block known enzymatic pathways of superoxide generation. Therefore, PF (0.1 mg protein/ml) was incubated with blockers of cyclooxygenase (indomethacin, 0.01 mM), xanthine oxidase (oxypurinol, 1.0 mM), nitric oxide synthase (L-NAME, 0.1 mM), cytochrome P-450 (17-octadecynoic acid, 5.0 µmol/l), flavoenzymes (DPI, 20 µmol/l), and the NADPH oxidase-blocking peptide PR-39 (0.1 mM). To evaluate the contribution of the mitochondrial electron transport chain, blockers of complex I (rotenone, 0.1 mM), complex II (TTFA, 10 µM), and complex III (antimycin A, 5 µM) were used. After 30 min of incubation with each of these compounds, at room temperature, EPR spectra were recorded as described above.

Spin-adduct quantitation. Quantitation of the observed DMPO-OH spin adduct was performed by simulation fitting, as previously described (17). This procedure provides accurate intensity quantitation and minimizes errors due to noise or baseline drifts. First, any baseline drift was removed by linear baseline correction. The DMPO-OH adduct was then simulated using the observed hyperfine splitting values a_N = a_H = 14.9 G, to match the corresponding measured spectrum. The simulated spectrum was calibrated with respect to an appropriate aqueous solution of nitroxide 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO, 1 µM) measured under identical conditions. A least-squares fit of the simulated spectrum to the measured spectra was performed to obtain the component weight. Finally, the component weights, calibration, and instrumental parameters were used to quantitate the measured spectrum (17). For the PF, the results are expressed as the concentration of DMPO-OH per milligram of protein.

Vascular reactivity assays. Six- to eight-week-old mice were killed by intraperitoneal injecton of pentobarbital sodium (100 mg/kg). The thoracic aortas were removed, freed from periadventitia, and cut in three fragments, each 3 mm in length. The aortic rings were mounted between horizontal stainless steel wires in a 15-ml organ bath containing Krebs-Henseleit buffer, pH = 7.40 (composition, in mM: 119.0 NaCl, 25.0 NaHCO₃, 10.0 glucose, 5.0 KCl, 2.0 CaCl₂, 1.2 MgSO₄, and 1.2 KH₂PO₄), which was continuously gassed with 95% O₂-5% CO₂. The lower wire was stationary, and the upper wire was connected to a force-displacement transducer (Kent Scientific) for measurement of isometric tension. The output from the force transducer was recorded on a MacLab 8 Bridge Amplifier (AD Instruments). The aortic rings were stretched progressively to achieve a resting tension of 0.60 g, which was determined by prior experiments to be the minimun tension facilitating the development and maintainance of maximal contractions to 30 mM KCl under these experimental conditions. Vessels were maintained under this resting tension for ~90 min, with the bath buffer changed every 20 min until each experiment was started.

Statistical analysis. All data are expressed as means ± SE. Comparisons among groups were performed by Student's t-test or one-way ANOVA, with Student-Newman-Keuls or Dunnett's multiple range tests. The significance level was 0.05. The Primer of Biostatistics computer program was used (version 3.01, 1992, McGraw-Hill).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Aortic rings. Aortic rings from wild-type mice did not give rise to any detectable EPR signal under baseline conditions (Fig.1A). With prior incubation of the vessels with DETC (0.1 mM) to inhibit Cu,Zn SOD, a small DMPO-OH adduct was observed (Fig. 1B). When vessels were exposed to exogenously added NADPH or NADH (0.5 mM each), prominent spectra exhibiting a 1:2:2:1 quartet signal were seen, characteristic of the DMPO-OH adduct (Fig. 1, C and E). This adduct was primarily derived from superoxide, because it was more than 80% quenched by prior incubation with the SOD_m M40403 (20 µM; Fig. 1, D and F). gp91^phox / mouse aortas present a very similar pattern of basal and NAD(P)H-stimulated radical adduct generation (Fig. 1, right).

View larger version (38K): [in this window] [in a new window]   Fig. 1.   Superoxide generation measured by electron paramagnetic resonance (EPR) spectroscopy using the spin-trap 5,5'-dimethyl-pyrroline-N-oxide (DMPO) in wild-type (left) and gp91phox / (right) mouse aortas. Spectra obtained under basal conditions show no adduct formation (A), and previous incubation with diethyldithiocarbonate (DETC, 0.1 mM) results in the appearance of a small DMPO-OH signal in both groups (B). After NADPH (0.5 mM) addition a DMPO-OH adduct is formed (C), which is quenched by the SOD mimetic (D). Addition of NADH (0.5 mM) results in formation of a similar spectrum (E), which is also abolished by the SOD mimetic (F). The amplitude and intensity of spectra obtained from both wild-type (WT) and transgenic animals were very similar. The figure shows typical spectra obtained in each group, recorded at 60 min of measurement. Experiments were repeated at least 3 times for each group.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Time course of superoxide-derived radical formation by intact vessels after addition of NADPH (A) or NADH (B). The time course and magnitude of DMPO-OH adduct formation after addition of NADPH or NADH, 0.5 mM each, to WT and gp91phox / mouse aortas is identical. Each point represents DMPO-OH concentration obtained in at least 3 different experiments (means ± SE).

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Quantitation of NADPH- and NADH-stimulated radical generation. DMPO-OH adduct spectra quantitation showed that the amount of radicals generated by aortas from WT and gp91phox / mice were identical under basal conditions or after incubation with DETC. There is no difference in radical generation from the two groups. Prior incubation with SOD mimetics decreases the EPR signal, leaving only trace signals. The data shown are the measured values seen after 60 min. Each bar represents mean ± SE of at least 3 experiments.

Vessel homogenates. To determine the chemical nature and cellular structures responsible for the spin-adduct production, PFs of vessel homogenates were prepared from a pool of aortas obtained from wild-type or gp91^phox / mice. These PFs contain mainly membranes and microsomes (31). Under basal conditions, PFs from gp91^phox / or wild-type mice aortas generated no detectable EPR signal. After addition of NADH or NADPH (0.5 mM each), formation of the characteristic DMPO-OH adduct spectrum was observed. This adduct was more than 90% quenched by SOD (250 U/ml) or the SOD mimetic M40403 (20 µM), suggesting that it was derived from superoxide. Catalase (400 U/ml) had no effect on the observed spectra (data not shown).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Time course of DMPO-OH generation by vascular particulate fractions. Particulate fractions from WT and gp91phox / aortas elicited similar amounts of DMPO-OH when exposed to NADPH (A) or NADH (B), 0.5 mM each. DMPO-OH generation reaches a plateau only after 60 min. Each point represents DMPO-OH concentration obtained in at least 3 different experiments (means ± SE).

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Quantitation of radical generation by particulate fractions of WT and gp91phox / mouse aortas. There is no difference in the magnitude of DMPO-OH generated by the two groups. Prior incubation with SOD (250 U/ml) quenched the signals approximately 90%, suggesting that the DMPO-OH adducts observed were derived from superoxide. Each point represents DMPO-OH concentration from spectra recorded after 60 min, obtained in at least 3 different experiments (means ± SE).

Enzymatic source of superoxide. To further characterize the enzymatic source of superoxide in the vessel wall stimulated by NADPH or NADH, experiments were performed using known inhibitors of superoxide-generating pathways (Fig. 6). The experiments reported here were obtained from vessel homogenates of wild-type mouse aortas only; however, similar results were obtained with homogenates from gp91^phox / mouse aortas (data not shown).

View larger version (33K): [in this window] [in a new window]   Fig. 6.   Effect of inhibitors of oxygen radical-generating pathways on radical generation from the particulate fraction WT mouse aortas. Inhibitors of cyclooxygenase (indomethacin, Indo, 10 µM), xanthine oxidase (oxypurinol, Oxy, 1.0 mM), cytochrome P-450 inhibitor (17-octadecynoic acid, 17-ODYA, 5 µM), the neutrophil NADPH oxidase peptide blocker PR-39 (0.1 mM), and nitric oxide synthase (N-nitro-L-arginine methyl ester, L-NAME, 0.1 mM) had no effect. Mitochondrial respiratory chain complex I, II, and III blockers (rotenone, ROT, 0.1 mM; thenoyltrifluoroacetone, TTFA, 10 µM; and antimycin A, ANTI, 5 µM, respectively) did not show any change in the signals either. Only the flavoenzyme blocker diphenyleneiodonium (DPI, 20 µM) inhibited the observed radical generation. Each bar represents mean ± SE of at least 3 experiments in each group. *P < 0.05 vs. no inhibitor added (control). The other groups do not have statistically significant difference vs. control.

Vascular reactivity. Experiments were conducted to assess the effect of the vascular oxidase on vascular contractile tone. Mouse aortas were contracted to EC₅₀ with phenylephrine and were maintained under this tension for at least 20 min. After this period each vascular ring was challenged with NADPH or NADH (1.0 mM each).

View larger version (25K): [in this window] [in a new window]   Fig. 7.   Vascular reactivity assay in response to NADPH. Aortic rings from WT mice contracted 20% above the half-maximum contraction obtained with phenylephrine when exposed to NADPH (1.0 mM). This NADPH-driven contraction starts immediately after the exposure and lasts at least 20 min. Incubation of the vessels with SOD mimetics (SODm) inhibits the contraction in a dose-dependent manner, suggesting that superoxide anion is involved in the phenomenon. Each point represents mean ± SE of at least 5 experiments each group. *P < 0.05 vs. NADPH alone, P < 0.05 vs. SOD 20 µM-treated rings, only after 10 min of measurement.

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Vascular reactivity of gp91phox / aortic rings to NADPH. Aortic rings from gp91phox / mice contract in response to NADPH (1.0 mM) in the same way of the WT mice aortic rings, suggesting that they possess a functioning vascular NADPH oxidase. Each point represents mean ± SE of at least 5 experiments.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Based on a broad range of experimental data (5, 22-26) a vascular NAD(P)H oxidase, resembling the enzyme present in phagocytic cells, has been reported to be the main source of superoxide in the vessel wall. This enzyme, or enzymes, has been suggested to catalyze the one-electron reduction of molecular oxygen, using NADPH or NADH as substrates and generating superoxide. Some of these data (22, 24, 25), however, were obtained with methods whose specificity or reliability to detect free radicals have recently been questioned (41). Therefore, we applied EPR spin-trapping measurement of superoxide and superoxide-derived free radicals to characterize the vascular NAD(P)H oxidase activity, because this technique can provide sensitive and specific detection of these radicals in biological systems (32, 34, 48, 49).

Initially, it was observed that simple exposure of vessels or vessel PF to either NADPH or NADH stimulates the generation of DMPO-OH adducts. DMPO-OH adducts can be generated by direct trapping of hydroxyl radicals by DMPO or they can originate from the trapping of superoxide with subsequent decay of the DMPO-superoxide adducts (DMPO-OOH) to form DMPO-OH (48). It is also known that in the presence of transition metal ions, DMPO can generate pyrrolidinoxyl radicals, including DMPO-OH (32). The adducts observed in our experiments were not quenched by catalase, indicating that they were not produced from direct trapping of H₂O₂-derived ·OH. Most importantly, they were quenched by Cu,Zn SOD (in particulate fractions) or a membrane-permeable SOD_m (in intact vessels), confirming that these signals were derived from superoxide. These observed DMPO-OH adducts were abolished by DPI but not by other pharmacological blockers. Therefore, these data confirm that a flavin-containing NAD(P)H oxidase present in the membrane fraction is the most important source of superoxide in the vascular wall.

These data are similar, in general, to those obtained by other authors (23, 26); however, we did not observe a preferential substrate for the enzyme(s). In previous studies, NADH appeared to be the preferential substrate for the vascular oxidase present in PF from calf pulmonary smooth muscle cells (23) or rat vascular homogenates (31), whereas NADPH-driven superoxide generation was reported to be predominant in the adventitia (26). Radical detection in these reports, however, were made using lucigenin-amplified chemiluminescence, and it has been reported that lucigenin itself, in the conditions used, can stimulate superoxide generation (41).

The EPR technique used here is sensitive and much more specific for detecting superoxide and superoxide-derived free radicals than other commonly used methods. However, even when EPR spin-trapping was used, different characteristics of the oxidase, with regard to its substrates NADPH or NADH, have been reported in different preparations. NADPH-stimulated radical generation was observed to be larger than that from NADH in membrane fraction from human vascular smooth muscle cells (37), whereas the opposite was seen in membrane preparations obtained from human endothelial cells (36). Our experiments were made using membrane fraction from the whole vessel, including the endothelium, media, and adventitial layer. Therefore, our data represent an overall measurement of radical generation by the whole vessel, including both endothelial and smooth muscle cell components, and this may account for differences observed from these prior studies. Nevertheless, it is interesting to speculate that distinct homologues of an NAD(P)H oxidase, with different structures and functional parameters, can coexist in the endothelial and smooth muscle cells of the vessel wall.

The main finding of this study is that the vascular oxidase does not possess a gp91^phox component identical to the phagocytic NADPH oxidase subunit. Previous studies about the structure and identity of the vascular oxidase and its similarility to the phagocytic oxidase have resulted in considerable controversy. mRNA for phagocytic NADPH oxidase components have been described in human endothelial cells (15) and fibroblasts (16); a cytochrome b₅₅₈ was cloned (10), and its activity was detected in rat vascular smooth muscle cells (40), rat microvascular endothelial cells (4), calf pulmonary artery smooth muscle cells (20), and human glomerular mesangial cells (30). Nevertheless, other authors (15) were unable to demonstrate a functioning cytochrome b₅₅₈ in human endothelial cells, although mRNA for its components (p22^phox and gp91^phox) were detected. Interestingly, in rat vascular smooth muscle cells, a functioning cytochrome b₅₅₈ was characterized without evidence of a gp91^phox component (40).

In our experiments, some important differences between the phagocytic and the vascular oxidase are apparent. First, the vascular oxidase works with NADPH and NADH as electron donors, whereas the phagocytic enzyme accepts electrons only from NADPH (3). Also, the simple exogenous addition of the substrates (NADH or NADPH) stimulated superoxide generation from vessels, in contrast to what happens to neutrophils, which need to be stimulated through complex pathways.

A second distinction comes from the fact that the PF of vascular cells contains the complete enzyme, because the CF generated no superoxide under basal conditions or after stimulation with NADH or NADPH, and the addition of the cytosol to the membrane fraction did not increase the signals. This is in contrast to the phagocytic oxidase, which needs cytosolic factors to be activated (27). Also, the fact that the peptide PR-39 does not affect the vascular oxidase suggests that either the vascular subunit has a different structure or the vascular enzyme is constitutively assembled, because the peptide, to exert its inhibitory action, needs to bind to the cytosolic fraction p47^phox, preventing its assembly with the membrane subunits.

To more specifically determine whether the vascular NADPH oxidase is distinct from the phagocytic enzyme, superoxide generation was characterized in aortas from gp91^phox / mice. These mice are well-characterized transgenic animals (29) that lack a functional gp91^phox subunit of the respiratory burst oxidase and therefore have a nonfunctioning phagocytic enzyme. Neutrophils obtained from these mice, in contrast to neutrophils obtained from wild-type mice, did not produce superoxide when stimulated by PMA. However, intact vessels and PFs from gp91^phox / mouse aortas generate superoxide in the same amount, with the same time course as wild-type mice aortas following exposure to NADPH or NADH. These data confirm that the vascular oxidase does not possess a gp91^phox subunit in its structure, and it is, therefore, structurally distinct and genetically different from the phagocytic enzyme.

Our results contrast from the previously reported results in isolated lungs, which showed the presence of gp91^phox in the pulmonary artery smooth muscle cells and lack of superoxide generation from lungs of gp91^phox / mice (2). In that study, radical production from the whole lung was measured, whereas in this study only aortic rings were studied. In addition, in the lung, an increase in radical generation was detected after exposure to PMA, which was not mirrored in our vascular preparations. Aortic rings from wild-type or gp91^phox/ mice stimulated with PMA did not show any increase in radical production. The possibility exists that the pulmonary and systemic circulations have distinct homologues of the oxidase. The recent cloning of homologues of gp91^phox in tumoral and smooth muscle cells (39), thyroid gland (6), and renal cortex (11) raises the interesting possibility that actually there is a family of enzymes, sharing functional and structural features, but unique properties that vary between tissues and species. This hypothesis could explain much of the conflicting data regarding the presence of neutrophil oxidase subunits in the vessel wall.

It is necessary to note that the oxidase activity observed in our experiments is due to a constitutive enzyme, because no stimulus was used to induce it. There have been reports that expression of the vascular oxidase (or at least, one of its isoforms) can be induced (13). Exposure of smooth muscle cells to angiotensin II increases p22^phox expression and superoxide generation in response to NADH addition (40). Tumor necrosis factor- also generated a similar response from these cells (7). Thrombin acts as a potent growth stimulator in vascular smooth cells in vitro and in vivo, and its mitogenic effects have been reported to be related to an increase in superoxide generation by an oxidase that has p47^phox as one of its constituents (28). Hypercholesterolemic rabbits develop an increase in NADH oxidase-derived superoxide production that interferes with vascular reactivity (43). More recently a genetically distinct oxidase (Mox1) that has homology with human and rat gp91^phox has been identified (39). This enzyme was induced in rat aortic smooth muscle cells by platelet-derived growth factor and, when stably transfected into NIH3T3 cells, increased their superoxide production. The preferential substrate for Mox1 (NADPH or NADH), however, has not yet been reported.

The range of physiological roles played by the vascular oxidase are still not completely understood. The enzyme(s) works as a PO₂ sensor in carotid body (1) and pulmonary vasculature (20), and it was also hypothesized that it can transduce signals related to cell growth (14), as reported for the vascular smooth muscle cell hypertrophy in response to angiotensin II, which is dependent on NAD(P)H oxidase-derived hydrogen peroxide generation (46). The recently described oxidase Mox1, when overexpressed, also induces cells to show a transformed appearance, with anchorage-independent growth that can lead to tumor formation in athymic mice (39).

In this work we showed that NAD(P)H oxidase can also play a role in the control of vascular tone, as reported previously (47). Aortic rings contracted 20% above the EC₅₀ to phenylephrine when exposed to high doses of NADPH (1.0 mM). Although these concentrations are far from the physiological range, they were used to disclose whether the activation of the vascular oxidase could affect the vascular tone. Arteries from both wild-type and gp91^phox / mice exhibit this NADPH-stimulated constriction, further demonstrating that gp91^phox is not required for the function of the enzyme(s). SOD_m decreased this contraction, suggesting that superoxide anion is the critical mediator triggering this NADPH-stimulated vasoconstriction.

It is interesting to note that, although NADH induces a similar amount of superoxide as that produced by NADPH in either PF or intact vessels, as previously reported, NADH exposure does not lead to vascular contraction (47). Compartmentalization of the radical generation can be the key feature responsible for this finding. Superoxide generated at distinct locations in the cell could have different accessibility to the radical scavenging systems or the machinery responsible for the vasoconstriction.

In summary, we confirm that vessels generate superoxide through a flavoenzyme present in the PF of the cells. Superoxide produced by activation of this enzyme has an important functional effect in that it induces vasoconstriction. The superoxide generation and function of the vascular oxidase is independent of the presence of functional gp91^phox; thus the vascular oxidase is structurally distinct from the phagocytic oxidase.