Oxygen Metabolism by Endothelial Nitric-oxide Synthase*

Nitric-oxide synthase (NOS) catalyzes both coupled and uncoupled reactions that generate nitric oxide and reactive oxygen species. Oxygen is often the overlooked substrate, and the oxygen metabolism catalyzed by NOS has been poorly defined. In this paper we focus on the oxygen stoichiometry and effects of substrate/cofactor binding on the endothelial NOS isoform (eNOS). In the presence of both l-arginine and tetrahydrobiopterin, eNOS is highly coupled (>90%), and the measured stoichiometry of O2/NADPH is very close to the theoretical value. We report for the first time that the presence of l-arginine stimulates oxygen uptake by eNOS. The fact that nonhydrolyzable l-arginine analogs are not stimulatory indicates that the occurrence of the coupled reaction, rather than the accelerated uncoupled reaction, is responsible for the l-arginine-dependent stimulation. The presence of 5,6,7,8-tetrahydrobiopterin quenched the uncoupled reactions and resulted in much less reactive oxygen species formation, whereas the presence of redox-incompetent 7,8-dihydrobiopterin demonstrates little quenching effect. These results reveal different mechanisms for oxygen metabolism for eNOS as opposed to nNOS and, perhaps, partially explain their functional differences.

Arginine ϩ 2O 2 ϩ 3/ 2͑NADPH ϩ H ϩ ͒ 3 NO ϩ citrulline ϩ 3/ 2NADP ϩ ϩ 2H 2 O REACTION 3 In addition to the reactions listed above, NOS also catalyzes the formation of reactive oxygen species (ROS) such as superoxide anion (O 2 . ) and H 2 O 2 (4 -6). Reaction 1 is the first step, Reaction 2 is the second step, and Reaction 3 shows the overall reaction. and/or Unlike Reaction 3, where activated molecular oxygen is split to oxidize L-arginine to citrulline, NO, and H 2 O, Reactions 4 and 5 do not require the cleavage of the O-O bond. NOS is said to be "coupled" when all the electrons from NADPH are utilized in the formation of monooxygenated products and H 2 O (Reactions 1-3) and "uncoupled" when electrons leak to molecular oxygen to form ROS (Reactions 4 and 5).
Spontaneous dismutation (Reaction 6) of O 2 . to H 2 O 2 (k ϭ ϳ2 ϫ 10 5 M Ϫ1 s Ϫ1 at pH 7.4) produces H 2 O 2 at a detectable level whether or not the NOS-catalyzed two-electron reduction of oxygen to form H 2 O 2 as a direct product occurs.
Diacetyldeuteroheme-substituted horseradish peroxidase (dHRP) is an artificial trapping agent first used by Ishimura and co-workers (7,8) to distinguish superoxide and hydrogen peroxide formation. dHRP rapidly (k Ͼ 10 6  trapped, and the initial product(s) of the uncoupled reaction(s) can be determined. We have modified the original dHRP trapping protocol for NADPH-consuming enzymes such as NOS and reported the successful combination of the dHRP trapping system with an NADPH regeneration system (9). Maintaining a constant low concentration of NADPH not only keeps NADPH from being oxidized by dHRP compound I but also keeps NADP ϩ "feedback inhibition" negligible. In this paper, we applied this modified dHRP trapping protocol to study the products of the eNOS reactions, both qualitatively and quantitatively. Even though Reactions 4 and 5 appear to be simple chemical reactions, quantitative analysis has always been a challenge given the spontaneous/catalyzed dismutation of O 2 . . EPR is the logical analytical tool for studying the formation of small radicals such as O 2 . ; however, quantification of O 2 . to determine stoichiometry and determination of rates is difficult or impossible using EPR (5,10). Our modified dHRP trapping protocol has been shown to be compatible with simultaneous measurement of oxygen concentration (9); in this paper we report the simultaneous measurement of oxygen consumption and dHRP compound III formation, revealing the uncoupled stoichiometry of eNOS. Both NO and ROS are highly reactive chemical entities in aqueous solution; the complexity of their coexistence in a reaction mixture makes it difficult to determine stoichiometry (11)(12)(13). To simplify the system, superoxide dismutase (SOD) and catalase can be added to suppress accumulation of ROS and recycle partially reduced oxygen to either H 2 O or O 2 . The presence of SOD accelerates Reaction 6 to a rate that is diffusionlimited (14), and catalase catalyzes the breakdown of H 2 O 2 to H 2 O and O 2 (15).
In a system that contains sufficient amounts of both SOD and catalase, oxygen in the uncoupled reaction is fully reduced to water. Regardless of whether O 2 . and/or H 2 O 2 is the initial product, the overall uncoupled reaction will be as follows.

REACTION 10
Considering all of the reactions listed above, it is apparent that molecular oxygen is the ultimate oxidizing reagent in both the coupled and uncoupled reactions by NOS; NADPH, on the other hand, is the ultimate electron donor that initiates all subsequent redox processes. . ] to be 1/1, when O 2 . is the initial product of the uncoupled reaction (Reaction 4). Tetrahydrobiopterin (BH 4 ) is an essential cofactor in the oxygenase domain and is proposed to have multiple roles in all three mammalian NOS isoforms (16 -18). One of the presently accepted functions of BH 4 is to act as a one-electron donor during reductive activation of the oxyferrous complex of the heme (18 -21). Although there are many publications about the roles of BH 4 in eNOS and how the absence of BH 4 promotes eNOS uncoupling (22)(23)(24), there is little quantitative evidence regarding its effect on oxygen metabolism. In this study, we report a decreased oxygen consumption rate, as well as a lower [O 2 ]/[NADPH] ratio when BH 4 is absent from the fully coupled eNOS reaction.
Enzymes-eNOS was expressed in Escherichia coli and purified as described by Martásek et al. (25). The enzyme preparation was concentrated to 50 M after being purified by a 2Ј5Ј-ADP-Sepharose 4B column and then dialyzed against 50 mM Tris-HCl (pH 7.4, with 100 mM NaCl and 5% glycerol) to remove 2Ј-AMP, arginine and BH 4 . NO Formation rate (hemoglobin capture assay) of the purified eNOS was determined to be 120 Ϯ 10 nmol/min/mg, using the extinction coefficient of 60 mM Ϫ1 cm Ϫ1 at 401 nm.
Cytosolic NADP ϩ -dependent isocitrate dehydrogenase (IDP-2) was prepared as described by Cunningham and Bash (26), and Ding and co-workers (27) with help from Dr. Lee McAlister-Henn's lab (The University of Texas Health Science Center at San Antonio). The His-tagged protein was purified from a nickel column and concentrated to 8 mg/ml. Activity was tested by monitoring NADPH formation at 340 nm. Purified IDP-2 has stoichiometrically bound NADP ϩ . The activity of the IDP-2, measured by monitoring the formation of NADPH at 340 nm with an extinction coefficient of 6220 M Ϫ1 cm Ϫ1 , was calculated to be 40 mol/min/mg. dHRP was prepared using the method described by Makino et al. (7,28). Native horseradish peroxidase was bought from Toyobo Inc. (Japan). Calmodulin was prepared by the method of Zhang and Vogel (29); SOD and catalase were purchased from Sigma-Aldrich.
dHRP Experiments-dHRP trapping of superoxide anion and hydrogen peroxide was monitored by an Agilent 8453 diode array UV-visible spectrophotometer. The standard reaction buffer consists of 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 200 M CaCl 2 , and 50 M diethylene triamine pentaacetic acid. The reaction was carried out in a glass cuvette, and measurements were taken every 10 s. A typical reaction mixture contained the standard reaction buffer and 100 M dHRP, 200 M MgCl 2 , 0.65 M eNOS (BH 4 -and arginine-free), 2 M IDP-2, and 10 mM DL-isocitric acid (trisodium salt) was added to start the reaction immediately prior to the first optical measurement. Where needed, calmodulin (7 M) and SOD (23 g/ml) were added as described in the figure legends. Control experiments were performed in the absence of eNOS.
Quantitative analysis of the uncoupled ] ratio was performed using simultaneous measurements of oxygen consumption and the abovementioned dHRP trapping of superoxide anion. The reaction was carried out in an air-tight cuvette (2.22 ml), and the oxygenated standard reaction buffer was premixed with 160 or 100 M dHRP, 200 M MgCl 2 , 2.3 M eNOS (BH 4 -and arginine-free), and 2 M IDP-2; 10 mM DL-isocitric acid was added to start the reaction. The rate of compound III accumulation was calculated using the 583-nm delta extinction coefficient of 3.8 mM Ϫ1 reported by Makino and Yamazaki (28), and the corresponding oxygen consumption was recorded by the oxygen sensing system described below.
Oxygen Consumption-The uptake of molecular oxygen was measured at 24°C using a fluorescence-based oxygen sensing system consisting of an SD2000 dual channel spectrophotometer combined with an LS-450 Blue LED light source and a Foxy-18G probe, all from Ocean Optics (Dunedin, FL). The reaction was carried out in an air-tight cuvette (1. It is noteworthy that excessive amounts of SOD and catalase were present in the oxygen consumption experiments to avoid accumulation of ROS. Each experiment contains multiple injections of NADPH, and the results were averaged. The resulting O 2 uptake at each addition was within Ϯ3%. At the end of each experiment, an excessive amount of sodium dithionite was added to ensure depletion of oxygen. At least two different preparations of eNOS were used to confirm stoichiometry in each instance. In the absence of L-arginine or NOHA, the reaction was totally uncoupled. When either L-arginine or NOHA was added, eNOS was partially uncoupled (more in the discussion section), and the overall oxygen consumption equals the weighted sum of both the coupled and uncoupled reactions.

RESULTS
The Uncoupled Reaction and the Effect of CaM-The basic uncoupled reaction, performed in the presence of dHRP, CaM, and eNOS, free of BH 4 and L-arginine, showed the formation of compound III of dHRP (Fig. 1A)  (ϳ4 min Ϫ1 ) under such conditions. The control experiment, performed in the absence of eNOS, showed no spectral changes over 10 min. The abovementioned reaction was then repeated in the absence of CaM. Exactly the same optical changes were observed (spectra not shown) but at much slower rate (ϳ1 min Ϫ1 ). A plot of spectral changes at 583 nm is shown in Fig. 1B; in the first 2 min, the reaction rates in the presence and absence of CaM were both linear, but the (ϩ)CaM rate is about three to four times faster than the (Ϫ)CaM rate. Our results are consistent with the CaM effect reported by Vásquez-Vivár et al. (30) using EPR spectroscopy.
Quantitative analysis of the uncoupled reaction was performed by simultaneous measurement of oxygen consumption and O 2 . formation (9) . formation. Although the oxygen concentration was continuously monitored by the oxygen sensing system described under "Experimental Procedures," a diode array UV-visible spectrophotometer recorded a series of spectra of the reaction mixture every 10 s. When a spectrum was recorded, the burst of light from the diode array spectrophotometer interfered with the oxygen sensor, leaving ticks visible on the oxygen concentration trace at 10-s intervals ( Fig. 2A); time alignment was achieved using these small ticks as an internal reference. Difference spectra of dHRP compound III were produced by subtracting the spectrum of ferric dHRP from those recorded every 10 s after the reaction started (Fig. 2B). The sharp isosbestic points in Fig. 2B confirm that compound III formation (Reaction 7) is the only reaction occurring during the initial 2-min time course. The oxygen consumption trace, recorded simultaneously on the same sample, is shown in Fig. 2A The Arginine Effect-L-Arginine was found to stimulate the oxygen consumption rate by eNOS ϳ10-fold, depending on the enzyme preparation. This is a surprising finding given our previous publication (9) indicating that the oxygen consumption rate by nNOS is strongly inhibited by L-arginine. A comparison is shown in Fig. 3A, in which the oxygen consumption rate by eNOS was stimulated from ϳ4 to ϳ40 min Ϫ1 , whereas the oxygen consumption rate by nNOS was inhibited from ϳ500 to ϳ140 min Ϫ1 .
In contrast, nonhydrolyzable L-arginine analogs, such as N -nitro-L-arginine methylester, N G -monomethyl L-arginine citrate, and L-citrulline (the product of Reaction 3) did not stimulate oxygen consumption by eNOS (Fig. 3, B and C). The significance of these experiments and the comparison of the L-arginine effect on eNOS and nNOS will be addressed in detail under "Discussion." The Tetrahydrobiopterin Effect-BH 4 stimulates the oxygen consumption rate of eNOS by ϳ3-fold in the presence of L-arginine ( Fig. 4A) but has no effect on oxygen consumption in the absence of L-arginine (data not shown). To evaluate the amount of ROS generated in a reaction mixture, SOD and catalase were introduced sequentially to the reaction mixture, whereas oxy-gen concentration was monitored. For example, the reaction could start in the presence of SOD, which keeps ROS in the form of H 2 O 2 (Reaction 6); after ϳ100 M of oxygen was consumed, catalase could be introduced to the reaction mixture, and H 2 O 2 would be rapidly converted to molecular oxygen (Reaction 9), which can be monitored/recorded by the oxygen sensing system. BH 4 almost totally inhibited ROS formation by eNOS, but ROS formation occurs in the absence of BH 4 or the presence of BH 2 , a redox-incompetent BH 4 analog (Fig. 4, B and C). Fig. 4B indicates that a significant amount of ROS was formed by BH 4 -free eNOS even in the presence of L-arginine (detailed conditions are indicated in the figure legend). Upon the addition of catalase, the evolution of oxygen was observed to be 20 -30 M. Repeating the experiment with BH 2 -bound eNOS produced a similar result (Fig. 4C), but repeating the experiment with BH 4 -replete eNOS yielded a negligible amount of oxygen evolution (Fig. 4D). Although these experiments are not strictly quantitative, they present a qualitative measure of ROS formation under different conditions. Because BH 4 itself is a strong reducing reagent and may react with ROS directly, the experiment presented in Fig. 4D was repeated with a range of BH 4 concentrations (500 M to 10 mM). The same result was obtained (data not shown), indicating that the diminished formation of ROS by BH 4 -replete eNOS, rather than the ROS scavenging activity by BH 4 itself, is responsible for the striking differences in the comparison of Fig. 4  (compare B and C with D). The literature indicates that free BH 4 and eNOS-bound BH 4 degrade at neutral pH at 0.017 and 0.006 min Ϫ1 , respectively (31), rates that are insufficient to compete with eNOS-catalyzed oxygen consumption. Fig. 5. In the absence of L-arginine or NOHA, the reaction was fully uncoupled, and the measured [O 2 ]/[NADPH] ratio was 0.50, the same as the theoretical value ( Table 1). The same ratio measured with L-arginine or NOHA was slightly lower than the theoretical value ( Table  1), indicating that eNOS was highly coupled under our experimental conditions. Repeating these experiments proved that our enzyme preparations consistently exhibited 92-96% coupling in the presence of L-arginine or 88 -91% in the presence of NOHA. Oxygen consumption rates are calculated from the initial slope of oxygen traces in Fig. 5 and are listed in Table 2. Unlike the [O 2 ]/[NADPH] ratio, enzyme preparations vary in terms of turnover numbers in a Ϯ10% range.

DISCUSSION
The Uncoupled Reaction of eNOS-eNOS uncoupling is known to be controlled by substrate/cofactor availability (30,32,33), and the uncoupled reactions play important roles under various physiological/pathological conditions, such as atherosclerosis and septic shock (34 -38). We applied the dHRP trapping technique with an NADPH regeneration system (9) to identify the initial product of the eNOS uncoupled reaction. O 2 . is the direct product detected in the absence of substrate (L-arginine and/or NOHA). CaM accelerates eNOS-mediated O 2 . generation by a factor of 3, which is consistent with the results reported by Masters and co-workers (30). In the absence of a trapping agent, H 2 O 2 formed by spontaneous dismutation of O 2 . will also be a detectable product.
In the absence of L-arginine, oxygen uptake by eNOS is very slow, ϳ4 min Ϫ1 in the presence of CaM, and even slower (ϳ1 min Ϫ1 ) in its absence. NADPH-cytochrome P-450 reductase, a flavin enzyme that shares ϳ60% sequence identity with the eNOS reductase domain, has an uncoupled oxygen consumption rate of ϳ1 min Ϫ1 (measured in the present system). 3 These results indicate that: 1) flavins in the NADPH-cytochrome P-450 reductase and eNOS are poor electron donors for molecular oxygen and 2) although slow, there are probably two sites for oxygen activation in eNOS, one in the reductase domain (presumably the flavins) (39) and the other in the oxygenase domain (presumably the heme), because binding of CaM to eNOS enables electron transfer from the reductase domain to oxygenase domain (1,40) and activates overall oxygen consumption.
L-Arginine and BH 4 Effects-The oxygen chemistry on the heme site of NOS has been generally assumed to follow that of cytochrome P-450 monooxygenases (1,(41)(42)(43). A simplified list of intermediates in the initial steps of heme-oxygen interaction can be represented as follows. The nNOS trace shown in this picture was taken from our previous publication for comparison (9). Oxygenated standard reaction buffer and 7 M CaM, 23 g/ml SOD, 1.7 M catalase, 5.5 M BH 4 , and 1.2 M eNOS were mixed before the addition of 440 M NADPH. A, L-arginine effect (500 M L-arginine after addition). B, N -nitro-Larginine methylester (L-NAME) and N G -monomethyl L-arginine citrate (L-NMMA) were introduced after the reaction started; no stimulation was observed. C, two additions of L-citrulline were introduced after the reaction started; no stimulation was observed. SEPTEMBER 28, 2007 • VOLUME 282 • NUMBER 39

REACTION 11
After these initial steps, activated oxygen can either heterolytically dissociate from heme to form superoxide anion, or follow the cytochrome P-450 scheme to complete the monooxygenation of L-arginine.

REACTION 13
The intermediates in Reaction 11 have been intensively studied with optical absorption spectroscopy, by stabilizing them under cryogenic conditions or by capturing them in real time with time-resolved measurements (19 -21, 31, 44 -47). With the heme known as the center for oxygen chemistry in NOS, we assume Reaction 12 is the major source for O 2 . formation from the uncoupled eNOS in the presence of CaM (5,9,10,30,34,39). However, the two intermediates (Fe III O 2 2Ϫ and Fe IV ϭO) in Reaction 13 have not been convincingly observed to date, and it is unclear whether different NOS isoforms follow the same scheme in Reaction 13.
We report that the most striking difference between the oxygen metabolism of eNOS and nNOS is the L-arginine effect, wherein L-arginine strongly stimulates oxygen consumption of eNOS and inhibits that of nNOS (Fig. 3A). It is well known that recombinant nNOS is 5-7 times faster than eNOS in terms of NO production, and the difference is generally attributed to the rates at which the reductase domains provide electrons to the heme (25,48). However, a comparison of the rates listed in Table 2 reveals a huge difference between eNOS and nNOS in oxygen consumption in the absence of L-arginine. So the question is: why is L-arginine-free eNOS so slow, and why is L-arginine-free nNOS so fast?
The slow oxygen consumption rate (ϳ4 min Ϫ1 ) reported in the current study is the result of either inefficient Fe II O 2 formation in the absence of L-arginine or a stabilized Fe II O 2 complex that does not easily dissociate to form O 2 . (Reactions 11 and 12). Tsai and co-workers (31) have studied the L-arginine effects on oxygen binding to eNOS in great detail, and their observations highly support the first hypothesis. Monitored at 427 nm, the accumulation of the Fe II O 2 complex was shown to be much higher in the reaction with L-arginine than without (31), meaning that the presence of L-arginine facilitates the binding of molecular oxygen to ferrous heme in eNOS. More detailed kinetic data demonstrate a 7-8-fold slower association constant, as well as a 20 -30-fold slower dissociation constant of oxygen in the presence of L-arginine, helping to explain the accumulation of Fe II O 2 complex. These kinetic data, although self-sufficient, are in agreement with the paradoxical L-arginine effects reported by other groups. On the one hand, the binding of L-arginine impedes the binding of diatomic molecules such as NO and CO as demonstrated by a variety of methods, including EPR (49 -51), and it is logical to expect anti-cooperative binding of L-arginine and oxygen. On the other hand, L-arginine conversely stabilizes the Fe II O 2 complex and slows down oxygen dissociation (52).
Up to now, there is no crystal structure for the O 2 -bound derivative of any of the NOS isoforms; thus, some details of oxygen binding geometry are still missing. Nevertheless, different groups have reported isoform-specific environments in the binding of diatomic molecules to nNOS and eNOS, using infrared/Raman spectroscopy and EPR (50,(53)(54)(55)(56)(57). All of these studies agree that despite the general similarity in structures, there are small differences among NOS isoforms that allow limited isoform specificity in the interaction between L-arginine and diatomic molecules, such as NO, CO, or O 2 .
A resonance Raman study by Rousseau and co-workers (58) presented evidence supportive of the differences in oxygen metabolism between eNOS and nNOS reported here. By monitoring the Fe-CO and C-O stretching modes of CO-ferrous complexes with bound L-arginine, the authors intriguingly showed that nNOS has a much more open substrate-binding pocket than eNOS. Regardless of the two possible interpretations proposed by this paper, the conclusion highly supports the tighter coupling of eNOS (Table 1) and the faster dissociation of superoxide from nNOS ( Table 2).
Complementary electron-nuclear double resonance studies was performed by Hoffman, Masters, and co-workers (59 -62), generating a series of publications regarding substrate binding to all three NOS isoforms. Based on spectral differences, the authors concluded that nNOS provides greater shielding of the charge on the guanidino nitrogen than eNOS does (63), which is one of the two interpretations in the previously mentioned resonance Raman study (58).   This differential L-arginine effect could be attributable to BH 4 . In the L-arginine-free eNOS, BH 4 does not accelerate superoxide dissociation (Fig. 3A), but with nNOS, it was reported that BH 4 accelerated the decay of the Fe II O 2 complex by a factor of 70, irrespective of the presence of L-arginine (52).
Even though BH 4 does not seem to directly affect oxygen binding in eNOS (31,50), it controls the coupling of eNOS and, to some extent, the fate of oxygen reduction. The imperfect coupling of eNOS in the presence of substrate and absence of BH 4 is evidenced not only in this study (Fig. 4), but also by other research groups using different approaches (33,64,65). In the presence of both substrate and BH 4 , however, eNOS is highly coupled (30,64). The results presented herein (Fig. 4) confirm that BH 4 is the "coupling switch" of eNOS; our approach is unique in that we quantitatively monitor oxygen consumption and utilize chemical reactions to convert accumulated ROS to visible and measurable molecular oxygen (Reactions 4, 6, 9, and 10). BH 2 , a redox-incompetent BH 4 analog, is unable to inhibit ROS formation by eNOS (Fig. 4C). Our results confirm the view that BH 4 takes a redox role in the catalytic cycle, most likely by providing the second electron to the heme (21). Electron transfer from BH 4 is proposed to be faster than that from the reductase domain and thus is consistent with the increased rate of oxygen uptake in the presence of BH 4 over that in its absence (Fig. 4A). Under pathological conditions, the limited availability of BH 4 has been shown to trigger eNOS uncoupling and increase the risk of cardiovascular diseases including atherosclerosis, coronary artery disease, and hypertension (66 -73). In addition, external stimuli, such as the chemicals in cigarettes, can convert BH 4 to BH 2 in the human body (74 -76). Our results support the view that the introduction or stabilization of BH 4 may be used for the treatment of endothelial dysfunction (70 -72, 77).
Stoichiometry-We report for the first time the quantitative measurement of eNOS oxygen stoichiometry. In the absence of L-arginine, the stoichiometry is simple, and the experimental values match the theoretical values (Table 1). Assessing the results from the simultaneous measurements (Fig. 2) and from the oxygen consumption experiment (Fig. 5A), it is clear that eNOS utilizes 1 mol of NADPH to provide electrons to 2 mol of molecular oxygen, generating 2 mol of superoxide anion as the product of the uncoupled reaction. When L-arginine or NOHA is added in the presence of BH 4 , however, the eNOS reaction is highly coupled. A comparison with nNOS data (9) illustrates two points: 1) eNOS is more coupled than nNOS, with either L-arginine or NOHA as substrate and 2) in both eNOS and nNOS, the enzymes are more coupled with L-arginine as substrate than with NOHA. The first point can be easily understood, given the physiological importance of the precise control of blood pressure (41), and the second one is at least partially because of the substrate-specific H-bonding interactions with the terminal oxygen, as recently reported by Li et al. (53).
In conclusion, we have demonstrated the CaM, L-arginine, and BH 4 effects on eNOS oxygen metabolism. We report, for the first time, the quantitative measurement of superoxide anion formation from eNOS. Our recombinant eNOS is 92-96% coupled in the presence of L-arginine, which strongly stimulates oxygen uptake. In the presence of either L-arginine or NOHA, the availability of BH 4 controls eNOS coupling and diminishes ROS formation by eNOS. This exquisite control is necessitated by its role in the vascular system in which the production of ROS must be prevented. It is verified that oxygen consumption by eNOS in the absence of substrate is extremely low, and only when substrate is supplied does NO production proceed in a highly coupled, physiologically significant process.