Nitroxides Increase the Detectable Amount of Nitric Oxide Released from Endothelial Cells*

Nitroxides are known to exert superoxide dismutase-mimetic properties and to decrease O·̄2- and H2O2-mediated cytotoxicity. However, the effect of nitroxides on ⋅NO homeostasis has not been studied yet. The present study investigates the effect of nitroxides on the detectable amount of ⋅NO released by 3-morpholinosydnonimine (SIN-1) and cultured endothelial cells. Cultured bovine aortic and atrial endothelial cells stimulated with 10 μm A23187 released a stable flux of ⋅NO, as detected by ⋅NO chemiluminescence. Addition of 100 units/ml SOD or 10 μmof the nitroxides 4-hydroxy-2,2,6,6-tetramethylpiperidine-N-oxyl (TEMPOL), 3-carboxy-proxyl, and 3-ethoxycarbonyl-proxyl, increased the chemiluminescence signal. The effect of these nitroxides on the amount of ⋅NO released from cell monolayers was dose-dependent, with the highest efficacy between 30 and 100 μm. EPR spin trapping in SIN-1 solutions revealed the formation of ⋅OH adducts from spontaneous dismutation of O·̄2 and concomitant reaction with H2O2. Both SOD and TEMPOL increased the signal intensity of the ⋅OH adduct by accelerating the dismutation of O·̄2. The results of this study demonstrate that the SOD-mimetic activity of nitroxides increases the amount of bioavailable ⋅NO in vitro.

Nitric oxide ( ⅐ NO) 1 is an ubiquitous endogenously produced free radical. The physicochemical properties allow ⅐ NO to serve as a biological messenger. ⅐ NO can exert cytoprotective and cytotoxic effects depending on its concentration, site of generation, and the reactions it undergoes (1)(2)(3). Released from endothelial cells, it acts as an endothelium-derived relaxing factor (EDRF) with anticoagulant and antithrombotic properties (4). Disturbances in ⅐ NO release and its decreased stability and bioactivity are proposed to be a major part of vascular diseases such as atherosclerosis, ischemia, or hypertension (5)(6)(7).
Like ⅐ NO, superoxide anion (O 2 . ) is a free radical with a relatively low overall reactivity (e.g. compared with the hydroxyl radical, ⅐ OH). It is formed as an intermediate in a variety of enzymatic reactions and is kept within a physiological concentration range by superoxide dismutase (SOD, EC 1.15.1.1) (8,9). It has been shown that SOD increases the half-life of EDRF released from isolated arteries (10). Elevated O 2 . production (e.g. via the xanthine oxidase, arachidonic acid, or NADH oxidoreductase pathway) and a concomitant increase in cytotoxicity are described for different pathological conditions (11). The reaction of O 2 . with ⅐ NO at a diffusion controlled rate not only depletes both radicals, it also leads to the formation of more toxic species, such as peroxynitrite (12,13). Thus, superoxide dismutase-mimetic compounds could be of therapeutic interest in the conventional context (e.g. as antiinflammatory agents; Ref. 14) but also by increasing the lifetime and biological activity of ⅐ NO. 3-Morpholinosydnonimine (SIN-1), a well known nitrovasodilator simultaneously releasing ⅐ NO and O 2 . (15), provides a model system for evaluation of the interaction of both species and the influence of SOD-mimetic compounds. Among SOD-mimetic compounds, low molecular weight copper or iron complexes were found to be very effective (16,17). Cytoprotection, at least partly due to SOD-like activity, was also reported for nitroxides, a class of free radicals widely applied as tools in electron paramagnetic resonance (EPR) spectroscopy (18). A subgroup of nitroxides, nitronyl nitroxides, can be used for ⅐ NO detection (19); however, reduction by O 2 .
and other reducing agents limits its application (20). As shown in Scheme 1, nitroxides can oxidize O 2 . to molecular oxygen (I).
The resulting hydroxylamine is EPR-silent and is oxidized back to the nitroxide by reducing another O 2 . to hydrogen peroxide (H 2 O 2 ) (II) (21). Cell culture studies revealed an inhibition of superoxidemediated cytotoxicity and mutagenicity by nitroxides similar to SOD (22,23). Although SOD activity of nitroxides could not be found by stopped-flow kinetic analysis (24), nitroxides have been identified as genuine SOD mimetics rather than O 2 . scavengers by direct and indirect physicochemical methods (25). Additionally, protection against oxidative damage independent of O 2 . and H 2 O 2 was found and proposed to result from nitroxide-mediated oxidation of redox-active trace metal ions (26). Because of the predominantly intracellular generation of O 2 . , membrane permeation by SOD mimetics or superoxide scavengers applied to biological systems is an important aspect. Moreover, stability and toxicity have to be considered. Nitroxides fulfill these requirements (27); they are relatively stable low molecular weight compounds with non-immunogenic properties, their toxicity is low, and, most important, their synthesis * This work was supported by a grant from the Schering Forschungsgesellschaft (to S. Z.), Deutsche Forschungsgemeinschaft Grant SFB 507, and Bundesministerium fü r Bildung und Forschung Grant WTZ 224.5 (to I. A. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
allows large variations of physicochemical properties, such as lipophilicity.
However, the effect of nitroxides on ⅐ NO generated by model systems (e.g. SIN-1) or endothelial cells has not been investigated so far. This study demonstrates that, similar to SOD (28,29), nitroxides augment the detectable amount of ⅐ NO released from SIN-1 and cultured endothelial cells by selectively removing O 2 . .
Quality of the endothelial cell cultures was verified by phase contrast microscope-detected cobblestone appearance at confluence, presence of factor VIII antigen, contents of alkaline phosphatase, and angiotensinconverting enzyme.

⅐ NO Chemiluminescence
The determination of ⅐ NO was performed using a ⅐ NO analyzer NOA270B (BAEC experiments) or NOA280 (BAtEC experiments) (Sievers, Boulder, CO) without use of a reducing agent. The output signal (mV) is proportional to the amount of ⅐ NO. The signal was recorded using a chart recorder (Gould, Cleveland, OH) and a modified high performance liquid chromatography data acquisition and calculation system (ACCESS CHROM, Perkin Elmer Nelson Systems, Cupertino, CA) and either expressed as mV (headspace) or as nanomolar concentration of ⅐ NO (supernatant). The ⅐ NO concentration (nM) calculation was performed only with samples from the supernatant solution, where ⅐ NO standards were used as a basis of calculation.

Detection of ⅐ NO Released from Cultured Endothelial Cells
The ⅐ NO release from endothelial cells was measured (a) in samples of the supernatant solution (for BAtEC) and (b) continuously in the headspace (for BAEC). All cell experiments were performed at 37°C.
Method a: Measurement of ⅐ NO in Supernatants (Liquid Samples)-Cells grown in six-well plates were washed twice with PSS and covered with 1 ml of PSS containing the required agents. After incubation for 30 min, 0.8 ml of the supernatant was injected into a special purge vessel with a gas-tight syringe (Hamilton, Reno, NV). ⅐ NO was expelled and transported to the reaction chamber by a stream of helium. The area under the resulting peak was calculated and calibrated with peak areas from ⅐ NO standards. Method b: Measurements of ⅐ NO in Headspace-Cells grown in flasks (25 cm 2 ) were washed as described for (a) and covered with 2 ml of PSS. Through a tube placed above the supernatant, headspace gas was drawn by vacuum continuously into the reaction chamber. Under these conditions, determination of ⅐ NO is based on the fixed distribution of gases between liquid and gas phase depending on their solubility. During the experiment the flask was shaken gently in a shaking incubator (Boekel Industries, Feasterville-Trevose, PA).

Detection of ⅐ NO released from SIN-1
For experiments with SIN-1, 3 ml of freshly prepared aqueous solution was placed into a sealed glass vessel and stirred with a magnetic stirrer. The rubber seal contained two tubes: one for headspace gas that was drawn into the reaction chamber and one inlet for gas exchange. The cell-free experiments were performed at room temperature (22°C).

⅐ NO Standard
⅐ NO released into the supernatant was quantified by comparison with ⅐ NO standards. Diluted ⅐ NO gases (65 and 6.8 ppm in N 2 ) were used as standards. For preparation of ⅐ NO solutions, deionized water was deaerated, saturated with argon, and gassed with a continuous stream of diluted ⅐ NO (65 ppm) for at least 20 min, resulting in a concentration of 130 nM at 22°C. There was no significant difference in the area under the curve using gaseous (6.8 ppm) or liquid standards containing the same amount of ⅐ NO. Standard curves were recorded each day. The detection limit was less than 1 pmol (1 nM ⅐ NO concentration for an injected sample volume of 1 ml) at a signal-to-noise ratio of 3 for single injections of ⅐ NO. The response was linear at least up to 1 M ⅐ NO concentration.

EPR Spectroscopy
EPR experiments were carried out at room temperature on a Bruker ECS 106 X-band spectrometer (equipped with a high sensitivity rectangular-mode cavity ER 4102 ST). The samples were placed into a flat quartz cell, and standard experimental conditions were as follows: modulation frequency, 100 kHz; modulation amplitude, 0.1 mT; field set, 347.5 mT; scan range, 10.0 mT; microwave power, 10 milliwatts.

Protein Determination
The protein content of BAtEC was measured for selected experiments with a kit for protein determination (per procedure no. TPRO-562, Sigma). Cells were lysed by incubation with 1% Triton X-100 (v/v) for 30 min. Bovine serum albumin was used as standard.

Cytotoxicity Detection
Cell death was evaluated by quantification of lactate dehydrogenase (LDH) release into the cell supernatant as an index for plasma membrane damage (34). The concentration of LDH was measured with a commercially available cytotoxicity detection kit (Boehringer Mannheim).

Calculations and Statistical Analysis
The dose response of SOD and nitroxides on ⅐ NO release was done in sets of n ϭ 4 -6 (n represents the number of experiments performed on different cell cultures or model systems). ⅐ NO concentration was expressed as mean Ϯ S.E. Significant difference between means of control and treatment groups was calculated by Student's t test; a value of p Ͻ 0.01 was accepted for statistically significant difference.

RESULTS
Detection of ⅐ NO Released from SIN-1-SIN-1 was chosen to study the interaction of ⅐ NO with O 2 . in a cell-free system. The ⅐ NO release from freshly prepared aqueous solutions of SIN-1 increased during the first 30 -40 min and reached afterwards a stable plateau (Fig. 1, inset) as detected by ozone-mediated chemiluminescence. During the plateau phase, SOD was added to remove O 2 . produced by SIN-1. The detectable amount of ⅐ NO was rapidly elevated by a factor of 16.7 (Fig. 1, trace a; Table I). SCHEME 1 A similar but less prominent effect was observed when the nitroxides were added under the same conditions (Fig. 1, traces b-d; Table I).
EPR Detection of Free Radicals Released from SIN-1-Addition of 1 mM SIN-1 to a 0.1 M DMPO solution resulted in the appearance of a spin adduct spectrum with a N ϭ a H ␤ ϭ 1.49 mT, indicating the formation of the DMPO-hydroxyl radical adduct (DMPO/ ⅐ OH, Fig. 2a). In the presence of either 100 units/ml SOD or 10 M TEMPOL (giving a strong triplet signal superimposition), a significant increase of the spin adduct signal intensity was found (Fig. 2, b and c). When catalase (0.5 mg/ml) was added to the reaction mixtures, formation of DMPO spin adducts was prevented both in the presence of SOD (Fig.  2d) and nitroxides (spectra not shown).
Detection of ⅐ NO Released from Endothelial Cells-Endothelial cell monolayers grown in a culture flask are not exposed to flow or shear stress; unstimulated ("basal") ⅐ NO release from BAEC was not detected in the headspace of this system. A sustained and reproducible ⅐ NO signal was measured after incubation with 1-10 M Ca 2ϩ ionophore A23187 (5 M, Fig. 3). ⅐ NO release was abolished by 0.1 mM N -nitro-L-arginine methyl ester (L-NAME) in the absence of exogenous L-arginine.
Similar to BAEC, exposure of BAtEC to 10 M A23187 in a six-well plate resulted in an increase of ⅐ NO concentration in the supernatant reaching steady state after approximately 15 min and lasting for at least 30 min (12.4 Ϯ 0.7 nM, corresponding to a production rate of approximately 2.3 pmol of ⅐ NO/ min/mg of protein).
⅐ NO Release from Endothelial Cells in the Presence of SOD-Addition of SOD resulted in an increase of the amount of ⅐ NO detected using both BAEC and BAtEC after stimulation with A23187. Fig. 3 displays a representative experiment with BAEC, whereas Fig. 4 illustrates SOD-induced concentrationdependent augmentation of detectable ⅐ NO released from BAtEC stimulated with different concentrations of A23187. Higher concentrations of SOD (up to 300 units/ml) did not cause a further increase of detectable ⅐ NO (data not shown).    Table I. DISCUSSION The study of the interaction of O 2 . and ⅐ NO is complicated by several factors; both species are unstable, the rate constant of their reaction is exceptionally high, and the concomitant formation of other reactive species has to be considered. These restrictions were taken into account by application of appropriate methods for detection of these radicals: ozone-mediated chemiluminescence (for authentic ⅐ NO) and EPR spectroscopy (for other radical species). This approach allowed us to study an effect of SOD-mimetic nitroxides that has not been investigated so far: the influence of these SOD-mimetic compounds on ⅐ NO released from SIN-1 and cultured endothelial cells. Two different modes of ⅐ NO analysis were used in this study: (a) headspace measurements for qualitative time-course assessment, and (b) discrete sample collection from the supernatant of cells at single time points. The assay of the ⅐ NO detection in solution represents absolute values for dissolved ⅐ NO, whereas headspace measurements allow resolution of the time course of the ⅐ NO generation within a period of 5 s.
Aqueous solutions of 1 mM SIN-1 resulted in a stable flux of ⅐ NO into the headspace after an initial lag phase of approximately 30 -40 min. Similarly, spontaneous ⅐ NO release measured by the conversion of oxyhemoglobin to methemoglobin has been described by Feelisch et al. (15). In contrast, using a chemiluminescence technique with a helium-flushed, gas-permeable tubing inserted into the SIN-1 solution for sample collection, Beckman et al. did not observe ⅐ NO release in vitro in the absence of SOD (35 Furthermore, the oxygen-dependent breakdown of SIN-1A might decline under conditions where no oxygen supply is guaranteed, as it would happen in an airtight sealed reaction vessel. Addition of 100 units/ml SOD to SIN-1 solution at the plateau phase of ⅐ NO release caused a significant 16.7-fold increase in the headspace concentration of ⅐ NO (Table I). This effect was also observed after addition of the nitroxides (10 M). The order of efficacy was as follows: SOD Ͼ TEMPOL Ͼ CP Ͼ ECP (Table I).
EPR spin trapping experiments using DMPO were performed to characterize the radical species formed during the decomposition of SIN-1 and to verify the SOD-mimetic activity of nitroxides (26,27 (38). Our data clearly demonstrate the SOD-mimetic activity of nitroxides in the SIN-1 model system. Furthermore, there is no indication of peroxynitrite-dependent ⅐ OH formation, which would be independent from catalase activity (35,39).
To study the effect of nitroxides on ⅐ NO produced by cultured endothelial cells, a stable and maintained ⅐ NO release was induced by stimulation of the cells with the Ca 2ϩ ionophore A23187 for up to 30 min. Within this time frame, there was no sign of decreased cell viability as measured by release of LDH into the extracellular space (data not shown). Increase of intracellular free calcium concentration is a stimulus not only for ⅐ NO production (i.e. activation of endothelial ⅐ NO synthase, ecNOS), but for other processes as well, including production of O 2 . (40,41), which could decrease the detectable amount of ⅐ NO   Table I).
The increased detectable concentration of ⅐ NO in the pres- ence of the nitroxides could be explained either by enhanced ⅐ NO release, decreased ⅐ NO removal, or a combination of both. Considering the chemistry of SIN-1 degradation (45), there is no reaction known by which SOD or the nitroxides would selectively increase ⅐ NO liberation without a concomitant increase in O 2 . release. Similarly, an activation of ecNOS (direct or indirect via increase of intracellular free calcium concentration or cofactors) leading to an elevated ⅐ NO biosynthesis is rather unlikely. Therefore, the most probable explanation is that the nitroxides increase the detectable amount of NO via reducing ⅐ NO degradation by removal of O 2 . from the system.
The EPR data clearly demonstrated an SOD-like action of the investigated nitroxides in the SIN-1 model. Thus, the most probable explanation for nitroxide-induced increase in detectable ⅐ NO release from endothelial cells is the selective removal of O 2 . by these compounds.
Genetically elevated amounts of endogenous SOD (46), SOD derivatives (47), or SOD-mimetic nitroxides (18) have been reported to reduce the cytotoxic effect of oxygen free radicals in vitro and in vivo. On the other hand, it must be mentioned that decreasing the O 2 . concentration alone by catalyzing its dismutation does not necessarily reduce an elevated cytotoxic potential (38). The spin trapping experiments point out that catalase activity is required for the further destruction of H 2 O 2 to nontoxic species.
In summary, this study demonstrates that the SOD-mimetic activity of nitroxides increase the amount of bioavailable and detectable ⅐ NO in vitro. Since O 2 . is a main cause of ⅐ NO degradation, its removal would be beneficial especially in situations when O 2 . is increased leading to the inactivation of ⅐ NO and formation of toxic species such as peroxynitrite (i.e. in arteriosclerosis or inflammation). Since the discovery of SODmediated increase in stability of EDRF or ⅐ NO in in vitro systems (28,29), the body of evidence is growing supporting a role of O 2 . and its effective removal by superoxide dismutation as determinants of ⅐ NO bioavailability under physiological and pathological conditions. Nitroxides exert the properties of potential pharmacological agents since they are (compared with SOD) relatively stable low molecular weight compounds without immunogenic properties and with low toxicity (26,48). Their structure allows synthetic modifications necessary for adaptation to the intended use. Nevertheless, it has to be determined in further in vitro and in vivo studies whether the SOD-mimetic properties of nitroxides could be of therapeutic importance against O 2 . -mediated lowering of bioavailable ⅐ NO.