Production of Nitric Oxide by Mitochondria*

The production of NO⋅ by mitochondria was investigated by electron paramagnetic resonance using the spin-trapping technique, and by the oxidation of oxymyoglobin. Percoll-purified rat liver mitochondria exhibited a negligible contamination with other subcellular fractions (1–4%) and high degree of functionality (respiratory control ratio = 5–6). Toluene-permeabilized mitochondria, mitochondrial homogenates, and a crude preparation of nitric oxide synthase (NOS) incubated with the spin trapN-methyl-d-glucamine-dithiocarbamate-FeIIproduced a signal ascribed to the NO⋅ spin adduct (g = 2.04; a N = 12.5 G). The intensity of the signal increased with time, protein concentration, andl-Arg, and decreased with the addition of the NOS inhibitorN G-monomethyl-l-arginine. Intact mitochondria, mitochondrial homogenates, and submitochondrial particles produced NO⋅ (followed by the oxidation of oxymyoglobin) at rates of 1.4, 4.9, and 7.1 nmol NO⋅ × (min·mg protein)−1, respectively, with aK m for l-Arg of 5–7 μm. Comparison of the rates of NO⋅ production obtained with homogenates and submitochondrial particles indicated that most of the enzymatic activity was localized in the mitochondrial inner membrane. This study demonstrates that mitochondria are a source of NO⋅, the production of which may effect energy metabolism, O2consumption, and O2 free radical formation.

Nitric oxide (NO ⅐ ) 1 is a free radical generated in biological systems by nitric oxide synthases (NOS). Because of its effect on neurotransmission, vasodilation, and immune response (1)(2)(3), NO ⅐ plays an important role in physiology, pathology, and pharmacology.
Studies with brain tissue and macrophage lysates have shown that NOS is localized exclusively in the soluble fraction (3)(4)(5)(6), and recent studies have indicated that the majority (Ͼ80%) of bovine endothelial NOS activity is bound to the particulate fraction of cell homogenates (7,8). Because the particulate fraction used in the studies was expected to contain plasma membranes, as well as microsomes, and, possibly, intracellular organelles, the actual subcellular location of the activity remained to be determined. Other lines of evidence have indicated the presence of NOS in the perinuclear region, in discrete regions of the plasma membrane of cultured endothelial cells, and in intact blood vessels (9,10); immunocytochemical studies have revealed the presence of a NOS, or an antigenically related protein, in mitochondria isolated from different tissues (11)(12)(13). The predominant association of this mtNOS with the mitochondrial membrane (11,12), and its co-localization with succinate dehydrogenase, a mitochondrial marker of the inner membrane (13), suggested that this enzyme has a particulate localization.
These studies as well as the presence of substrates and cofactors in mitochondria required for NOS activity such as L-arginine (L-Arg), L-Arg transporters, Ca 2ϩ , calmodulin, NADPH, and the availability of O 2 , led us to postulate mitochondria as a potential source of NO ⅐ production.
Following the use of a specific spin-trapping agent and the controlled oxidation of oxymyoglobin, NO ⅐ production was detected in purified mitochondrial preparations (intact mitochondria, permeabilized mitochondria, mitochondrial homogenates, and submitochondrial particles) and crude preparations of NOS (crude fraction) obtained from rat liver.
Given the important implication of a mitochondrial production of NO ⅐ for energy conservation mechanisms and free radical production, such production may serve as the basis for a new understanding of biochemical regulation, based on the ubiquitous distribution of mitochondria and the diffusibility of NO ⅐ through cellular membranes.

Biological Materials
Liver mitochondria were isolated from adult Wistar rats (180 -200 g) by differential centrifugation, essentially as described in Ref. 15 1 The abbreviations used are: NO ⅐ , nitric oxide; MGD-Fe, N-methyl-D-glucamine-dithiocarbamate-Fe II complex; NMMA, N G -monomethyl-Larginine; NIO, L-N 5 -(1-iminoethyl)ornithine; TEMPO or TEMPOL, 4-hydroxy-2,2,6,6-tetramethylpiperidine-1-oxyl; mtNOS, mitochondrial nitric oxide synthase; NOS, nitric oxide synthase; SMP, submitochondrial particles; EPR, electron paramagnetic resonance; G, gauss. serum albumin (fatty-acid free), 2 mM Hepes/KOH, pH 7.4) at 4°C. The homogenate was centrifuged at 600 ϫ g for 5 min in a JA-17 rotor. The supernatant was centrifuged at 10,300 ϫ g for 10 min. Intact, purified mitochondria were isolated using Percoll to remove contaminating organelles and broken mitochondria (16). Percoll was selected as the gradient medium because of its negligible osmolarity and chemical inertness (17,18). The pellet (mitochondrial fraction) was resuspended in 5 ml of MSHE supplemented with 20 ml of 30% Percoll in 225 mM mannitol, 1 mM EGTA, 25 mM Hepes, 0.1% bovine serum albumin (fatty-acid free). This solution was spun at 95,000 ϫ g in a Beckman Ti60 rotor for 30 min. The fraction with a density of 1.052-1.075 g/ml was collected and washed twice with MSHE at 6,300 ϫ g for 10 min to remove the Percoll. Purified mitochondria were washed twice with 150 mM KCl, followed by two washings with MSHE. Permeabilized mitochondria were prepared using a controlled treatment of purified mitochondria with toluene (19). Mitochondria were resuspended to 20 mg of protein/ml in MSHE supplemented with 8.5% (w/v) polyethylene glycol 8000 at 0 -4°C. Toluene (0.2%, v/v) was added to the suspension, which was gently inverted every 15-30 s over a 2-min period. Mitochondria were then pelleted by centrifugation at 15,000 ϫ g for 4 min at 4°C, and the supernatant was removed by aspiration. Mitochondria were washed twice, and resuspended, in MSHE containing polyethylene glycol. Submitochondrial particles (SMP) were obtained by sonication (20) of purified mitochondria and stored at Ϫ20°C at a protein concentration of 20 mg/ml. Mitochondrial homogenates were obtained by mechanical homogenization of the mitochondrial fraction subjected to three cycles of freeze-thawing. A crude fraction of NOS was obtained from purified rat liver mitochondria. Mitochondria from two to four livers were homogenized with Buffer A (1 mM EDTA, 5 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 50 mM Hepes, pH 7.5). This homogenate was centrifuged at 150,000 ϫ g for 1 h at 0-4°C. The mitochondrial membranes were washed with Buffer B (Buffer A plus 1 M KCl, 10% glycerol) and centrifuged at the same speed for 30 min. The pellet was treated with Buffer A plus 20 mM CHAPS at 4°C under continuous stirring. After 30 min, the suspension was centrifuged at 150,000 ϫ g for 30 min. This supernatant, concentrated in a Centricon™ cartridge (M r cut-off of 30,000; Amicon, Danvers, MA), was referred to as crude fraction (150,000 ϫ g supernatant).

Detection of Nitric Oxide
Electron Paramagnetic Resonance Assay-Aliquots of the samples, containing the NO ⅐ spin trap MGD (21,22), were transferred to bottomsealed Pasteur pipettes, and the spectra were recorded at 22°C with an X-band EPR spectrometer, operating at 9.77 GHz. Instrument settings are described in the legend for Fig. 1. The settings were selected using either sodium nitroferricyanide or nitrosoglutathione as NO ⅐ -releasing agents.
Oxymyoglobin Spectrophotometric Assay-The oxidation of oxymyoglobin (23) was followed at 581-592 nm (⑀ 581-592 ϭ 11.6 mM Ϫ1 cm Ϫ1 ) with a double-beam spectrophotometer Hitachi U-3110 with a multiple wavelength program at 22°C using 50 M oxymyoglobin. Superoxide dismutase (1 M) and catalase (1 M) were added to prevent interference by O 2 . and H 2 O 2 , respectively, that might be produced by the

Protein Determination
Protein was determined by the Lowry assay (24) using bovine serum albumin as standard.

Data Evaluation
All assays were done in duplicate and were repeated five to eight times in separate experiments using 2-4 rats/experiment. Data are presented as mean Ϯ S.E., in which the S.E. were between 10 and 12% of the mean values.

Assessment of Purity and Functionality of Rat Liver
Mitochondria-Rat liver mitochondria were isolated by differential centrifugation (15), purified by Percoll centrifugation (16), and washed with high ionic strength solutions. This procedure allowed the efficient removal of contaminating organelles, broken mitochondria, arginase (25), and adsorption artifacts, yielding a highly purified preparation. This is supported by the low degree of non-mitochondrial contamination (1-4%; Table  I), which was comparable with, and in some cases less than, that obtained with other purification procedures (18,26,27). Mitochondria isolated using this procedure differ from those obtained by differential centrifugation in that the former exhibited a higher respiratory control ratio, indicating functional integrity and membrane intactness (Table II).
Detection of Nitric Oxide in Mitochondrial Preparations by Spectroscopic Techniques: Detection of NO ⅐ by Spin Trapping/ EPR-Toluene-treated mitochondria with an increased permeability for the spin trap and external NADPH were incubated with the spin trap ((MGD) 2 /Fe II ) for 1 h at room temperature. A weak EPR signal consisting of a triplet (a N ϭ 12.5 G; g iso ϭ 2.04; intensity ratio 1:1:1; Fig. 1A) was assigned to the (MGD) 2 / Fe II -NO ⅐ complex (21,22) by comparison with the signal obtained with the NO ⅐ donor, nitrosoglutathione (Fig. 1E). In addition to the triplet, a line from a quartet signal was present assigned to the (MGD) 2 /Cu II complex produced by the reaction of free Cu II , in the homogenate or in the reaction solution, with the excess of MGD. The addition of L-Arg increased the signal by 30% (Fig. 1B), whereas N G -monomethylarginine (NMMA), the competitive inhibitor of NOS (30), decreased the signal by 50% (Fig. 1C) and 20% (Fig. 1D) in the absence and presence of L-Arg, respectively. Based on its sensitivity to NMMA, the formation of the (MGD) 2 /Fe II -NO ⅐ signal in the biological sample without L-Arg suggests an endogenous pool of Arg capable of sustaining NO ⅐ production through an NOS-catalyzed reaction (Fig. 1C). The NMMA-insensitive EPR signal suggested the presence of a labile pool of NO ⅐ . 2 Of note, the addition of 5 2 C. Giulivi, unpublished observations. g/ml calmodulin and/or 1 mM Ca 2ϩ did not significantly affect the signal intensity, indicating that NOS was fully active with the cofactors present in our preparations. Toluene-treated rat liver mitochondria, incubated with the spin trap for 8 h to increase the substrate and inhibitor concentrations in the mitochondrial matrix, exhibited signal intensities 2.3-and 2.7-fold higher than those found at 1 h, with and without L-Arg addition, respectively (Fig. 2, A and B). Preincubation of mitochondria with NMMA inhibited the signal formation by 50% and 15% in the absence and presence of L-Arg, respectively (Fig. 2, C and D). A weak (MGD) 2 /Fe II -NO ⅐ signal was also noted in the absence of mitochondrial homogenate (data not shown); this signal is likely to originate from NO ⅐ diffusion from ambient air (about 0.1 ppm).
Mitochondrial homogenates or a crude fraction of NOS incubated for 1 h with L-Arg and the spin trap showed the same three-line EPR spectrum observed with toluene-permeabilized mitochondria (Table III). The EPR signal intensities obtained with crude fraction were 5-7-fold higher than those observed with permeabilized mitochondria. The signal intensities were increased by L-Arg supplementation and decreased by NMMA addition to different extents, depending on the specific biological preparation (Table III). Similar results were obtained using L-N 5 -(1-iminoethyl)ornithine (NIO), another NOS inhibitor (data not shown). The high concentrations of NMMA required in these experiments are indicative of the competitive kinetics of the inhibition of NOS by NMMA in the presence of an endogenous pool of L-Arg, the latter most likely sustained by proteolytic activities present in the samples.
The iron-nitrosyl complex signal intensities obtained with toluene-treated rat liver mitochondria after 1 h of incubation showed a linear dependence on protein concentration. Double integration of the EPR signals and interpolation of calibration curves of the area of the low-field peak versus TEMPO or TEMPOL allowed the quantification of the EPR signals. This quantification permitted the calculation of an NO ⅐ production rate of 48 Ϯ 2 nmol/mg mitochondrial protein (r ϭ 0.99). The NO ⅐ experimentally detected by EPR accounted for approximately 16% of the rate of metmyoglobin formation, even when an excess of spin trap was used. This underestimation may be due to the metabolism of the spin trap by mitochondrial prep-

FIG. 2. EPR spectra of permeabilized mitochondria supplemented with NOS substrates and inhibitors after 8 h of incubation.
Toluene-permeabilized mitochondria (5.2 mg of protein) were added to reaction mixtures without (A and C) and with (B and D) 1 mM L-Arg, plus 10 mM NMMA (C and D) and then incubated for 8 h at room temperature. The EPR conditions and reaction mixtures were described under Fig. 1.

TABLE II
Comparison of mitochondrial functional parameters Mitochondria from rat liver (n ϭ 6) were prepared by differential centrifugation (differential centrifugation; Ref. 15) and purified by Percoll (16) 2 Fe II NO⅐ for different mitochondrial preparations Toluene-permeabilized mitochondria, mitochondrial homogenates, or crude fraction were incubated for the indicated times using 1 mM L-Arg and 1 mM or 5 mM (number between parentheses) of NMMA. The EPR signals were recorded using the instrumental settings described under Fig. 1 legend, and the signal intensities were expressed relative to no additions in toluene-treated mitochondria. The height intensities were evaluated from the signal peak at low magnetic field. arations to EPR silent species, probably of the type (MGD) 2 Fe II -NO ⅐ -X, where X ϭ halogen ions or NO 2 , as has been described for diethyldithiocarbamate (31,32). This notion is strengthened by the similar recovery (12%) of an EPR signal of a synthetic iron-nitrosyl complex (formed by incubating sodium nitroprusside, an NO ⅐ donor, and the spin trap) and the decrease in the signal intensities with protein concentrations above 12 mg. Measurement of NO ⅐ Production by the Oxidation of Oxymyoglobin-The production of NO ⅐ in the presence of L-Arg was measured in intact mitochondria, mitochondrial homogenates, and submitochondrial particles by following the NMMA-sensitive oxidation of oxymyoglobin to metmyoglobin (Fig. 3). The NMMA-insensitive rates of oxymyoglobin oxidation were 10 -20% of the total rate of metmyoglobin formation using mitochondria and mitochondrial homogenates, and 30% when using SMP. The higher unspecific oxidation in the latter instances may be attributed to a direct oxidation of oxymyoglobin by a component of the respiratory chain. 3 The rates of NO ⅐ production increased linearly with the protein concentration of mitochondrial preparations (Fig. 3). The specific rates were 1.36, 4.9, and 7.1 nmol ϫ (min⅐mg protein) Ϫ1 for intact mitochondria, mitochondrial homogenates, and submitochondrial particles, respectively. The activities obtained with mitochondrial homogenates and submitochondrial particles were higher than those obtained with intact mitochondria because the former were assessed under conditions for optimal NOS activity. A comparison of the rates of SMP and mitochondrial homogenates under identical conditions indicated that most of the activity was detected in the mitochondrial inner membrane (considering that 30% of the rat liver mitochondria protein corresponds to the inner membrane fraction (33), 25-35% of the particles had the "right side-in" conformation (34), and 80% of the activity (experimentally determined) was recovered after the sonication procedure) suggesting that NOS could be mainly (60 -80%) localized in this membrane fraction.
The production of NO ⅐ by intact mitochondria was followed by the NMMA-sensitive oxidation of oxymyoglobin in the presence of L-Arg (Fig. 4A). The rapid onset of the production was indicative of a fast transport of L-Arg into mitochondria, consistent with the reported L-Arg carriers found in isolated mitochondria from different tissues (35)(36)(37). The rate of NO ⅐ production by mitochondria yielded a classic hyperbolic response saturated with L-Arg concentrations above 20 M (Fig. 4A). The apparent K m for L-Arg was 5 M (1.1 nmol/mg of protein) and the V max was 0.38 M/min calculated from double-reciprocal plots (Fig. 4A, inset). These kinetic constants were consistent with those observed for permeabilized mitochondria (V max ϭ 5 nmol of NO ⅐ /min/mg of protein, and K m for L-Arg of 7 M; Fig.  4B), as well as with those reported for brain homogenates (as sources of nNOS) during the conversion of L-Arg to citrulline (K m ϭ 6 M for Arg and V max ϭ 0.15 mol/min/mg protein; Ref. 38). The K m for L-Arg was 20 -50 times lower than the matrix concentration of L-Arg (150 -300 M; Refs. 39 and 40), a value within the normal range of L-Arg in rat liver (20 -50 nmol/g wet weight; Ref. 25), when expressed per gram of liver wet weight (38 nmol/g wet weight). These concentrations indicate that mtNOS could be functionally active based on the requirements of L-Arg availability under normal conditions, and that if the kinetic properties of mtNOS in intact mitochondria are similar to those in vivo, it is unlikely that Arg levels play an important role in the regulation of mtNOS.
A slow rate of NO ⅐ production by intact mitochondria was also detected in the absence of added L-Arg (11 nM/min; Fig.  4A). This rate may reflect a slow but constant production of NO ⅐ by mitochondria using endogenous L-Arg, which may be increased under special pathophysiological conditions, such as those entailing an elevation of Ca 2ϩ .
The NO ⅐ production by intact mitochondria was not altered by D-Arg addition, was partially inhibited by NIO, and was totally inhibited by NMMA (Table IV). The lack of effect by the addition of D-Arg may be explained by the selectivity of the Arg transporter for the L-isomer, and the partial inhibition by NIO can be accounted for by a limited permeability of the inhibitor. Consistent with this possibility is a noted increase in the inhibition of of NO ⅐ production by the pre-incubation of mitochondria with NIO (Table IV).
In similar experiments, preincubation (10 -15 min) of SMP with 5 mM NMMA was found to inhibit the endogenous production of NO ⅐ , whereas the addition of L-Arg reversed such inhibition (Table V). DISCUSSION The following lines of evidence offer support for the mitochondrial generation of NO ⅐ .
First, the negligible contamination of the mitochondrial preparations with other subcellular fractions (Table I) and the integrity and functionality of these preparations (Table II) support the production of NO ⅐ by mitochondria.
Second, the production of NO ⅐ by mitochondria was demonstrated by two different spectroscopic assays: the formation of metmyoglobin, and spin trapping/EPR. To rule out species other than NO ⅐ reacting with oxymyoglobin (41)(42)(43)(44)(45), the rate of oxidation of oxymyoglobin was monitored under controlled conditions (with catalase and superoxide dismutase and measuring sensitivity to NMMA). However, because some NO ⅐ -derived oxides may still be able to produce this reaction, unequivocal identification of NO ⅐ was furnished by using the spin trap MGD, the iron-nitrosyl complex ESR signal of which is considered a "fingerprint" of NO ⅐ (21,22). The advantages of this technique (selectivity of the spin trap toward NO ⅐ and the lack of toxicity of the spin trap in biological systems) were initially limited by the free access of the spin trap to the biological source, the low recovery of the iron-nitrosyl adduct, and the presence of quenchers of NO ⅐ (e.g. hemoproteins, [Fe-S] clusters) that effectively compete with the spin trap MGD. These limitations were overcome by using millimolar concentrations of the spin trap (to effectively compete with other possible NO ⅐ quenchers) with toluene-treated mitochondria and mitochondrial homogenates (to allow free access of the spin trap to the biological source of NO ⅐ ). The use of submitochondrial particles and mitochondrial homogenates, albeit less physiologically relevant than intact mitochondria, devoid of the limitations noted above permitted the selection of optimal conditions for NO ⅐ production by mitochondria.
Third, the production of NO ⅐ is catalyzed by an enzyme, likely a NOS isoform, located at the mitochondrial inner membrane. This was inferred from three separate lines of evidence. (i) NO ⅐ production was modulated by NOS substrates (L-Arg) and inhibitors (NMMA, NIO, and D-Arg); (ii) the rate of NO ⅐ production by mitochondria and SMP versus L-Arg concentration followed a similar pattern to that described for NOS purified from different tissues; (iii) the higher specific activities in SMP or crude fraction (about 2 and 10 times higher, respectively) than those obtained with mitochondrial homogenates or permeabilized-mitochondria were indicative of an enzymic activity located at the inner membrane. Conclusive evidence that a NOS isoform was responsible for the NO ⅐ production was provided by the purification and characterization of the enzyme from purified rat liver mitochondria reported in the accompanying paper (46).
The rate of NO ⅐ production by rat liver mitochondria reported herein and by others 4 (47) and intact mitochondria (48). Given the role of NO ⅐ as cellular messenger, transmitter, and regulator (1-3), it could be hypothesized that this inhibition (or modulation) of mitochondrial respiration by NO ⅐ may represent a novel biochemical pathway regulating the supply of O 2 and energy to tissues under dynamic conditions. TABLE IV Inhibition of NO⅐ production by intact mitochondria Intact rat liver mitochondria (0.52 mg of protein/ml) were incubated in 1 ml of the reaction mixture containing 30 M L-arginine and 50 M oxymyoglobin. The suspensions were supplemented with 0.1 mM inhibitors, and the rate of NO⅐ production was quantified by measuring the oxidation of oxymyoglobin at 581-592 nm after 5 min of incubation. Mitochondria were preincubated with the inhibitors for 10 min, and then L-Arg was added.