Differential Sensitivity of Guanylyl Cyclase and Mitochondrial Respiration to Nitric Oxide Measured Using Clamped Concentrations*

Nitric oxide (NO) signal transduction may involve at least two targets: the guanylyl cyclase-coupled NO receptor (NOGCR), which catalyzes cGMP formation, and cytochrome c oxidase, which is responsible for mitochondrial O2 consumption and which is inhibited by NO in competition with O2. Current evidence indicates that the two targets may be similarly sensitive to NO, but quantitative comparison has been difficult because of an inability to administer NO in known, constant concentrations. We addressed this deficiency and found that purified NOGCR was about 100-fold more sensitive to NO than reported previously, 50% of maximal activity requiring only 4 nm NO. Conversely, at physiological O2concentrations (20–30 μm), mitochondrial respiration was 2–10-fold less sensitive to NO than estimated beforehand. The two concentration-response curves showed minimal overlap. Accordingly, an NO concentration maximally active on the NOGCR (20 nm) inhibited respiration only when the O2concentration was pathologically low (50% inhibition at 5 μm O2). Studies on brain slices under conditions of maximal stimulation of endogenous NO synthesis suggested that the local NO concentration did not rise above 4 nm. It is concluded that under physiological conditions, at least in brain, NO is constrained to target the NOGCR without inhibiting mitochondrial respiration.

Nitric oxide (NO) 1 is a diffusible biological messenger that subserves cell-to-cell signaling functions in most tissues. NO can also be cytotoxic and has been incriminated in many different pathologies, including atherosclerosis, septic shock, cancer, and neurodegenerative disorders (1). Although much has been learned about the mechanism of NO synthesis (2), the transduction pathways engaged by physiological NO signals to modify cell and tissue function remain to be clearly defined.
The established target is the guanylyl cyclase-coupled receptor, or NO GC R, 2 which exists in at least two different het-erodimeric isoforms (␣1␤1 and ␣2␤1). This is a metabotropic type of receptor equipped with a heme prosthetic group to which NO binds, triggering the formation of cGMP from GTP in the cyclase domain of the protein. Through this route, NO elicits many effects such as smooth muscle relaxation, inhibition of platelet aggregation, and synaptic plasticity (3,4). Knowledge of the NO concentrations that engage the NO GC R is important for understanding the receptor kinetics, for informing on the physiological NO concentrations likely to exist in tissues, and for developing realistic models of NO signaling. Currently, however, the information on this issue is incoherent. Studies on the purified ␣1␤1 receptor protein have suggested that the NO concentration giving half-maximal activation (the EC 50 ) is 250 nM (5). More recently, an EC 50 of 1.6 M has been obtained for the enzyme in an extract of rat aorta (6). The validity of this range appears to be supported by several studies that have used the NONOate, diethylamine/NO adduct (DEA/NO), which degrades to release NO with a half-life of 2.1 min (at 37°C). Typically the EC 50 of DEA/NO in standard assays of NO GC R activity is about 300 nM (7,8), and this has been assumed to approximate the potency of NO (8,9). In contrast to these estimates made on the purified protein or in cell-free extracts, the estimated EC 50 value for NO GC R activation by NO in intact cells from the brain ranges from 20 to 45 nM (10) down to 2 nM (11). Despite the variability on both sides, this comparison might suggest that, along with other functional differences, the NO GC R in intact cells possesses heightened sensitivity to NO (12).
Another putative target for NO is cytochrome c oxidase, which is the terminal component of the mitochondrial respiratory chain responsible for almost all cellular O 2 consumption. By competing with O 2 for binding to cytochrome c oxidase, NO is considered to regulate the rate of respiration and the tissue distribution of O 2 (13)(14)(15)(16). On the other hand, the ensuing inhibition of respiration, and thus, of ATP synthesis, could have pathological repercussions, particularly in tissues such as brain that rely almost entirely on oxidative phosphorylation to meet their energy requirements (17). An important consideration here is the NO concentration range that is active on cytochrome c oxidase relative to that which engages the NO GC R. Under physiological conditions (30 M O 2 ), the NO concentration required to achieve 50% inhibition of respiration (the IC 50 ) has been estimated to be 60 nM in brain synaptosomes (18) and 11 nM in isolated mitochondria (19). Hence, depending on which pairs of values are taken, concomitant activation of the NO GC R and inhibition of mitochondrial respi-* This work was supported by program grants from The Wellcome Trust and The Sir Jules Thorn Charitable Trust. 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.
‡ These authors contributed equally to this work. § Present address: Div. of Neurophysiology, National Institute for Medical Research, The Ridgeway, Mill Hill, London, NW7 1AA, UK.
ration would appear inevitable, or at least plausible.
Aside from the problems of variability in the published estimates of the apparent potency of NO toward both these targets, another pertinent issue is that, to determine the true potency of any agonist on its receptor, it is a prerequisite that the agonist be applied in known concentrations that are stable over the period of measurement. Because of the instability of NO in aerobic solutions, particularly in the hyperoxic solutions used in the laboratory, this has not been achieved for either target, raising further uncertainty about their true sensitivities. By devising a method for delivering known, constant NO concentrations, we address this deficiency and provide a quantitative comparison of the potency of NO for activating purified NO GC R and inhibiting mitochondrial respiration.

EXPERIMENTAL PROCEDURES
Constant NO concentrations were obtained by allowing a dynamic equilibrium to exist between NO release from a donor and NO inactivation by red blood cells. Red blood cells were prepared from whole rat blood as described (20) and kept on ice. This stock was diluted to a final concentration of around 2 million cells/ml in either: (a) buffer (pH 7.4) containing NaCl (130 mM), Tris/HCl (50 mM), MgCl 2 (3 mM), EGTA (0.1 mM), and 0.5% bovine serum albumin for NO GC R experiments or (b) a suspension of cells from the cerebellum (20 million cells/ml), prepared as described previously (10), for determination of the sensitivity of mitochondrial respiration to NO.
The NONOates diethylenetriamine/NO adduct (DETA/NO) and DEA/NO (both from Alexis, Nottingham, UK) were prepared in 10 mM NaOH and diluted 1:100 into the final incubation medium. The NO and O 2 concentrations were recorded with a 1-Hz sampling frequency at 37°C in a sealed, stirred vessel equipped with an electrochemical NO probe (ISO-NOP, World Precision Instruments, Stevenage, UK) and O 2 electrode (Rank Brothers, Cambridge, UK). The rate of O 2 consumption in the presence of fixed NO concentrations achieved following addition of DETA/NO was measured within the O 2 concentration range 20 -30 M. This rate was expressed as a fraction of the control rate measured beforehand within the range 30 -40 M O 2 . All measurements were adjusted for slight basal drift by deducting the gradient found on adding NaCN (1 mM) as described previously (18).
Purified NO GC R (soluble guanylyl cyclase; Alexis) was diluted in a buffer (pH 7.4) containing 10 mM Tris/HCl, 1 mM dithiothreitol, and 0.5% bovine serum albumin to give a stock concentration of 5 g/ml, which was stored on ice. For experiments with clamped NO concentrations, DETA/NO was added to red blood cells incubated at 37°C at time 0, substrate (1 mM GTP) at 45 s, and NO GC R (final concentration 0.05 g/ml) after 1 min. Addition of GTP or the receptor protein did not disturb the steady-state NO concentration, as measured by the electrochemical probe. Aliquots of the reaction mix were removed after various times and inactivated in boiling buffer (50 mM Tris, 4 mM EDTA, pH 7.4). For experiments with DEA/NO, NO GC R was added to a buffer containing: Tris/HCl (50 mM), MgCl 2 (3 mM), EGTA (0.1 mM), GTP (1 mM), dithiothreitol (3 mM), and 0.5% bovine serum albumin at 5 s before addition of DEA/NO. Aliquots were removed and inactivated in boiling buffer. The levels of cGMP in both cases were measured by radioimmunoassay.
For determining the activity of the NO GC R in brain tissue, 400 m thick slices of cerebellum from 8-day-old rats (prepared as in Ref. 21) were preincubated for ϳ2 h in an oxygenated (95% O 2 , 5% CO 2 ) artificial cerebrospinal fluid containing: NaCl (124 mM), KCl (3 mM), NaH 2 PO 4 (1.25 mM), MgSO 4 (1 mM), NaHCO 3 (26 mM), CaCl 2 (2 mM), and D-glucose (10 mM) at 37°C. The slices were exposed to DEA/NO in various concentrations or NMDA (in the presence of 100 M L-arginine, added 15 min beforehand), and 2 min later, slices were inactivated in boiling buffer (50 mM Tris, 4 mM EDTA, pH 7.4) and then sonicated. The protein content of the homogenate was measured by the bicinchoninic acid method with bovine serum albumin as standard and, following centrifugation, the cGMP content of the supernatant was determined by radioimmunoassay.

RESULTS
Delivering Clamped NO Concentrations-An apparatus for maintaining constant (clamped) NO concentrations has been designed (22), but it is expensive to construct and unsuited to most biological applications. We have found a cheap and simple solution to the problem based on the principle that a fixed NO concentration will be formed when a constant source of NO comes into equilibrium with a "sink" that consumes it. For a source, we used the diethylenetriamine/NO adduct DETA/NO, which decomposes slowly with first order kinetics (a half-life of 20 h at 37°C), thus releasing NO at an effectively constant rate for many hours. For a sink, we have exploited the ability of oxyhemoglobin (oxyHb) when located within red blood cells to consume NO at an appropriate first order rate, forming methemoglobin and nitrate (20). When DETA/NO (50 -500 M) was mixed with a suspension of red blood cells (2 million/ml), a fixed concentration of NO was attained within 1 min (Fig. 1a). The amplitude of the steady-state concentration varied faithfully with the rate of NO release so that a linear relationship was observed between the DETA/NO concentration and the steadystate NO concentration (Fig. 1b). At the upper extreme (750 M DETA/NO), the NO sink became exhausted after around 2 min, presumably due to oxyHb oxidation. Thus, the method is capa- ble of producing clamped NO concentrations covering the range 0 -75 nM and lasting at least several minutes. This simple reaction mixture was used to test NO GC R sensitivity, but in principle, introduction of an appropriate concentration of red blood cells in suspension into any biological system would enable NO clamping to be achieved.
Sensitivity of the NO GC R to NO-Addition of small volumes (10 l) of purified NO GC R to this pre-equilibrated mixture (1 ml) in the presence of substrate (GTP) and co-factor (Mg 2ϩ ) did not affect the NO concentration (not shown). Accordingly, the cGMP resulting from receptor activation accumulated linearly with time and with no visible delay, regardless of NO concentration (Fig. 1c). Furthermore, addition of cGMP showed that the product remained stable in the mixture (Fig. 1d). The full concentration-response curve (Fig. 2a) had a threshold of ϳ0.5 nM, and maximum activity was observed at ϳ20 nM. The curve was well described by the Hill equation with an EC 50 of 3.9 nM and a Hill constant of 2.1.
The high sensitivity of the purified NO GC R to NO found here contrasts with previous measurements variously reporting EC 50 values of 250 nM (5) or 1.6 M (6), possible reasons for which are discussed later (see "Discussion"). Also apparently conflicting with the present result are several other studies using the NONOate DEA/NO (a half-life of 2.1 min) whose EC 50 on the purified NO GC R following a 10-min incubation, about 300 nM (7,8), has been assumed to correspond to the EC 50 for NO (8,9). When we reproduced these conditions, the EC 50 for DEA/NO was indeed 300 nM (Fig. 2b). On measuring the profile of NO produced by 300 nM DEA/NO, however, it was found that the concentration peaked at 50 nM after 2 min and then fell below the probe detection limit after 5-6 min (Fig. 3a). Therefore, the EC 50 for DEA/NO could simply reflect NO being in active concentrations for about half of the measured time. In accordance with this possibility, when the activity was followed over the whole 10-min period, it varied as predicted should the true EC 50 for NO be 3.9 nM (Fig. 3a). Thus, cGMP accumulated at the maximal rate for the first 3-4 min, but then the rate progressively decreased as NO fell below 20 nM, falling to 0 after around 6 min, coincident with the disappearance of NO.
For this independent (but corroborating) estimate of the EC 50 for NO to be valid, the enzyme activity would have to respond in a dynamic manner as the NO concentration falls. If deactivation on removal of NO takes many minutes, as suggested by a study on the purified protein (Ref. 23, but see also Refs. 9 and 24), the potency of NO measured by charting the changes in NO GC R activity over time in response to DEA/NO (Fig. 3a) would be an overestimate. The deactivation rate was investigated by adding sufficient free oxyHb (10 M) to remove all free NO. Accumulation of cGMP ceased abruptly on the addition of oxyHb, consistent with deactivation occurring within the ϳ0.5 s of mixing time (Fig. 3b) and thereby validating this second estimate of the potency of NO.
Sensitivity of Mitochondrial Respiration to NO-The O 2 consumption of cells from rat brain (cerebellum) was measured in the presence of fixed NO concentrations and at an O 2 concentration of between 20 and 30 M, corresponding to the range typically found in the brain in vivo (17). As expected, concentration-dependent inhibition was observed, but the IC 50 value was about 120 nM (Fig. 4a,b), which is 2-10-fold higher than found in earlier work on brain synaptosomes and isolated mitochondria (18,19). As with the NO GC R, the Hill constant was 2, which is in agreement with findings using isolated mitochondria (19).
The question was then addressed from a different perspective: how low does the O 2 concentration need to be before an NO concentration that is maximally effective on the NO GC R (20 nM) exerts significant respiratory inhibition? For 50% inhibition by 20 nM NO, the O 2 concentration needed to fall to 5 M (Fig. 4c).
Estimating the Maximal Endogenously Produced NO Concentration-The foregoing results suggest that the NO concentration would need to be supramaximal for the NO GC R in order to inhibit cytochrome c oxidase significantly under normoxic conditions (Fig. 4d). It could be argued that the two proteins represent distinct high and low affinity targets, but for this hypothesis to be tenable, evidence is needed that the NO GC R can become saturated in response to endogenous NO synthesis. To examine this possibility, NO GC R activity was used as an intrinsic biosensor to provide a readout of the maximal tissue NO concentrations that can be generated endogenously.
In brain, neuronal NO synthase activity is coupled to activation of the NMDA type of glutamate receptor (4), and this pathway is particularly prominent in the developing cerebellum (21,25). exposed to increasing concentrations of exogenous NO, using DEA/NO as the donor. Because of the powerful NO inactivation pathway that is expressed in brain and other tissues, high concentrations of NO donors need to be applied to intact brain slices to supply the cells with active concentrations of free NO (11). At the peak of the response to DEA/NO (100 M; curve not shown), cGMP was 619 Ϯ 33 pmol/mg of protein (n ϭ 5), almost twice the level found with NMDA. The lower response found with NMDA was not because the agonist reduced the responsiveness of the NO GC R (e.g. through raising cytosolic Ca 2ϩ ; Ref. 26) because simultaneous exposure to NMDA failed to inhibit the response to DEA/NO (717 Ϯ 28 pmol/mg of protein; n ϭ 5). Because pharmacological activation of NMDA receptors in this way will lead to the switch-on of multiple NO sources throughout the slice, it follows that sources and sinks will rapidly come to equilibrium generating a steady-state "continuum" of NO throughout the tissue (27). Visible evidence for the validity of this assumption is provided by immunocytochemical data on cGMP accumulation in NMDA-stimulated cerebellar slices (28). Inspection of Fig. 2a suggests, therefore, that maximal endogenous NO synthesis raises the local NO concentration experienced by NO GC R in situ to about 4 nM. DISCUSSION By using clamped NO concentrations, we find that NO is about 100-fold more potent as an agonist for the purified NO GC R and 2-10-fold less potent as an inhibitor of mitochondrial respiration than reported previously, the net effect being that there is minimal overlap in the two concentration-response curves at physiological O 2 concentrations (Fig. 4d). This, together with evidence that the NO GC R is not saturated during maximal endogenous NO synthase activity and that even a saturating NO concentration for the NO GC R only affects respiration when the O 2 concentration is very low, signifies that, in the brain at least, the NO signaling pathway has evolved to target the NO GC R without simultaneously influencing mitochondrial function.
The explanation for the much higher previous estimates of Note that the range of NO concentrations that could be maintained in this way was larger than that shown in Fig. 1a because additional NO inactivation was provided by the cerebellar cells (11). b, concentration-response curve for inhibition of respiration by NO. The curve was obtained by binning the NO concentrations into regularly spaced, non-overlapping groups (each containing three to eight individual measurements). The degradation altogether. As shown here using DEA/NO, the EC 50 value derived in this way is misleading because it measures the donor concentration required to maintain NO in active concentrations for half the incubation period, not the potency of NO on its receptor. It follows that changes (or otherwise) in the EC 50 value for the donor cannot alone be taken as evidence for changes (or otherwise) in the sensitivity of the NO GC R. The recent conclusion, made on this basis, that association with membranes "sensitizes" the receptor to NO (29) may therefore be spurious as a lower rate of NO consumption (16,30) by membrane fractions as compared with cytosol could account for the differing apparent EC 50 values for DEA/NO (0.24 versus 0.48 M, respectively).
On the other hand, the EC 50 value measured here for the purified NO GC R (4 nM) is in good agreement with the value of 2 nM estimated for intact brain cells (11), especially considering the large extrapolation involved in arriving at this estimate. Higher EC 50 values (20 -45 nM) have also been reported for the same cells (10), but these relied on assumptions about the efficiency of NO release from a caged precursor, which must be regarded as questionable. Consequently, there are now no good grounds for proposing the existence of cellular factors that enhance the potency of NO on its receptor (12). Furthermore, in our hands, the rapid rate of deactivation of purified NO GC R (Յ0.5 s, Fig. 3b) is consistent with the subsecond deactivation rate measured in intact cells (10). Although the reasons for the much slower rates (from 5 s to 3 min) found for the purified NO GC R in other laboratories (9,23,24) and their variability remain uncertain, this finding dissipates the need to invoke the existence of factors in cells that enhance the rate of dissociation of NO from its receptor (9,10,12).
In addition to providing a measure of the potency of NO for the NO GC R, the shape of the concentration-response curve is also instructive. The current model of a single binding event to the heme prosthetic group (3) predicts a Hill slope of 1 (Fig. 2b, dashed line), whereas we found that the curve was steeper, with a Hill slope of 2. The most straightforward interpretation of a slope greater than 1 is that the receptor incorporates multiple agonist binding sites that act in a cooperative fashion. This possibility has been raised previously for the NO GC R based on analysis of activation kinetics (31), but we have recently argued that these data are equivocal and can be fitted by a simpler model (32). The shape of the equilibrium concentration-response curve of the NO GC R provides the first unambiguous evidence that a single NO binding event cannot underlie receptor activation. It follows that the simplest model must now incorporate two NO binding events, meaning that the NO GC R can exist in at least four states (unbound, a single NO-bound state, a double NO-bound state, and an active state) and that progressive activation of a population of receptors would proceed in three kinetic phases. Testing the predictions of this new model against existing equilibrium and kinetic data is clearly desirable.
In addition to NO GC R activation, there has been growing interest in the possibility that NO could function as a regulator of mitochondrial respiration, based in part on the assumption that at physiological O 2 concentrations, cytochrome c oxidase would be inhibited by NO concentrations similar to those required for NO GC R activation (13)(14)(15). Our experiments, however, showed that the two concentration-response curves were separated by a factor of 30 and that the IC 50 for respiratory inhibition by NO was 2-10-fold higher than previous estimates (18,19). The discrepancy probably relates to the former use of bolus-like additions of NO because when the brain cells used here were challenged with boluses of NO (at ϳ100 M O 2 ), the IC 50 measured during the subsequent decay of NO (120 nM; Ref. 11) matched that found for isolated mitochondria (also at ϳ100 M O 2 ) using a similar approach (19). Based on this result, it would have been predicted that the IC 50 in brain cells would have been about 11 nM at 30 M O 2 (19), whereas with clamped NO concentrations, it was 10-fold higher (120 nM). Presumably, inadvertent effects of the bolus itself and/or the use of ever-changing concentrations during the decay phase account for the different values. In any case, boluses of NO are unlikely to be found in vivo because the relatively slow switch-on of NO synthesis, which is dependent on Ca 2ϩ -calmodulin interactions or phosphorylation cascades (2), would be quickly balanced by NO inactivation (predicted to impose a half-life on NO of about 100 ms; Refs. 11 and 16), the result being a steady-state NO concentration analogous to that formed by the DETA/NO-red blood cell mixture (Figs. 1a and 4a) but much faster (a 100-ms time scale).
It was only with O 2 at very low concentrations that an NO concentration maximally active on the NO GC R (20 nM) could affect respiration. The O 2 concentration needed for 50% inhibition under these conditions (5 M) corresponds to one that, if sustained for more than a few seconds in brain tissue, is pathological (17). Indeed, initial signs of metabolic stress occur in brain at about 10 M O 2 (17), at which concentration 20 nM NO failed to affect respiration (Fig. 4c). Moreover, given the high K m value of the brain NO synthase for O 2 (350 M; Ref. 33), it is doubtful that 20 nM NO and low micromolar O 2 could coexist unless the pathway that inactivates NO (11) is compromised. However, although also O 2 -dependent (30), this pathway is as active at 30 M O 2 as it is at 200 M O 2 , indicating a much lower K m value than the NO synthase. 3 From these data, it is unlikely that NO concentrations in the range active on the NO GC R could simultaneously inhibit mitochondrial respiration in vivo. This raises the question of whether NO ever rises above this range physiologically to the extent that it could influence O 2 consumption. Our attempt to answer this by maximal stimulation of endogenous NO formation in brain slices suggested that this was not the case in that local NO rose only to an apparent concentration of 4 nM. This is consistent with studies in the brain in vivo where, even under conditions of abnormally elevated neuronal activity, cGMP rises to only a fraction of the maximum levels achievable (34). Furthermore, from a recent spectroscopic analysis of cytochrome c oxidase in rat brain in vivo, it has been concluded that endogenous NO does not inhibit this enzyme, either during physiological conditions or during reperfusion following transient ischemia (35). This finding accords with our data suggesting that NO rises only to about 1 nM in slices of rat striatum in vitro subjected to simulated ischemia and reperfusion (36). The NO-consuming pathway identified recently in brain (11) appears well suited to constrain NO concentrations to the low nanomolar range that selectively engages the NO GC R. Whether or not NO is similarly constrained in other tissues remains to be determined.
Finally, the NO concentrations active on the NO GC R have long been used as a guide to the physiological NO concentrations existing in tissues, and thus, to the exogenous concentrations that should be applied experimentally. Our results suggest a downward revision by about 2 orders of magnitude in this range. As a consequence, many findings made using NO concentrations previously considered physiological may be of more pathological relevance, although what constitutes a pathological NO concentration in vivo remains to be defined. 3 C. Griffiths and J. Garthwaite, unpublished observation.