Differences in three kinetic parameters underpin the unique catalytic profiles of nitric-oxide synthases I, II, and III.

We previously reported the existence of a special auto-regulation property of neuronal nitric-oxide synthase (NOS) based on NO near-geminate combination and partial trapping of neuronal NOS (nNOS) through a futile regenerating pathway. On this basis, we developed a kinetic simulation model that was proven to predict nNOS catalytic specificities and mutations effects (Santolini, J., Adak, S., Curran, C. M., and Stuehr, D. J. (2001) J. Biol. Chem. 276, 1233-1243; Adak, S., Santolini, J., Tikunova, S., Wang, Q., Johnson, J. D., and Stuehr, D. J. (2001) J. Biol. Chem. 276, 1244-1252). Here we show that the same model simulates and explains the distinct catalytic behaviors of inducible and endothelial NOS (iNOS and eNOS). Their marked differences were linked to variations in three basic parameters (rates of ferric heme reduction, ferric heme.NO dissociation, and ferrous heme.NO oxidation) that together control partitioning between futile and productive pathways and their relative rates. We also incorporated feedback inhibition into the kinetic model to account for potential rebinding of accumulated solution NO. The model accurately simulated the different relative impacts of both NOS.NO interactions (near-geminate combination of NO versus rebinding of solution NO) on catalytic behavior of each NOS isoform, including their speed and extent of heme.NO complex accumulation, K(m) for O(2), and propensity to transform NO into a higher oxide. Thus, individual catalytic behavior of any NOS can be understood through a single unified kinetic model. Because the model defines how different settings of individual kinetic parameters control regulation by two distinct NOS.NO interactions, it sheds light on mechanisms, structural features, and scope of NOS regulation and its physiologic impact.

whose subunits are comprised of a reductase domain containing FAD, FMN, and NADPH binding sites, and an oxygenase domain containing 6(R)-tetrahydrobiopterin (H 4 B), iron protoporphyrin IX (heme), and a binding site for the L-arginine (Arg) substrate (7,8). NO synthases catalyze two sequential mixedfunction oxidations, the first being Arg hydroxylation to form N -hydroxy-L-arginine (NOHA) as a bound intermediate (9 -11), and the second converting NOHA to citrulline and NO.
Three main NO synthases are expressed in mammals and differ in their functions, amino acid sequence, post-translational modification, and cellular location. Two NOS, neuronal NOS (nNOS or NOS-1) and endothelial NOS (eNOS or NOS-3), are constitutively expressed and involved in signal cascades. The third NOS is cytokine-inducible (iNOS or NOS-2) and functions as both a regulator and effector of the immune response. The counterpart to this diversity of location and function is a specific regulation of each isoform. For example, NO synthases differ significantly regarding Ca 2ϩ levels required to bind calmodulin (CaM), which triggers heme reduction and NO synthesis (12,13). They also have different capacities to be upor down-regulated by serine/threonine phosphorylation (14,15). A third difference involves regulation via heme⅐NO complex formation. In nNOS, we have established that heme⅐NO complex formation is an intrinsic feature that governs the rate of NO synthesis and shifts the enzyme apparent K m O 2 to a higher value (16,17). The model shown in Scheme 1 below can explain these effects (18). It incorporates the observation that ferric heme binds newly formed NO before it can leave the enzyme (19,20). This causes nNOS to partition between futile and productive cycles during catalysis (see Scheme 1), with only the productive cycle liberating NO (18). Key kinetic parameters for nNOS have been measured, including rates of heme reduction and NO dissociation, k cat , and oxidation of the ferrous heme⅐NO complex (19,21,22). These values are set such that a majority of nNOS exists as the ferrous heme⅐NO complex during steady-state NO synthesis (23,24). Computer simulation of the kinetic model in Scheme 1 reproduces this result and can accurately simulate the pre-steady-state and steady-state behaviors of nNOS mutants that display greater or diminished activity relative to wild-type enzyme (18,25).
The available data for iNOS and eNOS suggest remarkable differences in their NO modulation. For example, iNOS displays a more gradual time course and smaller extent of NO inhibition compared with nNOS, its heme⅐NO complex is predominantly ferric instead of ferrous, and NO scavengers increase its activity (26 -29). Indeed, much inhibition associated with its heme⅐NO complex formation, as well as complex buildup itself, is directly related to the concentration of NO that is achieved in the reaction (29). On the other hand, Masters and colleagues (30) showed that a minor amount of ferrous heme⅐NO complex forms in iNOS during the first seconds after initiating NO synthesis, suggesting that its regulation also involves the heme binding newly generated NO. Regarding eNOS, very little heme⅐NO complex forms during normal turnover and is associated with no detectable inhibition of catalysis (31). However, when NOHA was used in place of Arg as substrate to create a faster rate of NO synthesis, there was greater heme⅐NO complex formation and measurable catalytic inhibition, which was linked to NO accumulation in the medium. In addition, Adak et al. (32) recently showed that the eNOS heme forms a ferrous heme⅐NO complex after initiating catalysis if it is part of a chimeric NOS protein that contains the nNOS reductase domain, which supports more rapid heme reduction. Although our kinetic model was built to understand the particular catalytic features of nNOS, we presumed that it might also be useful in describing catalytic behaviors of other NOS isozymes. In this report, we directly tested this concept by measuring the missing kinetic parameters for iNOS and eNOS and then running simulations of the kinetic model. We also expanded our kinetic model to examine the impact of the external NO concentration, because this has an active role in regulating iNOS and eNOS activity. The results show that the kinetic model can accurately simulate or predict catalytic behaviors of eNOS and iNOS in the absence or presence of NO accumulation in solution. This provides a better understanding of how both types of heme⅐NO binding control NOS activity. It also shows how the kinetic parameters are set differently among NOS, and how, by setting them differently, a given NOS can take distinct advantage and emphasize control by one type of NO binding versus the other, or can minimize effects of NO all together. This provides a context to understand NOS isozyme differentiation and develops better links to their physiologic roles.

EXPERIMENTAL PROCEDURES
Materials-All reagents and materials were obtained from Sigma Chemical Co. or sources as previously reported (17,21,33).
Protein Expression and Purification-Wild-type iNOS (full-length and oxygenase domains) and eNOS (full-length) proteins were expressed in Escherichia coli and purified as previously described (31,(33)(34)(35). UV-visible spectra were recorded on an Hitachi U3110 spectrophotometer in the absence or presence of 20 M H 4 B and 1 mM Arg. Enzyme concentration was quantified using the absorption of the ferrous-CO adduct at 444 nm and an extinction coefficient of 74 mM Ϫ1 cm Ϫ1 (A 444 to A 500 ).
nNOS Activity Assays-NADPH oxidation and NO formation at steady state were measured at room temperature by spectroscopic assays as previously described (13).
Ferrous Heme⅐NO Dissociation Rate-Anaerobic solutions of ferrous, NO-bound iNOSoxy (20 M) or eNOS (10 M) were prepared in 40 mM EPPS buffer, pH 7.6, containing 2.5 mM Arg, 20 M H 4 B, and 1 mM DTT by adding successively 1.5 mM sodium dithionite and 200 M NO. Then a CO-saturated buffer (900 l), containing 40 mM EPPS, pH 7.6, 2.5 mM Arg, 20 M H 4 B, 1 mM DTT, and 1.5 mM dithionite, was maintained under positive CO pressure. The dissociation was triggered by adding 100 l of the ferrous heme⅐NO solutions to the latter buffer. The kinetics of NO release at 10°C was followed as the trapping of ferrous enzyme by CO, leading to buildup of 444-nm absorbance. Three different assays were achieved for each enzyme. The absorbance change at 444 nm was fitted to a monoexponential function using Origin software.
Ferric Heme⅐NO Dissociation Rate-Experiments were run according to methodology described by Scheele et al. (36), using oxyhemoglobin as a trapping agent for NO release (37). Anaerobic solutions of 2 M iNOSoxy or eNOS were prepared in 40 mM EPPS, pH 7.6, buffer containing 2.5 mM Arg, 1 mM DTT, 50 M H 4 B, and 0.1 mM EDTA. An NO-saturated solution was added to give 15 M NO, leading to formation of the ferric heme⅐NO complex. This solution was mixed rapidly with an aerobic solution containing 25 M oxyhemoglobin in the same buffer at 10°C. Rapid mixing was achieved with a Hi-Tech SF-61 stopped-flow instrument equipped with a Hi-Tech diode array detector. Characteristic isosbestic points were observed for conversion of oxyhemoglobin to methemoglobin, and absorbance changes at 401 and 415 were followed versus time. Three sets of ten scans were achieved with three distinct iNOSoxy and eNOS preparations. The kinetics of absorbance changes of each set was multifitted (rate parameters shared by all the kinetics) using the multifit facilities of Origin software. Biphasic kinetics were observed, an extremely fast phase (over 100 s Ϫ1 ) and a slower second phase (between 2 and 4 s Ϫ1 ). To differentiate reactivity of free NO versus NO that dissociates from the ferric NOS heme, we performed the same experiments at three different initial concentrations of NO (5, 10, and 15 M).
Ferrous Heme⅐NO Oxidation Rate-A solution of 8 M ferric iNOSoxy or 2 M ferric eNOS in a 40 mM EPPS, pH 7.6, buffer, containing 2.5 mM Arg, 20 M H 4 B, 1 mM DTT, and 0.5 mM EDTA was made anaerobic in a sealed cuvette and reduced by adding small amount of anaerobic sodium dithionite solution at 10°C. To avoid excess of dithionite, titration of the ferric enzyme was stopped as soon as the heme became reduced as judged by a shift in the Soret absorbance and the disappearance of absorbance at 650 nm. 200 M NO was subsequently added from a NO-saturated solution. The ferrous heme⅐NO enzymes were mixed rapidly in the stopped-flow instrument with an air-saturated buffer containing 40 mM EPPS, pH 7.6, 2.5 mM Arg, 20 M H 4 B, 1 mM DTT, and 0.5 mM EDTA. The kinetics of ferrous heme⅐NO oxidation at 10°C was followed at 436 nm (ferrous heme⅐NO decay) and at 395 nm (ferric heme buildup). Two sets of 10 traces were achieved for each enzyme. Characteristic isosbestic points were observed and indicated a singlestep process. Traces at both wavelengths were multifitted for each set as described above.
eNOS Ferric Heme Reduction Rate-An anaerobic solution containing 40 mM EPPS, pH 7.6, 20 M H 4 B, 450 M DTT, 2.5 mM Arg, 0.5 mM EDTA, and 100 M NADPH was made CO-saturated by bubbling. Ferric eNOS and CaM were made anaerobic in a sealed cuvette by alternate cycles of vacuum and nitrogen refilling. The cuvette was put under positive CO pressure and filled up to 1 ml with the CO-saturated buffer. Final concentrations of eNOS and CaM were 2 and 4 M, respectively. A solution of 200 mM Ca 2ϩ was made anaerobic by nitrogen bubbling. Reduction of ferric eNOS heme was triggered by adding 25 l of this solution and was monitored at 10°C by measuring the formation of the ferrous heme-CO complex at 445 nm. Experiments were done three times, and curves were independently fitted to a monoexponential function using Origin.
Ferric Heme⅐NO Reduction Rate-100 M NO from a NO-saturated buffer was added to an anaerobic solution of 3.5 M iNOS in 40 mM EPPS, pH 7.6, 2.5 mM Arg, 20 M H 4 B, and 450 M DTT. This enzyme solution was mixed rapidly in the stopped-flow instrument at 10°C with an anaerobic solution of 40 mM EPPS, pH 7.6, containing 2.5 mM Arg, 20 M H 4 B, 450 M DTT, and 200 M NADPH. Addition of NADPH triggered reduction of the ferric heme⅐NO species. Because spectral changes around the Soret peak are weak (a transition from 440 nm to 436 nm from ferric to ferrous heme⅐NO species) and partly masked by flavin reduction, the absorbance at 570 nm was monitored to follow heme reduction. Three sets of 10 traces were obtained. Kinetic traces in each set were multifitted as described above. Reduction of the eNOS ferric heme⅐NO complex was triggered by adding 25 l of an anaerobic 200 mM Ca 2ϩ solution into 1 ml of the ferric heme⅐NO solution with conventional mixing at 10°C. Absorbance change versus time was followed at 570 nm in an Hitachi spectrophotometer at 570 nm. Three different traces were obtained and were fitted to a monoexponential function using Origin.
iNOS and eNOS Pre-steady-state and Steady-state Stopped-flow Spectroscopy-Reactions were performed as described before using a Hi-Tech SF-61 stopped-flow instrument to record spectral transitions (18). One solution contained 40 mM EPPS, pH 7.6, 1 M eNOS, 2 M CaM, 2.5 mM Arg, 40 M H 4 B, 1 mM DTT, 0.5 mM EDTA, and 200 M NADPH. The enzyme reaction was initiated at 10°C by rapidly mixing this solution with an identical solution that contained 5 mM Ca 2ϩ but no enzyme. For iNOS reactions, the final enzyme concentration ranged between 2 and 4 M. The iNOS reaction was triggered by rapidly mixing the enzyme solution with an identical solution containing NADPH but no enzyme at 10°C. For eNOS, 96 spectra were recorded over 144 s. For iNOS, the collection times were over 7.2 and 14.4 s. Kinetics of absorbance change at 340, 395, 485, 440, and 580 nm were analyzed. Three sets of 10 mixing experiments were done for each enzyme. Each set was multifitted to derive apparent rates as described above.
Model Simulations-Three different models were used to run the simulations. Two of these (a reaction without substrate and a reaction without NO accumulation) were previously described in detail by Santolini et al. (18). The same set of equations and the same procedures were used here. The iteration step was typically 1 ms, and up to 100,000 points were calculated. The iteration step was increased up to 10 ms when an overall increase of the simulation time was desired (i.e. up to 16-min reactions). Percentages of each intermediate were determined at steady state. Initial rates were determined during the first 100 ms. Steady-state rates were determined once the steady-state phase was reached. The usual time range for this calculation was between 40 and 50 s; in the case of the simulated K m O 2 determination, for low oxygen concentrations, this time ranged was delayed up to 5 min. Experimentally observed oxidation rates for eNOS and iNOSoxy ferrous heme⅐NO complexes, which we derived at half air-saturated condition, were converted to bimolecular rate constants proportional to O 2 concentration for use in the simulations. For simulation of aerobic NO synthesis reactions, the concentration of O 2 was set at 280 M. NADPH and O 2 concentrations were kept constant during the entire simulation, to avoid secondary effects due to their exhausting. Many rate constants used for iNOS and eNOS were derived here experimentally as described in main text. The enzyme concentration used for iNOS and eNOS steady-state activity simulation was set to 1 M. The enzyme concentration for the K m O 2 of iNOS in the absence of substrate was set to 0.1 M.
A third model was designed to take solution NO rebinding into account. For that purpose we added two sets of rate equations corresponding to solution NO binding to the ferric or ferrous heme. Rate equations of ferric or ferrous heme were consequently modified. We used experimentally derived bimolecular rate constants for NO binding at 10°C as detailed in the main text. The simulation of this model was run with the same procedure as that of the two previous ones noted above. For short term experiments (within a minute) the concentrations of eNOS and iNOS were set to 1 M. For long term experiments, the concentrations of eNOS and iNOS were, respectively, set to 0.25 and 0.1 M, to match actual experimental conditions. The same concentrations were used to simulate how NO rebinding affects iNOS and eNOS K m O 2 .
A fourth model was built to simulate the activity of eNOS when NOHA is used as a substrate. It is derived from the one initially described (18) by simplifying the main forward pathway to its NOHA portion. Otherwise, the simulation of this model was the same as for the other ones. The rate constants are detailed in the main text. The concentration of eNOS used for the simulation was 0.25 M.

RESULTS
Kinetic Characterization of NO Interactions with iNOS and eNOS-Simulation of the kinetic model for NOS catalysis (Scheme 1 (18)) requires knowledge of a number of kinetic parameters. Although these were already determined for nNOS (19,(22)(23)(24)36), several were lacking for iNOS and eNOS. We therefore measured rates of NO dissociation from the heme, the rate of ferrous heme⅐NO oxidation, and the rate of ferric heme⅐NO reduction in iNOS and eNOS. Protocols we used to measure NO dissociation from the ferric or ferrous heme were adapted from Scheele et al. (36). The ferrous heme⅐NO complex was generated under anaerobic conditions and then diluted 10-fold into a sealed cuvette containing a CO-saturated buffer plus dithionite at 10°C. The excess dithionite quickly scavenges released NO while CO binds to the free ferrous heme to generate a complex absorbing at 445 nm. Because CO binding to ferrous NOS is fairly rapid (36,23), the rate of NO dissociation is equivalent to the rate of absorbance increase at 445 nm. Representative traces for dissociation of ferrous heme⅐NO iNOS and eNOS are shown in Fig. 1 (A and  B). Curves from three experiments each were fit to a monoexponential function and gave rise to k off values of 1.35 Ϯ 0.2 ϫ 10 Ϫ4 s Ϫ1 for iNOS and 6.3 Ϯ 1.0 ϫ 10 Ϫ4 s Ϫ1 for eNOS. These values are close to the k off determined for ferrous heme⅐NO nNOS (36) and confirm that NO dissociation is extremely slow from all three ferrous NOS isoforms.
We also determined the dissociation rate constant for the iNOS and eNOS ferric⅐NO complex at 10°C. Due to a complex binding process of NO to the heme (23,38), it appears that this value cannot be accurately derived by graphic analysis of apparent association rate constants. Scheele et al. (36) suggested a method to follow release of NO by its reaction with oxyhemoglobin to form methemoglobin. We performed stopped-flow experiments in which an anaerobic solution of ferric heme⅐NO iNOS or eNOS was mixed with an air-saturated oxyhemoglobin solution, with subsequent buildup of methemoglobin followed conjointly at 401 and 415 nm. Because excess NO in solution will also react with oxyhemoglobin, we also examined the validity of the method by performing our experiments at three different NO concentrations. This allowed us to discriminate two phases. A fast phase (rate over 100 s Ϫ1 ), whose amplitude and reaction rate depended on the initial concentration of added NO, corresponded to reaction of free NO with oxyhemoglobin. The slower phase gave the same reaction rate at all NO concentrations used, whereas its amplitude varied proportionally to the NO concentration. This indicates the slow phase represents dissociation of the NOS ferric heme⅐NO complex, whose initial concentration depended on the initial concentration of NO (see "Experimental Procedures").
Superimposition of 96 recorded spectra showed characteristic isosbestic points for oxyhemoglobin conversion to methemoglobin (data not shown). Three sets of 10 different kinetic traces were recorded. The slow phase of each set was multifitted to a monoexponential function using a common rate constant. Fig. 1C and D show representative kinetic traces at 415 nm for iNOS and eNOS, respectively. Ferric heme⅐NO dissociation rate constants we obtained were 2.0 Ϯ 0.4 s Ϫ1 for iNOS and 3.5 ϩ 0.5 s Ϫ1 for eNOS. These are a bit slower than values reported for ferric heme⅐NO dissociation in nNOS (around 5 s Ϫ1 ; 19, 36).
Another key-parameter in the kinetic model is the rate of ferric heme⅐NO reduction. To determine this, an anaerobic solution of ferric heme⅐NO iNOS or eNOS was prepared with H 4 B and Arg. Heme reduction in iNOS was triggered by rapidly mixing with anaerobic NADPH solution, whereas in eNOS it was triggered by mixing NADPH-reduced enzyme with anaerobic Ca 2ϩ solution to cause CaM binding. Because concurrent flavin reduction occurs in the iNOS reaction, spectral changes related to heme Soret absorbance could not be clearly discerned. We therefore monitored spectral change of the visible bands in that case. Spectra of iNOS and eNOS ferric and ferrous heme⅐NO complexes are shown in the insets of Fig. 2 (A  and B), respectively, whereas representative kinetic traces are shown in the same two main panels. Each set of traces was multifitted to a monoexponential function, and the data converge to values of 1.0 Ϯ 0.2 s Ϫ1 and 2.8 Ϯ 0.4 ϫ 10 Ϫ3 for reduction of the ferric⅐NO complexes of iNOS and eNOS at 10°C, respectively. Because ligand-free ferric iNOS and eNOS have similar heme reduction rates (39, 31), we conclude that SCHEME 1 bound NO does not influence their heme reduction kinetics, as is also true for nNOS (18).
The final missing kinetic parameter was the oxidation rate of the iNOS and eNOS ferrous heme⅐NO complex. We determined these values using a method as described by Adak et al. (22) for wild-type and mutant nNOS. Anaerobic ferrous heme⅐NO iNOS or eNOS was rapidly mixed with an air-saturated buffer at 10°C, and their oxidation to ferric enzyme was followed as absorbance decay at 436 nm or absorbance gain at 395 nm. A single-step oxidation process was indicated by isosbestic points in the overlaid spectral traces as shown in Fig. 2C for iNOS. Absorbance change at 436 or 395 nm followed identical exponential behavior in both enzymes. Representative kinetic traces at 395 and 436 nm for iNOS are shown in Fig. 2D. Three sets of ten kinetic traces were multifitted separately. The data converged to oxidation rates of 3.0 Ϯ 0.2 s Ϫ1 and 0.63 Ϯ 0.05 s Ϫ1 for iNOS and eNOS ferrous heme⅐NO complexes, respectively, at the half air-saturated condition (O 2 concentration ϳ140 M). These rates differ by almost 5-fold, and also differ from the rate of nNOS ferrous heme⅐NO oxidation that was determined under identical conditions (0.19 s Ϫ1 ) (22).
Modeling iNOS and eNOS Catalysis-For simplicity, we first focused only on the influence of near-geminate NO binding to the iNOS or eNOS ferric heme, which occurs immediately at the end of each catalytic cycle. Although NO that accumulates in solution during multiple turnover can also impact iNOS or eNOS catalysis (26 -29), we consider this influence at a later point. So, our basic model describing their pre-steady-state and steady-state behaviors was first built in the absence of NO accumulation, as we did for nNOS (18).
We determined by stopped-flow the patterns of pre-steadystate and steady-state behavior, in experiments run at 10°C, as previously described (18). NO synthesis was triggered by rapidly mixing an air-saturated solution of iNOS (or CaMbound eNOS) with the same buffer containing 200 M NADPH (or 5 mM Ca 2ϩ ). Spectral changes were determined from the rapid-scanning data. Fig. 3 shows the absorption change at three wavelengths as a function of time for the iNOS reaction. NADPH consumption was followed at 340 nm, whereas heme⅐NO complex buildup was followed at 436 and 575 nm. Three sets of 10 experiments were analyzed and converged to the same results. As indicated by the absorption traces at 340 nm, the rate of NADPH consumption by iNOS continuously decreases with time. In the same time, absorption at 436 or 575 nm also increases slowly after an initial rapid buildup. Rates of NADPH consumption during the initial phase (0 -0.5 s) and later phase (4 -10 s) were estimated by fitting to linear functions. Heme⅐NO buildup was fit to a two-exponential function. The amplitude of the initial exponential phase corresponded to a minority of the total iNOS (25%), and its rate was similar to that previously observed by Abu-Soud et al. (29). Our data are consistent with heme⅐NO complex buildup and inhibition of NADPH consumption occurring by two processes in iNOS (28 -30): (a) an immediate process that involves near-geminate binding of newly formed NO to the ferric heme and then subsequent reduction to the ferrous heme⅐NO complex and (b) a subsequent process that involves ferric heme binding NO that accumulates in solution.
Some of the kinetic parameters required to model iNOS catalysis were determined here. Otherwise, a k cat of 36 s Ϫ1 was determined at 10°C using NOHA as a substrate in singleturnover experiments. 2 The rate of iNOS ferric heme reduction (0.9 s Ϫ1 ) was from Presta et al. (39)  binding to ferrous iNOS was from Abu-Soud et al. (31). Computer simulations were run as described previously for nNOS (18; see "Experimental Procedures"). Fig. 4 (A and B) shows simulated results for 1 M iNOS in the first seconds after initiating NO synthesis. Simulated rates of NADPH consumption, NO and citrulline production, and the ratios between them are listed in Table I, along with the rate of heme⅐NO complex buildup. As explained previously (18), because of the mathematical procedure used in the simulation and kinetic uncertainties in the catalytic pathway, a precise simulation of the first milliseconds of the reaction remains unattainable, but is not necessary for the topics we pursue. So we fit the simulated data within the first 3 s to a monoexponential function as we did when modeling nNOS catalysis (18). The pre-steadystate and steady-state characteristics of iNOS catalytic activity obtained by our simulation match actual kinetic data obtained here or in previous reports (41,42). The rate of ferrous heme⅐NO complex buildup is the same (Table I). Consistent with actual measurements, the simulation has only 15% of total iNOS accumulating as a heme⅐NO complex in the steady state if solution NO does not accumulate. This explains the minor change between initial and steady-state activities in the simulation and the weak deflection in the simulated NADPH consumption rate (Fig. 4, Table I). The steady-state citrulline to NO ratio predicted by the simulation is greater than the value determined for nNOS, but matches values reported in the literature for iNOS (41,42). The simulation also confirms that iNOS has a slightly greater steady-state activity than nNOS, but its NO synthesis is more uncoupled to NADPH consumption (41).
We next used our kinetic model to simulate eNOS catalysis using kinetic parameters for eNOS that were either determined here or reported elsewhere. Abu-Soud et al. (24) achieved the kinetic characterization of O 2 binding to eNOS. The k cat was determined in single-turnover studies to be around 3 s Ϫ1 at 10°C. 2 Some uncertainty remained concerning the rate of heme reduction. Abu-Soud reported a value of 5 ϫ 10 Ϫ3 s Ϫ1 at 10°C under anaerobic conditions in the absence of CO (24). We checked this value by monitoring heme reduction using formation of the ferrous heme⅐CO complex as a marker. We obtained precisely the same rate (data not shown). Because the catalytic rate of eNOS is about 20-fold higher in aerobic reactions, this rate of heme reduction cannot reflect the actual rate under aerobic conditions. Although it is difficult to understand how the presence of oxygen could affect the rate of heme reduction, some studies have addressed this point in cytochromes P450 (43). Its ability to do so in eNOS is confirmed by the different heme reduction rates found by Miller et al. (44) and Abu-Soud et al. (31): Excepting different temperatures (23°C versus 10°C) their protocols were the same except that Miller performed experiments in air-saturated buffer and reported a heme reduction rate that was 400-fold higher than the one found by Abu-Soud et al. under anaerobic conditions (31,44). Because a direct measurement of ferric heme reduction is impossible under aerobic conditions (oxygen binding to the ferrous heme is too fast to distinguish a primary heme reduc- tion phase), an alternative method is to use the initial rate of NADPH consumption. From rapid mixing experiments in which CaM binding was used to activate heme reduction in eNOS that was pre-reduced with NADPH, we found that the initial rate of NADPH consumption is close to 0.2 s Ϫ1 . Because this measurement may not be exact, we ran simulations using heme reduction rates between 0.1 and 0.25 s Ϫ1 . All simulations mimicked eNOS catalytic behavior that was experimentally observed and gave rise to an overall NADPH consumption rate close to our experimentally determined value (Table II); the best fit was obtained for a heme reduction rate of 0.17 s Ϫ1 . This is 20-fold less than the heme reduction rate of nNOS, which is exactly the ratio that Miller et al. reported (44). Our simulation predicts minimal ferrous heme⅐NO complex buildup in eNOS during NO synthesis, about 2-3% of total enzyme, which matches experimental data suggesting between 5 and 10% (31,44). Therefore, in the simulation there is no inflection between the initial and steady-state rates of NADPH oxidation. Also, simulated NADPH/NO and NADPH/citrulline ratios are identical and match experimentally determined values (45,46). Thus, the kinetic model we developed for nNOS (18) also applies for eNOS and shows how the discrepancy between their catalytic behavior is related to their different kinetic parameters.
Modeling Effects of Solution NO-Our kinetic experiments, along with results of Griscavage et al. (26), Hurshman and Marletta (28), and Abu-Soud et al. (29), indicate that a second slower phase of heme⅐NO complex formation is possible in iNOS and eNOS that may be linked to NO accumulation in solution. We therefore modified our kinetic model to account for binding of solution NO to both ferrous and ferric heme in these enzymes. Rate constants for NO binding to iNOS were taken from Abu-Soud et al., who reported NO k on values for ferric or ferrous heme of 9.9 ϫ 10 5 and 6.7 ϫ 10 5 M Ϫ1 s Ϫ1 , respectively   (29), who varied NO buildup in solution using a superoxide generating system in his iNOS reactions. Fig. 5A shows the accumulation of solution NO as a function of time and at different rates of NO disappearance in our simulated iNOS reactions. The concentration of NO reaches a steady-state value after about 1 min in each simulation. This agrees with electrode measurements by Abu-Soud et al. (29) that showed the NO concentration reached a plateau after 1 min during iNOS catalysis and was maintained as long as sufficient NADPH remained. The steady-state NO concentration diminished in our simulations as the rate of NO decomposition was increased. This phenomenon was also observed experimentally by Abu-Soud et al. (29) when monitoring iNOS catalysis at different rates of superoxide production. We also performed kinetic simulations to look at the correlation that may exist between the extent of NO accumulation and the second slower phase of heme⅐NO complex buildup. For each pattern of NO accumulation in Fig. 5A we determined iNOS heme⅐NO content as a function of time (Fig. 5B). The simulation shows the clear presence of two different phases. The first phase involves formation of a ferrous heme⅐NO complex that occurs in the first seconds and whose magnitude does not depend on the rate or level of NO accumulation in solution. This initial phase corresponds to near-geminate combination of NO with the iNOS ferric heme followed by its reduction to a ferrous heme⅐NO complex (see Scheme 1). The second phase of heme⅐NO complex buildup, which occurs over the subsequent minutes, involves formation of a ferric heme⅐NO complex whose magnitude depends on the extent of NO accumulation. This reflects experimental results of Hurshman and Marletta (28) and matches the data of Abu-Soud et al. (29), who experimentally varied NO buildup during iNOS reactions. Thus, the simulation confirms a direct and proportional correlation between accumulation of solution NO and the extent of ferric heme⅐NO complex buildup in iNOS. The kinetic simulation model can also check relationships between NO accumulation and iNOS catalytic activities. In Fig. 5C, we show the NO concentration versus time in a simulated iNOS reaction with no NO disappearance rate included, and how this affects the rates of NADPH consumption or NO and citrulline accumulation. These rates all decrease as NO accumulates in the medium. This is consistent with accumulated NO causing the second phase of heme⅐NO complex formation in Fig. 5A. Importantly, the simulation shows that the rate of NO accumulation falls to near zero once micromolar NO concentrations are achieved in the reaction, although iNOS continues to oxidize NADPH and generate NO and citrulline. This is because an equilibrium is reached between NO production and NO rebinding to cycle through the futile catalytic pathway (see Scheme 1). Thus, the simulation can reproduce experimental results peculiar to iNOS that show the NO concentration in a reaction reaches a plateau despite continued release of citrulline and NO (29).  The concentration values noted in the panel are solution NO concentrations achieved at steady state. C, effect of time-dependent NO accumulation on iNOS steady-state activities. The rates of NADPH consumption and citrulline and NO production were determined through different simulations that utilized different time ranges to determine precisely each rate.
We performed an identical simulation of eNOS catalysis. Rate constants for NO binding were taken from Abu-Soud et al. (31), who reported k on values for ferric and ferrous heme of 8.2 ϫ 10 5 and 11 ϫ 10 5 M Ϫ1 s Ϫ1 , respectively. During the first minute the rates of NADPH consumption or citrulline and NO production were unaffected by the accumulation of NO. As shown in Fig. 6A, only 2-3% of eNOS formed a heme⅐NO complex in the first seconds, whereas after 1 min only 10% of the enzyme was complexed with NO. This matches our experimental results and those of Abu-Soud et al. (31), which showed no detectable variation in eNOS activity or heme⅐NO complex accumulation in the first minute, and is consistent with a low NO concentration achieved within this timeframe.
We ran the same simulations for a longer period (see "Experimental Procedures") to increase the concentration of solution NO. Fig. 6B shows a progressive accumulation of NO during the simulated reaction. Unlike the iNOS reaction, the NO concentration did not reach a plateau, even after 15 min. NO accumulation corresponded to heme⅐NO complex formation in eNOS. This reproduces results of Abu-Soud et al. (31), who observed that eNOS will form a detectable ferric heme⅐NO complex when sufficient solution NO concentration is achieved. Simulated rates of NADPH consumption or NO and citrulline production all decreased as the proportion of heme⅐NO complex increased (Fig. 6C). This mimics experimental results of Abu-Soud et al. (31) that showed a direct correlation between solution NO accumulation, proportion of heme⅐NO complex, and decrease in eNOS catalytic rates.
Oxygen Dependence as a Function of NO Accumulation-One characteristic that differentiates NOS isoforms is their apparent oxygen K m (K m O 2 ), which varies widely among the isoforms (from 2 M for eNOS to ϳ400 M for nNOS). Moreover, for iNOS and eNOS the apparent K m O 2 varies in the presence or absence of NO accumulation (29,31). We previously demonstrated that our kinetic model can simulate the 80-fold shift in K m O 2 observed for nNOS in the absence or presence of NO synthesis (18,16). We therefore checked if our kinetic simulation model could reproduce changes in apparent K m O 2 for iNOS and eNOS under different reaction conditions. Individual simulations were run at different O 2 concentrations, and the oxidation rate of the ferrous heme⅐NO complex at each O 2 concentration was set as a bimolecular rate constant proportional to the rate we measured under half air-saturated condition. Rates of NADPH consumption were calculated once the steady state was reached under each condition. Fig. 7A shows the simulated NADPH consumption rate for iNOS as a function of O 2 concentration under three experimental conditions. Curves were fitted to a single-site binding model to derive an apparent K m O 2 . In the absence of NO synthesis (no substrate), the apparent K m O 2 was 2.5 M, which matches the experimental K m O 2 determined by Abu-Soud et al. (31) in this condition. With NO synthesis from Arg but no NO accumulation, the apparent K m O 2 shifted to 7.5 M in our simulation. This shift is linked to near-geminate NO binding to the ferric heme as a consequence of catalysis and subsequent reduction of the complex, causing a proportion of iNOS to partition into the futile pathway during NO synthesis (see Scheme 1). For simulation of NO synthesis from Arg with NO accumulation, we obtained a further increase in the apparent K m O 2 to 25 M. This corresponds to a second shift reported by Abu-Soud et al. in similar circumstances (31). Although our simulated K m O 2 is not as high as the one experimentally reported (130 M), this value depends heavily on the experimental conditions, and the greater shift in K m O 2 observed by Abu-Soud et al. with NO accumulation was perhaps linked to a higher NO concentration achieved during their measurements. In any case, our simulations show that the two types of heme⅐NO binding (near geminate combination versus solution NO rebinding) lead to two distinct shifts in apparent K m O 2 for iNOS, as was observed experimentally by Abu-Soud et al. (29). Besides, our models showed a change in NADPH consumption V max (Fig. 7A). The decrease observed in the absence of substrate (1.5-fold lower) is linked to the simplification of the model to a single forward pathway that requires only 1 e Ϫ instead of three for the other models.
In the case of eNOS, Abu-Soud et al. (31) determined apparent K m O 2 values using two different substrates (Arg and NOHA). Importantly, NOHA supports a greater NO production rate in the steady state (31,47) and causes significantly greater NO to accumulate in solution (31). We therefore ran simulations of our kinetic model using NOHA or Arg as substrate and allowed for NO accumulation. Fig. 7B plots the reciprocal of NADPH consumption rates at different O 2 concentrations for each substrate. We obtained in the Arg simulation an apparent K m O 2 of 0.5 M, which is close to the K m O 2 experimentally determined by Abu-Soud et al. (31) and is consistent with very low NO concentration building up during the simulated reactions. For the NOHA simulation we obtained an apparent K m O 2 of 10 M, which was associated with an increase in NO accumulation during the measurement. A similar shift to around 20 M was experimentally observed by Abu-Soud et al. (31), and was associated with NO accumulation to about 1 M during his measurements. In addition, our simulation predicts a decrease of NADPH consumption V max from 0.32 s Ϫ1 with arginine as substrate to 0.16 s Ϫ1 with NOHA as substrate. Because the simulation for NOHA involves only 1 e Ϫ (against three in the case of arginine), this difference corresponds in fact to a 1.5-fold increase in activity when NOHA is used as the substrate instead of arginine. This has been also experimentally observed by Abu-Soud et al. (31). Thus, by modifying our kinetic model to account for both NO near geminate combination and rebinding of solution NO, we were able to simulate the shifts in apparent K m O 2 determined experimentally for eNOS. DISCUSSION We determined here missing kinetic data for iNOS and eNOS to test if their catalytic behaviors could be explained by a kinetic model that we recently developed for nNOS (18). Model simulations that used kinetic parameters particular for each NOS revealed that their individual catalytic behaviors are indeed described by the single kinetic model shown in Scheme 1. The model was able to accurately simulate the distribution of enzyme species in iNOS and eNOS during their approach to and after achieving steady-state catalysis. In addition, model simulations correctly predicted apparent K m O 2 values for iNOS and eNOS as well as the sensitivity of this parameter toward certain reaction conditions (for example, the absence or presence of substrate or buildup of NO in solution).
The kinetic model also revealed that the different catalytic behaviors of nNOS, iNOS, and eNOS are primarily due to differences in three kinetic parameters. These are the reduction rate of the ferric heme, dissociation rate of the ferric heme⅐NO product, and oxidation rate of the ferrous heme⅐NO complex. In the case of iNOS, ferric heme reduction is two to three times slower than in nNOS. This makes iNOS intrinsically less active than nNOS (i.e. ferric nNOS converts Arg to NO faster in a single turnover). However, because dissociation of the ferric heme⅐NO product is two times slower in iNOS, this makes iNOS similar to nNOS regarding their partitioning between productive and futile cycles (Scheme 1). Thus, the main difference between them resides in the 20-fold faster oxidation rate of the ferrous heme⅐NO complex of iNOS. This explains why a much smaller percentage of iNOS is present as a ferrous heme⅐NO complex in the steady state compared with nNOS and why iNOS exhibits less inhibition of NADPH consumption as the steady state is reached prior to NO buildup in solution. Because the lifetime of the ferrous heme⅐NO specie is diminished in iNOS relative to nNOS, it cycles faster through the futile cycle, and thus in unit time is also more likely to participate in the biosynthetic pathway to generate NO and partition again between futile and productive pathways. Diminished buildup of a ferrous heme⅐NO complex and concurrent faster cycling make steady-state NO production higher in iNOS than in nNOS when NO does not accumulate in solution (i.e. in the presence of oxyhemoglobin), even though iNOS is intrinsically less active than nNOS due to a slower rate of ferric heme reduction. Due to a faster rate of futile cycling in iNOS, its ratio of NADPH oxidized: NO formed in the steady state is somewhat greater than that for nNOS. This property also makes iNOS convert more NO molecules in unit time into higher oxide product (designated as nitrate in Scheme 1) than does nNOS (see nitrate production rates in the absence of solution NO listed in Table I).
In the case of eNOS, ferric heme reduction is much slower than in nNOS or iNOS, and this makes its intrinsic NO synthesis activity the lowest among the three. Because eNOS has a relatively fast rate of ferric heme⅐NO dissociation, its slow rate of ferric heme reduction ensures that the proportion of eNOS that cycles through the futile cycle is almost zero. This explains the near total absence of ferrous heme⅐NO complex buildup or inflection in NADPH consumption between the initial and steady-state phases of eNOS catalysis prior to solution NO accumulation. The lack of partitioning into the futile cycle also primarily explains how eNOS steady-state NO synthesis is only 4-fold slower than that of nNOS even though its ferric heme reduction is 40-fold slower. As explained previously (18), in nNOS the rate of steady-state NO synthesis is not simply a function of heme reduction rate but rather depends on the rates of the productive and futile cycles, which are populated in a ratio of about 1: 2 due to a unique setting of the three catalytic parameters noted above. This makes nNOS steady-state NO synthesis quite slower than its intrinsic rate (which is determined by its rate of ferric heme reduction (18,25)). In contrast, for eNOS its three kinetic parameters are set such that its steady-state rate of NO synthesis is practically the same as its intrinsic rate, which is also determined by its rate of ferric heme reduction. Fig. 8A simulates how increasing the heme reduction rate in eNOS up to the rate seen for nNOS (3-4 s Ϫ1 ) would effect buildup of its ferrous heme⅐NO complex. Because the oxidation rate of eNOS ferrous heme⅐NO complex is only slightly faster than that of nNOS, the simulation predicts that the proportion of ferrous heme⅐NO complex present in steady state would be only just below that of nNOS and be much greater than iNOS, at any given rate of heme reduction. In Fig. 8B, the simulation also predicts that, despite increased ferrous-heme⅐NO complex buildup, eNOS should gradually increase its steady-state NO release as a function of heme reduction rate within this range. Indeed, both effects as predicted by the simulation were recently observed in an NOS chimera comprised of an nNOS reductase domain and an eNOS oxygenase domain. Heme reduction reached 3.8 s Ϫ1 in this chimera and was associated with 59% of the eNOS heme being present as a ferrous heme⅐NO species during the steady state, and a 3.75-fold increase in steady-state NO synthesis (32). The magnitudes of these effects are quite similar to those predicted by the simulation in Fig. 8. Thus, our kinetic model predicts that increasing the heme reduction rate in eNOS would be a useful way to increase its NO synthesis activity in a biological setting. Indeed, we suspect this is how Akt-dependent phosphorylation of eNOS leads to a 3-fold increase in its steady-state rate of NO synthesis (25). Interestingly, the kinetic model also explains why increasing the heme reduction rate beyond a certain point is futile for both nNOS (18,25) and eNOS (Fig. 8B).
Our previous articles devoted to nNOS highlighted how its catalytic behavior is linked to near-geminate combination of NO with the ferric heme. In fact, near-geminate combination occurs in each NOS as a normal consequence of NO synthesis. As explained in this report, there also is a potential feedback inhibition for each NOS that depends on equilibrium binding of NO to the ferric heme. Thus, the global NO inhibition profile in any NOS will result from a balance between both processes. Although inhibition due to near-geminate combination is not influenced by external factors such as NO accumulation in solution, the feedback inhibition completely depends on it. The catalytic profile of any NOS will consequently depend on the setting of its catalytic and kinetic parameters and on the particular environmental conditions (concentration of enzyme and solution NO). This may explain discrepancies between previous reports regarding NOS activities and certainly allows for a broad range of flexible behaviors among NOS isoforms.
The degree to which solution NO can inhibit catalysis of any given NOS depends on its being able to bind to the NOS heme. This is a function of two parameters, namely, the kinetics of heme⅐NO binding and the amount of heme that is available for binding NO in the steady state. The kinetics of solution NO or O 2 binding to the heme in each NOS have been measured and appear similar (19,23,24,29,30). At the relative O 2 and NO concentrations present in our experiments, O 2 will far outcompete solution NO for binding to the ferrous NOS heme. Thus, it is the amount of available ferric heme that becomes important. This amount is set by the three kinetic parameters and markedly differs between the NO synthases. In the absence of solution NO accumulation, the simulation predicts that the amount of ferric eNOS and iNOS present at steady state (98 and 80%, respectively) is much greater compared with nNOS (22%). These percentages match well with those observed experimentally. Fig. 9 (A and B) simulates how the steady-state distribution of heme species will change as a function of solution NO concentration for iNOS and eNOS. In both enzymes the simulation predicts a decrease in the free ferric heme population and a coincident increase in the ferric heme⅐NO population but only a modest accumulation of the ferrous heme⅐NO specie. These simulated distributions match those observed experimentally when solution NO accumulates to micromolar levels during NO synthesis by each NOS (21,(27)(28)(29)31). Distribution of iNOS and eNOS into a ferrous heme⅐NO specie is minimal for two different reasons. In iNOS, a fast ferrous heme⅐NO oxidation rate (that is faster than the ferric heme⅐NO reduction rate) prevents the buildup of ferrous heme⅐NO specie during steady state. For eNOS, slow heme reduction ensures minimum partitioning of the enzyme into a futile cycle and leads to a species distribution without a significant amount of ferrous heme⅐NO.
How the steady-state NO synthesis activity of iNOS and eNOS changes as a function of solution NO concentration is simulated in Fig. 9C. Activity of each NOS is reduced with increasing solution NO, consistent with what is obtained experimentally when solution NO accumulates during NO synthesis (16, 26 -29, 31). This corresponds to a feedback inhibition where a portion of the enzyme is trapped in a NO-bound form, which statistically decreases the portion of enzyme likely to participate in synthesis of NO.
Our current work with eNOS and iNOS highlights some interesting concepts regarding NOS O 2 sensitivity. In the case of eNOS, heme reduction is too slow to drive the enzyme through the futile cycle and also to achieve sufficient NO concentrations for feedback inhibition. Therefore, heme reduction becomes the essential regulatory parameter and eNOS activity is directly and proportionally linked to the rate of heme reduction (14, 15, 48). As discussed above, this could provide means for external regulation, particularly through phosphorylation (14,15). These properties also cause eNOS to display almost no increase in its apparent K m O 2 when it produces NO. Rather, the apparent K m O 2 remains close to the K d value of the ferrous⅐oxygen complex (around 2 M). This makes eNOS activity insensitive to increases in O 2 concentration above 50 M, which may be of some importance for its physiologic role. The case of iNOS is totally different. Tuning of its kinetic parameters make iNOS partition significantly into the futile cycle and still remain susceptible to binding NO from solution. Thus, the apparent K m O 2 of iNOS is increased with NO synthesis and also undergoes a variable additional increase depending on the solution NO concentration (29). This property allows iNOS to function in the O 2 response of the human lung (49). The difference in the apparent K m O 2 between each NOS (without NO accumulation) may be linked to their different relative weights of the futile cycle (via ferrous heme⅐NO oxidation) and the synthetic pathway (oxygen binding to heme) in the overall oxygen requirement. Because eNOS does not significantly par-tition through the futile cycle, the apparent K m O 2 is directly linked to the synthetic pathway and derives directly from the dissociation constant of the ferrous heme⅐O 2 complex. Like nNOS, iNOS partitions partly through the futile cycle but its ferrous heme⅐NO oxidation is 30 times faster. The weight of the futile pathway in the determination of K m O 2 may therefore be less important for iNOS than for nNOS, leading consequently to a lower apparent K m O 2 for iNOS.
The particular kinetic settings of iNOS (both its heme reduction and ferrous heme⅐NO oxidation are fast) lead to a somewhat greater production of the alternative nitrogen oxide product (shown as nitrate in Scheme 1) even in the absence of NO accumulation in solution. In addition, because of its robust NO synthesis, NO typically accumulates in solution to a level where a good percentage of the iNOS ferric heme will bind solution NO. Significantly, in iNOS a majority of solution NO that rebinds to its ferric heme is cycled through the futile pathway. Thus, when the NO concentration reaches a value corresponding to an equilibrium between NO produced and NO recycled (low M), the system reaches another steady-state phase, characterized by the absence of net NO release, and thus, by a plateau in the steady-state NO concentration. This is exactly what is seen in experiments by Abu-soud et al. (29) and in our simulations. Under this condition, practically all enzyme catalysis is directed toward forming the alternative nitrogen oxide product (see Fig. 5C).
Although we depicted the alternative nitrogen oxide product as nitrate, this is not necessarily the case. Clearly, the product is not NO or NO Ϫ and instead results from a bi-molecular reaction between the ferrous heme⅐NO complex and O 2 . 3 This type of reaction has been described for several years, especially for oxyhemoglobin and myoglobin, and is supposed to give rise to nitrate (50,51). Significantly, deeper investigation led Arnold and Bohle (52) to stress the existence of an intermediate assumed to be peroxynitrite. They propose back donation of an electron from the ferrous-heme to the bound nitrosyl, which would allow it to react with O 2 more like bound nitroxide. The ability to reduce bound NO has already been described for numerous enzymes like cytochrome-c oxidase (53,54), superoxide dismutase (55), cytochrome c (56), and cytochrome P450 NOR (57)(58)(59)(60). Surprisingly, with the exception of P450 NOR, the redox potential of these enzymes does not seem to favor NO reduction, but the formation of nitroxyl and peroxynitrite was unambiguously probed (54,56). The range of heme midpoint potentials for iNOS (Ϫ260 to Ϫ300 mV (61)) approaches that of P450 NOR (Ϫ307 mV (57)). The relatively low heme midpoint potential values for iNOS compared with other NO synthases (61) should uniquely favor transition of its ferrous nitrosyl complex into a ferric nitroxyl-like complex (58). This hypothesis is validated by the ferrous heme⅐NO complex of iNOS exhibiting a resonance Raman NO stretching frequency of around 1555-1680 cm Ϫ1 , which is between the value for free NO (1840 cm Ϫ1 ) and free nitroxyl (1290 cm Ϫ1 ). Thus, the ferrous heme⅐NO complex of iNOS appears to bear significant ferric heme-nitroxyl character, and we suspect that this increases its susceptibility to oxidation. This is an important concept to explore, because oxidation the ferrous heme⅐NO complex can conceivably generate peroxynitrite as a primary product. This has already been suggested for cytochrome P450 2B4, which, when exposed to NO under aerobic conditions, led to nitration of a tyrosine residue near the heme (62). Peroxynitrite has also been suggested as an intermediate in the symmetric reaction of oxyhemoglobin with NO (63).
Heme midpoint potentials may primarily determine rates of 3 S. Adak and D. J. Stuehr, unpublished results.

FIG. 9.
Simulations showing how NO accumulation alters distribution of iNOS and eNOS heme-species and their catalytic activities. Distributions of heme species, NO concentrations, and citrulline production rates were obtained for different times of a simulation that allowed for NO accumulation. Enzyme concentrations were set at 0.1 M. A, iNOS. B, eNOS. Conditions are described under "Experimental Procedures." C, rate of citrulline production for iNOS and eNOS as a function of NO concentration, as calculated for the simulated reactions of panels A and B. Initial rates of citrulline production were 0.81 s Ϫ1 for iNOS and 0.08 s Ϫ1 for eNOS. ferrous heme⅐NO oxidation. For example, rates of ferrous heme⅐NO oxidation are faster for P450NOR and iNOS than for cytochrome c (E°Ј ϭ ϩ 270 mV) or hemoglobin (E°Ј ϭ ϩ 50 mV (64)). Thus, by virtue of its more negative heme midpoint potential, iNOS may be uniquely set to become a peroxynitrite generator during NO synthesis. We are currently exploring the validity of this hypothesis by investigating the relationship between heme midpoint potential and the kinetics and chemistry of ferrous heme⅐NO oxidation.
In summary, different settings of three kinetic parameters determine auto-regulation in each NOS, which in turn modify their catalytic characteristics regarding rate and efficiency of catalysis, and the amounts and types of nitrogen oxides that are produced. It will now be interesting to investigate how enzyme structural, electronic, and environmental features control these characteristics both statically and dynamically, and if the regulation is coupled to cellular processes.