Neuronal Nitric-oxide Synthase Mutant (Ser-1412 3 Asp) Demonstrates Surprising Connections between Heme Reduction, NO Complex Formation, and Catalysis*

Rat neuronal NO synthase (nNOS) contains an Akt-de-pendent phosphorylation motif in its reductase domain. We mutated a target residue in that site (Ser-1412 to Asp) to mimic phosphorylation and then characterized the mutant using conventional and stopped-flow spec-troscopies. Compared with wild-type, S1412D nNOS catalyzed faster cytochrome c and ferricyanide reduction but displayed slower steady-state NO synthesis with greater uncoupling of NADPH oxidation. Paradoxically, the mutant had faster heme reduction, faster heme-NO complex formation, and greater heme-NO complex accu-mulation at steady state. To understand how these behaviors related to flavin and heme reduction rates, we utilized three soybean calmodulins (CaMs) that supported a range of slower flavin and heme reduction rates in mutant and wild-type nNOS. Reductase activity and two catalytic parameters (speed and amount of heme-NO complex formation) related directly to the speed of flavin and heme reduction. In contrast, steady-state NO synthesis increased, reached a plateau, and then fell at the highest rate of heme reduction that was obtained with S1412D nNOS 1 CaM. Substituting with soybean CaM slowed heme reduction and increased steady-state NO synthesis by the mutant. We conclude the

constitutively expressed NO synthases (NOSs). The constitutive NOSs (endothelial (eNOS) and neuronal synthases (nNOS)) are Ca 2ϩ -responsive and have multiple functions in physiology and pathology (1,2). All NOSs are homodimers whose subunits contain an N-terminal oxygenase domain that binds heme, (6R)-tetrahydrobiopterin (H 4 B), and L-arginine (Arg), and a C-terminal reductase domain that binds FMN, FAD, and NADPH (3,4). A CaM binding motif is located between the oxygenase and reductase domain, and its occupancy triggers electron transfer between the FMN and heme (5). This enables NOSs to catalyze the NADPH-and O 2 -dependent oxidation of Arg to generate NO and citrulline, with Nhydroxyarginine (NOHA) being formed as an enzyme-bound intermediate (3,4).
The NOSs share many structural features but appear to differ in how their catalysis is regulated. For example, in rat nNOS we have established that heme-NO complex formation governs its rate of NO synthesis during the steady state by setting up a condition where the O 2 -dependent breakdown of the ferrous-NO complex becomes a rate-limiting step (6,7). This greatly shifts the apparent K m,O 2 value of the enzyme and thus has physiologic impact (8). In nNOS, ferrous heme-NO complex formation is rapid (6,9) and results from reduction of a ferric heme-NO species that is an immediate product of catalysis (10). In contrast, eNOS does not build up detectable ferrous heme-NO complex during steady-state catalysis (11), and its slow NO synthesis is associated with slow heme reduction (9,11).
To better understand the relationship between heme reduction rate, NO complex formation, and NO release, we developed a kinetic model for nNOS catalysis. The companion paper by Santolini et al. (12) describes our kinetic model in detail and demonstrates that it can accurately simulate initial and steady-state features of nNOS catalysis including heme-NO complex formation, a concomitant deflection in NADPH oxidation and NO synthesis, and an increase in enzyme apparent K m,O 2 values. Further simulations predicted that a complex relationship should exist between rates of heme reduction and NO synthesis during the steady state (12). Specifically, there should be an optimal rate of heme reduction beyond which NO synthesis is actually diminished, and nNOS appears to be set near its optimum heme reduction rate. Although this concept is of fundamental importance, only limited information is available on how heme reduction rate affects NO synthesis by nNOS. These include studies with CaM-troponin C chimeras (13), CaM point mutants that contain dysfunctional Ca 2ϩ -bind-ing sites (14), and NOS chimeras whose reductase domains are swapped (15). In the only study where heme reduction rates were measured (13), slower rates were obtained relative to CaM, and NO synthesis was proportional to the heme reduction rate (i.e. was slower). Unfortunately, no information was provided in these studies regarding nNOS heme-NO complex formation or how the enzymes might respond if heme reduction became faster than that obtained with CaM.
Given the above, we were interested in reports on serine/ threonine kinase Akt-dependent phosphorylation of Ser-1179 in bovine eNOS (16) and the analogous Ser in human eNOS (17). This residue is located in the NADPH-FAD module near the C terminus of the reductase domain. Cotransfection of the Akt-dependent kinase with eNOS led to phosphorylation at Ser-1179, increased NO production, and allowed the enzyme to be activated at lower Ca 2ϩ concentrations. Mutation of Ser-1179 to Asp, which mimics phosphorylation by permanently introducing a negative charge, also resulted in a 2-3-fold increase in NO synthesis and faster cytochrome c reduction (16 -18). These results suggest that the mutation may increase electron transfer in eNOS and can explain how eNOS is activated in endothelium in response to shear stress. An Akt phosphorylation motif (RXRXX(S/T)) is also present at the same site in rat nNOS and contains the corresponding residue Ser-1412. However, cotransfection of the Akt-dependent kinase with nNOS did not result in increased NO release (16), so it is puzzling why rat nNOS responded differently than eNOS.
To address this and to investigate further how electron transfer and catalysis are related, we generated the S1412D mutant nNOS and thoroughly characterized its electron transfer properties, heme-NO complex formation, and initial and steady-state catalytic behaviors. The experimental results and simulations using our kinetic model precisely explain the mutant phenotype and greatly expand our understanding of how heme reduction and NO complex formation combine to regulate NOS catalysis.
Molecular Biology-Restriction digestions, cloning, bacterial growth, and transformation and isolation of DNA fragments were performed using standard procedures. Site-directed mutagenesis was done using the Quick Change polymerase chain reaction in vitro Mutagenesis Kit from Stratagene. Incorporated mutations were confirmed by DNA sequencing at the Cleveland Clinic Core Facility. DNA containing the desired mutation was transformed into Escherichia coli for protein expression.
Oligonucleotides used to incorporate the S1412D mutation in nNOS were synthesized by Integrated DNA technologies. A silent mutation coding for a restriction site was incorporated into the oligonucleotides to aid in screening. The mutation (bold), silent restriction site (underlined), and the oligonucleotides used in the process were as follows: S1412D-BsiWI sense, GTCACCCTCCGTACGTATGAAGTGACCAAC-CGCCTTAGATCTGAGGACATCGCCTT; S1412D antisense, AAGGC-GATGTCCTCAGATCTAAGGCGGTTGGTCACTTCATACGTACGGA-GGGTGAC.
Expression and Purification of Wild-type and nNOS Mutant-Wildtype rat nNOS and the S1412D mutant both contained a His 6 tag attached to their N termini. They were overexpressed in E. coli strain BL21 (DE3) using a modified pCWori vector and purified by sequential chromatography on Ni 2ϩ -nitrilotriacetic acid and 2Ј,5Ј-ADP-Sepharose resins as described (19). The ferrous-CO adduct absorbing at 444 nm was used to quantitate heme protein content using an extinction coefficient of 74 mM Ϫ1 cm Ϫ1 (A 444 -A 500 ).
NO Synthesis, NADPH Oxidation, Ferricyanide Reduction, and Cytochrome c Reduction-Steady-state activities of wild-type and mutant nNOS were determined separately at 25°C using spectrophotometric assays that were described previously in detail (19). To assay catalytic activities versus Ca 2ϩ concentration, the standard assay mixtures were modified to contain 20 M EDTA, and different amounts of CaCl 2 were added.
H 2 O 2 -dependent NOHA Oxidation-H 2 O 2 -dependent nNOS oxidation of NOHA to nitrite was assayed in 96-well microplates at 25°C as described previously (19) with modification. The assay volume was 100 l and contained 40 mM EPPS, pH 7.6, 250 nM nNOS, or mutant, 1 mM NOHA, 1 mM dithiothreitol, 25 units/ml superoxide dismutase, 0.5 mM EDTA, and 4 M H 4 B. Reactions were initiated by adding 30 mM H 2 O 2 and stopped after 10 min by adding 1300 units of catalase. Nitrite was detected at 550 nm after adding the Griess reagent (100 l) and quantitated based on nitrite standards.
Heme and Flavin Reduction-The kinetics of flavin and heme reduction were analyzed at 10°C as described previously (13,19) using a diode array detector and stopped-flow apparatus from Hi-Tech Ltd. (model SF-61) equipped for anaerobic work. Heme reduction was followed by formation of the ferrous-CO complex, and the kinetics were determined by absorbance change at 444 nm. Reactions were initiated by rapid mixing an anaerobic, buffered, CO-saturated solution containing 50 M NADPH with an anaerobic CO-saturated solution containing wild-type or mutant nNOS (2 M), 40 mM EPPS buffer, pH 7.6, 10 M H 4 B, 0.3 mM dithiothreitol, 5 mM Arg, 4 M CaM, or 20 M soybean CaM proteins, and 1 mM Ca 2ϩ . Flavin reduction was monitored under the same conditions at 485 nm. Signal-to-noise ratios were improved by averaging at least 10 individual mixing experiments. The time course of absorbance change was fit to single or multiple exponential equations using a nonlinear least square method provided by the instrument manufacturer.
Kinetics of Ferrous Heme-NO Buildup and Concurrent NADPH Oxidation-Experiments were done at 10°C in a Hi-Tech 51MX stoppedflow instrument equipped with a variable wavelength detector. To initiate NO synthesis, an air-saturated solution that contained 40 mM EPPS, pH 7.6, 2 M nNOS or mutant, 0.4 mM dithiothreitol, 2 mM Arg, 20 M H 4 B, 4 M CaM, or 20 M soybean CaM proteins, 40 M NADPH, and 0.5 mM EDTA was rapid mixed with a buffered solution containing 2.4 mM Ca 2ϩ . Absorbance at 436 nm was monitored to follow ferrous heme-NO formation and at 340 nm to follow NADPH oxidation (7). The concentration of ferrous heme-NO complex formed during NO synthesis was estimated from the absorbance change at 436 nm using an extinction coefficient of 49,800 M Ϫ1 cm Ϫ1 (6), whereas the amount of NADPH oxidation was determined using an extinction coefficient of 6,220 M Ϫ1 cm Ϫ1 at 340 nm. Signal-to-noise ratios were improved by averaging six consecutive scans. Each experiment was performed three separate times.
Simulations-These were run as described in detail in the accompanying paper (12). Iterative computations were made using the 13 rate equations that describe the kinetic model. Simulations used the same rate constants listed in (12) unless specified otherwise. All rate constants were determined at 10°C.
To simulate the buildup of heme-NO complex using heme reduction rates determined for CaM and the ScaM proteins, the time step used was 50 s over a 3-s period, generating 60,000 points for each case. The curves were plotted and analyzed to determine a rate of complex formation in each case, which corresponds to the slow phase of ferrous heme-NO buildup. We removed the initial lag phase (up to 300 ms) from each simulated kinetic trace by using a floating origin: y ϭ A ϩ B exp(Ϫk(x Ϫ x o )). The Origin© software used was based on the Marquardt algorithm. To simulate percentage of ferrous-NO complex achieved at steady state, simulations were run using a 1-ms time step for each heme reduction rate obtained with CaM and the ScaM proteins. Ratios of NADPH oxidized per NO released were calculated under the same simulation conditions (see Ref. 12 for details). Simulated rates of NO release at different O 2 concentrations were calculated using a time step of 1 ms except for O 2 concentrations above 2 mM where the time step was 0.1 ms. The heme reduction rate was set to values we obtained with CaM in this report (3.9 s Ϫ1 for wild-type and 5.4 s Ϫ1 for S1412D mutant). For O 2 reaction with the Fe(II) heme-NO complex we used a pseudo-first order rate constant (0.19 s Ϫ1 ) that was reported for nNOS by Adak et al. (7) at a single O 2 concentration (half air-saturated) at 10°C. We then varied this rate constant in direct proportion to the simulated O 2 concentration.

RESULTS
The S1412D nNOS used in our study was purified to homogeneity and had the same molecular mass as wild-type nNOS (Fig. 1). This rules out potential artifacts due to contaminating reductase domain. The mutant displayed the same absorbance spectrum in the presence of Arg and H 4 B as wild type, and its ferrous CO-bound spectrum contained a Soret peak at 445 nm (data not shown). These properties indicate that mutation of Ser-1412 did not perturb enzyme structure, substrate, and prosthetic group binding or heme electronic environment.
NOS Catalytic Activities-We compared catalysis by wildtype and S1412D nNOS by measuring their rates of H 2 O 2 -dependent NOHA oxidation, cytochrome c or ferricyanide reduction, and NO production. The wild-type and mutant enzymes catalyzed H 2 O 2 -dependent oxidation of NOHA at similar rates (8.4 and 9.0 min Ϫ1 , respectively), consistent with their oxygenase domains being identical and this reaction not requiring electrons from the reductase domain. Both cytochrome c and ferricyanide reduction was faster in the mutant compared with wild-type in the presence or absence of Ca 2ϩ /CaM (Table I). The mutant did not synthesize NO without CaM and with CaM had a rate of NO synthesis that was 25% diminished compared with wild-type enzyme (Table I). We conclude that the mutant displays slower NO synthesis, but this cannot be explained by defects in oxygenase domain catalysis or electron transfer through the reductase domain.
Electron flux between the reductase and oxygenase domains can be a critical determinant of NO synthesis (9,11,(13)(14)(15). We therefore compared steady-state rates of NADPH oxidation by CaM-bound S1412D and wild-type nNOS under a number of different conditions (Table II). S1412D nNOS had faster rates of NADPH oxidation than wild-type in all cases. This is consistent with the mutant catalyzing faster electron transfer from its reductase domain to external acceptors such as ferricyanide or cytochrome c. However, it cannot account for the observed decrease in NO synthesis. In fact, the concomitant decrease in NO synthesis and increase in NADPH oxidation led to a more uncoupled activity during the steady state, with the calculated NADPH/NO ratio rising to 3.5 for the mutant instead of 2 for wild-type nNOS (Tables I and II).
Ca 2ϩ -dependent Activation of Wild-type and Mutant nNOS-Two different groups (16,17) reported that phosphorylation at the analogous residue in eNOS (Ser-1179) or substitution with Asp increased enzyme affinity toward Ca 2ϩ during CaM activation of NO synthesis. As shown in Fig. 2, at lower Ca 2ϩ concentrations the nNOS mutant showed higher steady-state NO synthesis rates compared with wild-type, but this disappeared sharply as Ca 2ϩ concentrations increased. The greatest difference in activity was seen at 13 M Ca 2ϩ (Fig. 2, upper inset), which was a suboptimal concentration for both enzymes. In contrast, when the experiment was repeated using cytochrome c reduction to measure activity, mutant activity remained consistently higher than wild-type across the same range of Ca 2ϩ concentrations. We conclude that the mutation causes enhanced response to Ca 2ϩ in nNOS, but this only occurs at suboptimal Ca 2ϩ concentrations and only when activity is measured by steady-state NO synthesis.
Heme and Flavin Reduction-We next examined the kinetics of flavin and heme reduction at 10°C by stopped-flow spectroscopy. Observed rate constants for heme and flavin reduction are listed along with other data in Tables III and IV. Traces in Fig. 3 panel A illustrate heme reduction in mutant and wildtype nNOS in response to CaM. Heme reduction was monophasic and achieved the same extent of reduction in both enzymes but was 1.4 times faster in the mutant. Thus, a slower steadystate NO synthesis activity in the mutant is actually associated with a faster rate of heme reduction. Panels B and C compare rates of flavin reduction in the presence or absence of CaM. Flavin reduction was biphasic in all cases and was faster in the mutant only in the absence of CaM. This indicates that faster cytochrome c reduction in the mutant is only associated with faster flavin reduction in the CaM-free state.
Relationships between Flavin and Heme Reduction and Enzyme Behavior-To understand better the mutant phenotype, we examined how different rates of flavin and heme reduction would impact mutant behaviors relative to wild type. For this purpose we utilized two native soybean calmodulins (ScaM1 and ScaM4) and a point mutant of ScaM1 (M144V), which were shown by Johnson and colleagues (20,21) to bind to nNOS with good affinity but support different rates of cytochrome c reduction and NO synthesis in the steady state. Our hope was that these "CaM analogs" would behave like CaM-troponin C chimeras in supporting different rates of flavin and heme reduction in mutant and wild-type nNOS (13). As shown in Tables III  and IV, this was the case because the ScaM proteins and CaM each supported different rates of flavin and heme reduction in the two enzymes. Table III shows how different rates of flavin reduction affect cytochrome c reductase activity in mutant and wild-type nNOS. Flavin reduction was biphasic in all cases, with each phase representing one-third to two-thirds of the total absorbance change depending on the CaM used in the experiment. Although the two nNOS started with different basal rates of flavin reduction, they achieved identical rates in response to each CaM and thus showed the same rank order of response (CaM Ͼ ScaM4 Ͼ M144V ScaM1 Ͼ ScaM1) and range of enhancement (100% with CaM to 50% with ScaM1). In both FIG. 1. Purification of S1412D nNOS followed by SDS-polyacrylamide gel electrophoresis. Lane 1, Coomassie-stained molecular mass markers; lane 2, wild-type nNOS standard; lanes 3-6, S1412D mutant after final 2Ј,5Ј-ADP purification step, after elution from the Ni 2ϩ -nitrilotriacetic acid resin; after ammonium sulfate precipitation, and after the crude extract, respectively. mutant and wild-type nNOS the cytochrome c reductase activities increased according to the same rank order observed for flavin reduction. As shown in Fig. 4, a linear relationship exists between rates of flavin reduction and cytochrome c reduction for both proteins. Moreover, Fig. 4 clearly shows that cytochrome c reduction was consistently greater in S1412D nNOS, although it displayed the same rate of flavin reduction as wild-type nNOS in response to any given CaM protein. Table IV shows that heme reduction by CaM and the ScaM proteins was stimulated according to the same rank order as for flavin reduction (CaM Ͼ ScaM4 Ͼ M144V ScaM1 Ͼ ScaM1), with heme reduction being consistently faster in S1412D nNOS in response to each protein. Because this exactly matches what we observed for cytochrome c reduction, it suggests that electron transfer to cytochrome c or the nNOS heme occurs through mechanisms that are affected in the same way by the S1412D mutation.
We next examined how different heme reduction rates affect the kinetics of heme-NO complex buildup, the percentage of heme-NO complex attained at steady state, and steady-state NO synthesis in mutant and wild-type nNOS. These three parameters are good determinants of NOS catalytic behavior (see Ref. 12). Results are summarized in Table IV. Heme-NO complex buildup was followed at 436 nm in the stopped-flow instrument. It occurred within the first seconds after initiating NO synthesis and was best described by a two-exponential function in all cases, with the slow phase representing the majority of absorbance change. As may be seen in Fig. 5, panel A, the fast and slow phases were both linearly proportional to the rate of heme reduction, with data from wild-type and mutant enzymes fitting well along the same two lines. This indicates that heme-NO complex formation is directly linked to the heme reduction rate. Fig. 5, panel B, shows that there was also a consistent increase in percentage heme-NO complex attained at steady state. Thus, it is clear that both rate and percentage of heme-NO complex buildup increases with heme reduction rate in wild-type and S1412D nNOS.
In contrast, the relationship between the heme reduction rate and steady-state NO synthesis was more complex. For wild-type nNOS the steady-state activity increased with heme reduction. In contrast, for S1412D nNOS the activity increased, reached a plateau, and then fell at the highest rate of heme reduction (5.4 s Ϫ1 , Table IV). Thus, SCaM4 supported a greater steady-state rate of NO synthesis in S1412D nNOS than did CaM itself, despite SCaM4 supporting a slower rate of heme reduction. When all steady-state NO synthesis rates were plotted versus speed of heme reduction (Fig. 5, panel C), the distribution suggested that an optimal rate of heme reduction exists for steady-state NO synthesis, and this optimum is exceeded in CaM-bound S1412D nNOS.
FIG. 2. Ca 2؉ dependence of S1412D and wild-type nNOS. Activities were measured at the indicated concentrations of added Ca 2ϩ and in the presence of excess CaM and 20 M EDTA as described under "Experimental Procedures." The upper panel plots NO synthesis activities, and the lower panel plots cytochrome c reductase activities. Insets plot the ratio of activities between S1412D and wild-type nNOS at each Ca 2ϩ concentration. Data are representative of two independent experiments done with different protein preparations. TABLE III Correlation between rates of cytochrome c reduction and flavin reduction in wild type and S1412D nNOS Cytochrome c reduction was measured at 25°C, and flavin reduction was measured at 10°C as detailed under "Experimental Procedures." Numbers in parentheses indicate the relative absorbance change for each phase of flavin reduction. Values represent the mean and S.D. for three measurements each.

System
Cytochrome c reduction The S1412D mutant is quite valuable because it helps demonstrate how rates of flavin and heme reduction relate to the initial and steady-state catalytic behavior of nNOS. In the steady state, the mutant displayed increased NADPH oxidase and cytochrome c reductase activities but somewhat slower NO synthesis. Remarkably, this differed from the analogous S1179D eNOS mutant, which displayed an increase in all three activities (16 -18). We first discuss how the S1412D mutation alters electron transfer in and out of the nNOS reductase domain, and then we discuss the effect of the mutation on heme reduction in light of our kinetic simulation model (see accompanying paper by Santolini et al. (12)) that can explain the behavior of the S1412D nNOS and why it differs from the analogous eNOS mutant.
Reductase Domain Electron Transfer-Before discussing the effects of the mutation, it is important to review some general properties of reductase domain catalysis. All NOS reductase domains are composed of an NADPH-FAD module and an FMN module (22). As established for nNOS by Mayer and co-workers (23) and Gachhui et al. (24), electron transfer to cytochrome c and ferricyanide occurs via direct electron transfer from the reductase domain and does not require heme reduction or superoxide generation. Cytochrome c binds with high affinity to the FMN module and accepts electrons exclusively from this site, whereas ferricyanide accepts electrons from either FAD or FMN (9,(23)(24)(25).
Our results are the first to show that modification of Ser-1412 alters both flavin reduction and reductase activities in any NOS. However, it is important to note that S1412D nNOS displayed faster flavin reduction relative to wild type only in the CaM-free state, and this was associated with faster reduction of cytochrome c and ferricyanide. Upon CaM binding, the mutant lost its advantage regarding speed of flavin reduction but still displayed greater cytochrome c reductase activity than wild type. This relationship held when any of three ScaM proteins substituted for CaM. We conclude that electron transfer to cytochrome c is directly related to the speed of flavin reduction (see Table IV). However, in CaM-bound nNOS Ser-1412 primarily influences this activity through a mechanism that does not involve changing the speed of flavin reduction. Although the details are not understood, we suspect that phosphorylation at Ser-1412 alters a physical interaction in the reductase domain that enhances electron transfer out of the FMN module. One mechanism would have the mutation make the FMN group more accessible to external electron acceptors, and we are currently testing this possibility. Because Ser-1412 is in the NADPH-FAD module and therefore "upstream" from the FMN module, the Asp mutation must alter an interaction between these two modules even in the presence of CaM. Recent data suggest the C terminus of the NADP-FAD module could be a candidate for interacting with the FMN module (26). In any case, the S1412D mutation seems to speed electron transfer to the nNOS heme and to cytochrome c by the same mechanism, because changes in these two parameters correlate precisely in the mutant and wild-type enzymes (see Tables III  and IV).
The S1412D mutation may increase electron transfer from NADPH to FAD in CaM-free nNOS by a different mechanism. We suspect the S1412D mutation partly relieves a repression that is exerted by the FMN module on the NADPH-FAD module. This repression of "upstream" electron transfer was first described by Adak et al. (19), who showed that several FMN module mutants had faster electron transfer between NADPH and FAD. The repression in wild-type nNOS is fully relieved by CaM binding or by FMN removal (19), and our current results show that the ScaM proteins can also relieve the repression but to lesser degrees than CaM. It is also clear that phosphorylation of Ser-1412 should only provide partial relief of repression compared with CaM binding or FMN removal. It is tempting to speculate that Ser-1412 is located at a site in the NADPH-FAD module that interacts with a repressor element from the FMN module. Although the repressor element has not been identi- B and C) in wild-type and S1412D nNOS. Stoppedflow traces for wild-type and S1412D nNOS are denoted by solid and dotted lines, respectively. The S1412D traces also contain a calculated line of best fit (dashed) in panels A and C. Heme reduction was detected by CO binding, and the kinetics were determined from the change in absorbance at 444 nm with time. CaM-bound enzymes were rapidly mixed with excess NADPH to trigger flavin and heme reduction at 10°C under anaerobic conditions as described under "Experimental Procedures." Flavin reduction was followed at 485 nm under anaerobic conditions at 10°C after mixing excess NADPH with CaM-bound (panel  Table III. Open and solid symbols denote wild-type and S1412D nNOS, respectively. The enzymes contained wild-type CaM (E and q), SCaM-4 (Ⅺ and f), SCaM-V144M (‚ and OE), and SCaM-1 (छ andࡗ). In the absence of CaM the flavin reduction rates are denoted by ƒ and . The lines represent the calculated best fit of values obtained for the CaM-and ScaM-bound enzymes. fied, a good candidate is a 40-amino acid insert present in the FMN module of nNOS that inhibits ferricyanide and cytochrome c reduction in the CaM-free state (27)(28)(29). Perhaps this element inhibits electron transfer between NADPH and FAD through interaction with Ser-1412.

FIG. 3. Kinetics of heme reduction (panel A) and flavin reduction (panels
Catalytic Control in S1412D nNOS-From the perspective of NO synthesis there is only one relevant change in the S1412D mutant, its faster heme reduction. But our analysis of catalytic parameters shows apparent contradictions. Indeed, how can a faster heme reduction give rise to greater heme-NO complex formation yet slower NO synthesis in the steady state? These paradoxical effects are in fact related to heme-NO complex formation during NO synthesis and can be understood within the context of our kinetic model for nNOS (12). An abbreviated kinetic model is illustrated in Fig. 6 and involves a forward pathway that generates the ferric nNOS heme-NO complex and citrulline as products (10). This heme-NO species then partitions between two paths to regenerate ferric enzyme. One involves NO dissociation from the ferric heme (step B), and the other involves reduction of the heme-NO species to form the ferrous heme-NO complex (step AЈ). Importantly, the reductive path is futile because little or no NO is released from the ferrous heme-NO complex before it reacts with oxygen (step C). Thus, partitioning between productive and futile pathways reflects relative rates of NO dissociation from the ferric heme versus its reduction to the ferrous heme-NO species. Because NO dissociation and heme reduction occur at similar speeds in nNOS (ϳ5 and 3.6 s Ϫ1 at 10°C, respectively), and oxidation of the ferrous heme-NO complex is comparatively slow (ϳ0.19 s Ϫ1 ), a majority of nNOS is present as the ferrous heme-NO complex during steady-state NO synthesis (6 -8).
Computer simulations of our kinetic model use rate constants derived for nNOS at 10°C that are described in the accompanying article (12). As shown there, simulations accurately reproduce the initial and steady-state behaviors of wildtype nNOS and the behaviors of the hyperactive mutant W409F (7,30). We therefore utilized the kinetic model to understand the pre-steady-state and steady-state behaviors of S1412D nNOS. Because this mutation is in the reductase domain, it is reasonable to assume that it only affects the speed of heme reduction. In fact, several data obtained with mutant and wild-type nNOS distributed along a single line when plotted as a function of heme reduction rate (see Fig. 5) strongly supported this assumption. The nNOS heme can receive an electron from the reductase domain at two different steps in the kinetic model as follows: during NO biosynthesis (step A in Fig.   FIG. 5. The rate of heme-NO complex formation (panel A), the amount of heme-NO complex attained during steady state (panel B), and the rate of steady-state NO synthesis (panel C) measured as a function of heme reduction rate. Data were obtained from Table IV. Graphs include all measurements for wild-type (open symbols) and S1412D nNOS. Panel A shows both fast and slow phases of heme-NO complex buildup with calculated lines of best fit. The enzymes contained wild-type CaM (E and q), SCaM-4 (Ⅺ and f), SCaM-V144M (⌬ and OE), and SCaM-1 (छ and ࡗ).

TABLE IV
Catalytic behaviors of wild type and S1412D nNOS as a function of the heme iron reduction rate Different heme reduction rates were achieved using CaM and three SCaM proteins. NO synthesis was measured at 25°C, and the other parameters were measured at 10°C. Measurements and analyses are detailed under "Experimental Procedures." The values are the mean and S.D. for three measurements. The numbers in parentheses indicate the relative proportion of absorbance change for each phase of ferrous-NO complex formation. 6) and after the ferric heme-NO complex forms (step AЈ in Fig.  6). We assumed that the rate of heme reduction for both steps was identical in S1412D nNOS, as appears to be the case for wild-type nNOS. 2 We then ran simulations of the kinetic model using heme reduction rates that were measured in response to CaM and the three ScaM proteins (see Tables III and IV).
Our experimental results showed that a linear relationship exists between the rates of heme reduction and heme-NO complex buildup in the first seconds of NO synthesis (Fig. 5). The fast phase of this reaction likely involves NO binding to ferric heme as soon as it is made (10,31). Thus, as the heme reduction rate increases, NO is made faster, and the ferric heme-NO product should form more quickly. In addition, since the heme-NO complex we followed here is ferrous (Soret peak at 436 nm), we suspect that reduction of the ferric heme-NO product also increases in speed exactly as does heme reduction in the biosynthetic pathway. To test these assertions, we utilized our kinetic model to simulate the kinetics of heme-NO complex buildup at different heme reduction rates, using a 50-s time step over a 3-s period (see "Experimental Procedures"). Simulated curves for four heme reduction rates are depicted in the inset of Fig. 7A. The lag phase we observe is due to our model having difficulty resolving the very rapid initial phase (the first 300 ms; discussed in Ref. 12). Despite this, we could fit the subsequent mono-exponential portion of each curve (which reflects the slow phase of heme-NO complex formation) to obtain simulated rates for heme-NO complex buildup at each heme reduction rate. As shown in panel A of Fig. 7, the simulated rates actually distribute along the same line as the experimental values for slow phase heme-NO complex buildup in S1412D and wild-type nNOS. Thus, results from the simulation are consistent with (a) a faster intrinsic NO synthesis for the S1412D mutant, (b) ferric heme-NO complex buildup prior to NO release, and (c) equivalent reduction rates for ferric and ferric-NO species within a given enzyme.
The amount of ferrous heme-NO complex attained at steady state also correlated with the rate of heme reduction (see Fig.  5). This parameter likely reflects enzyme partitioning between a productive regenerating cycle (release of NO from the ferric heme) and a futile one (reduction of the ferric-NO complex) (see Fig. 6). Thus, as heme reduction speeds up, the balance between these two paths should shift to favor more ferrous-NO complex formation and less NO release. This is well depicted by our model simulations. In Fig. 7, panel B, the measured and simulated percentages of ferrous heme-NO species present at steady state are plotted as a function of heme reduction rate. Although the simulated percentages are all slightly higher than the experimental values, both groups follow an identical distribution pattern that increases as heme reduction rate increases. Hence, an increase in the heme reduction rate will increase the percentage of enzyme that partitions into the futile cycle.
These two facets of enzyme catalysis (intrinsic rate of NO synthesis and enzyme partitioning toward the ferrous heme-NO complex) are both controlled by heme reduction, but they counterbalance one another to different degrees depending on the rate of heme reduction. Thus, their combined effect on steady-state NO synthesis is complex. This was noted in the accompanying paper by Santolini et al. (12) who simulated how steady-state NO synthesis varies with heme reduction rate. The analysis predicted that NO synthesis should increase, reach a maximum, and then fall as the rate of heme reduction increases. This pattern in fact matches what we observed here for wild-type and S1412D nNOS at different rates of heme reduction (see Fig. 5, panel C). Thus, our kinetic model can explain why we observed an optimum range of heme reduction for NO synthesis, and why the S1412D mutant has slower steady-state activity. Specifically, faster heme reduction in the mutant speeds its intrinsic NO formation (resulting in faster ferric heme-NO complex formation) but also causes a greater proportion of enzyme to partition into the futile cycle. Our data suggest that heme reduction rates between 3.5 and 4.5 s Ϫ1 strike an optimal balance between these opposing parameters to enable peak NO synthesis by nNOS. Heme reduction in CaM-bound S1412D nNOS is 5.4 s Ϫ1 and therefore is too fast to strike an optimal balance, resulting in slower NO synthesis during the steady state. However, when its heme reduction is slowed to 4.6 s Ϫ1 by using ScaM4 in place of CaM, a near optimal balance is regained, and the rate of NO synthesis is increased.
Simulation of our kinetic model using currently available rate constants actually predicted an optimal rate for heme reduction between 2 and 3 s Ϫ1 (12), which is slightly slower than the experimental optimum we observed here (about 3.8 s Ϫ1 , see Fig. 5, panel C). However, it is important to realize that the optimum rate of heme reduction depends on the following two facets of the kinetic mechanism, which themselves are functions of multiple reactions: 1) partitioning between productive and futile cycles, which depends on rates of ferric heme-NO formation, NO dissociation, and reduction of the ferric 2 J. Santolini, S. Adak, and D. J. Stuehr, unpublished results.

FIG. 7.
Model simulation of wild-type and S1412A nNOS behaviors. Panels A and B combine experimental values obtained for wildtype and S1412D nNOS that contained CaM or the ScaM proteins. Panel A overlays wild-type (ƒ), S1412D mutant (OE), and simulated (Ⅺ) rates of ferrous heme-NO complex buildup (slow phase, see "Experimental Procedures") as a function of heme reduction rate with a dashed line of best fit included for the simulated data. The inset shows simulated kinetics for heme-NO complex buildup at different heme reduction rates: a, 2 s Ϫ1 , b, 2.9 s Ϫ1 , c, 3.9 s Ϫ1 , d,) 5.4 s Ϫ1 . Panel B, overlays wild-type (ƒ), S1412D mutant (OE), and simulated (Ⅺ) percentages of ferrous heme-NO complex attained at steady state as a function of heme reduction rate. Panel C plots simulated rates of NO synthesis at steadystate versus O 2 concentration (log scale). See text for details. The inset shows a close up view of the simulated rates versus O 2 concentration (normal scale) within the physiologic range.
heme-NO complex, and 2) the relative rates of the productive and futile cycles, which depend on NO dissociation from the ferric heme and the reaction of O 2 with the ferrous heme-NO complex. Thus, the difference between simulated and experimental values for optimal rate of heme reduction is probably reasonable given such a complex system. Small inaccuracies in the kinetic constants if present can shift the optimum rate for heme reduction (see Fig. 5C in Ref. 12). Additional measurements and refinement of the simulation should reconcile the relatively small difference between the theoretical and experimental optimum and should also reveal mechanisms for "tuning" the optimum.
Kinetic simulations also predict that the rate of NADPH oxidation will reach a maximum as heme reduction rate increases in nNOS (12), but this parameter does not fall off as fast as NO synthesis. Thus, with the faster heme reduction there is a gradual decrease in catalytic efficiency. This is because faster heme reduction favors cycling through the futile cycle, which consumes NADPH without NO release and therefore leads to increased uncoupling between NADPH consumption and NO production. The simulation predicts S1412D nNOS should oxidize 3.65 NADPH per NO formed, which is quite similar to our experimental value of 3.5. These values are greater than the ratio for wild-type nNOS (the simulation predicts 2.8, the experimental value was 2.0) and signifies that a greater proportion of mutant enzyme cycles through the futile pathway during the steady state. In fact, the simulation predicts that 52% of S1412D nNOS cycles through the futile pathway in unit time, which is understandable given that its rate of heme reduction (5.4 s Ϫ1 ) is about the same as its NO release (5 s Ϫ1 ) and gives a partition coefficient near 1.
Another parameter that is affected by this complex regulation is enzyme apparent K m,O 2 . The apparent K m,O 2 measured during steady state is actually a combined function of ferrous heme affinity for O 2 and the oxidation rate of the ferrous heme-NO complex, which depends differently on O 2 concentration (8). This gives wild-type nNOS a surprisingly high apparent K m,O 2 that enables NO release in a graded manner across the entire physiologic concentration range for O 2 (0 -200 M) (8). We ran simulations to examine how the S1412D mutant and wild-type nNOS might differ in their steady-state NO synthesis as a function of O 2 concentration (Fig. 7, panel C). The rate of ferrous heme-NO oxidation was assumed to vary in direct proportion to the O 2 concentration, based on earlier data obtained with wild-type nNOS (8). The simulation predicts that the apparent K m,O 2 value of the mutant will shift to a somewhat higher value (738 M) compared with wild type (482 M) (using heme reduction rates of 3.9 and 5.4 s Ϫ1 , respectively). This makes sense, because the increased rate of heme reduction of the mutant causes a greater buildup of ferrous heme-NO complex during steady state. In air-saturated buffer (ϳ250 M O 2 ), the mutant rate of NO synthesis is predicted to be 84% that of wild type, which is close to the 78% value calculated from our experimentally determined rates. As shown in the inset of Fig.  7, panel C, the simulation predicts the mutant will always display slower NO synthesis throughout the physiological concentration range of O 2 . This is because oxidation of the ferrous heme-NO species remains rate-limiting for the futile cycle within this range of O 2 concentrations. However, as the O 2 concentration further increases to very high levels, the oxidation rate of the ferrous heme-NO complex is predicted to eventually become faster than the rate of heme reduction. At this point, the rate-limiting step in the futile cycle would switch to heme reduction, and under this circumstance the simulation predicts NO synthesis by the mutant would become faster than wild-type nNOS (Fig. 7, panel C).
Catalytic Control in eNOS and in Other nNOS Mutants-Our kinetic model may also explain why the analogous mutation in eNOS (S1179D) increased steady-state NO synthesis along with NADPH oxidation and cytochrome c reduction (16 -18). Basically, heme reduction in eNOS is much slower than in nNOS (Ͻ0.1 s Ϫ1 at 10°C; Ref. 11) and thus is in the range where increasing it should speed NO synthesis according to the kinetic model (12).
Two other reductase domain mutants have a similar catalytic phenotype as we observed here for S1412D nNOS. Both are nNOS deletion mutants, one missing the final 33 C-terminal residues (32) and the other missing a 42-residue insertion that is present in the FMN module (28). NADPH oxidation and cytochrome c reductase activities increased in both mutants relative to wild type, whereas their steady-state NO synthesis was either slower (28) or about the same (32). Rates of flavin and heme reduction were measured in the N-terminal deletion mutant under aerobic conditions and were found to be faster than wild type (32). However, neither group was able to explain the catalytic behavior of these mutants. We suspect that their behaviors reflect an increase in heme reduction rate above the optimum value for nNOS and will ultimately be understood based on our kinetic model.
Biologic Implications-There is no obvious advantage to speeding heme reduction in rat nNOS, because it already functions near its optimum rate and is predicted to have faster steady-state NO synthesis than the S1412D mutant throughout the physiologic O 2 concentration range. Indeed, our results suggest that phosphorylation will increase NO synthesis relative to native enzyme only within a narrow Ca 2ϩ concentration range that is suboptimal for full CaM binding. Thus, the Ser-1412 phosphorylation site in rat nNOS is of uncertain physiologic consequence. On the other hand, eNOS is set to respond differently to phosphorylation by increasing NO synthesis in response to a faster heme reduction rate. It will be interesting to see if auxiliary proteins that increase eNOS activity such as HSP-90 also act by speeding heme reduction. For nNOS, other mechanisms are available to increase its steady-state NO synthesis. For example, in nNOS mutants W409F and W409Y the ferrous heme-NO complex reacts 7 times faster with O 2 (7), and hyperactivity results despite their heme reduction being slower than wild type (7). 3 Although factors that influence ferrous heme-NO complex oxidation are not identified, our analysis suggests that modulation of this parameter would be a good strategy to up-regulate NO release by nNOS.