Arginine Conversion to Nitroxide by Tetrahydrobiopterin-free Neuronal Nitric-oxide Synthase

We studied catalysis by tetrahydrobiopterin (H4B)-free neuronal nitric-oxide synthase (nNOS) to understand how heme and H4B participate in nitric oxide (NO) synthesis. H4B-free nNOS catalyzed Arg oxidation toN ω-hydroxy-l-Arg (NOHA) and citrulline in both NADPH- and H2O2-driven reactions. Citrulline formation was time- and enzyme concentration-dependent but was uncoupled relative to NADPH oxidation, and generated nitrite and nitrate without forming NO. Similar results were observed when NOHA served as substrate. Steady-state and stopped-flow spectroscopy with the H4B-free enzyme revealed that a ferrous heme-NO complex built up after initiating catalysis in both NADPH- and H2O2-driven reactions, consistent with formation of nitroxyl as an immediate product. This differed from the H4B-replete enzyme, which formed a ferric heme-NO complex as an immediate product that could then release NO. We make the following conclusions. 1) H4B is not essential for Arg oxidation by nNOS, although it helps couple NADPH oxidation to product formation in both steps of NO synthesis. Thus, the NADPH- or H2O2-driven reactions form common heme-oxy species that can react with substrate in the presence or absence of H4B. 2) The sole essential role of H4B is to enable nNOS to generate NO instead of nitroxyl. On this basis we propose a new unified model for heme-dependent oxygen activation and H4B function in both steps of NO synthesis.

oxygenase domain with binding sites for heme, 6R-tetrahydrobiopterin (H4B), and Arg, and a C-terminal reductase domain with binding sites for FMN, FAD, and NADPH (4,5). A ϳ20amino acid consensus site for calmodulin (CaM) binding is located between the reductase and oxygenase domains of each NOS (6). The reductase domain transfers NADPH-derived electrons to the oxygenase domain in response to CaM binding, and this enables heme-dependent oxygen activation and stepwise conversion of Arg to NO and citrulline, with N-hydroxy-L-arginine (NOHA) being formed as an intermediate (4,5). Stoichiometry studies show that NOS consumes 1 mol of NADPH for Arg hydroxylation, and consumes an additional 0.5 mol of NADPH to oxidize NOHA to citrulline plus NO (7,8). In the absence of Arg, NOS heme reduction leads to superoxide and H 2 O 2 production (9,10).
The NOS heme is ligated to a cysteine thiolate (11)(12)(13) and thus is likely to activate oxygen for substrate oxidation as occurs in the cytochrome P450 monooxygenases (14). In fact, both steps of NO synthesis are envisioned to involve hemebased oxygen activation and catalysis. However, NOS also requires H4B, and given that H4B is not required for heme reduction (15), it has been puzzling why this cofactor is essential for NO synthesis. H4B affects NOS in many ways; it stabilizes dimeric structure (16,17), increases affinity for substrate (18,19), speeds decay of the NOS ferrous-O 2 complex (20), increases the rate of NADPH consumption (15,21), and controls the midpoint potential of the heme (15). However, only H4B's ability to destabilize the ferrous-O 2 complex appears linked to its essential role in NO synthesis.
Aromatic amino acid hydroxylases use H4B as an electron donor for oxygen activation (22,23), and a similar role for H4B in the NOS reaction has long been considered (24). Recent crystal structures of NOS oxygenase domains (11)(12)(13) show that H4B binds near the heme edge and forms a hydrogen bond between N3 of its pterin ring and a heme propionate, consistent with a possible role in heme reduction. Indeed, Hurshman et al. (25) recently used freeze-quench EPR to demonstrate that some H4B can convert to an H4B radical during reaction of ferrous NOS oxygenase domain with O 2 . Although this shows how H4B could participate in oxygen activation by donating an electron to the ferrous-dioxy heme, to what extent this takes place during normal NO synthesis is unknown. Clearly, electron donation from H4B is not required for NOHA oxidation, because H4B-free NOS catalyzes oxidation of NOHA to citrulline and nitroxyl (NO Ϫ ) in either an NADPH-or H 2 O 2 -driven reaction (26,27). In these same systems, Arg oxidation was not observed, leading the authors to conclude that H4B is essential for Arg hydroxylation. However, this is a troublesome proposal because it means that H4B's role in oxygen activation depends on the identity of the substrate, or implies that reactivity of NOS heme-oxy species somehow differ in the presence or ab-sence of H4B. In addition, it is still unclear how NOHA oxidation by H4B-bound NOS generates NO rather than NO Ϫ .
To address these issues, we extensively investigated Arg and NOHA oxidation by H4B-free neuronal NOS (nNOS). Our results reveal that nNOS does indeed catalyze oxidation of both Arg 2 and NOHA in the absence of H4B. This result, together with stopped-flow data, enable us to propose a simple, unified mechanism for heme-based oxygen activation, reactivity, and H4B participation in both steps of NO synthesis.

EXPERIMENTAL PROCEDURES
Materials-H4B was purchased from Schirks Laboratory (Jona, Switzerland) and stock solutions prepared in 3 M dithiothreitol (DTT). 2Ј,5Ј-ADP Sepharose 4B was purchased from Alexis Corp. NOHA was a gift from Dr. Bruce King (Wake Forest University, Winston-Salem, NC). All other reagents and materials were obtained from Sigma or from sources reported previously (28).
Expression and Purification of nNOS-Full-length nNOS containing a six-histidine tag at its N terminus was overexpressed in Escherichia coli using the PCWori vector and purified as reported previously (28). The protein as isolated was free of Arg and H4B and was low spin as judged by a Soret peak at 420 nm. The nNOS concentration was estimated based on the absorbance of its ferrous heme-CO adduct as previously reported (28). The nNOS oxygenase domain (nNOSoxy; amino acids 1-720) containing a six-histidine tag at its C terminus was also expressed in E. coli and purified in the absence of H4B as described previously (29).
NO Synthesis and NADPH Oxidation-The initial rate of NO synthesis by nNOS was quantitated at 25°C using the oxyhemoglobin assay for NO (28). nNOS was added to a cuvette containing 40 mM EPPS, pH 7.6, 150 g/ml CaM, 0.62 mM CaCl 2 , 10 units/ml superoxide dismutase, 0.3 mM DTT, 5 mM Arg, 4 M each of FAD and FMN, 100 units/ml catalase, and 10 M oxyhemoglobin. NO-mediated conversion of oxyhemoglobin to methemoglobin was monitored over time at 401 nm and quantitated using a difference extinction coefficient of 38,000 M Ϫ1 cm Ϫ1 . NADPH oxidation at 25°C was quantitated at 340 nm using an extinction coefficient of 6,220 M Ϫ1 cm Ϫ1 . The assay mixture was identical to the NO synthesis measurement except that oxyhemoglobin was absent.
H 2 O 2 -dependent Substrate Oxidation-H 2 O 2 -dependent nNOS oxidation of Arg or NOHA to nitrite was assayed in 96-well microplates at 25°C. Each well contained 40 mM EPPS, pH 7.6, 0.6 M nNOS, 1 mM Arg, 1 mM DTT, 25 units/ml superoxide dismutase, 0.5 mM EDTA, and with or without 100 M H4B. For NOHA oxidation the enzyme concentration was reduced to 0.06 M. 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 using the Griess reagent (100 l) and quantitated based on a nitrite standard curve.
NADPH-dependent Nitrite and Nitrate Synthesis-H4B-free nNOS (0.5 M) or H4B-bound nNOS (50 nM) was added to 96-well microplates containing 40 mM EPPS, pH 7.6, 150 g/ml CaM, 0.62 mM CaCl 2 , 0.3 mM DTT, 1 mM Arg or NOHA, 4 M each of FAD and FMN, 100 units catalase, and 25 units/ml superoxide dismutase to give a final volume 100 l. The reaction was started by adding NADPH to give 1.0 mM and stopped after 25 min by adding excess EDTA. Nitrate reductase (1 M) was added, and the plates were incubated for 2 h at room temperature to reduce nitrate. Excess NADPH was then oxidized by adding lactate dehydrogenase and pyruvate, and total nitrite plus nitrate was detected by adding Griess reagent and reading at 550 nm.
Optical Spectroscopy and Rapid Kinetic Measurements-Optical spectra were recorded on a Hitachi 3010 UV-visible spectrophotometer. Anaerobic spectra were recorded using septum-sealed quartz cuvettes that could be attached through a quick-fit joint to a vacuum system. The nNOS samples were made anaerobic by repeated cycles of evacuation and equilibrated with catalyst-deoxygenated N 2 . A separate buffer solution containing cofactors was evacuated and gassed with N 2 in a separate vessel and then transferred into the anaerobic cuvette. All transfers were made using gas tight syringes. The stopped-flow instrument was attached to a rapid-scanning diode array device (Hi-Tech model MG-6000) designed to collect 96 spectra in a specific time frame. Experiments for H 2 O 2 -dependent reactions involved mixing an anaerobic solution containing 6 M nNOSoxy and 1 mM NOHA with anaerobic buffer solution containing 2 mM H 2 O 2 . In some cases the enzyme solution also contained 10 M H4B. Experiments with O 2 -dependent reactions involved mixing an anaerobic solution of 16 M ferrous nNOSoxy containing 20 M H4B and 2 mM NOHA with aerobic buffer solution.
Ferrous Heme-NO Formation during Steady-state Catalysis-H4Bfree nNOS was diluted to 4 M in air-saturated 40 mM EPPS buffer, pH 7.6, containing 6.0 M CaM, 0.2 mM DTT or ␤-mercaptoethanol, 0.9 mM EDTA, 24 M NADPH, 1.0 mM Arg or NOHA, 0.5 mM glucose 6-phosphate, and 2 units of glucose-6-phosphate dehydrogenase; final volume was 0.5 ml. Reactions were started by adding 2 mM Ca 2ϩ and monitored by wavelength scanning at 25°C in the Hitachi U3010 spectrophotometer. In some cases, substrate was omitted and CO gas was bubbled into the cuvette after initiating catalysis as explained in the text. For stopped-flow rapid scanning experiments, conditions were identical except that an NADPH-regenerating system was not used; the final concentrations of NADPH and nNOS were 40 and 2 M, respectively, and the temperature was 15°C.
Citrulline and NOHA Measurement-Amino acids were derivatized by o-phthalaldehyde (OPA) and separated by reverse-phase HPLC, using an Waters model 510 instrument with a Waters 470 scanning fluorescence detector, Waters 715 Ultra Wisp Sampler Processor, and a Hewlett Packard ODS Hypersil 5-m 100 ϫ 21-mm C18 column (Hewlett Packard 79916 OD-552), equipped with a C18 guard column (2.1 mm, inner diameter, ODS Hypersil, 5 m, Hewlett Packard 79916KT-110). The injector was set to mix 60 l of an OPA reaction solution (4 mg of OPA dissolved in 0.5 ml of methanol to which 4.5 ml of 0.1 M sodium borate, pH 10, and 30 l of ␤-mercaptoethanol were added) with a 40-l reaction sample. After reacting for 2 min, the samples were automatically applied to the column, which was equilibrated with 5 mM ammonium acetate (pH 6.0) containing 20% methanol (v/v) (solvent A) run at 0.5 ml/min at room temperature. The elution conditions for the OPA derivatives were 0 -50% solvent B (methanol) over 9 min, followed by a linear increase to 100% methanol over the next 0.5 min, 100% methanol for 3 min, and a return to 100% solvent A over the next 0.5 min. Amino acid standards were used to quantify the samples. Citrulline, NOHA, and Arg standards had retention times of.6.2, 10.2, and 12.1 min, respectively, and the peaks were completely resolved. Amino acids were detected by fluorescence emission (excitation 360 nm and emission 455 nm) and quantitated based on authentic prepared standards.

NADPH-dependent Citrulline Production by H4B-free
nNOS-H4B-free nNOS generated detectable amounts of citrulline 3 and nitrite plus nitrate from Arg when the reactions contained relatively high concentrations of enzyme ( Fig. 1, Table I). NOHA was also detected as a product, indicating that it forms as an intermediate in the reaction catalyzed by H4Bfree enzyme. When NOHA was used in place of Arg as a substrate it also was converted to citrulline and nitrite plus nitrate by H4B-free nNOS ( Fig. 1, Table I), as shown previously with inducible NOS (26). Citrulline formation from Arg or NOHA was 10 or 15%, respectively, compared with the H4Breplete enzyme 4 in reactions run under otherwise identical conditions (Table I). A somewhat greater proportion of nitrate was generated in reactions catalyzed by the H4B-free enzyme. Control reactions run in the absence of substrate, nNOS, or NADPH did not generate detectable citrulline or nitrite plus nitrate. Including catalase and superoxide dismutase did not diminish product formation in any case. We conclude that H4B is not essential for nNOS to oxidize either Arg or NOHA to citrulline and nitrite plus nitrate in the NADPH-driven reaction. Our results are the first to show that the Arg reaction can occur in an H4B-free NOS. 2 As shown in Fig. 2, NADPH-driven citrulline formation from Arg or NOHA was time-dependent (left panel) and gave initial rates of 2 and 5 min Ϫ1 , respectively.
The reaction was also dependent on nNOS concentration until it reached a point where all NADPH was exhausted within the time of assay (above 500 nM nNOS, right panel). Importantly, citrulline production was associated with no detectable NO synthesis, even under assay conditions where a single NO per heme would have been detected. This implies the H4B-free nNOS formed a nitrogen oxide product other than NO that could oxidize to nitrite and nitrate.
We next determined the stoichiometric relationship between NADPH consumption and citrulline formation by H4B-free nNOS (Fig. 3). Reactions contained different amounts of NADPH, and the total citrulline produced was determined after all the NADPH was consumed. The slopes indicate that conversion of 1 mol of Arg or NOHA to citrulline was associated with oxidation of 16 and 6 mol of NADPH, respectively. For the same nNOS preparation made replete with H4B, the values were approximately 2 NADPH oxidized per citrulline from Arg, and 0.5 NADPH oxidized per citrulline from NOHA (data not shown), which are close to the theoretical minimum values (7). The uncoupling seen under H4B-free conditions was not due to structural changes, because H2B, which mimics all the structural effects of H4B (17,30), did not enhance coupling (data not shown). Our analysis indicates that NADPH oxidation in H4Bfree nNOS is uncoupled from either Arg or NOHA oxidation to citrulline in a multiple turnover setting.
Partitioning of the H4B-free nNOS during Catalysis-To investigate the mechanism of the H4B-independent reaction and identify the nitrogen oxide product, we utilized spectroscopy to observe the enzyme during catalysis. In Fig. 4 spectra of H4Bfree nNOS were recorded before or after initiating its aerobic NADPH oxidation at 25°C in the absence or presence of substrate. Sequential scans were recorded of the CaM-free resting ferric enzyme, after adding NADPH to reduce the flavins, and after adding Ca 2ϩ to trigger CaM binding, heme reduction, and catalysis. In the absence of substrate (panel A), the light absorbance spectrum of the CaM-and H4B-free ferric NOS was a mixture of high and low spin heme with a prominent heme Soret band at 420 nm, consistent with its lack of bound H4B (31). Adding NADPH caused losses in visible absorbance between 360 and 420 nm, 440 and 520 nm, and 560 and 680 nm, and buildup of a broad absorbance centered near 365 nm, consistent with reduction of nNOS flavins. After adding Ca 2ϩ to trigger steady state NADPH oxidation, we observed some decrease in absorbance between 390 and 420 nm and at 650 nm, consistent with heme reduction occurring in the substrateand H4B-free nNOS (15). The inset of panel A also shows significant heme reduction occurred as is evidenced by buildup of a 444-nm ferrous-CO species after the reaction was given CO gas. 5 As expected, the spectrum taken of the substrate-and H4B-free enzyme during steady state NADPH oxidation showed no evidence for buildup of a six-coordinate ferric or ferrous heme-NO complex, which display Soret absorbance bands at 436 and 440 nm, respectively (32). In the presence of Arg or NOHA (Fig. 4, panels B and C), the heme Soret band of the initial H4B-free ferric nNOS was broader with maximum at 398 nm, consistent with substrate binding and greater high 5 When the H4B-and Arg-free enzyme was reduced with dithionite, 95% of its heme formed a 444-nm complex with CO (data not shown). The residual peak at 420 nm in the inset to Fig. 4 therefore represents ferric low spin nNOS. This likely is the monomer that is present when nNOS is purified in the absence of Arg and H4B, whose heme cannot accept NADPH-derived electrons.  spin character (31). The spectra recorded just after initiating steady-state oxidation of Arg or NOHA showed a small buildup of heme-NO complex in both cases (data not shown). The proportion of this species grew over time as the O 2 was consumed in the cuvette. Spectra taken at 10 min (Fig. 4, panels B and C) clearly show the presence of six-coordinate ferrous-NO complex in the Arg and NOHA reactions, as evidenced by the shoulder near 436 nm and single broad absorbance peak near 570 nm (32). The nature of the heme-NO species is further defined by the difference spectra in panels B and C, which show absorbance maxima at 436 and 570 nm. Together, our results suggest that the immediate inorganic product of Arg or NOHA oxidation is nitroxide (NO Ϫ ), which binds to the ferric heme to form a ferrous heme-NO complex. We next examined the kinetics of ferrous-NO complex formation in the H4B-free nNOS and its effect on the NADPH oxidation rate. The left panels of Fig. 5 depict ferrous-NO complex formation during the initial phase of Arg or NOHA oxidation at 15°C. The reactions were started by rapid mixing a solution of Ca 2ϩ with a solution containing CaM, nNOS, substrate, EDTA, and excess NADPH. In both cases, the absorbance increase at 436 nm was best fit to a two-exponential equation, giving apparent rate constants that are listed in Table II. These values indicate that heme-NO complex formation was biphasic and had the same kinetics whether Arg or NOHA serve as substrate in the H4B-free enzyme. The rates obtained for the H4B-free nNOS are similar to the kinetics of heme-NO complex buildup in H4B-saturated nNOS (33, 34), but differ in two ways. First is the magnitude of absorbance gain at 436, which when normalized on a per heme basis indicate that the proportion of enzyme that forms the heme-NO complex during Arg or NOHA oxidation is small for H4B-free nNOS (approximately 10%) compared with H4B-saturated nNOS (approximately 70%; Ref. 33). Second, the relative absorbance change due to the fast phase of complex buildup in the H4B-free enzyme differs from the H4B-saturated enzyme (33,34). As shown in the right panels of Fig. 5, rates of NADPH oxidation were not slowed by ferrous-NO complex buildup in H4B free nNOS, consistent with the small proportion of complex that is observed (34). shows rates of citrulline formation and NO synthesis from Arg at different concentrations of enzyme. Assay conditions were identical to Fig. 1 except these reactions were started by adding 0.2 mM NADPH and were terminated after 5 min by adding 0.6 N hydrochloric acid. Citrulline was measured by a fluorometric HPLC method, and NO synthesis was measured by the oxyhemoglobin assay. Each point is the mean of three measurements.

FIG. 3. Correlation between citrulline formation and NADPH consumption by H4B-free nNOS with Arg (q) and NOHA (E).
Assay mixtures were identical to Fig. 1 except the reactions were started by adding different indicated concentrations of NADPH and the reactions run until all NADPH was consumed. Each point is the mean value of three measurements.

FIG. 4. Spectral traces of H4B-free nNOS before and after initiating NADPH-dependent catalysis.
Traces were recorded sequentially of the ferric enzyme (dashed), following addition of 24 M NADPH, 0.5 mM glucose-6-phosphate, and 2 units of glucose-6-phosphate dehydrogenase (dotted line), and then 10 min after adding excess CaCl 2 to initiate catalysis (solid line). Reactions were run at 25°C either without Arg or NOHA (A), with 1 mM Arg (B), or with 1 mM NOHA (C). The inset of panel A shows the spectrum obtained by adding CO gas to a reaction that was already initiated by excess Ca 2ϩ . The insets of panels B and C show difference spectra generated using the second and third scans recorded for each sample. heme to directly form reactive heme-oxy species. Although this system's ability to convert NOHA to citrulline is well established (35), it was reported not to work with Arg (26, 27) until quite recently (43). Fig. 6 shows that H4B-free nNOS generated detectable amounts of citrulline and NOHA from Arg in a reaction run at 25°C. Nitrite was also detected as a product in the Arg reaction (data not shown). Products were not observed in controls that were missing H 2 O 2 , enzyme, or Arg (data not shown). These results are consistent with H4B-free NOS catalyzing NADPH-dependent Arg oxidation, and suggest a similar mechanism operates in the H 2 O 2 -and NADPH-driven reactions. Table III compares product formation from Arg or NOHA by nNOS in the absence or presence of H4B. Added H4B caused only a 1.8 -2-fold increase in product formation in all cases. Similar small increases with H4B were observed previously in studies that used NOHA as substrate in the H 2 O 2 -driven reaction (35).
To better understand the mechanism, we utilized rapid-scanning stopped-flow spectroscopy to follow heme transitions during H 2 O 2 -driven NOHA oxidation by nNOSoxy under anaerobic conditions (Fig. 7). Ferric nNOSoxy was rapid-mixed with 2 mM H 2 O 2 at 10°C. In both the presence and absence of H4B (panels A and B), a product species with absorbance peak at 436 nm formed immediately. The difference spectra show a trough at 393 nm, peak at 436 nm, and a broad visible peak near 570 nm, identifying it as the ferrous-NO complex. This did not form in the absence of NOHA (panel C). Moreover, its spectrum clearly differs from the ferric-NO complex that forms as an immediate product during H4B-dependent aerobic oxidation of NOHA under single-turnover conditions (36), which displays a difference spectrum with peaks at 442, 550, and 580 nm (panel D). Thus, during H 2 O 2 -driven oxidation of Arg and NOHA the nNOSoxy formed a ferrous heme-NO complex in the presence or absence of H4B. DISCUSSION Despite progress in NOS structure-function (11)(12)(13), the essential role of H4B in catalysis has remained elusive. Our work with H4B-free nNOS shows that it can catalyze Arg oxidation in both NADPH-and H 2 O 2 -supported reactions, without generating any detectable NO. This leads us to conclude the following. 1) H4B is not essential for Arg hydroxylation or NOHA oxidation, although it improves coupling between NADPH oxidation and product formation in both cases. 2) The key function of H4B is to enable generation of NO rather than NO Ϫ . These conclusions are surprising, and their implications are discussed below.
H4B-independent Arg Oxidation-Although H4B-free NOS is known to oxidize NOHA in both NADPH-and H 2 O 2 -driven reactions, Arg hydroxylation in the NADPH-driven reaction was reported to absolutely require H4B (26,27). A discrepancy exists between the old and our new data primarily because we used nNOS at concentrations that were sufficient to clearly detect products even though the reaction is uncoupled with respect to NADPH oxidation. Our data help explain why nNOS maintains a residual Arg to citrulline activity in the presence of H4B antagonists (44). The different NOS isoforms also appear to catalyze H4B-independent reactions with varying efficiency. For example, we have observed Arg hydroxylation and citrulline synthesis with H4B-free iNOS in the NADPH-driven re-

TABLE II
Observed rate constants for ferrous heme-NO complex formation at 15°C Reactions were initiated by rapid mixing the enzyme reaction solutions described below with excess Ca 2ϩ to trigger CaM binding and heme reduction as described under "Experimental Procedures." The rates describe buildup of absorbance at 436 nm versus time and are the average obtained with two nNOS preparations. The data best fit to a two-exponential curve in all cases to generate two rate constants. The percentage of absorbance gain for each phase is indicated in parenthesis.

Experiment
Rate of ferrous-NO complex formation nNOS ϩ Arg ϩ H4B ϩ NADPH 11 (46%) 2.2 (54%) nNOS ϩ Arg ϩ NADPH 10 (75%) 3.0 (25%) nNOS ϩ NOHA ϩ H4B ϩ NADPH 11 (88%) 3.0 (12%) nNOS ϩ NOHA ϩ NADPH 11 (44%) 2.8 (56%) action, but these occur at about one-tenth the rate observed with nNOS. Mechanisms for NOS Arg hydroxylation are modeled after cytochrome P450 monooxygenase chemistry (Fig. 8). Transfer of an electron to the heme enables O 2 binding and formation of a detectable ferrous-dioxy species (I) (20,24,36,37). This species then obtains a second electron to form an iron-peroxo species (II), which decays to yield a perferryl FeO species (III) that is thought to hydroxylate the guanidino nitrogen of Arg. As shown, H 2 O 2 can be an alternative source of two electronreduced O 2 (35) and enables FeO formation without provision of electrons to the heme. H4B cannot directly participate in hydroxylation of Arg (11)(12)(13), and given that H4B is not required for heme reduction (15), it has been puzzling why this cofactor should be essential. Our current results show that H4B is in fact not essential, because the NOS heme hydroxylates Arg even in the total absence of H4B when provided with two NADPH-derived electrons or with H 2 O 2 . This leaves us to consider how NADPH oxidation becomes more coupled to Arg hydroxylation when H4B is bound.
When H4B-free nNOS catalyzes Arg hydroxylation, both electrons needed to generate the FeO species must transfer directly to the heme from the reductase domain. However, in H4B-bound nNOS this is not necessarily the case (Fig. 8). It is clear that the reductase domain gives the first electron to the heme irrespective of bound H4B (15), but the second electron could come either from the reductase domain or from H4B (Fig.  8). Our current work shows that the reductase domain can provide the second electron when H4B is missing, but this leads to uncoupling between NADPH oxidation and product formation. Recent work suggests that bound H4B can also furnish an electron to the ferrous-dioxy complex when the reductase do- main is missing (25). 5 Electron transfer from H4B in this type of experimental system is associated with hydroxylation of between 0.2 and 0.8 mol of Arg per heme (24,25,36,45), 6 implying that electron donation from H4B is at least partly coupled to Arg hydroxylation.
It is important to emphasize that the NOS ferrous-dioxy complex is unreactive toward Arg or NOHA (8,17,24) and will consequently decay to superoxide (which is unstable) and ferric enzyme if a second electron is not provided in a timely manner (Fig. 8). Thus, one way that H4B could improve coupling between NADPH oxidation and product formation is to provide an electron to the ferrous-dioxy species more quickly than can the reductase domain. Indeed, the uncoupling we observe between NADPH oxidation and product formation from Arg in H4B-free nNOS suggests that the reductase domain is not efficient in providing the second electron. Consider that the nNOS reductase domain transfers an electron to the ferric heme at a rate of about 3 s Ϫ1 under anaerobic conditions at 10°C (38), and the rate is not changed in the absence of H4B (17). Thus, heme reduction by the reductase domain is somewhat slower that the estimated rate of H4B radical buildup during reaction of an H4B-bound ferrous iNOS oxygenase with O 2 (11-20 s Ϫ1 ; Ref. 25). 6 This rate of H4B radical formation is also similar to the decay rate of the ferrous-dioxy complex in H4B-bound nNOS oxygenase (10 s Ϫ1 at 10°C), which otherwise decays slower in the absence of H4B (20). Thus, the available kinetic data already suggest that electron transfer from H4B can be connected to ferrous-dioxy reduction, and may be faster than electron transfer from the reductase domain. Once formed, the H4B radical is apparently quite stable in NOS (25), 6 and therefore would have time to be reduced back to H4B by an electron from the reductase domain before the next round of oxygen activation. That H4B can circumvent a kinetic problem in heme reduction during stepwise oxygen activation is an attractive possibility, because it would explain how H4B increases coupling between NADPH oxidation and product formation in NOS.
Citrulline and NO Ϫ Formation from NOHA-Pufahl et al. (35) first reported that H4B-free NOS could generate NO Ϫ from NOHA on the basis of citrulline, nitrite, and nitrate formation, and spectral evidence obtained in an anaerobic H 2 O 2 -supported reaction. A mechanism for NO Ϫ formation is easy to envision (Fig. 9): The iron-peroxo heme nucleophile that is generated with two NADPH-derived electrons or with H 2 O 2 (species II in Fig. 8) reacts with the electrophilic hydroxyguanidine carbon (39) to form a tetrahedral intermediate that breaks down to citrulline, water, ferric enzyme, and NO Ϫ . Our data show that a ferrous heme-NO complex builds up in H4B-free nNOS during NADPH-supported Arg or NOHA oxidation. Because heme-NO complex formation under this circumstance is associated with citrulline, nitrite, and nitrate production but no detectable NO synthesis, it probably occurs by NO Ϫ binding to the ferric heme. NO Ϫ generated from Angelli's salt will bind to ferric hemeproteins to form a ferrous heme-NO complex (40). In H4B-free nNOS, the ferrous heme-NO complex that forms during catalysis represents a much smaller proportion of the total nNOS compared with that observed during NO synthesis by the H4B-bound enzyme (33). This likely reflects differences in capture efficiency due to faster alternative reactions of NO Ϫ compared with NO (41) as well as the uncoupled nature of the H4B-free reaction. In any case, or data reveal that a ferrous heme-NO complex can form during catalysis through two different mechanisms depending on whether H4B is present in the enzyme or not. In H4B-free nNOS, NO Ϫ binds to the ferric heme to directly form the ferrous-heme-NO complex, whereas in H4B-bound nNOS, a ferric heme-NO species forms as an immediate product of catalysis (see panel D of Fig. 7 and Ref. 36) and is then reduced to the ferrous species by the reductase domain.
A New Model for H4B Function-Given that H4B is not required for Arg or NOHA oxidation, we conclude its sole essential function is to ensure that NOS produces NO rather than NO Ϫ . However, how is this accomplished? NO Ϫ is one electron reduced relative to NO; therefore, H4B somehow enables NOS to retrieve an electron from NOHA or a downstream product during catalysis of the second step (4,7,8). This explains why H4B-bound NOS only oxidizes 0.5 NADPH to generate NO from NOHA even though it represents a three-electron oxidation of NOHA. Fig. 9 shows a mechanism for H4B function in the second step of NO synthesis that is consistent with all findings to date. It has H4B donating an electron to the ferrousdioxy species as a central feature, because this generates an H4B radical that acts as an electron acceptor at a later point in the reaction to ensure release of NO instead of NO Ϫ . Electron donation by H4B in the second step is consistent with NOHA oxidation being uncoupled from NADPH oxidation in the H4Bfree enzyme (see Table I and Fig. 3). Although NOHA was initially proposed to reduce the ferrous-dioxy complex instead of H4B (4,7,24), arguments against this possibility exist (42). Moreover, NOHA does not donate an electron to the ferrousdioxy complex of either H4B-free or H2B-bound NOS (8,17,24,25). This finding is particularly damning in light of recent data showing that the structure of a NOS oxygenase dimer containing NOHA and H2B is essentially identical to a dimer containing NOHA and H4B (30). Thus, there is no structural basis for why the enzyme would allow NOHA to reduce the ferrous-dioxy complex only when H4B is present.
A second essential feature of our model is that a reaction product downstream from NOHA donates an electron back to the H4B radical (Fig. 9). Although NOHA in principle could serve as the reductant, recent evidence suggests it cannot be- cause the H4B radical remains long-lived, even when NOHA forms from Arg in the active site of NOS (25). 6 It is presently unclear which of the three downstream products would best reduce the H4B radical. However, a heme-based species is more attractive than free NO Ϫ because it would enable an electron to transfer back to H4B via the heme propionate, which is already implicated in electron transfer from H4B to the heme (13,30).
The tetrahedral intermediate that precedes NO Ϫ formation in Fig. 9 is presently the most attractive candidate, because it ensures NO Ϫ cannot be released, and helps explain why a ferric heme-NO complex is observed as the immediate product during a single turnover reaction by H4B-replete nNOSoxy (36). In any case, this electron transfer step must be relatively fast, because a relatively small amount of H4B radical (approximately 0.25 per heme) accumulates during NOHA oxidation by NOS oxygenase in a single turnover setting (25). The chief advantage of our model is that H4B would function identically in both steps of NO synthesis by acting as a kinetically preferred donor of the second electron to assist in O 2 activation at the heme. The central challenge will be to devise experiments that can isolate or kinetically distinguish the proposed electron donor and acceptor functions of H4B during the second step of the reaction (NOHA oxidation).

CONCLUSIONS
Our work with H4B-free nNOS reveals a simple unified model for heme and H4B function in both steps of NO synthesis. The three essential facets are as follows. 1) Heme-based oxidants that are generated in the NADPH-or H 2 O 2 -driven reaction are apparently identical, and are competent to react with either Arg or NOHA in the absence of H4B. 2) H4B likely functions in both steps of NO synthesis as a kinetically preferred donor of the second electron required for oxygen activation. This enables the enzyme to couple product formation to NADPH oxidation. 3) Electron donation from H4B in the sec-ond step enables it to perform its essential function as an electron acceptor at a later point during NOHA oxidation, thus insuring that the enzyme generates NO instead of NO Ϫ .