Rapid Kinetic Studies Link Tetrahydrobiopterin Radical Formation to Heme-dioxy Reduction and Arginine Hydroxylation in Inducible Nitric-oxide Synthase*

To understand how heme and (6R)-5,6,7,8-tetrahydro-l-biopterin (H4B) participate in nitric-oxide synthesis, we followed ferrous-dioxy heme (FeIIO2) formation and disappearance, H4B radical formation, and Arg hydroxylation during a single catalytic turnover by the inducible nitric-oxide synthase oxygenase domain (iNOSoxy). In all cases, prereduced (ferrous) enzyme was rapidly mixed with an O2-containing buffer to start the reaction. A ferrous-dioxy intermediate formed quickly (53 s−1) and then decayed with concurrent buildup of ferric iNOSoxy. The buildup of the ferrous-dioxy intermediate preceded both H4B radical formation and Arg hydroxylation. However, the rate of ferrous-dioxy decay (12 s−1) was equivalent to the rate of H4B radical formation (11 s−1) and the rate of Arg hydroxylation (9 s−1). Practically all bound H4B was oxidized to a radical during the reaction and was associated with hydroxylation of 0.6 mol of Arg/mol of heme. In dihydrobiopterin-containing iNOSoxy, ferrous-dioxy decay was much slower and was not associated with Arg hydroxylation. These results establish kinetic and quantitative links among ferrous-dioxy disappearance, H4B oxidation, and Arg hydroxylation and suggest a mechanism whereby H4B transfers an electron to the ferrous-dioxy intermediate to enable the formation of a heme-based oxidant that rapidly hydroxylates Arg.

Nitric oxide (NO) 1 is synthesized from L-arginine by the nitric-oxide synthases (NOS) (EC 1.14.13.39). The reaction consumes 1.5 NADPH and 2 O 2 for each NO formed from Arg and is catalyzed in two steps (for review see Refs. [1][2][3]. In the first step Arg undergoes mixed function hydroxylation to form water and N -hydroxy-L-Arg (NOHA) as an enzyme-bound intermediate. NOHA then undergoes mixed function oxidation in the second step to generate water, NO, and citrulline. Both reactions take place within the oxygenase domain of NOS, which contains iron protoporphyrin IX (heme), the cofactor (6R)-5,6,7,8-tetrahydro-L-biopterin (H 4 B), and the Arg binding site (1)(2)(3). Electrons derived from NADPH are provided to the oxygenase domain by an attached reductase domain that binds FMN, FAD, and NADPH (4).
How heme and H 4 B participate in NO synthesis is of wide interest because it involves a novel cooperation between these groups in biologic oxidation. Crystal structures of NOS oxygenase domains (5)(6)(7) show that the guanidinium group of Arg is held directly above the heme, consistent with the heme activating O 2 for substrate oxidation. Accordingly, the mechanisms for Arg hydroxylation in NOS have been modeled after cytochrome P-450 monooxygenase chemistry (Scheme 1). The transfer of an electron to the ferric NOS heme enables O 2 binding and formation of detectable ferrous-dioxy species (I, Fe II O 2 ) (8 -12). This species obtains a second electron to form the iron-peroxo species (II), which upon protonation and O-O bond scission generate water and iron-oxo species (III, FeO) that is thought to hydroxylate the guanidino nitrogen of Arg.
In contrast to close proximity between substrate and heme, the H 4 B cofactor binds away from Arg but next to the heme edge and forms a hydrogen bond between N 3 of its pterin ring and a heme propionate (5)(6)(7). Its position suggests that H 4 B cannot directly participate in O 2 binding or substrate oxidation but could function as an electron donor (6,12). In NOS, the first electron provided to the heme comes directly from the reductase domain, and this transfer does not require that H 4 B be present (13,14). However, either the reductase domain or H 4 B could conceivably provide the second electron to the Fe II O 2 species (I). Indeed, a recent study with the inducible NOS oxygenase domain (iNOSoxy) showed that bound H 4 B is oxidized to a radical when ferrous iNOSoxy reacts with O 2 (15). H 4 B radical formation was associated with some Arg hydroxylation, implying that these two processes might be related. Separate studies have characterized the spectral properties and formation and decay kinetics of the NOS Fe II O 2 complex (8 -12) or the kinetics of product formation from Arg in an NADPH-driven reaction (16). However, what kinetic and quantitative relationships that may exist between H 4 B oxidation, Fe II O 2 formation and disappearance, and product formation remain to be explored.
To address this issue, we combined stopped-flow, rapidquench, and rapid-freeze methods to analyze Arg hydroxylation during a single catalytic turnover by ferrous iNOSoxy. Our results reveal and define the temporal and quantitative links that exist between Fe II O 2 reactivity, H 4 B radical formation, and Arg hydroxylation and thus clarify how H 4 B and heme cooperate in the first step of NO synthesis. * This work was supported by National Institutes of Health Grants GM51491 and CA53914 (to D. J. S.) and GM58481 (to R. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Chemicals-[U-14 C]Arg was purchased from PerkinElmer Life Sciences. Its purity was confirmed by our high pressure liquid chromatography method (see below) and thus was used without further purification. H 4 B and dihydro-L-biopterin (H 2 B) were obtained from Schircks Laboratory (Jona, Switzerland). Dithiothreitol (DTT), Arg, ferricyanide, dithionite, and isopropyl alcohol were obtained from Sigma. 2,4-Dinitrophenyl acetate was purchased from Aldrich. NADPH and isopentane were obtained from Fisher.
Protein Expression and Purification-Mouse iNOSoxy (amino acids 1-498) containing a six-histidine tag at its C terminus was expressed in Escherichia coli BL21 using the PCWori vector and purified in the presence or absence of H 4 B as reported previously (5). Stopped-flow and rapid-quench experiments were performed using three separate protein preparations, whereas rapid-freeze experiments were performed using two separate protein preparations.
Preparation of Ferrous iNOSoxy-Concentrated iNOSoxy was placed in a cuvette and made anaerobic by several cycles of evacuation and purging with deoxygenated N 2 . An anaerobic solution that contained 50 mM HEPES, pH 7.5, H 4 B or H 2 B, and DTT was then added. The enzyme was reduced by the sequential addition of a dithionite solution from which the concentration was standardized against ferricyanide, and the heme reduction was monitored spectrophotometrically. The reduced enzyme solution was transferred into the driver syringe of various instruments using a gas-tight syringe.
Stopped-flow Spectroscopy-Rapid-scanning stopped-flow experiments were performed at 10°C using a HI-TECH SF-61 instrument equipped with a HI-TECH MG-6000 rapid-scanning diode-array detector. An anaerobic solution containing 50 mM HEPES, pH 7.5, 20 M ferrous iNOSoxy, 100 M H 4 B or H 2 B, and 1 mM DTT was rapidly mixed with an equal volume of air-saturated buffer (50 mM HEPES, pH 7.5). Ninety-six scans from 350 to 700 nm were collected within 0.28 s after each mixing. Data from 7 to 10 scans were compiled for global analysis using software provided by the instrument manufacturer as described elsewhere (8).
Rapid-quench Experiments-These experiments were performed in a HI-TECH RQF-63 instrument equipped with a temperature bath. The instrument was calibrated using alkaline hydrolysis of 2,4-dinitrophenyl acetate as described in the manufacturer's manual. The dead time of mixing plus quenching was less than 5 ms. For reaction times shorter than 200 ms, a continuous mixing mode was used, whereas for longer reaction times an interrupt mode (push-push) was used. In all cases, reactions were initiated by a 1:1 mixing event followed by mixing with a quench solution. Final dilutions were determined using a bromphenol blue standard and including the 1:1 mixing event that ranged from 3-fold to 4-fold, depending on the reaction aging time selected. Experimental conditions were as reported by Iwanaga et al. Analysis of Reaction Products-Rapid-quench samples were vortexed for 30 min and then centrifuged at 10,000 ϫ g for 10 min. Vigorous vortexing was essential to completely denature the protein and release all bound amino acids into the supernatant (16). An aliquot (100 l) of each supernatant was injected into a 250/3 Nucleosil 100 -5 SA cation exchange column (Macherey Nagel, Duren, Germany) that was equilibrated with 50 mM sodium acetate, pH 6.5, at a flow rate of 0.5 ml/min. Fractions (0.25 ml) were collected directly into scintillation vials to measure 14 C radioactivity. The elution times for radiolabeled Arg, NOHA, and citrulline were confirmed using authentic non-labeled compounds that were run under the same conditions and detected by thin layer chromatography on silica plates (17).
Rapid-freeze Experiments-These experiments were carried out using the HI-TECH RFQ-63 instrument equipped with a customized channel (HI-TECH) that could bypass the quench step and directly eject the aged reaction samples through a nozzle. A solution containing 50 mM HEPES, pH 7.5, 205 M ferrous iNOSoxy, 20 mM Arg, 0.5 mM H 4 B or H 2 B, and 0.3 mM DTT was mixed with O 2 -saturated HEPES buffer at 10°C. The reactions were aged at 10°C for various times after mixing and then shot into a funnel submerged in an isopentane bath maintained at Ϫ135 to Ϫ140°C. The frozen sample was packed into a 707-SQ EPR tube (Wilmad-Labglass, Buena, NJ) and stored in liquid N 2 until measurement. The dilution of iNOSoxy after mixing under each aging condition was determined using bromphenol blue standards.
EPR Spectra-EPR spectra were recorded in a Bruker ER 300 electron paramagnetic resonance spectrometer equipped with an ER 035 NMR gauss meter and a Hewlett-Packard 5352B microwave frequency counter. Temperature control was achieved using Oxford Instruments ESR 900 continuous-flow liquid helium cryostat and ITC4 temperature controller. All spectra were obtained at 150 K using a microwave power of 2 milliwatts, a frequency of 9.5 GHz, modulation amplitude of 10 G, and modulation frequency of 100 kHz. 20 scans/sample were accumulated to improve the signal to noise ratio. Spin quantitations were calculated by double integration as compared with a 500 M Cu-EDTA standard that was analyzed under the same measurement conditions. Radical concentrations versus time were fit to an "A to B to C" kinetic model (where "B" was the radical) using DeltaGraph software to calculate formation and decay rates.

RESULTS
We monitored a single catalytic turnover in iNOSoxy. A dithionite-reduced (ferrous) iNOSoxy containing Arg and H 4 B was rapidly mixed with an O 2 -containing buffer to initiate the reaction, which was then analyzed by one of three methods: rapid scanning to follow heme transitions, rapid quenching to follow product formation, or rapid freezing to follow buildup of the H 4 B radical. Fig. 1, A-F, contains representative data from each type of experiment.
Rapid scanning discerned three spectrally distinct species during the reaction: the beginning ferrous iNOSoxy, a transient intermediate, and an ending ferric iNOSoxy (Fig. 1A). The transient species had a Soret peak at 427 nm and visible spectral features that identify it as Fe II O 2 iNOSoxy (8 -10). Thus, the formation and decay of this intermediate probably represent the buildup of the Fe II O 2 complex and its subsequent reaction during the single turnover experiment.
Analysis of the kinetic spectral data showed that it only was fit well to an "A to B to C" kinetic model with monophasic transitions for formation and decay of the Fe II O 2 intermediate. This means that no spectrally distinct species were built up during the conversion of the Fe II O 2 intermediate to the ferric enzyme. Fig. 1B illustrates how concentrations of the three spectral species changed with time during the single turnover reaction. The Fe II O 2 intermediate formed rapidly and reached a maximum after 32 ms. Its conversion to the ferric enzyme was essentially completed by 300 ms after mixing. The calculated rates of formation and/or decay of each of the species are listed in Table I. To understand how H 4 B influenced this transformation, we used a H 2 B-saturated iNOSoxy in an otherwise identical stopped-flow experiment. We observed the same three spectral species (data not shown), but in this case conversion of the Fe II O 2 intermediate to the ferric enzyme occurred at a rate of 0.3 s Ϫ1 , which is 40 times slower compared with the H 4 Bcontaining enzyme.
Rapid-quench experiments revealed that H 4 B-containing iNOSoxy oxidized [ 14 C]Arg to NOHA but not to citrulline during the single turnover reaction (Fig. 1C). This conversion was substoichiometric with respect to heme (0.55 Ϯ 0.06 NOHAgenerated/heme, n ϭ 4) even when Arg and H 4 B achieved saturation binding (see "Experimental Procedures"). No NOHA was generated in reactions where H 2 B was substituted for H 4 B or when iNOSoxy was not reduced or was absent (data not shown). The time course of [ 14 C]Arg oxidation and NOHA buildup are illustrated in Fig. 1D. The estimated rates of [ 14 C]Arg disappearance and NOHA formation were essentially SCHEME 1. Model for oxygen activation in NOS.
identical and are listed in Table I. The conversion to NOHA was completed within 0.4 s after starting the reaction. The fact that some [ 14 C]Arg remained after the reaction was completed is consistent with our use of less [ 14 C]Arg than iNOSoxy in this experiment (see "Experimental Procedures") and the known binding affinity of Arg (K s ϭ 2-5 M).
Rapid-freeze experiments showed that a free radical signal (g ϭ 2.0) had built up in iNOSoxy during the single turnover reaction (Fig. 1E). Its spectral characteristics closely matched those of the H 4 B radical reported by Hurshman et al. (15). For example, the radical signal had a peak to trough line width of 40 G and evidence of a hyperfine structure. The free radical signal did not build up in an identical experiment that used H 2 B-saturated iNOSoxy (data not shown). These results confirm that bound H 4 B is oxidized by one electron during the single turnover reaction. The kinetics of H 4 B radical formation and decay are shown in Fig. 1F. Maximum intensity was achieved at about 125 ms where the radical concentration reached approximately 75% total iNOSoxy heme concentration.
Its estimated formation rate was 11 s Ϫ1 and decay rate was 0.7 s Ϫ1 (Table I). These kinetics of H 4 B radical buildup and decay in our system match those observed by Hurshman et al. (15). DISCUSSION We followed Fe II O 2 formation and disappearance, H 4 B radical formation, and Arg hydroxylation during a single catalytic turnover by iNOSoxy to understand their relationships and discern the mechanism of Arg hydroxylation. The kinetic results showed: Because H 4 B oxidation also occurred at the same rate as Arg hydroxylation and without buildup of additional heme-oxy intermediates, all steps required to form the ultimate oxidant and hydroxylate Arg must be as fast as (or faster than) the electron transfer from H 4 B. Thus, Fe II O 2 reduction is ratelimiting when one starts the reaction with ferrous enzyme. A similar situation holds true in many cytochrome P-450 enzymes where steps beyond Fe II O 2 reduction occur faster than the electron transfer to the heme (18,19).
Our results also show how coupled these processes are within the single turnover. For example, practically all bound H 4 B appeared to transfer an electron to the Fe II O 2 intermediate. This finding is evidenced by the high degree of radical buildup (0.75 per heme) despite its concurrent decay and by the Fe II O 2 species disappearing completely according to a monophasic rate. If some unreacted H 4 B had remained in the enzyme or if the transfer between H 4 B and Fe II O 2 was uncoupled, we should have observed less radical buildup and/or biphasic decay of the Fe II O 2 intermediate. A complete well coupled electron transfer between H 4 B and Fe II O 2 is consistent with their close proximity and the irreversible nature of the single turnover reaction.
Given the above results, one would expect that Arg hydroxy- was rapid-mixed with O 2 -containing buffer at 10°C to start the reaction. Subsequent transformations were followed either by stopped-flow rapid-scanning spectroscopy, rapid-quenching and high pressure liquid chromatography analysis, or rapid-freezing and EPR spectroscopy.  lation should match H 4 B radical formation and generate 1 NOHA/heme. However, we and others (8,12,15) typically observe substoichiometric NOHA formation in single turnover reactions ranging from 0.2 to 0.8 NOHA formed per heme. Here the estimated stoichiometry was about 0.6 NOHA formed per heme. Our current work shows that events leading up to and including the reduction of the Fe II O 2 intermediate by H 4 B were tightly coupled and complete. This rules out incomplete electron transfer from H 4 B as a possible explanation and instead implies that subsequent steps (i.e. conversion of the iron-peroxo intermediate to iron-oxo or its reaction with Arg) become uncoupled in the iNOSoxy reaction. Further studies should resolve this issue.
The ability of H 4 B to speed the "decay" of the Fe II O 2 species was first observed while conducting O 2 binding studies with the neuronal NOS oxygenase domain (20) and full-length nNOS (10). Of the multiple effects that H 4 B has on NOS, this one cannot be mimicked by H 2 B and is therefore linked to the tetrahydro reduction state (13). Our results suggest that H 4 B does not speed superoxide release from the Fe II O 2 complex as originally proposed (20). Rather, it reduces the Fe II O 2 complex, thus providing a faster route for its disappearance. In the single turnover reaction, Fe II O 2 "disappearance" represents its reduction and conversion to a heme-oxy species that can hydroxylate Arg very quickly. H 4 B also appears capable of reducing the Fe II O 2 intermediate in the absence of Arg (15,20). In this circumstance, H 2 O 2 could be released from the heme as it occurs in the cytochromes P-450 (18,19) and should be fast compared with superoxide release. This may explain how H 4 B decreases superoxide production in Arg-free NOS despite greatly increasing its rate of NADPH oxidation (21,22).
Because the NOS reductase domain can transfer NADPH electrons directly to the heme, it is puzzling why H 4 B should also be an electron donor. In fact, H 4 B is not essential for Arg hydroxylation, which can be catalyzed by H 4 B-free NOS in an NADPH-driven reaction, albeit being in an uncoupled manner relative to NADPH consumption (14 In the normal NADPH-driven reaction, the transfer of the first electron from the reductase domain is rate-limiting, whereas in the single turnover reaction the reduction of the Fe II O 2 intermediate by H 4 B is rate-limiting. The relative stability of the H 4 B radical should enable it to be reduced back to H 4 B by the reductase domain before it is needed for oxygen activation in the second step of NO synthesis (14).
In conclusion, our work demonstrates the kinetic and quantitative relationships among H 4 B oxidation, Fe II O 2 disappearance, and Arg hydroxylation in NOS. H 4 B appears to be a kinetically competent and complete source of the second electron to reduce the Fe II O 2 intermediate. It will now be important to identify the structural features of the enzyme that control these processes. The electron transfer between H 4 B and the Fe II O 2 intermediate should depend on their relative redox potentials and structural proximity. Crystallography has identified residues that surround H 4 B or stack with the NOS heme (5-7), and mutagenesis studies suggest that some of these residues help modulate H 4 B and heme function (24 -26). The kinetic approach described here should help define how these and other structural features enable the cooperation between heme and H 4 B during oxygen activation in NOS.