A Weak Fe–O Bond in the Oxygenated Complex of the Nitric-oxide Synthase of Staphylococcus aureus*

Little is known about the intermediates formed during catalysis by nitric-oxide synthase (NOS). We report here the characterization by resonance Raman spectroscopy of the oxygenated complex of the NOS from Staphylococcus aureus (saNOS) as well as the kinetics of formation and decay of the complex. An oxygenated complex transiently formed after mixing reduced saNOS with oxygen and decayed to the ferric enzyme with kinetics that were dependent on the substrate l-arginine and the cofactor H4B. The oxygenated complex displayed a Soret absorption band centered at 430 nm. Resonance Raman spectroscopy revealed that it can be described as a ferric superoxide form (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{Fe}^{\mathrm{III}}\mathrm{O}_{2}^{-}\) \end{document}) with a single νO–O mode at 1135 cm–1. In the presence of l-arginine, an additional νO–O mode at 1123 cm–1 was observed, indicating an increased π back-bonding electron donation to the bound oxygen induced by the substrate. With saNOS, this is the first time that the νFe–O mode of a NOS has been observed. The low frequency of this mode, at 517 cm–1, points to an oxygenated complex that differs from that of P450cam. The electronic structure of the oxygenated complex and the effect of l-arginine are discussed in relation to the kinetic properties of saNOS and other NOS.

Little is known about the intermediates formed during catalysis by nitric-oxide synthase (NOS). We report here the characterization by resonance Raman spectroscopy of the oxygenated complex of the NOS from Staphylococcus aureus (saNOS) as well as the kinetics of formation and decay of the complex. An oxygenated complex transiently formed after mixing reduced saNOS with oxygen and decayed to the ferric enzyme with kinetics that were dependent on the substrate L-arginine and the cofactor H 4  Nitric-oxide synthases (NOS) 2 produce nitric oxide (NO) from L-arginine. In mammals, NO is essential to biological processes such as vasodilatation, neurotransmission, and immune responses (1). Mammalian NOSs are homodimeric proteins (2)(3)(4). Each monomer comprises a carboxyl-terminal reductase domain where the FMN and FAD cofactors bind and an amino-terminal oxygenase domain where ironprotoporphyrin IX (heme) and tetrahydrobiopterin (H 4 B), two prosthetic groups essential to the hydroxylation reactions, bind (2)(3)(4).
The two-step reaction involves a first hydroxylation step that converts L-arginine to the N-hydroxy-L-arginine (NOHA) intermediate and a second hydroxylation step that converts NOHA to citrulline and NO (3,5). The active site of NOS consists of a heme coordinated to the sulfur atom of a cysteine amino acid on its proximal side and a substratebinding pocket on the distal side (2). The H 4 B-binding site is in the vicinity of the heme propionates and perpendicular to the heme. In mammalian NOS, H 4 B is required to transfer an electron, and possibly a proton (6), to activate the oxygenated complex and permit L-arginine hydroxylation (4,5,7). Crystal structures of mammalian NOS have shown that the substrate and cofactor H 4 B have extensive Van der Waals-and hydrogen-binding interactions with many residues as well as the heme propionates (2,4).
NOS are present in several bacteria. Four bacterial NOSs from Nocardia sp. (8), Deinococcus radiodurans (drNOS) (9), Bacillus subtilis (bsNOS) (10) and Staphylococcus aureus (saNOS) (11,12) have been characterized so far. Crystal structures are now available for the latter two (12,13). Their overall structures are very similar to those of the oxygenase domain of mammalian NOS, including the substrate-binding site. However, with a shorter amino terminus, bacterial NOS lacks the amino-terminal hook, zinc binding site, a portion of the cofactor binding site, and some of the residues that constitute the dimer interface of mammalian NOS. drNOS, bsNOS, and saNOS produce nitrite in reconstituted enzymatic systems (9,12). In addition, bsNOS has been shown to synthesize NO under single turnover conditions (10), indicating that it is indeed a NOS.
Recent studies have shown that saNOS, unlike mammalian NOS, does not need pterin to stabilize the iron-thiolate bond (11). However, pterins such as H 4 B and tetrahydrofolic acid interact with saNOS as shown by the conservation of the heme vibrational deformation modes in the ferric NO complexes of saNOS, 3 nNOS and iNOS (14,15). Recent results have revealed that drNOS can catalyze the regiospecific nitration of tryptophan (16). Surprisingly, this activity is completely inhibited by H 4 B, suggesting that it occurs at the cofactor site. H 4 B may not be synthesized by bacteria that contain NOS genes (2). Nevertheless, H 4 B and tetrahydrofolic acid can support the catalytic activities of bsNOS and drNOS (9,10). However, the identity of the native cofactor of bacterial NOS remains to be determined.
During catalysis, oxygenated complexes of NOS must be formed (5). The oxygenated complexes (Fe II O 2 ) of nNOS (17)(18)(19)(20), iNOS (21), and eNOS (22,23) have been characterized by optical absorption spectroscopy. At least two types of oxygenated complexes seem to exist: heme-oxyI and heme-oxyII (22). Heme-oxyII has a Soret absorption band near 430 nm and an intact O-O bond, as shown by resonance Raman spectroscopy (24). Heme-oxyI has a Soret absorption band near 420 nm (22) and was identified as an oxygenated complex based on the similarity of its magnetic circular dichroism spectrum to that of P450 cam (25). In eNOS, heme-oxyI is formed with H 4 B present but not the substrate. Heme-oxyII is formed with the substrate present or when both the substrate and H 4 B are absent (22). bsNOS and drNOS both form oxy-* This work was supported by the National Sciences and Engineering Research Council of Canada (Grant 250073), the Canadian Foundation for Innovation (equipment Grant 7128), and the Fonds Québé cois de la Recherche sur la Nature et les Technologies (Grant 78927 to M. C.). 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. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. S1-S4. 1  genated complexes with a Soret absorption band at 427 nm similar to the heme-oxyII complex of mammalian NOS (9,10). Although it is known that heme-oxyII is an oxygenated complex, its electronic configuration is not completely understood. In particular, the Fe-O bond has never been previously observed in NOS. We report here the characterization of the oxygenated complex of saNOS by resonance Raman spectroscopy, which provides important information about the electronic properties of this complex formed during catalysis by NOS.

EXPERIMENTAL PROCEDURES
Materials-H 4 B, NOHA, and D 2 O (99.9% purity) were purchased from Sigma-Aldrich. L-Arginine was from Alexis Biochemical (San Diego, CA). Argon and 16 O 2 gas were from Praxair (Mississauga, Ontario, Canada). 18 O 2 gas (99% purity) was from Icon (Mt. Marion, NY). Enzyme Preparation-saNOS was expressed in Escherichia coli from the cloned gene and was purified as described previously (11). Samples were maintained in 40 mM HEPES (pH 7.6), 150 mM NaCl and 1 mM dithiothreitol. Where indicated, L-arginine (1 mM or 100 M), NOHA (500 M), and citrulline (500 M) were added to the purified enzyme. For samples containing H 4 B, a concentration of 500 M was used. The association of cofactor H 4 B and substrates with saNOS were monitored by the displacement of dithiothreitol bound to the ferric enzyme using optical spectroscopy (11).
Stopped-flow Spectroscopy-Stopped-flow experiments were carried out with an Applied Photophysics (Leatherhead, UK) SX.18MV stopped-flow spectrometer installed inside an MBraun (Stratham, NH) LabMaster 100 glovebox where the oxygen concentration was maintained below 4 ppm. The dead time of the stopped-flow apparatus was 1.5 ms with the 10-mm path length cell. Anaerobic protein samples (5 M) were reduced with a 1.25 molar ratio of sodium dithionite over heme. Rapid mixing experiments with reduced saNOS and oxygen were carried out at 21°C. The O 2 concentration used was 10% except for the experiment with NOHA described below, for which 100% O 2 was used.
The kinetics of formation and decay of the oxygenated complex were followed at individual wavelengths in kinetic scanning mode. Kinetic traces were recorded at 5 nm intervals, from 380 to 450 nm, which generated optical spectra versus time data sets. The kinetic data were analyzed using the Specfit global analysis software (Spectrum Software Associates, Chapel Hill, NC) with a kinetic model involving three species: species A corresponded to the initial reduced saNOS, species B corresponded to the transient oxygenated complex, and species C corresponded to the final ferric form. From these analyses, the deconvoluted optical spectra, the fit at all wavelengths and the time course of appearance and decay of the three kinetic species were obtained.
The decay of the oxygenated complex to the ferric enzyme followed single exponential kinetics for some conditions. In other conditions, two exponential functions were required to fit the data (Table 1). When two exponential rates were apparent, the rates were obtained from single wavelength kinetic traces recorded over a longer time period than the kinetic scanning experiments to ensure an accurate determination of the slow reaction rate.
For single turnover experiments with NOHA, a reduced saNOS sample with 500 M NOHA and 500 M H 4 B was mixed with oxygensaturated buffer (100%), and the kinetic traces were recorded in kinetic scanning mode from 380 to 460 nm and from 510 to 610 nm at 5 nm intervals. The kinetic model included a fourth species, which corresponded to the Fe III NO complex formed after the oxygenated complex and before the ferric resting protein.
Construction of a Rapid Mixer-The continuous-flow rapid mixer was built according to the specifications of the rapid "T" mixer described by Takashi et al. (26). Briefly, two solutions were injected in the upper part of the "T" mixer. The solutions were injected using glass syringes connected to two PEEK tubes, which were in turn attached to Nanoport assemblies (Upchurch Scientific Inc., Oak Harbor, WA). The solutions were forced into the opposite ends of a 100-m-wide, 25-mdeep channel etched into the upper part of the "T" (mixer) (Custom Precision Technology, Cokato, MN). They were mixed when they engaged into a 5-cm-long quartz tube with a 250 ϫ 250-m-wide internal channel (Friedrich & Dimmock Inc., Milville, NJ). Two modifications to the design of Takashi et al. were made. First, a piece of glass was used in place of the metal plate forming the upper part of the "T" mixer. Second, a thicker Teflon sheet (250 m) containing a 250-M-diameter hole was used to seal the upper part of the mixer and the quartz tube. The position of the assembled mixer with respect to the exciting light source was controlled using a digital micrometer with a resolution of one micron (Newport Instruments, Mississauga, Ontario, Canada).
The efficiency of mixing was determined by the diffusion limited acid-base reaction between Na 2 HPO 4 and HCl followed by Raman spectroscopy (27). With this method, at the first observable point after mixing, 93% of HPO 4 2Ϫ was converted into H 2 PO 4 Ϫ with a flow rate of 3 ml/min Ϫ1 (supplemental data, Fig. S1). The efficiency of mixing decreased to 92 and 88% at 2 and 1 ml/min, respectively. A 3 ml/min flow rate was used thereafter. The theoretical dead time of the mixer was 0.28 ms. The actual dead time determined from the fluorescence signal generated by the formation of a hydroxyquinoline . Mg 2ϩ complex (28), was 0.49 Ϯ 0.04 ms (supplemental data, Fig. S2). Resonance Raman Spectroscopy with the Continuous-flow Mixer-Prior to mixing, the reduced form of saNOS (80 M) was prepared by equilibrating the enzyme with pure argon gas for at least 30 min at room temperature and by adding 1.25 molar equivalents of sodium dithionite. The concentration of the sodium dithionite stock solution was determined by the reduction of a small aliquot of the anaerobic saNOS sample. Complete reduction was assessed by optical spectroscopy. Air-saturated buffer was used for mixing experiments with 16 O 2 . Buffer with 20% 18 O 2 was prepared from buffer equilibrated with one atmosphere of 18 O 2 (2 ml) and diluted to 10 ml with anaerobic buffer. For the experiments with NOHA, buffers saturated with 100% 16 O 2 and 18 O 2 were used.
The rapid mixer was made anaerobic with a solution of 10 mM sodium dithionite followed by washing with anaerobic buffer to remove the excess dithionite. Two 10-ml syringes filled with oxygenated buffer and the 80 M reduced saNOS solution, respectively, were used. The output at 441.6 nm from a He/Cd laser (Liconix laser, Melles-Griot, Ottawa, Ontario, Canada), at ϳ12 milliwatts was focused on the sample inside the channel of the quartz cell. The resonance Raman spectra were acquired using previously described equipment (11) at 4 ms to 100 ms after mixing, depending on the sample. Several 30-s spectra (12 to 13) were acquired and averaged. All resonance Raman spectra were obtained at room temperature (25°C). The resonance Raman spectra were calibrated with the lines of indene. The resonance Raman spectrum of reduced myoglobin was recorded before each experiment to check for small differences between calibrations on different days. Equilibrium resonance Raman spectra were obtained as previously described (11).
nm and the fit to the three species kinetic model are shown in Fig. 1A. These wavelengths show the largest variations of absorbance. The good fit of the data to the kinetic model was shown by the very small amplitude of the residuals at all wavelengths (Ͻ0.0015 ⌬A) over the course of the reaction (Fig. 1D).
The deconvoluted absorption spectra of the reduced, oxygenated, and oxidized species were obtained (Fig. 1B). Reduced saNOS (solid line) displayed an absorption maximum at 412 nm, just like the spectrum of the starting material before mixing (11). The transiently formed oxygenated species (dashed/dotted trace) displayed a Soret absorption maximum at 430 nm, whereas the oxidized form (dotted trace), which is the species obtained at the end of the kinetic reaction, displayed a Soret absorption band centered at 398 nm. Because L-arginine was bound at the active site, the ferric spectrum was that of the five-coordinate, high spin form (11).
The variation in concentration of the three species as a function of time is shown in Fig. 1C. The oxygenated complex reached a maximal concentration of 2.4 M (96% of the total heme concentration) 15-30 ms after mixing. The rate of formation of the oxygenated complex was 106 s Ϫ1 (Table 1). Subsequently, the oxygenated species was converted to the oxidized form, with rate constants of 5.6 s Ϫ1 and 1 s Ϫ1 , respectively (Table 1).
In the absence of substrate, the appearance and decay of the oxygenated complex were faster, and the complex reached a concentration of 2.3 M 2.1 ms after mixing (supplemental data, Fig. S3). The rate of formation of the oxygenated complex was 1880 s Ϫ1 (Table 1). This complex decayed with a single rate constant of 39.6 s Ϫ1 ( Table 1).
The addition of 500 M H 4 B did not affect the rate of formation of the oxygenated complex, which remained at 2080 s Ϫ1 (Table 1). With both L-arginine and H 4 B, the rate of formation of the oxygenated complex (143 s Ϫ1 ) was similar to that with L-arginine alone (106 s Ϫ1 ). Although H 4 B had only a small effect on the rate of formation of the oxygenated complex, it increased the rate of decay in the presence of L-arginine, with rates of 18 and 5 s Ϫ1 , respectively, compared with rates of 5.6 and  1.0 s Ϫ1 with L-arginine alone ( Table 1). In the absence of L-arginine, H 4 B slowed the decay from a single rate of 39.6 s Ϫ1 to two rates of 32.8 and 3.6 s Ϫ1 , respectively (Table 1).

Rates of formation and decay of the oxygenated complexes of saNOS, drNOS, and bsNOS and the wavelength maxima of the Soret absorption band of the oxygenated complexes
Optical Spectra of the Oxygenated Complexes-For every condition, an oxygenated species with a Soret absorption band centered at 430 nm was formed (Table 1). For samples with no H 4 B, this was the only oxygenated complex detected. For samples containing H 4 B without L-arginine, the data did not fit the simple three-species model well, pointing to a kinetic model with a fourth species. A species with a very short half-life (5 ms) that displayed a Soret absorption maximum near 425 nm was transiently detected before it was converted at a rate of 89 s Ϫ1 to the oxygenated species with the Soret absorption maximum at 430 nm ( Table 1). The formation and decay of the two oxygenated species overlapped. The resonance Raman spectrum with H 4 B alone (see below) was thus obtained from a mixture of a majority of the 430 nm species (75%), some 425 nm species (23%) and 2% of the ferric form. Because the excitation wavelength was 442 nm, only the 430 nm species was characterized by resonance Raman spectroscopy.
For the sample containing both L-arginine and H 4 B, the kinetic model included three species. The deconvoluted oxygenated spectrum had a wide Soret absorption band consisting of two peaks centered approximately at 405 and 430 nm (Table 1). These species could not be resolved by including a fourth species in the kinetic model. Either both oxygenated species formed over the same time scale or a fraction of one species was converted to the other very rapidly before the slow decay to the ferric form. As the laser used for resonance Raman spectroscopy emitted light at 442 nm, only the species with the Soret absorption band at ϳ430 nm was characterized here.
Resonance Raman Spectra of the Oxygenated Complex of saNOS in the High Frequency Region-The resonance Raman spectra of the oxygenated form of saNOS were obtained using a home-built continuousflow mixer apparatus that had a dead time of 0.49 ms (supplemental data, Fig. S2). The high frequency region of the resonance Raman spectra was first acquired to verify that the oxygenated complex was formed. Reduced saNOS, without L-arginine and H 4 B, displayed a 4 line at 1349 cm Ϫ1 and a strong 3 line at 1467 cm Ϫ1 (Fig. 2C). The oxygenated spectrum was recorded 4 ms after mixing. At this time, 90% of the protein was complexed with oxygen based on the kinetics determined by stopped-flow spectroscopy. The oxygenated complex ( Fig. 2A) displayed 4 and 3 lines at 1373 and 1500 cm Ϫ1 , respectively. The frequency of the electron density marker line 4 indicated that the heme iron was ferric, like the oxygenated complex of nNOS ox (24) and P450 cam (29). The frequency of the 3 line was consistent with the presence of a ferric six-coordinate low spin complex.
The oxygenated spectrum ( Fig. 2A), differed from that of the ferric form (Fig. 2B), with a line of higher intensity at 1133 cm Ϫ1 and the absence of a shoulder at 1360 cm Ϫ1 on the left side of 4 at 1373 cm Ϫ1 . The resonance Raman spectrum of oxygenated saNOS displayed a line at 1467 cm Ϫ1 that corresponded to 3 of the reduced enzyme (11) (Fig.  1C). The 4 line of reduced saNOS at 1349 cm Ϫ1 was, however, absent from the spectrum of the oxygenated complex. Together, the low intensity of the 3 line at 1467 cm Ϫ1 and the absence of the 4 line at 1349 cm Ϫ1 indicated that the portion of saNOS that did not react with oxygen, or from which oxygen was photodissociated, was small.
Identification of the O-O Mode of the Oxygenated Complex of saNOS-To identify the vibrational modes that involved the oxygen molecule, the spectra of the oxygenated complex of saNOS were obtained with 16 O 2 and 18 O 2 . The resonance Raman spectrum of the oxygenated complex of L-arginine-free and pterin-free saNOS was obtained with 16 O 2 (Fig. 3A). The background signal from the quartz flow cell, shown in Fig. 3F, was acquired and subtracted from the raw spectrum shown in Fig. 3A to obtain the spectrum shown in Fig. 3B. The spectra C-E were also corrected to remove the quartz signal. An isotopic shift from 1135 cm Ϫ1 to 1071 cm Ϫ1 was observed with 16 O 2 and 18 O 2 , respectively (Fig. 3, B and C). The 16 O 2 minus 18 O 2 difference spectrum shows this shift more clearly (Fig. 4A) (Fig. 3D) (Fig. 3, B-E). This line was assigned to the in-plane heme vibrational mode 7 . An increase in the intensity of the line at 693 cm Ϫ1 was also observed. This line was assigned to the ␥ 15 out-of-plane mode of the heme compared with Mb and iNOSox (14,30).
The O-O lines were analyzed by spectral deconvolution. In the absence of L-arginine and pterin, the O-O lines were fitted to single Gaussian peaks of similar intensity and width at 1135 and 1071 cm Ϫ1 , respectively (Fig. 5A). In all other samples, the lines were best fitted to two peaks of mixed Gaussian and Laurentzian character. H 4 B alone had a small effect on the O-O frequency as seen by the appearance of small intensity vibrational modes at 1123 and 1062 cm Ϫ1 for the 16 (Fig. 5D and Table 2).
We observed an asymmetry in the surface area under the lines for the positive lines with 16 O 2 and the negative lines with 18 O 2 in the presence of L-arginine (Fig. 5, C and D). The fact that the area under the lines of the 1123/1135 cm Ϫ1 and 1062/1071 cm Ϫ1 lines was not reciprocal may be due to Fermi resonance caused by a heme vibrational mode near 1133 cm Ϫ1 overlapping the O-O mode at 1135 cm Ϫ1 (Fig. 3).
Identification of the Fe-O Mode of saNOS-The 16 O 2 minus 18 O 2 differential spectra of oxygenated saNOS presented in Fig. 4 (A-E) revealed the presence of a second isotope-sensitive vibrational mode. The line at 517 cm Ϫ1 in the 16 (Table 2). Unlike the O-O mode, the Fe-O mode was unaffected by the addition of L-arginine and H 4 B (Fig. 4, A-E).
To determine whether the heme-bond dioxygen was involved in hydrogen-bond interactions, the resonance Raman spectra of the oxygenated complexes of substrate-free and pterin-free saNOS were obtained in H 2 O and D 2 O. As shown in Fig. 4F, the H 2 O minus D 2 O difference spectrum displayed no shift of vibrational modes.
To detect the ␦ Fe-O-O mode, which is observed at 402 cm Ϫ1 for p450 cam (32), we obtained the 200-to 900-cm Ϫ1 region of the resonance Raman spectra of oxygenated saNOS in complex with L-arginine (supplemental data, Fig. S4). No isotope-sensitive vibrational mode other than the Fe-O mode could be observed.

Synthesis of NO under Single Turnover Conditions-
The hydroxylation of NOHA by saNOS containing H 4 B was followed by stopped-flow spectroscopy under single turnover condition. Fig. 6 shows the results of the global analysis of the kinetic scanning data in the Soret and visible regions of the spectra. The reduced species, with a Soret absorption maximum centered at 413 nm (Fig. 6C, solid line) rapidly bound dioxygen at a rate of 454 s Ϫ1 and formed the oxygenated complex with a Soret absorption band centered at 428 nm (Fig. 6C, dashed trace). The oxygenated complex was then converted to the Fe III NO complex (Fig. 6C, dotted line) that reached a maximal concentration 100 ms after mixing (Fig. 6E). The rate at which the oxygenated complex was converted to the Fe III NO complex was 22.2 s Ϫ1 ( Table 1).
The rapid formation of the Fe III NO complex was clearly observed at 445 nm with a kinetic trace shifted to a longer time scale with respect to the kinetic trace at 430 nm that monitored the formation of the oxygenated complex (Fig. 6B). The Fe III NO complex was identified by the Soret  absorption band centered at 438 nm (Fig. 6C, dotted trace) and maxima at 550 and 580 nm in the visible region (Fig. 6D, dotted trace). 3 The Fe III NO complex decayed to the ferric enzyme with the Soret absorption maximum at 398 nm (Fig. 6C, dotted/dashed trace) over 1 s with a rate constant of 3.9 s Ϫ1 (Fig. 6E, dotted/dashed trace) ( Table 1). Fig. 6A shows the raw optical spectra highlighting the spectrum at 3 ms (solid line), a spectrum at 13 ms indicating the formation of the oxygenated complex (dashed line), a spectrum at 53 ms showing a shift to longer wavelengths indicating the formation of the Fe III NO complex (dotted line), and the final ferric spectrum (dotted/dashed line).
To confirm the formation of the Fe III NO complex, the resonance Raman spectrum was acquired 100 ms after mixing ferrous, H 4 B, and NOHA saturated saNOS, with oxygen under single turnover condition. Stopped-flow results indicate that at 100 ms after mixing, the Fe III NO complex had reached a maximum concentration of 1.7 M (69% of the heme total). The remaining fractions were the oxygenated (11%) and ferric forms (20%). The resonance spectrum recorded 100 ms after mix-ing displayed an intense vibrational mode at 547 cm Ϫ1 (Fig. 7E) that we assigned to a ␦ Fe-N-O mode based on the comparison of this mode with that displayed by equilibrium Fe III NO forms of saNOS with and without substrates and cofactor. 3 It must be pointed out that the NO synthesized by saNOS cannot be labeled using 18 O 2 , because the oxygen atom of NO comes from NOHA itself.
We emphasize that ferric (Fig. 7G) and ferrous forms (Fig. 7H) of saNOS do not display a resonance Raman line at 547 cm Ϫ1 . Also consistent with the assignment of the 547 cm Ϫ1 line as a originating from the Fe III NO complex is the spectrum recorded at 5 ms after mixing (Fig.  7F), that displays a less intense line at 547 cm Ϫ1 as compared with the spectrum obtained at 100 ms after mixing (Fig. 7E). Based on the stopped-flow analysis (Fig. 6E), at 5 ms after mixing, 7% of the heme content was in the Fe III NO form, 83% in the oxygenated form, and 10% in the reduced form. Accordingly, the spectrum at 5 ms displays 16

7, F and H).
The relative contribution of each species to the resonance Raman spectra could not be further quantified, because the laser used to obtain the resonance Raman spectra enhanced each species differently. The Fe III NO form was the species enhanced preferentially due to the wavelength maximum of the Soret absorption band (440 nm) close to the frequency of the laser (441.6 nm), and the ferric and reduced forms were the less preferentially enhanced forms. The frequency of the ␦ Fe-N-O mode of the Fe III NO complex obtained at 100 ms was high at 547 cm Ϫ1 (Fig. 7E) and similar to equilibrium Fe III NO forms of saNOS (550 cm Ϫ1 ) 3 , saNOS/H 4 B (551 cm Ϫ1 ), 3 and saNOS/H 4 B/citrulline (549 cm Ϫ1 , Fig. 7, A-C) and different from that of saNOS/H 4 B in complex with NOHA (543 cm Ϫ1 , Fig. 7D) indicating that NOHA no longer occupied the active site of saNOS proteins that had synthesized NO.

Effect of L-Arginine and H 4 B on the Rate of Formation of the Oxygenated Complex-
The rate of oxygenation of saNOS was very fast at 1880 s Ϫ1 and did not change much with the addition of H 4 B. In contrast, L-arginine slowed the rate of formation of the oxygenated complex Ͼ17-fold to 106 s Ϫ1 (Table 1). A reduced rate of oxygenation with L-arginine has been observed with eNOS, with a ϳ50-fold reduction (22), but not with nNOS (17), reflecting differences among the NOS isoforms.
The kinetic data obtained in the presence of L-arginine alone and in the presence of both L-arginine and H 4 B were similar to those obtained by Adak et al. for drNOS (9) and bsNOS (10) ( Table 1). The rates of oxygenation in the range 60 to 143 s Ϫ1 are similar for all three bacterial NOS.

Effect of L-arginine and H 4 B on the Rate of Decay of the Oxygenated
Complex-The comparison of the autoxidation rates of L-arginine-free and pterin-free saNOS with the sample containing L-arginine shows that L-arginine slowed the autoxidation rate of the Fe III O 2 Ϫ complex 12to 40-fold. A stabilization of the oxygenated complex by L-arginine has also been observed with nNOSox (17,18) and eNOSox (22,33). For eNOSox, the slower rate of autoxidation with L-arginine has been suggested as favoring coupling with the electron transfer from H 4 B to activate the oxygen before the release of superoxide occurs (33). The fast rate of formation and autoxidation of the Fe III O 2 Ϫ complex displayed by saNOS may result in a significant release of superoxide, with potent deleterious effects. In fact, Berka et al. (33) reported that superoxide is generated at a level of 0.3 spin/heme with saNOS in the absence of L-arginine and H 4 B. This is comparable with the 0.5 spin/ heme produced by eNOSox under the same conditions (33). L-Arginine helps by slowing the autoxidation rate. NO Synthesis-The three mammalian NOS (6,35,36), as well as the bacterial NOS from Bacillus subtilis (10), synthesize NO in single turnover reactions. The hydroxylation of NOHA by saNOS under single turnover conditions described here shows that saNOS was able to couple electron transfer from H 4 B to oxygen activation and catalyze the NO synthesis reaction. NO was detected as a complex with the ferric enzyme based on the typical optical spectrum of the Fe III NO complex observed transiently during the course of the reaction. The Fe III NO complex appeared after the formation of the oxygenated complex and before the formation of the ferric, ligand-free enzyme. This sequence of events is exactly the same as that observed for iNOS (36), nNOS (35), and bsNOS (10) in single turnover reactions monitored by stopped-flow spectroscopy. The synthesis of NO by saNOS was confirmed by the observation of the ␦ Fe-N-O mode of a Fe III NO complex at 547 cm Ϫ1 100 ms after mixing ferrous saNOs/H 4 B/NOHA with oxygen.
Identification of the Oxygenated Complex of saNOS-In all conditions tested, an oxygenated spectrum with a wavelength maximum of the Soret absorption band at 430 nm was observed. This oxygenated species corresponded to heme-oxyII as defined by Marchal et al. (22). The high frequency region of the resonance Raman spectrum of the oxygenated complex of saNOS indicates that the heme iron is ferric with a 4 line at 1373 cm Ϫ1 (24,29). The complex is low spin as shown by the 3 line at 1500 cm Ϫ1 . The frequency of the O-O mode at 1135 cm Ϫ1 was far from the 1555-cm Ϫ1 value expected for a neutral O 2 but in the same range as that of superoxide complexed with metals (37). This indicates that the optical spectrum of the oxygenated complex of saNOS with a Soret absorption maximum centered at 430 nm corresponded to a heme-oxyII complex with a ferric superoxide (Fe III O 2 Ϫ ) character. The oxygen-   (Fig. 6E). A contribution from the reduced form (trace H) to the 100-ms spectrum was excluded as there is no strong line at 744 cm Ϫ1 in trace E, and the results from the stopped-flow analysis revealed that no reduced form remained 100 ms after mixing (Fig. 6E).
ated complex of saNOS was thus very similar to the heme-oxyII complex of nNOSox, which is six-coordinate, displays a Fe III O 2 Ϫ character, and has a Soret absorption band at 427 nm (24). An additional species with a blue-shifted Soret absorption maximum at 425 nm was observed with H 4 B alone and a species with a Soret absorption maximum at ϳ405 nm was observed with H 4 B and L-arginine combined. These species differ from the heme-oxyI species described by Marchal et al. (22), which has a Soret absorption maximum at ϳ420 nm. The 425 nm and the 405 nm intermediates of saNOS were not further characterized here.
The  (39), had a Fe-CO mode in the absence of L-arginine at a frequency similar to that observed in the presence of L-arginine. This observation led to the proposal that a residue or water molecule of the heme pocket may be responsible for the interaction with CO.
The adoption of the conformation with the O-O mode at 1123 cm Ϫ1 seemed further enhanced by H 4 B in combination with L-arginine. The effect was either additive or might be caused by small changes in the conformation of the enzyme that lead to changes in the heme pocket, which in turn favor the formation of the hydrogen bound between the substrate and the heme-bound dioxygen molecule. This effect could occur through the hydrogen bond interaction that both molecules form with the heme propionate-7. 3 In the mammalian nNOSox, binding of L-arginine causes the splitting of the O-O mode at 1135 cm Ϫ1 to two vibrational modes at ϳ1122 and 1135 cm Ϫ1 , respectively (40), which is similar to our observations with saNOS. However, the fraction of nNOSox that adopts the new conformation is much lower than with saNOS, because the majority of the O-O line remains at 1135 cm Ϫ1 . We conclude that the response to L-arginine binding by saNOS and nNOSox is similar, the difference being the amount of the form with O-O at 1123 cm Ϫ1 that is present when L-arginine is bound to the proteins.
Low Frequency of the Fe-O Stretching Mode-Compared with P450 cam (29,38) and thiolate model compounds (29), the frequency of the Fe-O mode of saNOS at 517 cm Ϫ1 was low ( Table 2). Several possibilities were considered to explain this observation.
We considered the possibility that a highly bent Fe-O-O structure would make the Fe-OO mode appear at a low frequency. A highly bent structure, such as that displayed by heme oxygenase, would behave as an Fe-O diatomic oscillator (41). This should lead to a larger 18 O 2 isotopic shift of the Fe-O mode than a linear structure, which would behave like an Fe-OO oscillator (41). The isotopic shift calculated for saNOS (30 cm Ϫ1 ) was indeed large. It was in fact larger than the expected shift for an Fe-O, two-body oscillator (23 cm Ϫ1 ). However, the oxygenated complex of P450 cam also displays large isotopic shifts (27-30 cm Ϫ1 , Table 2), and the crystal structures of the oxygenated complex show a normal Fe-O-O bending angle (129 -133°) for a heme protein (42,43).
Hydrogen bond interactions with the heme-bound proximal oxygen reduce the frequency of the Fe-O mode of some globins (44 -46). The resonance Raman spectrum of the oxygenated complex of saNOS in D 2 O showed no shift of vibrational modes with respect to the spectrum in H 2 O. This indicates that a hydrogen bond interaction with a water molecule is unlikely. However this result does not totally rule out a hydrogen bond with an amino acid residue if the proton involved is not exchangeable. However, hydrogen bond interactions with an amino acid would require a rearrangement of the heme pocket, because there are no residues within 6 Å of the heme-Fe in saNOS (12). A hydrogen bond with the proximal oxygen would possibly favor protonation of the proximal oxygen and thus promote synthesis of peroxide and uncoupling. A hydrogen bond to the proximal oxygen thus seems unlikely.
Alternatively, we looked at electronic effects on the Fe-O-O complex. There is a well established negative correlation of the Fe-XO and X-O modes for the Fe II NO and Fe II CO complexes of heme proteins (31,47). This negative correlation holds true for 5-coordinate Fe II O 2 complexes but may not hold for six-coordinate Fe II O 2 complexes of thiolatecoordinated compounds. Oxygen has two electrons that can be accommodated in a * nonbonding lone-pair orbital that bends the Fe-O-O unit or in the bounding orbital that retains a more linear conformation (31). Depending on the thiolate hybridization state, competition for the Fe d z 2 orbital by the thiolate may be strong, forcing the two electrons of the O 2 in the * antibonding orbital (38). This would bend the    (31) and 5C thiolate models (filled circles) (31). The line is the best linear fit to the set of frequencies for the 5C model compounds.
observed for six-coordinate thiolate-coordinated model compounds (see Table 2 of Vogel et al. (31)), P450 cam and the P450 cam D251Nputaredoxin complex (29,38) (Fig. 8). The frequencies of saNOS fit with the positive correlation between the Fe-O and O-O frequencies for 6C complexes with a thiolate bond (Fig. 8).
The low frequency of the Fe-O mode of saNOS may thus be explained by strong competition for the Fe d z 2 orbital by the thiolate, which would result in a weakening of the Fe-O bond, whereas the lower frequency of the O-O mode of saNOS with respect to P450 cam may be explained by the extent of back-bonding to the heme-bound dioxygen. It must be pointed out that the frequencies of the O-O mode of nNOSox with L-arginine at 1135 cm Ϫ1 and 1122 cm Ϫ1 (24) are nearly identical to those of saNOS with L-arginine (Table 2). It will be interesting to see whether mammalian NOS also display a week Fe-O bond.
Although L-arginine caused a shift to a lower frequency of the O-O mode, indicating an increase in back-bonding, no corresponding shift of the Fe-OO mode was observed. The hydrogen bond between L-arginine and the superoxide, or the indirect structural stabilizing effect of L-arginine, increases the extent of back-bonding, thereby lowering the frequency of the O-O bond. If only back-bonding is considered, this should make the Fe-O bond stronger. Because the Fe-O mode did not change in frequency, this indicates that competition by the thiolate increased with L-arginine present and was thus coordinated with the change in the amount of electron donation. Importantly, our results show that the reduction of the rate of autoxidation of the oxygenated complex by L-arginine did not arise from the strengthening of the Fe-O bond per se, because its frequency did not change with or without L-arginine.
Implications for Catalysis-The electronic structure displayed by saNOS with a weak Fe-O bond and a strong superoxide character may come at the price of a high rate of autoxidation of the oxygenated complex. The oxygenated complexes of saNOS and mammalian NOS are stabilized by L-arginine. However, the still quite high rate of autoxidation makes the presence of an electron donor close to the heme necessary to disfavor the autoxidation of the complex and favor oxygen activation (5,7,20,33). H 4 B plays this role in mammalian NOS. In saNOS, the rate of electron transfer to the heme remains to be determined, but unless it is very significantly faster than in mammalian NOS, an electron donor molecule at the cofactor site would be critical.
How the increase in the superoxide character of the heme-bound oxygen upon L-arginine binding affects the subsequent steps of oxygen activation is unclear. It is interesting that the interaction of the D251N mutant of P450 cam with putidaredoxin increases the amount of a population of the oxygenated complex with lower O-O and Fe-O frequencies (38) ( Table 2). This structural perturbation was proposed to be related to the function of putidaredoxin, which is the electron donor of P450 cam , and to involve an alteration of the electron-donating properties of the thiolate ligand.

CONCLUSIONS
We have shown that the formation and decay of the oxygenated complex of saNOS followed kinetics similar to those of mammalian NOS. L-Arginine reduced the rate of autoxidation of the oxygenated complex as with nNOSox and eNOSox. H 4 B, by providing an electron for oxygen activation, increased the rate of disappearance of the oxygenated complex in the presence of L-arginine and NOHA. saNOS could synthesize NO in a single turnover reaction from NOHA. The optical spectrum of the oxygenated complex, with a Soret absorption band at 430 nm, was similar to the heme-oxyII complex of nNOSox and eNOSox. In addition, nNOSox and saNOS display O-O modes at nearly identical fre-quencies. Therefore, the properties of the oxygenated complex of saNOS that we describe here might also provide insights into the properties of the oxygenated complex of mammalian NOS.