Reaction of Neuronal Nitric-oxide Synthase with Oxygen at Low Temperature

The reaction of reduced NO synthase (NOS) with molecular oxygen was studied at −30 °C. In the absence of substrate, the complex formed between ferrous NOS and O2 was sufficiently long lived for a precise spectroscopic characterization. This complex displayed similar spectral characteristics as the oxyferrous complex of cytochrome P450 (λmax = 416.5 nm). It then decomposed to the ferric state. The oxidation of the flavin components was much slower and could be observed only at temperatures higher than −20 °C. In the presence of substrate (l-arginine), another, 12-nm blue-shifted, intermediate spectrum was formed. The breakdown of the latter species resulted in the production ofN ω-hydroxy-l-arginine in a stoichiometry of maximally 52% per NOS heme. This product formation took place also in the absence of the reductase domain of NOS. Both formation of the blue-shifted intermediate and ofN ω-hydroxy-l-arginine required the presence of tetrahydrobiopterin (BH4). We propose that the blue-shifted intermediate is the result of reductive activation of the oxygenated complex, and the electron is provided by BH4. These observations suggest that the reduction of the oxyferroheme complex may be the main function of BH4 in NOS catalysis.

Nitric-oxide synthase (EC 1.14.13.39; NOS) 1 catalyzes the oxidation of L-arginine to L-citrulline and nitric oxide (NO) in two reaction cycles. The first one leads to the formation of N -hydroxy-L-arginine (NOHLA), and the second one leads to the formation of L-citrulline and NO (1,2). The latter reaction is not specific for NOS; the formation of L-citrulline and nitrite, an oxidation product of NO, has also been reported for cytochrome P450 (3). Similarly to cytochrome P450, NOS contains a thiolate ligated heme, NADPH is the initial electron donor, and the reaction requires molecular oxygen (O 2 ), which has to be activated and split for the oxidation of L-arginine and the formation of water (4,5). In contrast to P450, in NOS the entire electron transfer chain is contained in a single molecule, consisting of a flavin-containing reductase domain and a hemecontaining oxygenase domain (6). Furthermore, it requires a cofactor, (6R)-5,6,7,8-Tetrahydro-L-biopterin (tetrahydrobiopterin; BH 4 ), the role of which is still unclear but the presence of which is essential for NO synthesis (7,8). In addition, Ca 2ϩ / calmodulin is required for electron flow from the reductase domain to the oxygenase domain (9).
NOS catalysis involves the interaction of the reduced enzyme with molecular oxygen. Although little is known about this process for NOS, a wealth of data is available about the analogous P450 system (10). As shown in Scheme 1, reduced P450 is known to react rapidly with oxygen to form a transient oxygen complex, which has been stabilized at low temperature (11). Several biophysical studies have elucidated the electronic structure of this complex as a Fe(II)O 2 7 Fe(III)O 2 ⅐ equilibrium, which is poised in favor of the ferric form (Scheme 1, 1) (12)(13)(14). The subsequent reduction of this complex is thought to lead first to a peroxy compound (2) and eventually, after two protonation steps, to an iron(IV) porphyrin -cation radical (3). Both species 2 and 3 have been proposed to act as oxygenating agents (15). The breakdown of these species results in product release and formation of ferriheme via still unknown intermediates (4).
The corresponding reactions for NOS have not been reported, except for a very recent stopped-flow study, which demonstrated the transient formation in the oxidation of ferrous NOS by O 2 of a compound absorbing at 427 nm (16). In this work, we used a different approach to study the reaction between molecular oxygen and reduced neuronal NOS (nNOS). In order to trap possible intermediates, most of the experiments were carried out at subzero temperatures in the presence of 50% ethylene glycol as anti-freeze solvent (17). The same procedure was employed previously for the study of P450 oxygen complexes (11,18,19). Special attention was paid to the role of BH 4 in the reduction of the oxygen complex in the presence of L-arginine. Our results throw more light on the function of BH 4 in NOS catalysis.

EXPERIMENTAL PROCEDURES
Materials-Enzymes used for molecular biological procedures were from New England Biolabs (supplied by Biotrade, Vienna). The TA cloning kit from Invitrogen was from the same vendor. The Sequenase kit was obtained from Amersham Pharmacia Biotech (supplied by Med-Pro, Vienna). Oligonucleotides were synthesized by the Institute of Microbiology and Genetics at the Biocenter of the University of Vienna. Glutathione-agarose beads, thrombin, and all other chemicals were from Sigma. BH 4 was from Alexis Biochemicals (Switzerland); 7,8dihydro-L-biopterin (BH 2 ) was from Schircks Laboratories (Switzerland); calmodulin, L-arginine, and NOHLA were from Sigma; L- [2,3,4, H]arginine hydrochloride (57 Ci/mmol) was from Amersham Pharmacia Biotech and purified by HPLC before use. All other reagents were purchased from Merck except for CHAPS, which was from Boeh-* This work was supported by Fonds zur Förderung der Wissenschaftlichen Forschung in Ö sterreich Grants P 11478, P 10655, P 10859, and P 12191 and by the French-Austrian AMADEUS program. 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.
¶ Preparation of Nitric-oxide Synthase-Recombinant rat brain nNOS was purified from baculovirus-infected insect cells as described previously (20). The holoenzyme contained tightly bound BH 4 in the ratio of 0.5 per heme. BH 4 -free nNOS was obtained from Sf9 cells treated with 2,4-diamino-6-hydroxypyrimidine according to a published procedure (8). The nNOS oxygenase domain was cloned and expressed in Escherichia coli by a procedure to be published elsewhere. The molar concentrations of nNOS and its oxygenase domain were determined from absorption coefficients of the oxidized and the ferrous-CO complex according to Sono et al. (21).
Preparation of Cytochrome P450 -Cytochrome P450 CYP2B4 from rabbits, which had been under phenobarbital treatment for 1 week, was prepared according to the procedure of Coon et al. (22). The protein was kept frozen as a 69 M solution in 10 mM potassium phosphate buffer, pH 6.8, containing 20% glycerol and 1 mM EDTA. The purified enzyme was electrophoretically homogenous, and in the oxidized state it displayed an absorbance ratio of R 417/277 ϭ 1.06. Neither NOS nor cytochrome P450 showed P420 characteristics.
Low Temperature Spectroscopy-Base line-corrected absorbance spectra in the 350 -700-nm range were recorded with a Cary 3E (Varian) spectrophotometer in the double beam mode, using a 1.5-nm slit width. Data acquisition was in steps of 0.5 nm, with an acquisition time of 0.033-0.5 s/data point. The instrument was equipped with a homebuilt double sample compartment, which allowed spectral recordings at low temperatures (down to Ϫ50°C). Sample and reference cuvettes were placed in a bloc made of aluminum, which was thermostatted by circulation of ethanol, thermoregulated in a Haake F3-Q bath. Inside the cells, the temperature was monitored continuously with a thermocouple connected to an AOIP voltmeter. For thermal insulation, the sample compartment was surrounded by polyvinyl chloride walls, equipped with double quartz windows. Formation of ice and condensation of water on the windows was prevented by a flow of dry nitrogen.
Formation of Intermediate Oxygen Complexes-To prevent freezing of the sample, the experiments were done in mixed organic solvents. If not specified otherwise, we used ethylene glycol/water (1:1, v/v). This solvent did not change significantly the spectral properties of NOS, it did not induce a transition to the P420 state, and in its presence the enzyme was still active in a standard assay (23). The buffer was 50 mM potassium phosphate, pH 7.2, containing 1 mM CHAPS, 0.5 mM EDTA, and 1 mM 2-mercaptoethanol. It contained no calmodulin, and, unless specified, no BH 4 was added. For CYP2B4, the solvent was 50% glycerol, and the buffer was sodium phosphate, 50 mM, pH 7.4. The pH of these mixed solvents is known to vary only little as a function of temperature (17). The sodium dithionite stock solution (23 mM) was prepared in the same solvent. Prior to use, argon was bubbled through the solutions for 30 min. The enzymes were diluted in the oxygen-free buffers at final concentrations of 3 M in Teflon closed cuvettes in total volumes of 2 ml. They were reduced at 15°C by the addition of 20 l of a concentrated solution of sodium dithionite (final concentration, 230 M) using a Hamilton syringe. When reduction was complete, the temperature was decreased to the desired value (typically Ϫ30°C). 2-5 ml of precooled oxygen were then bubbled in the enzyme solution with a syringe. This procedure took about 5 s, and thereafter the spectra were recorded in intervals of 2 min.
Product Analysis-Reduction and oxygenation of NOS and its heme domain were carried out as described above, except for the presence of 500,000 cpm of 200 M L-[2,3,4,5-3 H]arginine. After incubation for 5-60 min at the specified temperature, the reaction was stopped by the addition of 75 l HCl (1 M) to the reaction mixture (1 ml). After centrifugation with a bench-top centrifuge, 100 l of the supernatant were injected into a 25-cm Nucleosil SA 100 -5 HPLC column from Macherey & Nagel. The arginine derivatives were separated at room temperature in 50 mM sodium acetate, pH 6.5, at a flow rate of 1.5 ml/min. Fractions of 750 l were collected for 30 min, and radioactivity was determined in each fraction by liquid scintillation counting. NOHLA was identified by comparison with an authentic standard (24). Its formation was quantified from its integrated elution peak relative to the total radioactivity. In one control experiment, we used NADPH and not dithionite to reduce nNOS. In that case, the phosphate buffer was replaced by triethanolamine, 50 mM, pH 7.4, and a procedure similar to that described by Abu-Soud et al. (25) was used. First, the heme iron of nNOS was reduced by 200 M NADPH in the presence of 14 M calmodulin and 1 mM calcium chloride in aqueous buffer at 15°C under the experimental conditions we had employed for reduction by dithionite. After dissociation of calmodulin by excess EDTA (1.8 mM final concentration), 200 M radiolabeled L-arginine, was added, and oxygen was bubbled through the solution. The production of NOHLA was then quantified as described above.

RESULTS
Oxygen Binding to Reduced nNOS in the Absence of Substrate-Under anaerobic conditions, 230 M dithionite reduced the NOS sample within 10 -15 min at 15°C. Spectrally, the reduction was characterized by a loss of the flavin absorbance, a shift of the Soret band from 400 to 410 nm, a shoulder at 455 nm, and a maximum at 554 nm. During the subsequent temperature decrease to Ϫ30°C, the spectrum of Fe(II) nNOS remained essentially the same, except for a slight sharpening of the Soret band. After the addition of oxygen at Ϫ30°C, the spectrum changed within 1 min. As shown in Fig. 1, the amplitude of the Soret band was slightly decreased, and its maximum was red-shifted to 416.5 nm. In the visible region, the main effect was a broadening of the absorption band. As shown in Fig. 2A, the oxygenated minus reduced difference spectrum showed a strong maximum at 426 nm, two sharp minima at 401 and 458 nm, and an isosbestic point at 415 nm. Furthermore, two broad maxima appeared around 500 and 600 nm, as well as a minimum at about 550 nm.
At Ϫ30°C, the slow decomposition of the oxygen complex led to an oxidized heme with reduced flavins (Fig. 1). As shown in Fig. 2A, the oxidized minus oxygen complex difference spectrum grew with maxima at 387.5, 418.5, 516.5, and 650 nm, minima at 442.5 and 558 nm, and an isosbestic point at 427 nm. The observed changes and the resulting spectrum are indicative of a mixture of high and low spin ferriheme but with a much higher fraction in the low spin state than before the experiment at 15°C. Finally, when the temperature was raised above Ϫ20°C, the flavins became also oxidized, as evidenced by a broad absorbance increase between 350 and 420, and between 470 and 550 nm (not shown). The latter reaction is probably due to a direct (nonphysiological) reaction of the reduced flavins with oxygen or with hydrogen peroxide resulting from the reaction of oxygen with dithionite.
As shown in Fig. 2, the spectral characteristics of formation and decomposition of the oxygen complex were similar to those observed with cytochrome P450. An exception was the trough at 458 nm in the difference spectrum (intermediate Ϫ reduced state, Fig. 2A), which was absent in P450. However, a comparison of the spectra of the reduced states shows that this spectral trough originates from the properties of the reduced state of NOS rather than from its oxygen complex. Indeed, the spectrum of reduced NOS, but not reduced P450, exhibits a shoulder at this wavelength. The differences observed during the autoxidation of the oxygen complex as compared with the analogous reaction in P450 (Fig. 2B), i.e. additional absorbance increases at 387.5, 516.5, and 650 nm, can be ascribed to the formation of a significant fraction high spin ferriheme.
The kinetics of oxygen binding and autoxidation were followed at 442.5 nm. Oxygen binding to reduced NOS was very rapid even at Ϫ30°C. Subsequent oxidation of the heme was monoexponential ( ϭ 6.5 min at Ϫ30°C). Oxidation of the SCHEME 1. Supposed reaction of cytochrome P450 with oxygen within its catalytic cycle. The numbers refer to species cited in the text. For simplicity the possible interaction of 2 with substrate R-H is not shown. flavins could only be observed at considerably higher temperatures ( ϭ 15.6 min at Ϫ11°C).
Oxygen Binding in the Presence of Substrate-The addition of L-arginine (200 M) resulted in the well known shift of the heme iron spin equilibrium toward more high spin. This effect persisted in the presence of ethylene glycol and at low temperatures. Reduction gave rise to a spectrum very similar to that of substrate-free reduced nNOS (Fig. 3A). In this case, too, mixing of reduced nNOS with oxygen at Ϫ30°C produced a detectable intermediate, but its Soret band was blue-shifted ( max ϭ 404.5 nm) (Fig. 3B). In the visible region, the intermediate spectrum was similar to that obtained in the absence of L-arginine. At Ϫ30°C, its decay kinetics resulted within 37 min in an oxidized type high spin enzyme. The difference spectra reflecting formation (intermediate Ϫ reduced state, Fig. 3C) and decomposition (reoxidized Ϫ intermediate) were strikingly different from those obtained in the absence of L-arginine (see above). The formation was characterized by a strong maximum at 375 nm and a double minimum at 417 and 453 nm. During the decomposition of the intermediate, a broad maximum appeared at 370 nm, with a shoulder at 420 nm and a minimum at 444 nm. BH 4 -free NOS-In the absence of L-arginine, binding of oxygen to BH 4 -free NOS at Ϫ30°C resulted in similar spectral changes as those observed with BH 4 -bound NOS (cf. "Discussion" and Table II); the autoxidation of the oxygen complex ( max ϭ 415 nm), resulted in low spin ferric heme. In contrast, the presence of L-arginine did not induce the blue-shifted oxygen complex observed in BH 4 -bound NOS, and the spectral characteristics of the oxygen complex were close to those observed in the BH 4 -free NOS in the absence of L-arginine. As for the BH 4 -bound NOS in the presence of L-arginine, the autoxidation of this complex resulted in high spin ferric heme.
The Oxygenase Domain-The spectral changes upon oxygen addition to the reduced oxygenase domain were recorded at Ϫ30°C as described above, in the absence and presence of BH 4 (20 M) and L-arginine (200 M). In all cases, a direct transition from the reduced to the oxidized state was observed, and no intermediate state could be detected. Whereas the cofactors did not significantly affect the spectrum of the reduced state, the final oxidized state was primarily low spin in the absence of substrate and high spin in the presence of L-arginine and/or BH 4 . As an example, the spectral changes in the presence of BH 4 and L-arginine are shown in Fig. 4. Clear isosbestic points were observed at 406, 483, 534, and 595 nm. At Ϫ30°C, the oxidation kinetics were relatively slow ( ϭ 12.4 min) without being affected noticeably by L-arginine and/or BH 4 .
Oxygen Complex Decomposition Products-One explanation for the blue-shifted oxygen complex of BH 4 -bound full-length NOS in the presence of L-arginine might be that it represents a later intermediate in the activation of oxygen, one electron reduced beyond the ferrous oxygenated species. Since such a state could in principal furnish the two electrons required for the oxidation of L-arginine to NOHLA and the concomitant reduction of oxygen to water, we decided to analyze the decomposition products of the reaction with radiolabeled L-arginine. Fig. 5 shows a typical HPLC profile of the reaction products. Only one radioactive reaction product was formed, and this had the same retention time as authentic NOHLA. Control experiments in the absence of enzyme with or without added BH 4 showed no product formation, and no NOHLA was detected unless the enzyme was reduced prior to the addition of oxygen. As shown in Table I, NOHLA formation was proportional to the NOS concentration in a stoichiometry between 0.3 and 0.5 per nNOS heme. NOHLA formation took place in both aqueous and mixed organic solvent. Moreover, the production of NOHLA was not specific to the presence of dithionite as the source of the first electron. When dithionite was replaced by NADPH (see "Experimental Procedures"), the yield of NOHLA production in aqueous buffer was 0.47 Ϯ 0.03 per heme, which is the same value as that obtained with an initial reduction by dithionite. It must be mentioned here that under similar conditions (reduction with NADPH, but with excess BH 4 ) Abu-Soud et al. (25) did not detect the formation of NOHLA. The reason for this difference is not clear. However, we are confident in our result; the experiment was carried out under argon atmosphere, and the result was reproducible.
As shown in Table I, the oxygenase domain alone was sufficient for this reaction. The presence of BH 4 , however, was essential. The oxidation state of the biopterin is also important; with the oxygenase domain, no NOHLA was produced when BH 4 was replaced by BH 2 . The low residual NOHLA production in the full-length holoenzyme in the presence of BH 2 is due to endogenous BH 4 , which is not readily displaced by added BH 2 (26). DISCUSSION The Oxygen Complex of Reduced nNOS in the Absence of Substrate-By the use of subzero temperature spectroscopy, it was possible to detect an intermediate complex between reduced nNOS and molecular oxygen. This technique was employed previously to trap the analogous complex of various cytochrome P450 forms. The spectral characteristics of the complexes of both enzyme classes are quite similar, and the differences observed during complex formation (a trough at 458 nm) are due to a spectral particularity of deoxyferrous nNOS. The shoulder at 458 nm persists in reduced nNOS in the absence of the reductase domain and is not affected by BH 4 . The shoulder, which was noted before by others (21), may be due to the presence of a sixth ligand in a fraction of the enzyme. The additional absorbance increase at 380 -390 nm suggests a hyperporphyrin spectrum, as has been reported for many ferrous complexes of cytochrome P450 (27,28). The identity of the putative ligand cannot be established on the basis of the spectral changes, since most exogenous ligands other than O 2 induce a red shift upon binding to ferrous P450 type heme.
The spectral similarity of nNOS and P450 oxygen complexes suggests that both complexes have a similar electronic configuration. Most evidence in P450 points to a structure of Fe(III)O 2 ⅐ , and this configuration is also expected for NOS (29). Since the same intermediate spectrum was obtained also in the absence of BH 4 , it may be concluded that, in substrate-free nNOS, BH 4 does not perturb the electronic structure of the oxygen complex.
Stability of the Oxygen Complex-The oxygen complex of nNOS is rather unstable, and it was necessary to lower the temperature to Ϫ30°C to prevent its immediate autoxidation. In that aspect, nNOS resembles microsomal cytochrome P450 (18). The oxygen complex of mitochondrial P450 scc is more stable (several hours at Ϫ30°C (19)). The most stable one, the complex of soluble P450 cam , can be detected without difficulty above 0°C (at 25°C, ϭ 90 s (14)). It appears thus that the protein environment plays a significant role in the stability of the oxygen complex. This idea is further supported by the fact that with the nNOS oxygenase domain the complex could not be detected, indicating that in this case autoxidation of the complex was much faster than its formation. Since reoxidation of the oxygenase domain was actually slower than that of the full-length enzyme, formation of the oxyferrocomplex must be considerably slower for the oxygenase domain. The origin of the effect remains to be established. One possibility might be that, in the full-length enzyme, the reductase domain enhances the oxygen association rate. Alternatively, the differences may somehow originate from the different expression systems used, since the full-length enzyme and the oxygenase domain were obtained from baculovirus-infected cells and from overexpression in E. coli, respectively.
The Oxygen Complex in the Presence of L-Arginine and BH 4 -The spectral shift of the nNOS oxygen complex in the presence of L-arginine and BH 4 may be interpreted in two ways. One possibility is that the substrate and BH 4 interact strongly with the Fe(III)O 2 ⅐ complex and modify sufficiently the energy of its electronic orbitals to induce the blue shift. Examples for an effect of substrates on the Fe(III)O 2 ⅐ complex are found in cytochrome P450 scc , which interacts with cholesterol, (22R)-hydroxycholesterol and (20R,22R)-dihydroxycholesterol. These substrates induce significant changes of the ferric spin state and thus of the P450 Soret absorbance spectrum. However, the max values of the respective oxyferrous complexes range between 423 and 416 nm, while that of the complex in the absence of substrate lies at 420 nm (31), so that the substrate-induced shift never exceeds 4 nm. In contrast, the nNOS oxygen complex in the presence of substrate is blue-shifted by 12 nm.
Alternatively, the blue-shifted intermediate spectrum in nNOS may reflect a further intermediate state (reduced by a second electron) in the course of the activation of oxygen. The observation that the decomposition of this enzyme state yields NOHLA, the formation of which requires two electrons, provides strong supporting evidence in favor of this possibility. Similarly, with cytochrome P450, product formation is only possible if the oxygen complex is reduced by a second electron.
Possible configurations of such a reduced oxygen complex are shown in Fig. 6. They comprise a peroxo-(or hydroperoxo-) Fe(III) (2), an oxyferryl porphyrin -cation (3), or a hydroxyferryl complex (4). Forms 2 and 3 have been characterized for different cytochromes P450 and peroxidases. An oxyferrous complex reduced by one electron maintaining an intact dioxygen bond was recently reported for the D251N mutant of P450 cam (32,33), which showed a red-shifted max of the Soret band (see Table II). In contrast, compound I of horseradish peroxidase (3), is characterized by a blue-shifted maximum (34). A similar blue-shifted intermediate (405 nm), reflecting form 3, was also reported for cytochrome P450 scc as an intermediate compound in the reaction with (22R)-hydroperoxycholesterol, and for cytochrome P450 cam as an intermediate in the reaction of the oxyferrous complex with reduced putidaredoxin (35)(36)(37). It thus appears that, relative to the oxyferrous form 1, form 2 is red-shifted, and form 3 is blue-shifted. Our interpretation of the blue-shifted NOS intermediate is therefore that it reflects form 3. Additional support for this hypothesis comes from the very recently determined crystal structure of inducible nitric-oxide synthase (38), which suggests a stabilization of form 3 by a stacking of aromatic residues with the heme. But clearly, the definite assignment of its electronic structure awaits further investigation.
The Role of BH 4 in the Reductive Activation of the Oxygen Complex-Our results clearly show that the decomposition of the oxygen complex of nNOS in the presence of both L-arginine and BH 4 results in the formation of NOHLA. This reaction requires that the Fe(III)O 2 ⅐ -L-arginine complex be reduced by one electron to Fe(III)O 2 Ϫ -L-arginine, which would then decompose in several steps to Fe(III) and NOHLA. But where does the second electron, necessary to reduce Fe(III)O 2 ⅐ -L-arginine stem from? For several reasons, it is highly unlikely that it originates from dithionite. (i) For the related cytochrome P450 system, the second electron is exclusively provided by a specific reductase, the analogous counterpart of the reductase domain in NOS. (ii) Replacement of dithionite by NADPH did not affect the yield of NOHLA formation. (iii) Reduction of nNOS by dithionite was much slower than the formation of NOHLA. In control experiments, we measured the reduction rate between 5 and 30°C (data not shown) and determined an activation energy of E a ϭ 115 kJ/mol. Extrapolation of the linear Arrhenius plot to Ϫ30°C yielded a prohibitively slow reduction rate ( ϭ 2 weeks), whereas NOHLA was formed in less than 5 min at the same temperature.
The fact that we performed our experiments in the absence of calmodulin, essential for electron transfer from the flavins to the heme (9) and that NOHLA formation was also observed with the isolated oxygenase domain rules out the flavins as the electron source. Theoretically, the two hemes within one dimer might be able to provide the electrons to account for the observed NOHLA-to-heme stoichiometry. However, since most  evidence suggests that the hemes operate independently (30,39) and that electron transfer between the hemes does not occur (40), this option is unlikely. This leaves BH 4 as the prime candidate for the donor of the second electron, in perfect agreement with the observation that NOHLA formation does not take place in its absence. The present results are therefore relevant with regard to the role of BH 4 in catalysis. It has long been suspected that the redox properties of BH 4 are crucial to its function, although direct evidence for its participation in the redox process is still lacking (reviewed by Mayer and Werner (41) and Hemmens and Mayer (42)). The best evidence to date in favor of a redoxactive role is the observation that redox-inactive BH 4 analogues can mimic the effects of BH 4 on NOS dimerization, substrate affinity, and heme spin state but fail to sustain L-citrulline production (26,(43)(44)(45). However, little information is available on the question of at which stage the putative BH 4 oxidation takes place and whether it occurs in the reaction with L-arginine or with NOHLA or in both steps. Concerning the latter question, it was reported that oxidized BH 4 -containing NOS can catalyze the conversion of substoichiometric concentrations of L-arginine into NOHLA, whereas further conversion to L-citrulline was strictly NADPH-dependent, suggesting that BH 4 may serve as an electron source for the first but not for the second step (46). However, in the present experiments, no NOHLA was formed unless the enzyme was previously reduced. The cause for the divergent results is unclear. A possible explanation might be that the enzyme preparation of Campos et al. (46) contained traces of reductant, e.g. reduced flavins.
The present results demonstrate that BH 4 is absolutely required for L-arginine oxidation. If our assignment of the blueshifted intermediate as the oxyferryl complex (3) is correct, the observation that its formation is only observed in the presence of BH 4 implies that BH 4 is required for oxygen activation. The simplest explanation and the one favored by us is that BH 4 furnishes the electron needed to reduce Fe(III)O 2 ⅐ (1) to Fe(III)-O-OH (2), which immediately transforms then to the Fe(IV)ϭO porphyrin -cation radical (3). Species 3 could then react with bound L-arginine via hydrogen abstraction to form species 4, the decomposition of which would liberate NOHLA and Fe(III). In the absence of BH 4 , the uncoupled reaction (47, 48) will occur, and compound 1 will dissociate to Fe(III) and O 2 . (Fig. 6).
In this way, our model also explains why the redox activity of BH4 is crucial for coupling NADPH oxidation to NO formation, although it is not required for O 2 reduction per se. Since only one electron is transferred in this reaction, we denote the resulting state of the biopterin formally as BH 3 . However, we cannot exclude the possibility that the state that is actually formed is the quinonoid-BH 2 (6,7-dihydro-L-biopterin), which is the usual product in enzymatic oxidations of BH 4 (41,49) and which can be reduced to BH 4 by NOS (50).
The NOHLA-to-heme ratios of Յ0.5 suggest that in nNOS as isolated, which contains approximately one equivalent of BH 4 per dimer, only the BH 4 -containing subunit produces NOHLA. This confirms that electron transfer between the two subunits in a NOS dimer does not take place (40) and agrees with the role postulated by us for BH 4 in catalysis. Similarly, in experiments with the oxygenase domain, where BH 4 was added (20 M), the NOHLA-to-heme ratio did not reach unity, although it was a little higher (0.52 instead of 0.36 under the same buffer conditions). This may be explained by an incomplete binding of BH 4 , which could result from a relatively high K d of the second BH 4 binding site and an increase of K d due to the presence of ethylene glycol. The latter idea is supported by our observation that with full-length nNOS the ratio is somewhat decreased in the presence of ethylene glycol.
Conclusion-The observations presented in this study provide important clues to the role BH 4 plays in NOS catalysis. The hypothesis that electron transfer from BH 4 is required for oxygen activation is the simplest explanation of our results. It does, however, need further corroboration, and it remains also to be established if the same holds true for the second step in NO catalysis, the oxidation of NOHLA to L-citrulline.