Reduction of quinonoid dihydrobiopterin to tetrahydrobiopterin by nitric oxide synthase.

Rat cerebellar nitric oxide synthase (NOS) purified from transfected human kidney cells catalyzes an NADPHdependent reduction of quinonoid dihydrobiopterin (qBH2) to tetrahydrobiopterin (BH4). Reduction of qBH2 at 25 microM proceeds at a rate that is comparable with that of the overall reaction (citrulline synthesis) and requires calcium ions and calmodulin for optimal activity; NADH has only 10% of the activity of NADPH. The reduction rate with the quinonoid form of 6-methyldihydropterin is approximately twice that with qBH2. 7,8-Dihydrobiopterin had negligible activity. Neither 7,8-dihydrobiopterin nor BH4 affected the rate of qBH2 reduction. Reduction is inhibited by the flavoprotein inhibitor diphenyleneiodonium, whereas inhibitors of electron transfer through heme (7-nitroindazole and N-nitroarginine) stimulated the rate to a small extent. Methotrexate, which inhibits a variety of enzymes catalyzing dihydrobiopterin reduction, did not inhibit. These studies provide the first demonstration of the reduction of qBH2 to BH4 by NOS and indicate that the reduction is catalyzed by the flavoprotein "diaphorase" activity of NOS. This activity is located on the reductase (C-terminal) domain, whereas the high affinity BH4 site involved in NOS activation is located on the oxygenase (N-terminal) domain. The possible significance of this reduction of qBH2 to the essential role of BH4 in NOS is discussed.

Rat cerebellar nitric oxide synthase (NOS) purified from transfected human kidney cells catalyzes an NADPHdependent reduction of quinonoid dihydrobiopterin (qBH 2 ) to tetrahydrobiopterin (BH 4 ). Reduction of qBH 2 at 25 M proceeds at a rate that is comparable with that of the overall reaction (citrulline synthesis) and requires calcium ions and calmodulin for optimal activity; NADH has only 10% of the activity of NADPH. The reduction rate with the quinonoid form of 6-methyldihydropterin is approximately twice that with qBH 2 . 7,8-Dihydrobiopterin had negligible activity. Neither 7,8dihydrobiopterin nor BH 4 affected the rate of qBH 2 reduction. Reduction is inhibited by the flavoprotein inhibitor diphenyleneiodonium, whereas inhibitors of electron transfer through heme (7-nitroindazole and Nnitroarginine) stimulated the rate to a small extent. Methotrexate, which inhibits a variety of enzymes catalyzing dihydrobiopterin reduction, did not inhibit. These studies provide the first demonstration of the reduction of qBH 2 to BH 4 by NOS and indicate that the reduction is catalyzed by the flavoprotein "diaphorase" activity of NOS. This activity is located on the reductase (C-terminal) domain, whereas the high affinity BH 4 site involved in NOS activation is located on the oxygenase (N-terminal) domain. The possible significance of this reduction of qBH 2 to the essential role of BH 4 in NOS is discussed.
Nitric oxide synthase (NOS) 1 catalyzes the conversion of arginine to nitric oxide and citrulline. The enzyme has attracted much interest because of the involvement of nitric oxide in a variety of physiological processes, such as vasodilation, cytotoxic activity, and intercellular signaling in the brain (1)(2)(3). All three isozymes of NOS have FAD, FMN, heme, and tetrahydrobiopterin (BH 4 ) as required prosthetic groups and either have calmodulin (CaM) tightly bound or require binding of this factor for activity. The C-terminal part of the protein, containing NADPH-, FMN-, and FAD-binding sites, is homologous to NADPH cytochrome P450 reductase, whereas the N-terminal part of the protein contains the heme-, BH 4 -, and argininebinding sites (4 -8). Based on direct studies of NOS and its similarity to cytochrome P-450 (9), it has been proposed that the pathway of electron flow in NOS is NOS is unique as a heme oxygenase in requiring BH 4 , and elucidation of the role of BH 4 is essential in understanding how this enzyme functions. In the aromatic amino acid hydroxylases, BH 4 acts as a cofactor, providing electrons for oxygenation; the oxidized biopterin, quinonoid dihydrobiopterin (qBH 2 ), is recycled to BH 4 by dihydropteridine reductase (10). Whereas it has been proposed that BH 4 functions in NOS in the same manner, neither the oxidation of BH 4 nor the reduction of dihydrobiopterin catalyzed by NOS has been demonstrated (11)(12)(13)(14). Giovanelli et al. (15) were unable to detect recycling of added BH 4 and suggested that added BH 4 was not acting as a stoichiometric reactant as it does in the aromatic amino acid hydroxylases. The possibility of recycling of enzyme-bound BH 4 was not excluded (16). On the basis of these results, it was suggested that BH 4 may function as an allosteric effector or to maintain some group(s) on the enzyme in a reduced state required for activity (15). This suggestion has been supported by more recent studies showing that BH 4 helps stabilize the active dimeric state of the enzyme (17,18), maintains a stable heme pocket that is accessible to oxygen (19), and prevents and reverses the inhibition of NOS by nitric oxide (20). Because it is not clear whether BH 4 undergoes redox reactions during NOS catalysis, we have examined directly the reduction by NOS of qBH 2 to BH 4 .

EXPERIMENTAL PROCEDURES
Materials-Beef liver catalase was obtained from Boehringer Mannheim. L-␣-Lysolecithin, N G -nitro-L-arginine (NNA), superoxide dismutase from bovine erythrocytes, glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides, and cytochrome c from horse heart were obtained from Sigma. Diphenyleneiodonium (DPI) was from Aldrich, and 7-nitroindazole (7NI) was from Biomol. CaM from bovine brain was purchased from Calbiochem. BH 4 , 7,8-dihydrobiopterin (7,, and (6R,S)-6-methyl-5,6,7,8-tetrahydropterin (6MPH 4 ) were obtained from B. Schircks Laboratories (Jona, Switzerland). L-[2,6- Corp. qBH 2 was prepared by oxidizing BH 4 with dichlorophenolindophenol (21) at pH 7.6 in 5.5 mM sodium bicine buffer (22), followed by removal of the dye by ether extraction. The qBH 2 solution was flushed with argon to remove the last traces of ether. Quinonoid 6-methyl dihydropterin (q6MPH 2 ) was prepared from 6MPH 4 using the same method. Details of the methods used to minimize the instability of the quinonoid dihydropterins are described below under "qBH 2 Reduction Assay." Enzymes-Recombinant rat phenylalanine hydroxylase (PAH) was prepared by minor modification (23) of the procedure of Citron et al. (24) and had a specific activity of 1.7 mol min Ϫ1 mg Ϫ1 . Rat brain NOS was purified from transfected human kidney 293 cells (6) by minor modifications (25) of the procedure of McMillan et al. (26). As was done previously (25), dithiothreitol was omitted from the gel filtration procedure. NOS isolated this way had a single band on an SDS-polyacrylamide gel and typically had a specific activity of about 220 nmol min Ϫ1 * 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 1 The abbreviations used are: NOS, nitric oxide synthase; BH 4 , 6-(Lerythro-1,2-dihydroxypropyl)-5,6,7,8-tetrahydropterin (tetrahydrobiopterin); 7,8-BH 2 , 7,8-dihydrobiopterin; CaM, calmodulin; DPI, diphenyleneiodonium; HPLC, high performance liquid chromatography; 6MPH 4 , (6R,S)-6-methyl-5,6,7,8-tetrahydropterin; NHA, N G -hydroxy-Larginine; 7NI, 7-nitroindazole; NNA, N G -nitro-L-arginine; PAH, phenylalanine hydroxylase; qBH 2 , quinonoid dihydrobiopterin; q6MPH 2 , quinonoid-6-methyl-dihydropterin. mg Ϫ1 as determined by citrulline formation at 25°C essentially as described by Bredt and Snyder (27). The BH 4 content of the purified NOS varied between 0.25 and 0.35 mol/mol subunit. Protein was determined by the bicinchoninic acid method (Pierce manual 23225X) with the use of a bovine serum albumin standard. qBH 2 Reduction Assay-Assay of qBH 2 reduction is complicated by the instability of this pterin. QBH 2 rearranges to 7,8-BH 2 (21), is converted to BH 4 , probably by dismutation (28), and is reduced nonenzymically by NADPH to BH 4 (21). Rearrangement to 7,8-BH 2 was minimized by preparing the stock solution in sodium bicine buffer, pH 7.6 (22) at 0°C and by using this solution within 4 min of its preparation. In agreement with the report of Matsuura et al. (28), we observed that nonenzymic dismutation of qBH 2 to BH 4 occurred predominantly during acidification (with perchloric acid). Accordingly, this period was kept as short as possible and constant for all samples. Nonenzymic reduction of qBH 2 to BH 4 was circumvented by decreasing the concentration of NADPH to 1 M; the low concentration of NADPH was kept constant by the inclusion of an NADPH regenerating system consisting of glucose-6-phosphate and glucose-6-phosphate dehydrogenase. This low NADPH concentration does not cause a large decrease in the rate of citrulline formation, which is approximately 60% of that measured by the Bredt and Snyder assay (27). This is in agreement with the low K m for NADPH of 0.2-0.5 M (29). 2 Reduction of qBH 2 was determined in a standard 20-l reaction mixture containing 25 mM sodium bicine buffer, pH 7.6, 0.6 mM EDTA, 0.9 mM CaCl 2 , 55 g/ml CaM, 1.5 mM glucose-6-phosphate, 0.12 units of glucose-6-phosphate dehydrogenase, and 1 M NADPH. Unless stated otherwise, the reaction was started by adding qBH 2 to a concentration of 25 M immediately after adding NOS (0.5-1.5 g). The reaction mixture was incubated at 25°C. Blank reactions in which NOS was omitted were used to correct for nonenzymic conversion of qBH 2 to BH 4 . Reactions were terminated after 30, 60, and 90 s by adding 4 l of 1.5 M HClO 4 followed by partial neutralization to pH 3 with 4 l of approximately 1.4 M KHCO 3 . The low pH is important for stability of BH 4 and to permit subsequent assay of the BH 4 content of the partially neutralized solution at a pH of 6.8 (see below). In order to compensate for small variations in concentration of the solutions of HClO 4 and KHCO 3 , the exact concentration of the latter solution required to achieve a pH of about 3 should be determined in a separate titration experiment. The precipitate was removed by centrifugation, and 25 l of the supernatant solution was immediately added to the mix described below in order to determine BH 4 . Unless stated otherwise, rates of BH 4 synthesis were determined from an average of the rates determined at 30 and 60 s. BH 4 formed in the above reaction was determined from the amount of [ 3 H]tyrosine formed in the BH 4 -dependent hydroxylation of [ 3 H] phenylalanine catalyzed by rat liver PAH. The reaction mixture (35 l) contained 0.1 mg ml Ϫ1 catalase, 130 mM potassium phosphate buffer, pH 6.8, 0.1 mM lysolecithin, 20 M [ 3 H]phenylalanine (7.5 Ci), and 1.1 g of PAH. The mixture was incubated for 30 min at 25°C; the reaction was subsequently stopped by adding 20 l of a solution containing 5% (w/v) trichloroacetic acid, 0.1 mM phenylalanine, and 0.1 mM [ 14 C]tyrosine (30 nCi), added to correct for losses during subsequent purification by high performance liquid chromatography (HPLC).
[ 3 H, 14 C]Tyrosine was purified by reverse phase HPLC (15). For amounts of [ 3 H]tyrosine less than 5 pmol, the product from the HPLC step was mixed with additional carrier tyrosine and further purified by crystallization (15 4 . The efficiency was usually approximately 70% and somewhat lower for amounts of BH 4 below 25 pmol. These procedures were also used for determination of the reduction products of 7,8-BH 2 and q6MPH 2 ; the reduction product of q6MPH 2 , 6MPH 4 , is also a substrate for PAH (30).
Simultaneous Measurement of qBH 2 Reduction and NOS Activity-Relative rates of conversion of arginine to citrulline and of qBH 2 reduction were determined in a single reaction mixture, the composition of which was identical to that described for assay of qBH 2  Inhibition by DPI-Because inhibition by DPI is known to be irreversible and time-dependent (31,32), DPI was first preincubated with NOS for a fixed time, followed by a 10-fold dilution of this preincubation mixture for determination of NOS activity. NOS (2.7 g) was preincubated at 25°C for 5 min in 2 l of standard assay mixture in which qBH 2 was omitted and the concentration of CaM was 450 g/ml and which contained 50 M DPI and 100 M NNA. The mixture was then diluted to 20 l with standard assay mixture containing 100 M NNA but no qBH 2 . Finally, the reaction was started by adding 25 M qBH 2 . Further analysis was performed as described above, except that BH 4 standards contained the same amount of DPI as the reaction mixture.
Inhibition by 7NI-The effect of 7NI (100 M) on qBH 2 reduction was determined in the standard assay mixture modified by the addition of 15 M arginine. Inhibition of NOS activity (citrulline formation) was determined in the standard assay mixture for qBH 2 reduction modified by the omission of qBH 2  Stimulation by qBH 2 and BH 4 of NOS Activity and N G -Hydroxyarginine Synthesis-NOS activity was measured as described previously (15) with the exceptions that the reaction mixture was buffered with 25 mM sodium bicine, pH 7.6, NADPH concentration was decreased to 1 M in the presence of the NADPH regenerating system described above, pterin was added at the concentrations specified, and incubation was for 10 min at 37°C. N G -Hydroxyarginine (NHA) formation in the absence of added NADPH during an incubation period of 5 min at 25°C was measured as described (25).
Cytochrome c Reduction Assay-Initial rates of cytochrome c reduction were measured spectrophotometrically at 550 nm. Conditions were identical to those for qBH 2 reduction except that qBH 2 was omitted, and the reaction (1 ml) contained 11 g of CaM and 100 M cytochrome c. Table I summarizes relative rates of pterin reduction under a variety of conditions. The addition of arginine (100 M) caused a small (25%) inhibition of the reduction rate, whereas NNA (100 M), which inhibits election transfer through heme (33), stimulated the rate between 25 and 50%. Because of the higher reduction rates in the presence of NNA, this compound was included in these assays. Reduction of qBH 2 was stimulated 4 -10-fold by Ca 2ϩ /CaM. Replacement of NADPH by NADH decreased the activity to 10%. Of the limited number of pterins tested, q6MPH 2 was the most active, showing twice the activity compared with qBH 2 . 7,8-BH 2 was neither active itself, nor did it inhibit qBH 2 reduction. Added BH 4 also did not inhibit qBH 2 reduction. Because large amounts of superoxide can be generated in the absence of arginine (34), we examined the effects of superoxide dismutase (1000 units/ml) plus catalase (20 g/ml) on the reduction; little (Ͻ10%) inhibition was observed. NADPH-cytochrome c reductase activity of NOS showed a similar stimulation (8-fold) by Ca 2ϩ /CaM and complete insensitivity to superoxide dismutase plus catalase at the concentrations de-  scribed above (data not shown).

Conditions for Pterin Reduction-
Comparative Rates of qBH 2 Reduction and Citrulline Formation-As shown in Fig. 1, the rate of reduction of qBH 2 at a concentration of 25 M catalyzed by NOS is comparable with that of citrulline formation measured under the same conditions. In contrast to citrulline formation, qBH 2 reduction was not linear with time, probably as a result of the decrease in qBH 2 concentration caused by its enzymic reduction and by its marked instability under the assay conditions.
The Effect of Different qBH 2 Concentrations on Its Rate of Reduction-The qBH 2 reduction rate increased with qBH 2 concentration measured in the range of 2.0 -50 M (Fig. 2). The effects of higher concentrations of qBH 2 were not examined, because they resulted in unacceptably high blank values; therefore, a precise K m was not determined. However, the data show that the concentration of qBH 2 required for a half-maximal rate of reduction of this pterin (EC 50 ) exceeds 20 M and is thus some 2 orders of magnitude greater than the concentration of BH 4 required for half-maximal activation of NOS (Fig. 3).
Effects of DPI, 7NI, and Methotrexate-The effects of three potential inhibitors were studied. Preincubation of NOS for 5 min with the flavoprotein inhibitor DPI (31, 32) at a concentration of 50 M inhibited qBH 2 reduction by 85%. 7NI is believed to bind to heme and interferes with BH 4 binding (14,35). This compound stimulated qBH 2 reduction 2-fold at a concentration (100 M) that inhibited the formation of citrulline from arginine by 80%. Methotrexate, a pterin analogue that is known to inhibit a variety of enzymes involved in dihydrobiopterin reduction (see "Discussion"), had no effect on qBH 2 reduction at a concentration of 100 M.
Stimulation of NOS Activity by BH 4 and qBH 2 - Fig. 3 compares the relative stimulation of the NOS reaction (citrulline synthesis) by added qBH 2 or BH 4 . The stimulation curve with qBH 2 is similar to that with BH 4 , except that it is shifted to higher concentrations and shows a decrease at higher concentrations of qBH 2 . This decrease in stimulation may be due to the presence of 7,8-BH 2 formed nonenzymically from qBH 2 under the assay conditions. 7,8-BH 2 has been shown to attenuate BH 4 stimulation (14, 16). qBH 2 had little or no stimulatory effect on the NOS-catalyzed conversion of arginine to NHA measured in the absence of added NADPH (data not shown).

DISCUSSION
The novel method described here allows for the sensitive and specific determination of BH 4 in the presence of relatively large amounts of qBH 2 . BH 4 formed during reduction of qBH 2 is coupled to the PAH-catalyzed conversion of an equivalent amount of [ 3 H]phenylalanine to [ 3 H]tyrosine. The procedure proved to be reproducible and easy to use and avoids HPLC of unstable BH 4 as required in other methods (36,37).
A proposed scheme for qBH 2 reduction by NOS is presented in Fig. 4. The strong inhibition of qBH 2 reduction by DPI demonstrates a role for flavin. Three enzymes, one of which is also a flavoprotein, have previously been reported to catalyze the pyridine-nucleotide-linked reduction of dihydrobiopterin. They are dihydropteridine reductase (38,39) and the flavopro- tein methylenetetrahydrofolate reductase (40,41), both of which catalyze qBH 2 reduction, and dihydrofolate reductase, which catalyzes the reduction of 7,8-BH 2 (42). Methotrexate is an extraordinarily potent inhibitor of dihydrofolate reductase (43) and causes appreciable inhibition of the other enzymes at the concentration of 100 M (38,44,45) used in our work. The sensitivity of these enzymes to methotrexate is in sharp contrast to the observed insensitivity of qBH 2 reduction by NOS to this pterin analogue. The absence of an inhibitory effect of methotrexate on qBH 2 reduction is especially relevant in the light of the recent report of a "dihydrofolate reductase module" on the oxidase domain of NOS (5).
What is the site of qBH 2 reduction? Does reduction proceed directly at the flavoprotein NADPH "diaphorase" site of NOS, or are electrons transferred from flavin to the high affinity BH 4 -binding site involved in the activation of NOS (14,15)? Two main findings indicate that the flavoprotein diaphorase is the site of qBH 2 reduction. First, BH 4 and 7,8-BH 2 , two compounds with a high affinity for the BH 4 -binding site (14,15), have no effect on qBH 2 reduction. Second, 7NI, which interferes with BH 4 binding to the high affinity site (14,35), stimulates qBH 2 reduction. This stimulation, as well as that by NNA, can be tentatively ascribed to the inhibition of reactions proceeding through heme (14,33,35), thereby increasing the availability of reducing equivalents for qBH 2 reduction. The high EC 50 (Fig. 2) for qBH 2 reduction compared with the relatively low K m values for binding and activation with BH 4 (Fig.  3, (13)(14)(15)) is consistent with qBH 2 reduction proceeding at the diaphorase site, rather than the high affinity BH 4 -binding site.
The NADPH diaphorase component of NOS is known to catalyze the oxidation of NADPH with a range of electron acceptors including cytochrome c, ferricyanide, and various dyes (11, 46 -49). Reduction of q6MPH 2 at a rate of approximately twice that of qBH 2 is consistent with this reported low specificity; the product of q6MPH 2 reduction, 6MPH 4 , shows little or no activation of NOS at concentrations optimal for activation by BH 4 (13,15).
There are conflicting data on whether superoxide is an intermediate in the NADPH-cytochrome c reductase activity of NOS (35,46,48,49). It was therefore of interest to examine the effects of superoxide dismutase on this reaction and also on the NOS-catalyzed reduction of qBH 2 . Our finding that superoxide dismutase had no effect on the NADPH-cytochrome c reductase activity of NOS is in agreement with other reports (35,46,48). Furthermore, the small inhibition of qBH 2 reduction by superoxide dismutase further suggests that superoxide is not involved in this reaction.
Our studies show that the reduction of both qBH 2 and cytochrome c by NOS is strongly stimulated by Ca 2ϩ /CaM. Other workers have also reported a strong stimulation of the reduction of the latter substrate by Ca 2ϩ /CaM; by contrast, for unexplained reasons, the NADPH-dependent reduction of ferricyanide and 2,6-dichlorophenolindophenol by NOS is stimulated to only a small extent or not at all by Ca 2ϩ /CaM (46,48,49). With respect to dependence on Ca 2ϩ /CaM, qBH 2 reduction therefore more closely resembles cytochrome c reduction than ferricyanide or 2,6-dichlorophenolindophenol reduction.
Under the conditions used in our earlier experiment (15), where the PAH reaction would have been expected to generate a steady-state concentration of about 0.05 M qBH 2 from the added 0.05 M BH 4 and the limits of detection were 2.5 pmol, there was no detectable regeneration of BH 4 in the presence of added NOS and NADPH. Because this result appears to contrast with our present data, it is important to understand why the earlier experiment did not detect any reduction of qBH 2 . The explanation, in all probability, is that in the earlier experiment, the amount of qBH 2 converted to BH 4 was below the limits of detection. Thus, from the data in Fig. 2, it can be estimated that 2 pmol of BH 4 would have been synthesized during the 30-min incubation period used in the earlier experiment; this amount would be considerably less if the instability of qBH 2 and the lack of linearity of BH 4 synthesis with time ( Fig. 1) are taken into account.
The significance of the qBH 2 reduction described here to the essential role of BH 4 in NOS remains to be clarified. Our results indicate that at the relatively high concentration of 25 M, qBH 2 reduction can proceed at a rate comparable with that of the overall reaction (citrulline synthesis) and that the efficiency of added qBH 2 in stimulating NOS activity approaches that of added BH 4 . QBH 2 seems to stimulate by first being converted to BH 4 and subsequent binding to the BH 4 -binding site. This postulate is supported by the observation that added qBH 2 causes little or no stimulation of NOS-catalyzed oxygenation of arginine to NHA (Fig. 4), a reaction that proceeds in the absence of added NADPH and consequently precludes qBH 2 reduction.
A central feature of the model proposed in Fig. 4 is that qBH 2 reduction proceeds on the C-terminal (reductase) domain, remote from the high affinity BH 4 -binding site of the N-terminal (oxidase) domain. Implicit in this model is the notion that if BH 4 recycling does occur, it may proceed by a pathway in which qBH 2 reduction occurs predominantly at the flavin site, followed by transfer of the BH 4 so formed to the high affinity BH 4 -binding site. A redox cycle would be completed by transfer of qBH 2 back to the flavin reduction site. Transfer of pterins between the flavin and BH 4 -binding sites may proceed either by dissociation from the enzyme or by a direct interdomain transfer. Our present studies also do not eliminate the possibility that cycling of pterin occurs exclusively at the BH 4binding site on the oxygenase domain. These possibilities are currently under investigation.