Cu2+ and Zn2+ inhibit nitric-oxide synthase through an interaction with the reductase domain.

Cu(2+) and Zn(2+) inhibit all of the NADPH-dependent reactions catalyzed by neuronal nitric-oxide synthase (nNOS) including ferricytochrome c reduction, NADPH oxidation, and citrulline formation. Cu(2+) and Zn(2+) also inhibit ferricytochrome c reduction by the independent reductase domain. Zn(2+) affects all activities of the full-length nNOS and the reductase domain to the same extent (estimated IC(50) values from 9 to 31 microm), suggesting Zn(2+) occupation of a single site in the reductase domain. Citrulline formation and NADPH oxidation by the full-length nNOS and ferricytochrome c reduction by the reductase domain are affected similarly by Cu(2+), with estimated IC(50) values ranging from 6 to 33 microm. However, Cu(2+) inhibits ferricytochrome c reduction by the full-length nNOS 2 orders of magnitude more potently, with an estimated IC(50) value of 0.12 microm. These data suggest the possibility that Cu(2+) may interact with nNOS at two sites, one composed exclusively of the reductase domain (which is perhaps also involved in Zn(2+)-mediated inhibition), and another that includes components of both domains. Occupation of the second (higher affinity) site could then promote the selective inhibition of ferricytochrome c reduction in full-length nNOS. Neither the inhibition by Cu(2+) nor that by Zn(2+) is dependent on calmodulin.

Nitric oxide (NO) 1 has been shown to be a participant in many physiological processes ranging from intercellular signal transduction to the immune response. NO-influenced events include smooth muscle relaxation, neurotransmission, circadian shifts, apoptosis, long term potentiation, synaptogenesis, development, and pathogen cytostasis (1)(2)(3)(4)(5). Nitric-oxide synthase (NOS; EC 1.14.13.39) catalyzes the biosynthesis of NO through an oxidative reaction on the amino acid L-arginine that ultimately results in carbon-nitrogen bond cleavage to yield citrulline and NO. The only known intermediate in this reaction is N G -hydroxy-L-arginine (6,7). NOS is a homodimer consisting of two domains (one oxygenase and one reductase) per polypeptide chain. The NOS oxygenase domain stoichiometrically binds protoporphyrin IX type heme and (6R)-5,6,7,8-tet-rahydro-L-biopterin (H 4 B). The heme iron is ligated to the NOS polypeptide chain through a cysteine thiolate; thus, the heme is a member of the cytochrome P450 class of enzymes and displays optical and magnetic properties typical of other known cytochrome P450s (8 -12). Together with the heme and H 4 B cofactors, the oxygenase domain forms the active site of the enzyme (13).
Three sets of crystal structures of the dimeric form of the oxygenase domain have been reported recently (14 -16). Although no structures are currently available for either the reductase domain or the full-length NOS, it is known that each reductase domain binds one equivalent of FMN and one equivalent of FAD (17,18). The reductase domain shuttles NADPHderived electrons to the active site heme to support catalysis in the oxygenase domain. Presumably, electron transfer proceeds in a manner analogous to that in cytochrome P450 reductase (CPR), another two-flavin reductase. Like CPR, the nNOS reductase domain contains an air-stable neutral semiquinone radical on its high potential flavin (known in CPR, and presumed in NOS, to be FMN) (8, 12, 19 -21). Another similarity between CPR and the neuronal NOS (nNOS) reductase domain is the ability of each to reduce artificial electron acceptors such as ferricytochrome c, ferricyanide, and 2,6-dichlorophenolindophenol (20,(22)(23)(24).
The oxygenase and reductase domains of NOS are linked by a region of the polypeptide that recognizes Ca 2ϩ -bound calmodulin (CaM). CaM binding accelerates both intradomain electron transfer (from NADPH through the reductase domain) and interdomain electron transfer (from the reductase to the oxygenase domain) (25)(26)(27). Because NOS has been heterologously expressed in Escherichia coli, a variety of genetic manipulations (including site-directed mutagenesis and the independent expression of each domain) are possible (28). Alternatively, CaM-responsive reductase domain can be obtained as a proteolytic byproduct of the full-length nNOS E. coli expression and purification procedure (21).
In this report, we have used the CaM-responsive nNOS reductase domain (generated by the latter method) in conjunction with the full-length nNOS to demonstrate that two known NOS inhibitors (Cu 2ϩ and Zn 2ϩ ) exert their effects on the reductase domain of NOS (29,30). These results are significant because they reconcile existing biochemical data (that certain transition metals, including Zn 2ϩ , inhibit NOS catalysis) with structural arguments, which suggest that Zn 2ϩ bound to the oxygenase domain may actually be pivotal (i.e. essential and noninhibitory) for proper H 4 B orientation and/or L-arginine binding (15,16). In addition, complementary results from this study have enabled us to propose a mechanism of ferricytochrome c reduction by the nNOS reductase domain that resolves the differences between previously reported sets of data (22,24,27). The proposed mechanism invokes the in situ production of superoxide to bridge the high potential flavin of the * This work was supported by the National Institutes of Health (Grant CA50414) and by the Howard Hughes Medical Institute. 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.
ʈ To whom correspondence should be addressed. reductase domain and the heme of ferricytochrome c. These results are also significant because they highlight a potential structural difference between the nNOS reductase domain and CPR, namely, that the high potential flavin of the nNOS reductase domain is not as surface-accessible as that of CPR.

Enzyme Overproduction and Purification
Full-length nNOS and the nNOS reductase domain were overproduced and purified as described previously (21). Briefly, four 1-liter cultures of DH5␣ E. coli cells containing the pCWori-nNOS expression plasmid were grown at 37°C to an OD 600 of 0.8 and subsequently cooled to 25°C prior to the induction of nNOS expression by the addition of 1 mM isopropyl-␤-D-thiogalactopyranoside. Cells were harvested by centrifugation 18 -21 h postinduction. The nNOS purification involves three steps: CaM affinity chromatography followed by ADP-affinity chromatography, and ultimately, gel filtration on Superdex-200 media. CaM-responsive reductase domain copurifies with the full-length nNOS up to the gel filtration step where they are resolved from one another, each to homogeneity.

Enzyme Activity Assays
Ferricytochrome c Reduction-Both the full-length nNOS and the reductase domain can reduce ferricytochrome c. The rate of this reaction was measured by monitoring the increase in absorbance at 550 nm (⌬⑀ ϭ 21,000 M Ϫ1 cm Ϫ1 ) over time. Each assay contained 2.5 mM CaCl 2 , 20 g of CaM, and 50 M cytochrome c in 50 mM HEPES buffer (pH 7.4) unless otherwise specified. Various concentrations of full-length nNOS or reductase domain were used depending on the nature of the experiment. Full-length nNOS assays also contained 5 M H 4 B and 50 M DTT, neither of which reduced ferricytochrome c at the specified concentrations. In addition, the indicated assays contained varying concentrations of transition metals, all added as their chloride salts. All ferricytochrome c reduction assays were performed at 25°C in a total volume of 500 L and were initiated by the addition of 500 M NADPH. SOD Activity-SOD was assayed following a method similar to that published by Beyer and Fridovich (31). In this assay, the activity of SOD is measured as the inhibition of superoxide-dependent reduction of ferricytochrome c. The xanthine/xanthine oxidase system is used to generate superoxide, and the reaction is monitored at 550 nm. The assay contained 200 M xanthine, 10 nM xanthine oxidase, 50 M cytochrome c, and 100 units of Mn-SOD. SOD assays (total volume of 500 L) were run at 25°C in 50 mM HEPES (pH 7.4). Xanthine oxidase activity was confirmed prior to the SOD assays by following the increase in absorbance at 293 nm resulting from the conversion of xanthine to urate. Xanthine oxidase assays (total volume of 1 mL) con-tained 200 M xanthine and 10 nM xanthine oxidase and were run at 25°C in 50 mM HEPES (pH 7.4).

Electronic Absorption Spectroscopy
All spectra and kinetic traces were recorded on a Varian Cary 3E spectrophotometer at a constant temperature of 25°C maintained by a Neslab circulating water bath.

RESULTS
The Inhibition of Ferricytochrome c Reduction by Cu 2ϩ and Zn 2ϩ -Inhibition of ferricytochrome c reductase activity is shown in Fig. 1. Cu 2ϩ and Zn 2ϩ each inhibit this activity in both the reductase domain and the full-length enzyme in a concentration dependent manner; however, the inhibition brought about by Cu 2ϩ is more pronounced in the presence of the oxygenase domain (with the full-length nNOS). That is, while Zn 2ϩ inhibited ferricytochrome c reduction by the fulllength nNOS and the independent reductase domain to similar extents, the inhibition observed with Cu 2ϩ was two orders of magnitude more potent with the full-length nNOS than it was with the reductase domain (the IC 50 values were 0.12 and 12 M, respectively).
The Effect of Cu 2ϩ and Zn 2ϩ on CaM-Ferricytochrome c reduction activities in both full-length nNOS and the reductase domain are greatly enhanced by association with Ca 2ϩ -bound CaM. We therefore had to rule out the possibility that the observed inhibition was simply a consequence of the transition metals either (i) competing for the required Ca 2ϩ or (ii) binding to one or more of the six auxiliary cation binding sites on CaM, which could distort its structure, thereby rendering it inactive. In the presence of 1 mM Ca 2ϩ , it had previously been shown that each of these auxiliary sites can bind Zn 2ϩ with a K d of approximately 833 M (32). This concentration of Zn 2ϩ greatly exceeds the levels used here; however, it is conceivable that, upon association with nNOS (or its reductase domain), CaM attains a conformation more amenable to Zn 2ϩ binding, which in turn results in the observed inhibition. Fig. 2 examines the possibility of CaM inhibition by using the reductase domain. The figure demonstrates that the inhibition of reductase domain mediated ferricytochrome c reduction by Cu 2ϩ and Zn 2ϩ does not, in fact, even require CaM. This result is consistent with the metals interacting with the reductase domain itself.
Inhibition of NADPH Oxidation by Cu 2ϩ and Zn 2ϩ -It was also possible that the observed inhibition of ferricytochrome c reduction was related to Cu 2ϩ and Zn 2ϩ modulating the ability of ferricytochrome c to accept electrons from the reductase domain. This could happen either by (i) metal binding to and altering the conformation of ferricytochrome c, somehow making it less susceptible to reduction or (ii) by a metal-induced disruption of the presumed interaction between nNOS and cytochrome c (24). To address this possibility, we performed NADPH oxidation assays on the full-length nNOS. In this case, the electron acceptor is the NOS heme, and ultimately, molecular oxygen. Therefore, it is an internal (interdomain) electron transfer that does not rely on any interaction with, or structural integrity of, an artificial acceptor molecule. Furthermore, inhibition of NADPH oxidation by Cu 2ϩ and Zn 2ϩ would suggest that inhibition of ferricytochrome c reduction was not caused by an effect by the metals on either cytochrome c itself, or cytochrome c-NOS docking. Fig. 3 shows that NADPH oxidation by the full-length nNOS is in fact inhibited by both Cu 2ϩ and Zn 2ϩ in a concentration-dependent manner. The estimated IC 50 values for the observed inhibition are 6 and 9 M, respectively.
Activity Correlation-Having demonstrated that the func-tion of the reductase domain is impaired by Cu 2ϩ and Zn 2ϩ , we sought to correlate the inhibition of NADPH oxidation and ferricytochrome c reduction with the previously reported inhibition of NOS-catalyzed citrulline formation by these metals (29,30). Such a correlation for each metal is shown in Fig. 4, and the IC 50 values for the inhibition of each activity are summarized in Table I. With Zn 2ϩ (Fig. 4A), the inhibition data for all three full-length nNOS activities are similar, which suggests that the ability of Zn 2ϩ to inhibit citrulline formation can be traced to an interaction that involves the reductase domain, exclusively. The situation with Cu 2ϩ (Fig. 4B) is not quite as simple, indicating that Cu 2ϩ may not interact with nNOS in precisely the same manner as Zn 2ϩ . Nevertheless, the results with Cu 2ϩ remain consistent with the conclusion that the observed inhibition of citrulline formation is also a consequence of an interaction between the added metal and the reductase domain. The Effect of Cu 2ϩ or Zn 2ϩ on Flavin Reduction-A stoppedflow kinetic analysis of flavin reduction shows that NADPH can efficiently reduce the low potential flavin of the reductase domain in the presence of an inhibitory concentration of either Cu 2ϩ (50 M) or Zn 2ϩ (100 M). Fig. 5A compares the rates of flavin reduction (as measured by the change in absorbance at 454 nm) observed under these conditions with that observed in the absence of exogenous transition metals. Data collected from reactions that contained no exogenous transition metals as well as those collected from reactions that included 100 M Zn 2ϩ were best fit to single exponential equations to generate pseudo first-order rate constants (k obs ) of 60 and 46 s Ϫ1 , respectively (at 25°C). Data collected from the reactions that contained Cu 2ϩ were not well fit by a similar equation; however, full spectra collected by diode array 2.5 s postmixing confirmed the reduction of the low potential flavin (Fig. 5B).
The Effect of Air on Ferricytochrome c Reduction-Electronic absorption spectra of the visible region of cytochrome c in the presence of CaM-bound reductase domain (Fig. 6A) demonstrate the dramatic effect of air (presumably mediated by oxygen) on the rate of ferricytochrome c reduction. Fig. 6B shows the anaerobic chemical reduction of ferricytochrome c, demonstrating that the emerging species in the upper panel is indeed the reduced form of cytochrome c and that the formation of this species does not require oxygen. In addition, we show in Fig. 7 that the effect of air on the rate of ferricytochrome c reduction is not exclusive to the CaM-bound reductase domain. Although  this effect is more pronounced in the presence of CaM (over a 20-fold difference in rate, Table II), anaerobic reactions performed in the absence of CaM still show a decrease in the rate relative to comparable aerobic reactions, from 0.54 to 0.20 mol min Ϫ1 mg Ϫ1 .
The Effect of SOD on Ferricytochrome c Reduction-Whether or not ferricytochrome c reduction by nNOS can be inhibited by superoxide dismutase (SOD) has been a source of considerable debate in the literature (22,24,27,33). Because we observed an air-dependent rate with the reductase domain, experiments addressing the effect of SOD were carried out. When assaying the nNOS reductase domain, the inclusion of up to 1000 units of SOD had virtually no effect on the rate of ferricytochrome c reduction (data not shown). In contrast to this result, 100 units of the same preparation of SOD completely inhibited the superoxide-mediated reduction of ferricytochrome c in the presence of 10 nM xanthine oxidase and 200 M xanthine (0.3 nmol min Ϫ1 inhibited to 0.0 nmol min Ϫ1 , data not shown).
The Effect of Zn 2ϩ on Anaerobic Ferricytochrome c Reduction-To test whether the inhibition by Cu 2ϩ and Zn 2ϩ was air-dependent, assays measuring the anaerobic rate of ferricytochrome c reduction in the presence of various amounts of Zn 2ϩ were carried out. The results are compared with similar assays performed under aerobic conditions in Table II. For these experiments, the use of Cu 2ϩ was avoided to circumvent the potential complication that under anaerobic conditions Cu 2ϩ could serve as a stable electron trap (Cu ϩ ), and in so doing, give an overestimation of the inhibition of ferricytochrome c reduction. Table II shows that Zn 2ϩ does in fact inhibit anaerobic ferricytochrome c reduction to an extent similar to that observed in aerobic assays. In fact, the potency of Zn 2ϩ seems, if anything, slightly greater in the anaerobic assays. The observed inhibition in the presence of 20 M Zn 2ϩ is approximately 50% in aerobic assays and approximately 75% in anaerobic assays. At 80 M Zn 2ϩ , the observed inhibition is 95% and 99%, respectively.

DISCUSSION
The data presented in this report all support the conclusion that Cu 2ϩ and Zn 2ϩ inhibit nNOS catalysis as a result of a direct interaction with the reductase domain of the enzyme. All three commonly assayed full-length NOS activities (citrulline formation, ferricytochrome c reduction, and NADPH oxidation) were inhibited by both Cu 2ϩ and Zn 2ϩ . Likewise, Cu 2ϩ and  cyt-c reduction 33 12 a Cu 2ϩ and Zn 2ϩ were both added as the dichloride salt. b cyt-c reduction, ferricytochrome c reduction. The reason for this difference is not clear; however, because the components of the assay are otherwise equivalent, it suggests that the oxygenase domain of NOS influences Cu 2ϩ binding to the reductase domain. There are many ways in which the oxygenase domain could exert such an effect. For example, it could contribute to Cu 2ϩ binding directly, by (for example) donating a metal ligand. Alternatively, the oxygenase domain could influence the structure of reductase domain by domaindomain contacts, perhaps thereby altering its conformation to a state more amenable to Cu 2ϩ binding. A third possibility is that the conformation of the full-length nNOS is modified by the docking of cytochrome c. In this case, the apparent high affinity for Cu 2ϩ (which is manifested by its effect on ferricytochrome c reduction) results from a composite of opposing structural effects; that is, the oxygenase domain and cytochrome c synergistically affect Cu 2ϩ binding to the reductase domain. This explanation is consistent with the data for Cu 2ϩ -mediated inhibition of citrulline formation and NADPH oxidation in the full-length nNOS (the estimated IC 50 values were 19 and 6 M, respectively), because cytochrome c is present in neither of these assays. Yet another possibility is that there is more than one binding site available to Cu 2ϩ in the full-length nNOS, one of which (that of apparent higher affinity) requires the presence of both domains. Because electrons are presumably routed in different directions depending on whether they serve to reduce the NOS heme or that of ferricytochrome c, it is conceivable that Cu 2ϩ could bind to one site (that of higher affinity) that would result in the selective inhibition of ferricytochrome c reduction without eliciting an inhibitory effect on electron transfer going in the other direction; hence, NADPH oxidation and citrulline formation (both of which are dependent on NOS heme reduction) are not affected. At higher concentrations of Cu 2ϩ , when the second (lower affinity) site would be occupied, electron transfer from the reductase domain is interrupted, manifested by the inhibition of all three activities. This not only explains the difference between the various activities presented in the Cu 2ϩ activity correlation curve but also the   All assays were run in the presence of 2.5 mM Ca 2ϩ and 20 g of CaM, and the activities are expressed as mol min Ϫ1 mg Ϫ1 . a The first column refers to the final concentration of Zn 2ϩ (added as ZnCl 2 ) in the assay reaction cuvette.
b Anaerobic samples were prepared by using repeated cycles of alternate evacuation and argon purging of a sealed reaction cuvette using a gas train constructed in our laboratory. difference between the magnitude of Cu 2ϩ -mediated inhibition of ferricytochrome c reduction observed with the full-length NOS as compared with that observed with the reductase domain. The independent domain would not have the first (higher affinity) site intact, and the observed inhibition would necessarily be a consequence of Cu 2ϩ interacting with the second (lower affinity) site. In support of this, the estimated IC 50 of Cu 2ϩ with the reductase domain was 12 M, which is similar to the IC 50 values observed for both citrulline formation and NADPH oxidation by the full-length NOS (19 and 6 M, respectively). In addition, the presence of a high affinity Cu 2ϩ binding site in the full-length nNOS is consistent with our previous findings that nNOS overproduced in E. coli can be purified with one equivalent of bound copper per subunit (30). Because the final step of the nNOS purification is gel filtration, one would not expect the second (lower affinity) site to be occupied until the enzyme is assayed in the presence of micromolar levels of Cu 2ϩ .
The situation with Zn 2ϩ is less complicated; all three fulllength NOS activities and the reductase domain activity are modulated similarly. The estimated IC 50 values of Zn 2ϩ required to affect these processes fall between 9 and 31 M, in the same range as the IC 50 values for the Cu 2ϩ effects caused by its interaction at the hypothesized second (lower affinity) site. Because Zn 2ϩ inhibition of the reductase domain correlates well with its inhibition of full-length nNOS activities (citrulline formation, NADPH oxidation, ferricytochrome c reduction), it is reasonable to conclude that an interaction between Zn 2ϩ and the reductase domain can account for all of its observed inhibitory effects on the full-length nNOS. This is of considerable significance because recent crystallographic studies of the NOS oxygenase domain have demonstrated that a Zn 2ϩ atom can bind at the dimeric interface of the enzyme (15,16), and this site had previously been hypothesized to be the site of Zn 2ϩ inhibition (29). Structural studies have led to the conclusion that Zn 2ϩ binding to the oxygenase domain is essential to the stability of the dimer and, in addition, serves to maintain the proper orientation of H 4 B and L-arginine. The current results are not inconsistent with such an argument because they point to a different site that can account for the observed NOS inhibition by Zn 2ϩ , i.e. in the reductase domain.
The precise mechanism of inhibition of nNOS and its reductase domain by Cu 2ϩ and Zn 2ϩ is not completely clear; however, the results presented here combined with previous results from our laboratory and others enable us to make some suggestions. From this report it is clear that an electron transfer reaction of the reductase domain is inhibited. This inhibition is not exclusive to the CaM-activated domain, nor does it require oxygen. In addition, we have ruled out the possibility that the observed inhibition is a consequence of the inhibition of NADPH-mediated reduction of the low-potential flavin. These results are consistent with previous studies using electronic absorption spectroscopy to show that in the presence of 100 M Cu 2ϩ (enough to completely inhibit citrulline formation by NOS), NADPH could not reduce the heme iron, although flavin reduction was observed (30). From the current data there are two possibilities: (i) inhibition of interflavin electron transfer or (ii) inhibition of electron transfer from the high-potential flavin to the acceptor molecule. The redox potentials of the NOS flavins are not currently known, and it is difficult to discern the extent of flavin reduction (i.e. to determine the percentage of two-, three-, and four-electron reduced states in a given population of NADPH-reduced enzyme by the optical spectrum of the reaction at steady state). So, even though it is unlikely that at steady state the low potential flavin is reduced by two electrons while the high potential flavin is reduced by one, if the redox potentials of the two semiquinones were comparable and metal binding somehow inhibits interflavin electron transfer, this situation could be possible. Additionally, it would be impossible to make a definitive distinction between FMNH ⅐ / FADH 2 and FMNH 2 /FADH ⅐ from absorption spectra of fulllength NOS. In short, the possibility of inhibition of interflavin electron transfer seems unlikely, but it cannot be eliminated. The more likely explanation (see (ii) above) is that Cu 2ϩ and Zn 2ϩ inhibit electron transfer from the high-potential flavin to the acceptor molecule (either the heme of nNOS or that of ferricytochrome c). There are many ways in which this could happen. Three possibilities are: 1) a metal-induced rearrangement of the nNOS polypeptide that (in 3-dimensional space) is between the high potential flavin and the acceptor molecule, perhaps thereby repositioning participating amino acid residues or setting up a local environment that is not conducive to electron transfer; 2) a structural perturbation that results in the movement of the high potential flavin further away from the acceptor molecule; or 3) the binding of the metal directly to the N5 and/or O4 of the isoalloxazine ring of the high potential flavin, sequestering the redox-active portion of the molecule, and thereby interfering with electron transfer. It is known that Cu 2ϩ can indeed bind to a flavin in this manner, and a structure of this species crystallized from aqueous solution has been solved (34). Zn 2ϩ can also bind to flavin in this manner, but its binding is more favored in hydrophobic environments; however, this environment may very well be provided by the reductase domain (34). An interesting possibility with particular relevance to Cu 2ϩ is that metal binding occurs with a semiquinone radical during the electron transfer process. In this instance, the complex may be more favorable because the Cu 2ϩ / semiquinone is resonance-stabilized with a Cu ϩ /oxidized flavin species (35,36). Such an hypothesis could be further substantiated by electron paramagnetic resonance experiments (performed anaerobically) in which the reductase domain was shown to reduce Cu 2ϩ following the anaerobic addition of NADPH.
It has previously been hypothesized that the reason for the observed CaM-dependent enhancement of ferricytochrome c reduction by full-length nNOS was that electrons could pass to the heme and subsequently be consumed by oxygen. This process would generate superoxide which could then diffuse in solution to the terminal acceptor, ferricytochrome c (22). It was later shown that a heme-ligand mutant (nNOS C415A) retained the property of CaM-dependent acceleration of ferricytochrome c reduction; however, the potential role of dioxygen, if any, was unclear (33). Our results showing the air enhancement of ferricytochrome c reduction by the reductase domain suggest that dioxygen can play a role in this process, and, that the high potential flavin of nNOS may be positioned further away from the acceptor molecule (and hence the surface of the enzyme) than that of CPR. This hypothesis is consistent with the results presented in our previous work which show that the rapid relaxation properties of the high potential flavin radical are imparted by elements of the reductase domain, implying that the flavin is somewhat buried (21). If the high potential flavin was at the surface of the enzyme, the radical spin would be expected to resemble that of an isolated free radical, and relax much more slowly. CPR, however, is believed to directly reduce its substrates by one electron from FADH ⅐ /FMNH 2 , transiently yielding FADH ⅐ /FMNH ⅐ , which rearranges to FAD/ FMNH 2 , donates another electron, and returns to FAD/ FMNH ⅐ , the species that is reduced by NADPH (19,20,23). For dioxygen to mediate the one electron reduction of ferricytochrome c, it is presumably reduced to superoxide by the hydroquinone of the high potential flavin of nNOS. However, the reaction was not inhibited by SOD, which suggests that superoxide does not escape the flavin "active site". Rather, superoxide must be generated in situ as dioxygen can serve to bridge the two redox centers.
In conclusion, we have shown that Cu 2ϩ and Zn 2ϩ interact with the reductase domain of nNOS to elicit their inhibitory effects. The underlying cause of all of the observed effects of these metals (inhibition of ferricytochrome c reduction, NADPH oxidation, and citrulline formation) is the inhibition of electron transfer, presumably that from the high potential flavin of the reductase domain to the acceptor heme iron of either ferricytochrome c or nNOS. In addition, we have given evidence that dioxygen facilitates electron transfer to ferricytochrome c.