Role of Reductase Domain Cluster 1 Acidic Residues in Neuronal Nitric-oxide Synthase

The nNOS reductase domain is homologous to cytochrome P450 reductase, which contains two conserved clusters of acidic residues in its FMN module that play varied roles in its electron transfer reactions. To study the role of nNOS reductase domain cluster 1 acidic residues, we mutated two conserved acidic (Asp918 and Glu919) and one conserved aromatic residue (Phe892), and investigated the effect of each mutation on flavin binding, conformational change, electron transfer reactions, calmodulin regulation, and catalytic activities. Each mutation destabilized FMN binding without significantly affecting other aspects including substrate, cofactor or calmodulin binding, or catalytic activities upon FMN reconstitution, indicating the mutational effect was restricted to the FMN module. Characterization of the FMN-depleted mutants showed that bound FMN was essential for reduction of the nNOS heme or cytochrome c, but not for ferricyanide or dichlorophenolindolphenol, and established that the electron transfer path in nNOS is NADPH to FAD to FMN to heme. Steady-state and stopped-flow kinetic analysis revealed a novel role for bound FMN in suppressing FAD reduction by NADPH. The suppression could be relieved either by FMN removal or calmodulin binding. Calmodulin binding induced a conformational change that was restricted to the FMN module. This increased the rate of FMN reduction and triggered electron transfer to the heme. We propose that the FMN module of nNOS is the key positive or negative regulator of electron transfer at all points in nNOS. This distinguishes nNOS from other related flavoproteins, and helps explain the mechanism of calmodulin regulation.

Synthesis of nitric oxide (NO) 1 by the neuronal NO synthase (nNOS) can activate as well as modulate many functions in mammalian physiology (1)(2)(3). The nNOS is inactive in its native form and requires Ca 2ϩ -promoted calmodulin (CaM) binding for activation (4). This enables nNOS to participate in signal transduction cascades by generating NO in response to increases in intracellular Ca 2ϩ (2). nNOS is a bidomain enzyme containing an N-terminal oxygenase domain with binding sites for heme, tetrahydrobiopterin (H 4 B), and L-arginine (Arg), and a C-terminal reductase domain with binding sites for FMN, FAD, and NADPH (5)(6)(7)(8)(9). A ϳ20 amino acid consensus CaMbinding site is located between the nNOS reductase and oxygenase domains (4,10). During NO synthesis the reductase domain transfers electrons from NADPH to the heme. This enables heme-dependent oxygen activation and stepwise conversion of Arg to NO and citrulline, with N-hydroxy-L-arginine (NOHA) being formed as an intermediate (3,11,12). CaM performs a critical role in the process by triggering electron transfer from the reductase domain flavins to the oxygenase domain heme (13,14). Recent evidence suggests this transfer occurs between reductase and oxygenase domains that are located on different subunits of the NOS homodimer (15). However, it is still unclear what protein residues facilitate electron transfer, and how CaM controls the domain interaction.
The NOS reductase domain actually belongs to a subset of related reductases that contain a N-terminal FMN-containing flavodoxin module that is linked to a C-terminal NADPH-and FAD-binding ferridoxin-like module (FNR) (16 -18). Other similar dual-flavin reductases include NADPH cytochrome P450 reductase, sulfite reductase flavoprotein, and the cytochrome P450BM3 reductase domain. Because of the structural homology, work done with these proteins serves to guide investigation of nNOS. In general, the FNR and FMN modules of these proteins appear to fold separately and function when expressed independently or after being separated by proteolysis (16 -18). A crystal structure of cytochrome P450 reductase has recently revealed the interactions that occur between the FNR and FMN modules of that enzyme (17).
All of these proteins (or their isolated reductase domains) share biochemical similarities in transferring NADPH-derived electrons to either hemeprotein acceptors or attached hemeprotein domains (11, 18 -21). Electron transfer typically proceeds from NADPH to FAD to FMN to hemeprotein, although this path has not been definitively demonstrated for nNOS. In cytochrome P450BM3, recent work suggests that its FMN module is capable of interacting with both the FNR and attached hemeprotein domain by means of a flexible linker, and this enables the FMN module to shuttle electrons between the FNR and heme during catalysis (22). All of these flavoproteins also transfer NADPH-derived electrons to artificial acceptors including cytochrome c, ferricyanide, and dichlorophenolindolphenol (DCIP) (11, 18 -21).
Despite the similarities, nNOS is distinguished from these flavoproteins by its ability to increase electron transfer rates to acceptors upon CaM binding (11,14,18). CaM binding is associated with an increase in tryptophan and flavin fluorescence (19,23), suggesting that CaM acts by inducing a conforma-tional change within the reductase domain. Stopped-flow analysis (9,14) and work with partially active CaM mutants (24,25) show that CaM stimulates electron transfer to the acceptors primarily by increasing the rate of NADPH-dependent flavin reduction. Importantly, CaM's ability to speed flavin reduction can occur independent of its triggering nNOS heme reduction (24 -26), suggesting that the effects involve different structural elements of CaM. Thus, while nNOS and related flavoproteins display many structural and biochemical similarities, the CaM activation component makes nNOS unique and suggests significant structure-function differences do exist.
Studies investigating the interaction between cytochrome P450 reductase and its cytochrome P450 or cytochrome c acceptor hemeproteins have implicated two clusters of acidic residues within the FMN module 207 Asp-Asp-Asp 209 (cluster 1) and 213 Glu-Glu-Asp 215 (cluster 2) in controlling interactions important for electron transfer (27). Both clusters reside in the FMN module of the reductase and are highly conserved among related flavoproteins including the NOSs (Fig. 1). Mutagenic (27) and chemical cross-linking (28) studies with cytochrome P450 reductase suggested Asp 208 of cluster 1 is important for electron transfer to its P450 acceptor hemeprotein. For example, N-demethylase activity of the Asn 208 mutant was inhibited by 63% without changing the reductase K m toward cytochrome P450 or NADPH (27). The mutation did not affect cytochrome c or ferricyanide reductase activities, indicating interaction between the reductase and these molecules is distinct from its interaction with cytochrome P450. Similar observations were reported in Anabaena flavodoxin (29), where mutagenic analysis of cluster I acidic residues Asp 144 and Glu 145 showed they were involved in flavodoxin-supported P450c 17 progesterone 17a-hydroxylase activity but not involved in cytochrome c reduction. Crystal structures of cytochrome P450 reductase (17) and Anabaena flavodoxin (30) show that cluster 1 residues are located near the surface, presumably positioned to interact with a positive surface patch on their hemeprotein acceptor.
Because protein sequence and functional data suggest that the nNOS reductase domain and cytochrome P450 reductase have a similar secondary and tertiary structure, we hypothesized that cluster 1 residues may also be important in controlling reductase-oxygenase domain interaction and electron transfer in nNOS. We utilized site-directed mutagenesis to assess the importance of conserved amino acids Phe 892 , Asp 918 , and Glu 919 with respect to FMN binding, electron transfer reactions, and catalytic activities of nNOS. Surprisingly, our data show that these three residues impact nNOS function primarily by stabilizing FMN binding to the reductase. Their mutation resulted in FMN-depleted forms of nNOS, which we used to investigate reductase domain function and how the FMN module participates in nNOS response to CaM.

EXPERIMENTAL PROCEDURES
Materials-Superoxide dismutase was obtained from Calbiochem and was of the ferrous manganese type. All other regents and materials were obtained from Sigma or from sources previously reported (9,15,31).
Molecular Biology-Wild type and mutant nNOS with a His 6 tag attached to the N-terminal of the protein were overexpressed in Escherichia coli strain BL21 (DE3) using a modified PCWori vector and purified as described (31,32). Restriction digestions, cloning, bacterial growth, and transformation and isolation of DNA fragments were performed using standard procedures (33). Site-directed mutagenesis was done using the Altered Sites I in vitro Mutagenesis Kit from Promega. Wild-type nNOS cDNA was cut from PCWori with NdeI and XbaI and cloned into the XbaI site of the pAlter-I mutagenesis vector. Incorporated mutations were confirmed by DNA sequencing at the Cleveland Clinic core facility. DNAs that contained the desired mutations were cloned into the NdeI and XbaI sites of the PCWori vector and transformed into E. coli BL21(DE3). Oligonucleotides used to construct sitedirected mutants in the nNOS were synthesized by Life technologies. Silent mutations that incorporate unique restriction sites were also added to aid in screening. Mutations and corresponding oligonucleotides are listed below, with silent mutations underlined and mutagenic codons in bold. The silent restriction site incorporated in the first three oligonucleotides was AflII and in the fourth was XhoI. D918A, pGGA-GAGGATTCTTAAGATGAGGGAGGGGGCTGAGCTTTGCGGAC; E919A: pGGAGAGGATTCTTAAGATGAGGGAGGGGGATGCGCTTT-GCGGGAC; D918A, E919A: pGGAGAGGATTCTTAAGATGAGGGAG-GGGGCTGCGCTTTGCGGAC; F892A: pGTACCCCCACGCCTGTGC-CTTTGGGCATGCGGTGGACACCCTCCTCGAGGAACTGGGA.
Expression and Purification of Wild-type and Mutant nNOS-Transformed bacteria were grown at 37°C in 3 liters of terrific broth supplemented with 125 mg/liter ampicillin and 20 mg/liter chloramphenicol. Protein expression was induced when the cultures reached an OD 600 of 0.8 to 1 by adding 1 mM isopropyl-␤-D-thiogalactoside, and the cultures were supplemented with 0.4 mM ␦-aminolevulinic acid. After further growth at room temperature for 24 h, the cells were harvested and resuspended in buffer A (40 mM HEPPS, pH 7.6, with 10% glycerol, 1 mM Arg, 150 mM NaCl, 10 M H 4 B, 3 mM ascorbic acid) containing 1 mM EDTA, 0.5 mg/ml each of leupeptin and pepstatin, 1 mg/ml lysozyme, and phenylmethylsulfonyl fluoride. Cells were lysed by freeze-thawing three times in liquid nitrogen followed by sonication for three 25-s pulses with a 1-min rest on ice between pulses, using a medium probe Sonicator Cell Disrupter (Model W-220F, Heat systems, Ultrasonics. Inc.). The cell lysate was centrifuged at 4°C for 30 min and the cell-free supernatant was precipitated by adding 50% (w/v) ammonium sulfate. The precipitant was centrifuged at 4°C for 30 min at 16,000 rpm in a JA-17 rotor and kept at Ϫ70°C. The ammonium sulfate precipitate was resuspended in a buffer A containing 1 mM phenylmethylsulfonyl fluoride. The resuspended solution was loaded onto a Ni 2ϩ nitrilotriacetic acid-Sepharose CL-4B column that had been charged with 50 mM NiSO 4 and equilibrated with buffer A containing 1 mM phenylmethylsulfonyl fluoride. The column was washed with 5 times of column buffer and 5 times of column buffer containing 40 mM imidazole. The nNOS protein was eluted with 160 mM imidazole in buffer A and active fractions were pooled and stored at 4°C overnight in the presence of 1 mM DTT. The fractions were next loaded onto a 2Ј,5Ј-ADP-Sepharose column equilibrated with 40 mM HEPPS buffer, pH 7.6, containing 10% glycerol, 0.5 mM Arg, 3 mM DTT, and 2 M H 4 B. After adsorption the column was washed with column buffer containing 450 mM NaCl, and the protein was eluted with column buffer containing 10 mM NADPH. Selected fractions were concentrated using a Centriprep-50, dialyzed against 40 mM HEPPS, pH 7.6, containing 10% glycerol, 2. heme protein content using an extinction coefficient of 74 mM Ϫ1 cm Ϫ1 (A 444 -A 500 ) (34).
Determination of Bound FAD and FMN-Bound FAD and FMN were released from nNOS or mutants by heat denaturation of the enzyme (95°C for 5 min in the dark). It is essential to use well sealed vials for this procedure in order to avoid loss of sample volume. Subsequently samples were cooled to 4°C and filtered to remove denatured protein.
Samples were injected into a Microsorb Cyano Analytical HPLC Column (5 mm ϫ 4.6 mm ϫ 15 cm) and subjected to isocratic elution with 5 mM ammonium acetate, pH 6.0, containing 20% (v/v) methanol at a flow rate of 1 ml/min. FAD and FMN had retention times of 4.1 and 7.6 min and the peaks were completely resolved. Flavins were detected by fluorescence emission and quantitated based on authentic freshly prepared FAD and FMN standards.
NO Synthesis-The initial rate of NO synthesis by nNOS and mutants was quantitated at 37°C using the oxyhemoglobin assay for NO (34). The nNOS (ϳ25 nM) was added to a cuvette containing 40 mM HEPPS, pH 7.6, containing 15 g/ml CaM, 0.62 mM CaCl 2 , 0.3 mM DTT, 5 mM Arg, 4 M each of FAD and H 4 B, 100 units/ml catalase, and 10 M oxyhemoglobin to give a final volume of 0.7 ml. The reaction was started by adding NADPH to give 0.2 mM. The NO-mediated conversion of oxyhemoglobin to methemoglobin was monitored over time as an absorbance increase at 401 nm and quantitated using the extinction coefficient of 38 mM Ϫ1 cm Ϫ1 .
NADPH Oxidation-The initial rate of NADPH oxidation at 25°C was quantitated spectrophotometrically at 340 nm using an extinction coefficient of 6.22 mM Ϫ1 cm Ϫ1 . The composition of the assay mixture was similar to that of the NO synthesis measurement except that oxyhemoglobin was absent unless specified otherwise.
Reduction of External Electron Acceptors-Wavelengths and extinction coefficient used to quantitate the NADPH-dependent reduction of cytochrome c, DCIP, and ferricyanide were 550 nm (21 mM Ϫ1 cm Ϫ1 ), 600 nm (20.6 mM Ϫ1 cm Ϫ1 ), and 420 nm (1.2 mM Ϫ1 cm Ϫ1 ), respectively. The composition of the assay mixture was 40 mM HEPPS, pH 7.6, 4 M FAD, 0.1 mg/ml bovine serum albumin, 10 g/ml CaM, 0.6 mM EDTA, 10 units/ml catalase, 10 units/ml superoxide dismutase, and cytochrome c, DCIP, or ferricyanide at 0.1, 0.1, or 1 mM, respectively. In some cases, 0.83 mM Ca 2ϩ was added to promote CaM binding to nNOS. After the addition of nNOS, the reaction was initiated by adding 0.1 mM NADPH. K m values for cytochrome c, CaM, and FMN were determined from experiments in which the concentration of these molecules was varied and by reciprocal analysis of the velocity versus concentration data.
Reduction of Heme Iron-All samples were equilibrated at 25°C under anaerobic conditions in buffer saturated with CO. The cuvette contained 3 M wild-type nNOS or mutants in CO-saturated 40 mM HEPPS buffer, pH 7.6, containing 0.5 mM DTT, 4 M H 4 B, 0.6 mM EDTA, 1 mM Arg, and 6 M CaM. Concentrated anaerobic NADPH solution was added to the sample to give 0.1 mM. CaM binding and heme reduction were initiated by adding 1 mM CaCl 2 .
Fluorescence Spectroscopy-Flavin fluorescence measurements were done using a Hitachi model F-2000 spectrofluorometer as described previously (9) with modifications. A 1-ml quartz cuvette with a path length of 1 cm was used for the experiments. The nNOS proteins were diluted to 2 M in 40 mM HEPPS, pH 7.6, containing 0.6 mM EDTA, 1 mM DTT, and 3 M CaM. Proteins were irradiated with 450 -460 nm light using an in-line 8% filter and their emission spectra were recorded between 450 and 700 nm. In some experiments flavin fluorescence emission at 530 nm was monitored versus time before and after consecutive addition of 1 mM Ca 2ϩ and 3 mM EDTA.
Flavin Reduction Kinetics-The kinetics of flavin reduction were analyzed under anaerobic conditions as described previously (14), using a stopped-flow apparatus (Hi-tech Ltd., model SF-51) equipped for anaerobic work. Wild-type nNOS and mutants were briefly treated with ferricyanide and desalted prior to use in these experiments in order to oxidize the residual air-stable flavin semiquinone that is present in nNOS after purification (9,34). Measurements were made under pseudo first-order conditions and initiated by rapid mixing a solution of 0.1 mM NADPH with a solution containing 3 M CaM-free or -bound nNOS or mutants in 40 mM HEPPS buffer, pH 7.6, containing 0.5 mM EDTA, 6.0 M CaM and in some cases, 1 mM Ca 2ϩ . The absorbance change was monitored at 485 nm. Signal to noise ratios were improved by averaging the 10 individual scans. The time course of absorbance change was best fit to a single or double exponential equation by use of a nonlinear least-squares method provided by the instrument manufacturer (14).

Mutant nNOS Expression and Prosthetic Group Content-
The two-step enzyme purification typically yielded about 8 mg of full-length heme-containing nNOS mutants per liter of culture, which is similar to our yield of wild-type nNOS expressed in the same system. Spectroscopic analysis showed that all mutants contained heme in a predominantly low spin state. The heme iron of each mutant was observed to shift to high spin upon addition of 20 M H 4 B and 1 mM Arg. Dithionite reduction of each mutant in the presence of Arg, H 4 B, and CO produced the expected 444-nm absorbance peak for the ferrous-CO complex in all cases (data not shown). These data show that the reductase domain mutations did not alter expression of the full-length enzyme or affect the properties of the heme-containing oxygenase domain. Flavin analysis showed that the mutants contained normal quantities of FAD (ϳ1 per subunit) but contained below normal or practically undetectable levels of FMN (Table I). The D918A,E919A double mutant and single mutants F892A and D918A had almost no bound FMN, while E919A contained almost half the saturating level of FMN. Thus, the cluster I point mutations reduced or prevented stable binding of FMN by the nNOS reductase domain.
NO Synthesis and NADPH Oxidation by the nNOS Mutants- Table II compares the catalytic activities (expressed as turnover number per heme) of wild-type and mutant nNOS enzymes with regard to NO synthesis and NADPH oxidation. NO synthesis from either Arg or NOHA was slower or absent in the mutants in relation to their FMN content, consistent with FMN being a critical component for electron transfer during NO synthesis in nNOS. CaM-stimulated NADPH oxidation by each mutant was also reduced or blocked in a pattern identical to NO synthesis. Because CaM stimulation of NADPH consumption requires heme reduction in nNOS when O 2 is the electron acceptor (14,35), this suggested that the FMN-de-  pleted mutants have defective electron transfer between their reductase domain and heme in response to CaM. Heme Iron Reduction-To test this possibility we compared NADPH-dependent heme iron reduction in the mutant and wild-type nNOS under anaerobic conditions. Heme iron reduction was followed spectrally over time as a buildup of the ferrous-CO complex, whose Soret peak absorbs maximally at 444 nm (8,13). The percentage of NADPH-dependent heme iron reduction was determined relative to complete reduction achieved by adding dithionite to the sample at the end of each experiment. As shown in Fig. 2, heme reduction in wild-type nNOS is fast and complete under these conditions as previously reported (35). Heme reduction rates in each mutant were slower and approached maximum levels that were approximately in proportion to their bound FMN content. Thus, inhibition of heme iron reduction due to FMN depletion likely explains why CaM-induced NO synthesis and NADPH oxidation is slow in the mutants.
Mutant Reduction of Cytochrome c, DCIP, and Ferricyanide-CaM binding speeds electron transfer from the nNOS reductase domain to artificial acceptors such as cytochrome c, ferricyanide, and DCIP. As summarized in Table III, CaM increased rates of cytochrome c reduction in all cases, although basal and CaM-stimulated rates were markedly lower in the mutants compared with wild-type. This suggests that the mutants respond to CaM, but can only reduce cytochrome c in proportion to their bound FMN content. In contrast, mutant basal rates of DCIP reduction were only slightly lower than wild-type, and instead of increasing in response to CaM they actually decreased in the three mutants that contained the least amount of bound FMN. Moreover, mutant basal rates of ferricyanide reduction were 2-or 3-fold greater than wild-type basal level, and did not change significantly in response to CaM. These findings indicate that reduction of DCIP or ferricyanide by nNOS do not require bound FMN, and for ferricyanide bound FMN inhibits its reduction.
Reconstitution of Catalytic Activities with Exogenous FMN-Cytochrome P450 reductase and cytochrome P450BM3 preparations that are rendered FMN-free by mutagenesis or dialysis can often be reconstituted with exogenous FMN (20,36,37). We attempted reconstitution of the FMN-depleted nNOS mutants by incubating the proteins for 2 min in buffer containing various concentrations of FMN, followed by assaying their NO synthesis or CaM-induced cytochrome c reductase activities (Fig. 3, A and B). Wild-type nNOS increased its NO synthesis and cytochrome c activities by about 30% in the presence of added FMN. Added FMN also enabled each mutant to recover NO synthesis and cytochrome c reduction activities. The percentage recovery achieved for NO synthesis versus cytochrome c reduction were similar for any given mutant, consistent with both reactions requiring bound FMN. When assayed at the highest FMN concentration the three mutants with lowest native FMN content reached maximal activities that were ϳ40 -50% the level of wild-type nNOS, while the mutant retaining 50% native FMN content (E919A) reached 85% wildtype activity. Incubating the mutants with FMN for longer times prior to assay did not increase the amount of recovered activity (data not shown). Similar results have been reported for FMN binding mutants of cytochrome P450 reductase and cytochrome P450BM3, which also did not fully reconstitute their activities in response to added FMN (36,37).
Apparent K m and V max values for mutant FMN reconstitution were determined by double reciprocal analysis of the NO synthesis data from Fig. 3 and are listed in Table IV. The apparent K m for FMN in wild-type nNOS was 90 nM. Mutant K m values were approximately 2 orders of magnitude greater than wild-type except for the partially FMN-replete E919A mutant, whose K m was increased by a factor of 10. Thus, the mutants all have reduced binding affinity for FMN, explaining why they contain little or no FMN after purification. As summarized in Table V, adding FMN to the mutants enabled them to respond more normally regarding each catalytic activity in the presence or absence of CaM. Ferricyanide reduction rates in CaM-free nNOS were lowered in the presence of added FMN,  suggesting bound FMN inhibits ferricyanide reduction in this circumstance. K m for CaM and Cytochrome c-We next investigated whether cluster 1 mutations effect nNOS affinity toward CaM or cytochrome c. Fig. 4 contains representative reciprocal plots of cytochrome c reduction rate versus cytochrome c concentration by wild-type nNOS and the D918A,E919A double mutant. Similar data were obtained for CaM titrations, using cytochrome c reduction as a marker for activity. Table VI summarizes apparent K m values for cytochrome c and CaM with wildtype and mutant nNOS proteins. The close similarity between mutants and wild-type indicate that our cluster 1 mutations did not effect nNOS interaction with either CaM or cytochrome c.
CaM-induced Flavin Fluorescence-CaM binding to nNOS or its isolated reductase domain causes a partially reversible increase in flavin fluorescence (9). We utilized our FMN-deficient mutants to determine if bound FMN was involved in the CaMinduced fluorescence increase. As shown in Fig. 5, wild-type nNOS increased its flavin fluorescence by a factor of 2 after Ca 2ϩ -promoted CaM binding, and the increase was partly reversed by dissociating CaM with excess EDTA, consistent with previous work (9). The E919A mutant underwent a similar change in fluorescence upon binding and dissociation of CaM, although the magnitude of change was less compared with control, consistent with this mutant being 50% FMN-deficient.
The three nNOS mutants that were mostly devoid of FMN did not increase their flavin fluorescence upon CaM binding under the same circumstances. Thus, bound FMN is essential for the fluorescence change associated with CaM binding to nNOS, suggesting the conformational change that occurs is restricted to the FMN module of the reductase.
Stopped-flow Analysis of Flavin Reduction-Because CaM binding causes a 15-fold increase in the rate of nNOS flavin reduction by NADPH (14), we measured flavin reduction rates in the nNOS mutants in the presence or absence of CaM. Experiments were done under pseudo first-order conditions using a 33-fold excess of NADPH, and flavin reduction was monitored at 485 nm. As shown in Fig. 6, the absorbance trace representing flavin reduction in our CaM-free wild-type nNOS best fit to a biphasic process with apparent rate constants of 9.2 s Ϫ1 (k 1 ) and 1.6 s Ϫ1 (k 2 ), respectively. CaM binding to wild-type nNOS increased the apparent rates to 54 and 14 s Ϫ1 (Table  VII), consistent with earlier reports (9,14). The first phase of the reaction (k 1 ) can be attributed to NADPH reduction of FAD, and the second phase (k 2 ) to electron transfer from FAD to FMN (38). In CaM-free nNOS the rate of FAD reduction is relatively slow and comparable to FAD reduction in monoamine oxygenase (2.5 s Ϫ1 at 4°C) (39), whereas in CaM-bound nNOS the rate of FAD reduction is comparable to mammalian cytochrome P450 reductase (55 s Ϫ1 at 20°C) (40), but slower than spinach ferridoxin (105 s Ϫ1 at 4°C) (41) or bacterial cytochrome P450BM3 (130 s Ϫ1 at 4°C) (38).
Also shown in Fig. 6 is the absorbance trace for the CaM-free D918A,E919A double mutant, which best fit to a monophasic process giving an apparent rate of 39 s Ϫ1 . The monophasic nature of the reaction is consistent with only FAD being bound in the protein and represents the rate of FAD reduction by NADPH. The apparent rate constants for all CaM-free and CaM-bound nNOS mutants are listed in Table VII. These data show that in the CaM-free state the three mutants devoid of FMN had rates of FAD reduction (k 1 ) that are four times faster than FAD reduction in wild-type nNOS, and their rates did not increase further with CaM. The partially FMN-replete E919A mutant had a CaM-free rate of FAD reduction that is between the two extremes, and the rate was slightly increased after CaM binding. Together, this indicates that bound FMN has a significant negative effect on FAD reduction rate in the CaMfree basal state. DISCUSSION Our work reveals that reductase domain cluster 1 residues Asp 918 , Asp 919 , and Phe 892 help stabilize FMN binding in nNOS. In related flavoproteins such as rat cytochrome P450 reductase and Anabaena flavodoxin, analogous cluster 1 residues facilitate electron transfer between the FMN module and hemeprotein partners, and mutations typically inhibit catalysis without affecting FMN binding (27, 29). Indeed, loss of FMN from these flavoproteins either requires special dialysis (20, 42,   (17,29). If nNOS cluster 1 residues are similarly positioned, they would need to influence FMN binding through relatively long-range interactions. Our cluster 1 mutants contained normal quantities of FAD and heme, and could bind NADPH, H 4 B, Arg, and CaM. Moreover, they all recovered considerable NO synthesis and cytochrome c reductase activity when supplied with exogenous FMN. This suggests the functional defects caused by mutation were primarily due to loss of FMN, and were restricted to the FMN module. However, because mutant activities approached but did not achieve wild-type levels even at the highest FMN concentrations, cluster 1 residues might have another effect on nNOS catalysis besides lowering FMN affinity. NO synthesis is related to the heme reduction rate in nNOS (9,44), thus a slower electron transfer rate to the heme could explain the residual inhibition we observe. Cluster 1 residues might effect heme reduction rate by stabilizing the interaction between the FMN module and hemeprotein domain, or by modulating the redox properties of the bound FMN, as occurs in some related flavoproteins (22). However, the relatively minor residual inhibition that we observed in our FMN-reconstituted mutants clearly indicates that these effects would be secondary relative to the mutational effect on FMN binding. We conclude that cluster 1 residues are not as critical for electron transfer functions of the nNOS FMN module as for other flavoproteins of the same class. Because cluster 1 mutations destabilized FMN binding in nNOS without greatly affecting other properties of the protein,   they can help define reductase domain function. For example, we observed that bound FAD was still reduced by NADPH in the FMN-free mutants while electron transfer to the nNOS heme or cytochrome c was blocked. Also, CaM binding to the FMN-free mutants did not enable reduction of these acceptors. This establishes: 1) electron flow from NADPH to FAD, FMN, then heme in nNOS, as occurs in structurally related flavoproteins that do not bind CaM (18,20,45).
2) The FMN module of nNOS is the exit point for electron transfer to the nNOS heme or to other hemeproteins whether CaM is bound or not. Regarding electron transfer to smaller molecules like DCIP and ferricyanide, rates in the FMN-and CaM-free nNOS mutants were either not inhibited or in the case of ferricyanide were much faster than wild-type. Thus, the NADPH-FAD module can transfer electrons to these smaller molecules independent of the FMN module, as occurs in related flavoproteins (18,20,45), and is likely an important electron donor to these molecules in CaM-free nNOS.
In the CaM-free state, stopped-flow analysis showed that FMN-depleted mutants had faster rates of FAD reduction than wild-type nNOS. Their observed rates were as much as 4-fold faster, and varied in reverse proportion to their residual FMN content. Because NADPH-derived electrons are transferred first to FAD and then to FMN, our work shows that bound FMN can actually inhibit an "upstream" electron transfer step in the nNOS reductase. This effect has not been observed previously in the related proteins cytochrome P450 reductase, sulfite reductase, or cytochrome P450BM3, and so may be unique to the NOS. Our finding that FMN removal increased the rate of ferricyanide reduction in CaM-free nNOS, while it either lowers or does not change the rate in related flavoproteins (20,36,37,43,46,47), supports this concept.
It is interesting that CaM binding to wild-type nNOS increased FAD reduction to a rate that was equivalent to the FMN-free nNOS mutants. Thus, CaM binding is functionally equivalent to FMN removal when one's view is restricted to FAD reduction. Two mechanisms are possible: bound CaM influences FAD reduction through a process that either relies on or does not involve the FMN module. The fact that CaM  figure), the FMN module can no longer suppress FAD reduction in the FNR domain. When CaM binds to nNOS (lower horizontal figure), the FMN module undergoes a conformational change (illustrated by square to circle transition) that relieves its repression of the FNR domain (A), increases the rate of FMN reduction (B), and triggers electron transfer to the heme (C). These changes are associated with increased electron transfer from the indicated modules to ferricyanide (FeCN 6 ) and to cytochrome c (dashed to solid arrow transition), and initiation of NO synthesis. binding did not further increase the rate of FAD reduction in FMN-free nNOS supports the first possibility. This suggests that bound CaM does not directly interact with the FNR module to increase FAD reduction, but instead acts indirectly through an effect on the FMN module that is functionally equivalent to FMN removal. This mechanism is consistent with CaM increasing flavin fluorescence only in the FMN module of nNOS.
Our current results suggest that the FMN module is the key response element in nNOS that regulates electron transfer at all points (summarized in Fig. 7). In the absence of CaM, the FMN module interacts with the FNR module to repress NADPH reduction of FAD (point A). In this circumstance, the FMN module accepts electrons from FAD (point B), but is unable to transfer electrons to the heme located in the oxygenase domain (point C). When CaM binds it causes a conformational change in the FMN module such that: 1) suppression of electron transfer in the FNR domain (point A) is relieved; 2) the rate of electron transfer from the FNR module to FMN (point B) is increased; and 3) the FMN module is able to transfer electrons to the oxygenase domain heme (point C). Removing FMN from its module by mutagenesis also relieves supression of FAD reduction within the FNR module, suggesting that the FMN module must be replete to have its effect. Regarding effects on catalysis, the relief of suppression at point A is associated with faster ferricyanide reduction, the increase in electron transfer at point B is associated with faster cytochrome c reduction, and the initiation of electron transfer at point C is associated with NO synthesis from Arg, or superoxide production in the absence of substrate. Together, these functions distinguish the nNOS reductase domain, and the FMN module in particular, from all other flavoproteins.
Our current model raises several interesting questions. For example, why does the nNOS FMN module slow electron transfer into the upstream FNR module of nNOS? Apparently, it must be important for the enzyme to slow electron flux through the reductase until a Ca 2ϩ influx into cells occurs and CaM binding signals the time for heme reduction and NO synthesis. Is suppression of FAD reduction linked to the FMN module's inability to transfer electrons to the heme in the absence of CaM? This is unlikely, because previous work with CaM mutants indicates that repression of either FAD reduction (24) or ferricyanide reduction (24 -26) can be relieved without activating electron transfer from FMN to the oxygenase heme. Thus, functions of the FMN module are separable, and may be controlled by different facets of CaM binding. Exactly how does CaM alter FMN module function and its interaction with the FNR module? Our fluorescence data suggest that CaM causes a conformational change exclusively in the FMN module that alters the FMN environment. Perhaps this alters the redox relationship between the flavins, or alters the FAD reduction potential relative to NADPH such that its reduction is thermodynamically more favorable. The mechanism may also involve a ϳ30 amino acid insert that is present in the FMN module of nNOS but absent in related flavoproteins, and has been proposed to regulate CaM affinity (48). These and other possibilities can now be addressed.