Lys842 in Neuronal Nitric-oxide Synthase Enables the Autoinhibitory Insert to Antagonize Calmodulin Binding, Increase FMN Shielding, and Suppress Interflavin Electron Transfer*

Neuronal nitric-oxide synthase (nNOS) contains a unique autoinhibitory insert (AI) in its FMN subdomain that represses nNOS reductase activities and controls the calcium sensitivity of calmodulin (CaM) binding to nNOS. How the AI does this is unclear. A conserved charged residue (Lys842) lies within a putative CaM binding helix in the middle of the AI. We investigated its role by substituting residues that neutralize (Ala) or reverse (Glu) the charge at Lys842. Compared with wild type nNOS, the mutant enzymes had greater cytochrome c reductase and NADPH oxidase activities in the CaM-free state, were able to bind CaM at lower calcium concentration, and had lower rates of heme reduction and NO synthesis in one case (K842A). Moreover, stopped-flow spectrophotometric experiments with the nNOS reductase domain indicate that the CaM-free mutants had faster flavin reduction kinetics and had less shielding of their FMN subdomains compared with wild type and no longer increased their level of FMN shielding in response to NADPH binding. Thus, Lys842 is critical for the known functions of the AI and also enables two additional functions of the AI as newly identified here: suppression of electron transfer to FMN and control of the conformational equilibrium of the nNOS reductase domain. Its effect on the conformational equilibrium probably explains suppression of catalysis by the AI.

more deshielded state, enabling more efficient electron transfer (36,37,40). Although much is known about how the CT of nNOS helps to regulate the FMN conformational equilibrium in conjunction with NADPH binding (36,37,41,42), what role the AI might play in regulating this critical equilibrium is completely unknown. It is also unclear which residues of the AI are key to its function regarding its effects on CaM binding and NOS catalytic activity. To address these issues, we mutated a conserved charged residue (Lys 842 ; Fig. 1) in the AI of full-length nNOS and in our nNOS reductase domain construct (nNOS red ) to test how charge neutralization and charge reversal at this site would affect AI function. Our results show that Lys 842 helps the AI to have its known macroscopic effects (antagonizing CaM binding and suppressing cytochrome c reductase activity) and moreover reveal new molecular level roles for the AI in suppressing the rates of flavin reduction and in governing the conformational equilibrium of the reductase domain to favor the closed or FMN-shielded conformational state. Together, our results define for the first time a multifaceted mechanism by which the AI inhibits catalysis by nNOS in the CaM-free state.
Generation of Mutants-The bacterial expression vector pCWori contained cDNA that coded for wild type (WT) rat nNOS cDNA with a His 6 tag attached to its N terminus (47,48) or nNOS red with its adjacent N-terminal CaM binding motif (Met 695 -Ser 1429 ) (37) as described previously. Site-directed mutagenesis was done using a QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA). The oligonucleotide pairs 5Ј-AGAGAAA-GAGCTACGCGGTCCGATTCA-ACA-3Ј (3Ј-TCTCTTTCTCGA-TGCGCCAGGCTAAGTTGT-3Ј) and 5Ј-AGAGAAAGAGCTACGAG GTCCGATTCAACA-3Ј (3Ј-TCTC-TTTCTCGATGCTCCAGGCTAA-GTTGT-5Ј) were used to generate the K842A and K842E mutants, respectively. Molecular biology techniques were performed using standard procedures or according to the manufacturer's instruction. Mutations were confirmed by DNA sequencing at the Cleveland Clinic genomics core facility. Expression and Purification of WT and Mutant Enzymes-WT and mutant enzyme expression was induced at room temperature over 1 or 2 days in Escherichia coli BL21(DE3) that harbored the pACYC-CaM plasmid as described previously (37). WT nNOS red and mutants were purified using sequential 2Ј,5Ј-ADP-Sepharose and CaM affinity chromatography as reported previously (37). The full-length nNOS and mutants were purified by sequential Ni 2ϩ -Sepharose and 2Ј,5Ј-ADP-Sepharose affinity chromatography as previously described (45,49).
Determination of Enzyme-bound Flavins-FAD and FMN were released from NOS samples by heat denaturation in a 95°C water bath for 5 min in the dark and measured using a high pressure liquid chromatograph with fluorometric detection (50).
Assay of Enzymatic Activities-The ferricyanide reductase activity was determined by monitoring the absorbance decrease at 420 nm using an extinction coefficient of 1.02 mM Ϫ1 cm Ϫ1 in assays containing 1 mM ferricyanide and 0.3 mM NADPH. Cytochrome c reductase activity was determined from the absorbance increase at 550 nm using a difference extinction coefficient of 21 mM Ϫ1 cm Ϫ1 in assays containing 0.1 mM cytochrome c, 0.3 mM NADPH. The rate of NO synthesis was determined from the NO-mediated conversion of oxyhemoglobin to methemoglobin using a difference extinction coefficient of 49 mM Ϫ1 cm Ϫ1 at 401 nm in reactions containing 10 M oxyhe-FIGURE 1. Top, amino acid sequence of the AI in Rat nNOS and bovine eNOS. The residues comprising the AI helix and the portion of the AI that is visible in the nNOS red crystal structure are indicated. The three canonical hydrophobic residues in the putative CaM binding motif of the AI are in boldface type. Lys 842 is marked with an asterisk. Bottom, ribbon diagram of the nNOS red crystal structure that highlights the relative position of the AI and Lys 842 in the FMN subdomain. The FMN subdomain is colored black, and the rest of the reductase domain is gray. Only the AI residues that are visible in the crystal structure are shown. The bound FAD and FMN molecules are shown as space-filling models. This figure was drawn using Protein Data Bank file 1TLL. moglobin, 0.1 mM NADPH, 10 M (6R)-tetrahydrobiopterin, and 1 mM L-arginine. The NADPH oxidation rate was measured from the absorbance decrease at 340 nm using an extinction coefficient of 6.22 mM Ϫ1 cm Ϫ1 . All assays were performed in cuvettes at 25°C in 40 mM EPPS buffer (pH 7.6) containing 10% glycerol, 10 units/ml superoxide dismutase, 100 units/ml catalase, 0.5 M FAD/FMN, and 50 M EDTA, using a Hitachi U-3110 spectrophotometer. In some reactions, 0.5 mM Ca 2ϩ and 5 M CaM were also added.
Kinetics of Heme and Flavin Reduction-Rapid mixing reactions were performed in a SF-51MX stopped-flow spectrophotometer (HiTech Ltd.), which had a dead time of 2 ms and was equipped with an anaerobic workbox. All stopped-flow experiments were performed at 10°C in 40 mM EPPS buffer, pH 7.6, containing 150 mM NaCl. The rate of heme reduction was determined by monitoring the absorbance increase at 444 nm, which indicated formation of the ferrous heme-CO complex as previously described (41,49). Reactions were initiated by rapid mixing an anaerobic CO-saturated solution containing 50 M NADPH with an anaerobic CO-saturated solution containing 5 M of WT or mutant nNOS in 40 mM EPPS buffer, pH 7.6, containing 10 M (6R)-tetrahydrobiopterin, 1 mM L-arginine, 10 M CaM, and 1 mM Ca 2ϩ .
The kinetics of flavin reduction were studied by rapid mixing stopped-flow analysis of reactions that mixed nNOS red enzymes with excess NADPH or 1 molar eq of NADPH under anaerobic conditions at 10°C using a stopped-flow apparatus and diode array detector (Hi-Tech Scientific KinetAsyst SF-61DX2) equipped for anaerobic analysis. In some experiments, single wavelength data were obtained in identical reactions run in a HiTech Ltd. SF-51MX equipped with photomultiplier detection equipped for anaerobic analysis. The procedures used were similar to those described previously (37). Briefly, the fully oxidized or 1-electron-reduced forms of nNOS red proteins were diluted to 5-10 M (for nNOS red ) and then rapidly mixed with a 10-fold molar excess of NADPH. In some cases, a 5-or 15-fold molar excess of NADPH was used. Also, in some cases, residual NADP ϩ present in the NADPH solutions or in the 1-electron-reduced enzyme samples was eliminated by adding an enzymatic NADPH-regenerating system (consisting of 2-6 units of glucose-6-phosphate dehydrogenase and 2-6 mM glucose-6-phosphate, depending on molar ratio of NADPH to enzyme) or by passing the NADPH-reduced proteins through a G-25 desalting column prior to use, respectively. To analyze CaM-bound enzyme samples, EDTA was replaced with CaCl 2 (2 mM) and CaM (15-20 M). Averaged spectra or absorbance traces were compiled from 7-10 reactions, and the traces at various wavelengths were fit to single or multiple exponential functions using software provided by the instrument manufacturer. Only the observed flavin absorbance change apparent in the traces was used for fitting (the fitting does not include any absorbance change that was calculated to take place in the mixing dead time). The percentage of flavin reduction that takes place in the instrument mixing dead time was calculated after obtaining absorbance traces from control stopped-flow reactions that rapidly mixed the same enzyme solutions described above with buffer that did not contain NADPH, as previously described in detail (37).
Anaerobic Pre-steady-state Cytochrome c Reduction-The rate of cytochrome c reduction by an excess of prereduced nNOS red proteins under various conditions was measured in the stopped-flow apparatus (SF-51MX, HiTech Ltd.) at 10°C as described previously (37). Briefly, each nNOS red protein (16 M) was prereduced using a slide projector bulb in a sealed cuvette in buffer containing EDTA (1 mM), 5-deazariboflavin (catalytic amount), glucose oxidase (10 units/ml), catalase (134 units/ml), and glucose (2 mM). To analyze CaM-bound samples, EDTA was replaced with glycine, and CaCl 2 (2 mM) and CaM (25 M) were included. The photoreduced proteins were incubated with (1 mM) or without NADPH and then rapidly mixed with cytochrome c (4 M) while recording the absorbance changes at 550 nm. Multiple absorbance traces were averaged and fit to a single exponential function.
EGTA Titration of CaM-bound WT nNOS and Mutants-The CaM-bound WT nNOS and mutants were titrated with EGTA in a range of final concentration from 1 to 50 M, and their rates of NO synthesis and concurrent NADPH oxidation were measured. The assay conditions were as described above. The concentrations of CaM-bound WT nNOS and mutants in the reaction mixture were 20 nM. An initial concentration of 20 M Ca 2ϩ was added into each of the reaction mixtures prior to the EGTA titration.

RESULTS
Protein Expression-The mutant proteins could be expressed in amounts that were similar to those of WT proteins, and they all contained 1 FAD and 1 FMN per NOS red molecule or per NOS heme.
Steady-state Catalytic Activities- Table 1 lists the NO synthesis, NADPH oxidation, cytochrome c reductase, and ferricyanide reductase activities we obtained for nNOS and the Lys 842 mutants. The WT and mutant enzymes (K842A and K842E nNOS) required CaM/Ca 2ϩ binding for NO synthesis. NO synthesis activity was lower only in the K842A mutant. Corresponding rates of NADPH oxidation during NO synthesis were comparable with WT nNOS although somewhat higher in the K842E mutant. In the absence of CaM/Ca 2ϩ , the K842A and K842E mutants had 3-and 4-fold greater rates of NADPH oxidation, respectively, than WT nNOS. Both nNOS mutants also had greater cytochrome c reductase activities (ϳ1.5-fold for K842A and ϳ2.9-fold for K842E nNOS) in the absence of CaM/Ca 2ϩ ( Table 1). The ferricyanide reductase activities of the mutants were similar to WT nNOS in the absence or presence of CaM/Ca 2ϩ . The Lys 842 mutations altered the cytochrome c reductase activity of nNOS red in a manner similar to that observed for full-length nNOS (Fig. 2). The activities of the CaM-free K842A and K842E mutants were ϳ1.9and ϳ3.1-fold greater than WT nNOS red , respectively. The activity of the CaM-free K842E mutant was 65% of the value obtained in its CaM-bound state. Together, the data indicate that Lys 842 helps the AI repress the cytochrome c reductase activity and the NADPH oxidase activity of nNOS in the CaM-free state and may also help to fully enable NO synthesis in CaM-bound nNOS.
CaM/Ca 2ϩ Concentration Dependence-Deleting the AI enables nNOS to bind CaM at lower Ca 2ϩ concentrations (25,26). We investigated the importance of Lys 842 by titrating the CaM-bound mutants with the Ca 2ϩ chelator, EGTA, and measuring their NO synthesis and NADPH oxidation activities under each condition relative to WT nNOS. All experiments contained 20 M free Ca 2ϩ at the start to normalize the free Ca 2ϩ concentration. As shown in Fig. 3, A and B, both of the Lys 842 mutants required greater concentrations of EGTA than WT nNOS to lose their CaM-dependent NADPH oxidation and NO synthesis activities. The results indicate that Lys 842 is an essential component of the AI that helps it to antagonize Ca 2ϩ -dependent CaM binding to nNOS.
Flavin Reduction Kinetics in Fully Oxidized nNOS Enzymes-NADPH-dependent flavin reduction is kinetically repressed in CaM-free nNOS (37). We utilized stopped-flow spectroscopy to determine how the Lys 842 mutations might affect the kinetics of flavin reduction in fully oxidized CaM-free nNOS (i.e. containing FAD and FMN) by an excess of NADPH. In this circumstance, flavin reduction should occur in the following sequential steps: a rapid hydride transfer from NADPH to FAD, two 1-electron transfers from FAD to FMN, and then further reduction of the FAD by a second molecule of NADPH or by other reduced nNOS red molecules in solution 4 (Scheme 1). The flavin reduction rates were calculated from the absorbance decrease obtained at 485 nm after mixing the oxidized enzymes with excess NADPH. Representative traces for the CaM-free WT and K842E nNOS proteins are shown in Fig. 4A. All traces could be fit reasonably well to a two-phase exponential equation. The observed rate constants obtained for WT nNOS and the two Lys 842 mutants are reported in Table 2. Both phases of flavin reduction were found to be faster in the mutants than in WT nNOS under the CaM-free conditions. In contrast, the CaMbound mutant and wild type enzymes had similar rates of flavin reduction.
We also examined the absorbance change at 600 nm that occurred during the flavin reduction experiments noted above,   JANUARY 29, 2010 • VOLUME 285 • NUMBER 5 because absorbance change at this wavelength can indicate the buildup and subsequent reduction of flavin semiquinone species during the flavin reduction reactions (Scheme 1). As shown in Fig. 4B, the absorbance at 600 nm increased rapidly and then decayed after mixing the CaM-free K842E or WT nNOS enzymes with excess NADPH. The rates of absorbance increase and decay at 600 nm were faster in the K842E mutant than in WT nNOS. The peak absorbance at 600 nm was reached more quickly in the K842E nNOS reaction (50 ms) than in the WT reaction (85 ms). This is consistent with the kinetic differences we observed at 485 nm ( Fig. 4A) and implies that the CaM-free K842E mutant has faster flavin semiquinone formation and fla-vin semiquinone reduction relative to WT nNOS during the reduction by excess NADPH.

Role of Lys 842 in AI Function in nNOS
Kinetics of Flavin Reduction in Fully Oxidized nNOS red Enzymes-We next performed a more extensive investigation of flavin reduction kinetics using the nNOS red protein constructs in order to extend our initial findings with the fulllength nNOS proteins. Reactions were run by mixing the fully oxidized nNOS red proteins with a 10-fold molar excess of NADPH at 10°C under anaerobic conditions in stopped-flow diode array and single wavelength spectrophotometers. Fig. 5 contains representative absorbance traces at 457 nm from reactions of the K842E and WT nNOS red proteins in their CaM-free and CaM-bound states. The traces fit best to a triple exponential equation, and the rates derived with the different nNOS red proteins are listed in Table 3. A significant proportion (20 -30%) of the total absorbance decrease at 457 nm occurred in the mixing dead time of the instrument in these reactions, consistent with previous reports (37) and with the reactions involving a rapid initial hydride transfer step from NADPH to FAD. The results indicate that the K842E (Table 3) and K842A nNOS red  The three individual rate constants obtained by fitting these data to multiple exponential functions are reported in Table 3. Final concentrations after mixing were as follows: nNOS red enzymes, 3.5-4.0 M; NADPH, 35-40 M.

TABLE 2 Heme reduction and flavin reduction of nNOS and the Lys 842 mutants
All reactions were carried out at 10°C under anaerobic conditions in a stopped-flow spectrophotometer as described under "Experimental Procedures." k 1 and k 2 refer to the rates for biphasic reactions. The values for flavin and heme reduction in wildtype and mutant NOSs are the mean Ϯ S.D. of 7-10 individual reactions and are representative of experiments done with two different enzyme preparations. The numbers in parentheses indicate the percentage of absorbance change attributed to the fast phase (k 1 ). ND, not detected.

Enzyme
CaM Flavin reduction Heme reduction (k) (data not shown) have generally faster rates of flavin reduction than does WT nNOS red in the CaM-free condition, similar to what we observed for the full-length nNOS proteins. The flavin reduction rates obtained in reactions of the CaM-bound nNOS red proteins indicate that they all were increased and became similar in this condition. Fig. 6, A and B, contains representative absorbance traces at 600 nm that were obtained during reactions of the fully oxidized, CaM-free, or CaM-bound forms of K842E and WT nNOS red proteins with a 10-fold molar excess of NADPH. The absorbance increase and decrease at 600 nm in this circumstance can indicate the rates of buildup and subsequent reduction of flavin semiquinone species that must form during the reductive reaction (Scheme 1). We monitored these reactions using a single wavelength detector over a relatively short time frame (main panels) and over a somewhat longer time frame (insets), in order to determine and compare the rates of change at 600 nm as accurately as possible. In all cases, we observed an initial rapid buildup of absorbance at 600 nm that reached a plateau and then decreased to achieve absorbance levels that remained above the initial value. 5 The remaining absorbance at 600 nm that is apparent at the end of the collection periods (relative to the initial values) may indicate a contribution from a transient NADP ϩ -FADH 2 charge transfer complex. Overall, this finding is consistent with maximal flavin radical formation occurring upon electron transfer from the FAD hydroquinone to FMN, followed by further import of electrons into enzyme flavins from additional NADPH as the reaction proceeds (Scheme 1). The absorbance buildup and decay phases fit reasonably well to monophasic transitions in all cases, and the observed rates are reported in Table 4. Companion diode array spectra that we obtained in replica reactions that mixed 10-fold molar excess NADPH with the fully oxidized, CaM-free nNOS red proteins are shown in supplemental Fig. S1. The diode array spectra show that the absorbance buildup at 600 nm was relatively small in all reactions, consistent with previous data obtained for WT nNOS red (37,51). Despite this, the CaM-free K842E nNOS red had a significantly faster buildup and decay of the 600 nm absorbance during its reduction with excess NADPH as compared with WT nNOS red . CaM binding to the WT protein increased its absorbance buildup and decay rates at 600 nm, as reported previously (37), whereas CaM binding to the K842E mutant caused only small additional rate increases.
The rates of flavin reduction in nNOS red can be dependent on the NADPH concentration (51) and can also be affected by NADP ϩ (52). To assess whether the kinetic differences in flavin reduction that we observed between K842E and WT nNOS red were due to a difference in their interactions with NADPH or NADP ϩ , we repeated the stopped-flow analyses described above but this time using a 5-or 15-fold molar excess of NADPH or using NADPH solutions that had all of their residual NADP ϩ converted to NADPH by inclusion of an enzymatic NADPH-regenerating system. In all cases, these modifications caused 15% or less change in the flavin transition rates at 600 nm (data not shown) that we had observed in our studies using a 10-fold molar excess of NADPH (Table 4). This suggests that the kinetic differences we observed between WT and K842E nNOS red were not due to possible differences in their interactions with NADPH or NADP ϩ .
Because of the relatively small amplitude changes we observed at 600 nm during the reductive reactions, we also compared the kinetics of the coincident absorbance change at 700 nm, which can indicate the buildup of a charge transfer complex that forms after the bound NADPH transfers a hydride to FAD (i.e. the NADP ϩ /FADH 2 species) (37). As shown in supplemental Fig. S2, there was a very fast buildup of absorbance at 700 nm that mostly preceded the buildup of absorbance at 600 nm in the same reaction and then a rapid decay in absorbance at 700 nm during further buildup of the 600 nm absorbance. This is consistent with the absorbance change at 700 nm reflecting a rapid hydride transfer to FAD and formation of a NADP ϩ /FADH 2 charge transfer complex. Together, the results are consistent with faster interflavin electron transfer in the CaM-free Lys 842 mutant proteins, which in turn enables a faster rate of flavin semiquinone formation during the reductive reaction.
Kinetics of Flavin Reduction Using 1 Molar Eq of NADPH-We also compared flavin reduction in the fully oxidized enzymes upon mixing with 1 molar eq of NADPH. In such cases, the reaction involves the initial hydride transfer from NADPH to FAD and then equilibration of the two electrons between the FAD and FMN cofactors. Because of the more positive midpoint potential of the FMN/FMNsq couple, the system will ultimately achieve a thermodynamic equilibrium where the primary species is the flavin disemiquinone with maximal absorbance at 600 nm.
Supplemental Fig. S3 and supplemental Table S1 contain representative spectroscopic and kinetic data collected in stopped-flow experiments with the CaM-free enzymes. Their reduction by 1 eq of NADPH generated initial spectral changes 5 We used a shorter time frame for the absorbance data collection in the experiments with K842E nNOS red in order to better catch the faster transitions that occur in this mutant. a Percentage of the total absorbance change occurring in the instrument dead time. b Individual rate constants are reported as follows: calculated rate (s Ϫ1 ) and percentage of the total absorbance change for this process in parentheses. JANUARY Table S1), consistent with the results we obtained using excess NADPH. The slow rates of flavin disemiquinone formation for either enzyme in the single equivalent reactions probably reflects a poor capacity of the NADP ϩ -FADH 2 complex to reduce the FMN cofactor (53). This observation highlights the importance of excess NADPH, as would normally be present, in driving flavin reduction in nNOS red .

Kinetics of Flavin Reduction in One-electron-reduced nNOS red
Enzymes-A characteristic absorption peak at 520 nm has been attributed to the FAD semiquinone species (FADH ⅐ ) in human nNOS (11,31). This provides a means to follow interflavin electron transfer in the one-electron-reduced form of nNOS red (which contains an air-stable FMN semiquinone) by monitoring the absorbance increase at 520 nm during reduction of the enzyme with NADPH (31) (Scheme 2). We first checked if the 520 nm absorbance peak previously associated with FAD semiquinone formation in human nNOS would also form in rat nNOS red . As shown in Fig. 7A, a 520 nm absorbance peak built up during the steady-state reaction of CaM-free nNOS red as it catalyzed NADPH oxidation. Identical spectral changes were observed during NADPH oxidation by K842A and K842E nNOS red (data not shown). We then generated the air-stable semiquinone form (i.e. FAD/ FMNH ⅐ ) of our CaM-free WT and mutant nNOS red enzymes for stopped-flow studies by incubating them with a 5-fold molar excess of NADPH and allowing them to airoxidize at room temperature for 20 min. As shown in Fig. 7, B and C, when the air-stable semiquinone form of WT or K842E nNOS red was mixed with a 10-fold excess of NADPH in the stopped-flow spectrophotometer, we observed an absorbance increase at 520 nm that indicated the conversion from FADH 2 / FMNH ⅐ to FADH ⅐ /FMNH 2 (subsequent to an initial, fast hydride transfer from NADPH to FAD, as summarized in Stopped-flow traces were collected at 600 nm after rapidly mixing 7-8 M CaM-free oxidized WT nNOS red or K842E nNOS red (A) and CaM-bound oxidized WT nNOS red or K842E nNOS red (B) with a 10-fold excess of NADPH at 10°C as described under "Experimental Procedures." Inset, plots obtained from collecting traces for longer time frames. Rate constants obtained by fitting the absorbance increase and decrease to single exponential functions are reported in Table 4. The calculated best fit curves are plotted as solid lines. Scheme 2). A similar reaction was also run to measure the rate of absorbance change at 520 nm in the one-electron-reduced, CaM-bound K842E nNOS red upon mixing with NADPH (supplemental Fig. S4). All of the absorbance traces at 520 nm fit well to a single phase exponential equation and gave observed rate constants of 13.5 Ϯ 0.4, 30.2 Ϯ 1.0, and 35.8 Ϯ 1.0 s Ϫ1 for the CaM-free WT, K842A, and K842E nNOS red , respectively, and 54 Ϯ 2.0 s Ϫ1 for the CaM-bound K842E nNOS red . Companion experiments showed that removal of residual NADP ϩ from the one-electron-reduced enzyme samples prior to their use in the stopped-flow study did not alter the rates of absorbance change we observed at 520 nm (data not shown). The results indicate that interflavin (FADH 2 to FMNH ⅐ ) electron transfer is also faster in the one-electron-reduced, CaM-free Lys 842 mutants than in the WT nNOS red , such that the rates observed in the CaM-free mutants approach that seen in the CaM-bound enzyme.
Conformational Equilibrium of the FMN Module-The cytochrome c reductase activity of CaM-free nNOS red is inversely dependent on the degree of FMN shielding, and the shielding is increased when NADPH is bound (36,37,41,42). It is unknown if the AI has a role in this process. We therefore investigated if the Lys 842 mutations impact the FMN shielding level in nNOS red by performing single turnover cytochrome c reduction reactions in the stopped-flow instrument (37,42). In these experiments, the rate of absorbance increase at 550 nm, which represents cytochrome c reduction by an excess of a prereduced nNOS red protein, is inversely proportional to the degree of FMN shielding and to the relative ratio of the FMN-shielded (or closed) versus FMN-deshielded (or open) conformation of the nNOS red . Fig. 8 contains representative absorbance traces collected for the WT (A) and K842E (B) nNOS red reactions that were run under three different experimental conditions: CaM-and NADPH-free, CaM-free and NADPH-bound, and CaM plus NADPH-bound. The first condition reports on the intrinsic set point of the conformational equilibrium for the given nNOS red protein being studied. The second and third conditions indicate the impacts of NADPH and CaM, respectively, on the conformational equilibrium. Identical experiments were also done with the K842A mutant (data not shown). All of the kinetic traces fit well to a single exponential equation. In the reactions with WT nNOS red , the traces show that bound NADPH slowed the rate of electron transfer to cytochrome c, whereas CaM increased the rate relative to the CaM-free enzyme (with or without NADPH), as reported previously (36,37,41). In contrast, the K842E mutant had a greatly blunted response to NADPH but had a relatively normal response to CaM. Calculated rate constants for the three enzymes are listed in   Ͼ Ͼ WT. In contrast to WT, neither mutant significantly increased its FMN shielding in response to NADPH binding. However, CaM decreased the level of FMN shielding in both mutants, such that their conformational equilibrium became similar to that of the CaM-bound WT nNOS red . These results reveal that the Lys 842 mutations and, by extension, the AI have a significant effect on the conformational equilibrium of nNOS red in the CaM-free condition. Heme Reduction in WT nNOS and Mutants-It is unclear how the AI might impact the heme reduction rate in nNOS. We compared the rates of heme reduction in the CaM-bound mutants and WT nNOS. The experiments involved mixing each CaM-bound nNOS protein with excess NADPH in COsaturated, anaerobic buffer in a stopped-flow spectrophotometer and monitoring the buildup of the ferrous heme-CO complex at 444 nm, which indicates heme reduction (43,44). The absorbance gains we observed at 444 nm were monophasic in all cases and gave the heme reduction rates listed in Table 2. The heme reduction rate was somewhat slower (66% of WT) in K842A nNOS but approached the normal value in K842E nNOS, consistent with their measured NO synthesis activities in Table 1. This indicates that interdomain (FMN to heme) electron transfer can in one case be negatively affected by mutating Lys 842 in the AI of nNOS.

DISCUSSION
The AI is unique to the NOS family of flavoproteins, and its mechanisms of action are of current interest. Previous studies showed that deleting the entire AI changes the catalytic profile of CaM-free nNOS and increases its affinity toward CaM binding (25,26). Our Lys 842 point mutants displayed these two characteristic properties of AI-deleted nNOS, namely an increased cytochrome c reductase activity under CaM-free conditions and an ability to bind CaM at lower Ca 2ϩ concentration. The Lys 842 mutants also had no detectable NO synthesis activity in the absence of CaM, which matches the findings in three of four reports for AI-deleted nNOS (26 -28) (one AI-deleted nNOS was reported to support 10 -30% maximal NO synthesis activity in the absence of CaM) (25). In the CaM-bound condition, we observed a partial loss of NO synthesis activity only in K842A nNOS, which correlated with a lower rate of heme reduction in this mutant. In comparison, among the four AIdeleted, CaM-bound nNOS reported in the literature, one had normal NO synthesis activity (27), one had a 40% reduction in activity (28), and two others had significantly lower activities (70 -80% reductions) (25,26), although in one case this was attributed to the mutant having a substoichiometric FMN content (26). On balance, this comparison suggests that Lys 842 enables AI function in nNOS, particularly the ability of AI to antagonize CaM binding and repress the cytochrome c reductase activity in the CaM-free state. Beyond this, we also found that the Lys 842 mutations altered other important aspects of catalysis that had not been examined before in any of the AIdeleted nNOS. These are discussed below.
Flavin Reduction Kinetics-The CaM-free Lys 842 mutants had faster overall rates of NADPH-dependent flavin reduction (as judged by absorbance changes at 457 or 485 nm) compared with WT nNOS. This implies that the mutations eased the kinetic repression on flavin reduction in CaM-free nNOS and thus reveal for the first time a role for the AI in this fundamental regulation.
Our data also indicate which part of the flavin reduction process (FAD reduction by NADPH versus the FAD to FMN electron transfer) is suppressed by the AI and is relieved by the Lys 842 mutations. Consider that the CaM-free Lys 842 mutants showed no increase in their ferricyanide reductase activity (Table 1). Based on previous work with nNOS (11,31,51,54), the ferricyanide reductase activity would have increased if the  Lys 842 mutations had increased the rate of FAD reduction by NADPH. Thus, the steady-state activity data suggest that the AI restricts its suppressive effect to the subsequent interflavin (FAD to FMN) electron transfer step. This model for AI function is supported by our stopped-flow studies that either monitored the formation of flavin semiquinones during NADPH reduction of the fully oxidized nNOS red proteins (600 nm) or monitored FADsq formation during NADPH reduction of the one-electron-reduced nNOS red proteins (520 nm). In both circumstances, we observed significantly faster rates of absorbance increase in the CaM-free Lys 842 mutants compared with CaM-free WT nNOS red . The spectroscopic data we collected for the fully oxidized proteins when reacted with either molar equivalent or excess NADPH suggest that the observed changes in amplitude at 600 nm were due to a blended contribution from flavin semiquinone formation and formation of an NADP ϩ -FADH 2 charge transfer complex. We were able to kinetically resolve the formation and decay of these two species to some extent (see supplemental Fig. S2). We also found that the absorbance changes attributed to flavin semiquinone formation (or FADsq at 520 nm) in the CaM-free mutants were almost as fast as the rates we observed in the CaM-bound nNOS red . This implies that the Lys 842 mutations incorporated into the AI relieved the kinetic suppression of interflavin electron transfer almost as well as CaM.

NADPH/CaM
The rates of flavin semiquinone formation we observed during reduction of the fully oxidized enzymes (600 nm data sets) were faster in magnitude than the rates of FADsq formation that we observed during reduction of the one-electron-reduced enzymes (520 nm data sets). This difference is probably due to a large difference in thermodynamic driving force that exists in the two circumstances regarding the interflavin electron transfer step. With the fully oxidized enzymes, the reaction involves electron transfer from FADH 2 to FMN to form the FMNsq, which is thermodynamically favorable and essentially an irreversible reaction (51,52). In contrast, with the one-electronreduced enzymes, the reaction involves electron transfer from FADH 2 to FMNsq to form FMNH 2 , which is nearly an equipotential reaction (51,52). This means that the back electron transfer (FMNH 2 to FADsq) will contribute substantially to the rate that we observe during the anaerobic reduction of the oneelectron-reduced proteins by NADPH. However, during steady-state catalysis, where an electron acceptor like cytochrome c is present to drive the equilibrium, the rate of the FAD to FMNsq electron transfer step is likely to be closer to the faster rates that we observed at 600 nm with the fully oxidized enzymes. In any case, a faster FAD-to-FMN electron transfer appears to take place in the CaM-free Lys 842 nNOS red mutants, and this reveals a role for the AI in suppressing this electron transfer step. Because the effect is associated with an increased cytochrome c reductase activity, it also supports the possibility that the rate of FMNH 2 formation helps to determine the steady-state reductase activity of CaM-free nNOS (11,31,32,40).
How might the Lys 842 mutations speed the observed rate of interflavin electron transfer in CaM-free nNOS red ? In the crystal structure, the isoalloxazine rings of the bound FAD and FMN are close enough to support a very fast electron transfer between them (39). However, in solution, we suspect that this conformer of nNOS red (and others like it) exists in a dynamic equilibrium with other conformers that have a less optimal FAD/FMN geometry for electron transfer, and the AI may alter the population distribution to favor the latter. The Lys 842 mutations probably alter the ability of the AI to impact the conformational equilibrium and consequently alter the observed "ensemble" rate of interflavin electron transfer that we can measure in the stopped-flow study. Any differences in flavin reduction kinetics between the full-length and reductase forms of the enzymes (compare Figs. 4 -6) may be due to structural influences of the attached NOSoxy domain that could conceivably influence the electron transfer between the FAD and FMN cofactors in the "closed" conformation of the NOS red . However, it is important to note that the Lys 842 mutations increased the rates of flavin reduction in both forms of the enzyme and thus must act independent of the NOSoxy domain.
Impact on the Conformational Equilibrium of nNOS red -Our data with the Lys 842 mutations revealed a new role for the AI in stabilizing the FMN-shielded conformation of nNOS red . This was unexpected and was not apparent from the crystal structure of nNOS red . How the AI does this is not yet clear. In contrast, the crystal structure of nNOS red shows a clear role for the CT in stabilizing the FMN-shielded conformation. The ␣-helix in the first half of the CT acts as a hairpin to help hold the FMN module in the shielded conformation, where it is in contact with the NADPH-FAD (FNR) subdomain (39). Our current data establish that both the AI and CT are required to stabilize the FMN module in the shielded conformation. Whether their effect involves a physical interaction between the CT and AI elements, as recently proposed (55), deserves further study. In any case, we suspect that the effect of AI on the conformational equilibrium may be a primary cause for the greater (i.e. less repressed) cytochrome c reductase activities of the CaM-free Lys 842 mutants and by extension of the AI-deleted nNOS enzymes as well (25)(26)(27)(28).
NADPH Effect on the nNOS red Conformational Equilibrium-We also found that NADPH no longer increased the level of FMN shielding in the Lys 842 mutants, in marked contrast to its ability to do so in WT nNOS red . This is surprising and provides the first evidence that the AI helps to transduce the effect of NADPH binding on the conformational equilibrium of nNOS red . This effect is potentially important because the conformational equilibrium helps to determine electron flux through CaM-free nNOS, as judged by the cytochrome c reductase activity (36,37,40). Previous work established that the CT and individual residues within it or near it are needed for NADPH to alter the conformational equilibrium (increase FMN shielding) in nNOS (37,41,42). We recently found that the NADPH effect was lost when the second half of the CT was deleted in nNOS red (this is the portion of the CT that lies beyond the helix and is not visible in the crystal structure) (42). Thus, the AI and the second half of the CT both appear to be necessary for NADPH to impact the conformational equilibrium. Whether this involves contact between the AI and CT elements (55) is still unclear.
Does the loss of the NADPH response explain why the CaMfree Lys 842 mutants have greater cytochrome c reductase activ-ities? At first this seems difficult to answer, because the Lys 842 mutations also changed the intrinsic set point of the conformational equilibrium in addition to eliminating the NADPH effect on the equilibrium (see Table 5). However, our related study can address this question (42). We found that deletion of the second half of the CT eliminated the NADPH effect without changing the intrinsic set point of the conformational equilibrium of nNOS red . Significantly, this deletion mutant showed no increase in cytochrome c reductase activity in the CaM-free state (42). This argues that the important effect of the Lys 842 mutations is not their blunting of the NADPH response but rather their ability to change the intrinsic set point of the conformational equilibrium to favor the FMN-deshielded state of nNOS red . On the basis of our related studies (40), this change in the conformational equilibrium is probably the primary way that the Lys 842 mutations cause an increased reductase activity in CaM-free nNOS red .
Impact on CaM-bound nNOS red -The CaM-bound forms of the Lys 842 mutants had cytochrome c reductase activities that were similar to that of CaM-bound WT nNOS. This is consistent with CaM having identical effects on nNOS red as do the Lys 842 mutations in AI (i.e. CaM also increases the rate of flavin reduction and decreases the extent of FMN shielding) (37). The AI-deleted nNOS also showed only minor differences in their cytochrome c reductase activities relative to WT nNOS in the CaM-bound state (25,27,28). Indeed, inhibition by the AI and CT are thought to be disabled upon CaM binding (24,28,36,37). Likewise, the effects of the Lys 842 mutations on AI repressor function become moot once CaM binds to nNOS red . However, the K842A mutation still caused a modest decrease in heme reduction rate and NO synthesis activity in CaM-bound nNOS. This suggests that the AI may still function in the CaMbound nNOS to facilitate an interaction between the FMN module and the oxygenase domain that enables heme reduction (25,26,28,51).
Structural Aspects and Potential Mechanisms-The crystal structure of nNOS red shows that Lys 842 is located on the exposed portion of the helical element that is the central feature of the first half of the AI (39). The AI helix contains a canonical CaM binding motif and in the structure makes contacts with both its parent FMN module and the adjacent FNR and connecting subdomains of nNOS red (Fig. 1) (39). Although the side chain of Lys 842 has low density in the crystal structure and could only be visualized in one of the two reductase structures in the unit cell, it is expected to either project outward or assume a conformation that makes hydrogen bond contact with the side chain of Asn 811 in the FMN subdomain. It is plausible that eliminating or reversing the positive charge at Lys 842 , which lies within the putative CaM binding motif of the AI helix, would diminish an antagonistic interaction of AI with CaM (39) and thus increase the apparent affinity of nNOS toward CaM as is observed. Mutation at Lys 842 might also alter the ability of the AI to directly interact with other subdomains or the CT in nNOS red or might alter the ability of the AI to influence the interactions of the FMN subdomain with its partner subdomains. These structural changes could in turn explain the effects we observed on the interflavin electron transfer kinetics and on the conformational equilibrium of nNOS red and in these ways can explain the increased reductase activity for the mutants in their CaM-free state. Our current results provide a starting point to further investigate these relationships and to define the multifaceted regulation by the AI.