Intra-subunit and inter-subunit electron transfer in neuronal nitric-oxide synthase: effect of calmodulin on heterodimer catalysis.

In neuronal nitric-oxide synthase (nNOS), calmodulin (CaM) binding is thought to trigger electron transfer from the reductase domain to the heme domain, which is essential for O(2) activation and NO formation. To elucidate the electron-transfer mechanism, we characterized a series of heterodimers consisting of one full-length nNOS subunit and one oxygenase-domain subunit. The results support an inter-subunit electron-transfer mechanism for the wild type nNOS, in that electrons for catalysis transfer in a Ca(2+)/CaM-dependent way from the reductase domain of one subunit to the heme of the other subunit, as proposed for inducible NOS. This suggests that the two different isoforms form similar dimeric complexes. In a series of heterodimers containing a Ca(2+)/CaM-insensitive mutant (delta40), electrons transferred from the reductase domain to both hemes in a Ca(2+)/CaM-independent way. Thus, in the delta40 mutant electron transfer from the reductase domains to the heme domains can occur via both inter-subunit and intra-subunit mechanisms. However, NO formation activity was exclusively linked to inter-subunit electron transfer and was observed only in the presence of Ca(2+)/CaM. This suggests that the mechanism of activation of nNOS by CaM is not solely dependent on the activation of electron transfer to the nNOS hemes but may involve additional structural factors linked to the catalytic action of the heme domain.

In neuronal nitric-oxide synthase (nNOS), calmodulin (CaM) binding is thought to trigger electron transfer from the reductase domain to the heme domain, which is essential for O 2 activation and NO formation. To elucidate the electron-transfer mechanism, we characterized a series of heterodimers consisting of one full-length nNOS subunit and one oxygenase-domain subunit. The results support an inter-subunit electron-transfer mechanism for the wild type nNOS, in that electrons for catalysis transfer in a Ca 2؉ /CaM-dependent way from the reductase domain of one subunit to the heme of the other subunit, as proposed for inducible NOS. This suggests that the two different isoforms form similar dimeric complexes. In a series of heterodimers containing a Ca 2؉ /CaM-insensitive mutant (delta40), electrons transferred from the reductase domain to both hemes in a Ca 2؉ /CaM-independent way. Thus, in the delta40 mutant electron transfer from the reductase domains to the heme domains can occur via both inter-subunit and intra-subunit mechanisms. However, NO formation activity was exclusively linked to inter-subunit electron transfer and was observed only in the presence of Ca 2؉ / CaM. This suggests that the mechanism of activation of nNOS by CaM is not solely dependent on the activation of electron transfer to the nNOS hemes but may involve additional structural factors linked to the catalytic action of the heme domain.
An important signaling molecule, nitric oxide (NO), 1 is en-dogenously produced from L-arginine (L-Arg) by a family of enzymes termed nitric-oxide synthases (NOSs). NOS has a bidomain structure consisting of an N-terminal oxygenase domain that has a cytochrome-P450 (P450)-like heme active site and a C-terminal reductase domain that is similar to NADPHcytochrome P450 reductase (reviewed in Refs. [1][2][3][4][5][6][7]. A calmodulin (CaM) binding site is located within the linker region between the two domains. The two constitutive isoforms of NOS, neuronal and endothelial (nNOS and eNOS), are regulated by the reversible binding of Ca 2ϩ /CaM, whereas inducible NOS (iNOS) contains CaM as an intrinsic factor and is Ca 2ϩindependent (reviewed in Refs. [1][2][3][4][5][6][7]. Many different mechanisms have been proposed for the NO synthesis reaction, but there is still no consensus of opinion. In all the mechanisms, the first step is reduction of the heme of the oxygenase domain by an electron generated by NADPH dehydrogenation in the reductase domain, followed by activation of molecular oxygen. It has also been suggested that the essential cofactor tetrahydrobiopterin (H 4 B) donates an electron to the ferrous-dioxy intermediate (7)(8)(9)(10). How are these electron transfers controlled and what are the electron-transfer routes in NOS? CaM binding regulates electron transfer both within the reductase domain and from the reductase domain to the heme domain (11)(12)(13)(14). Siddhanta et al. (15,16) reported that, in the iNOS dimer, which always binds CaM, electron transfer occurs from the flavins of one subunit to the heme domain of the other subunit during catalytic turnover. They therefore proposed a domain swap model for the structure of native iNOS. This also explains why homodimer formation is prerequisite for NOS activity. However, it was not certain that electron transfer occurs via the same mechanism in the nNOS and eNOS dimers, both of which are controlled by reversible Ca 2ϩ /CaM binding.
Sequence alignment between the three NOS isoforms reveals that both nNOS and eNOS contain an extra 40 -46 amino acids within their FMN binding subdomains, which are absent in iNOS. Our previous report and others demonstrated that removal of these amino acids affects the Ca 2ϩ /CaM-dependent activation of nNOS and eNOS, suggesting that they are autoinhibitory elements (14,(17)(18)(19). The heme of wild type nNOS was reduced with NADPH only in the presence of Ca 2ϩ /CaM, whereas the heme of the 40-amino acid deletion mutant (delta40) was reduced spontaneously upon addition of NADPH in the absence of Ca 2ϩ /CaM. The delta40 mutant had 10% of the activity of the wild type enzyme without Ca 2ϩ /CaM, which was increased 3-fold upon addition of Ca 2ϩ /CaM (14). Our goal is to understand how CaM binding regulates inter-subunit/ * This work was supported in part by Grants-in-aid (to T. S.) for Priority Area (biometallics) (11116201) and (to I. S.) for General Area (12680624) from the Ministry of Education, Science, Sports and Culture of Japan and by a Royal Society (United Kingdom) Fellowship (to S. D.). 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  intra-subunit electron transfer to mediate heme reduction and NO formation activity in both the wild type enzyme and the delta40 mutant.
In the present study, we have generated the following nNOS mutants: full-length E592A (a point mutation in the heme domain removing substrate binding capability), full-length E592Adelta40 mutant (combining the E592A mutation with deletion of 40-amino acids in the FMN binding subdomain), and the oxygenase domains of the wild type enzyme (Wildox) and the E592A mutant (E592Aox). To elucidate the route of electron transfer between the heme domain and the reductase domain in nNOS (i.e. intra-or inter-subunit), we prepared a series of heterodimers, each composed of a full-length subunit and an oxygenase domain subunit. We discuss the electron transfer routes from the reductase domain to the heme active sites in the heterodimers and the regulation of catalytic activities by Ca 2ϩ /CaM. Construction of the nNOS Mutants Expressing Plasmids-The cDNA for rat nNOS was kindly given by Dr. S. H. Snyder (Johns Hopkins University, School of Medicine). The expression plasmids for full-length nNOS of wild type and delta40 mutant were constructed as described previously (14,20). For E592A, site-directed mutagenesis was performed with the polymerase chain reaction-based strategy using the ODA-LA PCR kit from Takara Shuzo and a primer, 5Ј-ATG GGC ACA GCG ATC GGC GTC-3Ј. To clone Wildox containing the oxygenase domain (residues 1-720), the downstream primer used for PCR was 5Ј-GGTCTAGATTA(GTG) 6 GTTGGTGCCCTTCCACACG-3Ј, in which a 6xHis-tag was also attached to the C terminus of the Wildox mutant, and a stop codon and an XbaI site were included. We used the upstream primer, 5Ј-TGTGGAATTGTGAGC-3Ј, which corresponds to the sequence in the pKF19k plasmid. The desired PCR products were confirmed by sequencing using an automatic sequencer DSQ-2000L (Shimadzu Co., Kyoto, Japan). cDNA fragments containing wild type and the desired mutations were cloned into NdeI and XbaI sites of the vector pCWori ϩ and transformed into E. coli strain BL21(DE3) for expressing most mutants and the wild type enzyme (20).

Materials
Expression of Full-length nNOSs and the Oxygenase Domains-The nNOSs were expressed in BL21 E. coli cells, containing another plasmid, pGroESL, for expression of chaperone proteins (20 -22). The E. coli cells containing expressing plasmids were cultured at 37°C until the A 600 reached 1.4. Then, isopropyl-␤-D-thiogalactopyranoside, ␦-aminolevulinic acid, and riboflavin were added to the culture medium to make final concentrations of 0.5 mM, 0.45 mM, and 3 M, respectively. Further incubation with mild shaking at 25°C continued for another 20 h for the oxygenase domains or 40 h for the full-length enzymes.
Purification of Full-length nNOSs and the Oxygenase Domains-Full-length wild type and mutants of nNOS were purified using DEAE-TOYOPEARL, 2Ј,5Ј-ADP-Sepharose and calmodulin-Sepharose column chromatographies as described previously (14, 20 -22). For purification of Wildox and the other oxygenase domain mutants, E. coli cells were harvested by centrifugation at 3000 rpm and resuspended in buffer A (50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 mM EGTA, 10% glycerol, 10 M H 4 B, 1 mM DTT, 1 mM PMSF, 2 g/ml aprotinin, 2 g/ml leupeptin, and 2 g/ml pepstatin). The E. coli cells were crushed in buffer A by sonication on ice. After centrifugation at 100,000 ϫ g for 30 min at 4°C, ammonium sulfate was added to the resulting supernatant up to 40% saturation. The precipitates were collected and then dissolved in buffer B (50 mM sodium phosphate buffer, pH 7.8, 1 mM EDTA, 10% glycerol) containing 10 M H 4 B, 1 mM DTT, 1 mM PMSF, 2 g/ml aprotinin, 2 g/ml leupeptin, and 2 g/ml pepstatin. The solution was passed through a Sephadex G-25 column pre-equilibrated with the same buffer and was applied to an Ni-NTA-agarose column (Qiagen, Tokyo, Japan) pre-equilibrated with buffer B containing 20 mM imidazole, 10 M H 4 B, and 1 mM DTT. The column was washed with buffer B containing 20 mM imidazole. The oxygenase domain protein was eluted with buffer B containing 100 mM imidazole. After concentration, the enzyme fraction was quickly frozen in liquid nitrogen and stored at Ϫ80°C.
Before analysis, the buffers were changed to buffer C (50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 10% glycerol) containing 5 M H 4 B, and 20 M DTT using a Sephadex G-25 column. For preparation of H 4 B-free enzymes, all buffers used for the purification procedure contained 1 mM ␤-ME instead of H 4 B and DTT. All purified enzymes were more than 90% pure as judged by SDS-PAGE stained with Coomassie Blue R-250. The concentrations of the nNOSs were determined optically from the [Fe(II)⅐CO] Ϫ [Fe(II)] difference spectra using ⌬⑀ 444 -467 nm ϭ 55 mM Ϫ1 cm Ϫ1 . This ⌬⑀ value was estimated by the pyridine hemochromogen method (24) assuming that one heme is bound per monomer (14, 20 -22).
Preparation of Heterodimers-Essentially we followed the method developed by Dr. D. Stuehr (15,16). The full-length nNOSs and the oxygenase domain dimers were dissociated into monomers by incubation with 3 M urea in buffer C containing 1% Tween 20 and 2 mM ␤-ME for 1.5 h at 25°C and then for 1 h at 0°C. To generate heterodimers, the enzyme mixture was diluted five times with buffer C containing 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT, and incubated at 0°C for more than 3 h. Then the mixture was loaded on a Ni-NTA-agarose column equilibrated with buffer C containing 20 mM imidazole, 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT. The column was washed with a 10ϫ column volume of the equilibration buffer. The bound protein was eluted with buffer C containing 100 mM imidazole, 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT. The eluted fraction was then applied to a 2Ј,5Ј-ADP-Sepharose column equilibrated with buffer C containing 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT. After washing with a 10ϫ column volume of buffer C containing 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT, the heterodimer was eluted with buffer C containing 10 mM NADPH, 1 mM L-Arg, 10 M H 4 B, and 1 mM DTT.
Gel Filtration-To examine the dimer contents of the mutants, purified full-length NOSs and heterodimers were analyzed on a Superdex 200 HR10/30 size exclusion chromatography column (Amersham Pharmacia Biotech), equilibrated with 50 mM Tris-HCl (pH 7.5) buffer containing 0.2 M NaCl, 0.1 mM EDTA, and 0.1 mM L-Arg and connected to a fast-protein liquid chromatography system (Amersham Pharmacia Biotech). The molecular masses of the protein peaks were estimated relative to the molecular mass of standards proteins: ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), albumin (67 kDa), and ovalbumin (43 kDa).
Enzyme Assay-The rate of NO formation was determined from the NO-mediated conversion of oxyhemoglobin to methemoglobin, monitored at 401 nm using a methemoglobin minus oxyhemoglobin extinction coefficient of 49 mM Ϫ1 cm Ϫ1 (2). The NADPH oxidation rate was determined spectrophotometrically as an absorbance decrease at 340 nm, using an extinction coefficient of 6.22 mM Ϫ1 cm Ϫ1 . Unless otherwise indicated, assays were carried out at 25°C in 50 mM Tris-HCl (pH 7.5) buffer containing 10 M oxyhemoglobin, 0.1 mM NADPH, 5 M each of FAD and FMN, 10 g/ml CaM, 1 mM CaCl 2 , 100 units/ml catalase, 10 units/ml superoxide dismutase, 5 M H 4 B, 20 M DTT, and 0.05-0.1 M nNOS in the presence or absence of 0.1 mM L-Arg (15, 19 -21).
Optical Absorption Spectra-Spectral experiments under aerobic conditions were carried out on a Shimadzu UV-2500 spectrophotometer maintained at 25°C by a temperature controller. Anaerobic spectral experiments to measure heme reduction rates were conducted on a Shimadzu UV-160A spectrophotometer maintained at 15°C in a glove box under a nitrogen atmosphere with an O 2 concentration of less than 50 ppm. To ensure that the temperature of the solution was appropriate, the cell was incubated for 10 min prior to spectroscopic measurements. The experiments were repeated at least three times. Regression analyses were performed, and lines giving an optimum correlation coefficient were selected. Experimental errors were less than 10%.

Full-length E592A and E592Adelta40
Mutants-The resting Fe(III) forms of the E592A and E592Adelta40 mutants in the absence of L-Arg and H 4 B had broad Soret peaks at 400 nm, indicating a mixture of high and low spin heme types (Fig. 1). The position of the peak did not move upon addition of 1 mM L-Arg, whereas addition of 10 M H 4 B shifted the Soret peak position to 396 nm, indicating that the heme irons became mostly the high spin type. These results suggest that the mutations affected the substrate binding ability but did not change the H 4 B binding in terms of the absorption spectral character.
Under anaerobic conditions, the addition of NADPH to E592A caused a broad decrease in absorbance between 400 and 500 nm, indicating the reduction of the bound flavins of the enzyme. In the absence of Ca 2ϩ /CaM, no further change was observed over a period of 30-min incubation (Fig. 1A). Addition of Ca 2ϩ /CaM, however, initiated heme reduction as observed for wild type full-length nNOS and resulted in the appearance of a peak at 444 nm in the presence of CO. On the other hand, the E592Adelta40 mutant showed clear heme reduction with NADPH in the absence of Ca 2ϩ /CaM under anaerobic conditions (Fig. 1B). The Fe(II) heme⅐CO complexes of both the mutants had a peak at 444 nm as observed for wild type full-length nNOS.
To elucidate how much heme reduction with NADPH occurred in the absence or presence of Ca 2ϩ /CaM, we measured the absorbance change of the mutants at 444 nm with respect to time in a saturated solution of CO under anaerobic conditions. As summarized in Table I, after 30-min incubation with NADPH, heme reduction was about 30% complete in wild type nNOS and the E592A mutants in the absence of Ca 2ϩ /CaM. The rates of reduction were also slow. Note that Ca 2ϩ /CaMindependent heme reduction was observed even in the presence of 10 mM EDTA (data not shown). Addition of Ca 2ϩ /CaM to the system resulted in an increase in the reduction rate (10-fold) and the proportion of heme reduced (up to 90%). In contrast, for the delta40 and E592Adelta40 mutants, about 60 -85% heme reduction occurred in the presence of NADPH and in the absence of Ca 2ϩ /CaM, with comparable rates to that of the wild type enzyme in the presence of Ca 2ϩ /CaM. NO formation activities and NADPH oxidation rates are also summarized in Table I. For the wild type enzyme, NO formation activity could only be detected in the presence of Ca 2ϩ / CaM even though some NADPH oxidation and heme reduction occurred under the same conditions in the absence of Ca 2ϩ / CaM. Addition of Ca 2ϩ /CaM is therefore a trigger for NO formation in the wild type enzyme, consistent with previous re-ports. In the presence of L-Arg and Ca 2ϩ /CaM, NADPH consumption and NO formation are well coupled in the wild type enzyme.
For the delta40 mutant, significant NO formation was observed in the absence of Ca 2ϩ /CaM but not as much as expected from the heme reduction rate and ratio. Addition of Ca 2ϩ /CaM increased the NO formation activity and rate of NADPH oxidation in the presence of L-Arg. Coupling between NO formation and NADPH consumption was not as efficient in the mutant even in the presence of L-Arg and Ca 2ϩ /CaM as described previously for the delta40 mutant expressed in yeast (14). Addition of L-Arg did not lower the NADPH consumption rates for either the wild type enzyme or the delta40 mutant in the absence of Ca 2ϩ /CaM.
As expected, full-length E592A and E592Adelta40 mutants had no NO formation activities. The NADPH oxidation rates of these mutants were increased on addition of Ca 2ϩ /CaM, but were not subsequently lowered by addition of L-Arg, unlike the wild type enzyme. These results confirm that the mutants are insensitive to the substrate.
We also analyzed the dimer formation ability of these mutants using fast-protein liquid chromatography gel filtration. More than 75% of each of the delta40, E592A, and E592Adelta40 mutants were eluted as dimers after incubation with 1 mM L-Arg and 10 M H 4 B (data not shown).
Preparation of Heterodimers-To prepare heterodimers consisting of a full-length nNOS subunit and an oxygenase domain subunit with a six-histidine tag at the C terminus, we first treated the full-length dimers and oxygenase domain dimers with 3 M urea. Fig. 2 shows typical gel-filtration profiles for the E592Adelta40 mutant, the Wildox mutant, and E592Adelta40/ Wildox heterodimer. After urea treatment, about 30% of the E592Adelta40 was eluted as monomer, whereas only 10% of the Wildox was monomeric. Similarly, the monomeric forms of the other full-length nNOSs (wild type, E592A, and delta40) were obtained more easily than those of the corresponding oxygenase domain proteins. The monomer proportions were not increased even when the dimers were treated with higher concentrations of urea for longer time periods. More severe conditions destroyed the NOS proteins. After renaturation by 10ϫ dilution and incubation with buffer containing H 4 B and L-Arg, each heterodimer was separated and purified from the homodimers and the monomers by sequential chromatographies. Namely, at first, we isolated the enzyme fractions containing histidine tags (i.e. containing oxygenase domain subunits) using Ni-NTA column chromatography. We then isolated the enzymes containing NADPH binding sites (i.e. containing full-length subunits) using a 2Ј,5Ј-ADP-Sepharose column chromatography. Consequently, only heterodimers containing a full-length nNOS subunit and an oxygenase domain subunit are retained. The Fe(III), Fe(II), and Fe(II)⅐CO complexes of the heterodimers had similar spectra to those of the wild type enzyme. In the spectra of the Fe(II)⅐CO complexes, no absorption band at around 420 nm ascribed to the denatured complex, P420, was observed (as discussed later in Fig. 5), confirming that the heme domains were not altered by these treatments.
To determine the molecular size, we analyzed the isolated heterodimer fraction using gel filtration chromatography (Fig.  2). The E592Adelta40/Wildox heterodimer fraction was eluted as one peak at about 245 kDa, i.e. between the E592Adelta40 homodimer (320 kDa) and the Wildox homodimer (160 kDa). The other heterodimers, E592A/Wildox, Wild/E592Aox, delta40/Wildox, and delta40/E592Aox had similar gel filtration profiles. We also examined the components of the heterodimer fractions using SDS-PAGE after denaturation by heating with 2% SDS and 2% ␤-ME. As shown in Fig. 3, the results revealed that delta40/Wildox, E592Adelta40/Wildox, and delta40/ E592Aox heterodimer fractions contained approximately equal amounts of the full-length subunit (160 kDa) and the oxygenase domain subunit (80 kDa). The purity of these heterodimers was greater than 90% as judged by SDS-PAGE.
Catalytic Activities of the Heterodimers-The NO formation activities and heme reduction properties of the heterodimers are summarized in Table II. NO formation activity was observed with the renatured Wild/Wildox heterodimer only in the presence of Ca 2ϩ /CaM and was about 10% that of the wild type homodimer enzyme. NADPH reduced 34% of the heme in the presence of Ca 2ϩ /CaM under anaerobic conditions. The rate of heme reduction was about 10% that of the wild type homodimer. The E592A/Wildox heterodimer showed similar results for NO formation activity and heme reduction, whereas for the Wild/E592Aox heterodimer 35% heme reduction but no NO formation activity was observed in the presence of Ca 2ϩ / CaM. In contrast to the homodimer, the hemes of the heterodimers were not reduced by NADPH in the absence of Ca 2ϩ /CaM. We next analyzed a series of heterodimers formed with the delta40 mutant, in which 40 amino acids have been deleted. Interestingly, the hemes of the delta40/Wildox, E592Adelta40/ Wildox, and delta40/E592Aox heterodimers were reduced by NADPH even in the absence of Ca 2ϩ /CaM and the percentages of heme reduction reached 90%, 70%, and 78%, respectively (Figs. 4 and 5 and Table II). These results indicate that both the hemes in these heterodimers were reduced by NADPH in the absence of Ca 2ϩ /CaM under anaerobic conditions unlike in heterodimers containing the wild type reductase domain. The heme reduction rates were similar to those of the other heterodimers and slower than those of the wild type and delta40 homodimers ( Table I). Two of the three heterodimers, delta40/ Wildox and E592Adelta40/Wildox, had NO formation activities in the presence of Ca 2ϩ /CaM, which were similar to that of the Wild/Wildox heterodimer. Like the Wild/E592Aox heterodimer, In the presence (ϩ) or absence (Ϫ) of 1 mM Ca 2ϩ and 10 g/ml CaM. b Rate of conversion of oxyhemoglobin (10 M) to methemoglobin on NO synthesis at 25°C in the presence of 0.1 mM NADPH and 0.5 mM L-Arg.
c Heme reduction was estimated by percentage of the reduced heme after 30-min reaction with 0.1 mM NADPH in the presence of 0.5 mM L-Arg assuming the amount of heme reduction with dithionite is 100%. the delta40/E592Aox heterodimer had no NO formation activity under any conditions, indicating that only electrons transferred from the reductase domain of one subunit to the heme of the adjacent subunit are used for NO formation.

Effect of the E592A Mutation on Electron Transfer and Dimer
Formation-The crystal structures of the eNOS and iNOS heme domain dimers revealed that L-Arg binds to the active site of the heme domain with hydrogen bonding to Glu-371 of mouse iNOS (Glu-361 in human eNOS, corresponding to Glu-592 in rat nNOS) (23)(24)(25)(26). The E592A mutation resulted in a loss of sensitivity to L-Arg, characterized by a lack of spectroscopic perturbation on its addition (Fig. 1). This was corroborated by results from competitive imidazole binding experiments. Addition of L-Arg to the mutant in the presence of bound imidazole did not cause ligand displacement, or any spectral perturbation, unlike for the wild type enzyme (data not shown). Addition of H 4 B to the mutant caused a shift in the Soret peak position from 400 to 396 nm, suggesting that the mutation was not likely to affect H 4 B binding. As expected, the E592A mutant did not show any NO formation activity like the corresponding mutants of iNOS and eNOS (16,28). In the wild type enzyme, the NADPH oxidation rate was lowered on addition of L-Arg due to coupling with NO formation. In contrast, the rate of NADPH oxidation by the E592A mutant was not affected by L-Arg, consistent with its binding site being defective. For NO formation activity, NOS dimer formation is prerequisite. Gel filtration analysis indicated that the mutation did not affect either the percentage of homodimer formed by the full-length enzyme, or the percentage of heterodimer formed with the isolated oxygenase domain mutants. These results are consistent with the previous reports, which demonstrated that dimer formation does not require L-Arg in any of the NOS isoforms (27)(28)(29).
CaM Binding Is Essential for NO Formation but Not for Heme Reduction-It has been demonstrated that the binding of Ca 2ϩ /CaM is a trigger for electron transfer from the reductase domain to the heme in nNOS and eNOS (11,12). CaM binding also facilitates flavin reduction by NADPH and electron transfer from the reductase domain to exogenous electron acceptors such as cytochrome c or ferricyanide (13,14). Like the eNOS isoform, wild type nNOS has NO formation activity only in the presence of Ca 2ϩ /CaM, indicating a critical role for Ca 2ϩ /CaM in NOS activity. Although FMN to heme electron transfer is triggered by Ca 2ϩ /CaM binding, 38% of the heme of the wild type enzyme was reduced by NADPH in the absence of Ca 2ϩ / CaM under anaerobic conditions, albeit at a slow rate, as shown in Table I. Addition of Ca 2ϩ /CaM to the system resulted in the appearance of NO formation activity concomitant with a dramatic increase in the heme reduction rate. The E592A mutant also showed similar results for heme reduction by NADPH, except it could not support NO formation from L-Arg due to its inherent defective substrate binding ability, as mentioned above. It appears, therefore, that Ca 2ϩ /CaM and substrate binding are not absolutely essential for electron transfer from the reductase domain to the heme in nNOS. These results are also consistent with recent reports that CaM has very little influence on the flavin reduction potentials of nNOS (30) and that the reduction potential of the nNOS heme was not affected by substrate binding and is thermodynamically accessible to reduction by the flavins (31). On the other hand, the wild type enzyme absolutely required Ca 2ϩ /CaM binding for NO formation activity. CaM binding is likely to induce a conformational change for efficient electron transfer between the domains. It is thus suggested that there might be differences in active site conformation and/or the electron transfer route for the first electron required for initial heme reduction and subsequent electrons required for catalytic turnover.
Inter-subunit Electron Transfer in Wild type nNOS-Analysis of a heterodimer consisting of a full-length wild type subunit and a wild type oxygenase domain subunit revealed that only one heme in the dimer was reduced by NADPH in the presence of Ca 2ϩ /CaM (Table II). This is similar to the results obtained by Siddhanta et al. using iNOS heterodimers (15,16). They demonstrated that a single reductase domain is sufficient to support heme reduction and NO formation activity at one heme site. They also proposed a domain-swapping model for iNOS, in which electron transfer occurs only between the flavins and hemes located on adjacent subunits (16). iNOS irreversibly binds CaM and its activity is insensitive to Ca 2ϩ concentration. Therefore, it is very difficult to elucidate how CaM binding regulates NO synthesis and the individual electron transfer steps using iNOS as a model. In the case of nNOS, the heterodimers E592A/Wildox and Wild/E592Aox showed that only one of two hemes was reduced by NADPH in the presence of Ca 2ϩ /CaM as in the corresponding heterodimers of iNOS. Also, NO formation activity was observed only in the presence of Ca 2ϩ /CaM for the E592A/Wildox heterodimer (this has a deactivated full-length subunit and a wild type oxygenase domain subunit), whereas the Wild/E592Aox heterodimer did not show any activity despite containing a full-length wild type subunit (Table II). These results support the model proposed for iNOS, in which electrons for heme reduction flow from the flavins of the other subunit in the dimer (A-D in Fig. 6). It should also be noticed that heme reduction in the heterodimers occurred in the presence of Ca 2ϩ /CaM, but not in its absence, unlike in the case of the wild type and E592A full-length homodimers. There is clearly a difference between the heterodimers and homodimers in terms of the abilities of their respective reductase domains to transfer electrons to the heme domains and to drive NO synthesis. The wild type heterodimer also has a much lower catalytic rate than the corresponding homodimer. This suggests that the conformation of the heterodimer is not identical to that of the homodimer, which may explain why there appears to be no heme reduction without Ca 2ϩ /CaM in the heterodimers. If the two reductase domains of the homodimer are in close proximity, it is likely that loss of one will affect the conformation of the other. The location and/or position of the NOS reductase domains with respect to the oxygenase domains are, however, uncertain.
Inter-and Intra-subunit Electron Transfer in the delat40 Mutants-In our previous paper, it was shown that the nNOS delta40 mutant, which has a 40-amino acid deletion in the FMN subdomain, binds CaM at lower Ca 2ϩ concentrations and retains NO formation activity in the absence of Ca 2ϩ /CaM, supporting the idea that the insert is an autoinhibitory element (14). In the absence of Ca 2ϩ /CaM, both hemes in the delta40 homodimer were reduced under anaerobic conditions with a similar rate to that of the wild type enzyme in the presence of Ca 2ϩ /CaM (Table I). The deletion seems to activate both heme reduction and NO synthesis in a way similar to that of Ca 2ϩ / CaM binding, which activates the wild type enzyme. In the homodimer of the E592A delta40 mutant, 60% of the heme was reduced with NADPH in the absence of Ca 2ϩ /CaM. The addition of Ca 2ϩ /CaM resulted in further heme reduction up to 85%, as in the delta40 homodimer mutant ( Table I). Loss of substrate binding to the heme site did not appear to unduly affect heme reduction.
The deletion mutation appears to cause both hemes of the heterodimers to be reduced by NADPH in the absence or presence of Ca 2ϩ /CaM. These results indicate that, in the delta40 mutants, electrons can be transferred to the heme domains from the flavin located on either the same subunit (intrasubunit) or the adjacent subunit (inter-subunit), as shown in Fig. 6, E-H. It is unlikely that inter-heme electron transfer occurs in these heterodimers, because it would also be expected to occur in the other heterodimers (such as Wild/Wildox (B), E592A/Wildox (C), and Wild/E592Aox (D) in Fig. 6) in the presence of Ca 2ϩ /CaM.
The delta40 heterodimers were found to require Ca 2ϩ /CaM binding for NO formation activity, just like the wild type enzyme (Table I). The delta40 homodimer, however, retained about 30% of its activity in the absence of Ca 2ϩ /CaM (14). These results again suggest that there is a conformational difference between the homodimers and heterodimers. In this case, the effect of heterodimer formation is to prevent catalytically viable electron transfer from the delta40 reductase to the hemes in the absence of Ca 2ϩ /CaM (Fig. 6E). It is interesting that the reductase domain of the delta40 heterodimers is able to reduce both hemes in the absence of Ca 2ϩ /CaM but is unable to drive NO synthesis (Fig. 6, F-H). An optimal structure/ conformation necessary for NO formation activity may not be retained in the delta40 heterodimer in the absence of Ca 2ϩ / CaM, whereas, in the delta40 homodimer, that would be well preserved, because the homodimer has two reductase domains. These results support the idea that CaM binding and/or the presence of the two reductase domains are required for the formation of catalytically active dimer as proposed previously (32,33). Under assay conditions, it may be that electron transfer is too slow to drive NO synthesis in the delta40 heterodimers in the absence of Ca 2ϩ /CaM. The deletion mutation also had the effect of activating the reductase domain in the absence of Ca 2ϩ /CaM as a nonspecific reductase (13,14). It would not, therefore, be surprising if heterodimers containing this domain were more likely to undergo inter-dimer heme reduction.
It is possible that the route of the first electron transfer to the heme is different from the route of the second electron transfer, which is required for activation of the ferrous-dioxy intermediate during catalysis. Recent studies suggest that the second electron might be transferred from H 4 B, which is then regenerated by the reductase domain (7)(8)(9)(10). Ca 2ϩ /CaM binding may be required for this process to occur (in all but the delta40 homodimers). From the analysis of nNOS⅐NO complex formation, we found that the heme⅐NO complex of nNOS is easily reduced by NADPH in the presence of both L-Arg and H 4 B even in the absence of Ca 2ϩ /CaM (34). These results support the idea of different transfer routes and regulation mechanisms for the first and second electrons. To confirm this and understand the precise role of CaM binding, further experiments remain to be carried out.
Heme Reduction Rates-Lastly, the heme reduction rates FIG. 6. Hypothetical models of intra-and inter-molecular electron transfers in wild type nNOS and mutants. In the wild type homodimer (A), electrons cross over from one subunit to the other subunit for both heme reduction and NO formation catalysis. Similarly, electrons for heme reduction and for molecular oxygen activation cross over from one subunit to the other in a Ca 2ϩ /CaM-dependent way in the heterodimers (B, C, and D), which are composed of wild type and the E592A mutant subunits. In the delta40 heterodimers (F, G, and H), the first electron to reduce the heme transfers within the same subunit and/or between the subunits in a Ca 2ϩ /CaM-independent way, but the second electron, required to activate molecular oxygen, crosses over from one subunit to other subunit in a Ca 2ϩ /CaM-dependent way. In the delta40 homodimer (E), however, electrons transfer in a Ca 2ϩ /CaMindependent way. Black and white arrows indicate inter-molecular and intra-molecular electron transfer, respectively. The reductase domain containing the 40-amino acid deletion in the heterodimers (F, G, and H) might have a slightly different orientation with respect to the oxygenase domain from that of the reductase in the homodimer (E) in the absence of Ca 2ϩ /CaM. Filled squares indicate the extra 40-amino acid sequences in the full-length subunits. described in the tables are low compared with the relevant NO formation rates and the NADPH oxidation rates as reported for the wild type homodimer in our previous studies (20,22). The heme reduction described here was determined at 15°C under anaerobic conditions, whereas NADPH consumption and NO formation rates were determined at 25°C under aerobic conditions. In addition, note that the heme reduction described here reflects the introduction of the first electron to the heme iron and the slow equilibration between NADPH and heme during incubation. Although it is certain that the involvement of heme is essential for NO formation activity, it is not clear how subsequent electrons are delivered and circulated for the activation of O 2 during catalysis. For example, electron donation from H 4 B to the O 2 -bound heme has been implied (8 -10). Involvement of H 4 B in reduction of the NO⅐heme complex (a proposed dead-end complex) has been suggested (34), and electrons from the substrate itself may also be used to active O 2 (9,24). Further studies remain to be carried out to address these questions.
Summary-The following were suggested from the present study: 1) Electrons transfer from the reductase domain of one subunit to the oxygenase domain of the other subunit in wild type nNOS homodimers and heterodimers in a Ca 2ϩ /CaM-dependent way, as has been suggested for iNOS; 2) In the delta40 mutant heterodimer electrons transfer from the reductase domain of one subunit to both of the oxygenase domains in a Ca 2ϩ /CaM-independent way; 3) In the delta40 mutant heterodimers, electrons required for the NO formation activity transfer exclusively from the reductase domain of the delta40 mutant to the oxygenase domain of the other subunit in a Ca 2ϩ /CaM-dependent way. It is clear, therefore, that catalytically relevant electron transfer events occur in an inter-subunit manner in both the wild type and delta40 mutant forms of nNOS. This study has also showed that intra-subunit electron transfer is possible in the delta40 mutant, in contrast to the wild type enzyme. The 40 amino acids in the FMN binding subdomain appear to be very important in regulating electron transfer between the domains in nNOS and particularly in controlling the conformational integrity of the interdomain complex in the presence and absence of CaM.
Note Added in Proof-During the review of this paper, we became aware of a report describing inter-subunit electron transfer in wild-type nNOS by Dr. D. Stuehr's group (35). Their results are consistent with those reported in this paper.