Structural Elements Contribute to the Calcium/Calmodulin Dependence on Enzyme Activation in Human Endothelial Nitric-oxide Synthase*

Two regions, located at residues 594–606/614–645 and residues 1165–1178, are present in the reductase domain of human endothelial nitric-oxide synthase (eNOS) but absent in its counterpart, inducible nitric-oxide synthase (iNOS). We previously demonstrated that removing residues 594–606/614–645 resulted in an enzyme (Δ45) containing an intrinsic calmodulin (CaM) purified from an Sf9/baculovirus expression system (Chen, P.-F., and Wu, K.K. (2000) J. Biol. Chem. 275, 13155–13163). Here we have further elucidated the differential requirement of Ca2+/CaM for enzyme activation between eNOS and iNOS by either deletion of residues 1165–1178 (Δ14) or combined deletions of residues 594–606/614–645 and 1165–1178 (Δ45/ Δ14) from eNOS to mimic iNOS. We measured the catalytic rates using purified proteins completely free of CaM. Steady-state analysis indicated that the Δ45 supported NO synthesis in the absence of CaM at 60% of the rate in its presence, consistent with our prior result that CaM-bound Δ45 retained 60% of its activity in the presence of 10 mm EGTA. Mutant Δ14 displayed a 1.5-fold reduction of EC50 for Ca2+/CaM-dependence in l-citrulline formation, and a 2–4-fold increase in the rates of NO synthesis, NADPH oxidation, and cytochrome c reduction relative to the wild type. The basal rates of double mutant Δ45/Δ14 in NO production, NADPH oxidation, and cytochrome c reduction were 3-fold greater than those of CaM-stimulated wild-type eNOS. Interestingly, all three activities of Δ45/ Δ14 were suppressed rather than enhanced by Ca2+/CaM, indicating a complete Ca2+/CaM independence for those reactions. The results suggest that the Ca2+/CaM-dependent catalytic activity of eNOS appears to be conferred mainly by these two structural elements, and the interdomain electron transfer from reductase to oxygenase domain does not require Ca2+/CaM when eNOS lacks these two segments.

reductase domain, which harbors the FAD and FMN cofactors and the NADPH binding site, to the N-terminal oxygenase domain, which contains the heme catalytic center, the H 4 B cofactor, and the arginine binding sites (1)(2)(3)(4)(5)(6)(7)(8). An electron generated by NADPH is transferred in tandem to FAD and FMN and then to the heme, which has been proposed to be facilitated by CaM bound to a site situated between these two domains (9). The NOS family comprises three isoforms that share domain structures, sequence homology, and catalytic properties (10 -12). Despite these similarities, there are considerable differences among the NOS isoforms with respect to their cellular expressions, Ca 2ϩ -dependent CaM activation, and rate of electron transfer. Two isoforms, i.e. nNOS and eNOS, are constitutively expressed NOSs (cNOSs), but their expressed enzymes are latent until CaM binding is elicited by an elevated intracellular calcium level (13). In contrast, iNOS is absent or expressed in low abundance at the resting state, and its expression is induced by cytokines and endotoxins. The expressed iNOS is catalytically active, thought to be due to its high affinity for CaM binding even at a basal level of intracellular calcium (14). Among the three isoforms, eNOS has lower electron transfer rate and catalytic activity than nNOS and iNOS (15), suggesting other control mechanisms for eNOS catalysis.
The x-ray crystallographic analysis of the oxygenase domains of NOS isoforms has revealed a striking degree of conservation at the active-site structure (16 -18). The results from studies of chimeric enzymes in which the oxygenase domain was swapped to the reductase domain of another isoform suggested that divergence in the reductase domain rather than in the oxygenase domain accounted for the differences in Ca 2ϩ sensitivity and the rate of electron transfer between the cNOSs and the iNOS (15). An ϳ50-amino acid fragment ( Fig. 1) present in the FMN-subdomain of human eNOS (residues 594 -645) and nNOS (residues 834 -882), but absent in the corresponding part of iNOS, was proposed as an autoinhibitory element that impedes the electron transfer of cNOSs in the absence of CaM (19,20). Deletion of this region rendered the mutant enzymes less dependent on Ca 2ϩ concentration, with a faster rate of electron flow (21)(22)(23). We previously characterized an eNOS mutant (24) in which residues 594 -606 and 614 -645 were deleted (⌬45eNOS) with preservation of residues 607-613 because it was conserved between the sequences of cNOSs and iNOS (Fig. 1). The ⌬45eNOS contained an endogenous CaM bound to the protein isolated from an Sf9/ baculoviral expression system. This mutant was completely CaM-independent as well as significantly Ca 2ϩ -independent in L-citrulline formation and exhibited a higher rate of cytochrome c reduction in a CaM-independent manner (24). The results confirm that residues at 594 -606 and 614 -645 in eNOS control calcium sensitivity for CaM-dependent enzyme activation. However, because this mutant still requires Ca 2ϩ to achieve a maximal catalytic activity, the calcium requirement for electron transfer and NO production may be controlled by other intramolecular mechanisms.
Recent studies have implicated the C-terminal region of all NOS isoforms as an additional regulatory element in modulating electron transfer. The C terminus of cNOS contains a conserved serine (Ser 1178 in human eNOS and Ser 1417 in nNOS) with a kinase-dependent phosphorylation motif (RSRXX(S/T)) that has been noted in eNOS to be phosphorylated in response to a number of stimuli (25)(26)(27)(28)(29)(30)(31)(32). Phosphorylation has been shown to trigger eNOS activation at a lower Ca 2ϩ concentration and to increase the rate of NO production by 2-fold in vivo (25)(26). Mutation of this serine to Asp in eNOS (33) and nNOS (34), which mimics phosphorylation by introducing a negative charge, also causes a faster electron flow through enzymes. By sequence comparison with cytochrome P450 reductase (CPR), Roman et al. (35,36) proposed that the 21-42-amino acid C-terminal extension tail present in all NOSs but absent in CPR was involved in modulating electron transfer. Their experimental data demonstrated that deletion of the entire C-tail from rat nNOS (33 residues) or bovine eNOS (42 residues) greatly increased electron transfer into and between flavins in the absence of CaM. Paradoxically, their cytochrome c reductase activities were suppressed rather than enhanced by exogenously added CaM, and their CaM-induced NO synthesis activities were only 50% that of CaM-bound wild-type enzymes (36). In contrast, Lane and Gross (37) did not delete the entire C-terminal tail but instead partially removed the Ser 1179 and the subsequent 26 C-terminal amino acids from bovine eNOS (⌬27). This mutant exhibited a 5-fold reduction in EC 50 for calcium and a 2-4-fold increase in maximal catalytic activities. Both reductase and oxygenase activities of ⌬27 were enhanced 3-fold by exogenously added CaM (37). These findings underscore a complex control process of the C-terminal tail, especially with respect to the influence of CaM on electron transfer.
Sequence alignment reveals that a segment at the proximal C-terminal tail is conserved in the sequences of human eNOS (residues 1165-1178) and nNOS (residues 1404 -1417) but absent in iNOS. Besides the autoinhibitory loop of residues 594 -606/614 -645, this segment indicates another, more obvious dissimilarity between cNOSs and iNOS. We postulated that this conserved region in conjunction with the autoinhibitory loop might confer the dramatic differences in Ca ϩ2 sensitivity and the rate of electron flux between eNOS and iNOS. To test this hypothesis, we assessed the changes in electron transfer and NO synthesis caused by deletion of eNOS sequence (residues 1165-1178, ⌬14eNOS). We were particularly interested in learning whether combined deletions of ⌬594 -606/⌬614 -645 and ⌬1165-1178 would yield an eNOS mutant protein resembling iNOS in terms of the CaM requirement for reductase and oxygenase activities. Wild-type and mutant eNOSs expressed in an Sf9/baculovirus system were purified by adding an adequate amount of chelators to remove calcium, and the reductase and NO synthesis activities were measured. All of the purified proteins were free of the endogenous CaM. The results showed that a combined ⌬45/⌬14 deletion mutant had a significant increase in the rates of cytochrome c reduction, NADPH oxidation, and NO synthesis in a completely Ca 2ϩ /CaM-independent manner.
The wild-type eNOS (WTeNOS) cDNA was used as the template for ⌬14 construct, and the ⌬45 cDNA was used as the template for ⌬45/⌬14 construct. All primers were synthesized by Genosys Inc. (Woodlands, TX). Sequences of mutant cDNAs at junctional regions were confirmed by DNA sequencing at the core facility of the University of Texas Medical School at Houston.
Expression and Determination of Nitrate/Nitrite in Sf9 Culture Medium-The cDNAs of WTeNOS and deletion mutants (⌬45, ⌬14, and ⌬45/⌬14) were inserted into the EcoRI site of pVL1392 transfer vector, which was used to generate recombinant viruses in an Sf9/baculovirus system. The nitrate/nitrite accumulation in culture medium was measured using a colorimetric assay kit from Cayman Chemical Co. Ten million Sf9 cells were seeded in each T 75 culture flask, which was individually infected with 2 multiplicities of infection of recombinant viruses of WTeNOS and each mutant. Because of naturally low heme biosynthetic capability in the Sf9 cells, heme chloride (4 g/ml) was added into the culture medium at 48 h postinfection to enrich heme content for the expressed protein. The amount of nitrite/nitrate in the culture medium was determined at 72 h postinfection by adding Griess reagent. The absorbance at 540 nm was recorded by using a Dynatech MR5000 microplate reader, and NO 2 Ϫ /NO 3 Ϫ was quantified using NaNO 3 as standard.
Purification of CaM-free Proteins of WTeNOS and Deletion Mutants-To generate the CaM-free mutant proteins, cells were harvested at 72 h postinfection, washed twice with calcium-free phosphate-buffered saline, pH 7.2, and resuspended in Buffer A (25 mM Tris-HCl, pH 7.5, 0.2 mM dithiothreitol, 1 mM EDTA, 1 mM EGTA, 1 M pepstatin A, 1 M leupeptin, 1 M antipain, 1 mM phenylmethylsulfonyl fluoride, and 10% glycerol). Cells were sonicated four times for 30 s and then centrifuged at 15,000 ϫ g for 60 min at 4°C. The supernatant was loaded onto a 2Ј,5Ј-ADP-Sepharose affinity column (1.5 ϫ 5 cm) pre-equilibrated with Buffer A. The column was washed with 20 column volumes of Buffer A, and then with 10 column volumes of Buffer B (25 mM Tris-HCl, pH 7.5, 0.2 mM dithiothreitol, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 M NaCl, and 10% glycerol). The protein was eluted with Buffer B containing 20 mM 2Ј-AMP and concentrated by Centriprep-30 (Amicon). The concentrated protein was applied onto a gel filtration chromatography (1 ϫ 120 cm, Ultrogel AcA34) and eluted with a buffer containing 25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM dithiothreitol, and 10% glycerol.
SDS-PAGE and Immunoblotting-Protein concentration was estimated using an extinction coefficient of Soret absorption peak ϭ 100 mM Ϫ1 cm Ϫ1 for the NOS proteins (15) and also was determined by the method of Bradford (42). SDS-PAGE was performed on a 7.5% slab gel according to the Laemmli procedure (43) and stained by Coomassie Blue R250. For CaM immunoblot, the purified protein (15 g) was subjected to SDS-PAGE in a 15% gel under reducing condition and then transferred to a polyvinylidene difluoride membrane. The membrane was blotted by monoclonal antibody raised against CaM (Sigma, catalog no. C-7055). Goat anti-mouse IgG-horseradish peroxidase conjugate was used as secondary antibody detected by the ECL method (Amersham Biosciences).
Ca 2ϩ -dependent Measurement-To measure NOS activity at different free Ca 2ϩ concentrations, a 100 mM stock of Ca 2ϩ -EGTA (Molecular Probes, Inc.) was used to obtain the desired free Ca 2ϩ solution as calculated according to manufacturer's procedure using the K d value of (Ca 2ϩ -EGTA) ϭ 107.9 nM at 37°C in 10 mM MOPS, pH 7. Steady-state Catalysis-NO synthesis, NADPH oxidation, cytochrome c reduction, and ferricyanide oxidation were determined by measuring the optical absorbance change using a Shimadzu-2501 PC equipped with a temperature controller. A 10-mm light path cuvette was used unless indicated otherwise. The rate of NO formation was quantified from the NO-mediated conversion of oxyhemoglobin to methemoglobin by monitoring the absorbance increase at 401 nm using an extinction coefficient of 38 mM Ϫ1 cm Ϫ1 (44). Assays were carried out at 37°C in the absence and presence of 0.5 M calmodulin and 300 M CaCl 2 with a mixture (500 l) containing 25 mM Tris-HCl, pH 7.5, 0.2 mM EDTA, 10 M oxyhemoglobin, 100 M ␤-NADPH, 100 M H 4 B, 1 mM L-arginine, 10% glycerol, and 50 -100 nM enzymes. NADPH oxidation was measured as the decrease in absorbance at 340 nm using an extinction coefficient of 6.22 mM Ϫ1 cm Ϫ1 . Assays were performed at 37°C in the absence or presence of 0.5 M calmodulin and 300 M CaCl 2 with a buffer containing 25 mM Tris-HCl, pH 7.5, 0.2 mM EDTA, 10% glycerol, 100 M NADPH, 1 mM L-arginine, and 50 -100 nM enzymes. Cytochrome c reductase activity was determined at 37°C in a reaction mixture (500 l) containing 25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10% glycerol, 100 M cytochrome c, 100 M NADPH, and 30 nM enzyme with or without 0.5 M calmodulin and 100 M CaCl 2 . The reduced cytochrome c was monitored with the absorbance increase at 550 nm and quantified using ⌬ of 21 mM Ϫ1 cm Ϫ1 . Ferricyanide reduction was carried out in a reaction mixture similar to that described for the cytochrome c assay except that 3.2 mM ferricyanide and a 5-mm-light path cuvette were used. The amount of reduced ferricyanide was quantified using ⌬ of 1.02 mM Ϫ1 cm Ϫ1 at 420 nm.

Nitrite/Nitrate Accumulation in Sf9 Culture Medium-
Studies have shown that cells expressing iNOS spontaneously produce a large quantity of NO ⅐ measured as NO 2 Ϫ /NO 3 Ϫ in their culture medium at a resting state of Ca 2ϩ , whereas cells expressing cNOSs produces only a trace amount of nitrate in the absence of agonists (19). To determine whether the deletion mutants behaved like iNOS, the constructs of wild-type and mutant eNOSs were expressed in Sf9 cells under identical conditions for 72 h. Western blot with anti-eNOS antibody was used to estimate the amount of protein expressed and showed that both ⌬45 and ⌬45/⌬14 were expressed at a level approximately half that of wild-type and ⌬14eNOS (data not shown). Even with this constraint, the total amount of NO 2 Ϫ /NO 3 Ϫ from cells expressing ⌬45 and ⌬45/⌬14 was higher than that from cells expressing ⌬14 (81, 37, and 62 M for ⌬45, ⌬14, and ⌬45/⌬14, respectively), whereas that from cells expressing wild-type eNOS was barely detectable (Fig. 2). The results suggest that all mutants with either a single deletion or a combined deletion of residues 594 -606/614 -645 and 1165-1178 are active at the basal level of intracellular Ca 2ϩ but that ⌬45 and ⌬45/⌬14 are more sensitive to Ca 2ϩ than ⌬14.
Purification of CaM-free Proteins-To obtain a CaM-free protein, the wild-type and mutant eNOSs were expressed in Sf9 cells and purified by the presence of an adequate amount of EGTA and EDTA. The purified enzymes were shown to be near homogeneity with an appropriate molecular mass in Coomassie-stained SDS-PAGE (Fig. 3A.). All mutants displayed the absorbance spectra identical to that of WTeNOS (45), indicating that the deletions did not perturb enzyme structure (data ⌬45 and ⌬14 were prepared by deleting residues 594 -606/614 -645 and residues 1165-1178, respectively, from human eNOS. A combined deletion mutant (⌬45/⌬14) was also constructed. not shown). To further determine whether the WTeNOS and mutant enzymes contained an endogenous CaM, the purified proteins were subjected to SDS-PAGE followed by immunoblot with anti-CaM monoclonal antibody. None of the proteins purified in the presence of EGTA and EDTA contained an intrinsic CaM (Fig. 3B, lanes 1-4), whereas a ⌬45 protein prepared in the absence of chelators and run parallel in SDS-PAGE had a detectable CaM (Fig. 3B, lane 5).
Requirement for Ca 2ϩ and CaM in L-Citrulline Formation-Dependence of L-citrulline formation on CaM was titrated by adding increasing concentrations of CaM (Sigma, catalog no. P-1431) along with 39 M free Ca 2ϩ and other cofactors in the reaction mixtures. CaM concentration response curves for wildtype and mutant eNOS are shown in Fig. 4A. The CaM-free ⌬45 exhibited a constitutive activity in the absence of CaM, which was increased by increasing CaM concentration, whereas WTe-NOS and ⌬14 had a very low basal activity and required about 85 and 58 nM CaM, respectively, to reach half-maximal activity (EC 50 ). Interestingly, ⌬45/⌬14 was active in the absence of CaM, and its activity was reduced by about 25% in the presence of saturating Ca 2ϩ /CaM. Dependence on [Ca 2ϩ ] was similarly titrated in the presence of 300 nM CaM, and the response curves are shown in Fig. 4B. The desired concentration of free Ca 2ϩ was obtained by adding varied ratios of K-EGTA and Ca 2ϩ -EGTA as described previously (24). In the absence of Ca 2ϩ , ⌬45eNOS had a considerable level of activity, which was increased by exogenous Ca 2ϩ . WTe-NOS was inactive without added Ca 2ϩ , and the EC 50 for Ca 2ϩ was estimated to be ϳ160 nM. The activity of the ⌬14 mutant was barely detectable in the absence of Ca 2ϩ , and the EC 50 value was 90 nM. In contrast, the ⌬45/⌬14 mutant was fully active in the absence of calcium, and the activity was reduced by 20% at saturating CaM/Ca 2ϩ . The Ca 2ϩ and CaM response curves for all of the mutants coincided (Fig. 4, A versus B).
These results indicated that ⌬45/⌬14 became constitutively active and did not require Ca 2ϩ /CaM for L-citrulline formation.
Steady-state Enzymatic Activities of WTeNOS Versus Deletion Mutants-The electron flow in NOS is from NADPH to FAD, to FMN, and finally to the oxygenase heme. The steadystate catalysis of each subdomain, analyzed in the absence or presence of CaM, was expressed in nanomoles of product formation/nanomole of heme/min (min Ϫ1 ).
Ferricyanide Reduction-The NOS is able to reduce artificial electron acceptors such as cytochrome c and ferricyanide. Ferricyanide accepts electrons either from FAD or FMN, whereas cytochrome c accepts electrons exclusively from FMN (46). Ferricyanide reduction thus provides a useful tool for evaluating the effect of site-directed deletion on the FAD subdomain activity. As shown in Fig. 5A, the basal rate of ferricyanide reduction for all mutants was 1.6-fold higher than that of WTeNOS (2900 versus ϳ4700 min Ϫ1 for WTeNOS versus all mutants, respectively). In contrast, the CaM-stimulated maximal rates (6300 -6700 min Ϫ1 ) were similar for wild type and all mutants, indicating that the effect of deletion on ferricyanide reduction is more pronounced for the CaM-free than the CaMbound state. Cytochrome c Reduction-As electron transfer to cytochrome c occurs exclusively from FMN, cytochrome c assays were performed to determine the effects of mutating these elements on FMN subdomain (Fig. 5B). Unlike ferricyanide reduction, all mutants displayed the maximal rates of cytochrome c reduction severalfold greater than WTeNOS either in the CaM-free or CaM-bound enzyme. The addition of CaM increased cytochrome c reduction by 4.5-fold for both WTeNOS (90 versus 434 min Ϫ1 in the absence and presence of CaM, respectively) and ⌬14 (402 versus 1573 min Ϫ1 in the absence and presence of CaM, respectively), whereas CaM slightly enhanced the rate of ⌬45 (731 versus 787 min Ϫ1 in the absence and presence of CaM, respectively) and decreased the rate of ⌬45/⌬14 by 25% (1389 versus 1059 min Ϫ1 in the absence and presence of CaM, respectively). The results suggest that these two peptide regions play a role in controlling electron flow from the FMN moiety to the heme of cytochrome c.
NADPH Oxidation-NADPH is oxidized and consumed after electron flux from reductase to oxygenase domain where heme is reduced with O 2 as an electron acceptor (47). We measured the rate of NADPH oxidation in wild-type and mutant eNOSs (Fig. 6A). For mutant ⌬45, the rate of NADPH oxidation was increased 1.6-fold in the presence of CaM (33 versus 54 min Ϫ1 ). The basal rate of ⌬14 was slightly higher than that of WTeNOS (9.8 versus 4.2 min Ϫ1 ) but was markedly increased by the presence of CaM, at a rate far exceeding that of wild type (108 versus 32 min Ϫ1 ). In the absence of CaM, the rate of ⌬45/⌬14 was 18-fold higher than that of wild type (75 versus 4.2 min Ϫ1 ), but its rate was reduced by 36% when CaM was present (48 min Ϫ1 ). Changes in NADPH oxidation by deletion of these two regions parallel those in cytochrome c reduction (Fig. 6A versus Fig. 5B), suggesting that the control mechanism of these two peptide regions for electron flow from FMN to oxygenase domain is similar to that from FMN to an artificial electron acceptor, cytochrome c.
NO Synthesis-We next determined whether a higher rate of electron flow in the mutants correlated with a higher efficiency of NO production. We compared the rate of NO production between each mutant and WTeNOS. As shown in Fig. 6B, WTeNOS and ⌬14 produced a very low level of NO barely beyond the detection limit in the absence of CaM. With CaM, the maximal NO synthesis of ⌬14 was 2.5-fold higher than that of wild type (61 versus 27 min Ϫ1 ). In contrast, ⌬45 and ⌬14/⌬45 produced a significant amount of NO (17 versus 36 min Ϫ1 for ⌬45 and ⌬45/⌬14, respectively) in the absence of CaM. Upon the addition of CaM, there was a 1.8-fold increase in NO synthesis by ⌬45 (32 min Ϫ1 ) and a 40% reduction by the double mutant ⌬45/⌬14 (23 min Ϫ1 ). The rate of NO production in all three mutants is generally correlated with the rates of electron transfer from FMN to cytochrome c and oxygenase heme.

DISCUSSION
The molecular basis for the dramatic difference of calcium sensitivity in enzyme activation between the cNOSs and iNOS has not been fully elucidated. Three segments including the CaM binding sequence, the autoinhibitory loop in the FMN subdomain, and a C-terminal extension tail were reported to play a pivotal role in regulating Ca 2ϩ /CaM-mediated electron flux. Modifications at those sites affect enzyme activity by reducing Ca 2ϩ dependence relative to wild-type cNOSs. However, the CaM-binding domain, the autoinhibitory element, or the C-terminal tail alone does not confer Ca 2ϩ -independent activity similar to that of iNOS. Here we have focused on the aforementioned two regions (residues 594 -606/614 -645 and residues 1165-1178 in human eNOS) that exist in the reductase domain of cNOSs but have no counterpart in iNOS. We made the constructs by deleting either or both fragments from eNOS to mimic iNOS. Our prior work with deletion of residues 594 -606/614 -645 yields a mutant enzyme (⌬45) containing a trapped CaM isolated from an Sf9/baculovirus system, and because of this it is not clear whether the 45-residue segment actually serves a regulatory role in CaM dependence. We thus prepared CaM-free enzymes by adding an adequate amount of chelators during the purification process. The availability of CaM-free wild-type and mutant eNOSs allowed us to evaluate the effect of ⌬45 and ⌬14 mutations on CaM-dependent electron transfer and catalytic properties.
CaM-free ⌬45 had a lower heme content and lower catalytic activities than CaM-bound ⌬45 (24), indicating that the trapped CaM stabilized the mutant protein. Steady-state analysis demonstrated that ⌬45 in the absence of CaM catalyzed ferricyanide reduction, NADPH oxidation, and NO synthesis at rates higher than wild type. However, ⌬45 was regulated by Ca 2ϩ /CaM. Ca 2ϩ /CaM increased the rates by 30 -40% over those in the absence of Ca 2ϩ /CaM. The rate of cytochrome c reduction of ⌬45 was higher than wild type, but Ca 2ϩ /CaM had little influence on the rate of cytochrome c reduction by ⌬45. This is consistent with our previously published study (24) as well as that of Nishida and Ortiz de Montellano (21). Taken together, the results indicate that ⌬45 deletion significantly increased Ca 2ϩ -sensitivity with a parallel reduction in CaM requirement for electron transfer and catalytic activity. Thus, residues 594 -606/614 -645 control Ca 2ϩ /CaM binding and repress electron transfer.
The ⌬14eNOS with deletion of residues 1165-1178 did not show Ca 2ϩ /CaM-independent activity but required a lower Ca 2ϩ /CaM concentration in achieving maximal activity than WTeNOS. This is also supported by the spontaneous accumulation of a considerable amount of NO 2 Ϫ /NOO 3 Ϫ in the culture medium of Sf9 cells expressing ⌬14, indicating that the ⌬14 segment partially confers Ca 2ϩ requirement in eNOS catalysis. Truncation of ⌬14 increased CaM-induced NO synthesis and both basal and CaM-bound cytochrome c reduction 2-4-fold over wild type. CaM-stimulated NO synthesis and NADPH oxidation by ⌬14eNOS approach those of CaM-bound nNOS (21); thus the ⌬14-segment can be inferred to negatively control electron transfer through enzymes (36) that contributes to the relatively low catalytic activities of eNOS. Analysis of an x-ray crystal structure of the nNOS FAD domain lacking the Cterminal 22 residues (48) has revealed that the adjacent Cterminal 10 residues are disordered in the crystal structure, suggesting that the C-terminal tail probably lies between the flavins and/or between the FAD and NADPH, modulating the interflavin distance (49). As both orientation and distance between each redox partner influence the electron transfer rate, modification at the C-tail may either change the relative orientation or the distance between the two redox partners, resulting in a greater increase in electron flux. The phenotypic changes of ⌬14 are essentially identical to those observed with a bovine eNOS mutant (C⌬27) that is missing Ser 1179 and the subsequent 26 amino acids of the C terminus (37). In contrast, removal of the entire C-tail 42 residues from bovine eNOS (C⌬42) decreased the rate of maximal CaM-stimulated NO synthesis by ϳ50% compared with the wild type, despite a great increase in the rates of cytochrome c and ferricyanide reduction in the absence of CaM (36). The catalytic profile of C⌬42 was very similar to that of F1395S mutant in rat nNOS reported by Adak et al. (50), who showed that the aromatic side chain of Phe 1395 was involved in repressing electron transfer into and out of flavins in the CaM-free state and was required for CaM to relieve this repression. As this conserved Phe residue lies immediately upstream of C terminus, truncation of ⌬42 might have removed too large a portion of eNOS at this region, altering the conformation of this key Phe residue and therefore decreasing electron transfer efficiency between the FMN and the heme, resulting in a large reduction of NO production.
The above described studies demonstrated that the ⌬45 exhibited a considerable amount of reductase and oxygenase activities in the absence of Ca 2ϩ /CaM, whereas ⌬14 was largely enhanced by CaM in those reactions. However, the rates of cytochrome c reduction, NADPH oxidation and NO synthesis catalyzed by ⌬14 were 2-4-fold higher than those catalyzed by ⌬45 and wild type in the presence of CaM. The result supports the model of Roman et al. (36) that the ⌬45-segment plays a pronounced role in regulating Ca 2ϩ /CaM-dependence for interdomain activation, whereas ⌬14 is involved in limiting the rate of electron transfer into and out of FMN module. We extended our finding by combined deletion of both fragments, resulting in a mutant enzyme (⌬45/⌬14) with the basal rates of cytochrome c reduction, NADPH oxidation, and NO synthesis 2-4-fold over CaM-induced wild type. Intriguingly, CaM binding to this double mutant was observed to inhibit rather than enhance those reactions, indicating a completely Ca 2ϩ /CaM-independent electron transfer between FMN and the hemes of cytochrome c and eNOS oxygenase domain. This observation appears to contradict the viewpoint that CaM binding to its canonical binding site is essential for electron transfer from the reductase to oxygenase domain. Our results are consistent with the recent catalytic model (36). In the absence of CaM, the ⌬45 autoinhibitory loop and C-terminal tail function as barriers sitting near each side of flavin wall and/or between the FAD and NADPH, that almost completely block the electron transfer; therefore there is almost no detectable NO synthesis activity in CaM-free WTeNOS. When two segments are deleted, the barriers are removed such that the rate of electron transfer is greatly enhanced, even exceeding the rate of CaM-induced WTeNOS. The binding domains of these two elements on eNOS might either overlap or allosterically interact with CaM binding site(s). Once CaM binds to wild type, the CaM-driven conformational change partially lifts the inhibitory inserts to allow limited electron transfer within the reductase domain and from the reductase to the oxygenase domain. In fact, the wild-type and the double mutant (⌬45/⌬14) enzymes in the CaM-bound state are very similar in ferricyanide reduction, NADPH oxidation, and NO synthesis, indicating that the addition of CaM to the double mutant appears to retighten a loosened electron transfer, and those activities are reduced back to normal wildtype levels. Therefore, the primary mechanism of CaM to gate electron flux is to tether the redox partners of the interdomain or intradomain close together for electron transfer despite the presence or absence of these two inhibitory elements. Threedimensional structure information on the reductase domains of cNOSs is essential to define how the interaction among CaM, the autoinhibitory loop, and the C-terminal tail regulates eNOS catalysis.