Calmodulin-dependent Regulation of Inducible and Neuronal Nitric-oxide Synthase*

Neuronal and endothelial nitric-oxide synthases depend upon Ca2+/calmodulin for activation, whereas the activity of the inducible nitric-oxide synthase is Ca2+-independent, presumably due to tightly bound calmodulin. To study these different mechanisms, a series of chimeras derived from neuronal and inducible nitric- oxide synthases were analyzed. Chimeras containing only the oxygenase domain, calmodulin-binding region, or reductase domain of inducible nitric-oxide synthase did not confer significant Ca2+-independent activity. However, each chimera was more sensitive to Ca2+ than the neuronal isoform. The calmodulin-binding region of inducible nitric-oxide synthase with either its oxygenase or reductase domains resulted in significant, but not total, Ca2+-independent activity. Co-immunoprecipitation experiments showed no calmodulin associated with the former chimera in the absence of Ca2+. Trifluoperazine also inhibited this chimera in the absence of Ca2+. The combined interactions of calmodulin bound to inducible nitric-oxide synthase calmodulin-binding region with the oxygenase domain may be weaker than with the reductase domain. Thus, Ca2+-independent activity of inducible nitric-oxide synthase appears to result from the concerted interactions of calmodulin with both the oxygenase and reductase domains in addition to the canonical calmodulin-binding region. The neuronal isoform is not regulated by a unique autoinhibitory element in its reductase domain.

NO with an unpaired electron reacts with protein targets primarily through their thio or heme groups and acts as a messenger or modulator molecule in many biological systems (1)(2)(3)(4). It is produced from L-arginine with L-citrulline as a co-product in a reaction catalyzed by NOS 1 that requires NADPH, FAD, FMN, BH4, calmodulin, and heme (1)(2)(3)(4).
Three NOS isoforms were originally identified based on the tissue source: nNOS, eNOS, and iNOS (5). All NOSs contain three domains: an oxygenase domain at the N-terminal half and a reductase domain at the C-terminal half connected by a calmodulin-binding region in the middle of the molecule. All isoforms are highly related with at least 50 -60% identity and are classified into two categories based on their dependence on Ca 2ϩ for activity. When Ca 2ϩ binds to calmodulin, the complex binds to nNOS or eNOS and stimulates nitric oxide production. In contrast, when iNOS expression is induced upon stimulation of cells with cytokines or endotoxins, it is fully active, even when Ca 2ϩ levels in cells are low. The Ca 2ϩ -independent activity of iNOS is associated with calmodulin tightly bound to the enzyme (6).
Calmodulin binds proteins through IQ (IQXXXRGXXXR) motifs in a Ca 2ϩ -independent manner or through a canonical calmodulin-binding region in a Ca 2ϩ -dependent way (7). However, iNOS does not have a typical IQ motif. Moreover, all of the putative calmodulin-binding regions from nNOS, eNOS, and iNOS satisfy the criteria for properties of a canonical calmodulin-binding region, a basic amphipathic ␣-helical sequence containing 12 basic and hydrophobic residues (8). Calmodulin activates nNOS at two points in the electron transfer sequence: electron transfer into the flavins and interdomain electron transfer between the flavins and the hemes (9). The association of calmodulin with nNOS may tether the oxygenase and the reductase domain or some subdomains close together for electron transfer by which calmodulin acts as a molecular switch (10).
The calmodulin-binding sequence in iNOS is necessary but not sufficient for Ca 2ϩ -independent activity (11,12). Replacement of the calmodulin-binding sequence of eNOS or nNOS with the corresponding sequence from iNOS resulted in a chimera that was still Ca 2ϩ -dependent. It was proposed that, in addition to the canonical calmodulin-binding region, sites in the reductase domain confer Ca 2ϩ -independent binding of calmodulin, which is presumably responsible for the iNOS Ca 2ϩindependent activity. Alternatively, Salerno et al. (13) proposed an autoinhibitory segment in the FMN-binding domains of nNOS and eNOS, but not iNOS. Hence, calmodulin binding may displace this unique autoinhibitory segment, resulting in catalysis. These two models for Ca 2ϩ /calmodulin regulation of NOS activities were investigated by the characterization of chimeric enzymes made from nNOS and iNOS.

Construction of Mutant and Chimeric
NOSs-nNOS cDNA (a gift from Solomon H. Snyder) was cut with EcoRI and subcloned into the pCMV5 EcoRI site. Mouse iNOS cDNA (a gift from Richard C. Lyons) was cut by HindIII and BglII and subcloned into pCMV5. A XhoI site was introduced into nNOS cDNA at amino acids 796 -797 by oligonucleotide-directed mutagenesis without changing any amino acid. A XhoI site was also introduced into iNOS cDNA at amino acids 574 -575 without changing any amino acid by the two-step polymerase chain reaction method (14) through the ApaI-ApaI 900-base pair fragment of iNOS cDNA. nNOS-I 576 -1144 and nNOS-I 1-575 cDNAs were generated by exchanging the XhoI-XhoI fragments of XhoI-nNOS and XhoI-iNOS cDNAs. A HpaI site was introduced into XhoI-nNOS cDNA at amino acids 696 -697 or XhoI-iNOS cDNA at amino acids 475-476 to generate nNOS L696V (Leu changed to Val at residue 696) and iNOS L475V (Leu changed to Val at residue 475) for making other chimeric NOSs by the two-step polymerase chain reaction method. The specific activity and dependence on different [Ca 2ϩ ] for activity of nNOS L696V and iNOS L475V were the same as their respective wild-type NOSs (see below). The chimeric NOS constructs were made by two-step polymerase chain reaction using nNOS L696V , iNOS L475V , and appropriate NOS chimeras as templates and subcloning into proper vector fragments. Polymerase chain reaction products were sequenced to ensure no additional mutations occurred.
Cell Culture and Protein Expression-COS-7 cells (Life Technologies, Inc.) at 80% confluence in 100-mm tissue culture dishes were transfected with 5 g of cDNA by DEAE-dextran and chloroquine as described previously (15). The cells were shocked with 10% dimethyl sulfoxide in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum after removing chloroquine for a 3-min incubation at room temperature followed by a final wash step. The subsequent transfectants were cultured with Dulbecco's modified Eagle's medium (Cellgro Co.) plus 10% fetal bovine serum, streptomycin, penicillin, and fungizone (Life Technologies, Inc.) in the presence or absence of 1 mM L-NAME (Sigma), 5 M hemin (Sigma), and 100 M BH4 (Research Biochemical, Co.) to obtain active enzymes.
NOS Enzyme Activity Assays-To measure NOS activity at different Ca 2ϩ concentrations, EGTA/Ca 2ϩ buffers were prepared from buffers containing 2 mM EGTA, 100 mM Hepes, pH 7.4, with or without 300 M free Ca 2ϩ (method 1). The final free Ca 2ϩ concentrations were determined by spectrophotometry with the fluorescent dyes fluo-3 or Ca 2ϩ -Green 5N (Molecular Probes, Inc., Eugene, OR). Calibrated Ca 2ϩ concentrations (Molecular Probes) were used to confirm the free Ca 2ϩ concentrations measured.
Cells in one 100-mm tissue culture dish were harvested at 40 -48 h post-transfection in 100 l of lysis buffer (100 mM Hepes, pH 7.4, 80 M BH4, 5 mM dithiothreitol, and protease inhibitors: 0.54 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, and 1 g/ml pepstatin A) after trypsinization and washing cells twice with phosphate-buffered saline. For those transfected cells cultured in the presence of L-NAME, BH4, and hemin, the cells were incubated with regular media for 1 h before trypsinization. Sonication (30% output, 10 s) was used to lyse cells at 4°C. Lysates Activities were also measured in cell lysates in the presence of 100 M Ca 2ϩ and 800,000 cpm [ 3 H]arginine (60 Ci mmol Ϫ1 ) in 100 l of the reaction mixture described above (method 2). The lysates were pretreated with Dowex-50 resin equilibrated with lysis buffer to remove endogenous arginine. However, this pretreatment had negligible effects on NOS activities. The maximal activities of chimeric enzymes were calculated by Lineweaver-Burk analysis, assuming an average K m value of 6 M for L-arginine from published data (16 -25). The amount of protein expressed was estimated with appropriate standards from purified nNOS and iNOS by Western blot analysis using antibodies against peptides derived from amino acids 1422-1433 of human nNOS (differ by only one amino acid from rat nNOS) and from amino acids For catalytic activity, NOSs must be dimeric (21,26,27). Heme, BH4, and L-NAME were supplemented to cell growth media for expressing some of the NOS chimeras when needed. Transfectants were harvested in the lysis buffer containing dithiothreitol and BH4 for maintaining proper structure. More than 95% of iNOS expressed under these conditions was dimeric (data not shown) as analyzed by fast protein liquid chromatography (28,29).
Immunoprecipitation Assays-Lysates were prepared as above from COS-7 cells transfected with wild type or chimeric NOSs with the addition of 2 mM L-arginine and 0.5% Triton X-100 and then precleared with staphylococcal protein A-agarose beads (Pierce). The lysate was incubated with the indicated antibody, EGTA, CaCl 2 , or TFP (Sigma) for 2 h and then with staphylococcal protein A-agarose beads overnight at 4°C. The beads were washed sequentially in buffer A (lysis buffer containing 2 mM arginine, 0.5% Triton X-100, and 150 mM NaCl), buffer B (lysis buffer containing 2 mM arginine, and 150 mM NaCl), and buffer C (lysis buffer containing 2 mM arginine). For immunoblots, the cell lysates or immunoprecipitates were subjected to 6% SDS-PAGE and then transferred to nitrocellulose membrane (PROTRAN TM ; Schleicher & Schuell) at 100 V for 75 min in chilled transfer buffer. The expression of NOSs was detected by primary antibodies raised against the C terminus of human nNOS amino acids 1422-1433 or against the C terminus of mouse iNOS amino acids 1130 -1141 after incubation at 37°C for 2 h or 4°C overnight. The goat anti-rabbit IgG-horseradish peroxidase conjugate (Life Technologies, Inc.) was used as secondary antibody with an incubation at room temperature for 40 min and detected by the ECL method (Amersham Pharmacia Biotech). For the calmodulin immunoblot, NOS immuno-precipitates were subjected to 15% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (Millipore Corp.). The monoclonal antibody raised against calmodulin (catalog no. 05-173, Bio-Rad) was used as recommended. The goat anti-mouse IgG-horseradish peroxidase conjugate (Life Technologies, Inc.) was used as secondary antibody and detected by the ECL method (Amersham Pharmacia Biotech).

RESULTS
NOS Activities-To investigate mechanisms responsible for differences in calmodulin binding and Ca 2ϩ -dependent activity, a series of chimeras and point mutants derived from nNOS and iNOS were constructed and transiently expressed in COS-7 cells (Fig. 1). As described under "Materials and Methods," all expressed chimeric enzymes were catalytically active with activities ranging from 100 to 30% relative to iNOS. nNOS L696V and iNOS L475V containing a conserved point mutation designed to make some of the chimeric NOSs were also catalytically active, and their dependence on different [Ca 2ϩ ] were the same as their respective wild-type enzymes (Fig. 2). Thus, the introduced mutations did not affect the respective enzymes' sensitivity to different [Ca 2ϩ ] or maximal activity.
NOS Chimeras with Individual iNOS Oxygenase Domain, Calmodulin-binding Region, or Reductase Domain-Although the canonical calmodulin-binding sequences of nNOS (amino acids 725-755) and iNOS (amino acids 503-533) are not identical, they display features that are considered characteristic of Ca 2ϩ -dependent, calmodulin-binding regions (8,30,31). nNOS-I 503-533 with the calmodulin-binding region of iNOS and the respective reductase and oxygenase domains of nNOS did not show Ca 2ϩ -independent activity like iNOS (Fig. 3). However, this chimera was more sensitive to activation at lower [Ca 2ϩ ] than those required to activate nNOS. Based on results with nNOS-I 503-533 , the oxygenase and the reductase domains of iNOS alone were substituted into the respective regions of nNOS (nNOS-I 1-475 and nNOS-I 576 -1144 ). Results showed that neither one of these portions alone conferred significant Ca 2ϩindependent activity (Fig. 3). nNOS-I 1-475 , containing the iNOS oxygenase domain, was completely dependent on Ca 2ϩ for activity. Surprisingly, it was more sensitive to [Ca 2ϩ ] than nNOS. Likewise, nNOS-I 576 -1144 containing the iNOS reductase domain also increased the sensitivity of enzyme activity to [Ca 2ϩ ] (Fig. 3). A chimera containing only the C-terminal half of the iNOS reductase domain, nNOS-I 749 -1144 , also showed greater sensitivity to [Ca 2ϩ ] than nNOS (Fig. 3). These results indicate that neither the iNOS oxygenase domain, the iNOS calmodulin-binding region, nor the iNOS reductase domain alone confer significant Ca 2ϩ -independent activity similar to that of iNOS. However, all of these substitutions affect enzyme activity by increasing sensitivity to [Ca 2ϩ ] relative to nNOS.

NOS Chimeras with iNOS Calmodulin-binding Region and either iNOS Oxygenase or iNOS Reductase Domains-Chimeric
NOSs with combinations of iNOS oxygenase or iNOS reductase domains with the iNOS calmodulin-binding sequence were constructed to investigate the influences of these domains on Ca 2ϩ regulation. Results in Fig. 4 show that the iNOS calmodulinbinding region with either the iNOS oxygenase domain or the iNOS reductase domain (nNOS-I 1-533 and nNOS-I 504 -1144 ) sig-nificantly increased Ca 2ϩ -independent NOS activity compared with the chimeras containing only one of the domains (Fig. 3). Furthermore, there was also a similar increase in sensitivity to activation at different [Ca 2ϩ ] with both nNOS-I 504 -1144 and nNOS-I 1-533 (Fig. 4). Thus, the concerted actions of iNOS calmodulin-binding region with either the iNOS oxygenase or the reductase domains were similar in regard to Ca 2ϩ -dependent activities.
The Flanking Regions of iNOS Canonical Calmodulin-binding Region-In the case of the phosphorylase kinase ␥-subunit, calmodulin binds to two noncontiguous regions spanning 70 residues (one forms an ␣-helix with a canonical calmodulinbinding sequence, and the other forms a ␤-hairpin structure) located at the C-terminal end of the ␥-subunit, resulting in tight association of calmodulin as an integral ␦-subunit (32, FIG. 1. Domain scheme for nNOS, iNOS, and chimeric NOSs. NOSs contain three similar functional domains: the oxygenase, calmodulinbinding, and reductase domains. The calmodulin-binding segment spans about 100 residues in length and can be subdivided into the N-terminal flanking, canonical calmodulin-binding, and C-terminal flanking regions. The canonical calmodulin-binding sequences of nNOS and iNOS satisfy the criteria for a conventional Ca 2ϩ -dependent binding region (8,30). NADPH, FAD, and FMN consensus binding regions are located in the reductase domain. Heme, Arg, O 2 , and BH 4 bind to the oxygenase domain. Additionally, nNOS contains a PDZ domain located at its N-terminal end that is involved in cellular localization (59,60). Antibodies raised to peptides derived from the C-terminal amino acids 1422-1433 of human nNOS (red) and from 1130 -1141 of mouse iNOS (dark blue) were used in this study. Chimeric and mutant NOSs were constructed as described under "Materials and Methods." Indicated in the figure are nNOS, iNOS, nNOS L696V , iNOS L475V , iNOS-N 725-755 (iNOS residues 503-533 replaced with nNOS residues 752-755), and other chimeras (designated as nNOS with the indicated iNOS residues (I) replaced in the corresponding region of nNOS). Mauve, nNOS origin; blue, iNOS origin. 33). Secondary structure predictions on the iNOS and nNOS calmodulin-binding region and flanking regions (amino acids 475-575 and 696 -796, respectively) suggest a structure similar to those in phosphorylase kinase ␥-subunit for iNOS but not nNOS (data not shown).
To investigate if the flanking regions of the iNOS canonical calmodulin-binding sequence contribute to iNOS Ca 2ϩ -independent activity, nNOS-I 475-533 was expressed. Unfortunately, it was catalytically inactive (data not shown). Alternatively, a series of chimeric NOSs with swapping points between the canonical calmodulin-binding region and its flanking regions were expressed in COS-7 cells (Fig. 5). Replacement of the reductase domain of iNOS alone with that of nNOS (iNOS to nNOS-I 1-575 ) changed the enzymatic response to different [Ca 2ϩ ]. nNOS-I 1-575 had 50% activity in the presence of EGTA compared with that obtained in the presence of Ca 2ϩ. (Fig. 5). At the lowest measured free [Ca 2ϩ ] (5 nM), nNOS-I 1-575 was maximally activated, similar to iNOS. Further replacement of the C-terminal flanking region of the iNOS canonical calmodulin-binding region from nNOS-I 1-575 with that of nNOS (nNOS-I 1-575 to nNOS-I 1-533 ) had no additional effect. When the iNOS canonical calmodulin-binding region from nNOS-I  was replaced with that of nNOS (nNOS-I 1-533 to nNOS-I 1-503 ), the enzyme activity became completely Ca 2ϩ -dependent and was similar to that of nNOS. Replacement of the N-terminal flanking region of the canonical calmodulin-binding region of iNOS from nNOS-I 1-503 with that of nNOS (nNOS-I 1-503 to nNOS-I 1-475 ) did not significantly change the Ca 2ϩ sensitivity of the NOS activity. Thus, the immediate N-and C-terminal flanking regions of the iNOS canonical calmodulin-binding sequence appear not to be involved in Ca 2ϩ regulation of enzyme activity.
Concerted Domain Actions for Ca 2ϩ -independent Activity in iNOS-To confirm that the concerted actions of iNOS calmodulin-binding region with the iNOS oxygenase or reductase domains enhance Ca 2ϩ -independent activity, the Ca 2ϩ -dependent activities of the chimeric NOSs were assayed with different concentrations of components in the reaction buffers (method 2; see "Materials and Methods"). Ca 2ϩ -independent activity measured with NOS chimeras containing the iNOS calmodulin-binding region alone (nNOS-I 503-533 ), the oxygenase domain alone (nNOS-I 1-475 or nNOS-I 1-503 ) or iNOS reductase domain (nNOS-I 576 -1144 or iNOS-N 725-755 ) were 0% (Fig. 6). The chimeric NOSs containing the iNOS calmodulin-binding region with either the iNOS oxygenase (nNOS-I 1-533 , nNOS-I 1-575 ,) or iNOS reductase domains (nNOS-I 504 -1144 ) showed 63-73% Ca 2ϩ -independent activities (Fig. 6). Thus, these results confirm the importance of the iNOS oxygenase or reductase domains for conferring Ca 2ϩ -independent activity in concert with the iNOS calmodulin-binding region.
nNOS-I 576 -1144 , iNOS-N 725-755 , and nNOS-I 503-533 showed a small amount Ca 2ϩ -independent activity (10 -20%) in the assay with varying [Ca 2ϩ ] (method 1; Figs. 3 and 4), but no activity was detected in the other Ca 2ϩ -dependent assay in the absence of Ca 2ϩ (method 2; Fig. 6). The 10 -20% enzyme activity was not reduced when the EGTA concentration was increased to 10 mM with method 1 (data not shown). These two NOS activity assay systems differ in the concentrations of calmodulin, BH4, [ 3 H]arginine, dithiothreitol, and hemin as well as the reaction temperature (25 versus 37°C; see "Materials and Methods"). Each factor was examined individually or in combination with two or three factors, and no differences in Ca 2ϩ -independent activity were observed (data not shown). The reasons for the small activity differences in the two assay systems were not identified.
Role of Calmodulin in Ca 2ϩ -independent Activity-Inhibition of chimeric NOS activities by TFP, a calmodulin antagonist, and co-immunoprecipitation with calmodulin were used to investigate mechanisms responsible for Ca 2ϩ -independent activity of NOS enzymes. Results showed that nNOS activity was completely inhibited by 200 M TFP in the presence of Ca 2ϩ , while iNOS activity was resistant to TFP treatment up to 1 mM either in the presence or absence of Ca 2ϩ (Fig. 7). The activity of nNOS-I 504 -1144 containing the iNOS calmodulin-binding region and reductase domain was also resistant to TFP treatment in either the presence or the absence of Ca 2ϩ . However, 1 mM TFP abolished the activity of nNOS-I 1-533 , containing the iNOS calmodulin-binding region and oxygenase domain, in the absence but not in the presence of Ca 2ϩ (Fig. 7). The nNOS-I 1-575 chimera also had similar properties as nNOS-I 1-533 (data not shown). Furthermore, L-NAME (an inhibitor of NOS activity) abolished activities of all of the enzymes in 1 mM TFP in the presence or absence of Ca 2ϩ , indicating that these were NOS activities. Thus, TFP inhibited the Ca 2ϩ -independent activity derived from the concerted action of the iNOS calmodulinbinding region with the iNOS oxygenase domain but not with the iNOS reductase domain.
Calmodulin binding properties to nNOS-I 1-533 and nNOS-I 504 -1144 were investigated. Lysates of transfected COS-7 cells were immunoprecipitated with anti-nNOS or anti-iNOS antibodies in the presence of 100 M CaCl 2 or 2.5 mM EGTA, in the presence or absence of 1 mM TFP. Co-immunoprecipitation results showed that calmodulin dissociates in the absence of Ca 2ϩ from nNOS and nNOS-I 1-533 , containing the iNOS calmodulin-binding region and oxygenase domain (Fig. 8). The dissociation of calmodulin from nNOS-I 1-533 in the presence of EGTA was surprising, since this chimera had significant enzymatic activity under these conditions (Figs. 4 and 6). In the presence of Ca 2ϩ , TFP did not prevent calmodulin binding to nNOS or nNOS-I 1-533 (Fig. 8), although it inhibited nNOS activity but not nNOS-I 1-533 activity. In contrast, TFP had no effect in the presence or absence of Ca 2ϩ on the association of calmodulin with iNOS and nNOS-I 504 -1144 . Thus, the Ca 2ϩindependent activity derived from iNOS calmodulin-binding region and its reductase domain (nNOS-I 504 -1144 ) was associated with calmodulin binding. In contrast, the Ca 2ϩ -independent activity derived from iNOS calmodulin-binding region and iNOS oxygenase domain (nNOS-I 1-533 ) was not associated with high affinity calmodulin binding. DISCUSSION Synthetic peptides that bind calmodulin provide important structural information about molecular determinants for specific interactions (31, 34 -37). However, multiple regions of a protein may interact with calmodulin for effective regulation, and calmodulin binding per se is necessary but not sufficient for activation of many enzymes (38,39). Moreover, unique calmodulin-binding sequences with atypical structural properties may be overlooked (40,41). Previous results obtained on chimeras with the respective calmodulin-binding sequences from iNOS substituted in eNOS or nNOS showed the sequence was necessary but not sufficient for Ca 2ϩ -independent activity or calmodulin binding (11,12). Similar results are presented herein with chimera nNOS-I 503-533 activity, which was Ca 2ϩdependent. However, the sensitivity of nNOS-I 503-533 activity to [Ca 2ϩ ] was greater than that obtained with nNOS. When the calmodulin-binding sequence of nNOS was substituted into iNOS, the chimeric iNOS-N 725-755 activity was still Ca 2ϩ -dependent, but its sensitivity to activation by [Ca 2ϩ ] was also greater than nNOS. These results support the concept that there are sites in addition to the canonical calmodulin-binding sequence that contribute to Ca 2ϩ regulation of NOS activity.
As expected, substitution of the iNOS oxygenase or reductase domains alone into nNOS did not result in Ca 2ϩ -independent NOS activity, consistent with the view that the iNOS calmodulin-binding region is necessary, but not sufficient, for Ca 2ϩindependent NOS activity (11,12). However, substitution of either iNOS oxygenase or reductase domains increases the Ca 2ϩ sensitivity of the respective chimeric NOS activities with a greater effect observed with the reductase domain (nNOS-I 576 -1144 ). Even the nNOS-I 749 -1144 chimera, containing only the C-terminal half of the iNOS reductase domain, was more sensitive to [Ca 2ϩ ] for activity than nNOS. The functional interaction between iNOS calmodulin-binding sequence and its reductase domain in terms of Ca 2ϩ regulation was also recently reported (42). However, the functional interaction with the oxygenase domain is unique. These results support the concept that structural elements in addition to the specific calmodulinbinding sequences contribute to Ca 2ϩ regulation of NOS activities.
It was recently proposed that an autoinhibitory sequence exists in the reductase domain of nNOS and eNOS (13). These Ca 2ϩ -dependent NOSs contain unique sequences in their respective FMN-binding domains not shared by iNOS (residues 820 -880 in nNOS). Synthetic peptides derived from this insert in eNOS inhibited calmodulin binding and Ca 2ϩ activation by binding to either eNOS or nNOS. However, the insert from nNOS lacks a RRKRK motif thought to be important for autoinhibition in eNOS (13). In contrast to results obtained with the synthetic peptides from eNOS, synthetic peptides derived from the putative autoinhibitory region of nNOS provided only modest inhibition of calmodulin binding to nNOS and no significant inhibition of nNOS activity. Our results are also not consistent with a proposed autoinhibitory function for this insert in nNOS. For example, chimera nNOS-I 576 -1144 contains the reductase domain of iNOS, but its activity is inhibited in the absence of Ca 2ϩ . Also, chimeras containing the nNOS reductase domain with iNOS canonical calmodulin-binding region and oxygenase domain (nNOS-I 1-533 and nNOS-I 1-575 ) showed significant Ca 2ϩ -independent activities (50 -73%). Nishida and Ortiz de Montellano (42) recently reported that substitution of the reductase domain of nNOS into iNOS resulted in the retention of significant Ca 2ϩ -independent activity. The evidence that the insert in the FMN domain may function as an autoinhibitory element for eNOS is more compelling (13) and may point to a structural and functional distinction between nNOS and eNOS. Differences in the Ca 2ϩ regulation of a class of structurally related, calmodulin-dependent enzymes (skeletal and smooth muscle myosin light chain kinases) have been previously noted (43,44).
Ca 2ϩ -dependent activation of nNOS was associated with calmodulin binding, a finding similar to many previously published results. TFP inhibits the activities of other Ca 2ϩ /calmodulin-regulated enzymes, e.g. myosin light chain kinase, phosphodiesterase, calcineurin, etc. (45)(46)(47)(48). Previous reports showed that TFP abolishes nNOS but not iNOS activity, which is confirmed herein (49 -51). However, inhibition of nNOS activity by TFP in the presence of Ca 2ϩ is not associated with dissociation of calmodulin. The inactivation of nNOS by TFP may be similar to the inactivation of another calmodulin-dependent enzyme, myosin light chain kinase. TFP does not cause dissociation of calmodulin from the kinase, and inhibi- FIG. 7. Inhibition of NOS activities with trifluorperazine and L-NAME. NOS activities in lysates of COS-7 cells transfected with nNOS, iNOS, nNOS-I 1-533 , and nNOS-I 504 -1144 were measured in the presence of 300 M free Ca 2ϩ (Ⅺ) or 2 mM EGTA (f) by method 1 at the indicated concentrations of TFP and L-NAME (ࡗ). A, nNOS activity was sensitive to inhibition by TFP. B, iNOS activity was resistant to inhibition by TFP either in the presence or absence of Ca 2ϩ relative to nNOS. C, activity of nNOS-I 1-533 was resistant to inhibition by TFP in the presence but not in the absence of Ca 2ϩ . D, activity of nNOS-I 504 -1144 was resistant to inhibition by TFP in the presence or absence of Ca 2ϩ . The Ca 2ϩ -dependent and Ca 2ϩ -independent activities of iNOS, nNOS-I 1-533 , and nNOS-I 504 -1144 were both normalized to 100% in the absence of TFP, respectively. L-NAME (200 M) completely abolished nNOS, iNOS, nNOS-I 1-533 , and nNOS-I 504 -1144 activities in the presence of 1 mM TFP either in the presence or absence of Ca 2ϩ . Data shown are means Ϯ S.D. of at least three experiments. Error bars smaller than symbols are not shown.

FIG. 8. Binding of calmodulin to wild-type and chimeric NOSs.
Co-immunoprecipitation of calmodulin with different NOS enzymes was used to evaluate high affinity association of calmodulin. COS-7 cells were transfected with the indicated NOS constructs or with vector alone (mock controls), and their lysates were immunoprecipitated with anti-nNOS (␣ nNOS) or anti-iNOS (␣ iNOS) in the presence of 100 M Ca 2ϩ (C) or 2.5 mM EGTA (E) and in the presence or absence of 1 mM TFP. Immunoprecipitates were subjected to 6 and 15% SDS-polyacrylamide gel electrophoresis for Western blot analysis by anti-NOSs (upper panels) and by anti-calmodulin antibodies (lower panels), respectively. NOSs were not detected in mock controls, and calmodulin levels in mock controls were the same as those in immunoprecipitates of nNOS and nNOS-I 1-533 in the presence of EGTA with or without TFP treatment (data not shown). Similar results were obtained from nNOS L696V and iNOS L475V as their respective wild-type nNOS and iNOS (data not shown). tion is thought to be due to a distortion of the calmodulin structure necessary for activation (52,53).
Two chimeras responded differently to EGTA and TFP in terms of activity and calmodulin binding. In the case of nNOS-I 504 -1144 , EGTA alone or in combination with TFP did not result in dissociation of calmodulin, and there was no effect on enzyme activity. The iNOS segment containing amino acids 484 -726 with the calmodulin-binding region and the N-terminal portion of the reductase domain may contribute to Ca 2ϩindependent binding of calmodulin (11). This high affinity binding of calmodulin may account for one of the mechanisms responsible for Ca 2ϩ -independent activity of iNOS. It is conceivable that calmodulin binds to the calmodulin-binding region of iNOS with the N-and C-terminal lobes of calmodulin in close opposition forming a tunnel that engulfs the amphipathic ␣-helix of the iNOS calmodulin-binding sequence. This proposed model is similar to the known structure of calmodulin bound to a similar calmodulin-binding sequence from myosin light chain kinase (31,35,54). Using a synthetic peptide of the calmodulin-binding sequence from nNOS, Zhang et al. (55) showed that the structure of bound calmodulin was similar to that obtained with calmodulin bound to the myosin light chain kinase sequence. Although a similar study has not yet been performed with the calmodulin-binding sequence of iNOS, it is not unreasonable to expect a similar structure. Because calmodulin binding to a synthetic peptide of the calmodulin-binding sequence of iNOS is Ca 2ϩ -dependent (12,56), additional interactions with the reductase domain may result in Ca 2ϩindependent binding of calmodulin to iNOS. However, nNOS-I 504 -1144 activity was increased in the presence of Ca 2ϩ . Upon binding of Ca 2ϩ /calmodulin to nNOS, the oxygenase and reductase domains are tethered to project an effective electron transfer pathway for producing nitric oxide (9,10). Moreover, this tethering also increases the electron transfer rate from NADPH to flavin centers by a factor of 10 and 20 in eNOS and nNOS, respectively (9,57). Conceivably, the binding of Ca 2ϩ to the calmodulin bound to nNOS-I 504 -1144 may cause a conformational change sufficient to increase enzyme activity, resulting in a more efficient electron transfer pathway. These results are analogous to the effect of Ca 2ϩ on the activity of myosin light chain kinase in which calmodulin was cross-linked (58). In the absence of Ca 2ϩ , the kinase had 50% of the activity measured in the presence of Ca 2ϩ . Evidence was presented that the association of calmodulin was sufficient to stimulate enzyme activity, and the binding of Ca 2ϩ to calmodulin increased catalytic efficiency.
Results obtained with nNOS-I 1-533 , which contains the calmodulin-binding region and oxygenase domain of iNOS, were different. In the absence of Ca 2ϩ , the nNOS-I 1-533 chimera had 50% of the activity measured in the presence of Ca 2ϩ , similar to results obtained with nNOS-I 504 -1144 . However, calmodulin was not co-immunoprecipitated with the enzyme in the absence of Ca 2ϩ . One possible explanation for these results is that calmodulin binding is not necessary for activity. Activity was detected in the reconstituted activity of eNOS oxygenase and reductase domains expressed separately in the absence of calmodulin compared with its presence; however, the activity was low (57). A likely explanation is that calmodulin may bind in the presence of EGTA, but its affinity is lower compared with its binding affinity with iNOS and nNOS-I 504 -1144 . Washing of the immunoprecipitated protein may be sufficient to remove the weakly bound calmodulin. This idea is supported by the ability of TFP to inhibit the activity of nNOS-I 1-533 in the presence of EGTA but not Ca 2ϩ . When calmodulin is bound to the calmodulin-binding region of iNOS, there may be an additional interaction with the oxygenase domain that is stronger in the presence of Ca 2ϩ . However, full Ca 2ϩ -independent activity of iNOS requires calmodulin binding to the canonical calmodulin-binding region as well as to both the reductase and oxygenase domains. Additional investigations involving biophysical and structural studies are necessary to verify this model.