Mutations at Lysine 525 of Inducible Nitric-oxide Synthase Affect Its Ca2+-independent Activity*

Calmodulin binding to inducible nitric-oxide synthase may play an important role in its Ca2+-independent activity. Studies of inducible nitric-oxide synthase chimeras containing the calmodulin binding sequence of neuronal or endothelial nitric-oxide synthases show that the calmodulin binding sequence of inducible nitric-oxide synthase is necessary but not sufficient for the Ca2+-independent activity. The mutations at lysine 525 located at the C terminus of the calmodulin binding sequence of inducible nitric-oxide synthase were examined for the effects on the Ca2+-independent activity with chimeras containing the oxygenase or reductase domains of inducible or neuronal nitric-oxide synthases. Results show that the Ca2+-independent binding of calmodulin is not solely responsible for maximal Ca2+-independent activity of inducible nitric-oxide synthase. Lysine 525 of inducible nitric-oxide synthase may also play an important role in coordinating the maximal Ca2+-independent activity.

mers of NOSs (5) and increases the electron transfer rate within the reductase domain (6 -9), the monomers are incapable of generating NO. In contrast, CaM tightly associates with iNOS in a Ca 2ϩ -independent manner as an integral subunit. Accordingly, a Ca 2ϩ -independent electron transfer pathway exists in iNOS for its constitutive activity (10). Moreover, NOS chimeras, containing the CaM-binding region and the oxygenase domain or the reductase domain of iNOS with different affinities for CaM, confer significant Ca 2ϩ -independent activity. The CaM-binding region of iNOS alone is necessary but not sufficient for Ca 2ϩ -independent activity of iNOS or NOS chimeras with significant Ca 2ϩ -independent activity (11)(12)(13). The maximal Ca 2ϩ -independent activity of iNOS may be conferred by a range of structural features in all three domains of the enzyme, the CaM-binding site and oxygenase and reductase domains.
Chimeric and mutant CaMs have been used as informative probes to analyze the relative importance of each Ca 2ϩ -binding site for activation of regulated enzymes, e.g. myosin light chain kinase, phosphodiesterase, and nNOS (14 -17). Studies of the interactions of CaM mutants with nNOS indicated that CaM binding and stimulation of NO synthesis are distinct and separate events, although CaM binding to the CaM-binding sequence is necessary for catalysis (16,17). Specific interactions between CaM and nNOS beyond the CaM-binding sequence either induced or not induced by Ca 2ϩ binding are required for efficient electron transfer. Mutation or phosphorylation of the CaM-binding regions of nNOS or eNOS decreases Ca 2ϩ /CaM binding affinity (12, 18 -20) and thus enzyme activity (18,20). As yet, no specific residues or phosphorylation events have been reported to be a primary determinant for CaM binding to iNOS or for Ca 2ϩ -independent activity of the enzyme. Moreover, the Ca 2ϩ -independent, high affinity binding of CaM to iNOS makes it difficult to study the interactions between CaM and iNOS. The possibility of other metal ions involved in NOS activity is raised by the partial inhibitory effect of EGTA on iNOS activity (12,(21)(22)(23)(24)(25)(26)(27) and potentiation of the activities of nNOS and iNOS by non-heme iron (28).
Herein, we use NOS chimeras with or without point mutations to explore the role of lysine 525 of iNOS in the Ca 2ϩindependent activity. Lys-525 of iNOS resides at the C terminus of the CaM-binding region and is conserved in iNOSs from different animal species but not in nNOS or eNOS. The mutations at Lys-525 affect iNOS activity in the absence of Ca 2ϩ but not through Ca 2ϩ -independent binding of CaM in the CaM-NOS complex.

Construction of Mutant
NOSs-Residue lysine 525 of iNOS, a conserved residue at the C terminus of the CaM-binding sequence, and the corresponding residues in chimeras nNOS-I 1-533 , nNOS-I 504 -1144 , and nNOS-I 503-533 were mutated to glutamate or alanine in cDNAs of K/E iNOS, K/A iNOS, K/E nNOS-I 1-533 , K/A nNOS-I 1-533 , K/E nNOS-I 504 -* This work was supported in part by National Institutes of Health Grants HL26043 and HL06296 and the Bashour Research Fund. 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 U.S.C. Section 1734 solely to indicate this fact.
Other Methods-In addition to the DEAE-dextran method (13), Fu-Gene TM 6 transfection reagent (Roche Molecular Biochemicals) was also used for COS-7 cell transfection as recommended by the manufacturer. FuGene TM 6 transfection reagent, 36 l, was diluted into 864 l of serum-free Dulbecco's modified Eagle's medium (Life Technologies, Inc.) and incubated at room temperature for 5 min. The diluted Fu-Gene TM 6 transfection reagent was then added to 9 g of cDNA in 45 l and incubated at room temperature for 15 min. The resultant Fu-Gene TM 6/cDNA mixture was added to one 100-mm plate of COS-7 cells with 10 ml of fresh medium. At 16 -24 h post-transfection, the transfection media were replaced with fresh medium with addition of 2 mM L-NAME, 200 M BH 4 , or 10 M hemin in combination for expressing active enzymes. The transfected cells were harvested as described previously (13). Co-immunoprecipitation, trifluoperazine (TFP) treatment, and Western blot analysis were also performed as described previously (13). Anti-CaM monoclonal antibody was purchased from Upstate Biotechnology (Catalog no. 05-173).

RESULTS
Mutations in nNOS-I 1-533 -nNOS-I 1-533 comprises the oxygenase domain and the CaM-binding region of iNOS and the reductase domain of nNOS (Fig. 1A). nNOS-I 1-533 has 50 -70% activity in the absence of Ca 2ϩ (2-2.5 mM EGTA) compared with the presence of Ca 2ϩ (Fig. 2). This Ca 2ϩ -independent activity of nNOS-I 1-533 is inhibited by TFP (a CaM antagonist) in a concentration-dependent manner and is not associated with high affinity binding of CaM (see Figs. 7 and 8 in Ref. 13). The equivalent residue of iNOS lysine 525 in nNOS-I 1-533 residing in the C-terminal end of the CaM-binding region was mutated to glutamate in K/E nNOS-I 1-533 (Fig. 1). Interestingly, K/E nNOS-I 1-533 had no significant NOS activity in the FIG. 1. A, domain scheme for nNOS, iNOS, and chimeric NOSs. NOS enzymes contain oxygenase and reductase domains separated by a CaM binding sequence. The CaM-binding segment can be subdivided into the N-terminal flanking, canonical CaM-binding, and C-terminal flanking regions. NADPH, FAD, and FMN consensus-binding regions are located in the reductase domain, whereas heme, Arg, O 2 , and BH 4 bind to the oxygenase domain. Additionally, nNOS contains a PDZ domain located at its N-terminal end, which is a protein interaction module and is involved in cellular localization. Chimeric and mutant NOSs were constructed as described under "Materials and Methods." Shown in the figure are nNOS, iNOS, nNOS-I 504 -1144 , (nNOS residues 726 -1429 replaced with iNOS residues 504 -1144), nNOS-I 1-533 (nNOS residues 1-755 replaced with iNOS residues 1-533), nNOS-I 503-533 (nNOS residues 725-755 replaced with iNOS residues 503-533), and point mutants (K/E iNOS, K/A iNOS, K/E nNOS-I 1-533 , K/A nNOS-I 1-533 , K/E nNOS-I 504 -1144 , and K/E nNOS-I 503-533 derived from iNOS, nNOS-I 1-533 , nNOS-I 504 -1144 , and nNOS-I 503-533 with the corresponding lysine 525 of iNOS replaced by glutamate (K/E) or alanine (K/A), respectively). B, comparison of CaM binding sequences among NOS isoforms. The ϩ3 residue is indicated outside a 12-amino acid core sequence with a pair of aromatic or long chain hydrophobic amino acids (indicated by a pair of connecting arrows) that bind to hydrophobic pockets in the N and C domains of CaM. A lysine residue at the ϩ3 position from the C-terminal hydrophobic residue in the CaM-binding sequence is conserved in iNOS (bold and underlined) from mammal and chicken versions (36 -42). This is different from the corresponding residues Thr or Gln in mammal eNOS and nNOS as well as Arg from Drosophila nNOS. The representative sequences for mammalian NOSs are bovine eNOS, rat nNOS, and murine iNOS. There are variations in mammal iNOS CaM-binding sequences (listed under the corresponding residue in lower case), but no variations occur in mammal eNOS and nNOS. absence of Ca 2ϩ (2-2.5 mM EGTA) and required a higher concentration of Ca 2ϩ /CaM for half-maximal activation (Fig. 2). To determine whether the loss of activity of K/E nNOS-I 1-533 in the absence of Ca 2ϩ was due to the introduction of a negative charge or the loss of a positive charge, K/A nNOS-I 1-533 was constructed. Although K/A nNOS-I 1-533 had a lower concentration of Ca 2ϩ /CaM required for half-maximal activation than K/E nNOS-I 1-533 (Fig. 2B), it was also mainly dependent on Ca 2ϩ for NO production with marginal activity (10 -30%) in the absence of Ca 2ϩ (Fig. 2, A and B). Because the specific activities of K/A nNOS-I 1-533 and K/E nNOS-I 1-533 in the presence of Ca 2ϩ were comparable with nNOS-I 1-533 , the alanine or glutamate replacements at the corresponding residue of iNOS lysine 525 in nNOS-I 1-533 only affected the activity in the absence of Ca 2ϩ (2-2.5 mM EGTA). The residue equivalent to position 525 of murine iNOS is well conserved as lysine in all versions of iNOS but not in nNOS and eNOS (Fig. 1B). Thus, the positive charge of lysine 525 of iNOS may play a role in the activity measured in the absence of Ca 2ϩ with 2-2.5 mM EGTA.
Mutations in iNOS-Based on the results with K/E nNOS-I 1-533 and K/A nNOS-I 1-533 , the corresponding mutations were introduced into intact iNOS (Fig. 1A) to investigate the mutational effect on wild type iNOS with full Ca 2ϩ -independent activity. The mutations at lysine 525 decreased Ca 2ϩ -independent activity from 90% (iNOS) to 60% (K/A iNOS) and 45% (K/E iNOS), respectively (Fig. 3A). The activation responses to [Ca 2ϩ ] of K/E iNOS and K/A iNOS (Fig. 3B) were similar to chimeras nNOS-I 504 -1144 and nNOS-I 1-533 (Figs. 2B and 4B). Although K/E iNOS and K/A iNOS had specific activities in the presence of Ca 2ϩ similar to iNOS, both these mutant enzymes required Ca 2ϩ for full activation. These results suggest that mutations at lysine 525 in iNOS attenuate iNOS activity in the absence of Ca 2ϩ (2-2.5 mM EGTA).
Mutations in nNOS-I 504 -1144 and nNOS-I 503-533 -The mutational effect on another NOS chimera with significant Ca 2ϩindependent activity was examined. nNOS-I 504 -1144 comprises the reductase domain and the CaM-binding region of iNOS and the oxygenase domain of nNOS (Fig. 1A). This enzyme has 50 -70% activity in the absence of Ca 2ϩ (2-2.5 mM EGTA) as compared with the presence of Ca 2ϩ , which is similar to nNOS-I 1-533 . The activity of nNOS-I 504 -1144 in the absence of Ca 2ϩ is resistant to TFP inhibition and is associated with high affinity binding of CaM as found for iNOS and distinct from TFP inhibition of nNOS-I 1-533 (13). Results with K/E nNOS-I 504 -1144 containing a Lys to Glu mutation showed that the mutated chimera also had 60 -70% Ca 2ϩ -independent activity in the presence of Ca 2ϩ , and its [Ca 2ϩ ] dependence for activation was similar to non-mutated nNOS-I 504 -1114 (Fig. 4). Thus, the mutation had no effect on nNOS-I 504 -1144 activity in contrast to the effect observed in nNOS-I 1-533 or iNOS. These results suggest that the involvement of lysine 525 in iNOS activity in the absence of Ca 2ϩ (2-2.5 mM EGTA) occurs with the concerted nNOS-I 503-533 consists of nNOS with its CaM-binding region (residues 725-755) substituted for that of iNOS (residues 503-533) (Fig. 1A) and is dependent on Ca 2ϩ for activation but with a lower concentration of Ca 2ϩ /CaM for half-maximal activation than nNOS (Fig. 4). nNOS-I 503-533 and nNOS-I 1-533 share the same CaM-binding sequence but differ in the origin of the oxygenase domain. Because the concentration of Ca 2ϩ /CaM required for half-maximal activity of K/E nNOS-I 1-533 was greater than that for nNOS-I 1-533 , the corresponding mutation at nNOS-I 503-533 was predicted to also increase the concentration of Ca 2ϩ /CaM for required half-maximal activity compared with nNOS-I 503-533 . This proved to be the case (Fig. 4C). Accordingly, lysine 525 of iNOS may be in a position not only to influence activity in the absence of Ca 2ϩ (2-2.5 mM EGTA) but also to facilitate NO production at low Ca 2ϩ concentrations.
TFP Inhibition and Association of CaM with K/E iNOS, K/E nNOS-I 504 -1144 , and K/E nNOS-I 1-533 -Because the point mutation affects the Ca 2ϩ -independent activity differently in iNOS, nNOS-I 504 -1144 , and nNOS-I 1-533 , the mutational effects on other properties of these enzymes were further studied. TFP (a CaM antagonist) treatment and co-immunoprecipitation with CaM were used to further investigate the properties of K/E iNOS, K/E nNOS-I 504 -1144 , and K/E nNOS-I 1-533 compared with their respective parental enzymes. The activity of K/E iNOS was resistant to TFP inhibition in the presence of Ca 2ϩ but more sensitive in the absence of Ca 2ϩ compared with iNOS or with nNOS-I 504 -1144 ( In the absence of Ca 2ϩ , CaM binds to iNOS or nNOS-I 504 -1144 with higher affinity than to nNOS-I 1-533 (see Fig. 8 in Ref. 13). CaM binding properties to K/E iNOS, K/E nNOS-I 504 -1144 , and K/E nNOS-I 1-533 were also investigated herein to study the mutational effect on their affinity for calmodulin (Fig. 6). 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 plus 0.5% Triton X-100 for dissociating CaM. Co-immunoprecipitation results showed that CaM associated with K/E iNOS and K/E nNOS-I 504 -1144 either in the presence or absence of Ca 2ϩ . Also, CaM dissociated from K/E nNOS-I 1-533 , a Ca 2ϩ -dependent enzyme, in the absence of Ca 2ϩ as found for nNOS and the non-mutated nNOS-I 1-533 (13). Therefore, in contrast to Ca 2ϩ -independent activity and TFP inhibition, the mutation had no significant effect on the affinity of these enzymes for CaM.
Collectively, the mutation at iNOS lysine 525 did not significantly change the properties of K/E nNOS-I 504 -1144 in terms of activity in the absence of Ca 2ϩ (2-2.5 mM EGTA), high affinity association with CaM, and resistance to TFP treatment. In contrast, the mutation decreased activities of K/E iNOS and K/E nNOS-I 1-533 in the absence of Ca 2ϩ (2-2.5 mM EGTA) as well as affected the resistance to TFP treatment for both enzymes but had no effect on affinity for CaM. DISCUSSION Site-directed mutagenesis and crystal structures have provided insight into the functions of the various domains of some NOS enzymes (29 -33). Crystal structures of the dimeric oxygenase domains of eNOS and iNOS illustrate the interface interactions for dimerization and the formation of the active sites. The functional binding of heme, BH 4 , and L-arginine to NOS is thus well depicted and fits in an electron transfer link and relevant spatial interactions for catalysis of NO produc- tion. It was recently shown that a zinc ion coordinates pairs of symmetry-related cysteine residues at the interface of the oxygenase dimer for the FMN domain docking to stabilize BH 4 binding (31,32). These cysteine residues are from a pair of conserved tetrahedral motifs (CXXXXC) each from alternate oxygenase domains. A highly positively charged interaction surface for the FMN domain docking to the oxygenase domain is also proposed in the vicinity of the zinc-binding site. Given the difference in on-off rates of association of the oxygenase and reductase domain of iNOS (always associated) and nNOS (controlled by Ca 2ϩ /CaM association), the docking interactions of iNOS may have evolved differently from nNOS and accordingly account for the Ca 2ϩ -independent activity of iNOS. Replacing iNOS reductase domain with the counterpart of nNOS or vice versa, as in nNOS-I 1-533 or nNOS-I 504 -1144 , may change the docking interface interactions. These considerations may account for the inhibitory effect of EGTA on the activities of nNOS-I 1-533 and nNOS-I 504 -1144 .
The further loss of activity in K/E nNOS-I 1-533 in the absence of Ca 2ϩ with EGTA triggered the investigation into the mutational effect of Lys-525 on iNOS activity. The lack of the CaMbinding region in the crystal structures of both the iNOS and eNOS oxygenase domains (30 -32) hinders our understanding of how CaM is functionally involved. However, in the iNOS oxygenase structure (Fig. 1 in Ref. 32) the ␣-12 helix immediately preceding the CaM-binding region projects toward the caved docking interface, and thus the CaM-binding sequence is probably located around the interface proposed for FMN domain docking. Thus, CaM associated with the CaM-binding sequence may be embedded in the docking interface. If this is true, the equivalent lysine 525 in iNOS and nNOS-I 1-533 may be located at or close to a solvent-accessible surface at the docking site. Mutation of the equivalent lysine 525 in iNOS and nNOS-I 1-533 may change the accessibility of some effective sites to EGTA, e.g. the coordinated zinc, as perhaps occurred with K/A nNOS-I 1-533 and K/E nNOS-I 1-533 , enzymes with activities more sensitive to inhibition by EGTA. Although crystal structures of NOSs are consistent in general with the mutagenesis data, some discrepancy has been noted due to the preparations of proteins and conditions for crystallization (31,32). The zinc ion center of iNOS was not identified in early crystal studies because of the protein preparation (30 -32). Moreover, non-heme iron has been reported to bind nNOS and iNOS stoichiometrically and potentiate their activity (28). Therefore, it is possible that metal ions other than Ca 2ϩ or other EGTA-sensitive factors are involved in maximizing iNOS activity in the absence of Ca 2ϩ .
Because EGTA is a divalent metal ion chelator, the inhibitory effect of EGTA may involve other identified metal factors such as the coordinated zinc ion at the interface of oxygenase dimer for FMN domain docking. EGTA has a greater affinity for Ca 2ϩ (10 Ϫ11 M) than for zinc (10 Ϫ7 M). If the coordinated zinc becomes solvent accessible due to the mutation, EGTA chelation of the zinc ion may occur in the absence of Ca 2ϩ . Given the proposed role of the coordinated zinc ion in pterin binding and reductase docking, dissociation of the zinc ion from K/E nNOS-I 1-533 or K/A nNOS-I 1-533 is likely to cause attenuation in electron transfer from FMN domain to the pterin and heme for NO production. This inhibition would be released when the free Ca 2ϩ concentration increases because of competition for the EGTA, similar to the inhibition of nNOS and iNOS by high concentrations of zinc through competitive binding to a nonheme iron-binding site (28).
EGTA may inhibit iNOS activity to varied extents. Some investigators have reported previously that murine iNOS can be partially inhibited by EGTA (12,27,34) although others found no effect (11,13,35). Similar observations were also reported for rat and human iNOS (21)(22)(23)(24)(25)(26). Herein, our results show that Lys-525 of iNOS is in a position critical for activities of K/A nNOS-I 1-533 and K/E nNOS-I 1-533 in the absence of Ca 2ϩ with EGTA. It also plays an important role in maintaining maximal activity of iNOS in the absence of Ca 2ϩ . Immunoprecipitates were subjected to 6 and 15% SDS-polyacrylamide gel electrophoresis for Western blot analysis by anti-NOSs (upper panels) and by anti-CaM antibodies (lower panels), respectively. NOS protein was not detected in mock controls, and CaM levels in mock controls were the same as those in immunoprecipitates of K/E nNOS-I 1-533 in the presence of EGTA (data not shown). Results showed that replacement of lysine 525 of iNOS and the corresponding residues in NOS-I 504 -1144 and nNOS-I 1-533 with glutamate did not significantly change the relative affinities of these enzymes for CaM in the presence or absence of Ca 2ϩ , which is similar to their respective wild type and nonmutated chimeric enzymes (13).