Electron Transfer and Catalytic Activity of Nitric Oxide Synthases

The nitric oxide synthases (NOS) are single polypeptides that encode a heme domain, a calmodulin binding motif, and a flavoprotein domain with sequence similarity to P450 reductase. Despite this basic structural similarity, the three major NOS isoforms differ significantly in their rates of ·NO synthesis, cytochromec reduction, and NADPH utilization and in the Ca2+ dependence of these rates. To assign the origin of these differences to specific protein domains, we constructed chimeras in which the reductase domains of endothelial and inducible NOS, respectively, were replaced by the reductase domain of neuronal NOS. The results with the chimeric proteins confirm the modular organization of the NOS polypeptide chain and demonstrate that (a) similar residues establish the necessary contacts between the reductase and heme domains in the three NOS isoforms, (b) the maximal rate of ·NO synthesis is determined by the maximum intrinsic ability of the reductase domain to deliver electrons to the heme domain, (c) the Ca2+ independence of inducible NOS requires interactions of calmodulin with both the calmodulin binding motif and the flavoprotein domain, and (d) the effects of tetrahydrobiopterin and l-arginine on electron transfer rates are mediated exclusively by heme domain interactions.

Clear differences exist in the Ca 2ϩ and CaM dependence of the three NOS isoforms. The binding of CaM to nNOS and eNOS, the two constitutive isoforms, is Ca 2ϩ -dependent and reversible (10,11), whereas the binding of CaM to iNOS is essentially a Ca 2ϩ -independent, irreversible process (12). Thus, the catalytic activities of nNOS and eNOS are regulated by cellular Ca 2ϩ -levels, whereas the activity of iNOS is insensitive to the Ca 2ϩ concentration and is primarily regulated by the rate at which the protein is synthesized (12). In addition to the heme, CaM, and reductase domains common to all the isoforms, nNOS has an additional N-terminal domain thought to be involved in subcellular targeting (13,14), and eNOS is unique in that it has myristoylation and palmitoylation sites that target it to the membrane (15,16).
The catalytic activities of the NOS isoforms differ considerably, regardless of whether activity is measured as ⅐NO formation or cytochrome c reduction. Regardless of which of these two parameters is measured, the activities of iNOS and nNOS are considerably higher than that of eNOS. Thus, in our hands, the V max values for ⅐NO formation by recombinant iNOS (17), nNOS (18), and eNOS (19) were found to be ϳ800, ϳ400, and ϳ130 nmol min Ϫ1 mg Ϫ1 , respectively, and the rates of electron transfer to cytochrome c in the presence of CaM and Ca 2ϩ were ϳ45,000, ϳ44,000, and ϳ1,800 nmol min Ϫ1 mg Ϫ1 , respectively. In each case, the reductase domain was able to transfer electrons to cytochrome c at a rate much greater than the maximum rate of ⅐NO production, although the nature of the ratelimiting step(s) in ⅐NO formation and the reason(s) for the differences in the catalytic activities of the isoforms are unclear. The reduction of cytochrome c by nNOS and eNOS occurs in the absence of CaM and Ca 2ϩ , but the rate is greatly stimulated by the binding of these cofactors (20,21). In the case of iNOS, the cytochrome c reductase activity is not Ca 2ϩ -dependent. Expression of truncated eNOS and nNOS proteins consisting only of the CaM binding and reductase domains has confirmed that the CaM/Ca 2ϩ activation of electron transfer to cytochrome c does not require the heme domain (22)(23)(24).
The specific protein interactions that result in quasi-irreversible binding of CaM to iNOS but not nNOS or eNOS have not been defined. Chimeras in which mouse iNOS residues 503-532 were exchanged with the equivalent residues 725-754 of rat nNOS differed from the parent proteins in that both required an intermediate concentration of Ca 2ϩ to bind CaM and produce ⅐NO (25). Furthermore, truncation analysis of iNOS suggested that residues within the sequence 484 -726 that lie outside the canonical CaM binding sequence are required for Ca 2ϩ -independent CaM binding. In an independent study, Venema et al. (26) replaced the CaM binding domain of murine iNOS with that of bovine eNOS and vice versa. The NOS activity of the iNOS chimera proved to be partially Ca 2ϩdependent, whereas the eNOS chimera was CaM-independent but completely Ca 2ϩ -dependent. These results again suggest that the tight, essentially Ca 2ϩ -independent binding of CaM requires interactions in addition to those provided by the consensus iNOS CaM binding sequence. In contrast, surface plasmon resonance studies of the binding to CaM of peptides derived from the CaM binding domains of nNOS and iNOS led to the conclusion that the affinity of iNOS for CaM resides entirely in the CaM binding canonical sequence (27).
To identify contributions of the individual domains of the NOS polypeptide to (a) the differences in the rates of ⅐NO production by the three isoforms, (b) the basis for the Ca 2ϩindependent binding of CaM in iNOS, and (c) the effects of L-Arg and H 4 B on electron transfer rates, we constructed, expressed, and purified two chimeric proteins. In these chimeric proteins, the heme and CaM binding domains were contributed by either eNOS or iNOS, and the reductase domain by nNOS.

EXPERIMENTAL PROCEDURES
Materials-Bovine eNOS (7) and rat brain nNOS (6) cDNAs were kind gifts from William C. Sessa (Yale University) and Solomon H. Snyder (Johns Hopkins University), respectively. A mouse iNOS cDNA was provided by Stephen Black (University of California, San Francisco). The mouse iNOS was cloned into the pCWori vector and expressed and purified as described for the human hepatic iNOS isoform (17). The human CaM gene (28) was provided by Emanuel E. Strehler (Mayo Foundation). Enzymes used in DNA manipulation were from New England Biolabs (Beverly, MA). L-Arg was from Aldrich, H 4 B was from Alexis Biochemicals (San Diego, CA), and HEPES was buffer from Fisher. Bradford protein assay kits were from Bio-Rad. Recombinant human CaM was purified from Escherichia coli by published procedures (28). DNA purification kits and Ni 2ϩ -NTA agarose were purchased from QIAGEN (Chatsworth, CA). BL21 competent cells were from Novagen (Madison, WI). Unless otherwise indicated, all other reagents were from Sigma.
DNA Manipulations-One of the chimeric proteins was constructed by subcloning a PCR-generated fragment from pBluescriptKS/eNOS (7) into a mutant pBluescriptSK/nNOS cDNA. The eNOS fragment was obtained by PCR amplification using Vent R DNA polymerase (New England Biolabs, Beverly, MA) possessing a 3Ј to 5Ј proofreading exonuclease activity for higher fidelity than was available with Taq polymerase. The eNOS cDNA coding for amino acids Met-1 to Ser-528 was amplified using primer 1 (5Ј-GC CCCAGC CATATG GCAAAC TT-GAAA AGCGTG GGTCAG GA), which introduced a 5Ј NdeI site at the starting methionine (in boldface) and a G2A mutation for increased expression, and primer 2 (5Ј-CGGCC GGTCTC GCTAGC GTACAG GATG), which introduced a 3Ј NheI splice site. The nNOS cDNA (6) was mutated using the Transformer Mutagenesis Kit (CLONTECH); selection primer 3 (5Ј-GTC GACGGT ATTAAT AAGCTT GATATC), which converted a unique ClaI site in pBluescriptSK to a unique AseI site; and primer 4 (5Ј-CTG AAGGAC ACAGAT CATATG GAAGAG AACACG), which introduced a 5Ј NdeI site at the starting methionine. Simultaneously, the NheI splice point was created using mutagenesis primer 5 (5Ј-G ACCATT CTCTAC GCTAGC GAGACA GGCAAA TCAC), which resulted in a T761S mutation and yielded pBluescriptSK/nNOS NheI761. Sequence alignment using the GAP program (29) revealed that this amino acid is located in a stretch that is identical between nNOS and eNOS (Table I) and therefore was not expected to grossly affect the properties of the chimera. Indeed, the mutation is silent regarding the eNOS sequence and is a conservative mutation from threonine for nNOS. The chimeric gene construct was made by subcloning the NdeI-NheI eNOS fragment into pBluescriptSK/nNOS-NheI761. A 3Ј XbaI site in the multiple cloning domain was utilized to subclone the entire NdeI/XbaI chimeric gene into a pCWori expression plasmid that also possessed an insert coding for an N-terminal His 6 tag to aid in purification.
The IHC/NR construction was analogously produced from PCR-generated IHC from pBluescript/iNOS and PCR primers AGTCT CACAT ATGGC TTGCC CGTGC AAGTT TCTGT TCAA and CGGGC GTCGC TAGCA AAGAG GACTG TGGC to generate NdeI and NheI terminal restriction sites, respectively. The NheI splice site introduces a silent mutation at the conserved T539 of iNOS, which is homologous to Thr-761 of nNOS. Subcloning was then performed as for EHC/NR.
Protein Expression and Purification-The His 6 proteins were expressed in protease-deficient BL21(DE3) cells (Novagen) and purified as described previously (30).
Activity Assays-Cytochrome c reduction and NADPH oxidation rates were measured at 37°C on a Cary 1E spectrometer equipped with a Lauda circulating bath using the extinction coefficients ⑀ 550 nm ϭ 21 mM Ϫ1 cm Ϫ1 and ⑀ 340 nm ϭ 6200 M Ϫ1 cm Ϫ1 , respectively. NOS activity was measured at 37°C as the ⅐NO-dependent conversion of ferrous to ferric methemoglobin (31) using an extinction coefficient of ⌬⑀ 401-411 of 60 mM Ϫ1 cm Ϫ1 . Exogenous cofactors, when added, were 5 M FAD, 5 M FMN, 1 mM Ca 2ϩ , and a 3-fold molar excess of CaM over NOS monomer. For EGTA inhibition experiments, the EGTA stock solution was adjusted to pH 7.5 before addition.
Kinetic Parameter Determination-The values of K m and V max were determined for the EHC/NR chimera purified on ADP-agarose and CaM-Sepharose columns in the absence of H 4 B and L-Arg. For the IHC/NR chimera, the CaM-Sepharose column could not be used because CaM is already tightly bound to IHC/NR due to coexpression of CaM with the chimera. Imidazole was omitted from all buffers to avoid contamination of the final pure protein with bound imidazole. Buffer D 50 mM HEPES, pH 7.5, 50 mM EDTA, 500 mM NaCl, 10% glycerol, 5 mM 2-mercaptoethanol, and protease inhibitors (100 M phenylmethylsulfonyl fluoride, 1 M leupeptin, 1 M pepstatin, and 1 g/ml antipain) was used to elute the protein from the Ni 2ϩ -NTA column. After rebinding to ADP-Sepharose, the bound protein was washed with at least 5 column volumes of Buffer D. The purification was then continued as described earlier.
Gel Filtration Studies-Size exclusion chromatography was performed by FPLC on an HR200 Superdex (Pharmacia 10/30) column at room temperature using a flow rate of 0.5 ml/min of running buffer composed of 10% glycerol plus 150 mM phosphate-buffered saline and 2 mM dithiothreitol. When applicable, H 4 B was added to the protein solution to a concentration of 100 M. The running buffer contained 5 M H 4 B.

RESULTS
Chimera Construction-To explore the electron transfer mechanism in NOS, we constructed two chimeras in each of which the heme and CaM binding domains of one NOS isoform were fused to the reductase domain of a second isoform. In one chimera, the eNOS heme and CaM binding domains were fused with the nNOS reductase domain (EHC/NR), and in the other, the iNOS heme and CaM binding domains were fused to the nNOS reductase domain (IHC/NR).
The splice points for the chimeras were chosen to minimize whatever structural perturbations might arise from a change in the amino acid at the splice site and were located at regions with high primary sequence identity between isoforms, as determined by the GAP program from the GCG software package (29). The splice points, indicated by boldface type, are shown in Table I. The sequences of the splice points for the EHC/NR and IHC/NR chimeric proteins share the sequence of at least one of the parent isoforms, with no additions or deletions, and therefore no direct structural perturbation due to the splicing was expected.
Protein Expression and Purification-The chimeric proteins were expressed in E. coli under the same conditions used to express the wild-type proteins (17)(18)(19)32). Because there is no H 4 B or CaM in E. coli, the proteins are expressed in the absence of these cofactors, although the absence of CaM can be remedied by coexpression of the human CaM gene in the bacterial cells (17,19,33,34). As the chimeras were tagged with a polyhistidine peptide, their purification was achieved, as was done previously for the wild-type eNOS (19,32), nNOS (18), and iNOS isoforms (17), by affinity chromatography first on a Ni 2ϩ -NTA column and subsequently on a 2Ј,5Ј-ADP-agarose column (Fig. 1). The proteins were purified in either the presence or absence of H 4 B, yielding proteins that were designated, for example, H 4 B(ϩ)EHC/NR and H 4 B(Ϫ)EHC/NR, respectively. For the EHC/NR chimera, the final protein yield was 5 mg, and the specific activity was 500 nmol min Ϫ1 mg Ϫ1 . SDSpolyacrylamide gel electrophoresis analysis of the protein (Fig.  1, lane 4) indicated that the protein was obtained in a highly purified state. Furthermore, the molecular mass of the protein indicated by SDS-polyacrylamide gel electrophoresis (ϳ130 kDa) was consistent with the calculated EHC/NR molecular mass of 133 kDa. The active IHC/NR chimera could only be obtained when it was co-expressed with CaM. SDS-polyacrylamide gel electrophoresis analysis of CaM-coexpressed IHC/NR showed that the molecular mass of the purified protein was consistent with the calculated mass of 138 kDa (Fig. 1, lane 8).
The presence of H 4 B during the purification of EHC/NR or IHC/NR had no significant effect on the final protein yield.
Spectroscopic Properties-H 4 B(ϩ)EHC/NR and H 4 B(Ϫ)-EHC/NR are spectroscopically nearly indistinguishable. The only detectable difference is a slightly broader peak in the H 4 B-free protein (not shown). The Soret band at max ϭ 400 nm is similar to that of wild-type eNOS and nNOS (32) and indicates that the protein is primarily high spin. Addition of H 4 B and L-Arg converts the low spin shoulder to high spin. CaMcoexpressed H 4 B(ϩ)IHC/NR possesses a spectrum similar to that of EHC/NR. However, CaM-coexpressed H 4 B(Ϫ)IHC/NR exhibited a red-shifted Soret band at max ϭ 418 nm similar to that found for H 4 B(Ϫ)iNOS (17,30,35). These red-shifted Soret maxima, which are not observed with the other H 4 B(Ϫ)NOS isoforms (compare Refs. 17, 32, 36), 2 indicate that these proteins are primarily low spin. Addition of 100 M H 4 B produced nearly equal populations of high and low spin heme, and addition of 1 mM L-Arg continued the trend toward a high spin heme. Dithionite reduction of both chimeric proteins in the presence of CO produced the expected 444 nm absorbance maximum of the ferrous-CO complexes.
Dimer Formation-Dimer formation is required for catalytic ⅐NO production (37). Because dimer formation might be affected by altered protein contacts in the chimeras, the ability of both EHC/NR and IHC/NR to form dimers was evaluated. FPLC size exclusion chromatography showed that dimers were present in both H 4 B(ϩ) and H 4 B(Ϫ) EHC/NR independent of whether the samples were preincubated with H 4 B and of whether H 4 B was added to the FPLC elution buffers (Fig. 2). Nearly all (Ͼ90%) of the heme-containing protein was present as the dimer, with approximately 30% of the total protein in the monomeric state. With respect to this H 4 B-independent dimer-ization, EHC/NR behaved very similarly to wild-type human eNOS (19). In contrast, heme-containing but H 4 B-free IHC/NR eluted from the gel filtration Sephadex column as both a monomer and a dimer (Fig. 2, A and B). Although quantitation is difficult due to the similar retention times of the monomer and dimer, the proportion of dimer increased for both the hemecontaining (compare Fig. 2, A and B) and heme-free protein (compare Fig. 2, C and D) when the analysis was carried out in the presence of H 4 B. In this regard, IHC/NR behaves like native iNOS, which was shown earlier to exhibit similar H 4 Bdependent dimerization behavior (37,38), although we saw considerable dimerization even for H 4 B-and L-Arg-free iNOS (Fig. 2). The dimerization properties of the chimeras thus resemble those of the parent that contributes the heme and CaM domains.
⅐NO Formation and Cytochrome c Reduction-To evaluate the ability of the reductase and heme domains in the chimeras to interact productively and to determine whether CaM binding involves contacts with the protein outside the putative CaM binding domain, we measured the cytochrome c reduction and NOS activities of the chimeric proteins. For both chimeras, the omission of H 4 B during protein purification did not affect enzyme activity if the proteins were reconstituted with the cofactor. Exogenous flavins were required to achieve maximal activities for both chimeras, which suggests, as for the native isoforms (39), that some flavin loss occurred during purification despite the presence of flavins in the purification buffers.
The cytochrome c reductase activity of EHC/NR was as high as that of wild-type nNOS (Table II). CaM stimulated the EHC/NR cytochrome c reduction rate, as it does for the wildtype isoforms, but the magnitude of the enhancement was greater for the chimera than for the wild-type proteins. In the presence of exogenous flavins, the CaM-dependent enhancement for the chimera was 20 -60-fold, whereas the enhancement under similar conditions for eNOS and nNOS was ϳ10fold. The cytochrome c reductase activity of CaM-coexpressed IHC/NR was only slightly lower than that of wild-type nNOS or iNOS (Table II)  The kinetic parameters for EHC/NR in the presence of all cofactors indicate that the K m and V max values are somewhat higher for EHC/NR than for eNOS: L-Arg K m ϭ 8.8 M versus 3 M, and V max ϭ 500 versus 95-120 nmol min Ϫ1 mg Ϫ1 , respectively. For IHC/NR, L-Arg K m ϭ 9.2 M and V max ϭ 630 nmol min Ϫ1 mg Ϫ1 ; these values compare well with K m ϭ 12 M and V max ϭ 800 nmol min Ϫ1 mg Ϫ1 for recombinant hepatic iNOS purified in this laboratory (17).
The ⅐NO synthase activity of EHC/NR, which was much higher than that of wild-type eNOS, was comparable to that of the wild-type nNOS (Table II). The ability of this chimera to synthesize ⅐NO was abolished by exogenous EGTA and is therefore fully Ca 2ϩ -dependent. The specific activity of CaM-coexpressed IHC/NR approached that of wild-type nNOS and iNOS, despite the increased low spin state of the prosthetic heme group. The ⅐NO synthase activity of IHC/NR exhibited a Ca 2ϩ dependence intermediate between that of the constitutive and inducible isoforms. Whereas the activities of nNOS and eNOS are completely dependent of the Ca 2ϩ concentration, and that of iNOS is completely independent of it, IHC/NR exhibited a partial Ca 2ϩ dependence (Fig. 3). Whereas iNOS tolerated 5 mM EGTA with only 5-10% activity loss, IHC/NR showed a concentration-dependent drop in activity. Interestingly, the majority of the activity loss occurred with a K i of approximately 100 M EGTA, yet 40% of the activity was retained even in the presence of 5 mM EGTA.
NADPH Oxidase Activity-The effects of both L-Arg and H 4 B on CaM-dependent NADPH oxidation can be used to estimate the degree of coupling between ⅐NO synthesis at the heme domain and NADPH oxidation at the reductase domain. The effects of L-Arg and H 4 B on the catalytic coupling of nNOS, iNOS, eNOS, and the two chimeras were therefore examined to explore the effects of the domain swaps on the electron transfer properties. Wild-type nNOS, eNOS, and iNOS exhibited NADPH oxidation rates under turnover conditions that correlated with their ⅐NO synthesizing activity: 159 Ϯ 12, 237 Ϯ 13, and 57 Ϯ 1 mol NADPH oxidized per min per mol of NOS, respectively. NADPH oxidation at the reductase domain varies considerably as a function of the binding of L-Arg and H 4 B to the oxygenase domain, and the pattern of this variability is isoform specific.
nNOS exhibited a high degree of uncoupling in the absence of L-Arg, and this uncoupling was increased nearly 2-fold by the binding of H 4 B (Fig. 4). Addition of L-Arg alone or of L-Arg plus H 4 B decreased uncoupled electron transfer. Uncoupled turnover of nNOS has been reported, but the stimulation by H 4 B was not observed, probably because the baculovirus-expressed enzyme used in that study was already partially saturated with H 4 B (40). In a second study, the uncoupled turnover of L-Arg-free nNOS and reduction of this uncoupling by addition of L-Arg were described (20). The uncoupling pattern for iNOS differed from that for nNOS. Under all conditions, except the presence of both L-Arg and H 4 B, the rate of NADPH oxidation by iNOS was approximately 1 ⁄5 that of the fully reconstituted enzyme. eNOS, in agreement with its relatively low ⅐NO synthesizing and cytochrome c reducing activities (Table II), exhibited a lower NADPH consumption rate under turnover conditions than nNOS and iNOS. The binding of L-Arg to L-Arg-and H 4 B-free eNOS increased NADPH consumption by ϳ50%, and the binding of H 4 B alone decreased the rate by ϳ50%. The binding of both L-Arg and H 4 B, which results in catalytic turnover, increased NADPH consumption by 40%. Thus, NADPH consumption contrasts sharply for the three isoforms, both in terms of its magnitude under turnover conditions and with respect to its sensitivity to the presence or absence of L-Arg or H 4 B.
With the NADPH consumption profiles for the native isoforms in hand, the corresponding profiles for the two chimeras were determined. Under turnover conditions, EHC/NR exhibited a level of NADPH consumption similar to that of nNOS (Fig. 4), in agreement with its high ⅐NO synthesizing and cytochrome c reducing activities (Table II). The effects of L-Arg and H 4 B on NADPH consumption, however, were very similar to those for eNOS. L-Arg alone leads to a modest increase in uncoupled NADPH consumption, whereas H 4 B has the opposite effect. Thus, although the overall turnover rate reflects the high activity of the nNOS reductase domain, the regulation of the activity by cofactor binding is closer to that of the eNOS heme domain. In the case of IHC/NR, NADPH consumption is nearly identical to that of iNOS in that only under turnover conditions (i.e. in the presence of both L-Arg and H 4 B) was NADPH consumption elevated. A small difference between the iNOS and IHC/NR profiles is that H 4 B alone leads to a 60% a Exogenously added to assay system in addition to NADPH, H 4 B, and L-Arg. b As measured by the rate of cytochrome c reduction. c As measured by the hemoglobin assay for ⅐ NO. increase in the uncoupling of IHC/NR, whereas it has no effect on iNOS. The EHC/NR and IHC/NR NADPH consumption profiles resemble those of the parent that contributes the oxygenase domain. DISCUSSION Although the known mammalian NOS isoforms have the same enzymatic function and cofactor requirements, there are clear differences in their biochemical and biophysical properties. Major differences are the posttranslational modification of eNOS but not nNOS or iNOS and the differences in the Ca 2ϩ dependence of the inducible versus constitutive isoforms. More subtle differences include (a) the lower cytochrome c reducing and ⅐NO synthesizing activities of eNOS versus nNOS and iNOS (17)(18)(19); (b) differences in the quaternary structure requirements indicated by the fact that eNOS dimerization is nearly independent of H 4 B (19,32,41), whereas the dimerization of iNOS and nNOS is enhanced by H 4 B and L-Arg (37,38,42); and (c) marked differences in NADPH oxidase activity and its dependence on L-Arg and H 4 B (Fig. 4) (20, 40).
The dimerization of eNOS, iNOS, and nNOS appears to be primarily mediated by heme domain binding contacts (41), although immunocoprecipitation of truncated constructs suggests that in the case of eNOS there may also be binding contacts in the reductase domain (43). Although the dimerization of eNOS is independent of H 4 B (32, 44), it has been reported that H 4 B enhances but is not required for dimerization of iNOS (41) and nNOS (42). In our hands, iNOS, expressed and purified in the absence of H 4 B and L-Arg, eluted from a gel filtration column as two peaks of approximately equal intensity with the elution times expected for monomeric and dimeric iNOS (17) (Fig. 2). Gel filtration indicated that the H 4 B independence of eNOS dimer formation is preserved in EHC/NR. This finding suggests that sequences of the eNOS heme binding domain provide sufficient protein contact surfaces for dimer formation. Further support for this conclusion is provided by the observation that a truncated eNOS consisting of the heme and consensus CaM binding domains readily dimerizes. 3 Likewise, the dimerization properties of IHC/NR resemble those of iNOS, which provides the heme domain of the chimeric protein.
Thus, in contrast to EHC/NR and eNOS, the dimerization of both iNOS and IHC/NR is enhanced by H 4 B. Conversely, the results also establish that replacing the reductase domains of eNOS and iNOS with that of nNOS does not introduce interactions that interfere with the heme domain dimerization interactions.
Although they exhibit a similar dependence on Ca 2ϩ and CaM, the two constitutive NOS isoforms differ significantly in their ⅐NO synthase, cytochrome c reductase, and NADPH oxidase activities. nNOS has a high ⅐NO synthesizing activity reminiscent of that of iNOS (Table II), whereas the catalytic activity of eNOS is considerably lower. The iNOS reductase domain alone has been shown to exhibit high cytochrome c reductase activity in the absence of CaM (45). A truncated nNOS protein composed of the CaM binding sequence and reductase domain exhibited CaM-enhanced reductase behavior similar to that of native nNOS (22). The cytochrome c reductase activity of the IHC/NR chimera was comparable to that of iNOS but, significantly, the activity of the EHC/NR chimera was high and resembled that of nNOS or iNOS rather than that of eNOS. This indicates that the heme domain does not exert significant control over the cytochrome c reduction activity of the protein.
Heterologous expression of correctly folded, catalytically active IHC/NR in E. coli, as found for iNOS, requires coexpression with CaM. CaM-coexpressed IHC/NR has ⅐NO synthase and cytochrome c reductase activities comparable to those of both iNOS and nNOS (Table II). The retention of 40% activity even in the presence of 5 mM EDTA suggests that CaM is bound tightly to the IHC/NR chimera, but, in contrast to iNOS (12), not in a completely Ca 2ϩ -independent manner. Thus, replacement of the iNOS reductase domain by the corresponding domain of nNOS preserves all the elements required for high activity but introduces a perturbation to the complete Ca 2ϩ independence of iNOS activity.
In EHC/NR, replacement of the eNOS with the nNOS reductase domain conveys on the chimeric protein the higher electron transfer and ⅐NO synthesizing activities of nNOS. The Ca 2ϩ and CaM dependence of eNOS and nNOS are preserved in the EHC/NR chimera. These results, combined with the finding that the neuronal CaM binding/reductase domain without the oxygenase domain retains full electron transfer activity and CaM dependence (22), are consistent with assembly of the NOS proteins from discrete, independent domains. Thus, the neuronal reductase domain in EHC/NR imparts all of the features of nNOS that are required for high NOS activity. This high ⅐NO synthase activity of EHC/NR results directly from the increased rate of electron transfer made possible by replacement of the relatively ineffective eNOS reductase domain by the highly active nNOS reductase domain. The control of electron transfer to cytochrome c and heme is therefore internal to the reductase domain and is not differentially gated in the NOS isoforms by the heme domain.
The effects of L-Arg and H 4 B on NADPH oxidation provide further insight into the influence of reductase identity on the intramolecular transfer of electrons from NADPH to the heme. For both chimeras, the effect of L-Arg and H 4 B on the NADPH consumption profile matches that of the "parent" isoform contributing the heme domain. Thus, IHC/NR, like iNOS, exhibits low NADPH consumption under all but turnover conditions, and the EHC/NR NADPH consumption is higher than that of eNOS but comparable to that of nNOS. Nevertheless, for EHC/ NR, NADPH consumption was little affected by the binding of L-Arg or H 4 B, in agreement with the corresponding behavior of eNOS. For IHC/NR and iNOS, the low NADPH consumption in the absence of L-Arg and H 4 B, or of H 4 B alone, probably results from low dimer formation, because iNOS dimer formation was shown by Siddhanta et al. (46) to be necessary for the transfer of electrons from NADPH to the heme. The NADPH consumption results with the wild-type proteins are similar to those of Presta et al. (47) but differ from those of List et al. (40). As shown here, the NADPH consumption behavior for both chimeras is consistent in terms of level with the identity of the reductase, but the identity of the oxygenase domain defines the effects of L-Arg and H 4 B on the NADPH consumption. This agrees with the fact that the substrate and H 4 B bind in the oxygenase domain.
Our results with the IHC/NR chimera must be viewed in light of the conclusion of Nathan and co-workers (25) from studies of chimeric constructs of the neuronal and inducible isoforms, that sequences in iNOS in addition to the canonical 3 I. Rodríguez-Crespo, unpublished results. CaM binding domain are required to maintain the Ca 2ϩ independence of ⅐NO formation. The neuronal reductase domain in our chimeric construct may not possess these additional elements required for Ca 2ϩ -independent behavior because 60% of its activity was lost in the presence of 5 mM EGTA. However, the retention of nearly 40% of its activity at high EGTA concentrations suggests that CaM activation and binding are retained under low Ca 2ϩ conditions by IHC/NR to an extent greater than by the constitutive isoforms. The retention of activity even in the presence of 5 mM EGTA is reminiscent of the results of Venema et al. (26), who constructed chimera I501-523 eNOS, a derivative of eNOS in which the CaM binding sequence for iNOS was substituted for the corresponding sequence of eNOS. I501-523 eNOS retained 50% of its activity in the presence of 10 mM EGTA. However, our results for wild-type iNOS differ from those of Venema et al. (26) in that our protein retained Ͼ80% of its activity in the presence of 10 mM EGTA, whereas only 50% of the activity was retained with their protein.
Interestingly, the nNOS sequence encompassing residues 820 -880 aligns moderately well (29) with eNOS residues 585-657. This sequence is a candidate for a possible autoinhibitory domain common to many CaM-dependent enzymes (48). However, comparison of the sequences of nNOS and iNOS shows that the sequence is absent from the corresponding iNOSϳ598 -612 region, a lack that differentiates iNOS from the constitutive, Ca 2ϩ -dependent nNOS and eNOS isoforms (Fig. 5). In our IHC/NR chimeric construct, the potential autoinhibitory nNOS sequence was introduced into iNOS, yet significant Ca 2ϩ independence was retained. If this sequence serves as an autoinhibitory domain in nNOS and eNOS, its introduction (along with the entire nNOS reductase domain) into iNOS is not sufficient to engender a Ca 2ϩ dependence comparable to that of nNOS. It is nevertheless possible that it contributes to the partial loss of Ca 2ϩ independence observed with EGTA.
Our results are consistent with a domain-isolated function of the reductase module in NOS. In IHC/NR, the variable neuronal reductase activity substitutes for the highly active, CaMindependent iNOS reductase. In EHC/NR, the neuronal reductase similarly substitutes for the low-activity eNOS reductase, imparting an increased level of reductase activity that results in a substantially increased enzymatic activity. These results provide direct evidence that an increased reductase activity directly increases the ⅐NO synthase activity. This correlation of activities strongly suggests that the rate-limiting step involves electron transfer within the reductase domain rather than dioxygen bond breakage, substrate monooxygenation, substrate binding, or product release.