Characterization of the roles of the 594-645 region in human endothelial nitric-oxide synthase in regulating calmodulin binding and electron transfer.

It has been postulated that a segment (residues 594-645) inserted in the FMN subdomain of human endothelial nitric-oxide synthase (eNOS) plays a crucial role in controlling Ca(2+)-dependent CaM binding for eNOS activity. To investigate its functions, we expressed human eNOS in a baculovirus system with deletion of a 45-residue segment from this region (residues 594-606 and 614-645, designated as Delta45eNOS), and characterized the purified mutant enzyme. In contrast with wild-type eNOS, Delta45eNOS exhibited characteristics resembling inducible NOS (iNOS). It contained an endogenously bound CaM, which was essential in folding and stabilizing this mutant enzyme, and retained 60% of L-citrulline formation in 5 mM EGTA. We also produced four N-terminally truncated reductase domains with or without the 45-residue segment, and either including or excluding the CaM-binding sequence. Basal cytochrome c reductase activity of reductase domains without the 45-residue segment was up to 20 fold greater than that of corresponding insert-containing domains, and higher than CaM-stimulated activity of the wild-type enzyme. A series of mutants with smaller fragment deletion in this region such as Delta594-604, Delta605-612, Delta613-625, Delta626-634, Delta632-639, and Delta640-645 mutants were further characterized. The crude lysate of mutants Delta613-625 and Delta632-639 did not show activity in the presence of Ca(2+)/CaM, while other four mutants had activity comparable to that of WTeNOS. The purified Delta594-604 and Delta605-612 proteins had a 3-5-fold higher affinity for Ca(2+)/CaM, but their L-citrulline forming activity was still 80% dependent upon the addition of Ca(2+)/CaM. Both mutants exhibited a low level of the cytochrome c and ferricyanide reductase activities, which either did not respond to (Delta594-604) or slightly enhanced by (Delta605-612) the exogenous CaM. In contrast, activities of Delta626-634 and Delta640-645 like those of WTeNOS were largely Ca(2+)/CaM-dependent. Thus, our findings indicate that the N-terminal half of the 594-645 segment containing residues 594-612 plays a significant role in regulating Ca(2+)/CaM binding.

Nitric oxide ( ⅐ NO) is an important signal mediator in diverse physiological and pathological events of mammals (1)(2)(3). This molecule is coproduced with L-citrulline from L-arginine by three different nitric-oxide synthase (NOS) 1 isoforms, all of which share similar biochemical composition and enzymatic characteristics, and possess a bidomain structure (4 -8). The C-terminal half (reductase domain) contains the NADPH-, FAD-, and FMN-binding sites and the N-terminal half (oxygenase domain) contains the heme-, BH 4 -, and L-arginine-binding sites (9 -11). These two domains are connected by a calmodulinbinding region (6,8).
Despite structural similarities, the NOS isoforms can be grouped into two distinct classes according to their mode of expression and dependence on Ca 2ϩ . Constitutive NOS (cNOS) isoforms including endothelial NOS (eNOS) and neuronal NOS (nNOS), are constitutively expressed as latent enzymes that require a higher concentration of Ca 2ϩ for CaM binding and enzyme catalysis (12)(13). In contrast, cytokine-induced NOS (iNOS) is active at a basal level of intracellular [Ca 2ϩ ], presumably due to high affinity of CaM binding. The Ca 2ϩ /CaM-dependent NOS activity was assumed to be an intrinsic property determined by the primary sequences at the CaM-binding region. However, sequence analysis showed that all isoforms contained a similar canonical CaM-binding sequence rich in basic and hydrophobic residues, typical for Ca 2ϩ /CaM-dependent proteins (14). Furthermore, chimeric eNOS and nNOS in which the CaM-binding region was replaced with the corresponding region of murine iNOS (residues 503-532) did not exhibit fully Ca 2ϩ -independent NOS activity (15)(16)(17)(18).
Ruan et al. (15) performed truncation analysis and found an additional region spanning from residues 484 -726 in murine iNOS, which was necessary for Ca 2ϩ -independent NOS activity. Sequence alignment among NOS isoforms suggests that a ϳ50-residue segment present in the FMN-binding domain of human eNOS (residues 594 -645) and nNOS (residues 834 -882) but absent in that of iNOS represents a putative autoinhibitory element that impedes CaM binding to cNOSs at the basal level of intracellular [Ca 2ϩ ], a feature common to many Ca 2ϩ /CaM-dependent proteins (19). This notion was supported by inhibition of CaM binding and catalytic activities of eNOS and nNOS with synthetic peptides corresponding to this segment of eNOS, and reversal of inhibition by exogenous CaM (20). However, it remains unclear how the presence of this segment in the constitutive NOS isoforms affects the Ca 2ϩ / CaM dependence, and whether the segment has other roles in function of intact eNOS. To address these issues, we prepared a series of mutants by deleting this segment to various extent All primers were synthesized by Genosys Inc. (Woodlands, TX). Sequences of the mutant cDNAs at junctional regions were confirmed by sequencing using the dideoxy chain termination method (26).
Bacterial Expression-To obtain CaM-free ⌬45eNOS, WTeNOS and ⌬45eNOS were inserted into a bacterial expression pCW ori vector, which was cotransfected with the pgroESL or pACYC184-CaM plasmid (a generous gift from Dr. E. E. Strehler) in BL21 Escherichia coli cells according to the procedures reported by McMillan and Masters (11).
Determination of Nitrate/Nitrite in Culture Medium-The nitrate/ nitrite accumulation in culture medium was measured using a colorimetric assay kit from Cayman Chemical Co. (Alexis, San Diego, CA). Ten million Sf9 cells were seeded in each T 75 culture flask, each of which was individually infected with recombinant viruses (multiplicity of infection 2) for WTeNOS and deletion mutant expressions. Hemin chloride was added into the culture medium 24 h after infection. The amount of nitrite/nitrate in the culture medium was determined at the indicated time by adding Griess reagent. The absorbance at 540 nm was recorded using Dynatech MR5000 microplate reader, and NO 2 Ϫ / NO 3 Ϫ was quantified using NaNO 3 as standard.
Assays of Enzyme Activity-NOS activity was assayed by measuring conversion of L-[ 3 H]arginine to L-[ 3 H]citrulline as described by Bredt et al. (6) with slight modification. The reaction mixture containing 25 mM Tris-HCl, pH 7.5, 0.2 mM DTT, 0.5 M calmodulin, 100 M CaCl 2 , 100 M ␤-NADPH, 100 M BH 4 , 50 M L-arginine, 10% glycerol, and 1 Ci of L-[ 3 H]arginine was incubated with 1 g of enzyme at 37°C for 5 min. Cytochrome c reductase activity was determined at 37°C in a reaction mixture containing 25 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10% glycerol, 50 M cytochrome c, 0.5 M calmodulin, 100 M CaCl 2 , and ϳ3 g of enzymes. The reaction was initiated by addition of 100 M ␤-NADPH and the absorbance at 550 nm was recorded in a Shimadzu-2501 PC equipped with a TCC temperature controller. The reduced cytochrome c was quantified using ⌬⑀ red-ox of 21 mM Ϫ1 cm Ϫ1 . Ferricyanide reduction was measured at 37°C in a reaction mixture containing 0.5 mM ferricyanide and quantified using ⌬⑀ red-ox of 1.02 mM Ϫ1 cm Ϫ1 at 420 nm.
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 43. 7  Optical Spectroscopy-Optical spectra were recorded using a Shimadzu-2501 PC equipped with a TCC temperature controller. The ferrous heme-CO spectrum was obtained by flushing the sample with CO gas, followed by reducing the sample with a few grains of dithionite. The binding affinity of L-arginine for WTeNOS and deletion mutants was determined by perturbation difference spectroscopy as described by McMillan and Masters (27), and dissociation constant (K d ) value was estimated as described previously (9).
SDS-PAGE and Immunoblotting-Protein concentration was determined by the method of Bradford (28). SDS-PAGE was performed on 7.5% slab gel using 3% stacking gel in a Bio-Rad mini-gel apparatus according to the Laemmli (29) procedures and stained by Coomassie Blue R250. For CaM immunoblot, the purified protein (5 g) was subjected to SDS-PAGE in 15% gel under reducing condition and then transferred to polyvinylidene difluoride blot membrane in 25 mM potassium phosphate buffer, pH 7.0 at 20 V and 4°C overnight. The monoclonal antibody raised against CaM (catalog no. C-7055, Sigma) was used. The blot was detected by ECL method (Amersham Pharmacia Biotech) using goat anti-mouse IgG-horseradish peroxidase conjugate as secondary antibody.

RESULTS
Sequence alignment shows that human eNOS contains a 45-residue insert (residues 594 -606 and residues 614 -645, Fig. 1A) in the FMN subdomain (Fig. 1B), which is also present in nNOS but absent in iNOS or other related flavoproteins. An eNOS mutant with the deletion of this 45-amino acid segment designated as ⌬45eNOS (Fig. 1C) was constructed and expressed in Sf9 cells. When a comparable amount of protein was expressed, the NO 2 Ϫ / NO 3 Ϫ concentration in the cultured me-dium of the Sf9 cells 74 h after infection with ⌬45eNOS recombinant viruses was about 50-fold higher than that produced by Sf9 cells infected with WTeNOS recombinant viruses (82 M versus 1.5 M). These results suggest that ⌬45eNOS is active at the basal level of intracellular [Ca 2ϩ ] (Fig. 2). WTeNOS and ⌬45eNOS proteins were purified to near homogeneity and shown to have molecular masses of 133 and 123 kDa, respectively (Fig. 3, lanes 1 and 2). The absorbance spectrum of purified ⌬45eNOS exhibited a Soret peak at ϳ401 nm, a charge-transfer band at 647 nm, and a flavin absorbance shoulder between 450 nm and 475 nm (Fig. 4A). Addition of 1 mM imidazole shifted the Soret peak to 428 nm, and the low spin heme-imidazole complex could be converted back to high spin heme (397 nm) upon addition of 1 mM L-arginine (Fig. 4A). The K d value of ⌬45eNOS for L-arginine calculated from the spectral changes was 1.4 M (Fig. 4A, inset). The dithionitereduced, CO-saturated ⌬45eNOS displayed spectral peaks at 445 and 550 nm (Fig. 4B). These spectral properties are similar to those of WTeNOS (30), indicating that deletion of 45 residues did not perturb the heme electronic environment.
The cofactor requirement of purified ⌬45eNOS (Fig. 5, hatched bar) was compared with that of WTeNOS (Fig. 5, filled bar). In the presence of all essential cofactors, the specific activity of L-citrulline formation was170 nmol/min/mg for ⌬45eNOS versus 130 nmol/min/mg for WTeNOS. Omission of BH 4 resulted in a 65-75% reduction in L-citrulline formation for both WTeNOS and ⌬45eNOS, with the residual activity presumably due to the endogenous BH 4 in the purified proteins. However, omission of CaM from the reaction mixture completely eliminated L-citrulline formation for WTeNOS without an apparent effect for ⌬45eNOS. It was reported that murine iNOS retained 50 -85% of enzyme activity in 5 mM EGTA (15)(16)18), whereas eNOS activity was lost in the presence of 2 mM EGTA (5,16). Catalytic activity was thus measured in the presence of 5 mM EGTA. No activity was detected for the WTeNOS, but ϳ60% of activity was still retained for ⌬45eNOS. This behavior indicates that the activity of ⌬45eNOS is largely Ca 2ϩ -independent and WTeNOS is fully Ca 2ϩ /CaM-dependent (Fig. 5).
To determine whether the Ca 2ϩ /CaM independence of ⌬45eNOS was attributed to an intrinsically bound CaM, the purified proteins were subjected to SDS-PAGE, followed by immunoblot with anti-CaM monoclonal antibody. An endogenously bound CaM was found in ⌬45eNOS but not in WTeNOS (Fig. 6, lanes 1 and 2). We have tried to obtain CaM-free ⌬45eNOS to test whether ⌬45eNOS could generate ⅐ NO in the absence of CaM by using a bacterial expression system. WTeNOS and ⌬45eNOS were cloned into a bacterial expression pCW ori vector, which was cotransfected with the pgroESL or pACYC184-CaM plasmid into BL21 E. coli cells. WTeNOS purified from either system was similar and active, whereas ⌬45eNOS had to be coexpressed with pACYC184-CaM, suggesting that ⌬45eNOS requires CaM to properly fold and stabilize its conformational structure (data not shown).
Each of the constructs displayed almost a 1:1 FAD to FMN ratio (ϳ0.8 eq) (data not shown), indicating that removal of the 45-residue segment did not perturb FMN binding affinity. The absorbance spectra of the purified reductase domains had absorption peaks at 454 and 381 nm (Fig. 7, A-D). Both CaM/ FMN/FAD (Fig. 7A) and FMN/FAD (Fig. 7C) domains, which contained the 45-residue segment exhibited an additional prominent peak at 590 nm in the resting state, attributable to an air-stable flavin semiquinone radical, whereas both domains without the 45-residue segment (⌬45CaM/FMN/FAD and ⌬45FMN/FAD) did not exhibit the semiquinone peak in the resting state (Fig. 7, B and D). Addition of NADPH produced a spectrum characteristic of the air-stable semiquinone form in all reductase domains (Fig. 7, A-D). Presumably this portion is due to partial reduction of FMN or FAD in contrast to a complete reduction to flavin hydroquinone by dithionite.
The NADPH-dependent electron transfer rates to ferricyanide and cytochrome c were determined using the purified proteins in the absence (basal) or presence of Ca 2ϩ /CaM (Table  I). Cytochrome c and ferricyanide reductase activities in the WTeNOS were increased 3-5-fold in the presence of CaM. Both activities in ⌬45eNOS did not respond to exogenous CaM and were ϳ2-fold higher than CaM-stimulated activities of WTeNOS. CaM/FMN/FAD domain (residues 482-1204) displayed a similar rank order of basal and CaM-stimulated ferricyanide reductions. The cytochrome c reduction of this domain was enhanced 3.5 times by addition of Ca 2ϩ /CaM. Basal and CaM-stimulated cytochrome reductions in the CaM/FMN/ FAD domain were both 4-fold higher than that of WTeNOS and also higher than that previously reported for a slightly shorter reductase domain (residues 491-1204, Ref. 9). The reason for the later difference is unknown. The corresponding reductase domain lacking the 45-residue insert (⌬45CaM/FMN/FAD) had an even higher ferricyanide reductase activity, which was slightly stimulated by CaM. Adding Ca 2ϩ /CaM to this domain enhanced cytochrome c reduction by 3-fold. The FMN/FAD reductase domain had a low level of ferricyanide and cytochrome c reduction rates, which were not significantly changed by CaM, consistent with its not containing a CaM-binding sequence. The ⌬45FMN/FAD reductase domain had ferricyanide and cytochrome c reductase activities which were 3-and 20-fold, respectively, greater than those of the parent FMN/ FAD domain, and which was not enhanced by exogenously added CaM.
The functional significance of this segment was further evaluated using a series of mutants with smaller fragment deletion in this region such as ⌬594 -604, ⌬605-612, ⌬613-625, ⌬626 -634, ⌬632-639, and ⌬640 -645. We expressed these mutants in the Sf9/baculovirus system. At 72 h after infection with recombinant baculoviruses, the concentration of NO 2 Ϫ / NO 3 Ϫ in the culture media was determined. The ⌬594 -604 and ⌬605-612 mutants produced a considerable quantity of NO 2 Ϫ / NO 3 Ϫ close to ⌬45eNOS, while the other mutants produced either undetectable (mutant ⌬613-625 and ⌬632-639) or a small quantity (mutants ⌬626 -634 and ⌬640 -645) of NO 2 Ϫ / NO 3 Ϫ similar to WTeNOS (Fig. 8A), indicating that like ⌬45eNOS, mutants ⌬594 -604 and ⌬605-612 are active at the basal level of intracellular [Ca 2ϩ ]. The L-citrulline formation of cell lysates of each mutant was determined in the presence of optimal concentration of Ca 2ϩ /CaM and other essential cofactors. The results shows that all the mutants except ⌬613-625 and ⌬632-639 are at least as active as WTeNOS. Mutants ⌬613-625 and ⌬632-639 lost the activity completely despite equivalent expression of the proteins (Fig. 8B), suggesting a global conformational change in these two mutants; they were not further studied. Four active mutants (⌬594 -604, ⌬605-612, ⌬626 -634, and ⌬640 -645) were thus purified, and the spectra were consistent with those seen in WTeNOS and ⌬45eNOS (data not shown).
The Ca 2ϩ and CaM dependences of the purified mutant and wild-type enzymes were determined. Dependence of L-citrulline formation on CaM was titrated by adding increasing concentrations of CaM along with 100 M Ca 2ϩ and other cofactors in the reaction mixture. The CaM concentration-response curves are shown in Fig. 9A. ⌬45eNOS was active without exogenous CaM and addition of CaM did not increase the activity, while WTeNOS required about 50 nM of [CaM] to reach half-maximal activity (EC 50 ). The EC 50 values for ⌬594 -604 and ⌬605-612 mutants were ϳ15 and ϳ10 nM, respectively, whereas the EC 50 values for ⌬626 -634 and ⌬640 -645 were higher, 22 and 30 nM, respectively. The results suggest that these four shorter deletion mutants have an increased CaM binding affinity when compared with WTeNOS but the binding was weaker than ⌬45eNOS. This notion was supported by the absence of an intrinsically bound CaM in all shorter deletion mutants (data not shown).
Dependence on [Ca 2ϩ ] was similarly evaluated in the presence of 500 nM CaM and Ca 2ϩ dependence curves were shown in Fig. 9B. The procedure to obtain the desired free Ca 2ϩ by adding EGTA and Ca 2ϩ -EGTA was similar to the previously reported procedures (15,31). ⌬45eNOS retained ϳ60% activity in the absence of Ca 2ϩ , but activity increases with increasing Ca 2ϩ concentration. The WTeNOS was inactive without added Ca 2ϩ , and the EC 50 was estimated to be 150 nM. The ⌬594 -604 and ⌬605-612 mutants had a detectable but low activity in the absence of Ca 2ϩ and the EC 50 values were 25 and 35 nM, respectively. The ⌬626 -634 and ⌬640 -645 had no activity without addition of Ca 2ϩ , and the EC 50 values were estimated  FMN/FAD (panel A), ⌬45CaM/  FMN/FAD (panel B), FMN/FAD (panel  C), and ⌬45FMN/FAD (panel D). The solid lines denote the resting spectra of ϳ16 M proteins, and the dashed lines, the spectra after addition of 30 M NADPH. The dotted lines denote the spectra after addition of a few grains of dithionite. a Constructs are shown in Fig. 1. to be 65 and 55 nM, respectively (Fig. 9B). We also determined the cytochrome c and ferricyanide reductase activities for the shorter deletion mutants (Table I). Both reductase activities were enhanced 3-4-fold for the mutants ⌬626 -634 and ⌬640 -645, and 1.5-fold for the mutant ⌬605-612 by exogenously added Ca 2ϩ /CaM. In contrast, mutant ⌬594 -604 had a low level of ferricyanide and cytochrome c reduction rates, which were not significantly changed by CaM.
DISCUSSION Although x-ray crystallography results have shown that the oxygenase domains of iNOS and eNOS have similar structures (32,33), there are clear differences in the biochemical and biophysical properties between these two NOS isoforms. Salerno et al. (20) proposed a ϳ50-residue insert unique to cNOS isoforms accounting for the difference in Ca 2ϩ dependence of enzyme activation between cNOS and iNOS. They have shown that synthetic peptides derived from the insert of eNOS (residues 594 -645) are able to block CaM binding and enzyme catalysis (20). However, the exact role of this insert in intact eNOS remains unclear. In this study, we expressed the 45-residue deletion mutant of eNOS in the Sf9/baculovirus system and purified the expressed protein to homogeneity. The ⌬45eNOS mutant had the expected molecular mass, and exhibited optical spectral properties and dimeric structure comparable to the wild-type eNOS. Deletion of the 45-residue fragment from eNOS, hence, did not alter the global structure or the active center of eNOS but drastically changed its dependence on Ca 2ϩ for CaM binding, electron transfer, and catalytic activity.
The ⌬45eNOS expressed in Sf9 cells is highly active, producing a large quantity of ⅐ NO measured as NO 2 Ϫ / NO 3 Ϫ in the culture medium. The purified ⌬45eNOS contained endogenously bound CaM and retained 60% of L-citrulline formation activity even in the presence of 5 mM EGTA. The ⌬45eNOS, hence, resembles iNOS with respect to all the Ca 2ϩ /CaM dependence characteristics. These results support the notion that this 45-amino acid segment impedes binding of CaM to eNOS at the basal level of intracellular [Ca 2ϩ ] (20). Since intrinsically active ⌬45eNOS expressed in Sf9 cells produces a large quantity of ⅐ NO continuously, which is likely to be toxic to the cells (16), it is conceivable that a higher expression level and enzyme activity of ⌬45eNOS might be achieved by adding NOS inhib- itors. However, even with this constraint, the overall L-citrulline formation and cytochrome c activities of ⌬45eNOS were still 1.5 and 2.5 times higher as compared with those of CaMstimulated wild-type eNOS. Thus, removal of this insert resulted in not only an iNOS-like NOS activity but also an increased electron transfer rate. As both redox potential and orientation between each redox partner influence the electron transfer rate (34), deletion of this insert may either change the relative orientation or shift the redox potential between heme and FMN, or FMN and FAD redox centers in eNOS.
Studies have reported that the activating effect of CaM on nNOS can be fully accounted by its interaction with the reductase domains (35). The rate-limiting step for electron transfer in NOS was proposed to be within the reductase domain rather than in the oxygenase domain (18). As this 45-residue insert is also located in the FMN subdomain (20), to determine whether this insert is involved in electron transfer within the reductase domain, we generated a series of reductase domains either with or without this insert and the CaM-binding sequence. Our results showed that the full-length reductase domain (CaM/ FMN/FAD, residues 482-1204) and the reductase domain without the CaM-binding sequence (FMN/FAD, residues 511-1204) contained a considerable amount of air-stable semiquinone radical after isolation. In contrast, the other two reductase domains without the 45-residue insert (⌬45CaM/FMN/FAD and ⌬45FMN/FAD) did not show any flavin semiquinone radical after purification, implying that semiquinone radical in ⌬45CaM/FMN/FAD and ⌬45FMN/FAD reductase domains is less stable in the air, and the insert would play a role in stabilizing the semiquinone radical. Treatment of several reductase domains with a slight excess of NADPH produced an increase in absorbance at 590 nm, indicating the ability of electron transfer between NADPH and FMN/FAD domain. This behavior is similar to CPR (36 -38) and nNOS reductase domain (35,39). On the other hand, the basal and CaM-stimulated cytochrome c reductase activities of eNOS are much lower than the reported values of nNOS, iNOS, and related flavoprotein (9,11,18,35). The reason for a lower eNOS activity is unclear. It is especially intriguing that a difference exists between eNOS and nNOS. Although the 45-residue segment is present in both nNOS and eNOS, synthetic peptides derived from the nNOS insert exhibit only a slight inhibition to eNOS or nNOS activity (20), further supporting the notion of structural and functional differences between eNOS and nNOS.
Deletion of the 45-residue insert from eNOS reductase domain causes a significant increase in the basal cytochrome c and ferricyanide reducing activities by up to 20-fold when compared with the corresponding insert-containing proteins regardless of the presence of CaM-binding site. For example, the cytochrome c activity of ⌬45FMN/FAD domain containing neither the CaM-binding site nor the 45-residue insert is 5-fold higher than the CaM-stimulated activity of WTeNOS (Table I), and approaches to the level of iNOS (40). Removal of this insert results in a high reductase activity, suggesting that the insert plays a suppressive role contributing to the low intrinsic eNOS reductase activity, and the high electron transfer rate can occur within the reductase domain without CaM binding. The high reductase activity in the absence of CaM binding was also observed in the isolated iNOS reductase domain, which naturally did not have this insert, and its CaM-binding site was deleted (41). However, the cytochrome c reductase activity of ⌬45CaM/FMN/FAD protein, which contained the CaM-binding site and lacked of the 45-residue insert, was increased by a factor of 3 upon addition of Ca 2ϩ /CaM, consistent with the result that CaM binding can cause a conformational change in the reductase domain establishing a more suitable orientation for the redox partners that increases electron transfer rate from flavin to cytochrome c or to heme (35).
To dissect the involvement of this 45-residue insert in Ca 2ϩ / CaM-dependent ⅐ NO production, a series of mutants with deletion of smaller fragment in this region were expressed. Two mutants ⌬613-625 and ⌬632-639 completely lost activity despite the presence of Ca 2ϩ /CaM, suggesting a gross disturbance of their structures. Four active proteins, ⌬594 -604, ⌬605-612, ⌬626 -634, and ⌬640 -645, were purified and characterized. The results have indicated the following. 1) All four proteins did not show Ca 2ϩ -independent activity, but required lower Ca 2ϩ and CaM concentrations for maximal activity than WTe-NOS (Fig. 9). Because of having a higher affinity for CaM binding, cells expressing mutants ⌬594 -604 and ⌬605-612 continuously produce ⅐ NO and spontaneously accumulate NO 2 Ϫ / NO 3 Ϫ in the culture medium (Fig. 8A). Therefore, segment 594 -612 plays a more important role in regulating CaM binding and enzyme activity. This is not consistent with the peptide studies (20), which proposed RRKRK motif as essential for blocking CaM binding and cNOS activity. The RRKRK motif is situated in the segment 626 -634, but not in the segment 594 -612. Our results indicate that a free peptide might not truly reflect the specific interaction of corresponding residues in the intact protein. 2) In addition to having much lower Ca 2ϩ requirement for maximal ⅐ NO synthesis, deletion of residues 594 -604 unexpectedly resulted in a low level of ferricyanide and cytochrome c reduction rates, which was not stimulated by exogenous CaM. This observation led to the proposal that segment 594 -604 has two effects. First, it covers or allosterically perturbs the CaMbinding site in eNOS, thereby obstructing CaM binding, and hence deletion of 594 -604 greatly increases sensitivity to Ca 2ϩ /CaM for NO synthesis. Second, the activation of eNOS reductase activities by CaM requires the interaction of 594 -604 segment with CaM that causes a conformational change of the reductase domain and enhances electron transfer to the cytochrome c and ferricyanide. Thus, the 594 -604 deletion results in a basal level of reductase activities that no longer respond to the exogenous CaM. 3) Ca 2ϩ /CaM binding is required for mediating the electron transfer from the reductase to oxygenase domain, and the basal level of electron transfer within reductase domain is sufficient for ⅐ NO synthesis. Although the reductase activity of ⌬594 -604 is low and not sensitive to the exogenous CaM, its L-citrulline formation is Ca 2ϩ /CaM-dependent and approaches the level of WTeNOS upon Ca 2ϩ /CaM binding (Table I). Moreover, ⌬45eNOS activity could not be detected when it was expressed in bacteria without CaM (data not shown), probably because of perturbance of its global structure. The result suggests that CaM is crucial for proper folding and stabilization of ⌬45eNOS. These data were similar to the report that an active mouse iNOS in E. coli required coexpression with CaM (40).
Immunoblot analysis using an anti-CaM antibody in our study showed that ⌬45eNOS contained an endogenously bound CaM, while WTeNOS, other deletion mutants, and N-terminally truncated reductase domains did not. It appears that the tightly-bound CaM to NOS requires the absence of entire 45residue segment like ⌬45eNOS and iNOS. Removal of the 45-residue insert eliminates the steric hindrance, contributing to an enhanced CaM affinity at a basal level of intracellular [Ca 2ϩ ]. However, the ⌬45CaM/FMN/FAD domain, which contained the CaM-binding site but lacked the 45-residue insert, did not contain endogenous CaM, suggesting that CaM binding requires not only the canonical CaM-binding sequence but also sequences in the reductase domain and the oxygenase domain. Absence of oxygenase domain in ⌬45CaM/FMN/FAD protein would therefore weaken the interaction of CaM with eNOS, resulting in reversible CaM binding.
In conclusion, we have demonstrated by deletion experiments that the 45-amino acid insert in the FMN subdomain of eNOS influences CaM binding and Ca 2ϩ -dependent ⅐ NO production. It also plays a regulatory role in controlling electron transfer within the reductase domain, contributing to a low intrinsic reductase activity. Data from this report have provided valuable information on the regulation of Ca 2ϩ /CaM binding as well as the reductase and oxygenase activities of eNOS by a 45-residue region in the eNOS molecule.