Characterization of Recombinant Human Endothelial Nitric-oxide Synthase Purified from the Yeast Pichia pastoris *

Human endothelial nitric-oxide synthase (eNOS) was expressed in the methylotrophic yeast Pichia pastoris, making use of the highly inducible alcohol oxidase promoter. The recombinant protein constituted approximately 3% of total protein and was largely soluble (>75%). About 1 mg of purified eNOS was obtained from 100-ml yeast cell cultures by affinity chromatography of crude cell supernatants. The purified enzyme had aV max of 192 ± 18 nmol ofl-citrulline × mg−1 × min−1, had a K m forl-arginine of 3.9 ± 0.2 μm, and showed an absolute requirement for tetrahydrobiopterin (H4biopterin). NADPH oxidase activity was 136 ± 9 and 342 ± 24 nmol × mg−1 × min−1 in the absence and presence of 0.1 mm l-arginine, respectively, and not affected by H4biopterin. The protein contained 0.56 ± 0.06 equivalents of FAD and 0.79 ± 0.08 equivalents of FMN. On-line gel filtration/inductively coupled plasma mass spectrometry analysis confirmed that both iron (0.80 ± 0.09 mol/subunit) and zinc (0.43 ± 0.03 mol/subunit) were bound to the enzyme. Graphite furnace-atomic absorption spectroscopy yielded a value for bound iron of 0.84 ± 0.04 mol/subunit. The absorbance of the enzyme at 398 nm implied a heme content of 0.85 ± 0.09 mol/subunit, and the high pressure liquid chromatography heme assay gave an estimate of 0.71 ± 0.02 mol heme/subunit. Gel permeation chromatography yielded one single peak with a Stokes radius of 6.62 ± 0.7 nm, indicating that the native protein is dimeric. Upon low temperature gel electrophoresis the untreated protein appeared mainly as a monomer (88 ± 3%), but pretreatment with H4biopterin andl-arginine led to a pronounced shift toward dimers (77 ± 4%). Thus, in contrast to bovine eNOS (List, B. M., Klösch, B., Völker, C., Gorren, A. C. F., Sessa, W. C., Werner, E. R., Kukovetz, W. R., Schmidt, K., and Mayer, B. (1997) Biochem. J. 323, 159–165; Rodriguez-Crespo, I., Gerber, N. C., and Ortiz de Montellano, P. R. (1996) J. Biol. Chem. 271, 11462–11467), the human eNOS appears to be markedly stabilized by H4biopterin.

Nitric oxide, an important effector and signaling molecule in the nervous, immune, and cardiovascular systems, originates from L-arginine and oxygen in the reaction catalyzed by NO synthase (NOS 1 , E.C. 1.14.13.39) (reviewed in Refs. [1][2][3]. Three isozymes of NOS are known (neuronal, nNOS; endothelial, eNOS; and inducible, iNOS) that share the same domain structure and catalytic mechanism. All three isozymes catalyze NO synthesis at a cytochrome P-450-like thiolate-bound heme, located in the N-terminal oxygenase domain. Reducing equivalents for the reaction are transferred from NADPH to the heme by a C-terminal reductase domain containing one molecule each of bound FAD and FMN. This electron transfer requires the binding of calmodulin to a site between the reductase and the oxygenase.
Coupling of NADPH oxidation to NO synthesis requires a further cofactor, the pteridine derivative H 4 biopterin (4). It has not yet been determined exactly why NOS requires H 4 biopterin for activity. Although net oxidation to 7,8-dihydrobiopterin, as observed in other H 4 biopterin-dependent enzymes, does not occur (5), H 4 biopterin is suspected to act as a redox cofactor in the reaction, perhaps by means of a novel, transient one-electron chemistry (5)(6)(7). Less controversial than its possible catalytic role are its allosteric effects. These include a shift of the heme coordination equilibrium toward pentacoordinate (high spin) as opposed to hexacoordinate (low spin) forms (8 -10), modulation of the heme redox potential (11), synergistic binding of the pteridine and the substrate L-arginine (9,12), and a marked stabilization of the enzyme dimer (13)(14)(15).
The main differences between the isozymes concern their regulatory properties. The first major distinction is that whereas calmodulin binding and activation of nNOS and eNOS is Ca 2ϩ -dependent, iNOS binds calmodulin irreversibly and its activity is effectively Ca 2ϩ -independent under physiological conditions. Another difference is that nNOS, when less than fully saturated with L-arginine and H 4 biopterin, is able to catalyze uncoupled NADPH oxidation, with release of O 2 . from the heme site, at much higher rates than the other two isozymes (16,17). This reaction of nNOS is suspected to contribute to brain damage during ischemic episodes (stroke) (18), whereas available evidence suggests that under these conditions eNOS fulfills a protective role (19). Thus, precise discrimination between the isoenzymes seems to be a prerequisite for understanding the contribution of NOS to pathological processes, and by extrapolation, for drugs that modulate NOS activity to achieve useful effects (4).
In the case of eNOS, a number of questions related to isozyme-specific properties remain to be solved. First of all, it seemed important to demonstrate whether the suppression of uncoupled NADPH oxidase activity observed previously for the bovine enzyme (15) is preserved in the human enzyme. Secondly, although a relatively low specific activity compared with the other isozymes is to some extent an intrinsic property of eNOS (20), we were curious to see whether human eNOS expressed in eukaryotic cells would exhibit equally low or improved activity. A third question centers on the role of H 4 biopterin in eNOS dimerization. Previous studies of bovine eNOS found a substantial fraction of the characteristic SDSresistant dimer even in the absence of added H 4 biopterin, and addition of H 4 biopterin did not significantly increase this component (15,21). Also, the recent crystal structure of bovine eNOS oxygenase dimer did not reveal significant conformational differences between H 4 biopterin-bound and H 4 biopterinfree forms (7). These results might suggest that dimer stabilization by H 4 biopterin is not a feature of eNOS, which would be a major difference to the other isozymes.
A new feature of NOS structure was discovered by Raman et al. (7) in the course of solving the crystal structure of the bovine eNOS oxygenase dimer. A zinc ion was found at the dimeric interface, bound by two cysteine thiolates from each subunit. The tight, 4S-tetrahedral binding was reminiscent of structural, rather than catalytic zinc sites in other enzymes. Sitedirected mutation of one of the zinc-binding cysteines to alanine greatly reduced zinc incorporation but did not completely inactivate the enzyme, supporting a structural rather than catalytic role. Zinc was also found in nNOS (22), although not in the crystal structure of iNOS oxygenase domain as reported by Crane et al. (23). However, the recent paper of Fischmann et al. (24) confirmed the existence of zinc in human eNOS and iNOS expressed in Escherichia coli. Thus, it is not yet certain whether zinc is bound at this site in all NOSs or depends on expression conditions or perhaps on the expression system. Here we report results relevant to these issues obtained with a highly active human eNOS, obtained in high yield by overexpression in Pichia pastoris.

EXPERIMENTAL PROCEDURES
Materials-L- [2,3,4, H]Arginine hydrochloride (57 Ci/mmol) was from American Radiolabeled Chemicals, Inc. [ 3 H]-Arginine was further purified as described earlier (25). H 4 Biopterin and 4-amino-H 4 biopterin were obtained from Dr. B. Schircks Laboratories (Jona, Switzerland). Materials for molecular biology were from New England Biolabs, Life Technologies, Inc., and Qiagen. The EasySelect Pichia Expression Kit was from Invitrogen (Bio-Trade, Vienna, Austria). Myoglobin (product number M1882), carbonic anhydrase (product number C4831), and all other chemicals were from Sigma. The iron content of the myoglobin was determined by ICP-MS to be 0.71 mol/mol protein and the zinc content of carbonic anhydrase was found to be 0.83 mol/mol protein by the same method. Human eNOS cDNA was from John Parkinson (Berlex Biosciences, Richmond, CA).
Expression and Purification of eNOS-The mutant human eNOS (G2A) was reverted to wild type human eNOS by site-directed (Ala-2 to Gly-2) mutagenesis using the polymerase chain reaction-based mutagenesis as described (26) and confirmed by sequencing using the dideoxy chain termination method (27). The plasmid pcDNA3 containing the cDNA for human eNOS was subsequently double digested with HindIII and NotI. The recessed 3Ј-termini from the HindIII digest were filled by the Klenow fragment of E. coli DNA polymerase I in the presence of the appropriate dNTPs. The expression vector pPICZA (EasySelect Pichia Expression Kit, Invitrogen) was subsequently digested with EcoRI, followed by filling the recessed 3Ј-termini and with NotI. The 3.6-kilobase insert was ligated to restricted pPICZA. E. coli TOP10FЈ cells were transformed with the resulting ligation products and plated on LB/zeocin medium containing 1% tryptone, 0.5% yeast extract, 0.5% NaCl, pH 7.5, and 25 g/ml zeocin. The resulting transformants were tested by restriction analysis, and the positive clones were amplified to make larger amounts of DNA. The final DNA construct was linearized with PmeI, the DNA was transformed into the yeast P. pastoris KM71 (Mut 5 ), and the cells were plated on YPDS/ zeocin medium containing 1% yeast extract, 2% peptone, 2% glucose, 1 M sorbitol, and 100 g/ml zeocin, following the manufacturer's instructions. Transformants were further tested for growth on increasing concentrations (250, 500, and 1000 g/ml) of zeocin. An overnight culture (30°C) was inoculated from a single colony of the best clone in 10 ml of buffered minimal glycerol medium (containing 100 mM potassium phosphate, pH 6, 13.4 g/liter of yeast nitrogen base without amino acids, 400 g/liter biotin, 40 mg/liter L-histidine, and 1% (v/v) glycerol). The next day this culture was used to inoculate 750 ml of buffered minimal glycerol medium (1:200), and the culture was grown overnight at 30°C to an A 600 of 4 -6. The cells were then harvested and resuspended in 150 ml of buffered minimal methanol medium (containing 100 mM potassium phosphate, pH 6, 13.4 g/liter of yeast nitrogen base without amino acids, 400 g/liter biotin, 40 mg/liter L-histidine, and 5% (v/v) methanol) in the presence of 4 mg/liter hemin chloride and incubated for 24 h at 30°C to induce protein expression. Yeast cells were then harvested by centrifugation at 2000 ϫ g for 5 min at room temperature. The cells were resuspended at a concentration equivalent to an A 600 of 125 (based on the A 600 of the culture) in 50 mM Tris/HCl buffer, pH 7.4, containing 1 mM EDTA, 5% glycerol, 12 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, and 1 mM CHAPS. An equal volume of glass beads (0.5 mm) was added to the suspension, and the cells were broken by vigorous vortexing at 4°C for a total of 8 min in bursts of 30 s alternating with cooling on ice. The glass beads were separated by centrifugation at 800 ϫ g for 5 min. After a further clearing step at 1600 ϫ g for 5 min, the supernatant was centrifuged at 30,000 ϫ g for 15 min to yield an extract that was used for protein purification by affinity chromatography as described previously (28). The final elution was with 20 mM Tris/HCl buffer, pH 7.4, containing 100 mM NaCl and 4 mM EGTA. The enzyme was stored at Ϫ70°C in the presence of 1 mM CHAPS. Protein was determined with the Bradford method using bovine serum albumin as a standard (29).
Gel Filtration Chromatography-Gel filtration was performed on a Superose 6 HR 10/30 column connected to an Ä KTA Purifier 10 chromatography system (Amersham Pharmacia Biotech). The temperature was 10°C, the flow rate was 0.3 ml ϫ min Ϫ1 , and the eluent was 50 mM triethanolamine/HCl buffer, pH 7.4, containing 0.5 M NaCl and 0.5 mM EDTA. Where indicated, H 4 biopterin (0.1 mM) and L-arginine (1 mM) were added to the eluent and (30 min prior to injection) to the samples containing 35-90 g of NOS. Elution was monitored at 280 and 398 nm. The elution position of small amounts (Յ 5%) of heme-free monomeric eNOS was detected by subtraction of the 398-nm signal (heme absorption) from the 280-nm signal. Calculation of Stokes radii was performed as described previously (15).
Determination of Cofactors-Enzyme-bound flavins and H 4 biopterin were determined by reversed-phase HPLC and fluorescence detection using authentic FAD, FMN, and H 4 biopterin as standards as described (30), with the standards being treated identically to the enzyme samples. Heme was quantified by reversed-phase HPLC and UV-visible detection using myoglobin as standard, as described (30) with the modification that samples were acidified with 20 mM HCl before mixing 28 l of aqueous protein sample with 42 l of acetonitrile/trifluoroacetic acid (600:1, v/v). Calculation of molar cofactor/NOS ratios is based on a subunit molecular mass of 133 kDa.
Analysis of Enzyme-bound Metals-Iron and zinc bound to eNOS were determined by gel filtration with on-line detection by ICP-MS. Before analysis, the ICP-MS was tuned to give ϳ22000 cps (background 3 cps) for lithium and 32000 cps (background 2 cps) for yttrium at a doubly charged ratio Ce 2ϩ /Ce ϳ1.6% and an oxide ratio of CeO/Ce ϳ0.2% (at 1 g/liter lithium, yttrium, cerium). A Superose 6 HR 10/30 column was run at a flow rate of 0.3 ml ϫ min Ϫ1 on a Hewlett-Packard HP1100 ChemStation HPLC system incorporating a UV monitor set at 280 nm. The elution buffer was 0.05 M Tris/HCl, pH 7.4, containing 0.15 M NaCl. The outlet of the UV detector was connected directly to the Babington-type nebulizer of a Hewlett-Packard HP4500 inductively coupled plasma mass spectrometer. The isotopes chosen for monitoring were 57 Fe and 66 Zn. The background ICP-MS signals were essentially constant over 10 h of continuous operation, showing that the buffer did not adversely affect performance of the mass spectrometer. The autosampler of the HPLC system was set to an injection volume of 10 l. Known amounts (5-20 M) of carbonic anhydrase (for zinc) and myoglobin (for iron) were run to calibrate the peak areas of the MS traces. Graphite furnace-atomic absorption spectroscopy (GF-AAS) was performed on a Hitachi Z-9000 Zeeman-type instrument equipped with a pyrolytic cuvette. For iron, the following temperature program was used: drying, 50°C for 30 s, ramp from 80 -120°C over 60 s, hold at 120°C for 15 s; ashing, 800°C for 30 s; atomization, 2700°C for 10 s and cleaning, 3000°C for 3 s. Samples were prepared by gel filtration on a Superose 6 HR 10/30 column in 50 mM Tris/HCl, pH 7.4 containing 0.15 M NaCl, and buffer samples were collected from the column between runs for background correction. For calibration, an iron standard solution (Titrisol, product number 9972, Merck) was diluted to final concentrations of 0 -1 M in 50% (w/w) of the buffer in which the samples were prepared and 1% (w/w) HNO 3 . Samples of eNOS were diluted 2-fold with Nanopure water to final eNOS concentrations of 0.5-0.7 M. For both standards and NOS samples, 10-l samples were introduced into the cuvette followed by 10 l of 10% (w/w) nitric acid as external modifier. Measurements were repeated at least five times for each enzyme sample.
Determination of Enzyme Activity-NOS activity was determined as formation of L-  (33). The rate of reduction was calculated using the extinction coefficient of 21 mM Ϫ1 ϫ cm Ϫ1 for cytochrome c at 550 nm.
Gel Electrophoresis-Purified eNOS was analyzed as described previously by conventional (34) or low temperature SDS-PAGE (13). To test for formation of SDS-resistant dimers, 32-l samples of eNOS were incubated at 37°C for 10 min in the absence or presence of H 4 biopterin (0.2 mM) and L-arginine (1 mM). Incubations were terminated by chilling the samples on ice and adding 8 l of sample buffer containing 0.32 M Tris/HCl, pH 6.8, 0.5 M glycine, 10% SDS, 50% glycerol and 0.03% bromphenol blue. Samples containing 1-2 g of eNOS were subjected to low temperature SDS-PAGE on discontinuous 5% SDS gels, using the Mini Protean II system from Bio-Rad. Gels were stained either for protein with Coomassie Blue R250 or for heme with 3,3Ј-dimethoxybenzidine/H 2 O 2 using the method of Thomas et al. (35) as adapted by List et al. (15).

RESULTS
Expression of Human eNOS-Human eNOS was expressed in P. pastoris using the highly inducible P AOX1 promoter. Analyzing lysates of methanol-induced cells, a band with an apparent molecular mass of 131 kDa became visible on Coomassie Blue-stained SDS gels, in close agreement with the calculated value of 133 kDa (Fig. 1, lane B versus lane C). This band was recognized on Western blots by an antibody raised against bovine eNOS (data not shown). The recombinant protein was purified to apparent homogeneity by sequential chromatography of crude supernatants over 2Ј,5Ј-ADP-Sepharose and calmodulin-Sepharose (Fig. 1, lanes D and E, respectively). Yeast supernatants typically exhibited specific NOS activities of 4 -6 nmol of L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 . The representative purification shown in Table I led to the isolation of 1.2 mg of eNOS from 110 ml of cell culture. The specific activity of this particular preparation was 190 nmol of L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 . Addition of 1 mM CHAPS was essential to prevent precipitation during storage at Ϫ70°C.
Biochemical and Spectroscopic Characterization-The heme spectrum of eNOS displayed a Soret maximum at 398 nm, and the flavin absorbances at 456 and 480 nm appeared as shoulders (Fig. 2). The enzyme had a K m for L-arginine of 3.9 Ϯ 0.2 M and a V max of 192 Ϯ 18 nmol L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 (Table II). Calmodulin-dependent cytochrome c reduction was catalyzed with a specific activity of 2.81 Ϯ 0.14 mol ϫ mg Ϫ1 ϫ min Ϫ1 , more than 10-fold higher than L-citrulline formation. The protein contained 0.56 Ϯ 0.06 equivalents of FAD and 0.79 Ϯ 0.08 equivalents of FMN. Possible conversion of FAD to FMN during the HPLC assay was excluded by measuring FADonly standards. The pterin cofactor H 4 biopterin was not detectable with the HPLC method (detection limit was 30 nM, corresponding to 0.02 mol of H 4 biopterin/subunit). The absorbance of the pteridine-and substrate-free enzyme at 398 nm (assuming an extinction coefficient of 72 mM Ϫ1 ϫ cm Ϫ1 (36)) implied a heme content of 0.85 Ϯ 0.09 mol/subunit, and the HPLC heme assay gave an estimate of 0.71 Ϯ 0.02 mol heme/subunit (Table  II). On-line gel filtration/ICP-MS analysis confirmed that both iron and zinc were bound to the enzyme (Fig. 3). Comparison of the peak areas with those obtained for myoglobin and carbonic anhydrase led to values of 0.80 Ϯ 0.09 mol of iron/subunit and 0.43 Ϯ 0.03 mol zinc/subunit. GF-AAS yielded an iron content of 0.84 Ϯ 0.04 equivalents/NOS subunit (mean Ϯ S.E. from three preparations), in close agreement with the values from the UV absorbance and the on-line ICP-MS. We were not able to measure zinc in these samples by GF-AAS because the ashing temperatures required were too low to reduce background signals from the buffer components.
Citrulline formation was not detectable when calmodulin, NADPH, or H 4 biopterin had been omitted from the assay but was only slightly reduced, by about 15%, in the absence of added flavins (data not shown). Addition of H 4 biopterin to purified eNOS increased enzyme activity with an EC 50 of 0.23 Ϯ 0.04 M (mean Ϯ S.E., n ϭ 3; Fig. 4A). The potent pterin-site NOS inhibitor 4-amino-H 4 biopterin completely antagonized the effect of H 4 biopterin (Fig. 4B), inhibiting L-citrulline formation with an IC 50  Quarternary Structure of Human eNOS-Purified human eNOS was analyzed by gel permeation chromatography in the presence and absence of L-arginine and H 4 biopterin. Preincubation with L-arginine and H 4 biopterin had no effect and only one major peak, corresponding to the dimer, with a Stokes radius of 6.62 Ϯ 0.07 nm (mean Ϯ S.E., n ϭ 3) was observed (Fig. 5). In some chromatograms, traces of heme-free protein (Ͻ5% of total protein) were found with a Stokes radius of ϳ5.1 nm (data not shown), similar to that observed previously for bovine eNOS monomers (15).
SDS-PAGE at low temperature was used to investigate formation of SDS-resistant dimers of eNOS. Samples of eNOS that had been boiled in SDS sample buffer migrated as a single band with the mobility expected for the monomer (Fig. 6, lane  A). If the sample was not boiled before electrophoresis, another band of lower mobility appeared (Fig. 6, lane B). Following earlier reports (15,21,37) we interpret this band as the SDSresistant dimer of eNOS, which has anomalously high mobility (apparent molecular mass around 200 kDa instead of 266 kDa). In the absence of additions the dimer accounted for 12 Ϯ 3% (mean Ϯ S.E.; n ϭ 3 enzyme preparations) of the protein. Preincubation with L-arginine alone yielded 15 Ϯ 2% dimer (Fig. 6, lane C). Preincubation with H 4 biopterin alone stimulated SDS-resistant dimer formation to 68 Ϯ 2% (Fig. 6, lane  D), and with H 4 biopterin plus L-arginine a further increase, to 77 Ϯ 4% dimer, was observed (Fig. 6, lane F). The samples preincubated with H 4 biopterin also exhibited a minor band (Յ 4%) with slightly higher mobility than the monomer; we interpret this as partially folded monomer, based on the similar observation by Rodriguez-Crespo et al. (21) with bovine eNOS. Staining for heme showed that heme was retained only by the dimers (Fig. 6, lanes G-J).
Uncoupled NADPH Oxidation-The NADPH oxidase activity of human eNOS was assayed in the presence and absence of L-arginine, H 4 biopterin, and the enzyme inhibitors N G -methyl-L-arginine and N G -nitro-L-arginine to see how these factors affect uncoupled O 2 . production by the enzyme. In the absence of additions, eNOS exhibited a Ca 2ϩ /calmodulin-dependent NADPH oxidase activity of 136 Ϯ 9 nmol ϫ mg Ϫ1 ϫ min Ϫ1 (Table III). L-Arginine (0.1 mM) led to a 2.4-fold increase of NADPH oxidation (342 Ϯ 24 nmol ϫ mg Ϫ1 ϫ min Ϫ1 ). In keeping with the pronounced pterin dependence of the enzyme, L-citrulline formation was not detectable under these conditions. Saturation of the enzyme with 10 M H 4 biopterin had no stimulatory effect on NADPH oxidation measured with or without L-arginine but led to coupling of NADPH oxidation to Lcitrulline formation. Thus, in the presence of L-arginine and H 4 biopterin, the stoichiometry of NADP ϩ /L-citrulline formation was 2.05 (the generally accepted minimum stoichiometry for the reaction is 1.5). As reported previously for the porcine neuronal (38,39) and bovine endothelial NOS (15) isoforms, the NADPH oxidase activity of human eNOS was completely inhibited by N G -nitro-L-arginine but only slightly affected by N G -methyl-L-arginine (Table III). DISCUSSION There are only few reports in the literature describing the expression of recombinant NOS in yeast. Saccharomyces cerevisiae was used to express full-length iNOS (40) and fulllength nNOS (41,42), and P. pastoris was used to express the reductase domain of nNOS (43). Recombinant human eNOS was so far primarily overexpressed in E. coli (37) and baculovirus-infected insect cells (44,45) with specific activities of the   purified enzyme ranging from 80 to 170 nmol of L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 . We now successfully overexpressed human eNOS in the yeast P. pastoris and obtained from a 100-ml cell culture about 1 mg of purified enzyme with a specific activity of 190 nmol of L-citrulline ϫ mg Ϫ1 ϫ min Ϫ1 . The P. pastoris system is therefore 10-fold more efficient than the E. coli system without coexpression of calmodulin and 2-fold more efficient compared with coexpression with calmodulin (37). Moreover, the specific activities of human eNOS expressed in P. pastoris were similar in terms of L-citrulline formation and 2-fold higher in terms of cytochrome c reduction than those of the enzyme coexpressed with calmodulin in E. coli, demonstrating that calmodulin is not required for the expression of a functional and highly active eNOS in P. pastoris. Nonetheless, human eNOS expressed in eukaryotic cells exhibits a relatively low specific activity as compared with the other isozymes, supporting the idea that this is to some extent an intrinsic property of eNOS (20,46). One minor concern is our finding of less bound FAD than FMN, which is unusual because general experience with purified NOS indicates that FMN dissociates more easily under mildly denaturing conditions (47). 2 Some further optimization of the culture conditions (perhaps by adding riboflavin) might improve the flavin content of the enzyme.
Similar to the expression of NOS isoforms in E. coli, the expression of human eNOS in P. pastoris yielded an enzyme that contained no detectable amounts of H 4 biopterin. Addition of exogenous H 4 biopterin increased L-citrulline formation with an EC 50 of ϳ0.2 M, a value comparable with those of human eNOS expressed in a baculovirus system (44) and bovine eNOS in E. coli (21). Also, the potency of 4-amino-H 4 biopterin to antagonize the effect of H 4 biopterin was similar to reports on recombinant bovine eNOS (48).
In contrast to H 4 biopterin-free nNOS, which was shown to have the heme iron in the low spin state (9), the absorbance spectrum of the recombinant H 4 biopterin-free eNOS exhibited a maximum at 398 nm reflecting a high spin state of the heme iron. Similar to our data obtained with human eNOS, Martá sek 2 B. Hemmens and B. Mayer, unpublished observations. et al. (49) also reported that the heme iron of H 4 biopterin-free bovine eNOS was predominantly in a high spin state. These data suggest that in contrast to nNOS, eNOS does not require H 4 biopterin or L-arginine for the transition of the heme iron from low spin to high spin. Gel filtration chromatography of the native protein suggested that dimerization is independent of H 4 biopterin and thus confirms previous studies performed with human eNOS (37,50). Thus the enzyme needs neither H 4 biopterin (like iNOS) nor Ca 2ϩ /calmodulin (as has been reported for the oxygenase domain of eNOS (50)) for full dimerization under native conditions.

FIG. 4. Effect of H 4 biopterin (A) and 4-amino-H 4 biopterin (B) on enzyme activity of recombinant human eNOS.
Formation of an SDS-resistant dimer has been observed for all NOS isozymes (13-15, 21, 37). In most examples studied, the dimer stabilization was dependent on H 4 biopterin and has been interpreted as a structural correlate of the activation of NOS by the pteridine (13). We found a large increase of SDSresistant dimer of human eNOS because of H 4 biopterin binding. Addition of L-arginine in the presence of H 4 biopterin augmented the formation of SDS-resistant dimer, possibly reflecting inhibition of H 4 biopterin dissociation by L-arginine; L-arginine alone did not stabilize the dimer.
There is an obvious discrepancy between our results and the two studies on eNOS by Rodriguez-Crespo et al. (21,37). In the study on bovine eNOS expressed in E. coli, no increase in SDS-resistant dimer was observed on incubation of pteridinefree enzyme with H 4 biopterin. However, the gel shown in that paper does show an apparent increase in SDS-resistant dimer in a preparation purified in the presence of H 4 biopterin compared with a pteridine-free preparation. In the case of human eNOS purified from E. coli (21) essentially complete formation of SDS-resistant dimer was observed even in the absence of added H 4 biopterin, in nearly complete contrast to our results. The proposal that H 4 biopterin is not needed for dimer stabilization in eNOS was further supported by the finding of identical protein conformations in the H 4 biopterin-containing and H 4 biopterin-free crystal structures of the eNOS oxygenase dimer (7). A partial explanation may be provided by our observation that SDS-resistant dimer formation by human eNOS as isolated (12%) was significantly in excess of the H 4 biopterin content (Յ2%). This result was obtained with samples that had been stored at around 2 M. When samples of the same preparations that had been concentrated and stored at around 20 M were used for SDS-PAGE at low temperature, increased levels of SDS-resistant dimer (up to 30%) were observed without added H 4 biopterin (the preincubations for SDS-PAGE at low temperature were done at similar protein concentrations in both cases). SDS-resistant dimer formation by eNOS may thus be induced to some extent by other factors than H 4 biopterin, which does not, however, negate the strong effect of H 4 biopterin.
A possible reason for reduced and therefore H 4 biopterinsensitive dimer stabilization in our eNOS preparations might be a lack of the structural zinc that binds between the subunits. However, our human eNOS contains nearly one zinc ion per NOS dimer, with a Fe/Zn ratio of close to 2:1. The intersubunit ZnS 4 center in the heme domain of bovine eNOS (7,24) was first discovered just a year ago. The contrast to the previously determined structure of the iNOS oxygenase domain, in which no zinc but rather an intersubunit disulfide bond was found at this position (23), raised the question of whether bound zinc is a native feature of all NOSs or depends in some way on expression conditions or the expression system. Here, we have presented the first evidence known to us on incorporation of zinc into NOS in a eukaryotic expression system. Combining gel filtration with ICP-MS as an on-line detection method provides particularly robust results because it eliminates the problem of accurately estimating and subtracting non-protein-bound metals. Thus, bound zinc seems very likely to be a native feature of at least the constitutive isozymes of NOS.
In the presence of suboptimal concentrations of L-arginine and H 4 biopterin nNOS is able to catalyze uncoupled NADPH oxidation with release of O 2 . from the heme site, at much higher rates than eNOS and iNOS. Interestingly, the uncoupled NADPH oxidase activity of the human eNOS was considerably higher than that of bovine eNOS expressed in baculovirusinfected insect cells (15). Another striking difference between these two species of eNOS is that the rate of NADPH oxidation of the human isozyme was stimulated to 342 nmol ϫ mg Ϫ1 ϫ min Ϫ1 by the addition of L-arginine in the absence of H 4 biopterin. This shows that the enzyme has the potential to produce significant amounts of superoxide if cellular concentrations of H 4 biopterin, but not L-arginine, are depleted. Addition of saturating H 4 biopterin did not cause a further increase in NADPH oxidation rate but rather simply switched the outcome of the reaction to tightly coupled L-citrulline formation. To get more evidence for the ability of eNOS to produce superoxide, electron paramagnetic resonance spin trapping experiments directly measuring the O 2 . generation by eNOS were performed (45,51 In contrast to our results, however, superoxide production was not affected by L-arginine and completely inhibited by H 4 biopterin. One reason for this discrepancy may be a direct interaction of the spin trap with the enzyme or an interference of spin traps with reactions downstream of product formation, for example the prevention of peroxynitrite formation because of superoxide scavenging. However, it cannot be excluded that the differences between bovine and human eNOS may result from the use of different expression systems rather than reflecting species differences. The results presented in our study demonstrate that P. pastoris represents a convenient and efficient system for the expression of human eNOS. The discovery of one zinc/dimer in a NOS expressed in a eukaryotic system suggests that the intersubunit zinc site may be a structural feature of all NOS enzymes. Characterization of the recombinant enzyme revealed that human eNOS apparently differs from the bovine isoform in that the uncoupled NADPH oxidation activity is higher, and H 4 biopterin is not required for maximal NADPH oxidation. This could explain the beneficial effects of H 4 biopterin supplementation in endothelial dysfunction in humans, which would thus suppress eNOS-mediated superoxide formation (52). Uncoupling of the NOS reaction in endothelial cells was also suggested by Cosentino and Katusic (53), who reported that H 4 biopterin depletion results in oxidant accumulation and endothelial dysfunction in coronary arterial vessels. From our results we confirm that absence of H 4 biopterin allows eNOS to perform uncoupled oxygen reduction, but based on the increased NADPH oxidation in the presence of L-arginine we suggest that this reaction is controlled by switching to tightly coupled citrulline formation, if H 4 biopterin is present. Thus, eNOS-mediated uncoupled oxygen reduction is a feature of the enzyme that may be physiologically important in understanding the mechanism of endothelial dysfunction.