Expression and immunoaffinity purification of human inducible nitric-oxide synthase. Inhibition studies with 2-amino-5,6-dihydro-4H-1,3-thiazine.

Recombinant human inducible nitric-oxide synthase (rH-iNOS) was expressed in the baculovirus system and purified by a novel immunoaffinity column. rH-iNOS and its native counterpart from cytokine-stimulated primary hepatocytes exhibited similar molecular mass of 130 kDa on SDS-polyacrylamide gel electrophoresis, recognition by antipeptide antibodies, specific activities, and IC50 values for inhibitors. The active dimeric form exhibited a specific activity range of 114-260 nmol/min/mg at 37°C and contained 1.15 ± 0.04 mol of calmodulin/monomer. The enzyme exhibited a Soret λmax at 396 nm with a shoulder at 460 nm and contained 0.28-0.64 mol of heme/monomer. Dithionite reduction under CO yielded an absorbance maximum at 446 nm, indicating a P450-type heme. Imidazole induced a type II difference spectrum, reversible by L-Arg. 2-Amino-5,6-dihydro-4H-1,3-thiazine (ADT) was competitive versus L-Arg (Ki = 22.6 ± 1.9 nM), reversed the type II difference spectrum induced by imidazole (Kd = 17.7 nM), and altered the CO-ferrous absorbance of rH-iNOS. L-Arg did not perturb the CO-ferrous adduct directly, but it partially reversed the ADT-induced absorbance shift, indicating that both bind similarly to the protein but interact differently with the heme.

Nitric oxide (NO) is involved in the regulation of diverse biological functions (for reviews see Refs. [1][2][3]. Three nitricoxide synthases (NOS, 1 EC 1.14.13.39), termed neuronal NOS (n-cNOS or NOS1), endothelial cell NOS (ec-cNOS or NOS3), and inducible NOS (iNOS or NOS2), are capable of catalyzing the production of NO, citrulline, and NADP ϩ from Arg, molecular oxygen, and NADPH. Each isoform consists of an aminoterminal heme domain that binds the cofactor tetrahydrobiopterin (BH 4 ) and the substrate L-Arg, a consensus calmodulin (CaM)-binding domain, and a carboxyl-terminal reductase domain that binds NADPH as well as the flavins FAD and FMN (4 -8). Spectral studies have demonstrated that NOS inhibitors can interact with the heme domain (7)(8)(9). For n-cNOS, the binding of CaM has been demonstrated to facilitate the transfer of electrons from the reductase to the heme domain (9,10). Physical evidence suggests that the active form of the enzymes is a dimer (11,12).
n-cNOS and ec-cNOS are constitutively expressed under normal conditions, and their activities are regulated by the binding of CaM at elevated intracellular Ca 2ϩ concentrations (1)(2)(3). The expression of iNOS can be induced by inflammatory stimuli in a wide variety of cell types, including macrophages, hepatocytes, chondrocytes, and smooth muscle cells (13)(14)(15)(16). Unlike the cNOS isozymes, the activity of iNOS is reported to be independent of elevations in intracellular Ca 2ϩ concentrations (17). For murine iNOS, it has been demonstrated that CaM tightly binds the enzyme even at very low Ca 2ϩ concentrations (18). Therefore, once expressed, iNOS is believed to generate sustained levels of NO in vivo. The sustained generation of NO by iNOS may contribute to the pathology of a number of inflammatory diseases, e.g. septic shock, inflammatory arthritis, type I diabetes, and inflammatory bowel disease (1,3). Potent, isozyme-specific NOS inhibitors will aid in defining the role of iNOS in vivo and perhaps provide effective anti-inflammatory agents for use in human disease.
A number of NOS inhibitors have been characterized kinetically with respect to rodent iNOS (19 -23). However, rodent and human iNOS differ in several respects. For example, unlike cNOS enzymes, which are 90 -95% identical among mammalian species, the amino acid identity of murine and human iNOS is approximately 80% (14); human iNOS is significantly more sensitive to EGTA inhibition than murine iNOS (14,24); and the inhibitor, aminoguanidine, has been reported to be 20-fold more active against murine iNOS as compared with human iNOS (25). Since sources of native human iNOS are limited, the expression of a recombinant human iNOS that faithfully represents its native counterpart is essential for the characterization of inhibitors intended for clinical use.
To provide an abundant source of human iNOS, conditions were determined in the baculovirus system for expressing rH-iNOS with properties similar to its native counterpart. This system differs from other reports of rH-iNOS expression (26,27) in that the enzyme is provided an endogenous source of the key cofactor BH 4 and does not require exposure of the enzyme to exogenous inhibitors. Milligram quantities of highly stable rH-iNOS were purified by a novel immunoaffinity purification protocol that avoided the use of NADPH. The immunoaffinitypurified rH-iNOS was characterized and used to probe the mechanism of inhibition of a potent cyclic isothiourea.
2-Amino-5,6-dihydro-4H-1,3-thiazine (ADT) was synthesized by the method of Schoberl (28) but using a different protocol for removal of KBr and crystallization of ADT. After the removal of activated charcoal by filtration through a pad of Celite, the filtrate was concentrated to dryness in vacuo. The residue was evaporated to dryness once more from methanol. The residual solid was treated with 60 ml of methanol, filtered to remove Kbr, and then reduced to approximately 20 ml by evaporation in vacuo and allowed to cool to room temperature. The solution was diluted with 50 ml of ethyl acetate, and upon standing, crystals were formed, which were filtered, washed with ethyl acetate, and dried to isolate 3.2 g of ADT. An additional 0.97 g of ADT was obtained from the filtrate by repeating the concentration and cooling process.
Cell Culture-Spodoptera frugiperda (Sf9) insect cells were maintained between 5 ϫ 10 5 and 2 ϫ 10 6 cells/ml in spinner cultures at 27°C in Grace's medium (Invitrogen, San Diego, CA) supplemented with 10% fetal calf serum (Clonetech, Palo Alto, CA) and 50 g/ml gentamycin (Life Technologies, Inc.). The isolation, culturing, and induction of primary human hepatocytes or RAW 264.7 cells for the generation of native human iNOS or murine iNOS has been described previously (14,17). The large scale culture of primary human hepatocytes was modified to accommodate the use of Nunc (Naperville, IL) 6000-cm 2 cell factories. The hepatocytes were harvested 12 h postinduction by trypsinization and pelleting by low speed centrifugation. The hepatocyte cell pellet was resuspended in lysis buffer (described below) and subjected to three cycles of freeze-thawing. The lysate was clarified by ultracentrifugation (100,000 ϫ g) for 30 min.
Expression of Recombinant Human NOS Proteins in the Baculovirus System-For rH-iNOS, an XbaI/AflII restriction fragment of the human hepatocyte iNOS cDNA was blunt-end-ligated into the BamHI site of the transfer vector pVL941. Recombinant virus was generated by transfecting this construct into Sf9 insect cells using the BaculoGold transfection kit (Pharmingen, San Diego, CA). Recombinant virus was isolated and titrated by limiting dilution. Infected cells expressing rH-iNOS were identified by diaphorase staining (29). Briefly, medium was gently aspirated from the cells, which were then stained with 0.2 mM nitro blue tetrazolium in the presence of 0.2 mM NADPH, 0.2 M Tris-HCl, pH 7.2, and 0.2% Triton X-100. Large scale viral stocks were generated by sequential low multiplicity infections (multiplicity of infection of 0.01-0.1) with resulting titers ranging from 2 ϫ 10 7 to 8 ϫ 10 7 infectious particles/ml.
For the generation of rH-iNOS protein, Sf9 cells (1 ϫ 10 6 cells/ml) were infected at a multiplicity of infection of 4. Nitrite in the conditioned medium was measured at 540 nm with the Griess reagent (30), using sodium nitrite as a standard. For large scale growth, the cells were grown in 40-liter bioreactors at Kemp Biotechnologies, Inc. (Frederick, MD), maintained at 50% dissolved oxygen, and agitated at a rate of 70 rpm. Two h postinfection, ␦-aminolevulinic acid was added to a final concentration of 8 M, which increased the expression of active rH-iNOS by approximately 2-fold. Unless otherwise specified, cells were harvested at 34 h postinfection, pelleted, resuspended in lysis buffer (20 mM TES, pH 7.4, 1 mM dithiothreitol, 10% glycerol, 50 M BH 4 , 50 M FAD, 50 M FMN, 25 g/ml each of antipain, aprotinin, leupeptin, and chymostatin, 10 g/ml pepstatin A, 100 M phenylmethylsulfonyl fluoride, and 50 M phenanthroline) and homogenized. The lysate was clarified by ultracentrifugation (100,000 ϫ g) for 30 min. The resultant S-100 supernatant was filtered through successive filters to 0.45 , and frozen at Ϫ80°C. The crude S-100 lysate was stable when stored at Ϫ80°C for several months or at 4°C for several hours.
For the expression of recombinant human ec-cNOS, an EcoRI fragment containing the full-length human ec-cNOS cDNA (kindly provided by Dr. Ken Block, Harvard; Ref. 31) was blunt-ended and cloned into the blunt-ended BamHI site of pVL941. The generation of recombinant baculovirus and rH-ecNOS protein was performed similarly to rH-iNOS, except that the lysis buffer contained 10 mM CHAPS and 100 mM NaCl.
For the expression of recombinant human n-cNOS, the full-length cDNA was cloned by reverse transcription-PCR from human cerebellar mRNA, using oligonucleotide primers based on the published sequence (32). An EcoRI fragment containing the full-length cDNA was cloned into the EcoRI site of pVL1393 (Pharmingen, San Diego, CA). The generation of recombinant baculovirus and recombinant human n-cNOS protein was performed similarly to rH-iNOS.  (2)(3)(4)(5). Reactions were incubated at either room temperature or 37°C; aliquots were transferred into 100 l of stop solution (0.02 M sodium citrate, pH 2.2) and analyzed by HPLC. The HPLC separation has been described previously (33,34). The procedure was also performed as modified; a Polypore-SP (sulfopropyl) cation exchange column (2.1 ϫ 30 mm) from Applied Biosystems (Foster City, CA) was used for the separation of the analytes using Buffer A (0.02 M sodium citrate, pH 2.2, with 5% acetonitrile) and Buffer B (0.2 M sodium citrate, pH 3.0, with 5% acetonitrile) with a flow rate of 0.75 ml/min and under the following elution conditions: 0 -1 min, 15% B; 1.1-2.5 min, 90% B; 2.6 -4.5 min, 15% B. Injections were made every 4.5 min. The radioactivity detector was programmed to deliver a 1:3 (v/v) ratio of effluent to scintillation liquid.
Enzyme activity in crude recombinant enzyme preparations was determined after the removal of endogenous L-Arg by spin-filtration through Bio-Rad P-30 resin, preequilibrated in lysis buffer. In crude native human hepatocyte preparations, activity was determined in the presence of 60 mM Val to prevent interference due to endogenous arginase.
For K m determination, the L-Arg concentration was varied from 0.5 to 40 M. The data were fit to a hyperbolic function using Sigma Plot (version 4.17, Jandel Scientific, Cotra Madera, CA), and values were reported with standard errors. For IC 50 measurements, enzyme activity was monitored under the standard assay protocol at 1 or 5 M L-Arg in the presence of varying concentrations of inhibitor.
Generation of Antipeptide Antibodies-Three peptides were synthesized with the following sequences: immunogen peptide A (Cys-Arg-Nle-Orn-Ser-Leu-Glu-Met-Ser-Ala-Leu), probe peptide B (Tyr-Arg-Ala-Ser-Leu-Glu-Met-Ser-Ala-Leu), and consensus peptide C (Arg-Cys-Asn-Asp-Thr-Pro-Val-Phe-His-Glu-Met-Leu-Asn). Peptide A contained the 7-residue carboxyl terminus of human iNOS plus a multipurpose tetrapeptide extension (Cys-Arg-Nle-Orn). The Cys was used to couple the peptide to thyroglobulin using the sulfo-m-maleimidobenzoyl-N-hydroxysulfosuccinimide ester technique (35). The incorporation of Arg and Orn was intended to improve solubility, while norleucine was used to determine coupling efficiency by amino acid analysis. Polyclonal antibody production using this peptide conjugate was undertaken in rabbits by Research and Diagnostics Antibodies (Berkeley, CA) as described previously (36). Peptide B, which was used for the specific elution of the enzyme from the immunoaffinity column, also contained the 7-residue carboxyl terminus of human iNOS plus an amino-terminal tripeptide, Tyr-Arg-Ala. Peptide C was selected from a region of high amino acid sequence identity (residues 695-705 in rat n-cNOS) among all NOS isoforms and was used as an immunogen to generate a common antibody. The acetylated peptide B was prepared by dissolving the peptide in 1.0 ml of a 70:30 mixture of MeOH/acetic anhydride (v/v), incubated at 25°C for 10 min, diluted with an equal volume of H 2 O and evaporated to dryness. Reconstitution in H 2 O and evaporation was repeated twice to remove excess acetic acid. The resulting powder was dissolved in 0.075 M phosphate-buffered saline (PBS)-azide, pH 7.0.
Competitive Radioimmunoassay (RIA)-Peptide B was radioiodinated with Na 125 I by the iodogen method (Pierce) and purified by reverse phase (RP)-HPLC to yield a probe with a specific activity of 2,000 Ci/mmol. The antipeptide A antiserum is specific for iNOS and does not detect recombinant human ec-cNOS or recombinant human n-cNOS on immunoblots (data not shown). 100 l of the antiserum at a dilution of 1:100,000 was mixed with 100 l of buffer, standard, or an unknown sample and incubated overnight at 4°C. Prior to mixing with the antiserum, the samples were first boiled in 1 ϫ Laemmli buffer to denature proteinases capable of degrading the exogenously added peptide. 100 l of labeled probe was added and incubated overnight at 4°C. The peptide bound to the antiserum was then precipitated with 100 l of an anti-rabbit IgG serum at a 1:15 dilution for 90 min at room temperature. After collecting the precipitate by centrifugation the pellets were analyzed in a ␥-counter. The results were compared with a standard displacement curve using cold peptide B. Laemmli (37). Samples were visualized by either Coomassie Blue or silver staining or transferred to 0.45 M nitrocellulose for immunoblot analysis. The nitrocellulose membranes were blocked with 3% casein and incubated with primary anti-peptide A antibody at a 1:20,000 dilution, followed by an alkaline phosphataseconjugated anti-rabbit IgG-F c , and visualized with 5-bromo-4-chloro-3indolyl phosphate/nitro blue tetrazolium. Crude protein concentrations were determined using the method of Bradford (38) with bovine serum albumin as a standard. The concentration of purified rH-iNOS was determined either by the method of Bradford or by amino acid analysis, assuming a 1:1 stoichiometric complex of CaM and rH-iNOS. For amino acid analysis, samples were subjected to vapor phase hydrolysis for 24 and 72 h and analyzed on a Beckman 6300 analyzer using postcolumn derivatization with ninhydrin.

SDS-PAGE Analysis and Protein Concentration-SDS-PAGE was performed according to
Immunoaffinity Column Construction-18 ml of high titer antiserum (1:400,000) were diluted with an equal volume of 0.075 M PBS with 0.02% NaN 3 . Three ml of bovine thyroglobulin (3 mg/ml) in PBS-azide were added and gently mixed, and the solution was allowed to stand for 18 h at 4°C. Immune complexes were removed by centrifugation at 3000 rpm for 10 min. The supernatant was added to a 100-ml bed of Protein A-agarose packed in a glass column, capped, slurried and rotated end over end for 30 min at room temperature. The column was washed with 10 volumes of PBS-azide and mixed with 45 ml containing 100 mol of N-acetylated peptide B for 30 min. The column was drained and washed with 2 volumes of coupling buffer (0.2 M triethanolamine, pH 8.2, with 0.02% NaN 3 ; ImmunoPure IgG Orientation Kit, Pierce). 200 mg of dimethylpimelimidate in 50 ml of coupling buffer were added, and the column was resuspended and rotated end over end for 90 min at 25°C. The column was drained, and 50 ml of blocking buffer (0.2 M ethanolamine, pH 8.2, with 0.02% NaN 3 ) was added and mixed for an additional 10 min. The column was drained and then stripped of noncross-linked proteins using 0.2 M sodium citrate, pH 3.0. The final wash and storage consisted of PBS-azide.
Immunopurification of rH-iNOS-The entire procedure was performed at 4°C. A typical purification involved infusing 700 -900 ml of crude S-100 lysate from a 40-liter Sf9 infection onto a 75-ml immunoaffinity column using a Hamilton syringe pump at a flow rate of 3 ml/min. The wash steps consisted sequentially of 8 column volumes of the running buffer (20 mM TES, pH 7.4, 1 mM CHAPS, 100 M BH 4 , 0.1 mM dithiothreitol, and 0.25 M NaCl) and 2 volumes of the running buffer containing 0.1 M NaCl. For each elution, 30 ml of a 25 M solution of peptide B dissolved in running buffer with 0.1 M NaCl were added, and the column was resuspended, capped, and rotated end over end for 1.5 h. The column was drained and washed with an additional 20 ml of the peptide solution. The elution procedure was repeated two or three times, and all eluates were pooled. For subsequent purifications, the column was regenerated by stripping the bound peptide B using 0.1 M sodium citrate, pH 3.0. Columns stripped in this manner could be reused more than 10 times.
DEAE and Gel Filtration Chromatography-To concentrate the enzyme and to remove peptide B, the pooled eluates from the immunoaffinity column were diluted with 3 volumes of running buffer containing 0.01 M NaCl. The diluted sample was then pumped (1 ml/min) onto 0.5 ϫ 10-cm glass columns, each packed with 1 ml of TosoHaas TSK-GEL DEAE-5PW ion exchange resin (Analytical Sales and Service, Mahwah, NJ). Gradient experiments indicated that the enzyme eluted at 0.175 M NaCl. In all subsequent runs, the enzyme was eluted isocratically with 0.2 M NaCl. Gel filtration chromatography was performed according to a previously described method (11) on a TosoHaas TSK-GEL GMPW column (60 cm x 7.5 mm inner diameter) in running buffer containing 0.2 M NaCl at a flow rate of 0.4 ml/min.
RP-HPLC Separation-Samples for mass spectrometry, microsequencing, amino acid analyses, and stoichiometry determinations were chromatographed using an ABI 130A HPLC system with a 2.1 ϫ 100-mm C4 column maintained at 52°C. LC-ESI MS analysis was performed using a Finnigan TSQ-700 LC-ESI mass spectrometer (San Jose, CA), as described previously (40). A tryptic digest of rH-iNOS was separated on a C 18 reverse phase column (2.1 ϫ 100 mm) at a flow rate of 50 l/min with 0.075% aqueous trifluoroacetic acid and a gradient of 2-60% acetonitrile over 40 min. The effluent was fed directly to the electrospray interface of a Finnigan MAT TSQ-700 quadrupole mass spectrometer (41).
Cofactor Analysis-The amount of insect CaM (42) complexed with rH-iNOS in the starting material was determined by first separating CaM from rH-iNOS by RP-HPLC. The isolated insect CaM peak was hydrolyzed for amino acid analysis. Values for the amino acids Asp, Asn, Glu, Ala, Val, Ile, and Leu were determined and corrected for the experimentally derived recovery of (i) bovine CaM from RP-HPLC (0.87) and (ii) spiked norleucine from the amino acid hydrolysate (0.92). These amino acids were chosen after it was determined experimentally that their levels remained stable between 24 and 72 h of hydrolysis. Based on the known amino acid composition of Drosophila melanogaster CaM, the values for each of these amino acids were used separately to calculate the amount of CaM. The seven calculated values for CaM were then averaged, and the mean value was used to determine the stoichiometric ratio with rH-iNOS. The values for these amino acids were also subtracted from the corresponding amino acid values obtained from direct hydrolysis of the starting material. Using the known amino acid composition of rH-iNOS, each of these subtracted values was then used to separately calculate the amount of rH-iNOS monomer in the complex. The mean value was determined and used to calculate the stoichiometric ratio of rH-iNOS to CaM.
For pterin analysis, an aliquot of the immunoaffinity-purified enzyme was prepared using a DEAE-5PW column equilibrated in buffers free of BH 4 . Pterin content was measured as described previously (43).
Heme content was quantified using two methods (44,45). First, 200 l of the sample (ϳ1 mg/ml) were dissolved in 300 l of 0.1 M KOH, 3.7 M pyridine followed by the addition of several grains of dithionite. The pyridine hemochromogen absorbance was then measured at 556 nm. The absorbance was blanked against an identical sample containing no protein and normalized at 700 nm. The heme concentration was calculated using ⑀ 556 ϭ 34.7 mM Ϫ1 cm Ϫ1 . In the second method, the COferrous absorbance spectrum was generated by purging the sample in a septum-sealed cuvette with CO gas for 10 min; 10 l of a 300 mM sodium dithionite solution was added, and then the sample was purged again with CO gas for 5 min. The concentration of the CO-ferrous species was calculated using ⑀ 446 -476 ϭ 75 mM Ϫ1 cm Ϫ1 (45). The stoichiometry of heme was calculated using a subunit molecular mass of 147,798 Da for a 1:1 complex of iNOS to CaM.
Optical Spectroscopy-All spectra were recorded using a HP8452A diode-array UV/visible spectrometer (Hewlett Packard, Palo Alto, CA) with a thermostatted, multicell-transport cell holder attached to a circulating water bath. Sample spectra were blanked against 20 mM TES, pH 7.5, 4 M BH 4 , and the absolute absorbance between 190 and 800 nm was recorded at 15°C. The imidazole-ferric low spin spectrum of rH-iNOS was generated by the addition of 2 mM imidazole unless otherwise specified. Subsequently, 2 mM L-Arg was added to generate the high spin heme spectrum.
Titration of ADT (for structure, see Fig. 4) against the ferric form of rH-iNOS was accomplished by the stepwise addition of the appropriate stock solution via a syringe at 15°C. The absolute spectrum was recorded at each concentration and normalized at 700 nm. The final added volume was no more than 3% of the initial sample volume. The difference spectrum was generated by the subtraction of the resultant spectrum at each concentration of substrate or inhibitor from the initial spectrum. By plotting the changes in absorbance at the indicated wavelengths versus the concentration of substrate or inhibitor and fitting the data to Equation 1, the dissociation constant (K d ) can be calculated.
In Equation 1, y is the change in absorbance, x is the concentration of added inhibitor, a is the maximum change in absorbance at an infinite concentration of inhibitor, and b is the dissociation constant, K d . The addition of ADT to the CO-ferrous adduct of rH-iNOS was the same as described above for the ferric form of the enzyme but in a septum-sealed cuvette under argon. All transfers were done via a gastight syringe. The ADT solution (1 mM) was rendered anaerobic by bubbling the solution with argon for 10 min in a septum-sealed vial (Pierce).

Expression of Active rH-iNOS-rH-iNOS was expressed in
Sf9 cells using a recombinant baculovirus system. Time course studies indicated that total iNOS activity in S-100 lysates, as measured by conversion of L-[ 3 H]Arg to L-[ 3 H]citrulline, increased from 24 to 40 h postinfection and then declined (Fig.  1A). This activity was inhibited by NMA and was not detected in lysates from uninfected cells or cells infected with wild-type baculovirus (data not shown). Activity values in crude lysates ranged from 0.1 to 0.5 nmol/min/mg of total protein at room temperature. The amount of rH-iNOS protein in the lysates peaked at 43 h postinfection as determined by RIA (Fig. 1A), Coomassie Blue-stained SDS-PAGE, and immunoblot analysis with four NOS-specific antipeptide antibodies, including one that recognizes a 10-amino acid epitope that is highly conserved in all NOS proteins (amino acids 472-481; peptide C) and a second that recognizes the 7-residue carboxyl terminus unique to human iNOS (peptide A, data not shown). The progressive decrease in the ratio of enzyme activity to rH-iNOS protein level, as determined by RIA, suggested a decline in enzyme specific activity with time postinfection (Fig. 1B).
To further examine enzyme specific activity, S-100 lysates were prepared from cytokine-stimulated murine RAW cells or human hepatocytes, as well as Sf9 cells harvested at different times postinfection. Each sample was assayed for iNOS activity. Aliquots from each sample containing equal amounts of activity (1 pmol citrulline/min) were examined by immunoblot analysis using an antiserum raised to peptide C. Again, an increase was observed in the amount of immunoreactive rH-iNOS per unit of activity as a function of time postinfection (Fig. 1C), confirming the decline in enzyme specific activity as demonstrated by the ratio of enzyme activity per RIA unit (Fig.  1B). Samples harvested 22-27 h postinfection appeared to have specific activities similar to native human iNOS and murine iNOS preparations (Fig. 1C). were similar for partially purified rH-iNOS and native human iNOS (respectively), indicating that the active sites of both enzymes interacted similarly with each inhibitor. As reported previously (25), aminoguanidine also inhibited rH-iNOS and native human iNOS much less potently than murine iNOS (IC 50 ϭ 6.6 Ϯ 1.0 M). We tested the effect of EGTA on the recombinant and native enzymes at concentrations ranging from 0.03 to 2 mM. Both rH-iNOS and native human iNOS were inhibited up to 50 -70% by EGTA, while native murine iNOS was inhibited Ͻ10% at 2 mM EGTA. With each enzyme preparation, inhibition by EGTA was completely reversed by the addition of excess calcium (data not shown).

Comparison of Recombinant and Native Human iNOS-The
Purification of rH-iNOS by Immunoaffinity Chromatography-ADP-purified rH-iNOS preparations eluted and maintained in 5 mM NADPH were unstable, losing up to 50% of their activity following a 2-h incubation on ice. These preparations could be stabilized for up to 120 min at 37°C by dilution into assay buffer containing only 250 M NADPH or by removal of NADPH using sulfopropyl, DEAE, or gel filtration chromatography (data not shown). To isolate stable, concentrated rH-iNOS for spectroscopic studies without prior exposure to In B, the ratio of enzyme activity to RIA units was used as a measure of specific activity in the S-100 lysates. No ratio was calculated for the earliest time point, since the RIA value was below the limit of detection. In C, the relative specific activity of rH-iNOS was estimated by immunoblot analysis using the peptide C antipeptide antibody. Samples of native murine iNOS, native human iNOS, or rH-iNOS, harvested at the times indicated and each containing equal amounts (1 pmol/min) of iNOS activity, were loaded. NADPH, a novel immunoaffinity purification protocol was developed using a polyclonal antiserum generated against peptide A. The critical aspect of the column construction was that during the cross-linking process, the antigen-combining sites were protected from chemical modification by preincubation with acetylated peptide B. After removal of the protective peptide, the column specifically bound 2.7 nmol of peptide B/ml of resin, as quantified by RP-HPLC. In contrast, an unrelated peptide (residues 111-123 of the human interleukin-1␤ precursor) was not bound.
After passage of the filtered S-100 crude Sf9 lysate over the immunoaffinity column and extensive washing, the enzyme was eluted by incubation of the column with a 25 M solution of peptide B (Fig. 2). Similar results were obtained with higher concentrations of the peptide (50 -500 M). As shown in Table I, three 90-min incubations resulted in the elution of 21% of the applied activity with an additional 4% recovered after an overnight incubation. Recoveries from other purification runs ranged from 16 to 32%. Coomassie Blue-stained SDS-PAGE analysis showed a 130-kDa band in all of the specifically eluted fractions (Fig. 2, inset).
The peptide was removed, and the enzyme was concentrated approximately 200-fold by DEAE chromatography and isocratic elution with 0.2 M NaCl. This step provided highly concentrated enzyme solutions (0.75-3 mg/ml), with 60 -90% recovery of activity. Peptide B eluted with 0.5 M NaCl or 0.1 M sodium citrate, pH 3.0. Immunoaffinity-purified enzyme was stable for greater than 4 h at room temperature and greater than 18 h on ice. In contrast, 50% of the activity was lost following a 2-h incubation with 5 mM NADPH, the concentration used to elute the enzyme from ADP-Sepharose.
The specific activities of several immunoaffinity-purified rH-iNOS preparations were determined using two methods to measure protein concentration. By the method of Bradford, using bovine serum albumin as a standard, the specific activities ranged from 94 to 199 nmol/min/mg at 37°C with an average of 131 Ϯ 41 (S.D., n ϭ 6). These values were in good agreement with those obtained by amino acid analysis, which ranged from 114 to 260 nmol/min/mg at 37°C with an average of 164 Ϯ 57 (S.D., n ϭ 5). The K m determined for L-Arg was 2.30 Ϯ 0.25 M. In addition, two preparations of rH-iNOS with specific activities that differed by 3.3-fold gave indistinguishable IC 50 values for N G -methyl-L-Arg: 2.24 Ϯ 0.22 and 2.38 Ϯ 0.17 M, using 1 M L-Arg as substrate. In summary, rH-iNOS prepared by the immunoaffinity purification protocol was suitable for direct use in all subsequent enzyme analyses.
Gel Filtration Chromatography-Purified rH-iNOS isolated from all lysates harvested between 28 and 34 h postinfection migrated with a mass of 260 kDa by gel filtration chromatography. This protein peak corresponded with the peak of enzyme activity and confirmed the dimeric nature of active rH-iNOS (data not shown).
Physical Characterization of Purified rH-iNOS by Mass Spectrometry, Peptide Mapping, and Microsequencing-rH-iNOS separated by RP-HPLC yielded three major peaks (Fig.  3A). Each peak was analyzed by MALDI-TOF MS. Peak 1 afforded abundant molecular ions at m/z 615.4, corresponding to the protonated molecular mass of the monomeric form of CHAPS (data not shown). Peak 2 yielded singly (m/z 16,723) and doubly (m/z 8,362) protonated species (Fig. 3B) in excellent agreement with the predicted molecular mass for D. melanogaster CaM, 16,721.5 atomic mass units (42). Analysis of peak 3 showed a spectrum (Fig. 3C) with an abundance of singly charged molecular ions at m/z 131,673.4. This value is within 0.4% of the predicted mass of the primary sequence of human iNOS (131,160 atomic mass units) and is well within the experimental accuracy of the method for proteins with this mass range (0.1-0.5%). In addition, aliquots of peaks 2 and 3 were digested with trypsin, separated by RP-HPLC, analyzed by LC-ESI MS, and compared against a computer-generated tryptic digest of insect CaM or human iNOS. All of the predicted peptides for D. melanogaster CaM and 90% of the predicted peptides for iNOS were identified by this method (data not shown). The amino terminus of the protein could not be identified by LC-ESI MS or Edman sequencing, strongly suggesting that it is blocked.
Cofactor Analysis of Purified rH-iNOS-The stoichiometry of CaM versus rH-iNOS was calculated following RP-HPLC separation as described under "Experimental Procedures." The stoichiometry was calculated to be 1.15 mol of CaM/mol of rH-iNOS monomer (range, 1.12-1.19 Ϯ 0.04 (S.D., n ϭ 4)).
Following removal of excess BH 4 by DEAE chromatography, pterin analysis revealed that the reisolated enzyme was essentially saturated, with 11.6 pmol of BH 4   based on amino acid analysis as an alternative method of protein determination, were 0.28 -0.64 mol of heme/mol of rH-iNOS monomer, with an average of 0.41 (n ϭ 5).
Optical Spectral Characterization-Solutions at concentrations of 1 mg/ml or greater exhibited a yellowish color, consistent with the enzyme having heme and flavin cofactors. The rH-iNOS showed visible absorbance maxima at 278 and 396 nm, with an absorbance shoulder at 460 nm (data not shown). The CO spectrum upon reduction with dithionite yielded an absorbance maximum of 446 nm indicative of a cytochrome P 450 -type heme. Titration with up to 2 mM imidazole produced a type II spectral shift to 430 nm and a trough at 395 nm (data not shown), which was reversed by the addition of 2 mM L-Arg.
Inhibition Studies with ADT-Cyclic isothioureas have been reported to be potent inhibitors of rH-iNOS (46,47). However, mechanistic studies on this class of compounds have been limited due to insufficient quantities of pure human iNOS. Kinetic analysis demonstrated that ADT potently inhibited rH-iNOS with an IC 50 of 24.5 Ϯ 4.5 nM. The inhibition was not time-dependent and was competitive versus L-Arg (Fig. 4). The calculated K i for ADT is 22.6 Ϯ 1.9 nM. The IC 50 values for recombinant human n-cNOS and recombinant human ec-cNOS were 17.6 Ϯ 2.0 and 78.4 Ϯ 4.6 nM, respectively.
Optical difference spectrophotometry allows for the direct analysis of the interactions of ADT with the P 450 -like heme of rH-iNOS (48). The heme of NOS has been shown to exist in two states: high spin and low spin. As isolated, rH-iNOS exists predominately in the high spin state. The addition of ADT to purified rH-iNOS resulted in a spectral perturbation of only the enzyme in the low spin state (ϳ20%), inducing a type I difference spectrum (data not shown). For a quantitative determination of the ADT binding constant, all of the heme were converted to the low spin state by the addition of 1 mM imidazole. The addition of ADT to the low spin enzyme/imidazole complex resulted in a spectral shift from 430 to 395 nm (Fig. 5). This is indicative of a type I difference spectrum and is consistent with ADT being a competitive inhibitor of L-Arg. The plot of ADT concentration added against the magnitude of the spectral change between 395 and 430 nm yielded a saturation curve (Fig. 5, inset). Fit of the data gave an apparent K d of 300 Ϯ 33 nM. Correction of this value for the concentration of imidazole using Equation 2 (49) is as follows,  We next examined the effects of ADT on the CO-ferrous adduct of the enzyme, to further investigate the interactions of ADT with the heme site. The addition of 30 M ADT to the enzyme dramatically affected the CO-ferrous absorbance, by depleting the absorbance at 446 nm and increasing the absorbance between 395 and 430 nm (Fig. 6A). 1 M L-Arg alone did not perturb the absorbance of the CO-ferrous adduct. However, L-Arg (1 mM) partially reversed the ADT-induced spectral change with recovery of the P 450 absorbance (Fig. 6B). DISCUSSION The rH-iNOS described in these studies faithfully represents its native counterpart from stimulated human hepatocytes in each comparative analysis performed: (i) specific activity as estimated by immunoblot analysis (Fig. 1), (ii) M r on SDS-PAGE ( Fig. 1) gel filtration, (iii) IC 50 values for three inhibitors, and (iv) sensitivity to inhibition by EGTA. These results demonstrate that rH-iNOS is suitable for performing detailed spectral and mechanistic studies of inhibitors directed against the native human enzyme.
We previously identified CaM as a tightly bound component of the native iNOS from murine macrophages (18). The amount of CaM bound to rH-iNOS purified in the absence of exogenous calcium, as determined in four enzyme preparations, showed a stoichiometry of 1.15:1. This study confirms that rH-iNOS also binds CaM tightly, and it is the first to demonstrate stoichiometric binding of CaM to the inducible NOS isoform. These findings are in agreement with recently published studies showing that CaM binds peptides derived from the consensus CaM binding domain of murine iNOS with a stoichiometry of 1:1 (24,50). Our analyses also demonstrate that rH-iNOS, lysed in the presence of excess BH 4 , remained essentially replete with that cofactor after subsequent repurification by DEAE chromatography in BH 4 -free buffers, demonstrating the tight binding of this cofactor.
The availability of milligram quantities of purified rH-iNOS has allowed us to begin to determine the mechanism of inhibition of rH-iNOS by ADT, a representative of the cyclic isothiourea class of inhibitors, using kinetic and spectrophotometric methods. This class of compounds contains some of the most potent inhibitors of human iNOS described to date (46,47). A closely related analog of ADT, the 6-methyl derivative, inhibits murine iNOS with an IC 50 of 3.6 nM (46); however, detailed mechanistic studies with this or related compounds have not been reported for murine or human iNOS. Understanding how members of this class of inhibitors interact with rH-iNOS will provide crucial information for the design of potent and selective inhibitors. In the present study, detailed kinetic and spectroscopic analyses indicate that ADT is a competitive inhibitor versus L-Arg. This suggests that ADT can occupy the L-Arg binding site, probably at the guanidinium recognition subsite. However, the binding of ADT is 150-fold stronger than that of L-Arg, despite the lack of an amino acid moiety. It is conceivable that additional interactions are present between the rH-iNOS and parts of the dihydrothiazine ring structure that give rise to the increase in binding energy (ϳ3 kcal). The presence of a sulfur at the 2-position may allow for ligation to the ferric heme, but the optical spectrophotometric data do not support this view. ADT binding to the enzyme is characterized by a type I difference spectrum. Direct interactions of the sulfur with the heme would produce a type II or a modified type II difference spectrum, as observed for thiocitrulline with rat brain cNOS (51), or a split Soret absorbance as observed for isobornylmercaptan with cytochrome P 450 -cam (51).
ADT also perturbs the spectral absorbance of the CO-ferrous adduct of rH-iNOS, whereas L-Arg does not. CO is known to bind at the sixth axial ligand site of the ferrous heme, similar to that of imidazole binding to the ferric heme (4,5,52). This is also the site of dioxygen binding and presumably the site of the first hydroxylation of L-Arg to form N-hydroxyarginine (53). The loss of the P 450 absorbance indicates that CO may be displaced from the heme. It is conceivable that the binding of ADT to the subsite occupied by the guanidinium of L-Arg places it significantly close to the dioxygen site, such that it can displace CO from the sixth axial ligand site. This interpretation is consistent with the capacity of L-Arg to partially reverse the spectral changes induced by ADT, presumably by allowing CO to rebind to the enzyme. Although speculative, the effect of ADT on the CO-ferrous enzyme may indicate that the additional binding energies observed for ADT may be from its interactions with the dioxygen binding site. Further studies are necessary to substantiate this hypothesis.
In summary, we have presented a novel procedure to generate and purify milligram quantities of stable rH-iNOS that is structurally and catalytically similar to the native enzyme. Using this recombinant enzyme, we performed detailed mechanistic studies with a member of the cyclic isothiourea class of inhibitors, ADT. Kinetic and optical spectroscopic analyses of rH-iNOS with ADT have provided additional insights into the interaction of this class of inhibitors with the active site of the enzyme, demonstrating the utility of rH-iNOS expressed and purified by this methodology to characterize the interaction of inhibitors against this potentially key therapeutic target.