Arginine 120 of prostaglandin G/H synthase-1 is required for the inhibition by nonsteroidal anti-inflammatory drugs containing a carboxylic acid moiety.

The therapeutic action of nonsteroidal anti-inflammatory drugs (NSAIDs) is exerted through the inhibition of prostaglandin G/H synthase (PGHS), which is expressed as two isoenzymes, termed PGHS-1 and PGHS-2. From the crystal structure of sheep PGHS-1, it has been proposed that the carboxylic acid group of flurbiprofen is located in a favorable position for interacting with the arginine 120 residue of PGHS-1 (Picot, D., Loll, P. J., and Garavito, R. M. (1994) Nature 367, 243-249). Mutation of this Arg120 residue to Glu was performed and expressed in COS-7 cells using a vaccinia virus expression system. Comparison of microsomal enzyme preparations show that the mutation results in a 20-fold reduction in the specific activity of PGHS-1 and in a 100-fold increase in the apparent Km for arachidonic acid. Indomethacin, flurbiprofen, and ketoprofen, inhibitors of PGHS activity containing a free carboxylic acid group, do not exhibit any inhibitory effects against the activity of PGHS-1(Arg120-->Glu). Diclofenac and meclofenamic acid, other NSAIDs containing a free carboxylic acid group, were 50-100-fold less potent inhibitors of the activity of the mutant as compared with the wild type PGHS. In contrast, the nonacid PGHS inhibitors, 5-bromo-2-(4-fluorophenyl)-3-(4-methylsulfonyl)thiophene (DuP697) and a desbromo-sulfonamide analogue of DuP697 (L-746,483), were both more potent inhibitors of PGHS-1(Arg120-->Glu) than of the wild tyupe PGHS-1. Inhibition of PGHS-1(Arg120-->Glu) was time-dependent for diclofenac and time-independent for DuP697, as observed for the wild type enzyme, indicating that the mutation does not alter the basic mechanism of inhibition. Aspirin is an acid NSAID that inhibits PGHS-1 through a unique covalent acetylation of the enzyme and also showed a reduced rate of inactivation of the mutated enzyme. These data provide biochemical evidence of the importance of the Arg120 residue in PGHS-1 for interaction with arachidonic acid and NSAIDs containing a free carboxylic acid moiety.

enzyme that catalyzes the initial step in their formation is prostaglandin G/H synthase (PGHS) (2). 1 There are two isoforms of human PGHS, a constitutively expressed form termed PGHS-1 and an inducible isoform, termed PGHS-2, which can be induced in cells and tissues by growth factors and cytokines and also in several models of inflammation (3)(4)(5)(6)(7)(8). The PGHScatalyzed reaction consists of the bis-dioxygenation of arachidonic acid to form prostaglandin G 2 and the reduction of this metabolite to prostaglandin H 2 , which then serves as the precursor for a variety of biologically active prostanoids (1,2,9).
NSAIDs are the most widely utilized drugs for the treatment of inflammatory disorders such as arthritis. The major problem in the use of these compounds is their toxicity profile, which includes ulcerogenicity and renal dysfunction and has been attributed mainly to inhibition of the PGHS-1 isoform. Understanding the interaction of NSAIDs with PGHS can provide insight into the development of NSAIDs with enhanced therapeutic potential and a decreased capacity for toxicity. The recent determination of the x-ray crystal structure of sheep PGHS-1 has led to the development of a model for the topography of the NSAID binding site in PGHS-1 (20 -22). Docking flurbiprofen in the long hydrophobic channel of the proposed active site of PGHS shows the carboxylic acid group of flurbiprofen located in a favorable position for interacting with the guanidinium group of Arg 120 . The only other charged residue in the active site cavity and hydrophobic channel is a Glu 524 , which is proposed to form a salt bridge with Arg 120 .
Various studies on PGHS have utilized site-directed mutagenesis to demonstrate a Tyr 385 residue essential for PGHS catalysis (23), essential histidines involved in heme binding of PGHS (24), residues glycosylated on PGHS (25), and the residue on PGHS acetylated by aspirin (15,16). This study reports on the mutagenesis of Arg 120 to glutamate in human PGHS-1 in order to delineate the interaction between NSAIDs and PGHS. Our observations show that PGHS-1(Arg 120 3 Glu) has a decreased affinity for arachidonic acid and is no longer sensitive to inhibition by several NSAIDs containing a carboxylic acid moiety.

MATERIALS AND METHODS
Mutagenesis of PGHS-1 to PGHS-1(Arg 120 3 Glu)-General recombinant DNA techniques were performed as described elsewhere (12,15,26). A 700-base pair BamHI-AccI DNA fragment, encoding the first 220 codons of human PGHS-1, was excised from plasmid pTM1-hPGHS-1 (12) and subcloned into the multiple cloning site of pBluescript (Stratagene, La Jolla, CA), yielding pBS-Bam-Acc-PGHS-1. Mutagenesis of Arg 120 of PGHS-1 to Glu was performed using an oligonucleotide-directed mutagenesis kit (Amersham Corp.), single-stranded DNA generated from pBS-Bam-Acc-PGHS-1, and a 33-mer oligonucleotide (5Ј-CTGGTACTCACAGTGGAATCCAACCTTATCCCC-3Ј). The single amino acid change of arginine residue 120 to glutamate is encoded by the underlined three nucleotide mismatch of the synthesized oligonucleotide. Confirmation of the desired mutation was by dideoxy DNA sequencing using a T7 DNA polymerase sequencing kit (Pharmacia Biotech Inc.). The mutated 700-base pair BamHI-AccI DNA fragment was then subcloned back into pTM1-PGHS-1, resulting in pTM1-PGHS-1(Arg 120 3 Glu). Homologous recombination between the transfer vector pTM1-PGHS-1(Arg 120 3 Glu) and wild type vaccinia virus (VV) was performed following standard protocols elsewhere (12,27) and resulted in the recombinant VV:PGHS-1(Arg 120 3 Glu). All procedures for the infection of COS-7 cells with VV:PGHS-1(Arg 120 3 Glu) and the helper virus VV:TF7-3 and the preparation of microsomal protein fractions from infected cells have been described elsewhere (12). Essentially, monolayer cultures of confluent COS-7 cells were co-infected with either vv:hPGHS-1 or vv:hPGHS-1(Arg 120 Glu) and the helper virus VV:TF7-3, each at a multiplicity of infection of 1. At 24 h postinfection, cells were harvested by scraping, washed twice with phosphate-buffered saline, resuspended in lysis buffer (100 mM Tris, pH 7.4, 10 mM EDTA, 2 mM phenylmethonylsulfonyl fluoride, 2 g/ml leupeptin, 2 g/ml aprotinin, and 2 g/ml soybean trypsin inhibitor), and then disrupted using a microsonic cell disruptor (Cole-Parmer). Samples were centrifuged for 10 min at 1,000 ϫ g at 4°C, and the resulting supernatant was centrifuged for 60 min at 100,000 ϫ g at 4°C. The 100,000 ϫ g microsome fraction was resuspended in 100 mM Tris, pH 7.4, 10 mM EDTA to yield a protein concentration of 1-8 mg/ml. Protein concentrations were determined using the Pierce Coomassie protein assay reagent as described by the manufacturer's instructions.
Assay and Inhibition of Prostaglandin G/H Synthase Activity-Assays for the synthesis of PGE 2 from arachidonic acid by microsomal fractions preparared from COS-7 cells co-infected with VV:PGHS-1(Arg 120 3 Glu) and VV:TF7-3 have been previously described (12,15). Briefly, enzyme assays were performed at protein concentrations of 15 g/ml in a volume of 210 l containing 0.1 M Tris-HCl (pH 7.4), 10 mM EDTA, 1 mM reduced glutathione, 0.5 mM phenol, and 1 M hematin. NSAIDs were dissolved in dimethyl sulfoxide and preincubated with the assay mixture containing the microsomes for 15 min at room temperature. The reaction was initiated by the addition of 5 l of arachidonic acid dissolved in ethanol to obtain a final concentration of 20 M. The reaction was terminated after 3 min by the addition of 20 l of 1 M HCl and neutralized with an equivalent volume of 1 M NaOH. PGE 2 product synthesis was quantitated using a PGE 2 radioimmunoassay (DuPont NEN). Apparent K m values were for PGHS-1 and PGHS-1(Arg 120 3 Glu) was determined from the product of PGE 2 production by the microsomal preparations at a short reaction time of 30 s as a function of arachidonic acid concentration. The K m analysis was performed using a program by J. S. Easterby (Hyperbolic Regression Analysis of Enzyme Kinetic Data) and is presented as double reciprocal plot. The time dependence of inhibition was determined using the assays for PGE 2 production as described, and the inhibitor was preincubated for 0 -30 min with PGHS-1 or PGHS-1(Arg 120 3 Glu) before arachidonic acid addition. Inhibition was plotted as the percentage of PGHS activity remaining at the designated time points in the presence of inhibitor as compared with the control. The inhibitors indomethacin and flurbiprofen were obtained from Cayman Chemical Co., diclofenac was from Sigma, and DuP697 and L-746,483 were synthesized by the Department of Medicinal Chemistry (Merck Frosst, Canada).

Expression of PGHS-1 and PGHS-1(Arg 120 -Glu) in COS-7
Cells-Arg 120 of PGHS was mutated to a glutamate residue to evaluate the importance of this arginine residue in the inhibition of activity of PGHS by acid NSAIDs and nonacid PGHS inhibitors. PGHS-1 and PGHS-1(Arg 120 3 Glu) were expressed in COS-7 cells by co-infection with the helper virus VV:TF7-3 and either of the recombinants, VV: PGHS-1 or VV:PGHS-1(Arg 120 3 Glu). At 24 h postinfection, the level of expression of these PGHS forms was analyzed by preparation of 100,000 ϫ g membrane pellets from the infected cells and immunoblot analysis using an anti-PGHS-1 antiserum (Fig. 1). The level of recombinant protein expression of PGHS-1 and PGHS-1(Arg 120 3 Glu) was equivalent as determined by scanning laser densitometry of the immunoblots (data not shown). The similar levels of recombinant PGHS-1 and PGHS-1(Arg 120 3 Glu) protein in the microsomes permit the measurement of enzymatic activity and the evaluation of inhibitor potency at the same protein concentration without a difference in partitioning of substrate or inhibitors by the microsomal membranes. Similar microsomal preparations of PGHS have been described in the literature for analysis of various mutants of PGHS (15,16,23,25).
Effect of Mutation of Arg 120 on Arachidonic Acid Oxidation by PGHS-Microsomal preparations from VV:PGHS-1-infected COS-7 cells were previously demonstrated to produce primarily PGE 2 when incubated with arachidonic acid (12). The specific FIG. 1. Immunoblot analysis of PGHS-1 and PGHS-1(Arg 120 -Glu) expressed in COS-7 cells utilizing a vaccinia virus expression system. COS-7 cells were infected with the indicated viral constructs at a multiplicity of infection of 1:1, and the cells were harvested after 26 h. Microsomal proteins from these cells were subjected to SDS-polyacrylamide gel electrophoresis on 10% precast Tris-glycine acrylamide gels (Novex) and electrophoretically transferred to nitrocellulose membranes. Immunoblot detection was performed using a PGHS-1 polyclonal antiserum (29) and 125 I-protein A as described previously (28). Total microsomal protein at 2 and 5 g from PGHS-1 expressing cells is shown in lanes 1 and 2, respectively. Total microsomal protein at 2, 5, and 10 g from hPGHS-1(Arg 120 3 Glu) expressing cells is shown in lanes 3, 4, and 5, respectively. The immunoblot was exposed to Kodak XAR-2 film at Ϫ80°C for 4 h. The migration position of molecular mass standards is depicted.
activity of the mutant PGHS-1(Arg 120 3 Glu) was 20-fold lower than that observed for the wild type PGHS-1. Interestingly, the peroxidase activity of PGHS-1(Arg 120 3 Glu) was reduced to a similar extent as the cycloxygenase activity of this mutant as measured with a peroxidase assay using hydrogen peroxide as a substrate and guaiacol as a reducing agent (30). Although the PGE 2 production of the mutant was low, this amount of synthesis was 40-fold higher than that of mock infected COS cells, and product formation was dependent on exogenous arachidonic acid, thus allowing the further characterization of its activity. The time course of PGE 2 product formation over a range of arachidonic acid concentrations was performed for both PGHS-1 and PGHS-1(Arg 120 3 Glu) and showed similar hyperbolic curves with product formation plotted as a function of time. The time course for the mutant contained no detectable lag phase as compared with the wild type, even though the cyclo-oxygenase and peroxidase activity of the mutant PGHS-1 were decreased. Also, 0.1 or 1 M 13-hydroperoxyoctadecanoic acid did not stimulate arachidonic acid-dependent PGE 2 production by either the mutant or wild type enzyme preparation. The production of PGE 2 by microsomal preparations of PGHS-1 and the Arg 120 3 Glu mutant was evaluated as a function of arachidonic acid concentration at a short reaction time of 30 s, and Lineweaver-Burk plots for PGHS-1 and PGHS-1(Arg 120 3 Glu) are shown in Fig. 2. The apparent K m for arachidonic acid of PGHS-1(Arg 120 3 Glu) was 60 M, about 100-fold higher than that of hPGHS-1 (K m ϭ 0.5 M). The same K m values were obtained using a 3-min enzymatic reaction (data not shown). This large shift in the apparent K m suggests that the Arg 120 is involved in binding arachidonic acid but is not absolutely required for catalytic activity. Microsomal assays performed at pH 5.0 resulted in a loss of only 20% of the activity for hPGHS-1(Arg 120 3 Glu), whereas the activity of hPGHS-1 was decreased 20-fold to a level similar to that of the mutant form (data not shown). Although several interpretations can be envisioned for these results, they are consistent with the ionization of the carboxylic acid of arachidonic acid being more important for the binding of substrate to the wild type as opposed to the mutant enzyme.
Inhibition of hPGHS-1 and hPGHS-1(Arg 120 -Glu) by NSAIDs-If NSAIDs with an acidic carboxylic acid moiety, such as indomethacin, flurbiprofen, and diclofenac, interact with the Arg 120 of PGHS-1 through an ionic interaction, then these NSAIDs should show a reduced inhibitory potency against hPGHS-1(Arg 120 3 Glu). Conversely, the potency of PGHS inhibitors that do not contain a carboxylic acid, such as DuP697 (17), should be less affected by the mutation of the arginine residue.
PGHS inhibitors such as indomethacin, flurbiprofen, and diclofenac inhibit this enzyme by an essentially irreversible time-dependent mechanism (17,18). The inhibitory potencies of these NSAIDS were determined using a preincubation time of inhibitor with enzyme to allow the time-dependent inhibition to develop prior to initiation of the reaction with 20 M arachidonic acid. Flurbiprofen, indomethacin, and ketoprofen are potent inhibitors of PGHS-1 with IC 50 values of 3.2, 90, and 60 nM, respectively ( Fig. 3 and Table I). In contrast, no significant inhibition of PGHS-1(Arg 120 3 Glu) at concentrations up to 24 M was observed with indomethacin, flurbiprofen, or ketoprofen. The fenamic acid NSAIDS, diclofenac and meclofenamic acid, retained some inhibitory effects against PGHS-1 ( 120 Arg-Glu), with IC 50 values of 1.5 and 5.0 M for the Arg 120 3 Glu mutant, respectively. Although, these latter NSAIDs inhibited PGHS-1(Arg 120 3 Glu), the potency was decreased 40 -100-fold as compared with the inhibition of wild type PGHS-1. The acid NSAID ibuprofen was also tested and found to inhibit PGHS-1 with an IC 50 of 100 M, whereas no detectable inhibtion of PGHS-1(Arg 120 3 Glu) was observed at 240 M ibuprofen. Clearly, all of the acid NSAIDs tested are significantly more potent for inhibition of the wild type PGHS-1 as compared with the mutant PGHS-1(Arg 120 3 Glu).
The DuP697 compound inhibits PGHS-1 by a competitive reversible inhibitory mechanism (17,18). DuP697 displayed significant potency for inhibition of both PGHS-1 and PGHS-1(Arg 120 3 Glu) with IC 50 values of 1.2 and 0.1 M, respectively (Fig. 4A). Also, a desbromo-sulfonamide analogue of DuP697 (L-746,483) (31) was found to inhibit both PGHS-1 (IC 50 ϭ 10 M) and PGHS-1(Arg 120 3 Glu) (IC 50 ϭ 0.84 M). The increased potency of inhibition against the mutant by DuP697 and L-746,483 appears to be related to the higher K m of the mutant for arachidonic acid. These compounds are competitive with arachidonic acid, and because the inhibition was performed at a substrate concentration (20 M arachidonic acid) approaching the K m of the PGHS-1(Arg 120 3 Glu) mutant (K m ϭ 60 M) but higher than the K m for PGHS-1 (K m ϭ 0.5 M), the IC 50 for inhibition of activity of the mutant is expected to be lower than that of the wild type. This was confirmed by performing assays with the wild type PGHS-1 at lower arachidonic acid concentration (0.6 M) approaching the K m of the microsomal preparation of PGHS-1 (K m ϭ 0.5 M). At this substrate concentration approaching the K m of PGHS-1, DuP697 results in an IC 50 of 40 nM, a value similar to that observed for inhibition of the mutant (data not shown). In sharp contrast to acid NSAIDs, these results demonstrate that the nonacid PGHS inhibitors, DuP697 and L-746,483, retain their potency for inhibition of the mutant PGHS-1(Arg 120 3 Glu).
Experiments were performed to confirm the type of inhibition by diclofenac and DuP697 for PGHS-1(Arg 120 3 Glu) by incubating microsomal preparations for 0 -30 min with the inhibitors prior to initiation of the reaction with arachidonic acid (Fig. 5). The results from these experiments demonstrate that DuP697 at the given inhibitor concentrations inhibits both PGHS-1 and PGHS-1(Arg 120 3 Glu) in a time-independent manner. In sharp contrast, diclofenac inhibits both the wild type and mutant enzyme in a time-dependent mechanism. These results are consistent with previous reports of the mechanism of inhibition of PGHS-1 by DuP697 and diclofenac and indicates that the mutation did not alter the mechanism by which these compounds inhibit the activity.
Aspirin is an acid NSAID that is a time-dependent inhibitor of PGHS-1 but whose mechanism of action is through acetyla-  Various NSAIDs and recently developed inhibitors of PGHS were tested for the inhibition of PGE 2 production by microsomal preparations of PGHS-1 and PGHS-1 (Arg 120 3 Glu). The IC 50 values were obtained from experiments in which the inhibitor was preincubated with these enzyme preparations for 15 min before a 3-min incubation with 20 M arachidonic acid.  4. Inhibition of PGE 2 production by the nonacid inhibitors DuP697 and L-746,483. Microsomal preparations were prepared from COS-7 cells expressing PGHS-1 and PGHS-1(Arg 120 3 Glu) and were assayed for inhibition of PGE 2 by the inhibitors DuP697 (A) and L-746,483 (B) as described in the legend to Fig. 3. These compounds were tested for inhibition of PGHS-1 activity (q) and PGHS-1(Arg 120 3 Glu) activity ( ). tion of PGHS. Microsomal preparations of PGHS-1 and PGHS-1(Arg 120 3 Glu) were preincubated with 1 mM aspirin for 0 -90 min before the addition of arachidonic acid and measurement of enzymatic activity (Fig. 6). Aspirin inhibited PGHS-1 in a time-dependent mechanism with a complete inhibition at 10 min independent of the arachidonic acid concentration used for the experiment (0.5 or 20 M). Interestingly, aspirin at all time points tested only inhibits the mutant PGHS-1(Arg 120 3 Glu) by 40% after a 90-min preincubation. This is further evidence that an acid NSAID that inhibits PGHS through a novel covalent modification requires an initial interaction with the Arg 120 residue.

DISCUSSION
The present results demonstrate that the Arg 120 residue of hPGHS-1 is essential for the potent inhibitory effects of the acid inhibitors flurbiprofen, indomethacin, diclofenac, ketoprofen, and meclofenamic acid. Also, the less potent PGHS inhibitors, ibuprofen and aspirin, are acid NSAIDs that also have a decreased potency for inhibition of PGHS-1(Arg 120 Glu) as compared with PGHS-1. Interestingly, Arg 120 is not essential for interaction with the nonacid DuP697 and L-746,483 compounds. The decreased or lack of inhibition of PGHS-1(Arg 120 3 Glu) by acid NSAIDs supports the model proposed by Garavito and his colleagues in which the Arg 120 residue of PGHS-1 interacts with the carboxylic acid group of flurbiprofen (20). Based on the x-ray crystal structure of sheep PGHS-1, there are only two charged residues that are found within the hydrophobic channel leading up to the cyclo-oxygenase active site, Arg 120 and Glu 524 . In the absence of arachidonic acid or an NSAID, Arg 120 and Glu 524 may form a salt bridge. The importance of this arginine residue for PGHS activity is demonstrated by both a large increase in the K m of the mutant for arachidonic acid and the decrease in both the cyclo-oxygenase and the peroxidase activities of the enzyme. The conversion of the positive charged arginine to a negative charged glutamate residue abolishes the possible interaction between the carboxylic acid group of acid NSAIDs and arachidonic acid with the guanidinium group of Arg 120 of PGHS. This ionic interaction may be essential for the formation of a tight binding inhibitor complex leading to the time-dependent inhibition of PGHS by inhibitors such as indomethacin and flurbiprofen. The presence of the common arginine residue (PGHS-1 and PGHS-2) for the interaction of the carboxylic acid group from both arachidonic acid and certain NSAIDs with PGHS is consistent with the competitive binding of inhibitor preceding the formation of the irreversible enzyme-inhibitor complex (19).
The inhibition by DuP697 and L-746,483, which are nonacid PGHS inhibitors, is not affected by the mutation, implying that the interaction of DuP697 and PGHS-1 involves amino acids other than Arg 120 whose spatial arrangement is not affected by the substitution of Glu for Arg. In addition, the time-independent inhibition of PGHS-1 by DuP697 is maintained in the mutant PGHS-1(Arg 120 3 Glu), and the time-dependent inhibition by diclofenac is similar for the mutant as compared with the wild type, albeit at a higher concentration of inhibitor. The mechanism of inhibition of the Arg 120 3 Glu mutant by DuP697 and diclofenac shows that some of the essential features for the binding of inhibitors at the active site are preserved and confirms the selectivity of the interaction between Arg 120 and NSAIDs possessing a free carboxylic acid group. Aspirin is a unique acid NSAID that inhibits PGHS-1 through a time-dependent covalent acetylation. The slower time-dependent inhibition of the mutant PGHS-1(Arg 120 3 Glu) by aspirin demonstrates the importance of the interaction of aspirin with Arg 120 for its ability to inhibit PGHS.
It has been suggested that PGHS-1 provides prostaglandins for use under normal physiological conditions, whereas PGHS-2 is involved in inflammatory responses (7,8). NSAIDs in current therapeutic use inhibit both isoforms of PGHS with no or little selectivity and have associated toxicities (i.e. ulcero- genecity) that have been attributed primarily to their inhibition of PGHS-1 (7,8,(11)(12)(13)(14). Newly developed PGHS inhibitors, such as NS-398, DuP697, and L-745,337, which display selective inhibition of PGHS-2, retain their anti-inflammatory, analgesic, and anti-pyretic properties but are much more gastric sparing than conventional NSAIDs (7,8,32,33). Most of the conventional NSAIDs contain a carboxylic acid group. Surprisingly all of the selective inhibitors of PGHS-2 reported to date lack a carboxylic acid group but contain a sulfonyl group. Further understanding of the mode of binding of these NSAIDs with both PGHS-1 and PGHS-2 should provide insight into developing better NSAIDs with a decreased potential for toxicity.