Human cyclooxygenase-1 activity and its responses to COX inhibitors are allosterically regulated by nonsubstrate fatty acids.

Recombinant human prostaglandin endoperoxide H synthase-1 (huPGHS-1) was characterized. huPGHS-1 has a single high-affinity heme binding site per dimer and exhibits maximal cyclooxygenase (COX) activity with one heme per dimer. Thus, huPGHS-1 functions as a conformational heterodimer having a catalytic monomer (E(cat)) with a bound heme and an allosteric monomer (E(allo)) lacking heme. The enzyme is modestly inhibited by common FAs including palmitic, stearic, and oleic acids that are not COX substrates. Studies of arachidonic acid (AA) substrate turnover at high enzyme-to-substrate ratios indicate that nonsubstrate FAs bind the COX site of E(allo) to modulate the properties of E(cat). Nonsubstrate FAs slightly inhibit huPGHS-1 but stimulate huPGHS-2, thereby augmenting AA oxygenation by PGHS-2 relative to PGHS-1. Nonsubstrate FAs potentiate the inhibition of huPGHS-1 activity by time-dependent COX inhibitors, including aspirin, all of which bind E(cat). Surprisingly, preincubating huPGHS-1 with nonsubstrate FAs in combination with ibuprofen, which by itself is a time-independent inhibitor, causes a short-lived, time-dependent inhibition of huPGHS-1. Thus, in general, having a FA bound to E(allo) stabilizes time-dependently inhibited conformations of E(cat). We speculate that having an FA bound to E(allo) also stabilizes E(cat) conformers during catalysis, enabling half of sites of COX activity.

and C 10 E 6 detergents were purchased from Anatrace (Maumee, OH). BCA protein reagent was from Pierce. Polyacrylic acid (sodium salt) 5100 was from Hampton Research Corp. [1-14 C] AA (1.85 GBq/mmol) was from American Radiolabeled Chemicals. Hexane, isopropanol, and acetic acid were HPLC grade from Thermo Fisher Scientifi c, Inc. Complete protease inhibitor was from Roche Applied Science. Restriction enzymes were from New England Biolabs, Inc. Nickel-NTA (nitrilotriacetic acid) was from Qiagen. A sample of tobacco etch virus (TEV) protease was generously provided by Dr. Ming Lei ( 38 ). All other materials were analytical grade from Sigma Chemical Co.

Expression, purifi cation, and assay of huPGHS-1
A recombinant huPGHS-1 was engineered with a sequence that contains a octa-histidine tag and a TEV protease cleavage site near the N-terminus but downstream of the signal peptide ( Fig. 1 ). The coding region was incorporated into a baculovirus that was used to infect insect cells from which the recombinant enzyme was purifi ed. The pFastBac expression vector was generated according to the instructions of the manufacturer for the Invitrogen Bac-to-Bac expression system. After 3-4 days of infection with the recombinant baculovirus, the cells were harvested, washed with PBS, and stored at Ϫ 80°C. The method for the purifi cation of huPGHS-1 was essentially the same as in a previous report for ovPGHS-1 ( 22 ). Protein was quantifi ed using BCA reagent. The purity of the recombinant huPGHS-1 was determined by SDS-PAGE and by Western blot analysis. The specifi c activity of purifi ed huPGHS-1 for the preparations used in the experiments reported here averaged 19 U/mg. One unit of COX activity is defi ned as 1 mol of O 2 consumed/min at 37°C in a standard assay mixture containing 100 M AA.
The COX activity of huPGHS-1 was measured as described previously ( 39 ), except that huPGHS-1 was routinely preincubated with two molar equivalents of heme at 37°C for 10 min before being assayed. COX activity was measured polarographically at 37°C in glass chambers containing 3 ml of 0.1 M Tris HCl, pH 8.0, 100 M AA, 1 mM phenol, and 5 M heme. A Yellow Springs Instruments Model 53 oxygen monitor was used to monitor O 2 consumption, and kinetic traces were recorded using DasyLab (DasyTec) software ( 22 ). Reactions were initiated by adding enzyme to the assay chamber, unless otherwise indicated. In experiments involving COX inhibitors, the enzyme preparations were usually preincubated with heme, then preincubated with the inhibitor plus or minus an FA, and an aliquot was then added to the O 2 electrode assay chamber. The rates reported were maximal rates occurring after a lag phase. The lag time is defi ned as the time required for the COX activity to reach a maximum after initiating the reaction ( 24 ), which, under optimal conditions, is about 10 s.

Cleavage of huPGHS-1 by TEV protease
Purifi ed recombinant huPGHS-1 [2.0 mg (1.0 mg/ml)] was treated with TEV protease (0.2 mg) at 4°C for 24 h in 20 mM Tris HCl, pH 8.0, containing 40 mM KCl, 5% glycerol, and 0.02% C 10 E 6 . The sample was mixed with Ni-NTA resin equilibrated in 20 mM Tris HCl, pH 8.0, containing 40 mM KCl, 5% glycerol, and 0.02% C 10 E 6 . The reaction solution was fi ltered slowly through the Ni-NTA resin twice, the resin was harvested and washed with 2.0 ml of buffer, and was harvested again. The yield of unbound huPGHS-1 was 1.1 mg.
In this report, we describe interactions of huPGHS-1 with COX inhibitors and nonsubstrate FAs. Our results provide support for the concept that huPGHS-1, like huPGHS-2, operates as a conformational heterodimer. Two additional, more-specifi c observations are a ) that nonsubstrate FAs weakly inhibit huPGHS-1, and b ) that responses of huPGHS-1 to nsNSAIDs and COX-2 inhibitors are potentiated by common FAs that are not COX substrates. A novel fi nding is that FAs, when combined with time-independent nsNSAIDs like ibuprofen (IBF) elicit a time-dependent inhibition. We speculate that this refl ects an FA-induced kinetic delay that underlies the ability of PGHSs to exhibit half of site activity and to be regulated allosterically by ambient FAs in cells. Both the COX and POX activities of PGHSs require heme for activity. Purifi ed huPGHS-1 had little or no activity when assayed in the absence of heme but underwent a relatively rapid time-dependent activation upon preincubation with heme ( Fig. 2A ). These results indicated that the purifi ed enzyme is largely in an apo form, as is typically observed with purifi ed PGHSs ( 24,26,42,43 ) and is activated at relatively low heme/protein ratios. We analyzed in further detail the affi nity and stoichiometry of heme binding to apo-huPGHS-1 ( Fig. 2B-D ). Titration of apo-huPGHS-1 with heme yielded a major UV/VIS spectral peak with a maximum at 412 nm, consistent with the formation of a heme/PGHS-1 complex ( Fig. 2B ) ( 24,26,42,43 ). The data indicate that there is one high-affi nity heme binding site per PGHS-1 dimer. In the representative titration shown in Fig. 2B , there was a higher affi nity heme site with a K d 1 = 0.110 ± 0.005 M (n = 3) and a second lower affi nity site(s) ( K d 2 = 6.7 ± 1.1 µM) ( Fig. 2C, D ). We presume that the higher affi nity site represents binding of heme to a POX site of huPGHS-1 and that lower affi nity binding occurs to the second POX site and/or another site(s); removal of the octa-histidine tag on the N-terminus of the recombinant enzyme using TEV protease had no signifi cant effect on the heme binding spectra (data not shown). In parallel with the spectroscopic titrations, COX activity assays were performed with huPGHS-1 at different heme-to-protein ratios ( Fig. 2E ). Maximal COX activity occurred with about one heme per dimer (i.e., 0.95 mol heme/mol dimer). This is consistent with there being one high-affi nity heme binding site per huPGHS-1 dimer. Similar results were found in earlier studies of ovPGHS-1 by Kulmacz and Lands ( 26 ) and more recently with huPGHS-2 in our laboratory ( 24 ). 0.1 M Tris HCl, pH 8.0, at 37°C essentially as described previously ( 24 ). Organic extracts of the reactions were subjected to HPLC on a Luna C18 ( 2 ) column (5 m, 250 × 4.6 mm, Phenomenex) mounted on a Shimadzu HPLC system equipped with a radio-HPLC detector (IN/US system, ␤ -RAM model 4), and the separated radioactive reaction products and starting material were quantifi ed ( 24 ).

Titration of huPGHS-1 with heme
Difference absorption spectra of apo-huPGHS-1 titrated with heme were obtained using a Shimadzu UV-2501 PC scanning spectrophotometer as described previously ( 24 ). Aliquots of heme solutions in 20 mM Tris HCl, pH 8.3, containing 40 mM KCl, 0.02% C 10 E 6 , and 2% DMSO, were added to a quartz cuvette containing 200 µl of 5 M apo-huPGHS-1 in the same buffer solution without DMSO at room temperature ( 40,41 ). The increase in absorbance at 412 nm was transformed to a binding fraction and then plotted using Grapher TM 7 software in a linear form using the Scatchard equation in which the slope is equal to the Ϫ 1/ K d for heme binding.

Characterization of recombinant huPGHS-1 and its interaction with heme
A recombinant huPGHS-1 was engineered with a sequence that contains both an octa-histidine tag and a TEV protease cleavage site near the N-terminus and downstream of the signal peptide ( Fig. 1A ). The coding region was incorporated into a baculovirus that was used to infect insect cells from which the recombinant enzyme was purifi ed. Figure 1B shows an SDS-PAGE of the purifi ed recombinant huPGHS-1. The electrophoretic mobility of the huPGHS-1 with respect to the standards corresponded to a protein with a molecular mass of 73 kDa, similar to that of ovPGHS-1 prepared from ovine seminal vesicles. The purity of the huPGHS-1 was estimated by densitometry to be greater than 90%. The average specifi c activity for three representative preparations of purifi ed huPGHS-1 was 19 ± 1.8 U/mg, and the K m of huPGHS-1 for fi city of huPGHS-1 is similar to that of ovPGHS-1 ( 22 ). AA and DHLA are oxygenated with greatest effi ciencies. Addition of 100 M H 2 O 2 to the COX assay mixtures caused increases in the rate of oxygenation of EDA, ADA, EPA, and DHA, as has been observed for EPA previously with PGHSs from various sources ( 6 ).

FA substrate specifi city of huPGHS-1
A number of PUFAs are oxygenated by PGHSs ( 6,22,31 ). Characterization of the products has shown that DHLA, AA, EPA, and ADA are converted to endoperoxides and monohydroxy FAs in different proportions, whereas ␣ -LA, ␥ -linolenic acid, 11 cis , 14 cis -eicosadienoic acid (EDA), and DHA are converted primarily to monohydroxy FAs ( 22,23,31,(44)(45)(46). PGHS-1 has a more-restrictive specifi city than PGHS-2 ( 5, 47 ). As shown in Fig. 3 , huPGHS-1 will oxygenate a number of PUFAs. The speci-   binding to the enzyme using a Scatchard analysis in which the slope is equal to -1/ K d (Grapher TM 7 software); the value of K d 1 in this experiment was 0.11 ± 0.005 M, and the value of K d 2 was 6.7 ± 1.1 M. E: huPGHS-1 (0.98 M) was preincubated with the indicated concentrations of heme for 10 min at 37°C, and then the COX activity was measured using an O 2 electrode assay; in this case, 5 M heme was not included in the standard assay mixture. Values in E are from triplicate determinations. Error bars indicate ± SD.
In  Fig. 4 and Tables 1 and 2 are consistent with the idea that both AA and nonsubstrate FAs bind to an allosteric (E allo ) site of huPGHS-1 and that when added in excess, nonsubstrate FAs can replace AA in the allosteric site. Thus, it appears that huPGHS-1, like huPGHS-2 ( 24 ), has a high-affi nity E allo site to which AA can bind but where oxygenation does not occur. When nonsubstrate

The effects of nonsubstrate FAs on the COX activity of huPGHS-1
Nonsubstrate FAs have either no major effect or a substantial stimulatory effect on huPGHS-2 ( 22,24 ). Members of this same group of FAs were generally found to have a slight inhibitory effect on the COX activity of huPGHS-1 ( Fig. 4 ). OA is the most-effective inhibitor, causing 25-30% inhibition. That nonsubstrate FAs cause only incomplete inhibition at relatively high nonsubstrate FA-to-AA ratios suggests that nonsubstrate FAs do not compete effectively with AA for binding to the COX active site of the enzyme.
The average specifi c activity of purifi ed huPGHS-1 is about 20 U/mg, so under optimal, saturating substrate conditions, 1 mol of enzyme can convert about 1,500 mol of AA to product per min. However, as shown in Table 1 , when 1 µM [1-14 C]AA was incubated with 1 µM huPGHS-1 for 8 min at 37°C, 11% of the radioactive AA remained in an unreacted form. If a bolus of unlabeled AA was added to the reaction mixture 4 min after the reaction had been initiated, the amount of unreacted radioactive AA decreased to 2.4%. Thus, the enzyme remained active, but was unable to oxygenate all of the AA. Table 1 also shows that, 1.0%, 2.4%, and 4.9% of the AA remains when 1 M [1-14 C]AA is incubated for 4 min with 0.1, 1.0, or 2.0 M huCOX-1, respectively , and 60 M cold AA is added after the initial 4 min (and the reaction continued for another 4 min). These latter data show that the amount of unreacted AA is proportional to and increases with the amount of huPGHS-1 added to the reaction mixture. The results are consistent with a model ( 24 ) in which unreacted [1-14 C]AA is bound to the enzyme in a latent form.  ) was incubated with the indicated concentration of huPGHS-1 at 37°C for 4 min. Where indicated, 60 µM unlabeled AA was added 4 min after initiating the reaction, and the incubation was continued for another 4 min. The reactions were stopped by the addition of ethyl acetate/acetic acid (20:1), and an aliquot of the organic phase was subjected to radio-reverse-phase (radio-RP -HPLC) to separate the radioactive products and unreacted AA, as described in Experimental Procedures. The results are shown as the percentage of 14 C label remaining in the HPLC fraction eluting with unreacted AA. a Signifi cant difference from control to which no unlabeled AA was added at 4 min in Student t -test ( P < 0.005). C]AA (1 µM) was incubated with huPGHS-1 (1 µM) at 37°C for 4 min, then unlabeled AA, PA, or OA was added, and the incubation was continued for another 4 min. The reactions were stopped by the addition of ethyl acetate/acetic acid (20:1), and an aliquot of the organic phase was subjected to radio-RP-HPLC to separate the radioactive products and unreacted AA, as described in Experimental Procedures. The results are shown as the percentage of 14  huPGHS-1 by ASA ( Fig. 5 ). For example, treatment of apo-huPGHS-1 ( Fig. 5A ) versus holo-huPGHS-1 ( Fig. 5B, C ) with 0.5 mM ASA under similar conditions caused 20% and >80% COX inhibition, respectively. We recently found that nonsubstrate FAs affect the responses of huPGHS-2 to various inhibitors including ASA. PA, OA, SA, and ⌬ 11c 20:1 potentiated the inhibition of holo-huPGHS-1 by ASA ( Fig. 5D ).

Effects of nonsubstrate FAs on responses of huPGHS-1 to time-dependent COX inhibitors
FBP, DICLO, NAPX, INDO, and meclofenamic acid (MECLO) are time-dependent inhibitors of huPGHS-1 ( 30 ). As shown in Fig. 7A , various nonsubstrate FAs modestly potentiate the inhibition by FBP of huPGHS-1. Similarly, PA increases the time-dependent inhibition of huPGHS-1 by DICLO and NAPX ( Fig. 7B, C ) but has no effect on the response of huPGHS-1 to INDO or MECLO ( Fig. 7D, E ). For an FA to potentiate the effect of an inhibitor (e.g., FBP, DICLO, and NAPX) or at least not to interfere with the action of an inhibitor (e.g., INDO and FAs such as PA or OA bind the allosteric monomer of huPGHS-1, they slightly decrease the catalytic effi ciency of the partner, catalytic monomer toward AA.

Heme and nonsubstrate FAs potentiate inhibition of huPGHS-1 by ASA
ASA causes irreversible inhibition of PGHS-1 by acetylating the sidechain of Ser-530 in one monomer of ovPGHS-1 ( 35 ). A similar experiment with purifi ed holo-huPGHS-1 and [ 14 C]acetyl-salicylic acid indicated that ASA acetylates a single monomer of huPGHS-1 (data not shown). Earlier reports showed that heme is required for the effi cient acetylation of ovPGHS-1 by ASA ( 48 ). Similar results were observed in comparing the inhibition of apo-and holo- Fig. 6. ASA inhibition of huPGHS-1 is attenuated by IBF, CBX, and NAPX. Purifi ed apo-huPGHS-1 was preincubated with two molar equivalents of heme at 37°C for 10 min, then incubated with 100 µM ASA with or without an inhibitor (100 µM IBF, 0.78 µM CBX, or 100 µM NAPX) at 37°C for 30 min. Finally, an aliquot of the preincubation mixture was added to an oxygen electrode assay chamber, and COX activities were assayed polarographically, as described in Experimental Procedures, with 100 µM AA as substrate. Addition of the preincubation mixture to the assay chamber resulted in a 75-fold dilution of the contents of the preincubation mixture. The results are presented as a percentage of the COX activity of the heme-activated huPGHS-1. The control is without ASA or inhibitor. The results are from a single experiment involving triplicate determinations. The error bars indicate ± SD. Signifi cant differences between ASA plus inhibitor versus ASA alone are denoted with asterisks. *** P < 0.001, **** P < 0.0001. Purifi ed apo-huPGHS-1 was incubated with ASA at 37°C for 60 min, and COX activities were measured polarographically using a standard assay mixture with 100 µM AA, as detailed in Experimental Procedures. The results are shown as a percentage of starting COX activity. B: Effect of preincubation with heme on huPGHS-1 COX activity. Purifi ed apo-huPGHS-1 (2.4 M) was preincubated with two molar equivalents of heme at 37°C for 10 min, then incubated with 0, 0.5, or 4.0 mM ASA at 37°C for 60 min and assayed as in A. C: Time-dependent inactivation of huPGHS-1 by ASA. Purifi ed apo-huPGHS-1 (1.25 M) was preincubated with two molar equivalents of heme as in B, then incubated with 0 or 0.5 mM ASA at 37°C for the indicated times and then assayed for COX activity as in A. D: Effects of various nonsubstrate FAs on inhibition of huPGHS-1 by ASA. Purifi ed apo-huPGHS-1 was activated by preincubation with heme, as in B, incubated with 50 µM ASA with or without a nonsubstrate FA at 37°C for 30 min, and then assayed for COX activity, as in A. The control contained neither ASA nor a nonsubstrate FA. The results are from a single experiment involving triplicate determinations. The error bars indicate ± SD. Asterisks denote signifi cant differences between ASA plus FA versus ASA alone in the Student t -test. * P < 0.05, ** P < 0.01, *** P < 0.001. PA, palmitic acid; SA, stearic acid, OA, oleic acid. together led to a nonadditive, approximately 20% inhibition; additionally, when huPGHS-1 pretreated in this way was assayed in the presence of 100 M AA plus 100 µM IBF, the COX activity was also decreased by about 25% over that observed with enzyme that had been preincubated with PA or IBF alone. The time dependence elicited by preincubation of huPGHS-1 with PA and CBX or MEF, while apparent, was less pronounced than that observed with IBF. In the cases of IBF ( Fig. 8A ) and CBX ( Fig. 8C ), both of which can interfere with ASA inhibition of huPGHS-1 ( Fig. 6 ), the overall evidence is consistent with these inhibitors functioning by binding to the E cat monomer of huPGHS-1.
We investigated the unusual time dependence seen with IBF plus nonsubstrate FAs in greater detail fi rst by performing preincubations of enzyme with IBF plus PA for 0-60 min ( Fig. 9A ). An obvious time dependence occurred during the fi rst 30 min of the preincubation of 100 M IBF plus 25 M PA, with a maximum decrease in activity of 20-25% of control values. There was a dose dependence in the effects of both PA and IBF ( Fig. 9B, C ). Increasing the concentration of IBF beyond 100 M or the concentration MECLO), the FA must bind to a different site on the enzyme than the inhibitor. It should be noted that when tested at concentrations three to fi ve times higher than those used in Fig. 7 , each inhibitor by itself caused a complete, time-dependent loss of COX activity.

A combination of a time-independent COX inhibitor and a nonsubstrate FA induces a time-dependent inhibition of huPGHS-1
The observation that nonsubstrate FAs can potentiate the effects of some time-dependent COX inhibitors led us to investigate potential interactions between FAs and timeindependent inhibitors. We were surprised to fi nd that when PA was preincubated with IBF, mefenamic acid (MEF), or CBX, an inhibition was observed in each case that was incomplete but exhibited time dependence ( Fig. 8 ); similar results were obtained when OA was substituted for PA ( Fig. 8C ). For example, when huPGHS-1 was preincubated with PA or IBF alone and then assayed upon dilution into an assay mixture containing 100 M AA, greater than 95% of the starting COX activity remained. In contrast, preincubation of huPGHS-1 with PA and IBF The results in each panel are from a single experiment involving triplicate determinations. The error bars indicate ± SD. Signifi cant differences between results obtained with an inhibitor alone versus the inhibitor plus a nonsubstrate FA are denoted with asterisks. * P < 0.05, ** P < 0.01.

Comparisons of huPGHS-1, ovPGHS-1, and huPGHS-2
There is greater than 90% sequence identity between ovPGHS-1 and huPGHS-1 ( 5,51 ). Thus, it is not surprising that the two enzymes are functionally and quantitatively similar with respect to heme stoichiometry and affi nity ( 26 ), stoichiometry of ASA acetylation ( 35 ), dependence on heme for effi cient ASA inhibition ( 48 ), and partial inhibition by nonsubstrate FAs including PA, SA, and OA ( 22 ). There are modest differences in substrate specifi cities, with huPGHS-1 exhibiting relatively less activity toward EDA, ADA, and DHA and more activity toward EPA ( 22 ).
We recently provided evidence that both huPGHS-2 ( 22-25 ) and ovPGHS-1 ( 22,25,28 ) function as conformational heterodimers of the E allo /E cat form. Our present results with huPGHS-1 are consistent with and expand upon these and earlier studies ( 26,27 ), indicating that heme, nonsubstrate FAs, and certain COX inhibitors bind with different affi nities to one of the two monomers comprising each huPGHS-1 dimer. Using the general model developed for huPGHS-2 ( 24 ), one monomer of PA beyond 25 M did not affect the magnitude of the inhibition; however, the critical micelle concentration of PA is approximately 25 M under the preincubation conditions ( 24 ).
Further substantiating the time-dependent feature of the combined IBF/PA inhibition was the observation that the time-dependent inhibition was reversed in a time-dependent manner. In the experiment depicted in Fig. 9D , huPGHS-1 was preincubated with IBF plus PA, and the enzyme sample was then added to an assay chamber to which AA was included at zero time or added 10, 30, or 90 s later. The COX activity was completely recovered within 30 s. We would emphasize that there is a 10 s kinetic lag time before maximal COX activity can be measured, even when AA is present in the assay mixture when the enzyme is added. Thus, the 0 s time point in Fig. 9D actually corresponds to a 10 s delay during which COX activity, if inhibited, could be recovered. We suspect, based on the rapid recovery of activity observed between 0 and 30 s ( Fig. 9D) , that were we able to measure activity at a real zero time, the enzyme would be inhibited by considerably more than 20-25%. Purifi ed apo-huPGHS-1 was activated with heme as described in the legend to Fig. 3 , then incubated with the indicated concentrations of nonsubstrate FAs and/or inhibitor at 37°C for 30 min. Finally, COX activities were assayed polarographically as described in Experimental Procedures, with 100 µM AA as substrate and, where shown, the indicated concentration of COX inhibitor was included in the assay chamber; note that addition of the preincubation mixture to the assay chamber resulted in a 75-fold dilution of the contents of the preincubation mixture. The results are presented as a percentage of the COX activity of the heme-activated huPGHS-1. The control in each panel is without inhibitor or nonsubstrate FA. The results in each panel are from a single experiment involving at least triplicate determinations. The error bars indicate ± SD. Signifi cant differences between results obtained with an inhibitor alone versus the inhibitor plus nonsubstrate FA are denoted with asterisks. * P < 0.05, ** P < 0.01. strate FAs compete with AA for binding to the E allo subunit of huPGHS-1.
Unlike nonsubstrate FAs, AA can also bind effectively to E cat . Using the model shown in Fig. 10 , the K m value for AA and the values shown in Table 2 for unreacted, E allo bound AA at various nonsubstrate FA concentrations, we can calculate the K d values for the binding of AA and nonsubstrate FAs to E allo and E cat , as described previously ( 24 ). We estimate the K d for AA binding to E allo to be 0.35 µM, similar to the value (0.26 M) determined for E allo of huPGHS-2 ( 24 ). The K d value for the binding of PA (and OA) to E allo is calculated to be about 60 times higher than that for AA, or approximately 20 µM. This estimate is based on the concentrations of PA (and OA) that causes 50% inhibition of AA binding to E allo when the effective concentration of AA is about 0.0758 µM ( Table 2 ). This K d value for PA is in the same range (7.5 M) as that estimated for PA binding to E allo of huPGHS-2 ( 24 ). The K m for AA with huPGHS-1 was determined to be 5.1 M, which, based on our model, is the K d for AA binding to E cat ; again, we assume that FAs do not bind effi ciently to E cat because huPGHS-1 is not completely inhibited, even at high concentrations of nonsubstrate FAs.
We speculate that nonsubstrate FAs are important in the coordinate regulation of PGHS-1 versus PGHS-2 activities when the two isoforms are coexpressed at comparable levels in cells. Thus, at low AA concentrations and ambient levels of the major nonsubstrate FAs (i.e., PA, SA, and OA), huPGHS-1 activity would be expected to be somewhat depressed; in contrast, huPGHS-2 activity would be potentiated ( 22,24 ). At relatively higher AA concentrations, such as those seen immediately following activation of cytosolic PLA 2 , huPGHS-1 activity would be increased relative to that of PGHS-2 ( 52 ). Obviously, this concept needs to be tested in intact cells.

Effects of nonsubstrate FAs on responses of huPGHS-1 to ASA and other time-dependent COX inhibitors
Nonsubstrate FAs either modestly potentiate or have no effect on the time-dependent inhibitory actions of the six time-dependent inhibitors tested in our study. The results with ASA, FBP, DICLO, and NAPX, whose actions are promoted by nonsubstrate FAs, are easily rationalized if the nonsubstrate FAs act allosterically to facilitate the binding and/or the subsequent actions of inhibitors on huPGHS-1 ( 24 ) ( Fig. 10 ). Although less defi nitive, the fact that nonsubstrate FAs fail to interfere with the actions of INDO or MECLO suggests that FAs do not compete with the site to which either inhibitor binds; furthermore, we observed that slightly higher concentrations of INDO and MECLO (as well as ASA, FBP, DICLO, and NAPX) cause complete time-dependent inhibition of the enzyme, consistent with these inhibitors acting directly on E cat ( 24 ). That NAPX functions via E cat of huPGHS-1 is also consistent with the observation that NAPX can interfere with the action of ASA, which acts via E cat . In contrast to what is observed with huPGHS-1, NAPX, and FBP appear to function as allosteric inhibitors of huPGHS-2 by binding to E allo . We speculate that all time-dependent COX inhibitors of binds heme and is denoted as the catalytic subunit (E cat ), and the other monomer is the allosteric subunit (E allo ) ( Fig. 10 ).

Effects of nonsubstrate FAs on huPGHS-1 catalysis
Nonsubstrate FAs have a modest inhibitory effect on AA oxygenation by huPGHS-1. The magnitude of the inhibition (15-30%) is similar for all the FAs tested. There is no dose-dependent increase in inhibition at concentrations above 5 M FA, suggesting that the inhibition of huPGHS-1 by nonsubstrate FAs does not involve direct competition with AA at the E cat site of huPGHS-1 but rather involves binding of nonsubstrate FAs to E allo ( Fig. 10 ) That PA and OA can replace bound, unreacted AA from E allo of huPGHS-1 ( Tables 1, 2 ) is further evidence that nonsub- , and then assayed as in A. With respect to B and C, the relative COX activities were 101 ± 2.3% and 98.6 ± 2.8%, respectively, when 100 µM IBF or 25 µM PA was preincubated alone with huPGHS-1 for 30 min at 37°C. D: Recovery of COX activity following pretreatment with IBF plus PA. huPGHS-1 (1 M) was activated with heme and then preincubated at 37°C for 30 min with 100 M IBF plus 25 M PA. Aliquots of the sample were added to a standard assay solution to which AA (100 M fi nal concentration) was added 0, 10, 30, or 90 s later. This involved a 75-fold dilution of the preincubation mixture. Signifi cant differences between results at 0 s and later times are denoted with asterisks. *** P < 0.001. inhibition was rapidly reversed (i.e., within 10-30 s) and the lag time required to achieve maximal activity is about 10 s, we suspect that the maximal time-dependent IBF inhibition was signifi cantly more than 25%. One plausible explanation is that PA is acting via E allo to expose the time-dependent component of IBF inhibition of huPGHS-1 that normally cannot be observed in the time frame ( у 1 min) in which time dependence has usually been measured ( 24 ) ( Fig. 10 ). We speculate that preincubating huPGHS-1 with IBF plus PA converts the enzyme to a time dependently inhibited conformer represented by the E allo -PA/ + *E cat -IBF form ( Fig. 10 ). When diluted into an assay chamber containing 100 M AA, PA and then IBF dissociate, leading to formation of a maximally active E allo -AA/E cat -AA form.
It is apparent from previous studies that there must be a transient time dependence associated with half of the sites of COX activity. For example, in order for the G533A/Native huPGHS-2 mutant heterodimer (having one monomer with a Gly533Ala mutation and a native monomer) to exhibit the same activity as the native homodimer ( 25 ), the mutant heterodimer would have to become at least transiently stabilized, with all of the G533A monomer in the E allo form and all of the native subunit in the E cat form ( 5 ). This could be facilitated were an FA bound to E allo able to stabilize an E allo -FA/E cat complex long enough for it to bind substrate AA (E allo -FA/E cat -AA) without having the FA dissociate to E allo /E cat . PGHS-1 function by binding E cat , whereas with PGHS-2, at least some time-dependent COX inhibitors (e.g., NAPX and FBP) function via E allo .
Combining IBF, MEF, or CBX with a nonsubstrate FA induces a time-dependent inhibition of huPGHS-1 IBF, MEF, and CBX are time-independent inhibitors of ovPGHS-1 ( 35,53 ). IBF and MEF are reported to be time-independent inhibitors of huPGHS-2 ( 34 ), whereas CBX is a time-dependent inhibitor of this isoform ( 11 ). Empirically, time dependence means that the COX activity of the enzyme can be decreased in a time-dependent manner by preincubation with the inhibitor; typically, dilution of the enzyme inhibitor complex leads to a time-dependent restoration of activity. Using these criteria, we fi nd that when tested alone, IBF, MEF, CBX, and nonsubstrate FAs are time-independent inhibitors of huPGHS-1 ( Fig. 8 ). Thus, a surprising fi nding was that when huPGHS-1 was preincubated with a combination of IBF and PA, the enzyme lost 20% of its activity, and the degree of inhibition with IBF plus PA was enhanced during a 30 min preincubation period. Moreover, when enzyme that had been preincubated with PA plus IBF was diluted into an assay chamber lacking the AA substrate and then AA was added 30 s later, the activity was recovered; thus, the timedependent inhibition was time dependently reversible.
The maximal inhibition that we could detect with IBF plus PA was incomplete (20-25%). However, because the Fig. 10. Model depicting interactions among huPGHS-1 and COX inhibitors and nonsubstrate FAs. E allo / E cat represents a huPGHS-1 heterodimer where E cat is the catalytic subunit to which heme is bound and E allo is the allosteric subunit that lacks bound heme. AA is arachidonic acid, PA is palmitic acid, a representative nonsubstrate FA, I is a generic COX inhibitor that can bind to E cat . Initially, all inhibitors bind E allo /E cat reversibly to form E allo /E cat -I. With typical time-dependent inhibitors, E allo /E cat -I undergoes a conformation change to a more-stable E allo /*E cat -I form that reconverts to E allo /E cat -I only slowly. However, the rates for the interconversion of E allo /E cat -I and E allo /*E cat -I are unique to each inhibitor, with the equilibrium lying in the direction of E allo /E cat -I with "reversible" inhibitors like IBF and toward E allo /*E cat -I with "irreversible" inhibitors like INDO. The presence of nonsubstrate FAs such as PA stabilizes E allo /*E cat -I. We speculate that this occurs through a process involving formation of an E allo -FA/ + *E cat -I complex. We further suggest that nonsubstrate FAs must stabilize E allo -FA/ + *E cat -I for all inhibitors, including those previously considered to be rapidly reversible, time-independent inhibitors (e.g., IBF). The AA in the E cat -AA can be oxygenated to form PGG 2 . Additional explanations of the various forms and their relative activities are indicated in the fi gure. K d values for AA and PA binding to E cat and E allo were calculated as described in the text. We make the assumption that nonsubstrate FAs do not bind effi ciently to E cat (i.e., K d > 50 M)