Acetylation by Aspirin (Prostaglandin Synthase/Prostacyclin/Thromboxane A2/Human Platelets/Acetylsalicylic Acid) JOHN W

Proc. Natl. Acad. Sci. USA Vol. 75, No. 10, pp. 5181-5184, October 1978 Medical Sciences

Sensitivity of fatty acid cyclooxygenase from human aorta to acetylation by aspirin (prostaglandin synthase/prostacyclin/thromboxane A2/human platelets/acetylsalicylic acid) JOHN W. BURCH, NANCY LEWIS BAENZIGER, NANCY STANFORD, AND PHILIP W. MAJERUS* Division of Hematology-Oncology, Department of Internal Medicine, Washington University School of Medicine, St. Louis, Missouri 63110 Communicated by P. Roy Vagelos, July 28, 1978

ABSTRACT The rate of acetylation of fatty acid cyclooxy- from Nu Chek Prep, Elysian, MN. Nonidet P40 was purchased genase (prostaglandin synthase, EC 1.14.99.1) by [acetyl-3H- from Particle Data Laboratories, Elmhurst, IL. aspirin was measured in microsomes from human aortas and Indomethacin; coronary arteries and intact and disrupted human platelets. We bovine hemoglobin, type I; reduced glutathione; and epi- also measured the inhibition by aspirin of prostacyclin gener- nephrine bitartrate were purchased from Sigma Chemical Co., ation from exogenous arachidonic acid in shredded human St. Louis, MO. All other chemicals were reagent grade. aorta. Cyclooxygenase in human aorta and coronary artery Acetylation of Vascular Microsomes. Frozen aortas or microsomes is approximately 1/250th as sensitive to aspirin as coronary arteries were shredded with a hand-held enzyme in intact platelets, and 1/60th as sensitive to aspirin as vegetable enzyme measured in a platelet microsomal preparation. On the shredder or crushed with a hammer. Care was taken to keep basis of the in vitro data presented, we predict that small oral the tissues frozen (i.e., in contact with dry ice) during these doses of aspirin are sufficient to inhibit platelet prostaglandin procedures. The shredded or broken tissue was homogenized production but are not sufficient to substantialFy affect aorta in ice-cold 0.154 M KCl, 10 ml per g of tissue, with a Polytron or coronary artery prostaglandin production. homogenizer (five 1-min bursts at setting 7). The resulting homogenate was centrifuged 5 min at 4400 g. The supernate Inhibition of prostaglandin and thromboxane synthesis in was filtered cheesecloth 'and a platelets interferes with platelet function, as measured in Vvo through particulate fraction (1) or in vitro (2). This finding has led to clinical trials to de- containing microsomes was prepared by centrifugation at termine whether inhibitors of fatty acid cyclooxygenase 180,000 X g (max) for 30 min. The pellet of microsomes was (prostaglandin synthase, EC 1.14.99.1), the enzyme responsible rinsed with distilled water and resuspended by brief homoge- for the conversion of arachidonic acid to prostaglandin G2, will nization in phosphate buffer,t pH 6.5, or in the phosphate alter the mortality rate after myocardial infarction. Several trials buffer supplemented with 50 ,tg of hemoglobin per ml, 5 mM using aspirin and sulfinpyrazone in men who have survived one epinephrine bitartrate, and 3 mM reduced glutathione. One myocardial infarction suggest that these drugs decrease the milliliter of buffer was added for microsomes from each 0.5 death rate (3-6). The wisdom of this therapeutic approach, to 1 g of shredded whole tissue. The resulting microsomal sus- however, has been challenged (7, 8). The major product of pension (1 ml) was incubated at 370 with [acetyl-3H]aspirin. arachidonic acid transformation via cyclooxygenase in arterial Reactions were stopped at various times by adding 8 ml of cold tissue, prostacyclin (PGI2), is postulated to be a naturally oc- buffer (40) and placing the samples on ice. The microsomal curring antithrombotic agent (8). Accordingly, inactivation of suspensions were collected by centrifugation as described above cyclooxygenase in blood vessels could promote thrombosis. and solubilized in 0.5% Nonidet P40. The radioactivity incor- Aspirin blocks prostaglandin production (9, 10) by covalently porated into cyclooxygenase was determined after acetone acetylating fatty acid cyclooxygenase (11). The enzyme in precipitation of the Nonidet P40-solubilized protein, solubili- human zation in sodium dodecyl sulfate, and polyacrylamide gel platelets is extremely sensitive to small doses of the drug electrophoresis in the presence of sodium dodecyl sulfate as in vivo and in vitro (12). We now demonstrate that fatty acid described previously (12). In experiments utilizing indo- cyclooxygenase from human thoracic aorta and human coro- methacin, ibuprofen, or flurbiprofen, these drugs were incu- nary artery is much less sensitive to aspirin in vitro, as measured bated with the microsome suspensions for 5-10 min at room by acetylation of the enzyme with [acetyl-3H]aspirin or by temperature prior to addition of aspirin. Experiments utilizing bioassay of PGI2 produced by aorta after aspirin treatment. mixtures of aortic microsomes and platelet particulate fractions MATERIALS AND METHODS were also incubated for 5-10 min at room temperature prior to the addition of aspirin. Human ascending aortas and coronary arteries were obtained In experiments using vascular microsomes, "nonspecific" at autopsies performed less than 12 hr after death. The speci- acetylation by aspirin was measured as radioactivity incorpo- mens were cleaned of adventitia, immediately frozen in liquid rated into the cyclooxygenase peak from [acetyl-3H]aspirin in nitrogen, and stored at -80°. Washed human platelets and a the presence of a large excess of unlabeled aspirin. Nonspecific platelet particulate fraction were prepared (12). Prostacyclin, acetylation was measured at a single time in each experiment flurbiprofen, and ibuprofen were gifts from Upjohn, Kalama- and was assumed to be linear with time (11). In two experiments zoo, MI. [Acetyl-3H]aspirin, specific activity 177 Ci/mol, was frozen shredded aorta was thawed in aspirin-containing solution synthesized (11). Unlabeled aspirin was purchased from Merck, and the enzyme was thereby acetylated before microsome West Point, PA, and stored as a 100 mM solution in ethanol. preparation. These experiments gave the same results, indi- ['4C]Serotonin, 40-60 Ci/mol, was purchased from New En- gland Nuclear, Boston, MA. Arachidonic acid was purchased Abbreviation: PGI2, prostacyclin. * To whom reprint requests should be addressed. The publication costs of this article were defrayed in part by page t Buffers used in the present experiments. Phosphate buffer, pH 6.5: charge payment. This article must therefore be hereby marked "ad- 4.3mM K2HPO4/4.3 mM Na2HPO4/24.4 mM NaH2PO4/113 mM vertisement" in accordance with 18 U. S. C. §1734 solely to indicate NaCl/5.5 mM glucose. Tris buffers: 15 mM Tris-HCl (pH 7.4 or 8.6 this fact. at 25°)/139 mM NaCI/5.5 mM glucose. 5181 Downloaded by guest on September 27, 2021 5182 Medical Sciences: Burch et al. Proc. Natl. Acad. Sci. USA 75 (1978) cating that microsome preparations did not affect sensitivity to aspirin. Acetylation of Platelets or Platelet Particulate Fractions. Washed intact platelets were suspended in phosphate buffer, pH 6.5, or in Tris buffer,t pH 7.4, and were incubated with [acetyl-3H]aspirin at 370. The platelet particulate fractions were resuspended by brief sonication in either of the above buffers E containing 50 ,tg of hemoglobin per ml, 5 mM epinephrine 0. bitartrate, 3 mM reduced glutathione, and 5 mM NaOH (to CT 80 neutralize the epinephrine bitartrate). They were also incubated with [acetyl-3H]aspirin at 370. At intervals, 2 X 108 platelets or the equivalent amount of particulate fraction was removed, 40 - added to sufficient sodium dodecyl sulfate/2-mercaptoethanol solution to result in respective final concentrations of 5% and 0.1 M, and placed on ice. Subsequently, the samples were boiled for 5 min and radioactivity incorporated-into cyclooxygenase 0.1 10 20 30 40 was determined by sodium dodecyl sulfate/polyacrylamide Gel slice gel electrophoresis (11). FIG. 1. Sodium dodecyl sulfate/polyacrylamide gel electropho- PGI2 Synthesis in Aortic Tissues. Frozen shredded aortic resis ofhuman aorta microsomes. Microsomes from 1 g ofhuman aorta tissue was suspended in Tris buffer, pH 7.4, with or without 200 were incubated with 200 jAM [acetyl-3H]aspirin for 90 min. ,uM aspirin (0.2% ethanol final concentration in either case), incubated at 370 at a concentration of 0.2-0.8 g wet weight/ml, death (31-88) or the time from death to autopsy. There were and then chilled. Tissue pellets were re-isolated by centrifu- no marked differences in the degree of atherosclerosis in the gation at 40 for 15 min at 14,500 g, washed once by resuspension various samples. The time course of acetylation obtained in in Tris buffer,t pH 8.6, and re-isolated. To generate PGI2, several experiments using microsomes from human aortas or pellets were then suspended in Tris buffer, pH 8.6, (0.8-1.2 g human coronary arteries, or using intact human platelets, is wet weight/ml) containing 25 ,gM arachidonic acid, stirred 8 shown in Fig. 2. Note that the rapid acetylation of cyclooxy- minutes at 37°, then chilled, and the tissues were collected by genase in platelets was obtained at 1/10th the aspirin concen- centrifugation as above. Supernates from this step were frozen tration used with aorta and coronary artery. The reactions with in liquid nitrogen and stored at -80°. The PGI2 thus generated aspirin under these conditions are pseudo-first-order, because was measured by the inhibition of thrombin-induced [14C]- they are carried out in the presence of a large excess of aspirin serotonin release in washed human platelets as described pre- relative to enzyme. The negative slope of each line equals the viously (13). In this assay, incubation of platelets with 200-M,1 apparent rate constant, kapparent, which is directly proportional aliquots of supernate increased the thrombin concentration to aspirin concentration (11). We have defined a "potency" required for serotonin release; the resultant shift in the thrombin term as kapparent divided by aspirin concentration to compare dose-response curve, AT5o, is defined as the dose of thrombin the ease of acetylation by aspirin of cyclooxygenase from dif- required for 50% serotonin release in the presence of sample ferent tissues. This term is directly proportional to the true rate + the dose required for 50% serotonin release in control platelets constant of the reaction between cyclooxygenase and aspirin, in the same experiment. This assay is linear at PGI2 concen- and is independent of the aspirin concentrations used in indi- trations from 0.1 to 1.5 ng/ml. A PGI2 standard incubated with vidual experiments. Its measurement requires an estimate of platelets yielded a ATso of approximately 8/ng in these experiments. The putative PGI2 in the supernate was labile at pH 4 and stable at pH 8.6 as shown previously (13, 14). RESULTS Incubation of human aortic microsomes with [acetyl-3H]aspirin results in the acetylation of a single major protein species as 0.5 . demonstrated by sodium dodecyl sulfate/polyacrylamide gel electrophoresis (Fig. 1). We conclude that the protein is cy- 0.31.0 clooxygenase because: (i) the results are similar to those previously reported for sheep seminal vesicles (15) and human c0.2 -. platelets (11). (ii) Electrophoresis of mixtures of acetylated 0 platelet and aortic microsomes produced a single radioactive CU *'*I protein migrating in the same position as samples from either 0.1 platelets or aorta alone. (iii) Incubation of aortic microsomes and [acetyl-3H]aspirin in the presence of inhibitors of cy- or clooxygenase (100,uM indomethacin, flurbiprofen, ibupro- 0.05 1. ~ fen) resulted in a 50-60% decrease in tritium incorporation. (iv) 10 40 80 120 160 The acetylation was saturable, because incubation of aortic Minutes microsomes and [acetyl-3H]aspirin in the presence of a 10- to FIG. 2. Time course of cyclooxygenase acetylation by aspirin. 100 fold excess of unlabeled aspirin, reduced incorporation to Aorta and coronary artery microsomes were incubated with 200 ,4M [acetyl-3H]aspirin, platelets with 20 AM [acetyl-3H]aspirin. Identi- a mean of 8%. fications and calculated potencies (see text) are: &, human aorta, We studied tissues from 12 cadavers (7 males, 5 females). In potency 19 M-1 min-; 0, human aorta, potency 20 M-1 min';O, 10, cyclooxygenase acetylation was observed. The total incor- human coronary artery, potency 40 M-1 min-; 0, human aorta, po- poration ranged from 200 to 4600 cpm/g of aorta. We have no tency 48 M-1 min-; U, human aorta, potency 96 M-1 min'; A, explanation for the variation; it was not related to the age at human intact platelets, potency 19,600 M-1 min-. Downloaded by guest on September 27, 2021 Medical Sciences: Burch et al. Proc. Natl. Acad. Sci. USA 75 (1978) 5183

the total amount of active cyclooxygenase present in the reac- results. A single experiment was carried out on microsomes tion (to calculate the fraction unacetylated enzyme). The latter obtained from-a pooled sample of coronary arteries from four is difficult to measure accurately in unpurified suspensions such cadavers. The mean potency obtained from three time points as those used in the present experiments because at long incu- was 55 M-' min1. Several experiments using aorta from nine bation times and high aspirin concentrations nonspecific cadavers yielded a mean potency of 75 M-' min1. Similar acetylation becomes significant. In the experiments shown in results were obtained when shredded aorta was incubated with Fig. 2, total active cyclooxygenase was estimated by successive [acetyl-3H]aspirin and microsomes were prepared and solu- approximations, using a least-squares linear regression method bilized subsequent to the aspirin treatment. In contrast to these to give the best-fit straight line for the experimental data. To values, intact platelets from 10 individuals were more sensitive avoid the potential error such an estimate may introduce into to aspirin (potency = 18,600 M-1 min'). The acetylation of the data, potencies were also calculated in a different fashion a platelet particulate fraction prepared in a fashion similar to for comparison of the effects of aspirin in different tissues (Fig. that used for the aorta preparations yielded a potency of 4300 3). In the latter instance maximum acetylation was estimated M-1 mink (three individuals). Therefore, cyclooxygenase in as the net tritium incorporation into cyclooxygenase at 160 min. microsomes from human coronary arteries and aortas is ap- This estimate of maximal acetylation is somewhat low, as can proximately 1/250th as sensitive to aspirin as enzyme in whole be seen by comparison with Fig. 2, and artificially elevates the human platelets. Vascular enzyme is approximately 1/60th as calculated potency values for experiments with aorta or coro- sensitive to aspirin as enzyme in disrupted platelet prepara- nary artery microsomes. The tissues may be less sensitive to tions. aspirin than reported. A similar analysis was carried out for the Acetylation of cyclooxygenase in mixtures of disrupted calculations involving disrupted and intact platelets. There is platelets and aortic microsomes (10-200 tiM aspirin) demon- little nonspecific acetylation in the particulate fraction of strated that the decreased ability of aspirin to acetylate aortic human platelets (11), because the acetylation reaction of platelet cyclooxygenase was not due to the presence of a diffusable in- cyclooxygenase is much more efficient. Therefore, such an hibitor of acetylation present in the aorta preparation. estimate of total active cyclooxygenase is more accurate in this We also measured the effect of aspirin on PGI2 production tissue, and the corresponding potencies are also more accurate. from exogenous arachidonic acid by shredded aorta to establish A comparison of the potencies presented in Fig. 2 with those the sensitivity of cyclooxygenase to aspirin by a completely presented in Fig. 3 indicates that the two methods give similar different method (Table 1). Aortas from different cadavers synthesized 1.3-11.8 ng of PGI2 per g wet weight, as determined by comparison with a PGI2 standard. PGI2 formed by shredded aorta was about 10% of that produced by fresh rings of small branches of human gastric or colonic arteries (16). This 30,000 could be due either to a difference between aorta and these arteries or to differences in the experimental design. Although PGI2 production is not a direct measure of cyclooxygenase activity, we have observed that the total amount produced at 1 0,000F a given arachidonate concentration is proportional to the amount of cyclooxygenase. We calculate a mean potency of aspirin for human aortic cyclooxygenase of 85 M-1 min-1, which is in agreement with that determined by acetylation with 3,0001 [acetyl-3H]aspirin. More importantly, when human aortic tissue

.C 1,0001F Table 1. Aspirin inhibition of PGI2 synthesis in human aorta Cyclo- PGI2 oxygenase Poten- 0 0. 3001 Aspirin produced, activity cy, treatment, ng/g wet remaining, M-1 Exp. min AT5o wt tissue % min1

*-. 1 0 10.45 7.9 100 1001- 80 2.90 2.2 28 80 2 0 2.83 1.3 100 *- 35 1-77 0.8 63 69 *- 301 3 0 3.26 1.3 100 60 1.36 0.5 42 69 4 0 6.08 6.6 100 40 2.82 3.1 46 98 Coronary Aorta Disrupted I ntact 0 9.13 11.8 100 platelets platelets 5 artery 35 4.24 5.4 46 110 FIG. 3. Aspirin sensitivity of cyclooxygenase from human platelets, aortas, and coronary arteries. Each point represents a single Samples of shredded human aorta were incubated with or without determination of cyclooxygenase acetylation measured by sodium 200MAM aspirin, and the capacity to synthesize PGI2 from exogenous dodecyl sulfate/polyacrylamide gel electrophoresis at a specific time arachidonic acid was determined. The percent cyclooxygenase activity and aspirin concentration. The fraction enzyme unacetylated was remaining was calculated as % activity = (zAT50 for aspirin-treated determined as 1 - (cpm observed/cpm at maximum acetylation). Log aorta/AT5o for control aorta) X 100%. Potency was determined by (fraction unacetylated) was plotted versus time and the slope was using the fraction of cyclooxygenase activity remaining as described considered as kapparent. Potency = (kapparent/aspirin concentration)- in the legend for Fig. 3. Downloaded by guest on September 27, 2021 5184 Medical Sciences: Burch et al. Proc. Natl. Acad. Sci. USA 75 (1978) was incubated for 35 min with 10 AM aspirin, conditions under vascular enzyme would require an aspirin dose of 32 g per day. which platelet cyclooxygenase is over 99% inactivated, PGI2 Assuming that intact aorta is proportionally more sensitive to production was not inhibited. aspirin than microsomes as is true in platelets, then 8 g of aspirin per day might be required to inhibit arterial cyclooxy- DISCUSSION genase. We have demonstrated that, under the conditions described The effectiveness of aspirin in altering the natural history of above, aspirin acetylates a single major protein in the micro- diseases such as myocardial infarction or stroke can only be somal fraction of human aorta and coronary artery homoge- assessed by clinical trials. If aspirin proves to be a useful drug nates. This protein has characteristics that identify it as fatty in such diseases, a dose of aspirin sufficient to inhibit platelet acid cyclooxygenase: it comigrates on sodium dodecyl sulfate function yet not interfere with PGI2 production in vascular gel electrophoresis with human platelet cyclooxygenase; its tissue is probably obtainable. acetylation is specific (i.e., saturable); and its acetylation is in- We thank the Washington University Clinical Unit for Recent Ex- hibited approximately 50% by the cyclooxygenase inhibitors pirations for assistance in procuring aortas and coronary arteries. This indomethacin, flurbiprofen, and ibuprofen. These inhibitors research was supported by Grants HL 14147 (Specialized Center for will fully inhibit cyclooxygenase in the particulate fraction of Research in Thrombosis) and HL 16634 from the National Institutes human platelets; the reason only 50% inhibition of acetylation of Health. is seen in aortic tissue may be due to the high aspirin concen- 1. Mielke, C. H., Jr., Kaneshiro, M. M., Maher, I. A., Weiner, J. M. trations and long incubation times needed to acetylate the en- & Rapaport, S. I. (1969) Blood 34, 204-215. zyme. For example, indomethacin is a reversible inhibitor of 2. O'Brien, J. R. (1968) Lancet i, 779-783. cyclooxygenase (17); its ability to block acetylation can be 3. Elwood, P. C., Cochrane, A. L., Burr, M. L., Sweetnam, P. M., Williams, G., Welsby, E., Hughes, S. J. & Renton, R. (1976) Br. overcome by high aspirin concentrations and long incubation Med. J. 1, 436-440. times (17). 4. Boston Collaborative Drug Surveillance Group (1974) Br. Med. The aortic cyclooxygenase is acetylated more slowly than the J. 1, 440-443. platelet enzyme (Fig. 2). Refractoriness of vessel wall cy- 5. The Coronary Drug Project Research Group (1976) J. Chronic clooxygenase has also been suggested by experiments studying Dis. 29, 625-642. experimental thrombosis in rabbits treated with intravenous 6. The Anturane Reinfarction Trial Research Group (1978) N. Engl. aspirin (18). Previous evidence (summarized in ref. 12) has J. Med. 298,289-295. suggested that the high suceptibility to aspirin of platelet cy- 7. Marcus, A. J. (1977) N. Engl. J. Med. 297,1284-1285. be Thus diflunisal 8. Moncada, S. & Vane, J. R. (1977) in Biochemical Aspects of clooxygenase may unique. [5-(2,4-difluo- Prostaglandins and Thromboxanes, eds. Kharasch, N. & Fried, rophenyl)salicylic acid], an antiinflammatory agent closely J. (Academic, New York), pp. 155-177. related to aspirin, inhibits cyclooxygenase from platelets and 9. Smith, J. B. & Willis, A. L. (1971) Nature (London) New Biol. sheep seminal vesicles equally, while aspirin is 40-fold more 231,235-237. potent in inhibiting the platelet enzyme (19). 10. Ferreira, S. H., Moncada, S. & Vane, J. R. (1971) Nature (Lon- If PGI2 functions in vivo as a naturally occurring an- don) New Biol. 231, 237-239. tithrombotic agent, therapeutic attempts to alter platelet 11. Roth, G. J. & Majerus, P. W. (1975) J. Clin. Invest. 56, 624- function by modulation of prostaglandin and thromboxane 632. production should, ideally, be designed to preserve vascular 12. Burch, J. W., Stanford, N. & Majerus, P. W. (1978) J. Clin. Invest. 61,314-319. production of PGI2. This has led Moncada and Vane (8) to 13. Baenziger, N. L., Dillender, M. J. & Majerus, P. W. (1977) Bio- suggest that cyclooxygenase inhibitors may not be ideal anti- chem. Biophys. Res. Commun. 78,294-301. platelet drugs. Our studies demonstrate that this objection to 14. Johnson, R. A., Morton, D. R., Kinner, J. H., Gorman, R. R., the use of aspirin as an antiplatelet drug is probably invalid. McGuire, J. C., Sun, F. F., Whittaker, N., Bunting, S., Salmon, That the decreased sensitivity of aortic cyclooxygenase is a J., Moncada, S. & Vane, J. R. (1976) Prostaglandins 12, 915- property of the tissue and not due to the experimental proce- 928. dure is suggested both by the fact that platelets processed in a 15. Roth, G. J., Stanford, N. & Majeris, P. W. (1975) Proc. Natl. Acad. manner similar to aorta remain more sensitive to aspirin and Sci. USA 72,3073-3076. by the fact that aspirin is equally potent in inactivating cy- 16. Moncada, S., Higgs, E. A. & Vane, J. R. (1977) Lancet i, 18- re- 20. clooxygenase from lamb aorta frozen immediately after 17. Stanford, N., Roth, G. J., Shen, T. Y. & Majerus, P. W. (1977) moval and lamb aorta stored and processed as described here Prostaglandins 13, 669-675. (data not shown). Aspirin, 160 mg per day, will completely 18. Kelton, J., Carter, C., Buchanan, M. R. & Hirsh, J. (1978) Clin. inhibit platelet cyclooxygenase in human subjects (12). As- Res. 26, 350A. suming that the sensitivity of aortic microsomes reflects that 19. Majerus, P. W. & Stanford, N. (1977) Br. J. Clin. Pharmacol. 4, of whole tissue in vivo, a similar degree of inhibition of the 15S-18S. Downloaded by guest on September 27, 2021