The FASEB Journal • Research Communication
Stronger inhibition by nonsteroid anti-inflammatory drugs of cyclooxygenase-1 in endothelial cells than platelets offers an explanation for increased risk of thrombotic events Jane A. Mitchell,* Ruth Lucas,* Ivana Vojnovic,† Kamrul Hasan,* John R. Pepper,* and Timothy D. Warner†,1 *Cardiothoracic Pharmacology, Unit of Critical Care Medicine, Royal Brompton Hospital, Imperial College School of Medicine, London, UK; and †The William Harvey Research Institute, Barts and the London, Queen Mary’s School of Medicine and Dentistry, London, UK Recent data have suggested that regular consumption of nonsteroid anti-inflammatory drugs (NSAIDs), particularly selective inhibitors of cyclooxygenase-2 (COX-2), is associated with an increased risk of thrombotic events. It has been suggested that this is due to NSAIDs reducing the release from the endothelium of the antithrombotic mediator prostaglandin I2 as a result of inhibition of endothelial COX-2. Here, however, we show that despite normal human vessels and endothelial cells containing cyclooxygenase-1 (COX-1) without any detectable COX-2, COX-1 in vessels or endothelial cells is more readily inhibited by NSAIDs and COX-2-selective drugs than COX-1 in platelets (e.g., log IC50ⴞSEM values for endothelial cells vs. platelets: naproxen –5.59ⴞ0.07 vs. – 4.81ⴞ0.04; rofecoxib – 4.93ⴞ0.04 vs. –3.75ⴞ0.03; nⴝ7). In broken cell preparations, the selectivities of the tested drugs toward endothelial cell over platelet COX-1 were lost. These observations suggest that variations in cellular conditions, such as endogenous peroxide tone and substrate supply, and not the isoform of cyclo-oxygenase present, dictate the effects of NSAIDs on endothelial cells vs. platelets. This may well be because the platelet is not a good representative of COX-1 activity within the body as it produces prostanoids in an explosive burst that does not reflect tonic release from other cells. The results reported here can offer an explanation for the apparent ability of NSAIDs and COX-2-selective inhibitors to increase the risk of myocardial infarction and stroke.—Mitchell, J. A., Lucas, R., Vojnovic, I., Hasan, K., Pepper, J. R., Warner, T. D. Stronger inhibition by nonsteroid anti-inflammatory drugs of cyclooxygenase-1 in endothelial cells than platelets offers an explanation for increased risk of thrombotic events. FASEB J. 20, 2468 –2475 (2006) ABSTRACT
Key Words: prostacyclin 䡠 thromboxane A2 䡠 thrombosis 䡠 cyclooxygenase-2 Nonsteroid anti-inflammatory drugs (NSAIDs) such as aspirin, ibuprofen, naproxen, and rofecoxib 2468
have, in one form or another, been used for many thousands of years for the relief of inflammation and pain. However, NSAIDs can also produce a number of well-characterized side effects. In particular, their use is linked to an increased risk of gastrointestinal adverse events, the most serious being obstructions, perforations, ulcers, and bleeds (1). In the U.S. the annual death rate from NSAID-induced gastrointestinal bleeds is estimated to be 0.08% (2). NSAIDs owe their therapeutic benefits, as well as their side effects, to inhibition of the enzyme cyclo-oxygenase (COX; refs. 3, 4). COX exists in two isoforms: COX-1 and COX-2. In general terms, COX-1 is the constitutive isoform present in many tissues associated with the production of prostanoids for physiological functions, whereas COX-2 is the inducible isoform that appears at sites of inflammation and is associated with the production of prostanoids involved in inflammatory responses. In the last decade or so COX-2-selective drugs such as celecoxib, rofecoxib, and lumiracoxib have been produced with the aim of targeting COX-2 in inflammatory sites while leaving COX-1 largely unaffected. This is consistent with the idea that inhibition of COX-1 underlies the gastrointestinal side effects of NSAIDs and that NSAID drug selectivity toward inhibition of COX-1 over COX-2 correlates with their ability to cause gastrointestinal side effects (5). Indeed, in controlled clinical trials selective inhibitors of COX-2, including lumiracoxib and rofecoxib, produce fewer severe gastrointestinal adverse events than traditional (nonselective) NSAIDs (3, 4). COX products are important to the normal functioning of the cardiovascular system, particularly prostaglandin (PG) I2 produced by endothelial cells and thromboxane (Tx) A2 formed by platelets (3, 4). PGI2 1 Correspondence: The William Harvey Research Institute, Barts and the London, Queen Mary’s School of Medicine and Dentistry, Charterhouse Square, London EC1M 6BQ, UK. E-mail:
[email protected] or
[email protected] doi: 10.1096/fj.06-6615com
0892-6638/06/0020-2468 © FASEB
is antithrombotic and vasodilating; it also reduces vascular remodeling and limits cholesterol uptake (6, 7). In contrast, TxA2 is prothrombotic and vasoconstricting, and promotes vascular remodeling (6, 7). Drugs that inhibit TxA2 production therefore reduce platelet reactivity whereas drugs that inhibit PGI2 production increase it. In clinical use, low-dose aspirin works by inhibiting COX in platelets and reducing TxA2 production while having relatively less effect on PGI2 production. This profile of activity is associated with an increase in bleeding time and a reduction in the risk of heart attack and stroke (8 –10). In contrast, in humans COX-2-selective drugs reduce whole-body production of PGI2 while leaving TxA2 much less affected (11, 12). Because of this ability to reduce whole-body production of antithrombotic PGI2, supported by observations that COX-2 knockout mice have greatly reduced levels of urinary PGI2 metabolite (13), it has been suggested that the COX-2-selective drugs could increase the risk of thrombotic events. This hypothesis is supported particularly by the increased rate of thrombotic events recorded in placebo-controlled trials of COX-2-selective inhibitors (14, 15). The mechanisms underlying the inhibition of PGI2 production by COX-2-selective drugs have not actually been directly identified, as it has been assumed that this effect alone demonstrates the predominance of COX-2 in endothelial cells. However, with the notable exception of the kidneys and sometimes lungs, COX-2 expression is rarely detected in either large conduit vessels or microvessels such as those in the brain, breast, gut, joint, liver, mouth, prostate, or skin in the absence of local inflammation or disease (4, 16 –26). Furthermore, observational studies suggest that similar to COX-2-selective drugs, traditional NSAIDs such as ibuprofen and diclofenac may also increase the risk of cardiovascular events (4, 27–30). We hypothesized, therefore, that rather than differences in COX isoform, it could be differences in intracellular conditions, such as the relative levels of lipid peroxides, that are at a much higher level in platelets than in the endothelium (31, 32), or the supply of arachidonic acid substrate (33, 34) that influenced NSAID potencies. In the current study we used human blood vessels, human endothelial cells, and human platelets and have taken a classical pharmacological approach to show that despite endothelial cells containing COX-1, their production of PGI2 is inhibited by both COX-2-selective inhibitors and traditional NSAIDs more strongly than the production of TxA2 by platelets. We suggest that by examining COX activity in endothelial cells, platelets, and cells expressing COX-2, it will be possible to identify drugs with the desired anti-inflammatory and analgesic properties of traditional NSAIDs but without the accompanying risk of gastrointestinal and cardiovascular side effects. NSAIDs IN ENDOTHELIAL CELLS
MATERIALS AND METHODS COX activity in combination assay of human endothelial cells and platelets Primary cultures of human aortic endothelial cells were cultured in 96-well plates for COX activity experiments and in 6-well plates for the generation of samples for Western blot. For COX activity assays, freshly taken human whole blood, platelet-rich plasma (PRP; prepared by centrifugation of whole blood, 200 g, at room temperature for 7 min), aspirintreated (10⫺4 M, 6 h), whole blood, or culture medium was added to the endothelial cells. In some experiments whole blood was added to empty 96-well plates. Test agents (all 10⫺10 to 10⫺3 M) or vehicle (0.1% v/v dimethyl sulfoxide) were then added and the plates were incubated for 60 min, followed by addition of the calcium ionophore, A23187 (5⫻10⫺5 M). After a further 30 min incubation period, samples were centrifuged (1500 g, 4°C, 5 min) and the incubation medium was removed and immediately frozen. Concentrations of 6-keto-PGF1␣ (as a measure of PGI2 formation within the endothelial cells) and TxB2 (as a measure of TxA2 formation by platelets) in samples were then determined by RIA. COX activity in homogenates of human endothelial cells and human platelets One T75 flask each of confluent cultures of human aortic endothelial cells was washed with PBS, then scraped into ice-cold assay buffer (PBS containing EGTA, 10⫺3 M, and phenylmethysulfonylfluoride, 10⫺3 M). Cells were then homogenized with 10 strokes in a glass TeflonTM homogenizer in 7 ml of assay medium. To prepare platelet homogenates, 50 ml of human whole blood was collected into trisodium citrate (3.15% w/v) and centrifuged (200 g, room temperature, 7 min). The platelet supernatant was then removed, treated with 300 ng/ml PGI2, and centrifuged (1000 g, 15 min, room temperature). The platelet pellet was resuspended in 7 ml of assay buffer and homogenized as for endothelial cells; 100 l of homogenate was then added to individual wells of a 96-well plate together with 5 ⫻ 10⫺3 M of epinephrine and test agents (all 10⫺10 to 10⫺3 M) or vehicle (0.1% v/v dimethyl sulfoxide). Plates were then incubated for 1 h before addition of 3 ⫻ 10⫺5 M arachidonic acid and incubation for another 15 min. Diclofenac 10⫺3 M was then added and the samples centrifuged (1500 g, 4°C, 5 min); the incubation medium was removed and immediately frozen. Concentrations of PGE2 and 6-keto-PGF1␣ (as a measure of PGI2 formation within the endothelial cells) or TxB2 (as a measure of TxA2 formation by platelets) in samples were then determined by RIA. COX activity in whole human vessels Human saphenous vein was obtained during coronary artery bypass surgery. Informed consent was obtained from patients under the approval of the Royal Brompton Hospital Research Ethics Committee. Vessels were cleared of connective tissue and dissected into sections ⬃ 5 mm wide. Care was taken to preserve endothelial integrity. Segments of vessel were incubated for 60 min in plasma derived from the patient’s blood in the presence of either aspirin (10⫺4M), ibuprofen (1.5⫻10⫺4 M), rofecoxib (3⫻10⫺6 M), or vehicle. After 60 min COX activity was stimulated by the addition of arachi2469
donic acid (3⫻10⫺5 M) and incubation for another 30 min. Plasma was removed from the incubates and production of PGI2 was determined by RIA as above. To control for the carryover of prostanoids from tissue preparation, some incubates were made in the presence of diclofenac 10⫺3M. Vessels demonstrating high levels of immunoreactive 6-keto-PGF1␣ under such conditions were excluded from further analyses. Effect of hydroperoxides on NSAID inhibition of endothelial cell PGI2 production To test the effects of hydroperoxides on the potencies of NSAIDs against endothelial cell PGI2 production, primary human endothelial cells were cultured in 96-well plates. Once confluent, medium was replaced with fresh medium with the addition of test drugs, acetaminophen (10⫺4 M), aspirin (10⫺4 M), ibuprofen (10⫺4 M), rofecoxib (3⫻10⫺6 M), or vehicle together with the cell-permeable organic hydroperoxide, t-butylOOH (10⫺4 M), for 60 min. Calcium ionophore (A23187, 5⫻10⫺5 M) was then added and incubation continued for 30 min, then the reaction was stopped by the addition of diclofenac (10⫺3 M). Conditioned medium was removed to measure 6-keto-PGF1␣ as above. Western blot analysis for cyclo-oxygenase-1 and cyclooxygenase-2 Western blot analysis was performed as described previously (32). Cells were either left untreated or incubated for 24 h with interleukin (IL)-1 (1 ng ml⫺1) to induce COX-2 protein expression in 6-well culture plates. Cells were washed twice with cold PBS before protein was extracted into buffer (Tris 5⫻10⫺3 M, ethylenediaminetetraacetic acid 10⫺2 M, Triton X-100 1% v/v, phenylmethylsulfonyl fluoride 10⫺3 M). The resulting cell extract was boiled with gel loading buffer containing 2 mg ml⫺1 bromphenol blue in a ratio of 1:1. Approximately 20 g of protein as determined by Bradford protein assay was loaded onto a 7.5% SDS separating gel. After electrophoretic separation (1 h at 100 V; Bio-Rad, Hercules, CA, USA), samples were transferred to nitrocellulose (1 h at 100 V) and probed with a specific COX-1 or COX-2 antibody (Ab) raised in rabbit. The blot was then incubated with anti-rabbit immunoglobulin (Ig) E (raised in goat) linked to horseradish peroxidase and visualized using enhanced chemiluminescence (Amersham, Bucks, U.K.).
RESULTS COX expression in human aortic endothelial cells Western blot analysis demonstrated that primary cultures of human aortic endothelial cells contained COX-1, with no detectable expression of COX-2 (Fig. 1). For control, incubation of endothelial cells with IL-1 (10 ng ml⫺1, 24 h) was found to induce the expression of readily detectable COX-2. Human platelets contained detectable levels of COX-1 but not COX-2. Effects of NSAIDs on production of PGI2 and TxA2 in endothelial cell platelet combination assay In control incubations, whole blood together with endothelial cells produced 28.02 ⫾ 1.43 ng ml⫺1 TxB2 after addition of A23187; PRP incubated with endothelial cells produced 28.19 ⫾ 1.03 ng ml⫺1 TxB2 and endothelial cells incubated with medium produced 0.68 ⫾ 0.10 ng ml⫺1 TxB2 (n⫽21 for each). Incubation of whole blood with aspirin (10⫺4 M, 6 h) reduced the production of TxB2 by ⬎95% (1.3⫾0.2 ng ml⫺1 TxB2; n⫽18). After exposure to A23187, endothelial cells released 8.48 ⫾ 0.50 ng ml⫺1 6-keto-PGF1␣ when incubated with medium, 8.08 ⫾ 0.50 ng ml⫺1 when incubated with blood, and 8.62 ⫾ 0.76 ng ml⫺1 when incubated with PRP (n⫽21). As PRP had an equally neutral effect as blood on endothelial production of 6-keto-PGF1␣ and since neutrophils within whole blood can produce PGI2 and influence platelet activity (35; unpublished observations), additional studies were conducted using a combination of endothelial cells and PRP. In these combination assays, addition of aspirin, celecoxib, ibuprofen, naproxen, or rofecoxib caused concentration-dependent reductions in the production of 6-keto-PGF1␣ and TxB2 (Fig. 2; Table 1). All compounds tested inhibited production of 6-keto-PGF1␣ more potently than the production of TxB2, particularly celecoxib and rofecoxib. The rank ordering of potency of the drugs tested on COX-1 in the platelet was aspirin ⬎ naproxen ⬎ ibuprofen ⬎ rofecoxib ⫽ celecoxib; the rank ordering of potency of the drugs on COX-1 in endothelial cells was aspirin ⬎ celecoxib ⫽
Materials Human aortic endothelial cells were purchased from TCS CellWorks Ltd. (Bucks, UK). Arachidonic acid sodium salt, calcium ionophore (A23187), epinephrine, ibuprofen, indomethacin, and naproxen were from Sigma Chemical Co. (Poole, Dorset, UK) as were antibodies to 6-ketoPGF1␣, TxB2 (stable breakdown products of PGI2 and TxA2, respectively) and PGE2. Antibodies to COX-1 and COX-2 were purchased from Cayman Chemicals (Ann Arbor, MI, USA). [3H]-6KetoPGF1␣, [3H]-PGE2 and [3H]-TxB2 were obtained from Amersham International (Amersham, Bucks, UK). IL-1 was purchased from R&D Systems (Abingdon, UK). Celecoxib and rofecoxib were gifts from Boehringer-Ingelheim KG (Ingelheim am Rhein, Germany). 2470
Vol. 20
December 2006
Figure 1. Expression of COX-1 and COX-2 in 3 samples of human aortic endothelial cells cultured under control conditions (lanes c1, c2, c3) or after incubation with IL-1 (10 ng/ml, 24 h) (lanes i1, i2, i3).
The FASEB Journal
MITCHELL ET AL.
Figure 2. Inhibition in combination assays (human aortic endothelial cells and PRP incubated together) by aspirin (A), ibuprofen (B), naproxen (C), celecoxib (D), and rofecoxib (E) of endothelial cell PGI2 production (unfilled squares, measured as 6-keto-PGF1␣) and platelet TxA2 (filled squares, measured as TxB2). Data are means ⫾ sem of n ⫽ 7 observations for each point. Broken lines and asterisks show published data for inhibition of COX-2 by IL-1-treated A549 cells (see ref. 51).
ibuprofen ⫽ naproxen ⬎ rofecoxib (Fig. 2, Table 1). The rank order of potencies previously demonstrated for these compounds against COX-2 within IL-1treated A549 cells is rofecoxib ⫽ celecoxib ⬎ aspirin ⬎ ibuprofen ⬎ naproxen (IC50 values respectively 0.31, 0.34, 2.13, 19.56, and 35.26⫻10⫺6 M; Fig. 2).
6-keto-PGF1␣, n⫽13) was inhibited by ⱖ 50% in the presence of aspirin (10⫺4M), ibuprofen (1.5⫻10⫺4M), or rofecoxib (3⫻10⫺6M).
Effects of NSAIDs on production of PGI2 and TxA2 in homogenates of endothelial cells and platelets
Addition of t-butylOOH to endothelial cell incubates for 60 min reduced the inhibition of 6-keto-PGF1␣ production caused by acetaminophen (10⫺4 M) or rofecoxib (3⫻10⫺6 M) but did not affect the inhibition caused by aspirin (10⫺4 M) or ibuprofen (10⫺4 M) (Fig. 5).
Similar to observations made with intact cells, each NSAID tested caused concentration-dependent inhibitions of COX activity in the homogenate preparations. The differences in potency of the NSAIDs between COX-1 in endothelial cells and COX-1 in platelets was lost following homogenization (Fig. 3). Characterization of cyclo-oxygenase in human blood vessels We confirmed our observations from studies with endothelial cells using intact human blood vessels. Western blot analysis demonstrated that human internal mammary artery, saphenous vein, or radial artery contained detectable levels of COX-1 but not COX-2 (Fig. 4A). While human blood vessels did not contain any detectable COX-2, the basal production of PGI2 by vessel segments incubated in plasma (3.46⫾1.42 ng TABLE 1. IC50 values for drugs against the production of 6-ketoPGF1␣ and TxB2 in coincubations of human aortic endothelial cells and platelet rich plasma (n⫽7 for all) 6-keto-PGF1␣ Drug
Aspirin Ibuprofen Naproxen Celecoxib Rofecoxib
TxB2
m
log M ⫾ se
m
log M ⫾ se
0.28 1.62 2.54 1.32 11.71
–6.56 ⫾ 0.05 –5.79 ⫾ 0.03 –5.59 ⫾ 0.07 –5.88 ⫾ 0.07 –4.93 ⫾ 0.04
4.35 30.57 15.50 207 178
–5.36 ⫾ 0.03 –4.52 ⫾ 0.03 –4.81 ⫾ 0.04 –3.68 ⫾ 0.04 –3.75 ⫾ 0.03
NSAIDs IN ENDOTHELIAL CELLS
Effect of hydroperoxides on NSAID inhibition of endothelial cell PGI2 production
DISCUSSION The potential of NSAIDs, including COX-2 selective drugs, to increase the risk of heart attack and stroke has become a recent worldwide concern (3, 4). Indeed, this risk is now receiving more attention than the wellcharacterized ability of NSAIDs to precipitate gastric ulcers and other gastrointestinal adverse events. When selective inhibitors of COX-2 were developed, their potential to cause cardiovascular side effects was not considered; the main objective was to produce drugs that retained the anti-inflammatory and analgesic effects of traditional NSAIDs without their gastrointestinal side effects. Newer drugs such as rofecoxib were therefore tested in animal and human tests of NSAID efficacy and NSAID-induced gastrointestinal damage, but were not examined in models of cardiovascular safety (3, 4, 36, 37). However, the latest available findings indicate that we must now extend our range of concerns about NSAID toxicity and include cardiovascular risk as an element in the development of new drugs. In the current study we have modeled in vitro the local environment of blood and blood vessel in vivo. With the exception of studies examining the effects of t-butylOOH, we have used human plasma as an incubation medium to take account of the influences of blood proteins on NSAID potency (38). We have established in our primary endothelial cells or in human 2471
Figure 3. Inhibition by aspirin (A), ibuprofen (B), naproxen (C), celecoxib (D), and rofecoxib (E) of prostanoid production by homogenates of: endothelial cells, PGI2 production (unfilled squares, measured as 6-keto-PGF1␣); platelets, TxA2 production (filled squares, measured as TxB2). Data are means ⫾ sem of n ⫽ 4 – 6 observations for each point.
vessels removed surgically and immediately snap-frozen that COX-1 is the predominant isoform present. Indeed, we could not detect the presence of COX-2 protein in any of our cell or tissue samples in the absence of exogenously applied proinflammatory stim-
Figure 4. A) Expression of COX-1 and COX-2 in 2 samples of human internal mammary artery (IMA), 2 samples of human internal mammary artery (IMA), and 1 sample of human radial artery compared with the human megakaryocyte cell line (MEG-01) and A549 cells incubated with IL-1b (10 ng/ml, 24 h). B) Inhibition by aspirin (10⫺4M), ibuprofen (1.5⫻10⫺4M), or rofecoxib (3⫻10⫺6M) of PGI2 production (measured as 6-keto-PGF1␣) by segments of human blood vessel. Data are means ⫾ sem of n ⫽ 14 –15 observations for each column. 2472
Vol. 20
December 2006
uli. Despite the absence of detectable COX-2 in our endothelial cells, we found that selective inhibitors of COX-2 and traditional NSAIDs reduced their production of PGI2 more potently than platelet production of TxA2 occurring at the same time in the same incubate. Modeling plasma concentrations resulting from administration of standard doses of the five test drugs in humans (39 – 43) into our system suggested a general level of inhibition of PGI2 production (67⫾12%, n⫽5) that accords well with those reported from in vivo studies (3, 4). Similar modeling shows that at standard dosing, the calculated inhibition of TxA2 production varies from ⬎75% for ibuprofen and naproxen to ⬍10% for celecoxib and rofecoxib. It is important to reinforce the observation that although all the drugs tested inhibited endothelial PGI2 production much more actively than platelet TxA2 production selectivity, this could not be accounted for by endothelial cell expression of COX-2, as none was detected by Western blot analysis. Furthermore, as discussed below, the selectivity of drugs for endothelial cell COX-1 is lost when “local environment ” is removed by homogenization. Finally, if endothelial cell PGI2 production was
Figure 5. Effects of t-butyl-OOH (10⫺4 M, filled bars) on the inhibition under control conditions (unfilled bars) of PGI2 production (measured as 6-keto-PGF1␣) caused by acetaminophen (10⫺4 M), aspirin (10⫺4 M), ibuprofen (10⫺4 M), and rofecoxib (3⫻10⫺6 M) in human aortic endothelial cells. Data are n ⫽ 3 from 1 experiment representative of 4 separate experiments.
The FASEB Journal
MITCHELL ET AL.
underpinned by the activity of COX-2, this would have been matched by the pharmacological selectivity of the compounds used (i.e., neither the individual potencies of the compounds nor the relative potencies accord with COX-2 underlying the PGI2 production we detected). This is a key point as it indicates that drugs could be developed that are selective for COX-2 at the site of inflammation without affecting COX-1 in the endothelium. It has been proposed that endothelial cells in culture, as used in this study, lack COX-2 because they are not exposed to the mechanical forces such as shear that are constantly present within the blood vessel lumen. However, shear of endothelial cells in vitro causes complex changes, including both increases and decreases, in the expressions of COX-1 and COX-2 (4, 5, 44); the COX-1 gene even contains the sequence of shear stress-responsive elements within its promoter region (45). In fact, the constant shear to which endothelial cells within the body are constantly exposed promotes a reduction in proliferation and metabolic rate that is consistent with a resting phenotype. Furthermore, exposure of endothelial cells in vitro to shear for 7 days is not associated with any increase in COX-1 or COX-2 gene transcription (46). As outlined above, immunohistochemistry provides little evidence for COX-2 expression within the majority of healthy blood vessels. To extend our studies using primary human endothelial cells in culture, we used healthy human blood vessels to more closely model the in vivo production of PGI2 by the circulatory system. We found that in these vessels, as in the primary cultured endothelial cells, both traditional NSAIDs and COX-2-selective compounds inhibited the production of 6-keto-PGF1␣ despite the absence of detectable COX-2 protein. Notably, in COX-1 knockout mice, celecoxib fails to increase blood pressure as it does in wild-types (13). Increases in blood pressure caused by COX-2-selective drugs and traditional NSAIDs could be a causative factor underlying increases in thrombotic events (47), and it would be intriguing if these are linked to inhibition of COX-1 rather than COX-2. It is interesting that we found aspirin to be more active on endothelial than platelet COX-1 since, as determined by urinary measurements, aspirin affects whole-body TXA2 production more strongly than PGI2 (8). However, the effects of aspirin on different parts of the circulation are strongly influenced by its pharmacokinetics (i.e., aspirin’s short circulating half-life), and these are important contributors to its platelet selectivity (48). Furthermore, others have reported during skin bleeding that local vascular productions of PGI2 and TXA2 are equally sensitive to inhibition by low-dose aspirin (49). Taking account of aspirin’s instability, this could indicate selectivity toward the vascular wall. The reason why COX-1 in endothelial cells is more sensitive to inhibition by some NSAIDs than COX-1 within the platelet is not clear and is largely beyond the scope of this current study. However, it may well be that the differences in intracellular environment and subNSAIDs IN ENDOTHELIAL CELLS
strate supply between the endothelium and the platelet dictate the different potencies of these drugs. This would explain the loss in endothelial cell selectivity found in our experiments using broken cell preparations. When endothelial cells and platelets are treated together with NSAIDs (as would be the case in vivo), the production of PGI2 is inhibited more strongly that that of TxA2. One potential intracellular factor of importance is the relative levels of lipid peroxides in platelets vs. the endothelium. We know from studies using acetaminophen that platelets have a much higher level of lipid peroxides than the endothelium (31, 32). Furthermore, we recently showed that the potencies of other NSAIDs, including rofecoxib, are also reduced when the platelet environment is modeled by increased endogenous lipid peroxides in cells expressing COX-2 (32). To test this possibility in our endothelial cells, we examined the effects of the cell-permeable organic hydroperoxide, t-butylOOH against the effects of aspirin (an irreversible COX inhibitor), ibuprofen (a traditional NSAID), rofecoxib (a COX-2-selective NSAID), and acetaminophen (as a positive control; ref. 31). These experiments demonstrated that in our COX-1containing endothelial cells the inhibitory effects of rofecoxib, but not aspirin or ibuprofen, were significantly reduced when endogenous lipid peroxide levels were elevated. Thus, the large disparity in endogenous lipid peroxide levels in endothelial cells vs. platelets could explain endothelial selectivity for some of the NSAIDs tested, and why these compounds and acetaminophen have strong inhibitory effects on the production of PGI2 but not TxA2 in vivo (11, 12, 50). Intracellular levels of the COX substrate arachidonic acid may also influence the potencies of NSAIDs as many act against COX-2 as simple, competitive substrate inhibitors (33). Indeed, in low substrate conditions even highly selective COX-2 inhibitors can inhibit the activity of COX-1 (34). It may well be, therefore, that despite their experimental attractiveness, platelets are poor representatives of COX-1 activity within the body. Activated platelets make large amounts of TxA2 in a dramatic burst, with accompanying high levels of substrate and high levels of lipid peroxides. PGI2 and other COX-1 products are more often produced tonically with lower accompanying substrate and lipid peroxide levels. This suggests we must be cautious in attributing differences in NSAID potency against PGI2 and TxA2 production in vivo to drug selectivity when enzyme environment can vary so profoundly. In summary, we show that despite expressing COX-1 in a similar manner without detectable COX-2, endothelial cells and platelets are differentially sensitive to the inhibitory effects of a number of NSAIDs such that endothelial cell production of PGI2 is inhibited more actively than platelet production of TxA2. This observation can offer an explanation as to why NSAIDs, including rofecoxib and celecoxib, might promote a prothrombotic environment. Finally, systems such as those described could assist in the development of 2473
NSAIDs that spare both the stomach and the endothelium.
16.
This research was supported by grants from the British Heart Foundation, a Fellowship grant from the Spanish Government, and European Community FP6 funding (LSHM-2004 – 0050333, Eicosanox). This publication reflects only the authors’ views. The European Community is not liable for any use that may be made of information herein.
17.
18.
19.
REFERENCES 1. 2.
3.
4. 5. 6. 7. 8.
9.
10.
11.
12.
13.
14.
15.
2474
Wallace, J. L. (1996) NSAID gastroenteropathy: past, present and future. Can. J. Gastroenterol. 10, 451– 459 Lazzaroni, M., and Bianchi Porro, G. (2004) Gastrointestinal side-effects of traditional non-steroidal anti-inflammatory drugs and new formulations. Aliment. Pharmacol. Ther. 20 (Suppl. 2), 48 –58 Grosser, T., Fries, S., and FitzGerald, G. A. (2006) Biological basis for the cardiovascular consequences of COX-2 inhibition: therapeutic challenges and opportunities. J. Clin. Invest. 116, 4 –15 Mitchell, J. A., and Warner, T. D. (2006) COX isoforms in the cardiovascular system: understanding the activities of non-steroidal anti-inflammatory drugs. Nat. Rev. Drug Discov. 5, 75– 86 Warner, T. D., and Mitchell, J. A. (2004) Cyclooxygenases: new forms, new inhibitors, and lessons from the clinic. FASEB J. 18, 790 – 804 Davidge, S. T. (2001) Prostaglandin H synthase and vascular function. Circ. Res. 89, 650 – 660 Dogne, J. M., Hanson, J., and Pratico, D. (2005) Thromboxane, prostacyclin and isoprostanes: therapeutic targets in atherogenesis. Trends Pharmacol. Sci. 26, 639 – 644 FitzGerald, G. A., Oates, J. A., Hawiger, J., Maas, R. L., Roberts, L. J., 2nd, Lawson, J. A., and Brash, A. R. (1983) Endogenous biosynthesis of prostacyclin and thromboxane and platelet function during chronic administration of aspirin in man. J. Clin. Invest. 71, 676 – 688 Johnson, E. S., Lanes, S. F., Wentworth, C. E., III, Satterfield, M. H., Abebe, B. L., and Dicker, L. W. (1999) A metaregression analysis of the dose-response effect of aspirin on stroke. Arch. Intern. Med. 159, 1248 –1253 Sanmuganathan, P. S., Ghahramani, P., Jackson, P. R., Wallis, E. J., and Ramsay, L. E. (2001) Aspirin for primary prevention of coronary heart disease: safety and absolute benefit related to coronary risk derived from meta-analysis of randomised trials. Heart 85, 265–271 McAdam, B. F., Catella-Lawson, F., Mardini, I. A., Kapoor, S., Lawson, J. A., and FitzGerald, G. A. (1999) Systemic biosynthesis of prostacyclin by cyclooxygenase (COX)-2: the human pharmacology of a selective inhibitor of COX-2. Proc. Natl. Acad. Sci. U. S. A. 96, 272–277 Catella-Lawson, F., McAdam, B., Morrison, B. W., Kapoor, S., Kujubu, D., Antes, L., Lasseter, K. C., Quan, H., Gertz, B. J., and FitzGerald, G. A. (1999) Effects of specific inhibition of cyclooxygenase-2 on sodium balance, hemodynamics, and vasoactive eicosanoids. J. Pharmacol. Exp. Ther. 289, 735–741 Cheng, Y., Wang, M., Yu, Y., Lawson, J., Funk, C. D., and Fitzgerald, G. A. (2006) Cyclooxygenases, microsomal prostaglandin E synthase-1, and cardiovascular function. J. Clin. Invest. 116, 1391–1399 Bresalier, R. S., Sandler, R. S., Quan, H., Bolognese, J. A., Oxenius, B., Horgan, K., Lines, C., Riddell, R., Morton, D., Lanas, A., et al. (2005) Adenomatous Polyp Prevention on Vioxx (APPROVe) Trial Investigators. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N. Engl. J. Med. 352, 1092–1102 Solomon, S. D., McMurray, J. J., Pfeffer, M. A., Wittes, J., Fowler, R., Finn, P., Anderson, W. F., Zauber, A., Hawk, E., and Bertagnolli, M.; Adenoma Prevention with Celecoxib (APC) Study Investigators. (2005) Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N. Engl. J. Med. 352, 1071–1080
Vol. 20
December 2006
20.
21.
22. 23.
24.
25.
26.
27.
28.
29.
30.
31.
32. 33. 34.
Komhoff, M., Grone, H. J., Klein, T., Seyberth, H. W., and Nusing, R. M. (1997) Localization of cyclooxygenase-1 and -2 in adult and fetal human kidney: implication for renal function. Am. J. Physiol. 272, F460 –F468 Baber, S. R., Champion, H. C., Bivalacqua, T. J., Hyman, A. L., and Kadowitz, P. J. (2003) Role of cyclooxygenase-2 in the generation of vasoactive prostanoids in the rat pulmonary and systemic vascular beds. Circulation 108, 896 –901 Iadecola, C., Forster, C., Nogawa, S., Clark, H. B., and Ross, M. E. (1999) Cyclooxygenase-2 immunoreactivity in the human brain following cerebral ischemia. Acta Neuropathol. (Berlin) 98, 9 –14 Costa, C., Soares, R., Reis-Filho, J. S., Leitao, D., Amendoeira, I., and Schmitt, F. C. (2002) Cyclo-oxygenase 2 expression is associated with angiogenesis and lymph node metastasis in human breast cancer. J. Clin. Pathol. 55, 429 – 434 Yu, H. G., Li, J. Y., Yang, Y. N., Luo, H. S., Yu, J. P., Meier, J. J., Schrader, H., Bastian, A., Schmidt, W. E., and Schmitz, F. (2003) Increased abundance of cyclooxygenase-2 correlates with vascular endothelial growth factor-A abundance and tumor angiogenesis in gastric cancer. Cancer Lett. 195, 43–51 Tabernero, A., Schneider, F., Potenza, M.A., Randriamboavonjy, V., Chasserot, S., Wolf, P., Mitolo-Chieppa, D., Stoclet, J. C., and Andriantsitohaina, R. (2003) Cyclooxygenase-2 and inducible nitric oxide synthase in omental arteries harvested from patients with severe liver diseases: immuno-localization and influence on vascular tone. Intensive Care Med. 29, 262–270 Kang, R.Y., Freire-Moar, J., Sigal, E., and Chu, C. Q. (1996) Expression of cyclooxygenase-2 in human and an animal model of rheumatoid arthritis. Br. J. Rheumatol. 35, 711–718 Graupera, M., Garcia-Pagan, J. C., Abraldes, J. G., Peralta, C., Bragulat, M., Corominola, H., Bosch, J., and Rodes, J. (2003) Cyclooxygenase-derived products modulate the increased intrahepatic resistance of cirrhotic rat livers. Hepatology 37, 172–181 Sonis, S. T., O’Donnell, K. E., Popat, R., Bragdon, C., Phelan, S., Cocks, D., and Epstein, J. B. (2004) The relationship between mucosal cyclooxygenase-2 (COX-2) expression and experimental radiation-induced mucositis. Oral Oncol. 40, 170 –176 Wang, W., Bergh, A., and Damber, J. E. (2005) Cyclooxygenase-2 expression correlates with local chronic inflammation and tumor neovascularization in human prostate cancer. Clin. Cancer Res. 11, 3250 –3256 O’Grady, A., O’Kelly, P., Murphy, G. M., Leader, M., and Kay, E. (2004) COX-2 expression correlates with microvessel density in non-melanoma skin cancer from renal transplant recipients and immunocompetent individuals. Hum. Pathol. 35, 1549 –1555 Hippisley-Cox, J., and Coupland, C. (2005) Risk of myocardial infarction in patients taking cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs: population based nested case-control analysis. Br. Med. J. 330, 1366 Andersohn, F., Suissa, S., and Garbe, E. (2006) Use of first- and second-generation cyclooxygenase-2-selective nonsteroidal antiinflammatory drugs and risk of acute myocardial infarction. Circulation 113, 1950 –1957 Kearney, P. M., Baigent, C., Godwin, J., Halls, H., Emberson, J. R., and Patrono, C. (2006) Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomised trials. Br. Med. J. 332, 1302–1308 Gislason, G. H., Jacobsen, S., Rasmussen, J. N., Rasmussen, S., Buch, P., Friberg, J., Schramm, T. K., Abildstrom, S. Z., Kober, L., Madsen, M., and Torp-Pedersen, C. (2006) Risk of death or reinfarction associated with the use of selective cyclooxygenase-2 inhibitors and nonselective nonsteroidal antiinflammatory drugs after acute myocardial infarction. Circulation In press Boutaud, O., Aronoff, D. M., Richardson, J. H., Marnett, L. J., and Oates, J. A. (2002) Determinants of the cellular specificity of acetaminophen as an inhibitor of prostaglandin H(2) synthases. Proc. Natl. Acad. Sci. U. S. A. 99, 7130 –7135 Lucas, R., Warner, T. D., Vojnovic, I., and Mitchell, J. A. (2005) Cellular mechanisms of acetaminophen: role of cyclo-oxygenase. FASEB J. 19, 635– 637 Gierse, J. K., Koboldt, C. M., Walker, M. C., Seibert, K., and Isakson, P. C. (1999) Kinetic basis for selective inhibition of cyclo-oxygenases. Biochem. J. 339, 607– 614 Riendeau, D., Percival, M. D., Boyce, S., Brideau, C., Charleson, S., Cromlish, W., Ethier, D., Evans, J., Falgueyret, J. P., Ford-
The FASEB Journal
MITCHELL ET AL.
35.
36.
37.
38. 39.
40. 41. 42.
43.
Hutchinson, A. W., et al. (1997) Biochemical and pharmacological profile of a tetrasubstituted furanone as a highly selective COX-2 inhibitor. Br. J. Pharmacol. 121, 105–117 Gonzalez-Correa, J. A., Arrebola, M. M., Guerrero, A., MunozMarin, J., Ruiz-Villafranca, D., Sanchez de La Cuestaqq, F., and De La Cruz, J. P. (2005) Influence of vitamin E on the antiplatelet effect of acetylsalicylic acid in human blood. Platelets 16, 171–179 Chan, C. C., Boyce, S., Brideau, C., Ford-Hutchinson, A. W., Gordon, R., Guay, D., Hill, R. G., Li, C. S., Mancini, J., and Penneton, M. (1995) Pharmacology of a selective cyclooxygenase-2 inhibitor, L-745,337: a novel nonsteroidal anti-inflammatory agent with an ulcerogenic sparing effect in rat and nonhuman primate stomach. J. Pharmacol. Exp. Ther. 274, 1531–1537 Bombardier, C., Laine, L., Reicin, A., Shapiro, D., BurgosVargas, R., Davis, B., Day, R., Ferraz, M. B., and Hawkey, C. J., et al. VIGOR Study Group. (2000) Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N. Engl. J. Med. 343, 1520 –1528 Warner, T. D., Vojnovic, I., Bishop-Bailey, D., and Mitchell, J. A. (2006) Influence of plasma protein on the potencies of inhibitors of cyclooxygenase-1 and -2. FASEB J. 20, 542–544 Benedek, I. H., Joshi, A. S., Pieniaszek, H. J., King, S. Y., and Kornhauser, D. M. (1995) Variability in the pharmacokinetics and pharmacodynamics of low dose aspirin in healthy male volunteers. J. Clin. Pharmacol. 35, 1181–1186 Davies, N. M., and Anderson, K. E. (1997) Clinical pharmacokinetics of naproxen. Clin. Pharmacokinet. 32, 268 –293 Davies, N. M. (1998) Clinical pharmacokinetics of ibuprofen. The first 30 years. Clin. Pharmacokinet. 34, 101–154 Davies, N. M., McLachlan, A. J., Day, R. O., and Williams, K. M. (2000) Clinical pharmacokinetics and pharmacodynamics of celecoxib: a selective cyclo-oxygenase-2 inhibitor. Clin. Pharmacokinet. 38, 225–242 Davies, N. M., Teng, X. W., and Skjodt, N. M. (2003) Pharmacokinetics of rofecoxib: a specific cyclo-oxygenase-2 inhibitor. Clin. Pharmacokinet. 42, 545–556
NSAIDs IN ENDOTHELIAL CELLS
44.
45.
46.
47.
48. 49.
50. 51.
Topper, J. N., Cai, J., Falb, D., and Gimbrone, M. A., Jr. (1996) Identification of vascular endothelial genes differentially responsive to fluid mechanical stimuli: cyclooxygenase-2, manganese superoxide dismutase, and endothelial cell nitric oxide synthase are selectively up-regulated by steady laminar shear stress. Proc. Natl. Acad. Sci. U. S. A. 93, 10417–10422 Okahara, K., Sun, B., and Kambayashi, J. (1998) (1998) Upregulation of prostacyclin synthesis-related gene expression by shear stress in vascular endothelial cells. Arterioscler. Thromb. Vasc. Biol. 18, 1922–1926 Dekker, R. J., van Soest, S., Fontijn, R. D., Salamanca, S., de Groot, P. G., VanBavel, E., Pannekoek, H., and Horrevoets, A. J. (2002) Prolonged fluid shear stress induces a distinct set of endothelial cell genes, most specifically lung Kruppel-like factor (KLF2). Blood 100, 1689 –1698 Singh, G., Miller, J. D., Huse, D. M., Pettitt, D., D’Agostino, R. B., and Russell, M. W. (2003) Consequences of increased systolic blood pressure in patients with osteoarthritis and rheumatoid arthritis. J. Rheumatol. 30, 714 –719 Pedersen, A. K., and FitzGerald, G. A. (1984) Dose-related kinetics of aspirin. Presystemic acetylation of platelet cyclooxygenase. N. Engl. J. Med. 311, 1206 –1211 Kyrle, P.A., Eichler, H. G., Jager, U., and Lechner, K. (1987) Inhibition of prostacyclin and thromboxane A2 generation by low-dose aspirin at the site of plug formation in man in vivo. Circulation 75, 1025–1029 Green, K., Drvota, V., and Vesterqvist, O. (1989) Pronounced reduction of in vivo prostacyclin synthesis in humans by acetaminophen (paracetamol). Prostaglandins 37, 311–315 Warner, T. D., Giuliano, F., Vojnovic, I., Bukasa, A., Mitchell, J. A., and Vane, J. R. (1999) Nonsteroid drug selectivities for cyclo-oxygenase-1 rather than cyclo-oxygenase-2 are associated with human gastrointestinal toxicity: a full in vitro analysis. Proc. Natl. Acad. Sci. U. S. A. 96, 7563–7568 Received for publication May 30, 2006. Accepted for publication July 5, 2006.
2475