Direct Inhibition of Cyclooxygenase-1 and -2 by the Kinase Inhibitors SB 203580 and PD 98059

The kinase inhibitors SB 203580 and PD 98059 have been reported to be specific inhibitors of the 38- and 42/44-kDa mitogen-activated protein kinase (MAPK) pathways, respectively. In this study, the two inhibitors were found to decrease platelet aggregation induced by low concentrations of arachidonic acid, suggesting that they also interfere with the metabolism of arachidonic acid to thromboxane A2. In support of this, SB 203580 and PD 98059 inhibited the conversion of exogenous [3H]arachidonic acid to [3H]thromboxane in intact platelets. Measurement of platelet cyclooxygenase-1 activity following immunoprecipitation revealed that SB 203580 and PD 98059 are direct inhibitors of this enzyme. Both compounds were shown to inhibit purified cyclooxygenase-1 and -2 by a reversible mechanism. In addition, SB 203580 (but not PD 98059) inhibited platelet aggregation induced by prostaglandin H2 and the conversion of prostaglandin H2 to thromboxane A2 in intact platelets. SB 203580 also inhibited this pathway in platelet microsome preparations, suggesting a direct inhibitory effect on thromboxane synthase. These results demonstrate that direct effects of the two kinase inhibitors on active arachidonic acid metabolites have to be excluded before using these compounds for the investigation of MAPKs in signal transduction pathways. This is of particular relevance to studies on the regulation of cytosolic phospholipase A2 as these two MAPKs are capable of phosphorylating cytosolic phospholipase A2, thereby increasing its intrinsic activity.

The flavone compound PD 98059 (2-(2-amino-3-methoxyphenyl)oxanaphthalen-4-one) is a specific inhibitor of the mammalian MAPK kinase (MEK) (7). It acts by binding to the inactivated form of MEK, thereby preventing its phosphorylation by c-Raf or MEK kinase (8). Depending on cell type and stimulation, the IC 50 determined in intact cells ranges from 2 to 10 M, values that are similar to those determined for inhibition of activation of MEK1 by Raf in vitro (7,9). Activation of MEK2 is inhibited with an IC 50 of 50 M (8). Kinetic properties suggest an allosteric mechanism for inhibition, which might be the reason for its high selectivity toward MEK (7,8).
We have previously used SB 203580 and PD 98059 to investigate the involvement of p38 mapk and p42/p44 mapk , respectively, in the regulation of cytosolic phospholipase A 2 and the activation of human platelets (10,11). During the course of these studies, we have not observed an action of either drug consistent with inhibition of kinases other than their known targets. However, we have obtained evidence that SB 203580 and PD 98059 interfere with the conversion of arachidonic acid to the pro-aggregatory metabolite thromboxane (Tx) A 2 (11). In platelets, the metabolism of arachidonic acid by the cyclooxygenase pathway results in the production of several active eicosanoids. The prostaglandin (PG) endoperoxide PGH 2 activates platelets (12), whereas its stable product, PGE 2 , can have pro-aggregatory or inhibitory effects depending on concentration (13). PGH 2 is the substrate for thromboxane synthase and is converted by this enzyme to TxA 2 in platelets. In this study, we have investigated whether the reduction of TxA 2 formation by SB 203580 and PD 98059 is due to inhibition of p38 mapk and p42/p44 mapk , respectively, or to a direct effect on arachidonic acid metabolism.

MATERIALS AND METHODS
Reagents and Antibodies-SB 203580 and PD 98059 were kindly provided by Dr. J. C. Lee (SmithKline Beecham, King of Prussia, PA) and Dr. A. R. Saltiel (Parke-Davis), respectively. SB 203580 was also obtained from Alexis Corp. (Nottingham, United Kingdom). Purity of these compounds was Ͼ98%. Polyclonal anti-cyclooxygenase-1 antibody, purified cyclooxygenase-1 and -2, and enzyme immunoassay (EIA) kits for measuring PGE 2 were purchased from Cayman Chemical Co., Inc.(Ann Arbor, MI). Polyclonal rabbit anti-p42 mapk antibody was from Santa Cruz Biotechnology, and rabbit anti-p38 mapk antiserum was a gift from Dr. R. J. Ulevitch (Scripps Research Institute, La Jolla, CA). 5-[1,2-3 H]Hydroxytryptamine (specific radioactivity of 25 Ci/mmol), [5,6,8,9,11,12,14, H]arachidonic acid (specific radioactivity of 212 Ci/mmol), and radioimmunoassay and EIA kits for measuring TxB 2 * This work was supported by the British Heart Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ These two authors contributed equally to this work. § To whom correspondence should be addressed. Fax: 44-1865-271853; E-mail: angelika.borsch@pharmacology.oxford.ac.uk. 1 The abbreviations used are: MAPK, mitogen-activated protein kinase; SAPK, stress-activated protein kinase; MEK, MAPK/extracellular signal-regulated kinase kinase; Tx, thromboxane; PG, prostaglandin; EIA, enzyme immunoassay.
were supplied by Amersham Pharmacia Biotech. [␥-32 P]ATP (specific activity of 3000 Ci/mmol) was obtained from NEN Life Science Products. Arachidonic acid (Sigma) was dissolved in ethanol and stored at Ϫ20°C. PGH 2 (Calbiochem), dissolved in hexane/isopropyl alcohol (9:1), was stored at Ϫ70°C; before experimentation, the solvent was evaporated under a stream of N 2 , and PGH 2 was dissolved in acetone (100 g/ml) (14). Furegrelate (Sigma) was dissolved in Tyrode's buffer. Prostacyclin was kindly donated by Wellcome Laboratories (Beckenham, Kent, UK). Bovine thrombin and protein A-Sepharose CL-4B were obtained from Sigma. Collagen was purchased from Nycomed Arzneimittel (Munich, Germany). All other reagents were of analytical grade.
Preparation of Washed Platelets-50 ml of blood was drawn on the day of experimentation from healthy volunteers by venipuncture using 7.5 ml of acidic citrate dextrose (120 mM sodium citrate, 110 mM glucose, and 80 mM citric acid) as anticoagulant. All solutions were prewarmed to 30°C. Platelet-rich plasma was obtained by centrifugation at 200 ϫ g for 20 min, and platelets were collected by centrifugation at 1000 ϫ g for 10 min in the presence of 0.1 g/ml prostacyclin. The platelet pellet was gently resuspended in Tyrode's buffer (20 mM HEPES, 135 mM NaCl, 3 mM KCl, 0.25 mM Na 2 HPO 4 , 12 mM NaHCO 3 , 1 mM MgCl 2 , and 5 mM glucose, pH 7.3) and 150 l of acidic citrate dextrose solution. Platelets were washed with a mixture of 25 ml of Tyrode's buffer and 3 ml of acidic citrate dextrose and centrifuged at 1000 ϫ g for 10 min in the presence of prostacyclin. The pellet was resuspended in 1 ml of Tyrode's buffer, and the volume was adjusted to give 4 ϫ 10 8 cells/ml. Platelets were allowed to rest for 30 min at 30°C prior to experimentation.
Activation of p42 mapk and p38 mapk was measured after immunoprecipitation from platelet lysates using 5 l of anti-p42 mapk antibody or 2 l of anti-p38 mapk antiserum. Proteins were separated on 10% SDSpolyacrylamide gels that contained 0.5 mg/ml myelin basic protein co-polymerized with the acrylamide. Kinases were denatured in 6 M guanidine HCl and renatured for 16 h at 4°C as described previously (16). Gels were incubated in kinase buffer (50 mM Tris, pH 8.0, 5 mM MgCl 2 , 1 mM EGTA, 5 mM dithiothreitol, 50 M ATP, and 20 Ci/ml [␥-32 P]ATP) for 1 h at 37°C, extensively washed in 5% trichloroacetic acid and 1% Na 4 O 2 P 7 , and dried. After autoradiography, the region of MAPK was cut from the gel and Cerenkov-counted for radioactivity.
Release of 5-[ 3 H]Hydroxytryptamine-Platelets were incubated with 10 Ci of 5-[ 3 H]hydroxytryptamine for 1 h at 30°C, washed, and resuspended in Tyrode's buffer. Stimulation of platelets was stopped with an equal volume of 6% (v/v) glutaraldehyde solution. Platelets were pelleted by centrifugation at 13,000 ϫ g for 15 min, and the radioactivity of the supernatant was determined by liquid scintillation spectrometry in a Beckman scintillation counter to a 5% level of significance.
Preparation of Platelet Microsomes-Washed platelets were suspended at 2 ϫ 10 9 cells/ml in 25 mM HEPES, pH 7.0, containing 25 mM NaCl, 100 mM KCl, 2 mM MgSO 4 ⅐7H 2 O, 12 mM trisodium citrate, 10 mM glucose, 5 mM dithiothreitol, 1 mM benzamidine HCl, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM EGTA, and microsomal fractions were prepared as described (17). The suspension was sonicated on ice four times for 5 s and then centrifuged at 1500 ϫ g. This process was repeated on the pellet, and both sonicates were centrifuged at 13,000 ϫ g for 30 min at 4°C to remove insoluble material. The supernatant was centrifuged at 150,000 ϫ g for 60 min at 4°C, and the pellet, representing the microsomal fraction, was suspended in 20 mM Tris, pH 8.0, containing 1 mM EDTA, 1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, and 1 mM benzamidine HCl. Protein concentrations were determined by the Bradford reaction (Bio-Rad) using bovine serum albumin as a standard. Microsomes were stored at Ϫ70°C. Thromboxane synthase activity was verified by the capacity of the platelet microsomes to convert PGH 2 into TxB 2 . The amount of TxB 2 produced was determined by EIA as described by Maclouf (15). In some experiments, microsomes were preincubated for 15 min at 37°C with The reaction was stopped by acidification (addition of 1 N HCl to reach pH 4), and lipids were extracted by incubation overnight at 4°C with 3 volumes of ethyl acetate. Samples were concentrated by evaporation and applied to silica thin-layer chromatography plates (Merck). Lipids were separated by ascending thin-layer chromatography in ethyl acetate/isooctane/acetic acid/H 2 O (110:50:20:100, v/v/v/v) as described (18). 1-cm fractions of the silica plate were scraped off, and the radioactive products were quantitated by liquid scintillation spectrometry in a Beckman scintillation counter. Radioactive lipids were identified compared with the migration of standards. The 12-lipoxygenase product 12-[ 3 H]hydroxyeicosatetraenoic acid migrated close to [ 3 H]arachidonic acid, and radioactivity from this lipid was added to untransformed [ 3 H]arachidonic acid for the calculation of metabolite formation.
Cyclooxygenase Assays-For immunoprecipitation of cyclooxygenase-1, platelets were lysed in ice-cold buffer (50 mM Tris-HCl, pH 8.0, 30 mM n-octyl glucoside, 1 mM EDTA, and 1 mM benzamidine HCl) (19). Lysates were precleared with protein A-Sepharose CL-4B, and cyclooxygenase-1 was immunoprecipitated using 5 l of polyclonal anticyclooxygenase-1 antibody and 25 l of protein A-Sepharose CL-4B slurry overnight at 4°C. Immunoprecipitates were recovered by microcentrifugation and washed twice in the ice-cold lysis buffer. Immunoprecipitates were resuspended in 20 l of 50 mM Tris, pH 8, containing 0.1% bovine serum albumin and 1 mM phenol at 37°C and were incubated with buffer, Me 2 SO (2%; control), SB 203580 (20 M), or PD 98059 (20 M) for 10 min followed by hematin (1 M) for 1 min. Cyclooxygenase-1 was stimulated with 25 M arachidonic acid containing 1 Ci of [ 3 H]arachidonic acid at 37°C. After 10 min, the reaction was terminated by the addition of 3 volumes of ice-cold Tris-buffered saline and centrifugation for 1 min at 5000 ϫ g. Lipids were extracted from the supernatant in 3 volumes of ethyl acetate and separated by thin-layer chromatography as described above. Cyclooxygenase-1 bound to protein A-Sepharose was solubilized in Laemmli sample buffer and subjected to SDS-polyacrylamide electrophoresis. Cyclooxygenase-1 immunoprecipitation was verified by immunoblotting.
For measuring transformation of arachidonic acid using purified cyclooxygenase, cyclooxygenase-1 and -2 were diluted to give PGE 2 production within the linear range of the EIA kit. Cyclooxygenase was incubated with buffer, Me 2 SO, SB 203580, PD 98059, or indomethacin for 15 min, and the reaction was initiated by the addition of arachidonic acid (25 M) as described above. Formation of PGE 2 was measured using EIA according to the instructions of the manufacturer.

Inhibitory Effects of SB 203580 and PD 98059 on Platelet
Responses-Platelet aggregation induced by collagen is dependent on the formation of thromboxane A 2 from arachidonic acid by the action of cyclooxygenase and thromboxane synthase. We have reported earlier that SB 203580 and PD 98059 inhibit platelet aggregation induced by low concentrations of collagen. PD 98059 is more powerful than SB 203580 in inhibiting this response (11). Its inhibitory action was only overcome by collagen at 20 g/ml (Fig. 1A), whereas the effect of SB 203580 was overcome at 5 g/ml collagen (11). In contrast to collagen, aggregation stimulated by thrombin (0.1 unit/ml) was not dependent on the release of TxA 2 , as demonstrated by preincubation with the cyclooxygenase inhibitor indomethacin (Fig. 1B). The presence of neither PD 98059 (20 M) nor SB 203580 (20 M) altered the aggregation traced to thrombin (Fig.  1B). These concentrations of PD 98059 and SB 203580 are sufficient to inhibit thrombin-induced activation of p42 mapk and p38 mapk , respectively (10,20). The inclusion of Me 2 SO did not have any significant effect on platelet aggregation as tested with thrombin (0.1 units/ml) and collagen (10 g/ml) (data not shown). SB 203580 and PD 98059 partially inhibited aggregation stimulated by the Ca 2ϩ ionophore A23187 (2 M) (data not shown), a platelet stimulus that induces profound liberation of arachidonic acid from phospholipids (21). In contrast, aggregation induced by A23187 in the presence of the cyclooxygenase blocker indomethacin was not altered in the presence of PD 98059 or SB 203580 (data not shown). Thus, SB 203580 and PD 98059 impaired platelet aggregation only under conditions where formation of TxA 2 contributed to the response.
Thromboxane formation stimulated by collagen (2 and 5 FIG . 1 g/ml), as measured by EIA, was partially inhibited by SB 203580 and completely blocked by PD 98059 (Fig. 2), which is in agreement with the effects of the compounds on aggregation.
In addition, release of 5-hydroxytryptamine stimulated by collagen in the absence of indomethacin was reduced by both compounds, but could be overcome at higher concentrations of collagen (Fig. 3).
Platelet Activation by Arachidonic Acid-To investigate the effect of the compounds on the metabolism of arachidonic acid, we incubated platelets with low concentrations of arachidonic acid, which induce aggregation as a consequence of conversion to TxA 2 . SB 203580 and PD 98059 inhibited platelet aggregation induced by 0.2 M arachidonic acid with approximate IC 50 values of 3 and 0.8 M, respectively (Fig. 1C). Aggregation stimulated by 1 M arachidonic acid was reduced by 50% in the presence of SB 203508 (20 M) and was blocked by PD 98059 (20 M) (Fig. 1C). Moreover, both compounds inhibited formation of TxB 2 from arachidonic acid (1 M) as measured by EIA (Table I). These observations demonstrate that both inhibitors interfere with the metabolism of arachidonic acid to TxA 2 .
Arachidonic acid (0.2 and 1 M) did not stimulate platelet p42 mapk and caused weak activation of p38 mapk relative to activation by thrombin (Fig. 4). However, activation of MAPKactivated protein kinase-2, the in vivo substrate of p38 mapk , was not significantly enhanced above basal levels by arachidonic acid (data not shown). The TxA 2 receptor agonist U46619 has been reported to cause weak phosphorylation of p42 mapk and p38 mapk (20). Thus, arachidonic acid and its metabolites seem to be only weak stimulators of MAPKs.
Because the kinase inhibitors could interfere with the release of endogenous arachidonic acid through an effect on the activity of cytosolic phospholipase A 2 , it was important to determine the transformation of arachidonic acid to TxB 2 independent of the release of endogenous thromboxane. For this reason, we incubated platelets with   (Fig. 5A). This was in contrast to incubation with the thromboxane synthase inhibitor furegrelate (10 M) (19,22) (Fig. 5B). Similarly, in the presence of furegrelate, PD 98059 fully inhibited the formation of all three products (data not shown). These results confirm that SB 203580 and PD 98059 inhibit platelet cyclooxygenase.
To test the possibility that both compounds cause direct inhibition of cyclooxygenase, we immunoprecipitated cyclooxygenase-1 from unstimulated platelets and measured its activity in vitro. Under resting conditions, neither p42 mapk nor p38 mapk is activated (11,16 (Fig. 6). There was no significant formation of [ 3 H]PGE 2 in the presence of either SB 203580 (20 M) or PD 98059 (20 M) (Fig. 6). The presence of Me 2 SO had no significant effect on enzyme activity (data not shown).
Human platelets contain the constitutively expressed cyclooxygenase-1, but not the inducible cyclooxygenase-2 (23). To further characterize the inhibition of cyclooxygenase by the two kinase inhibitors, we measured the effect of increasing inhibitor concentrations on purified cyclooxygenase-1 and -2. SB 203580 and PD 98059 inhibited cyclooxygenase-1 with approximate IC 50 values of 2 and 1 M, respectively (Fig. 7A). Inhibition was completely reversible for both compounds, as tested by preincubation of cyclooxygenase with 3 M inhibitor before a 10-fold dilution and incubation with substrate (data not shown). Cyclooxygenase-2 was also inhibited by both compounds in a reversible manner. IC 50 values were ϳ2 and 4 M for SB 203580 and PD 98059, respectively (Fig. 7B). Inclusion   FIG. 4. Activation of p42 mapk and p38 mapk . Platelets were incubated with buffer (basal), arachidonic acid (aa; 0.2 and 1 M), or thrombin (1 unit/ml) for 2 min at 37°C under stirred conditions. p42 mapk and p38 mapk were immunoprecipitated from platelet lysates and run on SDS-polyacrylamide gels containing myelin basic protein (MBP). Gels were denatured, renatured, and incubated with kinase buffer. After extensive washing, gels were dried. Phosphorylated proteins were visualized by autoradiography, cut from the gels, and Cerenkov-counted for incorporation of 32 P into myelin basic protein. of the solvent Me 2 SO did not reduce enzyme activity. The degree of inhibition of cyclooxygenase-1 and -2 by SB 203580 and PD 98059 (20 M) was similar to that by a maximally effective concentration of the cyclooxygenase inhibitor indomethacin (data not shown). Inhibition by SB 203580 could be overcome by increasing arachidonic acid concentrations; in the presence of SB 203580 (3 M), the K m for arachidonic acid was increased without a change in the V max , suggesting a competitive mechanism of inhibition (data not shown). In contrast, PD 98059 (3 M) decreased the V max even at the highest concentration of arachidonic acid used (50 M) (data not shown).
Platelet Activation by PGH 2 -To distinguish between the effects of the kinase inhibitors on cyclooxygenase and thromboxane synthase, we stimulated platelets with the cyclooxygenase product PGH 2 . SB 203580 inhibited PGH 2 -induced platelet aggregation, whereas PD 98059 did not alter the response (Fig. 1D). In addition, the formation of TxB 2 from 1 g/ml PGH 2 was inhibited by SB 203580, but not by PD 98059, as determined by EIA (Table I).
It was important to determine the activity of thromboxane synthase independent of the stimulation of receptor-activated pathways and of activation of cytosolic phospholipase A 2 . We therefore measured the transformation of PGH 2 to TxB 2 on platelet microsomes, a membrane fraction that contains thromboxane synthase (17). Microsomes were obtained from unstimulated platelets, which means that p42 mapk and p38 mapk were not activated. PGH 2 added to microsomes was transformed to TxB 2 as determined by radioimmunoassay (Table I).
The presence of PD 98059 in this assay did not significantly alter the response, whereas the addition of SB 203580 blocked TxB 2 formation (Table I). These results demonstrate that SB 20350 (but not PD 98059) acts as a direct inhibitor of thromboxane synthase. DISCUSSION Several studies have confirmed the selectivity of the kinase inhibitors SB 203580 and PD 98059 against a variety of kinases. SB 203580 was tested against 16 protein kinases and two phosphatases and was found to be selective against p38 and p38␤ (SAPK2a and SAPK2b), but to have no inhibitory effect on SAPK3, SAPK4, p42/p44 mapk , c-Jun N-terminal kinase (SAPK1), or any of the upstream kinases (3,6). Similarly, the selectivity of PD 98059 has been described in several studies (7-9). We (Refs. 11 and 20 and this study) and others (24) have observed that platelet responses dependent on the formation of TxA 2 are inhibited by these compounds. However, since SB 203580 has been developed from drugs that are inhibitors of cyclooxygenase and lipoxygenase, it is uncertain whether the inhibitory actions of SB 203580 on platelet responses can be interpreted as effects of p38 mapk or whether they are due to inhibition of arachidonic acid metabolism. Furthermore, we have also observed inhibition of platelet activation by PD 98059 under conditions that have little effect on p42 mapk activation. For these reasons, we set out to investigate the effects of these compounds on the arachidonic acid cascade.
We obtained evidence for direct inhibition of platelet cyclooxygenase by SB 203580 from a variety of approaches: first, inhibition of platelet aggregation induced by stimuli that are dependent on TxA 2 to elicit aggregation; second, inhibition of TxB 2 release; third, inhibition of the conversion of [ 3 H]arachidonic acid to 3 H-labeled metabolites after the addition of [ 3 H]arachidonic acid to intact platelets; and fourth, inhibition of the activity of immunoprecipitated cyclooxygenase-1 as measured in vitro. In addition, we found direct effects of SB 203580 on thromboxane synthase using PGH 2 as agonist, as the following responses were inhibited: platelet aggregation, TxB 2 formation from intact platelets, and conversion of PGH 2 to TxB 2 in platelet microsomes.
PD 98059 was a more powerful inhibitor of thromboxane-dependent platelet aggregation than SB 203580. In intact platelets, PD 98059 was a stronger inhibitor of cyclooxygenase (approximate IC 50 for inhibition of arachidonic acid-induced platelet aggregation ϭ 0.8 M) than of MEK activation by Raf (IC 50 ϭ 2-10 M in vivo) (7, 10). Alessi et al. (8) have previously pointed out that relatively high concentrations of PD 98059 (20 -50 M) have to be used on intact cells to completely block the activation of p42/p44 mapk , and our studies on the activation of p42/p44 mapk in human platelets have confirmed these results (10). However, a 10-fold lower concentration of PD 98059 was sufficient to inhibit platelet aggregation, suggesting that this inhibition is independent of p42 mapk . In fact, p42 mapk was not activated by arachidonic acid under the conditions of the aggregation experiment. Moreover, PD 98059 does not alter phosphorylation of cytosolic phospholipase A 2 in collagen-activated platelets (11). The interference with collagen-induced aggregation would therefore seem to reflect its inhibitory effect on cyclooxygenase. In contrast to SB 203580, PD 98059 had no direct effect on thromboxane synthase.
Inhibition of purified cyclooxygenase-1 and -2 was reversible for both compounds. The IC 50 values for purified cyclooxygenase-1 agree with the values determined in platelet aggregation experiments. Cyclooxygenase-2 was slightly less susceptible to inhibition by SB 203580 and PD 98059 compared with cyclooxygenase-1. Marshall et al. (25) have previously analyzed the effects of an analogue of SB 203580 on cyclooxygenase activity; inhibition by 2-(4-methylthiophenyl)-3-(4-pyridyl)-6,7dihydro-5H-pyrrolo[1,2-a]imidazole was reversible and competitive. They speculated that this compound and other nonsteroidal anti-inflammatory drugs such as acetoamidophenol and phenylbutazone act as radical scavengers and thereby decrease the availability of hydrogen peroxide intermediates during the enzymatic reaction (25). It is possible that a similar mechanism applies to SB 203580.
The rate-limiting step for the liberation of TxA 2 in human platelets is the formation of arachidonic acid from membrane phospholipids by the activity of cytosolic phospholipase A 2 (26). Cytosolic phospholipase A 2 is regulated by changes in the intracellular Ca 2ϩ concentration (27) and by phosphorylation of serine 505 (28) and possibly on other serine residues (29,30). SB 203580 and PD 98059 are important tools to investigate MAPK-mediated phosphorylation of cytosolic phospholipase A 2 and the regulation of arachidonic acid release. However, as we show in this report, both compounds interfere directly with arachidonic acid metabolism through the cyclooxygenase pathway. Moreover, in other cases where arachidonic acid metabolites regulate the cell's response, a clear distinction between the effects of kinase inhibition and cyclooxygenase inhibition by SB 203580 and PD 98059 has to be made. For routine application of the inhibitors, we suggest that the compounds be used in conjunction with a cyclooxygenase blocker, such as indomethacin or aspirin, in order to be able to interpret results unambiguously.