© Birkhäuser Verlag, Basel, 2001 Inflamm. res. 50 (2001) 262–269 1023-3830/01/050262-8 $ 1.50+0.20/0
Inflammation Research
Original Research Papers In-vitro test system for the evaluation of cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) inhibitors based on a single HPLC run with UV detection using bovine aortic coronary endothelial cells (BAECs)* G. Dannhardt and H. Ulbrich Institute of Pharmacy, University of Mainz, Staudinger Weg 5, D-55099 Mainz, Germany, Fax +49 6131 3923062, e-mail:
[email protected] Received 13 June 2000; returned for revision 13 November 2000; accepted by G. Geisslinger 28 November 2000
Abstract. Objective and Design: The aim of this study was to develop a new, whole-cell test system which is easy to handle and requires a standard equipment for the parallel screening of COX-1 and COX-2 inhibitors. Materials: Bovine aortic endothelial cells (BAECs). Treatment and methods: Unstimulated bovine aortic coronary endothelial cells (BAECs) were used as a source of COX1 and BAECs pretreated with ASA (100 mM) and activated with phorbol myristate acetate (PMA) were used as a source of COX-2. The time- and concentration-dependent induction of COX-2 expression in the BAECs was evaluated by a kinetic profile (HPLC analysis) and detected by Western-Blot analysis using polyclonal antibodies against COX-1 and COX-2. Results: In BAECs, diclofenac and meloxicam showed balanced inhibition of COX-1 (IC50: 0.01/0.4 mM) and COX-2 (IC50: 0.03/0.6 mM). Indomethacin inhibited COX-1 more potently than COX-2 (IC50: 0.008/0.04 mM). Aceclofenac inhibited COX-2 more potently than COX-1 (IC50: 3.0/ 7.3 mM). DFU and Cl-SC57666 [16] inhibited COX-2 (IC50: 0.04/0.001 mM) highly selectively but did not inhibit COX-1 (IC50: >100 mM). Conclusions: In summary an assay has been developed, for the determination of IC50-values for inhibitors of COX-1/2 on cells of the same origin, in line with values in the literature. Moreover, new insights have been gained into the relationship of COX-1/2 and lipoxygenase pathways in BAECs by detecting 15- and 12-HETE: Inhibition of COX-1 by the NSAIDs mostly resulted in an enhancement of 15-HETE and 12-HETE release. In contrast inhibition of COX-2 decreased 15-HETE release. Correspondence to: G. Dannhardt * Dedicated to Prof. Dr. Dr. med. Dr. h.c. E. Mutschler, on the occasion of his 70th birthday.
Key words: BAECs – Cyclooxygenase-1/2 – Whole cell in vitro test system – NSAIDs
Introduction The two isoforms of cyclooxygenase (COX) metabolise arachidonic acid to bioactive prostanoids [1, 2]. COX-1 is expressed constitutively in many tissues; it is responsible for the production of prostaglandins which regulate homeostatic functions in the gastric mucosa, platelets and the vascular endothelium [3]. Vascular endothelial cells (EC) play a central role in the regulation of the vascular tone, platelet aggregation and inflammatory cell functions. During inflammation COX-1 levels do not increase. In contrast, the expression of COX-2 is induced by a number of inflammatory cytokines, growth factors, and mitogens [4]. COX-2 has therefore been implicated in prostaglandin generation at inflammatory sites, while COX-1 mediates prostaglandin production of normal cellular activity. It was concluded that the established nonsteroidal antiinflammatory drugs (NSAIDs) cause gastrotoxicity because of the selective inhibition of COX-1, rather than COX-2. This point of view is controversial since it was found that under physiological conditions COX-2 is expressed constitutively in the kidney [5], spinal cord [6] and brain [7]. In addition, up-regulation of COX-2 has been found to occur in the stomach [8]. Nevertheless there is great interest in developing new COX-2-selective inhibitors. Several in vitro assays have been developed to investigate the selectivity of NSAIDs for the cyclooxygenases COX-1 and COX-2. However, these test systems vary greatly with regard to the experimental conditions. Variables that have been altered include the source of the enzyme used (animal or human), the method of enzyme preparation (purified
Vol. 50, 2001
COX-1/-2 assay using BAECs
enzyme vs. native or recombinant enzyme vs. whole cell), the source of arachidonic acid (endogenous or exogenous) and different incubation times for COX-1 and COX-2. Our previously published in vitro cell assays for the determination of COX-1/5-LOX and COX-1/thromboxane synthase (TXS) used bovine platelets, neutrophils and pig platelets as the enzyme source [9, 10]. Moreover we used an alternative membrane assay (hen’s egg chorioallantoic membrane) as a model for inflammation [11]. These assays show good reproducibility and comparability with human data. The advantage of our new test system is that it makes parallel screening of COX-1/COX-2 inhibitors possible, using only one cell type as enzyme source. Unstimulated BAECs do not contain detectable amounts of COX-2 protein [12]. However, treatment with well known COX-2 inducers caused a significant increase in COX-2, but not COX-1 despite mRNA upregulation of both [13]. We report on the validation of an in vitro assay using unstimulated bovine aortic coronary endothelial cells (BAECs) as a source of COX-1 and BAECs pretreated with ASA (100 mM) and activated with phorbol myristate acetate (PMA) as a source of COX-2. After optimising the experimental conditions for COX-2 expression in BAECs, the effects of some known NSAIDs were investigated for COX-1 and COX-2 activity. In addition the ratio of PGH-synthases and 15- and 12-lipoxygenase metabolites can be assessed in one HPLC run. Materials and methods Cell cultures Culture techniques were adapted from procedures by Gimbrone [14] and Jaffe [15]. Segments of bovine thoracic aorta, 20 to 30 cm in length, were obtained from cows from the heart-lung block of organs passing along in the assembly line tray at a meat-processing plant in Alzey, Germany. The aortae were transported to the laboratory in a sterile beaker containing ice cold Dulbecco’s phosphate buffered saline (pH 7.4, KH2PO4 2 g/l, KCl 0.2 g/l, NaCl 8.00 g/l, Na2HPO4 ¥ 7 H2O 2.16 g/l), benzylpenicillin (100 U/ml) and gentamycin (50 mg/ml) and all subsequent operations were carried out in a Jouan laminar flow hood. Adipose tissue and fat were removed from the aorta, which was opened longitudinally, and fixed on a plate with the intimal face upwards. The endothelial surface was incubated for 40 min at 37 °C in an atmosphere of 95% air and 5% CO2 with 6 ml dispase (480 mg /100 ml H2O). Detached endothelial cells were collected in a sterile 50 ml tube containing 25 ml growth culture medium (medium 199 containing 20% heat-inactivated fetal bovine serum, benzylpenicillin (100 U/ml), streptomycin (100 g/ml), and L-glutamine (2 mM). After centrifugation at 165 ¥ g for 10 min the pellet was resuspended in culture medium. The cells were placed in a 75 cm2 cell culture flask and maintained in a 95% air 5% CO2 humidified incubator at 37 °C. The culture medium was replaced every three days and changed to minimal medium (medium 199 containing 10% heat-inactivated fetal bovine serum, benzylpenicillin (100 U/ml), streptomycin (100 g/ml), and L-glutamine (2 mM)) 48 h before the experiments were performed. The cells were identified as endothelial cells due to their ability to form a typical cobblestone appearance when confluent and take up fluorescently labelled acetylated low-density lipoprotein. Upon reaching confluence (7–10 days) the cells were subcultured by detaching them with 6 ml 0.25% trypsin-EDTA solution. Experiments were performed when the cells reached approximately 95% confluency on the second passage. For each experiment a minimum of five cultures were tested in at least three separate experiments.
263
Western-blot-analysis of COX-1 and COX-2 The cells were incubated both in the presence and in the absence of 20 nM PMA for 6 h. To solubilize the COX proteins, the cells were scraped with a rubber-policeman, sonicated and incubated for 10 min at 0 °C with extraction buffer (50 mM Tris-HCl pH 8.0 containing 5 mM EDTA), 1% Nonidet P-40, pepstatin A 10 mg/ml, leupeptin 10 mg/ml and antipain 1 mg/ml) while being gently shaken. The cell extracts were centrifuged at 12,000 ¥ g for 2 min and were then boiled for 3 min at ratio of 1:1 with sample buffer (Tris 0.2 mM, SDS 8% w/v, glycerol 40% v/v, 2-mercaptoethanol 20% v/v and bromphenol blue 0.2 % w/v). The samples were diluted until they had equal amounts of protein for each condition, and 30 ml of protein lysate were separated by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulphate (SDS) (SDS-PAGE). A 11% w/v resolving gel and a 5 % w/v stacking gel were used. The proteins were transferred onto nitrocellulose membranes (Schleicher & Schuell, Göttingen, Germany) with a semidry transfer unit (Biorad, Hercules, CA, USA). The transfer was performed in a 25 mM Tris, 192 mM glycine buffer, pH 8.3, containing 20% methanol for 1h at 300 mA. Blot efficiency was proved by reversible staining with Ponceau red in 0.3% trichloroacetic acid to visualise proteins. Nonspecific antibody binding to the nitrocellulose was prevented by incubating the filter overnight at 4 °C with block solution (dried minimal-fat milk 5% w/v and Tween-20 0.25% v/v in DPBS solution). Blots were further incubated with polyclonal antibody serum raised to ovine COX-1 developed in rabbits (Cayman Chemical, Ann Arbor, USA) or with a rabbit polyclonal antibody serum raised to murine COX-2 (Cayman Chemical, Ann Arbor, USA). The blots were then incubated with a goat anti-rabbit IgG developed in sheep, linked to alkaline phosphatase conjugate (Kirkegaard Perry, Labaratories, Maryland, USA). Finally, the blots were developed for approximately 5 min with a Sigma Fast BCIP/NBT Buffered Substrate Tablet containing 5-bromo-4-chloro-3-indolyl phosphate (BCIP) 0.15 mg/ml, nitro blue tetrazolium (NBT) 0.30 mg/ml, Tris 100 mM and MgCl2 5 mM; (pH 9.2). The detection limit of protein in the cell extracts was 1 – 10 ng of protein.
Materials All the reagents used were of analytical grade and obtained from the following sources: acetonitrile, methanol, phosphoric acid (Merck, Darmstadt, Germany); phorbol myristate acetate (PMA) (Calbiochem, Eschborn, Germany); aceclofenac (Council of Europe, European Pharmacopoeia, Strasbourg Cedex, France); 1-[2-(4-chlorophenyl) cyclopenten1-yl]-4-(methylsulfonyl)benzene was synthesized in accordance with the literature [16]; DFU (MSD, München, Germany); nordihydroguaiaretic acid (NDGA), indomethacin, diclofenac, NS 398, acetylsalicylic acid, medium 199, ETYA, ETI, fetal calf serum, penicillin-streptomycin solution, benzylpenicillin, Ca-ionophore A 23187 (free acid), amphothericin B solution, gentamicin sulphate, L-glutamine solution, salts for buffer solutions (Sigma, München, Germany); dispase, trypsin-EDTA solution (Biochrom, Berlin, Germany); HPLC reference compounds: 6keto PGF1a , TXB2 , PGE2 , 12-HHT, 15-HETE, 12-HETE, 5-HETE, PGA2 , LTB4 , PGI2 , AA (Paesel, Frankfurt am Main, Germany) stored at – 75 °C. All cell culture materials and additives were purchased from Costar, (Cambridge, MA, USA) and Falcon, Becton Dickinson Labware Europe, (Meylan Cedex, France).
Inhibition of COX-1 in BAECs Cyclooxygenase-1 assay. BAECs in 75 cm2 flasks were washed three times with 5 ml Dulbecco’s Phosphate Buffered Saline (DPBS) (137 mM NaCl, 1,47 mM KH2PO4 , 81 mM Na2HPO4 , 27 mM KCl. Then the cells were incubated with 2975 ml DPBS, 12.5 ml of drug in DMSO (dimethylsulphoxide) solution or only DMSO and 1000 ml CaCl2 (10 mM in 0.8% w/v NaCl) for 15 min at 37 °C in an atmosphere of 95 % air 5 % CO2 . The cells were stimulated by adding 1000 ml DPBS and 12.5 ml of
264 a solution calcium ionophore A 23127 (4.2 mg/ml, 20 mM = final concentration) and the incubation was continued for 35 min (in the kinetic tests incubation time varied). 5 ml of a mixture of acetonitrile and methanol (1:1 v/v) containing 10 mg NDGA (3,3 mM) as oxygen scavenger and internal standard were added. The cells were counted after treatment with trypsin-EDTA solution using a hemocytometer or a coulter counter. The flasks were cooled for 10 min in an ice bath. The supernatants were stored at –20 °C until analysed. For the COX-1 assay the 12-HHT peak at 235 nm and that of 6-keto PGF1a at 200 nm was used.
Stimulation and inhibition of COX-2 in BAECs Cyclooxygenase-2 assay. Untreated BAECs do not contain detectable amounts of COX-2 protein [17] (Fig. 5). In contrast, BAECs activated with PMA contain a protein of approximately 70 kDa, which was recognised by a specific antibody against COX-2. The expression of the COX-1 isoenzyme at the protein level is not affected. BAECs were incubated for 30 min with 100 mM ASA or vehicle (as a positiv control) to inhibit irreversibly the COX-1-activity. After removal of the supernatants, cells were washed 3 times with DPBS in order to remove any residual ASA. The medium was changed and cells were activated for 6 h with 20 nM PMA to induce the COX-2. After 6 h the flasks were washed 3 times with buffer. The cells which had expressed the COX-2 were preincubated with DPBS, with a solution of drug in DMSO (dimethylsulphoxide) or DMSO alone for 15 min at 37 °C in an atmosphere of 95% air 5% CO2 . The subsequent incubation steps were performed as described above. For the COX-2 assay we used the 12-HHT peak at 235 nm and that of 6-keto PGF1a at 200 nm.
Sample preparation Cyclooxygenase-1/2 assay. An octadecyl reversed-phase column (ODS) was used to extract eicosanoids from the incubation media. The columns were pre-washed with 5 ml methanol twice, 5 ml water and 5 ml solution of Na2 EDTA 0.2%. The supernatants were diluted with 10 ml of water and extracted on the ODS column. The AA metabolites were eluated with 850 ml methanol three times and diluted with 2550 ml water and then subjected to HPLC.
Reversed-phase HPLC analysis Cyclooxygenase-1/2 assay. We adapted the RP-HPLC method of Tordjman [18] to investigate drug interactions with AA metabolism: Samples of two ml were injected on a Nucleosil 7 C 18 column (see apparatus). Elution was performed at 25°C with a flow rate of 2 ml/min. For the initial 28 min after injection, the acetonitrile concentration (v/v) was 27 %, pH 2.8, during the time between the minute 29 and the minute 70 the acetonitrile concentration (v/v) was 51%, and for the next ten minutes the acetonitrile concentration was increased to 100% to wash the column. The AA metabolites were quantified by peak area ratio. UV absorbance spectra of the column eluate were monitored at wavelengths of 200 to 400 nm. For the COX-1/2 assay we used the 12-HHT peak at 235 nm and that of 6-keto PGF1a at 200 nm. The 15- and 12-LOX activity was determined using 15- and 12-HETE peaks respectively, and detection at 235 nm.
G. Dannhardt and H. Ulbrich
Inflamm. res.
calculated using the data analysis and graphics program “Grafit” version 3.09b 1989–1996 Erithacus software by Robin Leatherbarrow. The results were expressed as mean ± standard error of the mean (SEM).
Apparatus Solid phase extraction was performed on a Baker-12 SPE system using octadecyl reversed-phase extraction columns 500 mg, 6 ml, (Mallinckrodt Baker, Griesheim, Germany). The automated RP-HPLC system Waters, (Eschborn, Germany) consisted of two Waters 510 pumps with a programmable pump control module (PCM), a Waters photo diode array detector 996, a Waters autosampler 717 plus and a column oven K 3. The separation of the AA metabolites was performed on a Nucleosil 100 C-18 column (7 mm, 250 ¥ 4.6 mm), Schambeck SFD GmbH, (Bad Honnef, Germany). A Millipore Q plus PF Millipore, (Eschborn, Germany) was used to deionize and purify the water. A Leica DMIRB microscope (Bensheim, Germany) was used for examining and counting the cells. A coulter counter Modell Industrial D (Coulter Electronics LTD, Beds., England) was used for counting the cells.
Results Identification of arachidonic acid metabolites and kinetic studies for the assessment of COX activities in BAECs The peaks from a typical stimulation of BAECs with Caionophore A 23187 comigrated with PG and HETE standards. In addition to the retention times we used PDA-Spectra overlays to verify the identity of each peak. The prostaglandin fraction contained 6-keto PGF1a (tr = 9.1 min) and in some cultures a small amount of PGE2 . Untreated BAECs released only a small amount of 6-keto PGF1a . The major peak generated in BAECs comigrated with 12-HHT (tr = 41.6 min) and had a UV maximum at 235 nm. The HETE fraction was also compared to standards chromatographed under identical conditions. These peaks were identified as 15-HETE (tr = 52.1 min) and 12-HETE (tr =56.1 min) with UV absorbance at 235 nm indicating the presence of a conjugated diene moiety. The stimulation times were optimised by kinetic measurements: The release of the eicosanoids after stimulation increased over 60 min (Fig. 1). Similar results concerning the formation of 12-HHT, 15-HETE, 12-HETE, and 6-keto
Statistical analysis The test compounds were tested in concentrations from 10–5 –10–9 mol/l. The selectivity of the test compounds was evaluated by calculation of the percentage inhibition of the 6-keto PGF1a and 12-HHT production. Potency estimates are presented as IC50-values. The IC50-values were
Fig. 1. Time-course analysis of the biosynthesis of 6-keto PGF1a (), 12-HHT (¥), 15-HETE () and 12-HETE () by BAECs after stimulation with Ca-ionophore A 23187 (20 mM). The values are the mean of at least six stimulation experiments, ± SEM is shown by the bar.
Vol. 50, 2001
COX-1/-2 assay using BAECs
PGF1a were obtained from HUVEC by Ibe et al. [19] and Hollenberg et al. [20]. They showed that the release of 15HETE and 12-HETE paralleled the release of 12-HHT. In all the experiments the synthesis of 15-HETE exceeded that of 12-HETE. Since the reduced release of AA, for example by inhibition of PLA2 , or the inhibition of the TXS may be also responsible for this effect. The second COX-1 assay uses 6keto PGF1a as a product of the prostacyclin synthase, making the procedure independent from the TXS activity. Under these conditions 6-keto PGF1a concentrations are high enough for UV detection at 200 nm, whereas the TXS product TXB2 cannot be detected. We found almost the same IC50-value when we looked at the 6-keto PGF1a or the 12-HHT peak for the tested NSAIDs in the same HPLC run. Consequently diminished production of 12-HHT does not depend on inhibition of TXS or PLA2 but reflects COX-1 activity.
265
Fig. 3. Viability profile of BAECs, treated with ASA (100 mM) for 30 min, LPS (1 mg/ml) for 12 h and PMA (20 nM) for 6 h). The values are the mean of at least six incubation experiments, ± SEM is shown by the bar.
Accumulation of AA metabolites by BAECs incubated with PMA We found that BAECs responded to LPS and PMA by a timedependent enhancement of 6-keto PGF1a and 12-HHT (Fig. 2). This fact corresponds to the studies of Akasereenont et al. [21] who used LPS over a range of concentrations and different incubation times to activate COX-2. Miralpeix et al. [22] showed that treatment of the HUVEC-C line with PMA resulted in a dose- and time-dependent higher production of 6-keto PGF1a , 12-HHT, 11-HETE and 15-HETE. In contrast to Akasereenont et al. incubation of BAECs with LPS (1 mg/ml) for 12 h did reduce cell viability (viability: 67% of control). As a result of these initial experiments, a single, fixed molarity of PMA (20 nM) over 6 h was used in the subsequent work (Fig. 3). This concentration did not affect the viability of the cells and the amounts of 6-keto
Fig. 4. Time-course for the accumulation of 6-keto PGF1a (blank bar) and 12-HHT (black bar) in the supernatants of BAECs treated with ASA (100 mM) for 30 min and untreated BAECs (A 23187) at different time points. The cells were incubated with vehicle for 15 min and stimulated for 35 min with Ca-ionophore A 23187 (20 mM) (A 23187) and pretreated with ASA (100 mM) for 30 min, treated with vehicle for 15 min and stimulated for 35 min with Ca-ionophore A 23187 (20 mM).The supernatants were analysed after different time points (1–6 h). Data are expressed as mean S.E.M from the triplicate determinations of 3 separate experimental days (n = 5).
PGF1 and 12-HHT formed were high enough to use them for UV detection. To exclude any contribution of BAECs COX-1 activity to prostaglandin production in response to PMA within the test system, the BAECs were pretreated with acetylsalicylic acid (100 mM), followed by several washing steps. ASA treated cells accumulated very small amounts of cyclooxygenase metabolites over 6 h (Fig. 4). The cells which were treated for 6 h with acetylsalicylic acid (100 mM) showed no morphological differentiation. Characterisation of cyclooxygenase isoforms present in BAECs
Fig. 2. Unstimulated BAECs produce 6-keto PGF1a (blank bar) and 12HHT (black bar) after preincubation with vehicle for 15 min and stimulation with Ca-ionophore A 23187 (20 mM) for 35 min, indicating the COX-1 activity of BAECs (control). The incubation with ASA (100 mM) for 30 min inhibits irreversibly the COX-1 enzyme and the production of 12-HHT and 6-keto PGF1a . PMA (20 nM) incubation over 6 h induced COX-2 and caused an increase of 6-keto PGF1a and 12HHT after preincubation with vehicle for 15 min and stimulation with Ca-ionophore for 35 min (20 mM). The incubation of PMA-stimulated BAECs with Cl-SC5766 for 15 min and stimulation with Ca-ionophore A 23187 (20 mM) for 35 min inhibits 90% of the 12-HHT and 6-keto PGF1a production, indicating the inhibition of the COX-2 enzyme. The values are the mean of at least six incubations.
Extracts of untreated or ASA-treated (100 mM) BAECs contained no detectable amounts of cyclooxygenase-2-protein, as determined by Western blot analysis Fig. 5, lane 1 and 2, detectable level is 1–10 ng of protein [21]. In contrast, after incubation for 6 h with PMA (20 nM) BAECs contained a protein of approximately 70 kDa, which was recognised by a specific antibody to cyclooxygenase-2 (Fig. 5; lane 4 and lane 5). The induction of cyclooxygnase-2 protein by PMA in BAECs was abolished by dexamethasone (2 mM) (lane 3). These facts demonstrate the up-regulation of cyclooxygenase-2 in this model.
266
Fig. 5. Western Blot using polyclonal antibodies to cyclooxygenase-2 of cell extracts from untreated and PMA-activated BAECs. Extracts of untreated (lane 1) or ASA-treated (100 mM) (lane 2) BAECs contained no detectable amounts of cyclooxygenase-2-protein, detectable level is 1 – 10 ng of protein [21]. In contrast, after incubation for 6 h (lane 4 ) or 12 h (lane 5) with PMA (20 nM) BAECs contain a protein of approximately 70 kDa, which was recognised by a specific antibody to cyclooxygenase-2. The induction of cyclooxygnase-2 protein by PMA in BAECs was abolished by dexamethasone (2 mM) (lane 3). Equal amounts of protein were loaded in all lanes.
Effects of NSAIDs on COX-1/2-activity in untreated and PMA-activated BAECs We used six different NSAIDs to determine IC50-values (Table 1) and concentration-response curves (an example is given in Fig. 6) for both COX-1 and COX-2. The rank order and potency of data shown in Table 1 are in line with several previously published IC50-values for COX-1/2 inhibitors [24, 25].
G. Dannhardt and H. Ulbrich
Inflamm. res.
COX-1 assay. To examine the effects of NSAID concentrations from 10–5 –10–10 mol/l on COX-1 intact BAECs were used. The transformation of AA to COX-1 products was initiated by addition of Ca-ionophore A 23187 (20 mM) in the presence of calcium. After 35 min the enzyme reactions were stopped by adding a mixture of methanol-acetonitrile containing nordihydroguaiaretic acid (NDGA) as an oxygen scavenger and an internal standard. The test solution was diluted with water and the products were isolated using octadecyl-reversed phase extraction. In contrast to Mitchell et al. [12] the methanol eluate (2.55 ml) was not concentrated before HPLC analysis, but was diluted with water (2.55 ml). The solvent strength of the resulting mixture was weaker than that of the eluent, allowing the injection of relatively large sample volumes. This was an improvement compared to the previously described procedures [12] to avoid subjecting of the sensitive samples to stress of concentration; in addition, the time of analysis was reduced. COX-2 assay. Endogenous COX-1 was irreversibly inhibited by pretreatment of BAECs with acetylsalicylic acid [23] and COX-2 was induced by PMA. The activated BAECs were incubated and the arachidonic acid metabolites were quantified by peak area ratio. UV absorbance spectra of the column eluate were monitored for wavelengths at 200 to 400 nm. For the COX-1/2 assay we used the 6-keto PGF1a at 200 nm and the 12-HHT peak at 235 nm. As an indicator of 15and 12-lipoxygenase activity we used 15- or 12-HETE peaks respectively, and detection at 235 nm.
Effects of NSAIDs on 15- and 12-HETE-production in untreated and PMA-activated BAECs HPLC analysis of eicosanoids produced by untreated BAECs showed a decrease of 6-keto PGF1a and 12-HHT after incubation with various NSAIDs in a concentration dependent manner, indicating COX-1 inhibition. Interestingly we found that all tested COX inhibitors markedly increased the formation of both 15-HETE and 12HETE (Fig. 7). Treatment of PMA-activated BAECs (COX-2 assay) with the NSAIDs caused a dose-dependent decrease of 6keto PGF1a , 12-HHT and 15-HETE but not 12-HETE (data not shown), indicating that 6-keto PGF1a , 12-HHT and 15HETE are products of the COX-2 isoenzyme. The mean values of at least two separate experiments are presented. These data are in agreement with those of Miralpeix and Wohlfeil [13, 22].
Discussion Fig. 6. Inhibitory effect of indomethacin on COX-1 and COX-2 activity. (B) The effect of indomethacin (0.01– 100 mM) on COX-2 was evaluated by measuring the production of 6-keto PGF1a () and 12-HHT () by BAECs pretreated by ASA (100 mM) for 30 min and activated with PMA (20 nM) for 6 h; (A) inhibitory effect of indomethacin (0.001-100 mM) on production of 6-keto PGF1a () and 12-HHT () by unstimulated BAECs (COX-1). The cells were treated with indomethacin for 15 min and stimulated for 35 min with Ca-ionophore A 23187. The values are the mean of at least six stimulation experiments, ± SEM is shown by the bar.
The time- and concentration-dependent induction of COX-2 expression in BAECs was evaluated by a kinetic profile. The optimal conditions were fixed at a molarity of 20 nM/ml PMA and an incubation time of 6 h. Various NSAIDs (diclofenac, indomethacin, meloxicam, aceclofenac, DFU, ClSC57666) in different concentrations (10–0.0001 M) were used to validate our assay: The potencies of the different COX-1/2 inhibitors were expressed as IC50-values.
Vol. 50, 2001
COX-1/-2 assay using BAECs
267
Fig. 7. Increase of 15- and 12-HETE levels of untreated BAECs after NSAID [1 mM] incubation. The cells were treated with vehicle (control) or NSAID for 15 min and stimulated for 35 min with Ca-ionophore A 23187 (20 mM). The values are the mean of at least six stimulation experiments, ± SEM is shown by the bar.
Fig. 8. Decrease of 15-HETE concentrations in PMA-activated BAECs treated with NSAIDs. After treatment with ASA (100 mM) for 30 min, the COX-2 was induced with PMA (20 nM) for 6 h. Then cells were treated with vehicle (control) or NSAID for 15 min and stimulated for 35 min with Ca-ionophore A 23187. The values are the mean of at least six stimulation experiments, ± SEM is shown by the bar.
The rank order of potency against COX-1 is indomethacin > diclofenac > meloxicam > aceclofenac > Cl-SC57666 > DFU. IC50-values ranged from 8 nM to > 100 mM. A similar ranking was found using human platelets 24 or using transfected CHO cells as a source of COX-1 (Riendeau et al. [25]). The ability of indomethacin, diclofenac, meloxicam, DFU and Cl-SC57666 to inhibit COX-2, was assayed in BAECs in which COX-2 activity was induced by PMA stimulation. The IC50-values for the nonselective inhibitors indomethacin, diclofenac and meloxicam were 0.004, 0.03 and 0.6 mM respectively. The selective COX-2 inhibitors DFU and ClSC57666 inhibited COX-2 activity in these cells with IC50values of 0.04 and 0.001 mM respectively. The data are consistent with those reported by Riendeau et al. [25]. To test the hypothesis that variation of the carboxylate moiety of known NSAIDs can generate COX-2 selective inhibitors [26], we tested aceclofenac (2-[2,6-dichlorophenyl) amino phenylacetoxyacetic acid). According to its structure it could be a prodrug of diclofenac. Therefore we tested the time-dependent stability of aceclofenac by incubating 20 mg with BAECs at 37 °C. The incubation mixtures were analysed by HPLC. In contrast to Yamazaki et al. [27] all data suggest that, under the conditions used, aceclofenac is not converted to diclofenac or other COX-1/COX-2 inhibiting metabolites (data not shown). This is in agreement with the different COX-1/2 inhibiting potencies of aceclofenac and diclofenac (IC50: 7,3/3,0 mM (Ace) vs. IC50: 0.01/0.03 m mM (Diclo)). Contrary to Kalgutkar et al. [26], esterization of the carboxylate moiety of diclofenac (to yield aceclofenac) decreased its ability to inhibit COX-1 and COX-2. Lipoxygenases (LOX) are lipid-peroxidating enzymes that are implicated in the pathogenesis of a variety of inflammatory disorders. 15-LOX produces 15 (S)–HPETE, which is converted to 15 (S)HETE. The 15-LOX subfamily is able to oxidize low density lipoproteins (LDL), which thereafter play a central role in atherosclerosis [28]. 15- and 12-HETE release in human rheumatoid arthritis (RA) type B synoviocytes is modulated by various cytokines including IL-4 and IL-1 that are important in the pathophysiology of RA [29]. 12 (S)-HETE has been found to play a major role in many cellular processes, such as thrombocyte aggregation, chemotactic stimulation of leukocytes, synaptic transmission, tumor cell metastasis, and cellular apoptosis [30]. Elevated levels of 12 (S)-HETE are also connected with immune disorders, such as inflammatory bowel disease and psoriasis.
Table 1. Inhibition of BAECs COX-1 and PMA-induced BAECs COX-2 by cyclooxygenase inhibitors Inhibitor
IC50-values (µM), ± SEM COX-1
Indomethacin Diclofenac Aceclofenac Meloxicam DFU Cl-SC57666
n=3 n=4 n=3 n=2 n=2 n=2
0.008 0.010 7.30 0.40 > 100 > 100
COX-2 ± 0.001 ± 0.001 ± 0.07 ± 0.20 – –
n=3 n=4 n=3 n=2 n=3 n=5
0.04 0.03 3.00 0.60 0.04 0.0010
ratio COX-2/COX-1 ± 0.02 ± 0.09 ± 0.69 ± 0.30 ± 0.04 ± 0.0018
5 3 0.4 1.5 – –
268
According to these data, reduced 15- and 12-HETE concentrations may contribute to the antiinflammatory effects of NSAIDs in the therapy of RA. A model which simultaneously allows determination of prostaglandins and HETEs could be useful for the investigation of new mechanisms of action of NSAIDs. Interestingly we found that all COX-1/2 inhibitors [1 mM] tested markedly enhanced the formation of both 15-HETE compared with control (100 %) (indomethacin 20 %, diclofenac 81 %, aceclofenac 26 %, meloxicam 30 %) and 12-HETE (indomethacin 158 %, diclofenac 81%, aceclofenac 35 %, meloxicam 40 %) in nonstimulated BAECs in the COX-1 assay. These findings support the hypothesis of the so-called shunt to the 15- and 12-LOX pathways by inhibiting COX-1. The stimulation of BAECs, the mobilisation of calcium and the activation of phospholipase A2 are the initial steps necessary for the biosynthesis of PGs and leukotrienes. A consequence of COX-1 inhibition is the shift to other AA-utilizing pathways. In the COX-2-assay there is no shunt to other arachidonic acid utilising pathways for example lipoxygenases. All the NSAIDs [1 mM] tested markedly inhibited the release of 15HETE from ASA-treated and PMA-activated BAECs compared with control (100%) (indomethacin 22%, diclofenac 57%, DFU 51%, Cl-SC57666 39%, aceclofenac 10 mM: 40%, meloxicam 25%). These data are consistent with those of Miralpeix and Wohlfeil [13, 22]. To clarify the differences in the release of HETEs in the COX-1 and COX-2 assay the stereochemistry of 15-HETE, the effects of PMA activation on lipoxygenases and the characterisation of the biochemical alteration of LOX induced by high doses of aspirin, is needed to be studied. Several of the drawbacks and problems of COX-1/COX2 assays have been resolved in this assay. The duration of incubation with the drugs, which is a critical issue with regards to the time dependency of COX inhibition, are the same in both assays and prostaglandin release is measured from endogenous AA stores to avoid artificial conditions. The whole-cell in vitro assay presented uses intact endothelial cells, which are target cells for the anti-inflammatory activity of NSAIDs. COX-2, which is not constitutively expressed in an artificial cell line, was induced by simulating an inflammatory process with PMA. In summary, we have achieved the development of a method allowing the determination of IC50 values for inhibitors of COX-1/2, which eliminate the drawbacks of other assays published. The inhibiting potencies evaluated are consistent with those in the literature. In addition, the effects of NSAIDs on LOX activity may also be studied. Molecular cloning studies and structured variations are under investigation to provide new insights into the relationship of COX-1/2 and LOX pathways in BAECs. Reference [1] Xie WL, Chipman JG, Robertson DL, Erikson RL, Simmons DL. Expression of a mitogen-responsiveness gene encoding prostaglandin synthase is regulated by mRNA splicing. Proc Natl Acad Sci USA 1991; 88: 2692–6. [2] Kujubu DA, Fletcher BS, Varnum BC, Lim RW, Herschman HR. TIS10, a phorbol ester tumor promoter-inducible mRNA from Swiss 3T3 cells, encodes a novel prostaglandin syn-
G. Dannhardt and H. Ulbrich
[3] [4] [5]
[6] [7] [8] [9] [10]
[11]
[12]
[13]
[14] [15] [16]
[17]
[18]
[19] [20] [21]
[22]
Inflamm. res.
thase/ cyclooxy-genase homologue. J Biol Chem 1991; 266: 12866–72. De Witt DL. Prostaglandin endoperoxide synthase: Regulation of enzyme expression. Biochim Biophys Acta 1991; 1083: 121–34. Herschman HR. Prostaglandin synthase 2. Biochim Biophys Acta 1996; 1299: 125–40. Kömhoff M, Gröne HJ, Klein T, Seyberth HW, Nüsing RM. Localization of cyclooxygenase-1 and -2 in adult and fetal human kidney: implication for renal function. Am J Physiol 1997; 272: F460–68. Beiche F, Scheuerer S, Brune K, Geisslinger G, Goppelt-Struebe M. Up-regulation of cyclooxygenase-2 mRNA in the rat spinal cord following peripheral inflammation. FEBS-Lett 1996; 390: 165–9. Parfenova H, Eidson TH, Leffler CW. Upregulation of COX-2 in cerebral microvascular endothelial cells by smooth muscle cell signals. Am J Physiol 1997; 273: C277–88. Zimmermann K, Weber AA, Sarbia M, Schrör K. Cyclooxygenase-2 is constitutively Expressed in Stomach Mucosa. Arch of Pharm 1998; 357: 33. Dannhardt G, Lehr M. In-vitro Evaluation of 5-Lipoxygenase and Cyclooxygenase Inhibitors using Bovine Neutrophils and Platelets and HPLC. J Pharm Pharmacol 1992; 44: 419–24. Dannhardt G, Flemmer L, Hartmann RW, Kleber A, Schulze E. Spectrofluorimetric Quantification of Malondialdehyde for Evaluation of Cyclooxygenase-1/Thromboxane Synthase Inhibition. Arch. Pharm. Med. Chem. 1998; 331: 359–64. Dannhardt G, Kreher M, Nowe U, Pies A. Method for Testing Non-Steroidal Anti-Inflammatories: The Modified Hen’s Egg Chorioallantoic Membrane Test (HET-CAM Test) Compared to Other Procedures. Arch. Pharm. Med. Chem. 1996; 329: 301–10. Mitchell JA, Akarasereenont P, Thiemermann C, Flower JF, Vane JR. Selectivity of nonsteroidal antiinflammatory drugs as inhibitors of constitutive and inducible cyclooxygenase. Proc Natl Acad Sci USA 1993; 11693–97. Wohlfeil ER, Campbell WB. 25-Hdroxycholesterol enhances eicosanoid production in cultered bovine coronary artery endothelial cells by increasing prostaglandin G/H synthase-2. Biochim Biophys Acta 1997; 1345: 109–20. Gimbrone MA. Culture of vascular endothelium. In: Spaet TH. editor Progress in Hemostasis and Thrombosis. vol. 3. New York: Grune and Stratton, 1976: 1–28. Jaffe EA, Hayer LW, Nachman RL. Synthesis of antihemophilic factor antigen by cultured human endothelial cells. J Clin Invest 1973; 52: 2757–64. Li JJ, Anderson GD, Burton EG, Cogbum JN, Collins JT, Garland DJ et al. 1,2-Diarylcyclopentenes as selective cyclo-oxygenase-2 inhibitors and orally active anti-inflammatory agents. J Med Chem 1995;38:4570–8. Akarasereenont P, Mitchell JA, Appleton I, Thiemermann C, Vane JR. Involvement of tyrosine kinase in the induction of cyclo-oxygenase and nitric oxide synthase by endotoxin in cultured cells. Br J Pharmacol 1994; 113: 1522–8. Tordjman C, Serkiz B, Rambaud F, Bonnet J, Volland JP. High performance liquid chromatographic quantitation of cyclooxygenase and lipoxygenase metabolites of arachidonic acid from rat polymorphonuclear leukocytes. J Chromatogr. 1990; 532: 135– 43. Ibe BO, Falck JR, Johnson AR, Campbell WB. Regulation of prostacyclin and HETEs in human endothelial cells. Am J Physiol 1989; 256: C 1168–75. Hollenberg SM, Tong W, Shelhamer JH, Lawrence M, Cunnion RE. Eicosanoid production by human aortic endothelial cells in reponse to endothelin. Am J Physiol 1994; 267: H2290–6. Akarasereenont P, Mitchell JA, Appleton I, Thiemermann C, Vane JR. Comparison of the induction of cyclooxygenase and nitric oxide synthase by endotoxin in endothelial cells and macrophages. Eur J Pharmacol 1995; 273: 121–8. Miralpeix M, Camacho M, Lopez-Belmonte J, Canalias F, Beleta J, Palacios JM et al. Selective induction of cyclo-oxygenase-2 activity in the permanent human endothelial cell line HUV-EC-C:
Vol. 50, 2001
[23]
[24]
[25]
[26]
COX-1/-2 assay using BAECs
biochemical and pharmacological characterization. Br J Pharmacol 1997; 121: 171–80. Patrignani P, Panara MR, Greco A, Fusco O, Natoli C, Iacobelli S, et al. Biochemical and pharmacological characterization of the cyclooxygenase activity of human blood prostaglandin endo-peroxidesynthases. J Pharmacol Exp Ther 1994; 271: 1705–12. Kawai S. Cyclooxygenase selectivity and the risk of gastrointestinal complications of various non-steroidal anti-inflammatory drugs: A clinical consideration. Inflamm Res 1998; 47: 102–6. Riendeau D, Percival MD, Boyce S, Brideau C, Charleson S, Cromlish W. et al Biochemical and pharmacological profile of a tetrasubstituted furanone as a highly selective COX-2 inhibitor. Br J Pharmacol 1997; 121: 105–17. Kalgutkar AS, Crews BC, Rowlinson SW, Marnett AB, Kozak KR, Remmel, PP et al. Biochemically based design of cyclooxygenase-
269
[27]
[28] [29]
[30]
2 (COX-2) inhibitors: Facile conversion of nonsteroidal antiinflammatory drugs to potent and highly selective COX-2 inhibitors. Proc Natl Acad Sci 2000; 97: 925–30. Yamazaki R, Kawai S, Matsuzaki T, Kaneda N, Hashimoto S, Yokokura T, et al. Aceclofenac blocks prostaglandin E2 production following its intracellular conversion into cyclooxygenase inhibitors. Eur J Pharmacol 1997; 329: 181–7. Feinmark SJ, Cornicelli JA. Is there a role for 15-lipoxygenase in atherogenesis? Biochem Pharmacol 1997; 54: 953–59. Liagre B, Vergne P, Rigaud M, Beneytout JL. Arachidonate 15lipoxygenase of reticulocyte-type in human rheumatoid arthritis type B synoviocytes and modulation of its activity by proinflammatory cytokines. J Rheumatol 1999; 26: 1044–51. Dailey LA, Imming P. 12-Lipoxygenase: Classification, Possible Therapeutic Benefits from Inhibition and Inhibitors. Curr Med Chem 1999; 6: 389–98.
