Involvement of MAP kinases in the control of cPLA2 and arachidonic acid release in endothelial cells

Atherosclerosis 156 (2001) 81 – 90 www.elsevier.com/locate/atherosclerosis

Involvement of MAP kinases in the control of cPLA2 and arachidonic acid release in endothelial cells Ingibjo¨rg J. Gudmundsdo´ttir a, Haraldur Halldo´rsson a,b, Kristı´n Magnu´sdo´ttir a, Gudmundur Thorgeirsson a,b,* a

Department of Pharmacology, Uni6ersity of Iceland, PO Box 8216, 128 Reykja6ik, Iceland b Department of Medicine, Landspı´talinn, Uni6ersity Hospital, Reykja6ik, Iceland

Received 17 January 2000; received in revised form 8 June 2000; accepted 17 August 2000

Abstract Cytosolic Phospholipase A2 (cPLA2) has been implicated in receptor-mediated release of arachidonic acid from membrane phospholipids, the limiting step in prostacyclin and other eicosanoid production. Its activity is controlled by Ca++ levels and enzymatically regulated phosphorylation. The purpose of this study was to assess the importance of phosphorylation of cPLA2 in human umbilical vein endothelial cells and to identify the kinases involved. Inhibitors were used to study the pathways leading to phosphorylation and activation of mitogen activated protein kinases (MAP-kinases) and cPLA2, as well as release of arachidonic acid and prostacyclin production after stimulation with different agonists. We have found that agonists that release arachidonic acid, including histamine, thrombin, AlF− 4 , and pervanadate, all activate the MAP kinases ERK, p38 and JNK and cause phosphorylation of cPLA2. Agonist specific differences in the signal transduction pathways included variable contribution of tyrosine phosphorylation, protein kinase C and ERK activity, and different effects of pertussis toxin. Treatment with PD98059 (inhibitor of ERK-activation) or SB203580 (inhibitor of p38) caused partial decrease in arachidonic acid release and cPLA2 activity. In contrast the nonspecific protein kinase inhibitor staurosporin completely inhibited cPLA2 activity. We conclude that in endothelial cells arachidonic acid release is largely mediated by cPLA2 through agonist-specific pathways. The MAP kinases ERK and p38 both have demonstrable but not major effect on agonist stimulated arachidonic acid release and the data suggest that an additional unidentified kinase also has a role. © 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Endothelium; MAP-kinases; cPLA2; AA-release

1. Introduction Metabolites of arachidonic acid, the eicosanoids, constitute a large and expanding family of biologically active compounds that include prostaglandins, thromboxanes and leukotrienes. Prostacyclin is among the most active compounds released by endothelial cells, causing both vasodilation and inhibition of platelet aggregation. The limiting step in prostacyclin production is Phospholipase A2 mediated liberation of arachidonic acid from membrane phospholipids. Recent work indicates that arachidonic acid may play a role as a signal molecule in its own right, further underscoring * Corresponding author: Tel.: + 354-568-0866; fax: + 354-5680872. E-mail address: [email protected] (G. Thorgeirsson).

the physiological importance of arachidonic acid release [1,2]. Phospholipase A2 comprises a family of enzymes that catalyze the hydrolysis of the sn-2 fatty-acyl bond of phospholipids to liberate free fatty acids [3]. Cytosolic Phospholipase A2 (cPLA2), an 85 kD enzyme which preferentially liberates arachidonic acid, has been implicated in receptor mediated eicosanoid production and intracellular signal transduction processes [4]. The importance of cPLA2 has recently been demonstrated by the generation of knockout mice, which showed markedly reduced production of both prostaglandins, leukotrienes and platelet activating factor in peritoneal macrophages [5,6]. The activity of cPLA2 is controlled by submicromolar changes in the cytosolic concentration of Ca++ and by an enzymatically regulated phosphorylation. Increased intracellular Ca++ induces the

0021-9150/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 0 2 1 - 9 1 5 0 ( 0 0 ) 0 0 6 3 1 - 6

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

82

association of cPLA2 with cellular membranes, and a wide variety of stimuli have been found to induce its phosphorylation [4]. The main phosphorylation sites are ser-505 and ser-727 [7]. The importance of these

phosphorylations for activation of cPLA2 as well as the identity of the kinase(s) have not been fully elucidated. After Lin and coworkers demonstrated that the MAP kinase ERK could phosphorylate cPLA2 in vivo [8], the

Table 1 Involvement of kinases and G-proteins in agonist stimulated arachidonic acid releasea Agonist

Thrombin (1 U/ml)

Histamine (11 mmol/l)

Pervanadate (20 mmol/l)

Pretreatment [ 3H] Arachidonic acid release as % of total radioacti6ity in lipid fraction None 4.14 2.95

19.8

[ H] Arachidonic acid release as % of agonist alone 9 SEM Genistein (100 mmol/l) 599 13 Staurosporine (0.2 mmol/l) 129925 CGP 41251 (2 mmol/l) 99910 TPA (200 ng/ml, 20 h) 125920 Pertussis toxin (100 ng/ml, 20 h) 789 4

36 9 10 893 27 913 34 9 14 59 92

AlF− 4 (30 mmol/l)

3.15

3

86 926 98 923 73 95 67 921 92 9 3

70 9 30 23 9 13 22 96 25 914 8 94

a Data for cells without pretreatment are expressed as percentage of total radioactivity in the lipid fraction. Data for pretreated cells are expressed as percentage of the response to the agonist without pretreatment 9 SEM from three experiments, each done in duplicate.

Table 2 Involvement of kinases and G-proteins in agonist stimulated prostacyclin productiona Prostacyclin production Agonist Pretreatment 6 -keto-PGF1h (ng/ml) None Prostacyclin production as % of agonist Genistein (0.1 mmol/l) Staurosporine (0.2 mmol/l) CGP 41251 (2 mmol/l) TPA (200 ng/ml, 20 h) Pertussis toxin (100 ng/ml, 20 h) PD98059 (20 mmol/l)

Thrombin (1 U/ml)

34.4 alone 9 SEM 0 85914 699 19 1399 52 989 4 0

Histamine (11 mmol/l)

29.0 0 80 93 62 92 96955 111 910 0

Pervanadate (20 mmol/l)

70.0 13 9 6 219 11 69 927 98 937 112 919 0

AlF− 4 (30 mmol/l)

49.5 191 392 694 54 918 13 92 0

a

Data for cells without pretreatment are expressed as ng/ml of 6-keto-PGF1a measured by radioimmunoassay. Data for pretreated cells are expressed as percentage of the response to the agonist without pretreatment 9 SEM from three experiments, each done in duplicate. Table 3 Involvement of kinases and G-proteins in agonist stimulated ERK activationa ERK activation Agonist

Thrombin (1 U/ml)

Histamine (11 mmol/l)

Pervanadate (20 mmol/l)

AlF− 4 (30 mmol/l)

Pretreatment c.p.m. on filter None

8024

12192

35402

8424

ERK acti6ation as % of agonist alone 9 SEM Genistein (100 mmol/l) 259 11 Staurosporine (0.2 mmol/l) 49 2 CGP 41251 (2 mmol/l) 82919 TPA (200 ng/ml, 20 h) 130926 Pertussis toxin (100 ng/ml, 20 h) 73913

18 95 2 92 44 9 21 78 9 7 151 939

110 96 57910 106 96 7694 101 96

291 191 692 32 9 18 42 917

a Data for cells without pretreatment are expressed as c.p.m. on filter. Data for pretreated cells are expressed as percentage of the response to the agonist without pretreatment 9SEM from three experiments, each done in duplicate.

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

83

Fig. 1. Effect of protein kinase inhibitors on the phosphorylation of ERK 1/2. Confluent endothelial cells were stimulated for 10 min with thrombin (3 U/ml), histamine (33 mmol/l), pervanadate (15 mmol/l) or for 12 min with AlF− 4 (30 mmol/l). The inhibitors, PD98059 (25 mmol/l), CGP41251 (2.5 mmol/l), staurosporine (0.5 mmol/l) or genistein (0.1 mmol/l) were added at least 10 min before addition of the agonists. Lysates were electrophoresed and blotted as described under Section 2 and detected using antibody against ERK1 which also recognises ERK2. Cells treated with staurosporine and any of the agonists gave identical results to control cells (not shown).

Fig. 2. Tyrosine phosphorylation of the p38 MAP kinase by different agonists. Confluent endothelial cells were stimulated for 10 min with thrombin (3 U/ml), histamine (33 mmol/l) or for 12 min with AlF− 4 (30 mmol/l), A23187 (1 mmol/l), TPA (200 ng/ml) or pervanadate (15 mmol/l). Lysates were electrophoresed and blotted as described under Section 2 and detected using antibody against the tyrosine phosphorylated moiety of p38.

ERKs have often been assumed to be the physiological activators of cPLA2 after receptor stimulation. However, it has recently become clear that this is not always the case [9,10]. Protein kinase C (PKC) has also been implicated in the phosphorylation of cPLA2 but its effect is probably indirect. Both PKC dependent and independent pathways of MAP kinase and cPLA2 phosphorylation have been identified [10,11]. In platelets cPLA2 is phosphorylated by p38 MAP kinase as well as an unidentified kinase, [12,13] whereas ERK has a limited role [12–15]. In endothelial cells, involvement of ERK in cPLA2 activation and arachidonic acid release has been demonstrated in response to basic fibroblast growth

factor [16] and lysophosphatidylcholine [17]. p38 MAP kinase is found to mediate actin reorganisation and cell migration in human umbilical vein endothelial cells (HUVEC) [18], but its involvement in arachidonic acid release has not been studied. The purpose of this study was to investigate the pathways leading to arachidonic acid release and prostacyclin production in human endothelial cells exposed to different agonists. By the use of selective inhibitors of MAP-kinase pathways [19], tyrosine phosphokinases, tyrosine phosphatases, PKC and G proteins we have examined the involvement of the signalling components causing phosphorylation and activation of the MAP kinases and cPLA2.

84

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

2. Materials and methods

2.1. Materials Morgan’s medium 199, foetal calf serum, antibiotics and genistein were from Gibco; cod serine collagenase was obtained from the Science Institute, University of

Iceland; PD98059 and palmityltrifluoromethyl ketone (PACOCF3) were from Biomol; SB203580 from SmithKline Beecham; the staurosporine analog CGP41251 was a generous gift from Ciba-Geigy (Summit, NJ); thrombin, histamine, A23187, TPA (12-O-tetradecanoylphorbol 13-acetate), staurosporine, pertussis toxin, arachidonic acid, catalase, aprotinin, pepstatin, leupeptide and phenylmethylsulfonylfluoride from Sigma; 1-stearoyl-2[1-14C] arachidonyl phosphatidylcholine (50.0 mCi/mmol), [3H] arachidonic acid, nitrocellulose membranes, ECL western blotting detection reagents and films, and p42/p44 MAP kinase enzyme assay system from Amersham International; rabbit polyclonal antibodies against MAP kinases (ERK1/2 and p38), mouse monoclonal cPLA2 antibodies, antirabbit and anti-mouse IgG-HP from Santa Cruz Biotechnology; antibody against phosphorylated p38 MAP kinase from New England BioLabs; pervanadate was prepared as described by Pumiglia and coworkers [20]. The pervanadate solution was used within half an hour.

2.2. Endothelial cell culture

Fig. 3. Activity of cPLA2. (A) Effect of agonists on the in vitro activity of cPLA2. Confluent endothelial cells were stimulated for 10 min with thrombin (3 U/ml) or for 12 min with pervanadate (20 mmol/l). (B) Effect of protein kinase inhibitors on the activation of cPLA2 and effect of potato acid phosphatase on its activity. The inhibitors, genistein (0.1 mmol/l) and staurosporine (0.5 mmol/l) were added 10 min before stimulation with thrombin (3 U/ml). The phosphatase (1 mg/ml) was added to the supernatant and incubated for 5 min at 20°C before the enzyme assay (black columns). The in vitro cPLA2 assay was performed as described under Section 2. The data are averages of duplicates from representative experiments and are expressed as percentages of cPLA2 activity in untreated cells.

Fig. 4. Mobility shift of cPLA2. Effect of agonists and protein kinase inhibitors. The inhibitors, genistein (0.1 mmol/l) or staurosporine (0.5 mmol/l) were added to confluent endothelial cells  10 min before stimulation for 10 min with thrombin (3 U/ml) or 12 min with pervanadate (20 mmol/l). Lysates were electrophoresed and blotted as described under Section 2 and detected using antibody against cPLA2.

As previously described, endothelial cells were cultured from human umbilical veins by a modification of the method of Jaffe and co-workers [21]. After harvesting the cells by 0.1% collagenase digestion, they were seeded on 80 cm2 culture flasks (for cPLA2 assay) or 35 mm culture dishes (for electrophoresis, arachidonic acid release, prostacyclin production and MAP-kinase activity) in Morgan’s medium 199 containing 20% foetal calf serum and antibiotics (penicillin, 100 units/ml and streptomycin, 100 mg/ml). The culture dishes were incubated at 37°C in humidified air with 5% CO2. The medium was changed 24 h after seeding the cells and every third day thereafter until the cell culture reached confluence. Only primary cultures were used.

2.3. Arachidonic acid release Confluent cells were incubated for 24 h in medium 199 containing 20% foetal calf serum, antibiotics, and 1 mCi of (3H) arachidonic acid per milliliter. Before the experiments, the cells were washed twice with medium containing BSA (1mg/ml, fatty acid free), and kept in this solution with or without the different inhibitors. After 20 min, a portion of the medium was removed and an equal volume of a solution containing the agonist was added to give the indicated final concentrations. At the times indicated, another portion of the medium was removed. All portions were counted in a scintillation counter for quantification of arachidonic acid and its labelled metabolites. The rest of the medium was used for measurement of prostacyclin production.

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

85

Table 4 The effects of PD98059 and SB203580 on agonist stimulated arachidonic acid releasea Inhibitor

PD98059 (20 mmol/l)

SB203580 (20 mmol/l)

PD98059+SB203580 (20 mmol/l each)

191 9 26 6298 48 9 5 67 914 67 9 16 65 99 55 9 8

202 924 77913 4992 45 9 10 18 9 2 99 9 709 8

Agonist [3H]Arachidonic acid release as % of agonist alone 9 SEM None 1159 7 Thrombin (1 U/ml) 7997 Histamine (11 mmol/l) 629 2 Pervanadate (20 mmol/l) 6296 AlF− 3595 4 (30 mmol/l) TPA (100 ng/ml) 20913 A23187 (0.4 mmol/l) 7096

a Data for pretreated cells are expressed as percentage of the response to the agonist without pretreatment 9 SEM from three experiments, each done in duplicate.

2.4. Prostacyclin production To measure PGI2 production of the cells, the medium was subjected to a radioimmunoassay for 6-oxo-PGF1a, a stable catabolite of PGI2, as previously described [21].

2.5. MAP kinase acti6ity A p42/p44 MAP kinase enzyme assay system was used to measure ERK activity after stimulation of cells. The assay is based on the catalysed transfer of a (32P)-radiolabelled-g-phosphate group of ATP to a peptide that is selective for p42/p44 MAP kinase. After stimulation, cells were washed in lysis buffer (mmol/l): Tris 10, NaCl 150, EGTA 2, orthovanadate 1, PMSF 1, mercaptoethanol 0.1%, leupeptin 10 mg/ml, aprotinin 10 mg/ml and pepstatin 10 mg/ml. Cellular debris was precipitated at 21 000×g for 30 min, 15 ml of the supernatant then added to 10 ml substrate and 5 ml magnesium-(32P)-ATP buffer. After mixing and brief centrifuging, the mixture was incubated for 30 min at 30°C, and 10 ml of stop reagent added to quench the reaction. The phosphorylated peptides were separated from the unbound phosphate on a specific binding-paper and the radioactive incorporation counted in a scintillation counter.

mogenizer, centrifuged at 21 000× g for 30 min at 4°C, and the supernatant used for the enzyme assay by the method of Lin et al. [22]. Assay substrate was prepared by drying 14C- arachidonic acid-phosphatidylcholine under N2 in an Eppendorf vial, resuspending in DMSO by vortexing, adding 10 mmol/l Tris-HCl pH 7.4 to the mixture and sonicating for 5 min. The reaction was started by mixing 2 ml of substrate, 4 ml of 50 mmol/l CaCl2 and 34 ml of sample and the mixture incubated at 37°C for the required time (10 or 20 min). The reaction was stopped by the addition of 40 ml quench solution (2% v/v acetic acid in ethanol with 0.5 mg/ml arachidonic acid). For separation, 5× 10 ml of each sample were spotted onto a heat activated TLC plate. The plate was put in a pre-equilibrated tank with the non aqueous phase of a mixture of ethyl acetate, iso-octane, acetic acid and water (55:75:8:100 v/v/v/v). After evaporation, the lipids were visualized by I2 staining. The band containing free arachidonic acid was then scraped and quantified by scintillation counting.

2.6. cPLA2 acti6ity For in vitro determination of cPLA2 activity, the cells were cultured to confluence in 80 cm2 flasks. After washing, the cells were kept in 5 ml medium 1999 inhibitors for 10 min, the agonists added and the cells incubated for additional 10 min. The medium was then removed and the cells washed in homogenizing buffer (mmol/l): sucrose 250, HEPES 50, EDTA 1, EGTA 1, PMSF 1, aprotinin 0.02 and leupeptin 0.02. A total of 1 ml of homogenizing buffer was then added and the cells scraped off the flask with a rubber policeman and homogenized with 15 strokes in a Potter-Elvehjem ho-

Fig. 5. The effect of PD98059 and SB203580 on the activation of cPLA2. The inhibitors (20 mmol/l) were added to the cells 10 min before stimulation for 10 min with thrombin (3 U/ml). The in vitro cPLA2 assay was performed as described under Section 2. The data are expressed as percentage of cPLA2 activity in untreated cells and are the means 9SEM of four determinations, each done in duplicate.

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

86

Fig. 6. The effect of PD98059 and SB203580 on the mobility shift of cPLA2. The inhibitors (25 mmol/l) were added to the cells at least 10 min before stimulation with thrombin (3 U/ml) or histamine (33 mmol/l) for 10 min, or pervanadate (20 mmol/l), TPA (200 ng/ml), AlF− 4 (30 mmol/l) or A23187 (1 mmol/l) for 12 min. Lysates were electrophoresed and blotted as described under Section 2 and detected using antibody against cPLA2.

2.7. Electrophoresis and immunoblotting For the mobility shift experiments of ERK and cPLA2 and the detection of phosphorylated p38, confluent endothelial cells were stimulated with agonists for the indicated times in 1 ml of Morgan’s medium 199. Inhibitors were added 10 min before the agonist. The reactions were stopped by aspirating the medium and adding 200 ml of SDS sample buffer, after which the samples were boiled for 5 min and centrifuged for 5 min before electrophoresis on 8 or 10% SDS-PAGE. Separated proteins were then transferred to nitrocellulose in a semi dry transfer unit (Hoefer) for 90 min (ERK and p38) or 4 h (cPLA2) at 40 mA. After blocking with 1% BSA in wash buffer: Tris 0.01 mmol/l, pH 7.5, NaCl 0.1 mmol/l and 0.1% Tween 20 overnight at 4°C, the nitrocellulose membranes were probed with antibodies to ERK, cPLA2 or to phosphorylated p38 and the immunocomplexes detected with horseradish peroxidase conjugated to rabbit or mouse IgG, using the ECL substrate system. 3. Results

3.1. Agonist dependent signal transduction pathways for endothelial arachidonic acid release and prostacyclin production To study the involvement of kinases in the arachidonic

acid release induced by various agonists, we used genistein to inhibit protein tyrosine kinases on one hand, and three types of treatments affecting PKC on the other. They were: long term treatment with TPA to downregulate PKC; CGP41251, a specific inhibitor of PKC; and the unspecific protein kinase inhibitor staurosporine. The selective G-protein inhibitor pertussis toxin was used to evaluate G-protein involvement in signal pathways by various agonists. Table 1 shows the arachidonic acid release by endothelial cells when stimulated with thrombin, histamine, pervanadate or AlF-4 after various inhibitory pretreatments. The effects of the pretreatments varied between the agonists. The response to thrombin was dependent on tyrosine phosphorylation but totally independent of PKC. The arachidonic acid response to histamine was the opposite to that of thrombin, dependent on PKC but virtually unaffected by genistein. The effect of pertussis toxin treatment on the histamine response was minimal while the effect on the thrombin response was more marked (22%). The protein tyrosine phosphatase inhibitor, pervanadate, is a powerful stimulator of endothelial cells, inducing tyrosine phosphorylation of many proteins, generation of inositol phosphates and production of prostacyclin [23]. Genistein inhibited the arachidonic acid response to pervanadate and the response was also strongly dependent on PKC activity. After CGP 41251 inhibition or long term TPA treatment there was only about one third of the pervanadate activity left, and after staurosporine treatment only 8%. Of the four agonists was most tested the arachidonic response to AlF− 4 sensitive to PKC inhibition, leaving only one fourth of the response. The stimulatory effect of this direct Gprotein activator was completely knocked out by pertuspathway was also affected by sis toxin. The AlF− 4 genistein, although much less than by other inhibitory treatments. Table 2 shows the prostacyclin production by endothelial cells subjected to the same manipulations as shown in Table 1 for arachidonic acid release. When the results are compared the main difference is that genistein blockage of arachidonic acid release was agonist dependent while genistein completely blocked all stimulated prostacyclin production. Genistein also inhibited prostacyclin production from exogenously added arachidonic acid (not shown), suggesting that it might inhibit cyclooxygenase. Table 3 shows results from the in vitro activity assay on ERK obtained from endothelial cells under the same conditions as above (Tables 1 and 2). Again, the effects of inhibitors on stimulation varied greatly between agonists. Pervanadate was the most resistant to all inhibition except to that of long-term TPA treatment, whereas AlF− 4 was most sensitive to all inhibitory ma-

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

nipulations. Both thrombin and histamine activation of ERK were dependent on tyrosine kinase(s). As with arachidonic acid release, ERK activation by histamine was PKC dependent whereas thrombin induced ERK activation was not. There was a striking difference between these two agonists after pertussis toxin pretreatment. It had a clear inhibitory effect on the thrombin stimulated response while enhancing the histamine response. When the results from arachidonic acid release and ERK activity are compared, the most notable differences between the two responses are seen with staurosporine which totally inhibited the ERK response to thrombin and histamine without any inhibitory effect on arachidonic acid release. Fig. 1 depicts the effect of protein kinase inhibitors on the phosphorylation of ERK 1 and 2 after stimulation with thrombin, histamine, pervanadate and AlF− 4 . Results are in full agreement with the ERK activity results in Table 3 apart from the pervanadate induced mobility shift which was totally inhibited by staurosporine. As expected, inhibition of ERK activation by PD98059 prevented mobility shift by all agonists. Fig. 2 shows the electrophoretic results obtained by the use of an antibody to the phosphorylated moiety of the MAP kinase p38. These results show that all the agonists tested caused phosphorylation of p38.

3.2. Agonist mediated acti6ation of cPLA2 in HUVEC. To study the contribution of phosphorylation in the activation of cPLA2 we investigated the effects of agonists and inhibitors on its activity in vitro under constant Ca++ concentration. Fig. 3(A) shows that thrombin and pervanadate increased the in vitro activity of cPLA2. Compared to control, thrombin (3 U/ml) caused 2.4-fold and pervanadate (20 mmol/l) 3.4-fold increase. Treatment with 0.1 mmol/l of PACOCF3, a specific inhibitor of cPLA2, caused 95% inhibition (data not shown), indicating that the activity measured was due to the cytosolic PLA2. The increased activity was clearly caused by phosphorylation, as it was reversed by treatment with potato acid phosphatase (Fig. 3(B)). The mobility shift data in Fig. 4 show that a small fraction of unstimulated cPLA2 is phosphorylated, whereas a major portion of the enzyme becomes phosphorylated after thrombin or pervanadate stimulation. The effects of staurosporine and genistein on agonist stimulated in vitro activity and phosphorylation of cPLA2 are shown in Fig. 3 (B), Fig. 4, respectively. Staurosporine inhibited all the in vitro activity caused by thrombin, and genistein most of it. Staurosporine inhibited the mobility shift caused by both thrombin and pervanadate, genistein inhibited only that by thrombin.

87

3.3. MAP kinase mediated modulation of arachidonic acid release, prostacyclin production and cPLA2 in HUVEC To further investigate the signals mediating arachidonic acid release after stimulation with various agonists, we employed inhibitors of ERK activation and p38 activity, PD98059 and SB203580 respectively (Table 4). While PD98059 had no effect on the basal release, it caused some inhibition of the arachidonic acid release triggered by all agonists and the responses to AlF− 4 and TPA were most sensitive to the inhibitory effects. SB203580 caused B 50% inhibition in the response to all agonists but doubled the arachidonic acid release of unstimulated cells. When both inhibitors were present, their effect was less than additive after stimulation with thrombin, histamine or A23187. For the remaining three agonists, the effect was additive. Despite the limited inhibition of arachidonic acid release, PD98059 caused total inhibition of prostacyclin production as shown in Table 2. The effect of these inhibitors on cPLA2 activity obtained from thrombin stimulated cells is shown in Fig. 5. PD98059 inhibited around 46% of the thrombin induced activation and SB203580 had a similar inhibitory effect. Surprisingly, when the inhibitors were added together, their inhibitory effect was abolished. The results from mobility studies of cPLA2 are presented in Fig. 6 showing that SB203580 and PD98059 caused partial inhibition of cPLA2 mobility shift induced by all the agonists tested. When the inhibitors were both present, the inhibition was complete for all agonists.

4. Discussion The principal findings of this study are: 1. cPLA2 is activated in HUVEC following treatment with several agonists which cause prostacyclin production, including thrombin, histamine, and the tyrosine phosphatase inhibitor pervanadate. 2. The MAP kinases ERK and p38 have demonstrable but not a major role in agonist induced arachidonic acid release. While inhibitors of ERK activation and p38 partly inhibit cPLA2 in the fixed calcium concentration environment of an enzyme assay, combination of such inhibitors did not fully inhibit its activity. In contrast, the nonspecific protein kinase staurosporin fully inhibited cPLA2 activity, suggesting the involvement of yet another kinase. 3. The signal transduction pathways involved are agonist dependent. Activation of arachidonic acid release and prostacyclin production includes differential contribution of PKC, ERK and tyrosine phosphorylation depending on the agonist involved.

88

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

The contribution of cPLA2 to the agonist induced release of arachidonic acid in HUVEC is supported by the general concordance between the results measuring arachidonic acid release, activity of cPLA2 and its mobility shift. In addition, treatment with the selective inhibitor PACOCF3 completely blocked all enzyme activity in the assay showing that cPLA2 is the enzyme involved. Secondly, the electrophoretic mobility shift is caused by phosphorylation and is detected with an antibody specific for cytosolic PLA2. Correspondingly, the enzyme was deactivated in the in vitro assay after treatment with potato acid phosphatase. These results as well as the agreement between the different study methods strongly suggest that cPLA2 is the enzyme responsible for the bulk of arachidonic acid release. Studies with endothelial cells stimulated with lysophosphatidylcholine on one hand [16] and basic fibroblast growth factor on the other [17] resulted in an analogous conclusion linking arachidonic acid release to cPLA2. The first MAP kinase to be implicated in cPLA2 phosphorylation was ERK [8]. We have observed that the agonists that cause release of arachidonic acid in HUVEC also cause ERK activation. As a first attempt to establish involvement of ERK in activation of cPLA2, we compared the effects of various pretreatments on arachidonic acid release and ERK activity (Tables 1 and 3). The results help in delineating the pathways engaged by the different agonists, but do not suggest a major involvement of ERK in regulating arachidonic acid release. Recently it has become clear that in addition to the ERKs, phosphorylation of cPLA2 can also be mediated by p38 MAP kinase(s) [9,10]. Since all the agonists we have tested activate p38, this MAP kinase clearly is a candidate for cPLA2 kinase in endothelial cells. To study directly the involvement of ERK and p38 we used the inhibitors PD98059 and SB203580. Each inhibitor caused only a limited inhibition of arachidonic acid release after stimulation with a variety of agonists. Furthermore, the effects of the inhibitors when combined were less than additive after stimulation with thrombin, histamine or A23187. Similar results were obtained with thrombin induced activation of cPLA2 (Fig. 5). These results, however, contrast with the mobility shift data shown in Fig. 6 which demonstrate marked inhibition of the mobility shift by each inhibitor and a complete inhibition after a combination treatment. In thrombin stimulated platelets, a similar discrepancy between the mobility shift and the activity of cPLA2 after treatment with SB203580 was noted by Kramer and coworkers [13]. In agreement with their activity data, they found by the use of P32 a residual 50% phosphorylation of cPLA2 after inhibitor treatment. They concluded that in addition to the p38-mediated phosphorylation at ser505, cPLA2 activation in thrombin stimulated platelets is mediated by phosphorylation at another site which

does not cause a mobility shift, and is due to the activity of yet another unidentified kinase. A recent phosphopeptide mapping of cPLA2 from thrombin stimulated human platelets revealed ser-505 and ser-727 as the phosphorylation sites [7]. Treatment with SB203580 which totally blocked cPLA2 mobility shift caused only 50% inhibition of phosphorylation on each of the sites, casting doubt on the reliability of the mobility shift assay for determination of the levels of cPLA2 phosphorylation. HUVEC has clearly, a far more complicated signaling pathway than the platelet. However, recent work by Muhalif et al. [24] demonstrated partial inhibition of arachidonic acid release in smooth muscle cells by MAP-kinase antisense nucleotides and the inhibitor of ERK activation, PD98059, while inhibition of CaM-kinase II completely abolished the release of arachidonic acid. In view of these findings, the most plausible interpretation of our cPLA2 activity and arachidonic acid release data in endothelial cells is that cPLA2 is phosphorylated by both ERK and p38 as well as a third kinase, and that this kinase might become more active in the presence of inhibitors of ERK and p38. This is supported by the observation depicted in Fig. 3(B) that the nonspecific protein kinase inhibitor staurosporin completely inhibited the cPLA2 response to thrombin in the fixed calcium concentration environment of the enzyme assay. Although we have found that JNK is phosphorylated in HUVEC after treatment with both thrombin and histamine, its contribution is ruled out since we have also found that PD98059 completely inhibits JNK activation under these circumstances (data not shown). The results obtained by the use of inhibitors are complicated by their limited specificity. As our results show, PD98059 as well as genistein completely inhibited agonist stimulated prostacyclin production (Table 2), while inhibition of arachidonic acid release was B50% under the same conditions (Tables 1 and 4). Furthermore, the same inhibitors also prevented conversion of exogeneously added arachidonic acid to prostacyclin (data not shown). Similar secondary effects of these inhibitors have been described in other cells [12,15,25] and recently both SB203580 and PD98059 were shown to be direct inhibitors of cyclooxygenase [26]. These effects are frequently not taken into account in the interpretation of results [27,28]. The nonspecific inhibitor staurosporine did not inhibit cellular arachidonic acid release after stimulation with thrombin or histamine despite inhibition of ERK activation and the in vitro activity and mobility shift of cPLA2 (Tables 1 and 3Fig. 4). This discrepancy may be explained by our previous observation that staurosporine greatly enhances thrombin mediated production of inositol phosphates in HUVEC [23]. An enhanced Ca++ signal intracellularly may therefore

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90

compensate for the lack of phosphorylation after inhibition with staurosporine. This would not be noted in the cPLA2 assay in which the Ca++ concentration is maintained at a constant level. Notable in our results is that thrombin and histamine signalling seem to follow different pathways, although both bind to 7-span membrane receptors that activate Phospholipase Cb (PLCb) through G-protein mediation. We have previously shown that thrombin but not histamine activates PLCg through tyrosine phosphorylation [23]. The greater inhibitory effect of the tyrosine kinase inhibitor genistein on thrombin mediated compared to histamine mediated arachidonic acid release could be accounted for by this difference (Table 1). In this study, arachidonic acid release following receptor stimulation was only modestly affected by the inhibitors of tyrosine kinases and PKC, whereas the responses to pervanadate and AlF− 4 were inhibited to a greater extent. Furthermore, receptor stimulation with thrombin was more affected by genistein or pertussis toxin treatment than was stimulation with histamine, whereas inhibition or down regulation of PKC had more inhibitory effect on the histamine stimulation of arachidonic acid release. Although modest, the effects of these manipulations suggest differences in signalling between thrombin and histamine somewhat analogous to the differences in pathways described by Hawes et al. between Gi and Gq mediated signaling in COS cells [29]. However, as subsequently pointed out by the same group of investigators, the pathways mediated by the a and the bg subunits of trimeric G-proteins may in some cell types show early convergence on PLCb and therefore can neither be distinguished by inhibitors of PKC nor inhibitors of tyrosine kinases [30]. Recently Sexl and coworkers, studying ERK activation in HUVEC after exposure to adenosine receptor agonists and isoproterenol also observed differences in the pathways employed [31]. Finally, we have previously shown that endothelial response to leukotriene C4 is unique in that pretreatment with pertussis toxin increases both inositol phosphate formation and arachidonic acid release [32]. Thus, although the details of the pathways leading from G-protein coupled receptors to MAP kinases and arachidonic acid release have not been fully elucidated, our results and those of others suggest that different pathways are operating simultaneously and are utilized to a different extent by the various agonists. In conclusion, our findings show that in endothelial cells arachidonic acid release is largely mediated by cPLA2 and that agonist specific pathways leading from G-protein coupled receptors through MAP kinases to activation of cPLA2 are involved. The MAP kinases ERK and p38 both have demonstrable but not major effect on agonist stimulated arachidonic acid release, and the data suggest that an additional unidentified kinase also has a role.

89

Acknowledgements This work was supported in part by the Research Fund of the University of Iceland, the Research Fund of the National University Hospital and the Icelandic Science Council.

References [1] Daniels I, Lindsay MA, Keany CIC, Burden RP, Fletcher J, Haynes AP. Role of arachidonic acid and its metabolites in the priming of NADPH oxidase in human polymorphonuclear leukocytes by peritoneal dialysis effluent. Clin Diagn Lab Immunol 1998;5:683 – 9. [2] Huang S, Konieczkowski M, Schelling JR, Sedor JR. Interleukin-1 stimulates Jun N-terminal/stress-activated protein kinase by an arachidonate-dependent mechanism in mesangial cells. Kidney Int 1999;55:1740 – 9. [3] Dennis EA. The growing phospholipase A2 superfamily of signal transduction enzymes. Trends Biochem Sci 1997;22:1 – 2. [4] Kramer RM, Sharp JD. Structure function. and. regulation. of. Ca2 + -sensitive. cytosolic. phospholipase A2 (cPLA2). FEBS Lett 1997;410:49 – 53. [5] Uozumi N, Kume K, Nagase, et al. Role of cytosolic phospholipase A2 in allergic response and parturition. Nature 1997;390:617 – 22. [6] Bonventre JV, Huang Z, Taheri MR, O’Leary E, Li E, Moskowitz MA, Sapirstein A. Reduced fertility and post-ischaemic brain injury in mice deficient in cytosolic phospholipase A2. Nature 1997;390:622 – 5. [7] Bo¨ rsch-Haubold AG, Bartoli F, Asselin J, et al. Identification of the phosphorylation sites of cytosolic phospholipase A2 in agonist-stimulated human platelets and HeLa cells. J Biol Chem 1998;273:4449 – 58. [8] Lin LL, Wartmann M, Lin AY, Knopf JL, Seth A, Davis RJ. cPLA2 is phosphorylated and activated by MAP kinase. Cell 1993;72:269 – 78. [9] Kramer RM, Roberts EF, Hyslop PA, Utterbach BG, Hui KY, Jakubowsky JA. Differential activation of cytosolic phospholipase A2 (cPLA2) by thrombin and thrombin receptor agonist peptide in human platelets. J Biol Chem 1995;270:14816 –23. [10] Bo¨ rsch-Haubold AG, Kramer RM, Watson SP. Cytosolic phospholipase A2 is phosphorylated in collagen and thrombin stimulated human platelets independent of protein kinase C and mitogen-activated protein kinase. J Biol Chem 1995;270:25885 – 92. [11] Qiu Z, Leslie CC. Protein kinase C-dependent and -independent pathways of mitogen-activated protein kinase activation in macrophages by stimuli that activate phospholipase A2. J Biol Chem 1994;269:19480 – 7. [12] Bo¨ rsch-Haubold AG, Kramer RM, Watson SP. Phosphorylation and activation of cytosolic phospholipase A2 by 38-kDa mitogen-activated protein kinase in collagen-stimulated human platelets. Eur J Biochem 1997;245:751 – 9. [13] Kramer RM, Roberts EF, Ulm SL, Bo¨ rsch-Haubold AG, Watson SP, Fisher MJ, Jakubowski JA. p38 mitogen-activated protein kinase phosphorylates cytosolic phospholipase A2 (cPLA2) in thrombin-stimulated platelets. J Biol Chem 1996;271:27723 – 9. [14] Bo¨ rsch-Haubold AG, Kramer RM, Watson SP. Inhibition of mitogen-activated protein kinase kinase does not impair primary activation of human platelets. Biochem J 1996;318:207 –12. [15] McNicol A, Philpott CL, Shibou TS, Israels SJ. Effects of the mitogen-activated protein (MAP) kinase inhibitor 2-(2%-amino-2%-

90

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

I.J. Gudmundsdo´ ttir et al. / Atherosclerosis 156 (2001) 81–90 methoxyphenyl)-oxanaphtalen-4-one (PD98059) on human platelet activation. Biochem Pharm 1998;55:1759 – 67. Sa G, Murugesan G, Jaye M, Ivashchenko Y, Fox PL. Activation of cytosolic phospholipase A2 by basic fibroblast growth factor via a p42 mitogen-activated protein kinase-dependent phosphorylation pathway in endothelial cells. J Biol Chem 1995;270:2360 – 6. Wong JT, Tran K, Pierce GN, Chan AC, OK, Choy PC. Lysophosphatidylcholine stimulated the release of arachidonic acid in human endothelial cells, J Biol Chem 1998; 273:6830 – 6836. Rousseau S, Houle F, Landry J, Huot J. p38 MAP kinase activation by vascular endothelial growth factor mediates actin reorganization and cell migration in human endothelial cells. Oncogene 1997;15:2169 –77. Cohen P. The search for physiological substrates of MAP and SAP kinases in mammalian cells. Trends Cell Biol 1997;7:353 – 61. Pumiglia KM, Lau LF, Huang CK, Burroughs S, Feinstein MB. Activation of signal transduction in platelets by the tyrosine phosphatase inhibitor pervanadate (vanadyl hydroperoxide). Biochem J 1992;286:441 –9. Halldo´ rsson H, Kjeld M, Thorgeirsson G. Role of phosphoinositides in the regulation of endothelial prostacyclin production. Arteriosclerosis 1988;8:147 –54. Lin L-L, Lin AY, Knopf JL. Cytosolic phospholipase A2 is coupled to hormonally regulated release of arachidonic acid. Proc Natl Acad Sci 1992;89:6147 –251. Helgado´ ttir A, Halldo´ rsson H, Magnu´ sdo´ ttir K, Kjeld M, Thorgeirsson G. A role for tyrosine phosphorylation in generation of inositol phosphates and prostacyclin production in endothelial cells. Arterioscler Thromb Vasc Biol 1997;17:287 – 94. Muthalif MM, Benters IF, Uddin MR, Malik KU. Calcium/ calmodulin dependent protein kinase IIa mediates activation of mitogen activated protein kinase and cytosolic phospholipase A2

.

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

in norepinephrine induced arachidonic acid release in rabbit aortic smooth muscle cells. J Biol Chem 1996;271:30149 –57. Ozaki Y, Yatomi Y, Jinnai Y, Kume S. Effects of genistein, a tyrosine kinase inhibitor, on platelet functions. Biochem Pharm 1993;46:395 – 403. Bo¨ rsch-Haubold AG, Pasquet S, Watson P. Direct inhibition of cyclooxygenase-1 and -2 by the kinase inhibitors SB203580 and PD98059. J Biol Chem 1998;273:28766 – 72. Wheeler-Jones CPD, May MJ, Houliston RA, Pearson JD. Inhibition of MAP kinase (MEK) blocks endothelial PGI2 release but has no effect on von Willebrand factor secretion or E-selectin expression. FEBS Lett 1996;388:180 – 4. Patel V, Brown C, Goodwin A, Wilkie N, Boarder MR. Phosphorylation and activation of p42 and p44 mitogen-activated protein kinase are required for the P2 purinoceptor stimulation of endothelial prostacyclin production. Biochem J 1996;320:221 – 6. Hawes BE, van Biesen T, Koch WJ, Luttrell LM, Lefkowitz RJ. Distinct pathways of Gi- and Gq-mediated mitogen-activated protein kinase activation. J Biol Chem 1995;270:17148 –53. Della Rocca GJ, van Biesen T, Daaka Y, Lutrell DK, Lutrell LM, Lefkowitz RJ. Ras-dependent mitogen activated protein kinase activation by G protein-coupled receptors. Convergence of Gi- and Gq-mediated pathways on calcium/calmodulin, Pyk2, and S kinase. J Biol Chem 1997;272:19125 – 32. Sexl V, Mancusi G, Ho¨ ller C, Gloria-Maercker E, Schuetz W, Freissmuth M. Stimulation of the mitogen-activated protein kinase via the A2A-adenosine receptor in primary human endothelial cells. J Biol Chem 1997;272:5792 – 9. Geirsson A, Halldo´ rsson H, Magnu´ sdo´ ttir K, Kjeld M, Thorgeirsson G. Potentiating effects of pertussis toxin on leukotriene C4 induced formation of inositol phosphate and prostacyclin in human umbilical vein endothelial cells. J Cell Physiol 1998;177:103 – 8.