Stimulation of Platelet-activating Factor Synthesis in Human Endothelial Cells by Activation of the de Novo Pathway PHORBOL 12-MYRISTATE 13-ACETATE ACTIVATES 1-ALKYL-2-LYSO-sn-GLYCERO-3-PH 0 SPHATE : ACETYL-COA ACETYLTRANSFERASE AND DITHIOTHREITOL-INSENSITIVE 1-ALKYL-2-ACETYL-sn-GLYCER 0 L : CDP-CHOLINE CHOLINEPHOSPHOTRANSFERASE

Human umbilical vein endothelial cells (HUVEC) produce platelet-activating factor (PAF) by a remodeling pathway involving a phospholipase Az followed by an acetyl-CoA-dependent acetyltransferase which acetylates a lyso-PAF intermediate to form PAF and is stimulated by a variety of agents that generate inflammatory and allergic responses. A second route for PAF synthesis in mammalian tissues is a de novo pathway, which requires the participation of three enzymes: 1alkyl-2-lyso-sn-glycero-3-phosphate (alkyllyso-GP): acetyl-coA acetyltransferase, 1-alkyl-2-acetyl-snglycero-3-phosphate phosphohydrolase, and dithiothreitol (DTT)-insensitive 1-alkyl-2-acetyl-sn-glycerol (alkylacety1-G):CDP-cholinecholinephosphotransferase. In the present study we show that protein kinase C activation by phorbol 12-myristate 13-acetate (PMA) induces PAF production in HUVEC by an increase of both alkyllyso-GP:acetyl-CoA acetyltransferase and DTT-insensitive alkylacetyl-G:CDP-choline cholinephosphotransferase. PAF synthesis, labeled precursors (r3H]acetate  and [rnethyl-3H]choline) incorporation, and both enzyme activities of the de novo pathway increase concomitantly in response to different doses of PMA. PMA does not activate the enzymes of the remodeling pathway. We conclude that both remodeling and the de novo pathway for PAF synthesis are

Incorporation of Radioactive Precursors into PAF-HUVEC (100mm diameter culture wells, 2.1 k 0.7 X lo6 cells, n = 22) were labeled at 37 "C with 80 pCi of [3H]acetate or with 20 pCi of [ m e t h~l -~H ] choline in 4 ml of M199-BSA for 30 or 60 min, respectively, and then stimulated with PMA. The incubation was stopped as detailed before, and the radioactivity incorporated into PAF fraction counted after separation by TLC. As for the PAF (see before), both radiolabeled lipids that co-migrated with a PAF standard on TLC were shown to be authentic PAF by the following evidence: 1) an identical elution volume on high performance liquid chromatography; 2) degradation of 98% of the radiolabeled products by phospholipase A, from Crotalus adamantus, confirming that acetate had been incorporated at the sn-2 position (21); 3) degradation of 15% of the radiolabeled lipids by treatment with lipase A1 from Rhizopus arrhizus, confirming that 85% of the products had an ether linkage at the sn-1 position (21); 4) typical bioactivity causing aggregation of washed rabbit platelets (6), using a calibration curve with synthetic PAF for each series of assays.
The complete characterization of PAF has been performed in three individual experiments out of four. The specificity of platelet aggregation was inferred from the inhibitory effect of 5 p M CV-3988 and 5 p~ BN 52021 (l), two well known PAF antagonists. The biological activity was sensitive to previous treatment with phospholipase A, and resistant to the treatment by lipase AI.
["CIAA Labeling and Agonist-stimulated Release of I 4 C Radioactiuity-HUVEC monolayers (100-mm diameter culture wells) were labeled for 24 h with 1.5 pCi of [14C]AA in 7.5 ml of growth medium and then washed twice with M199-BSA and three times with M199 to remove unincorporated radioactivity (22). Total incorporation was 1,830,000 k 240,000 cpm/dish. Experimental dishes were incubated with PMA or thrombin in 6 ml of M199-BSA. During 20-min incubation, aliquots of the medium were taken to determine the percentage of total cellular radioactivity released.
The assay conditions to measure lyso-PAF:acetyl-CoA acetyltransferase have been described in detail (11, 18). Alkyllyso-GP:acetyl-CoA acetyltransferase was measured according to Lee et al. (8). 150 pg of lysate protein were incubated for 5 min at 23 "C in 0.5 ml of 0. of alkylacetyl-G = 0.94) to identify alkylacetyl-GP, and chloroform/ methanol/acetic acid, 964:1, v/v (RF of alkylacetyl-G = 0.43) to quantify alkylacetyl-G which can be rapidly synthesized from the former product due to the high alkylacetyl-GP phosphohydrolase activity (9). Alkyllyso-GP:acetyl-CoA acetyltransferase was expressed as nanomoles of [3H]acetate transferred to alkyllyso-GP including the radioactivity found in the hydrolysis product of alkylacetyl-GP. Alkylacetyl-G represented no more than 15% of the total products.
DTT-insensitive alkylacetyl-G:CDP-choline cholinephosphotransferase was measured according to the procedure described by Renooij and Snyder (23). 100 pg of lysate protein were incubated for 10 min at 37 "c in 0.5 ml of 0.1 M Tris-HC1, pH 8.0, containing 0.5 mM EGTA, 0.5 mg of BSA, 5 mM DTT, 10 mM MgCl,, 200 PM alkylacetyl-G , 0.5 pCi of CDP-[methyl-'*C]choline, 100 p~ unlabeled CDPcholine. Enzymatic activity was linear as a function of lysate protein (up to 200 pg) and incubation time (up to 30 min). Reaction was stopped by lipid extraction followed by TLC purification and radioactivity counting of lipids co-migrating with PAF as specified above.

RESULTS
Biosynthesis of PAF in Response to PMA in HUVEC-PMA induced the synthesis of PAF in a clear-cut dosedependent manner (Fig. 1, panel A). Maximal stimulation of PAF synthesis was observed at 250 nM PMA, and no further increase was observed at 1 PM PMA both after 2 and 20 min of incubation. The synthesis started shortly after the stimulus (2 min) and was maintained nearly constant over a 20-min period . H7 (~-(5-isoquinolinylsulfonyl)-2-methylpiperazine), a known inhibitor of protein kinase C (24), elicited a dosedependent block of PMA-induced PAF synthesis (not shown). PMA-induced PAF synthesis was also evaluated by measuring the incorporation of [3H]acetate and of [methyL3H]choline into the fraction co-migrating with synthetic PAF. Incubation of HUVEC in the presence of PMA enhanced the incorporation of both labels into the fraction co-migrating with synthetic PAF (Fig. 1, panels B and c). The dose-response pattern of this incorporation showed that maximal response was reached at 250 nM PMA, both after 2 and 20 min of incubation. A time dependence of labels incorporation was also observed at each PMA concentration (Fig. 1, panels B and C). Less than 15% (range, 6-15%) of [3H]acetate-labeled lipid product was degraded by treatment with phospholipase A1 suggesting that the major acetylated compound produced by HUVEC activated by PMA and co-migrating in a TLC system with PAP is the 1-0-alkyl form rather than the 1-acyl form.
Agonist-stimulated Release of 14C Radioactivity in HU-VEC- Table I shows the effect of 250 nM PMA on the release of I4C radioactivity from [14C]AA-labeled cells. No release was observed over a 20-min incubation period, indicating that phospholipase A2 did not seem to be influenced by PMA in that experimental condition. Thrombin (0.5 unit/ml), used as a positive control, elicited a significant release of 14C radioactivity (Table I).
Enzyme Activities of the de Novo and Remodeling Pathways in PMA-stimulated HUVEC-The activities of both alkyllyso-GP:acetyl-CoA acetyltransferase and DTT-insensitive alkylacetyl-G:CDP-choline cholinephosphotransferase resulted in an increase in HUVEC treated with PMA. The increase of both enzyme activities was evident as early as 2 min after PMA addition, was maintained over a 20-min period, and reached the maximum at 250 nM PMA concentration, both after 2 and 20 min of incubation (Fig. 2, panels A and B). So far, PAP synthesis, labeled precursor incorporation, and both enzyme activities of the de novo pathway increased concomitantly in response to different doses of PMA (Figs. 1

and 2).
No differences in lyso-PAFacetyl-CoA acetyltransferase activity were observed between control and PMA-stimulated cells at all of the times and concentrations tested.

DISCUSSION
We report here that PMA stimulates the de mu0 biosynthesis of PAF in HUVEC through the activation of alkyllyso-GP:acetyl-CoA acetyltransferase and DTT-insensitive alkyl-      Figs. 1 and 2). PMA promotes the incorporation of labeled acetate into ~-O-alkyl-2-lyso-glycero-3-phosphocholine forming l-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine, which was first identified and termed PAF (revised in Refs. 1 and  2). PMA also induces the synthesis of a small amount of 1acyl PAF form, as shown in other cells (25, 26) including HUVEC (27), differently stimulated. Recently Garcia et al. (28) reported a high degree of synthesis of 1-acyl-2-acetyl-snglycerophosphocholine in HUVEC. This difference could be due to the different experimental conditions or to the different metabolic pathways involved. To rule out that [3H]acetate incorporation into PAF could be an index of the activation of the remodeling pathway by PMA rather than of the recruitment of the acetate moiety into the de m u 0 pathway, we measured the enzymatic activity of lyso-PAF:acetyl-CoA acetyltransferase and the release of I4C radioactivity from ["C] AA-labeled HUVEC. PMA does not modify the basal activity of this enzyme nor does it promote the release of I4C radioactivity. Indeed, the phospholipase Az-mediated release of AA (22) and the Ca2+-induced activation of lyso-PAF:acetyl-CoA acetyltransferase (29), which are elicited by inflammatory agents such as thrombin in HUVEC (29), are absent in PMAstimulated cells. These data agree with previous ones showing the lack of PMA effect on the remodeling pathway in rat alveolar macrophages (4) and in human vascular endothelial cells (30).
As to DTT-insensitive alkylacetyl-G:CDP-choline cholinephosphotransferase, our findings are in keeping with previous studies on neutrophils, which showed 1-oleyl-2-acetylglycerol-dependent cholinephosphotransferase activation and PAF synthesis (17), and the requirement of well determined concentrations of alkylacety1-G for optimal expression of the PMA-elicited PAF synthesis (16). In addition, we report the novel finding that PMA exerts a regulatory role on the activity of alkyllyso-GP:acetyl-CoA acetyltransferase. The detailed biochemical characterization of both alkyllyso-GP:acetyl-CoA acetyltransferase and DTT-insensitive alkylacetyl-G:CDPcholine cholinephosphotransferase has been reported (8, 10). In most tissues, the specific activity of alkyllyso-GP:acetyl-CoA acetyltranferase is the lowest among the three enzymes of the de mu0 pathway (8-10); therefore, it appears that this enzyme would be the rate-limiting step in the production of PAF. However, it has been shown that only the decrease of the DTT-insensitive cholinephosphotransferase (and not of the acetyltransferase) accounts for the decreased rate of PAF synthesis during the development of renal necrosis (31). This observation, together with the finding that PAF generated by chick retinas upon stimulation with neurotransmitters is due to an increase in the DTT-insensitive alkylacetyl-G:CDPcholine cholinephosphotransferase (11, 12), points to an important regulatory role of this enzyme in the de nouo synthesis of PAF. Our results suggest that, in HUVEC, both alkyllyso-GP:acetyl-CoA acetyltransferase and DTT-insensitive alkyl-acetyl-G:CDP-choline cholinephosphotransferase (whose specific activities are very similar) are regulatory enzymes and that changes in their activity may be due to PKC-induced enzyme phosphorylation mediated by PMA. While inflammatory stimuli ( i e . thrombin)-induced increase of intracellular Caz+ results in activation of the remodeling pathway (25), it most probably inhibits the synthesis of PAF through the de mu0 pathway, since Ca2+ is an inhibitor of both alkyllyso-GP:acetyl-CoA acetyltransferase (8) and DTT-insensitive alkylacetyl-G:CDP-choline cholinephosphotransferase (10).
PKC-induced activation of PAF synthesis through the de m u 0 pathway may take place when HUVEC are stimulated by agents which can activate PKC in the absence of any Ca2+ transient, i.e. cis-fatty acids (32), elevated glucose concentration (33), and selective oxidative modification of the enzyme regulatory domain (34).