Inhibition of Thromboxane A Synthesis in U937 Cells by Glucocorticoids LACK OF EVIDENCE FOR LIPOCORTIN 1 AS THE SECOND MESSENGER*

The mechanism of inhibition of eicosanoid synthesis by glucocorticoids has been investigated using differ- entiated U937 cells as a model. These cells synthesize thromboxane Aa (TXAz) in response to a variety of agonists, and synthesis of TXAz initiated by certain stimuli was inhibited by pretreatment of the cells with glucocorticoids. The inhibitory response was specific for glucocorticoid steroids and required receptor oc- cupancy based on both the rate of onset of the inhibitory activity and the correlation between potency and receptor affinity of various analogs. The inhibitory response was also specific for the agonist used to ini-tiate TXAz synthesis. Both lipopolysaccharide- and zymosan-induced TXAz synthesis were inhibited by increasing concentrations of dexamethasone (>80%, ICao 10 nM), while synthesis initiated by addition of either exogenous arachidonic acid or the Ca2+ ionophore A23187 was unaffected over the same concentration range. The latter result indicates that the dexamethasone block is upstream from release of esterified arachidonic acid. Attempts to localize the block more accurately showed that although dexamethasone was not acting as a generalized inhibitor of transcrip- tion or translation, its ability to inhibit TXAz synthesis was mimiced by the activity of actinomycin D and cycloheximide. The role of the purported inhibitor radioactivity in the layer was quantified by liquid scintillation counting, and the TXB2 concentration of the medium was measured

The mechanism of inhibition of eicosanoid synthesis by glucocorticoids has been investigated using differentiated U937 cells as a model. These cells synthesize thromboxane Aa (TXAz) in response to a variety of agonists, and synthesis of TXAz initiated by certain stimuli was inhibited by pretreatment of the cells with glucocorticoids. The inhibitory response was specific for glucocorticoid steroids and required receptor occupancy based on both the rate of onset of the inhibitory activity and the correlation between potency and receptor affinity of various analogs. The inhibitory response was also specific for the agonist used to initiate TXAz synthesis. Both lipopolysaccharide-and zymosan-induced TXAz synthesis were inhibited by increasing concentrations of dexamethasone (>80%, ICao 10 nM), while synthesis initiated by addition of either exogenous arachidonic acid or the Ca2+ ionophore A23187 was unaffected over the same concentration range. The latter result indicates that the dexamethasone block is upstream from release of esterified arachidonic acid. Attempts to localize the block more accurately showed that although dexamethasone was not acting as a generalized inhibitor of transcription or translation, its ability to inhibit TXAz synthesis was mimiced by the activity of actinomycin D and cycloheximide.
The role of the purported phospholipase inhibitor protein lipocortin 1 in mediating the dexamethasone inhibition of TXAz synthesis was studied by examining the effect of dexamethasone on lipocortin 1 metabolism. Under conditions which gave maximal inhibition of lipopolysaccharide-or zymosan-stimulated TXAZ synthesis, dexamethasone had no effect on the steady state level of lipocortin 1 mRNA or protein, indicating that lipocortin 1 induction by dexamethasone is not responsible for the observed inhibition. Furthermore, lipocortin 1 was not secreted from the cells under any conditions examined, and the intracellular form had a relatively long half-life (>21 h). The lack of induction of lipocortin 1 by dexamethasone and the fact that it is not released from the cells are both inconsistent with the properties previously described for lipocortin-like activities and indicate that lipocortin 1 is not a glucocorticoid second messenger in this experimental model. Although the data are consistent with a mechanism involving inhibition of a factor that activates TXAZ synthesis, we cannot rule out a mechanism involving * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
4 To whom correspondence should be addressed. glucocorticoid induction of a phospholipase inhibitor protein distinct from lipocortin 1.
Glucocorticoids inhibit prostaglandin synthesis in a variety of cell types but do not inhibit prostaglandin H synthase activity (1,2). Studies on the mechanism of glucocorticoids suggest that they act by inducing the synthesis of a protein that inhibits phospholipase A2 and hence blocks mobilization of esterified arachidonic acid (3). Dexamethasone-inducible phospholipase AS inhibitor proteins have been partially characterized from perfused lung and conditioned medium from macrophages (macrocortin), neutrophils (lipomodulin), renomedulary interstitial cells (renocortin), and thymus (4-8).
Macrocortin, lipomodulin, and renocortin have similar physical and immunological properties (9,10) and have been renamed lipocortin (11). Characterization of the biological properties of these inhibitor proteins has given rise to the following operational definition for lipocortin-like activity; a protein that (a) is synthesized and secreted in response to glucocorticoids, ( b ) inhibits pancreatic phospholipase A2 in uitro, (c) inhibits eicosanoid synthesis when added to cells, ( d ) exhibits antiinflammatory activity in standard models of acute inflammation, and (e) is regulated by phosphorylation (3,9,12). Using inhibition of phospholipase A2 in uitro as an assay, lipocortin has been purified from cell-free supernatants conditioned by dexamethasone-stimulated macrophages and its structure defined using a tandem protein sequencing/ molecular cloning approach (13,14). Examination of the tissue distribution of lipocortin led to the isolation and cloning of a homologous phospholipase A2 inhibitor, lipocortin 2 (15). Lipocortins 1 and 2, as defined by molecular cloning, were isolated solely on the basis of their ability to inhibit phospholipase A2-catalyzed hydrolysis of labeled Escherichia coli membranes. Investigations into the mechanism of this inhibition have shown that the inhibitory activity is due to association of the inhibitor with the phospholipid substrate and not due to direct binding to phospholipase A2 (16,17). Furthermore, this inhibitory property is shared with a number of other related proteins which exhibit Ca2+-dependent binding to phospholipid vesicles (18). Recombinant lipocortin 1 has been reported to inhibit leukotriene Cr-induced TXA; synthesis by perfused lung (19), and the lipocortin 1 gene has been reported to be dexamethasone-inducible in rat peritoneal macrophages (14).
Aside from these reports, little is known concerning the relationship between lipocortin 1 (as defined by molecular cloning) and the active principle described by the operational definition described above. In this report, we elected to study the regulation of TXA2 synthesis in differentiated U937 cells by glucocorticoids. Lipopolysaccharide (LPS)or zymosanstimulated TXA2 synthesis is inhibited by glucocorticoids by a process that requires receptor occupancy. The dexamethasone-induced block has been mapped upstream of PGH synthase and appears to be upstream from arachidonic acid mobilization (phospholipase). Attempts to correlate the dexamethasone inhibitory response with induction of lipocortin 1 (mRNA or protein) or lipocortin 1 secretion, were unsuccessful. The evidence accumulated to date suggest that dexamethasone does not act by inducing lipocortin 1 but rather by suppressing the expression of a LPS/zymosan-inducible gene(s) that in turn activates the cells to synthesize TXA2. Alternatively, a mechanism involving glucocorticoid induction of an inhibitor protein cannot be ruled out based on the present data.

Materials
U937 cells were obtained from the American Type Culture Collection and cultured in RPMI 1640 medium supplemented with FungiBact and 10% FCS (Irvine Scientific) in 5% COZ at 37 "C.

Methods
Differentiation of U937 Cells"u937 cells were maintained in suspension culture in RPMI 1640 supplemented with 1 X FungiBact, 10% heat-inactivated fetal calf serum. For differentiation, cells were plated at 2.0 X 10' cells/ml in medium containing 100 nM PMA and allowed to attach for 48 h. The cells were then fed with PMA-free medium and cultured overnight prior to use.
Effects of Steroid Treatments on Thromboxane A Synthesis-TXAz synthesis by PMA-differentiated U937 cells was initiated by washing the monolayers twice with serum-free RPMI 1640 followed by addition of fresh medium containing 1% FCS and either E. coli LPS (10 rg/ml), heat-activated zymosan A (500 pg/ml), A23187 (10 pglml), or arachidonic acid (10 pM). After 4 h (LPS or zymosan) or 15 min (A23187 or arachidonic acid), the medium was removed and thromboxane BP (TXBZ) levels determined using a TXBz-specific enzyme immunoassay (EIA) as described (20). Steroid treatments were performed by adding the appropriate steroid to RPMI 1640 medium containing 1% FCS from an ethanol stock (0.1% ethanol final concentration) and pretreating the cells for varying lengths of time. The medium was removed, replaced with steroid-free medium containing the appropriate agonist, and TXAZ synthesis quantified as described above.
Glucocorticoid Receptor Determination-Glucocorticoid receptors were quantified using a whole cell binding assay. PMA-differentiated U937 cells (1.5 X IO6 cells/well) were incubated with RPMI 1640 medium containing 1% FCS and various amounts of I3H]dexamethasone or [3H]dexamethasone plus a 500-fold molar excess of unlabeled steroid for 3 h at 21 "C. The medium was removed, the cell layer washed three times with 5 ml of phosphate-buffered saline (Caz+/ Me-free), and the cells collected on Fiberglas filters. Bound [3H] dexamethasone was quantified by liquid scintillation counting.
Effect of Transcription and Translation Inhibitors on TXAZ Synthesis-Dose-response curves for actinomycin D and cycloheximide/ emitine were determined to establish the optimum dose for inhibition of transcription and translation, respectively. PMA-differentiated U937 cells were pretreated with various concentrations of drug in RPMI 1640, 1% FCS for 30 min, and the rate of transcription (for actinomycin D) or translation (cycloheximide/emitine) was measured. For transcription, the cells were labeled with I3H]uridine (10 pCi/ml) in the same medium for 1 h while translation was measured by labeling with [3H]leucine (10 pCi/ml in leucine-free medium) for 1 h. In each case, the labelings were performed in the continued presence of drug and the rates estimated based on incorporation of 3H into total cellular trichloroacetic acid-insoluble material. The effect of these inhibitors on TXAz synthesis was performed by pretreating the cells with the appropriate dose of drug in RPMI 1640, 1% FCS for 30 min followed by stimulation with the desired agonist in the continued presence of drug. TXBZ levels were then determined by EIA as above.
Lipocortin 1 Metabolism-Lipocortin 1 metabolism was studied by metabolic labeling of PMA-differentiated U937 cells with [%]methionine followed by immune precipitation with a lipocortin 1-specific polyclonal antiserum (21). Cells were labeled using [36S]methionine (100-500 pCi/ml) in methionine-free RPMI 1640 medium containing 10% FCS for various times and either harvested or chased in medium containing excess unlabeled methionine. For harvest, the culture medium was removed, dialyzed at 4 "C against 25 mM NH4HC03 (pH 7.8), and lyophilized. The washed cell layers were lysed in 2.0 ml of 25 mM Tris-HC1 (pH 7.41, 0.1% Nonidet P-40, 5 mM MgClz, 1 mM EGTA, 0.25 mM phenylmethylsulfonyl fluoride, 10 pg/ml aprotinin at 4 'C for 20 min, centrifuged (1000 x g, 10 rnin), and the supernatant used for immune precipitation. An aliquot of the cell lysate or culture medium fraction (0.3 ml) was incubated with 10 pl of lipocortin 1specific rabbit antihuman lipocortin 1 antiserum for 30 min on ice, and the immune complexes were then precipitated using activated Pansorbin (Calbiochem). The washed pellets were dissociated in SDS-PAGE loading buffer, run on 15% Lamelli gels and 36S-labeled protein bands were detected by fluorography with salacylic acid.
Lipocortin 1 was also detected by Western blotting. Cell lysates, prepared as described above, or pure lipocortin 1 (purified from human placenta as described (21)) were run on nonreducing 15% SDS-PAGE, electroblotted to nitrocellulose (0.2 A, 5 h), and probed with a 1/400 dilution of lipocortin 1-specific antiserum. Immune complexes were detected using 9-protein A followed by autoradiography.
Isolation of Human Lipocortin 1 cDNA-Two lipocortin 1-specific oligonucleotide probes, complementary to nucleotides 254-278 (CTCCAGGAAACAGGAAAGCCCCTGG) or 910-934 (TGATAT-CATTCATGTCAATTTCAGA) of the human lipocortin 1 cDNA sequence were synthesized using an Applied Biosystems Instruments 380B synthesizer. The oligonucleotides were purified by preparative denaturing PAGE, desalted on C18 Sep Pak cartridges, and labeled using T4 polynucleotide kinase and [-p3'P]ATP. U937 cells were treated with dexamethasone (1.0 p~) and PMA (0.1 p~) for 4 h and total cellular RNA prepared by LiCl precipitation as described (22). Poly(A+) RNA was prepared by two rounds of chromatography on oligo(dT) cellulose and used to synthesize cDNA by oligo(dT)-primed first strand synthesis using avian myeloblastosis virus reverse transcriptase followed by RNase H/avian myeloblastosis virus reverse transcriptase-mediated second strand synthesis (23). The cDNA was tailored sequential treatment with T4 polymerase and EcoRI methylase followed by addition of EcoRI linkers and size selection on Bio-Gel A-50m (24). A portion of the size-selected cDNA was ligated into dephosphorylated EcoRI-digested X g t l O arms, packaged in uitro, and plated on E. coli C600 hfl-. Lipocortin 1 cDNA clones were isolated by hybridization with the lipocortin 1-specific oligonucleotide probes. A 1.3-kb EcoRI fragment, corresponding to nucleotides 16-1326 of the human lipocortin 1 cDNA sequence was subcloned into pUC19 and completely sequenced by the dideoxy chain termination method (25). The nucleotide sequence was identical to human lipocortin 1 (14).
Quantification of Lipocortin 1 mRNA-Analytical RNA isolations (10' cells/sample) were performed by cell lysis in 4 M guanidine isothiocyanate, 8% 8-mercaptoethanol exactly as described (26). Aliquots of total RNA (10 pg) were denatured by treatment with glyoxal, fractionated on a 1% agarose gel with buffer recirculation, electroblotted to Nytran, and immobilized on the filter by UV irradiation.
Lipocortin 1 cDNA (1.3-kb EcoRI fragment) was labeled using Klenow fragment and random primers (27) and hybridized (lo7 dpm/ml) to the blot as described by Church and Gilbert (28). Labeled bands were detected by autoradiography and quantified by densitometry.

RESULTS
Characterization of Cellular Model-Differentiated U937 cells were chosen as a cellular model to study the effect of glucocorticoids on arachidonic acid mobilization. Since the validity of this model depends on the presence of glucocorticoid receptors in these cells, receptors were quantified by [3H] dexamethasone binding in a whole cell receptor binding assay. U937 cells or PMA-differentiated U937 cells (dU937) were incubated with various amounts of [3H]dexamethasone in the absence or presence of a 500-fold molar excess of unlabeled dexamethasone for 3 h at room temperature and the cellassociated steroid quantified by liquid scintillation counting. The binding curve shown in Fig. 1 indicates a specific binding component in dU937 cells that saturates at about 5 nM. Undifferentiated U937 cells had approximately %fold less specific binding capacity (data not shown). Scatchard analysis of the binding data obtained from dU937 cells reveals a single class of glucocorticoid receptors with a Kd of 1.9 nM and a receptor density of 16,600 receptors/cell. Because dU937 cells have a higher receptor density, they were used in all subsequent experiments.
Preliminary metabolism experiments using dU937 cells prelabeled with [3H]arachidonic acid followed by stimulation with a variety of agonists showed that thromboxane A ( T U * ) was the major eicosanoid synthesized, while smaller amounts of prostaglandin E were also detected ( 4 0 % of the TXAZ level). Undifferentiated U937 cells produced only 2% of the TXAz synthesized by dU937 cells, again requiring the use of dU937 cells for all subsequent experiments. The kinetics of agonist-stimulated TXAZ synthesis by dU937 cells are shown in Fig. 2. Cells were treated with either lipopolysaccharide (10 pg/ml), zymosan (500 pg/ml), A23187 (1 pg/ml), or exogenous arachidonic acid (10 p~) for various times and TXAz synthesis was quantified by its hydrolysis product TXBZ by enzyme immunoassay. TXAz synthesis induced by LPS or zymosan (top panel) occurred following a lag of 2 h and continued at a near linear rate for 7 h. The rates of TXAz synthesis were more rapid when either A23187 or arachidonic acid were used (bottom panel), exhibiting linear release for 5 min followed by a slow release lasting for 1 h. Based on these results, a 4-h stimulation (LPS and zymosan) or a 15-min stimulation (A 23187 and arachidonic acid) was used in all subsequent experiments.
Effect of Glucocorticoids on TXAZ Synthesis-Since LPS was the strongest agonist for initiating TXAz synthesis, the effect of representative steroids on the LPS response was determined, and the results are shown in Fig. 3. dU937 cells were pretreated with either dexamethasone, aldosterone, testosterone, or 17/3-estradiol at concentrations ranging from to lo-' M for 2 h, stimulated with LPS in the absence of steroid, and TXAz release quantified by EIA. Pretreatment with the glucocorticoid dexamethasone caused a dose-dependent inhibition of TXAz synthesis with an ICW of 10 nM and a maximal inhibition of 82% (top p a n e l ) . Aldosterone, testosterone, or 17fi-estradiol were without effect over the same dose range. To further test the specificity of the glucocorticoid effect, we examined a series of glucocorticoid analogs for their  inhibitory potency, and the results are shown in Fig. 3 (bottom panel). This experiment was performed as described above using fluocinolone acetonide, dexamethasone, methylprednisolone, and hydrocortisone in the pretreatments. Fluocinolone acetonide was the most potent analog with an ICs0 of 0.8 nM, while the other analogs were less active. The fact that the dose-response relationships parallel the known receptor affinities for these analogs indicates that the inhibitory effect is specific for glucocorticoids and requires receptor occupancy. Furthermore, the data shown in Fig. 3 show that the glucocorticoid effect is irreversible for at least 4 h after steroid removal. Whether this is due to the steroid remaining associated with the cells or the synthesis of a factor that is mediating the inhibitory response is unknown.
Mechanism of the Glucocorticoid Inhibition-The mechanism of the glucocorticoid inhibition of TXAz synthesis was further investigated by initiating TXAz synthesis using a series of agonists that activate different portions of the TXA, biosynthetic pathway. dU937 cells were pretreated with dexamethasone at various concentrations for 2.0 h and then stimulated with either LPS or zymosan (4 h), or A23187 or exogenous arachidonic acid (15 min) in the absence of steroid. As shown in Fig. 4, when stimulated with either LPS or zymosan, the cells produced nearly equivalent amounts of TXA2, and the dexamethasone dose-response curves for inhibition of TXA2 synthesis were parallel. When dexametha- Dexamethasone inhibition of TXAt synthesis initiated by various agonists. Triplicate wells of U937 cells were differentiated and cultured as described in Fig. 1. The cells were then pretreated with various concentrations of dexamethasone for 2 h as described in Fig. 3. They were then stimulated with either 10 pg/ml LPS (0) or 500 pg/ml zymosan (A) for 4 h or with 10 pM arachidonic acid (0) or 1 pg/ml A23187 (A) for 15 min in steroid-free medium, and the TXB2 content of the medium was determined by EIA. Results are expressed as mean f S.E. sone-treated cells were challenged with either the Ca2+ ionophore (A23187) or exogenous arachidonic acid for 15 min, they only synthesized about 40% as much TXA2 as LPS-or zymosan-treated cells. The amount of TXAz synthesized upon treatment with either A23187 or exogenous arachidonic acid was weakly inhibited by pretreatment with dexamethasone (~1 0 % ) over the same concentration range where >80% inhibition was noted for either LPS or zymosan. These results indicate that the dexamethasone block appears to be upstream from PGH synthase (reversed by exogenous arachidonic acid) and possibly upstream from release of esterified arachidonic acid (reversed by A23187).
The mechanism of the dexamethasone-induced inhibition was further investigated by determining the kinetics of the inhibitory response as a function of dexamethasone concentration using LPS as the stimulus. dU937 cells were treated with various concentrations of dexamethasone for 0-4 h and then stimulated with LPS for 4 h in the absence of steroid. TXAz synthesis was quantified, and the results are shown in Fig. 5. The rate of induction of the inhibitory effect was directly proportional to the initial dexamethasone concentration in the medium. Maximal inhibition of TXAz synthesis was observed after 15 min (1 pM), 60 rnin (0.1 pM), or 120 min (0.01 p~) , while maximal inhibition at 0.001 PM was not achieved even after 4 h of treatment. At near physiological concentrations of glucocorticoid ( 10-7-10-8 M) the inhibitory effect requires 1-2 h to reach a maximum, a time frame consistent with the requirement for protein synthesis for expression of the inhibitory response. To determine whether transcription or translation was required for the dexamethasone response, the effect of various macromolecular synthesis inhibitors on both LPS-stimulated TXAz synthesis and the dexamethasone inhibition of the LPS response was determined. The results summarized in Table I show that concentrations of cycloheximide or emitine that give 94 and 99% inhibition of translation, respectively, completely inhibited the LPS-stimulated TXAz synthesis in the absence of dexamethasone. A concentration of actinomycin D that inhibits transcription 89% also completely inhibited LPS-stimulated TXAz synthesis, independent of dexamethasone addition. Because the LPS response was blocked by transcription and translation inhibitors in the absence of dexamethasone, it was not possible to demonstrate a reversal of the glucocorticoid response. Although dexamethasone and the macromolecular

Effect of various inhibitors on macromolecular synthesis and LPS-stimulated TXAZ synthesis
Triplicate wells of PMA-differentiated U937 cells (1.5 X lo6 cells/ well) were pretreated with the noted concentration of drug for 30 min (actinomycin D, cycloheximide, and emitine) or 120 min (dexamethasone). The medium was then replaced with medium containing drug and either [3H]uridine (10 pCi/ml) (a), I3H]leucine (10 pCi/ml) (b), or [36S]methionine (c) and the incubations continued from 1 to 2 h. Trichloroacetic acid-insoluble radioactivity in the cell layer was then quantified by liquid scintillation counting, and the TXB2 concentration of the culture medium was measured by EIA. The results are expressed as percent inhibition of cells not treated with drug.

Synthesis by Glucocorticoids
synthesis inhibitors both prevented LPS-stimulated TXA2 synthesis, dexamethasone did not inhibit bulk translation and only weakly inhibited transcription at a dose that maximally inhibited TXA2 synthesis.
Because the LPS-stimulated cells had been treated with cycloheximide for a total of 4.5 h, it was necessary to establish that the inhibition of LPS-stimulated TXAz synthesis was due to inhibition of translation and not a loss of biosynthetic enzymes. The half-life of TXAz synthesizing capacity was estimated by comparing TXAz synthesis stimulated by LPS uersw A23187 as a function of time of cycloheximide treatment. dU937 cells were pretreated with cycloheximide for varying amounts of time (ranging from 30 to 240 min), and then TXA2 synthesis was initiated with either LPS or A23187 (added 15 min prior to harvest of the LPS-treated cultures) in the continued presence of cycloheximide. TXBz levels were then determined, and the results are summarized in Table 11. The half-life of TXA2 synthesizing capability (the sum of TXA2 synthase, PGH synthase, and phospholipase) for A23187 was approximately 270 min while, aside from a slight stimulation at early times, the LPS-induced response was inhibited at all times tested. From these results we conclude that the observed inhibition of LPS-stimulated TXAz synthesis by protein synthesis inhibitors is not due to a loss of TXA2 synthase, PGH synthase, or even phospholipase but is due to the requirement of protein synthesis for expression of the TXA, synthesis aspect of the LPS response.
Role of Lipocortin 1 in Mediating the Dexamethasone Response-Previous studies have suggested that dexamethasone inhibits arachidonic acid mobilization by inducing the synthesis of a phospholipase Az inhibitor protein, and lipocortin 1 has been proposed as a candidate for the glucocorticoid second messenger. The possible role of lipocortin 1 in mediating the dexamethasone response in dU937 cells was investigated by studying the effect of dexamethasone on lipocortin 1 metabolism. dU937 cells were metabolically labeled with [35S]methionine for either 2 or 4 h in the absence or presence of dexamethasone M ) . The incorporation of [35S]methionine into protein was not affected by the addition of dexamethasone to the labeling medium (Table I). The lipocortin 1 content of both the culture medium and cell layer was then determined by immune precipitation using a lipocortin 1specific rabbit polyclonal antiserum under conditions of antibody excess. The immune precipitated lipocortin 1 was analyzed by SDS-PAGE, detected by fluorography, and the results are shown in Fig. 6 (top panel). A single 37-kDa band was detected in samples derived from the cell layer (lanes [1][2][3][4] but not the culture medium (lanes [5][6][7][8]. Furthermore, the

TABLE I1 Kinetics of cycloheximide inhibition of LPSand A23187-stimulated TXA, synthesis
Triplicate wells of PMA-differentiated U937 cells (1.5 X IO6 cells) were pretreated for 30 min with fresh medium or with medium containing 10 pg/ml cycloheximide. The medium was then removed and replaced with identical media containing LPS (10 pglml). At the appropriate times (30, 120,240 min), culture medium was harvested and the TXB2 concentration determined by EIA. For the A23187-stimulated samples, the pretreatments were continued for the desired treatment time followed by 15-min treatment with A23187 (1 pg/ml) in the same media, and the TXB2 content of the medium was determined by EIA.  (lanes 3,4, 7, or 8). Lanes 9 and 10 are the same as lane 1 with either preimmune serum or no antibody controls, respectively. The culture medium and cell lysates were divided and immunoprecipitated using a lipocortin 1-specific polyclonal antiserum. Immune complexes were dissociated, run on 15% SDS-PAGE, and the bands visualized by fluorography (top panel). Lunes 1-4 are from whole cell lysates, and lanes 5-8 are from the culture medium. Pulse-chase labeling is shown in the bottom panel. Cells were cultured and labeled as described above. After a 2-h labeling, the medium was replaced with fresh medium containing 2 mM methionine and chased for 0 h (lanes 1 and  7), 1 h (lanes 2 and 8), 2 h (lanes 3 and 9), 4 h (lanes 4 and IO) 5 and l l ) , or 21 h (lanes 6 and 12). Radiolabeled lipocortin 1 from both the culture medium (lanes 1-6) and cell layer (lanes 7-12) was immunoprecipitated and chromatographed as described above.
amount of lipocortin 1 detected appeared to be unchanged by dexamethasone treatment (compare lanes 1/2 and 3/41. The lack of lipocortin 1 release into the culture medium (lanes 5-8) was unexpected since it was originally purified from cellfree supernatants conditioned by dexamethasone-treated rat peritoneal macrophages. To address this discrepancy, the metabolism lipocortin 1 was further studied by pulse-chase labeling. dU937 cells were pulse labeled with [35S]methionine for 2 h and chased in medium containing excess unlabeled methionine for various intervals. Labeled lipocortin 1 was detected by immune precipitation as described above, and the results are shown in Fig. 6 (bottom panel). As before, there was no lipocortin 1 released into the medium (lunes I$), even at prolonged chase times (21 h). The cell associated lipocortin 1 pool appeared metabolically inert, with a half-life of greater than 21 h under these conditions. The apparent lack of induction by dexamethasone, the lack of secretion into the culture medium, and the long half-life of the intracellular form are not consistent with the properties of lipocortin-like activities reported previously. To verify the lack of dexamethasone induction of lipocortin 1, we attempted to measure induction under conditions that we had previously determined were maximal for inhibition of LPS-stimulated TXA2 synthesis. dU937 cells were glucocorticoid starved by treatment overnight with RPMI 1640, 10% steroid-free FCS and then treated with RPMI 1640,1% steroid-free fetal calf serum with or without 0.1 p~ dexamethasone for various times. At the appropriate time point, replicate dishes were harvested and analyzed for lipocortin 1 mRNA (Northern blot) or for lipocortin 1 protein (Western blot). The results are shown in Fig.  7. The top panel shows the steady state levels of lipocortin 1 mRNA detected by hybridization with a 32P-labeled 1.3-kb EcoRI fragment of human lipocortin cDNA as a probe. The resting level of lipocortin 1 mRNA (1.4 kb) is readily detected (lane I) and it increased slightly with time in culture (lunes 2, 4, 6, and 8). Addition of dexamethasone to the culture medium gave results comparable to the untreated cultures (lanes 3,5, 7, and 9). Similar results were obtained when the protein was measured directly by Western blotting of cell lysates containing equal amounts of protein (middle panel). A slight but significant decrease in the level of lipocortin 1 during the treatment was noted, and this decrease was unaffected by dexamethasone (compare even and odd numbered lanes). The densitometric quantification of both the lipocortin 1 mRNA levels (normalized to tubulin mRNA) and lipocortin 1 (by comparison to a blot containing dilutions of pure lipocortin 1) is shown in the bottom panel. Since the levels of lipocortin 1 are not changing in response to dexamethasone under conditions where maximal inhibition of LPS-stimulated TXA2 synthesis is observed, we conclude that lipocortin 1 induction is not responsible for the observed decrease in arachidonic acid mobilization. An alternate possibility is that the absolute amount of lipocortin 1 is not changing but the protein is,undergoing a modification (phosphorylation) during LPS stimulation that is blocked by dexamethasone. In this regard, we have not been able to detect phosphorylated lipocortin 1 in immune precipitates prepared from cells prelabeled with 32PO:-and stimulated with either LPS or PMA (data not shown).

DISCUSSION
The exact mechanism by which glucocorticoids inhibit arachidonic acid mobilization and eicosanoid biosynthesis is unknown. We have developed a cellular model that is responsive to glucocorticoids in an attempt to further define the mechanism of inhibition of eicosanoid biosynthesis. The model is based on a human histocytic lymphoma cell line (U937 cells) that can be terminally differentiated into cells possessing a variety of phenotypic markers characteristic of monocytes/macrophages by treatment with PMA (29). PMAdifferentiated U937 cells were chosen over their undifferentiated counterparts for the following reasons; 1) they possess glucocorticoid receptors and have approximately 3-fold more receptors than undifferentiated cells, 2) they synthesize TXA2 as the major product in response to a variety of agonists while the undifferentiated cells produce very low amounts of eicosanoid products, and 3) they express at least two families of phospholipase A2 inhibitor proteins, namely lipocortins and antiinflammatory protein (13,14,30). The glucocorticoid receptor density in the PMA-differentiated U937 cells is similar to that reported for human peripheral blood monocytes (31). Increases in glucocorticoid receptor number upon Synthesis by Glucocorticoids terminal differentiation of HL60 cells have been reported (32), and the increase is comparable to that observed in U937 cells. Arachidonic acid metabolism in these cells is also consistent with previous reports (33, 34) with regard to both the detection of TXA, as the major product and the large increase in biosynthetic capacity after differentiation with PMA. The increased eicosanoid biosynthetic capacity upon differentiation of U937 cells by dimethyl sulfoxide has been reported and appears to be due to increased levels of phospholipase (35). Similar studies in HL60 cells indicate that induction of PGH synthase is responsible for increased prostaglandin synthesis capacity upon terminal differentiation (36). Inhibition of prostaglandin synthesis by glucocorticoids in macrophages (5,37-40), neutrophils (6), renomedullary interstitial cells (7, 41), and fibroblasts (2, 42) has been well documented. A mechanism involving glucocorticoid induction of a protein that inhibits phospholipase A2 and hence arachidonic acid mobilization and prostaglandin synthesis has been proposed (3-7). Evidence supporting this mechanism includes the release of the inhibitor from the cells by dexamethasone treatment and the ability of macromolecular synthesis inhibitors to block glucocorticoid-induced inhibitor synthesis. Several phospholipase A2 inhibitor proteins have been isolated based on their ability to inhibit porcine pancreas phospholipase A2 in vitro, and their complete structures have been defined by molecular cloning (14,15,30). Although these phospholipase A2 inhibitor proteins have been regarded as glucocorticoid second messengers, the exact relationship between these inhibitors and inhibition of eicosanoid synthesis has not been studied systematically. We have verified the inhibitory effect of glucocorticoids in our cellular model and have further defined the mechanism by which they block TXA2 synthesis by utilizing a variety of agonists to activate the cells to produce TXA2. Inhibition of TXA2 synthesis was specific for glucocorticoid steroids and also for the nature of the stimulus. Agonists that cause general cell activation (LPS or zymosan) stimulated maximal synthesis of TXA2 by a process that was strongly inhibited by glucocorticoid analogs of various potencies (43). Incomplete inhibition of agoniststimulated TXA2 synthesis by dexamethasone was due to background synthesis of TXA2 when the culture medium was changed. The quantity synthesized in control cultures, presumably due to synthesis initiated by exogenous arachidonic acid and/or serum factors, consistently matched the amount of TXA, that was not blocked by maximal concentrations of dexamethasone (data not shown). Agents which activate only a portion of the eicosanoid biosynthetic pathway (A23187 or arachidonic acid) gave lower yields of TXA2, but the synthesis was not affected by dexamethasone. One interpretation of these results is that the block occurs upstream from both PGH synthase and release of esterified arachidonic acid. Alternatively, the LPS/zymosan-stimulated activation of TXA2 synthesis may be coupled to a different set of biosynrun on a denaturing gel (glyoxal), electrotransferred to a Nytran membrane, and probed with a 32P-labeled 1.3-kb EcoRI fragment of human lipocortin 1 as described under "Experimental Procedures." Lipocortin 1 transcripts were detected by autoradiography and the signal normalized to an internal standard mRNA (tubulin). In the middle panel, postnuclear supernatants containing 20 pg of protein/ lane were run on 15% SDS-PAGE, electroblotted to nitrocellulose, and probed with lipocortin 1-specific polyclonal antiserum (1/400 dilution). Antigen-antibody complexes were visualized using Iz5Iprotein A followed by autoradiography. Quantification was performed by running dilutions of pure lipocortin 1 from placenta in adjacent lanes followed by densitometry of the autoradiography signals. The thetic enzymes (phospholipase/PGH synthase) than the enzymes that make TXAz when given exogenous arachidonic acid or A23187. In the second case, the LPS/zymosan-coupled system would be dexamethasone-sensitive, while the other set of enzymes would be insensitive to the inhibitory effects of dexamethasone. At present, it is not possible to distinguish between these two mechanisms.
Assuming that A23187 activates phospholipase indirectly by Ca2+ influx, the lack of inhibition by dexamethasone was unexpected in the context of a phospholipase-inhibitor protein complex, unless complex formation is reversed by increased levels of intracellular Ca". Kinetic analysis of the inhibitory response revealed that at physiological concentrations of glucocorticoids, maximal inhibition of eicosanoid synthesis was achieved within 1-2 h, and the effect was not reversible for at least 4 h after steroid removal. These observations are consistent with the requirement for transcription and/or translation for expression of the inhibitory response. It was not possible to reverse the inhibitory effect of glucocorticoids using either actinomycin D or cycloheximide since these agents also blocked TXAz synthesis initiated by either LPS or zymosan. The lipocortin hypothesis predicts that induction of a phospholipase inhibitor protein gene (lipocortin 1) and concomitant inhibition of arachidonic acid mobilization is responsible for the observed inhibition of eicosanoid synthesis. We did not observe any correlation between induction of the steady state levels of lipocortin 1 mRNA or lipocortin 1 protein and the glucocorticoid inhibition. Expression of the lipocortin 1 gene has been reported to be induced severalfold in rat peritoneal macrophages from dexamethasone-treated rats, but the increase was not correlated with a block in eicosanoid synthesis or increases in lipocortin 1 protein levels (14). Under conditions where maximal inhibition of TXAz synthesis occurs, there are no changes in the lipocortin 1 content of the cells. We could not detect lipocortin 1 secretion (constitutive or dexamethasone-induced) from dU937 cells or release from human neutrophils upon degranulation (data not shown). The lipocortin 1 cDNA sequence (14) does not predict a signal sequence as would be expected for a secretory protein. Addition of a purified lipocortin 1 preparation that was active as a phospholipase A2 inhibitor in the in vitro assay to dU937 cells did not block LPS-induced TXAz synthesis in the dose range 0.25-25 pg/ml.' These results are not consistent with points a and c of the operational definition presented in the Introduction. Our results are consistent with a recent report demonstrating the lack of correlation of lipocortin 1 as an inhibitor of zymosan-stimulated arachidonic acid release from macrophages under conditions where glucocorticoids are active (44). Alternatively, recombinant lipocortin 1 has been reported to inhibit leukotriene C4-induced TXA, synthesis from perfused lung (19) and prostacyclin synthesis by arterial rings (45). The reasons for the discrepancy between inhibitory activity of recombinant lipocortin 1 and the apparent lack of activity of natural lipocortin 1 purified from placenta are unknown since they have not been compared in parallel (21, 44). Finally, the significance of pancreatic phospholipase A2 inhibition in vitro (point b) and the antiinflammatory activity of lipocortin 1 has been questioned (16,18,44). Since the original lipocortinlike activities were detected in cell-free supernatants from steroid-treated cells, the relationship between lipocortin 1 (as defined by molecular cloning) and previously reported phospholipase inhibitor proteins is unclear. Involvement of other members of the lipocortin family in mediating the glucocorticoid inhibition in this system can not be ruled out. Inclusion * M. A. Petro, unpublished work. of these proteins into the lipocortin family is based on amino acid sequence homology with lipocortin 1 (46) and the ability to inhibit phospholipase Az-catalyzed hydrolysis of E. coli membranes by binding to the substrate micelles (16). Since lipocortin 1 was the first member of this family and does not appear to be a glucocorticoid second messenger, it is unlikely that other members of the lipocortin family are responsible for the inhibitory action of glucocorticoids on eicosanoid synthesis. If the mechanism of glucocorticoid suppression of eicosanoid synthesis does involve a steroid-inducible gene, the possibility remains that the protein factor has yet to be defined.
Although the dexamethasone block has been mapped upstream from phospholipase, the lack of understanding of the signal transduction pathways involved in macrophage activation by either LPS or zymosan has complicated our efforts to map the inhibition more accurately. LPS stimulation of the murine macrophage cell line RAW.264.7 leads to the synthesis of prostaglandins and the accumulation of lysophosphatidylinositol (47) suggesting that cell activation by LPS may be coupled to a phosphoinositol-specific phospholipase Az. The LPS-induced accumulation of lysophosphoinositol is partially blocked by cycloheximide, consistent with our observation that protein synthesis is required for LPS-induced TXA, production (47). Activation of macrophage eicosanoid synthesis by zymosan has been studied in more detail (48). Initiation of synthesis requires binding of the zymosan particles to the cell surface (presumably via mannose receptors) but can be dissociated from phagocytosis (49). As with LPS, stimulation of eicosanoid synthesis by zymosan is blocked by inhibitors of transcription or translation (50). In fact, eicosanoid synthesis initiated by both LPS and zymosan is inhibited by both dexamethasone and protein synthesis inhibitors, while neither of these agents affect A23187-or arachidonic acid-induced TXAz synthesis. The parallel activities of dexamethasone and transcription/translation inhibitors suggests that glucocorticoids may be acting by suppression of transcription and/or translation. Dexamethasone does not inhibit bulk transcription or translation under conditions where it is maximally active against eicosanoid synthesis, but it may act by inhibiting the expression of a single or discrete set of genes that are required for autocrine stimulation of the cells to produce TXAz. LPS-inducible macrophage products include lymphokines such as IL1 and TNF and both of these factors are known to stimulate prostaglandin synthesis in a number of cell types. In mouse peritoneal macrophages, IL1 production leads to autocrine stimulation of prostaglandin E synthesis which in turn down-regulates IL1 and TNF synthesis (51). Furthermore, it has been recently reported that dexamethasone inhibits LPS inductioa of ILlp in both peripheral blood monocytes or U937 cells (52, 53) and of TNF induction in mouse macrophages (54,55). We have demonstrated dexamethasone inhibition of LPS-induced increases in both ILlp and TNFa transcripts in our system, and the concentration dependence is similar to the inhibition of TXAz ~ynthesis.~ Down-regulation of ILl/TNF production by glucocorticoids and a resultant loss of prostaglandin synthesis could account for the observed inhibition of TXA2 synthesis and would simultaneously prevent the synthesis of two families of inflammatory mediators, ILl/TNF and eicosanoids. Although our results are consistent with dexamethasone inhibition of the synthesis of an activator of TXAZ synthesis, based on the available data we cannot rule out the induction of an inhibitor of TXAz synthesis. Further investigations of the dexamethasone inhibition of TXAZ synthesis and its relationship to both L. J. Robinson, unpublished observations.