Mitogen-activated Protein (MAP) Kinase Regulates Production of Tumor Necrosis Factor-α and Release of Arachidonic Acid in Mast Cells

Aggregation of the high affinity IgE receptor (FcεRI) in a mast cell line resulted in activation of the p42 and the stress-activated p38 mitogen-activated protein (MAP) kinases. Selective inhibition of these respective kinases with PD 098059 and SB 203580 indicated that p42 MAP kinase, but not p38 MAP kinase, contributed to the production of the cytokine, tumor necrosis factor-α, and the release of arachidonic acid in these cells. Neither kinase, however, was essential for FcεRI-mediated degranulation or constitutive production of tumor growth factor-β. Studies with SB 203580 and the p38 MAP kinase activator anisomycin also revealed that p38 MAP kinase negatively regulated activation of p42 MAP kinase and the responses mediated by this kinase.

Stimulation of mast cells by aggregation of membrane IgE receptors (Fc⑀RI), leads to recruitment of the tyrosine kinase Syk and activation of Syk-dependent signaling cascades (1,2). These cascades include activation of phospholipase C and sphingosine kinase for mobilization of calcium ions and PKC 1 (3,4) and the activation of p42 MAP kinase cascade through Ras (2,5). These cascades lead ultimately to secretion of intracellular granules, a response primarily driven by the increase in [Ca 2ϩ ] i and activation of PKC (6), and a cPLA 2 -mediated release of arachidonic acid. The activation of cPLA 2 is dependent on increase of [Ca 2ϩ ] i and phosphorylation by MAP kinase (2,7,8).
Stimulated mast cells also produce a variety of cytokines that include interleukins 1, 3, 4, 5, and 6 as well as TNF␣ and granulocyte-macrophage colony-stimulating factor (9,10). Typically, increased expression of cytokine mRNA and protein is detectable 30 min to several hours after the addition of stimulant (11). These cytokines, particularly TNF␣, are thought to mediate pathologic inflammatory reactions (10) and protective responses to bacterial infection (12). The production and release of TNF␣ are regulated through signals transduced by calcium and PKC, although there are indications that additional Fc⑀RI-mediated signals may operate for optimal production of TNF␣ in cultured RBL-2H3 mast cells. Compared with antigen, other stimulants are relatively weak inducers of TNF␣ production when doses of stimulants are matched for maximal stimulation of degranulation (13). Also, concentrations of Ro31-7549 that block PKC, secretion of granules, and release of TNF␣ only partially block production of TNF␣ (13).
The present objective was to determine whether stimulation of MAP kinases induces additional signals for production of TNF␣. A linkage between these events has not been established in mast cells. Antigen-induced stimulation of p42 MAP kinase coincides with the activation of its upstream regulators, Ras, Raf, and MEK-1 (2,5), and persists through the period when production of TNF␣ would be most apparent (14). As noted in this paper, however, RBL-2H3 cells also possess the mammalian homologue of the yeast HOG-1 protein kinase, p38 MAP kinase. We have utilized the MEK-1 inhibitor, PD 098059 (15, 16), and the p38 MAP kinase inhibitor, SB 203580 (17), to evaluate the role of these MAP kinases in the production of TNF␣ and, for comparison, the release of arachidonic acid, degranulation, and production of TGF␤. Release of arachidonic acid is thought to be dependent on phosphorylation of cPLA 2 by MAP kinase, although the identity of the MAP kinase is uncertain (18). Degranulation and TGF␤ production were assumed to be MAP-kinase-independent responses (7,19). We show that, while p42 MAP kinase regulated production of both TNF␣ and arachidonic acid, p38 MAP kinase negatively regulated the activation of p42 MAP kinase and the responses mediated by this kinase.  (20) and Adams et al. (21), respectively, and purified by column chromatography and recrystallization. These compounds were determined to be Ͼ95% pure on the basis of high * 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   Cell Culture and Measurement of Stimulatory Responses-The RBL-2H3 cell line was maintained in complete growth medium (minimum essential medium) supplemented with 15% fetal calf serum, glutamine, antibiotic, and antimycotic agents. Trypsinized cells were plated into 150-mm culture dishes or six-well Costar cluster plates and were incubated overnight in complete growth medium with O-dinitrophenolspecific IgE (0.5 g/ml) and, for measurement of arachidonic acid release, [ 14 C]arachidonic acid (0.1 Ci/ml).

Reagents-Reagents
Cultures were washed the next day and replenished with the required medium. For the assay of hexosaminidase or [ 14 C]arachidonic acid, experiments were performed in a PIPES-buffered medium (25 mM PIPES, pH 7.2, 159 mM NaCl, 5 mM KCl, 0.4 mM MgCl 2 , 1.0 mM CaCl 2 , 5.6 mM glucose, and 0.1% fatty acid-free fraction V bovine serum albumin). For [ 32 P]phosphorylation of proteins, cultures were incubated for 90 min with 32 P-labeled orthophosphate in PIPES-buffered medium exactly as described (7). For all other assays, experiments were performed in complete growth medium supplemented with 15% fetal calf serum (for measurement of TNF␣), 5% fetal calf serum (for measurement of TGF␤), or 0.1% bovine serum albumin and 25 mM Hepes, pH 7.2 (for assay of MAP kinases and separation of proteins by immunoprecipitation and electrophoresis). The inhibitors were added either 30 min (PD 098059) or 15 min (SB 203580 and indomethacin) before stimulation of cultures with antigen (DNP-BSA) as described in the figure legends.
Measurement of Degranulation, Release of [ 14 C]Arachidonic Acid, and Production of TNF␣ and TGF␤-Release of the granule marker, hexosaminidase, was determined by colorimetric assay of medium and cell lysates by previously described procedures (6). For measurement of release of arachidonic acid, cells were labeled to equilibrium with [ 14 C]arachidonic acid before the addition of inhibitors and antigen as described above. Reactions were terminated by placing cultures on ice and rapidly removing medium. The medium was briefly centrifuged (Beckman Microfuge for 30 s) to remove extraneous cells. Both medium and cell lysates (in 0.1% Triton X-100) were assayed for hexosaminidase (6) and radiolabel (22). Values were expressed as the percentage of intracellular hexosaminidase or radiolabel that was released into the external medium, and they were corrected for spontaneous release from unstimulated cells. It should be noted that, in RBL-2H3 cells, arachidonic acid is metabolized in part to leukotriene C 4 /B 4 and prostaglandin D 2 via the 5-lipoxygenase and cyclooxygenase pathways, respectively (23,24). Release of radiolabel, as measured in this paper, was an estimate of total release of [ 14 C]arachidonic acid and its metabolites. The cytokines were assayed as described elsewhere (19). Whole cell lysates were prepared by freezing and thawing the cultures three times. TGF␤ was assayed with a human TGF␤ enzyme-linked immunosorbent assay kit, which utilized a mouse monoclonal anti-human antibody that cross-reacted with rat TGF␤. TNF␣ was assayed with a murine TNF␣ enzyme-linked immunosorbent assay kit, which utilized a monoclonal hamster anti-murine antibody that reacted with mouse or rat TNF␣ and -␤. The limits of detection for these assays were 25 pg of TGF␤/10 6 cells and 6 pg of TNF␣/10 6 cells. Values were corrected for spontaneous release in the absence of stimulant (Յ3% for release of hexosaminidase, Յ2% for release of arachidonic acid, and undetectable release of TNF␣) except for TGF␤, which was produced constitutively in RBL-2H3 cells (19).
Assay of MAP Kinase Activity in Cell Extracts-After stimulation of cultures in six-well cluster plates, the cultures were washed once, and the medium was removed. The plates were then placed on ice before the addition of 510 l of a Tris buffer (25 mM Tris, pH 7.5, 25 mM NaCl, 1 mM Na 3 VO 4 , 2 mM EGTA, 1.5 mM dithiothreitol, 2.5 mM p-nitrophenyl phosphate, and 20 g/ml leupeptin and aprotinin). Cells were disrupted by freezing and thawing three times. The lysate was centrifuged (15,800 ϫ g for 10 min), and 450 l of the supernatant fraction was mixed with 50 l of ethylene glycol and 80 l of washed phenyl-Sepharose. The phenyl-Sepharose was washed beforehand with 300 l of the Tris buffer. The mixture was kept on ice for 5 min for binding of MAP kinase to the beads. After centrifugation, the phenyl-Sepharose beads were washed with 1 ml of 10% (v/v) ethylene glycol and then with 30% (v/v) ethylene glycol. Finally, MAP kinase was eluted by incubating the beads with 75 l of 60% ethylene glycol for 5 min on ice. After centrifugation of the suspension, 15 l of supernatant was incubated (15 min, 30°C) in a solution that contained 50 mM Tris (pH 7.5), 10 mM MgCl 2 , [␥-32 P]ATP (10 Ci/mmol, 37 kBq/tube), and 25 g of MAP kinase substrate peptide (peptide 94 -102 of bovine myelin basic protein). The phosphorylated peptide was isolated by centrifugation of the incubation mixture through phosphocellulose membrane (SpinZyme; Pierce), which was then washed twice with 500 l of 75 mM H 3 PO 4 for the assay of radioactivity.
Assay of MAP Kinase and MEK Activities by Immunoprecipitation- A, cultures were labeled with [␥-32 P]pyrophosphate and then stimulated as described above. MEK was immunoprecipitated from cell lysates and separated by electrophoresis for autoradiography. B, unlabeled cultures were stimulated as noted above. MEK was immunoprecipitated and then assayed by phosphorylation of p42 mapk -glutathione S-transferase fusion protein (extracellular signal-regulated kinase) as substrate with [␥-32 P]ATP. 32 P-Labeled substrate was separated by electrophoresis for detection by autoradiography as described under "Materials and Methods." C, cell lysates were analyzed by Western blotting for p42 MAP kinase. The lower band, previously identified as p42 MAP kinase (p42 mapk ), is indicated. The additional retarded band represents the activated tyrosine-phosphorylated form of p42 MAP kinase (2,14). D, unlabeled cells were stimulated as described above. MAP kinase was immunoprecipitated with antibody against MAP kinase R2 and assayed for MAP kinase activity using myelin basic protein (MBP) as the substrate. Phosphorylated protein was separated by electrophoresis and detected by autoradiography. The mixture was incubated at 30°C for 12 min. The reaction was terminated by the addition of 30 l of 2 ϫ SDS sample buffer. MEK was immunoprecipitated with anti-MEK antibody and assayed similarly except that p42 MAP kinase glutathione S-transferase fusion protein (1 g/assay) was used as substrate for phosphorylation. Proteins were separated by 12% SDS-PAGE. Radioactive proteins were detected by autoradiography.
Electrophoretic Separation and Immunoblotting of [ 32 P]MEK, p42 MAP Kinase, and cPLA 2 -The preparation of cell lysates and immunoprecipitates, analysis of proteins by SDS-PAGE, and transfer to nitrocellulose paper were performed as described elsewhere (2, 7) with the following exception: cPLA 2 was separated on NOVEX 10% Tris/glycine gels for 3 h at 35 mA and 4°C as described by Kramer and co-workers (25). Previously described procedures were used for isolation and detection of [ 32 P]MEK (7). Otherwise, proteins were detected by the immunoblotting technique with antibodies against MEK, p42 MAP kinase, cPLA 2 , or anti-phosphotyrosine. Secondary antibodies included horseradish peroxidase-conjugated antibody against rabbit IgG or mouse IgG. Finally, proteins were visualized by the ECL System (Amersham Corp.) or by autoradiography.

RESULTS
MEK Inhibitor PD 098059 Inhibits the Activity of p42 MAP Kinase, Release of Arachidonic Acid, and Production of TNF␣-As shown in Fig. 1, the MEK inhibitor PD 098059 attenuated antigen-induced [ 32 P]phosphorylation of MEK (panel A) and the activation of MEK as determined by in vitro assay of immunoprecipitated MEK (panel B). Activation of p42 MAP kinase was also attenuated, as indicated by the change in electrophoretic migration of p42 MAP kinase (panel C) or by the assay of MAP kinase activity of immunoprecipitated p42 MAP kinase (panel D). The extent of these inhibitions was dependent on the concentration of PD 098059. As shown in Fig.  2, the suppression of MAP kinase activation by PD 098059 (panel A) was associated with similar dose-dependent suppression of arachidonic acid release (panel B) and TNF␣ production (panel C). The suppression of the latter two responses was highly correlated (r Ͼ 0.95). All three responses were inhibited by ϳ50% with 10 M PD 098059. As will be described later, activation of cPLA 2 was also inhibited by PD 098059. These results suggested that release of arachidonic acid and production of TNF␣ were both regulated by p42 MAP kinase.
The Effect of PD 098059 on Degranulation and TGF␤ Production-To test the selectivity of PD 098059, we next examined the effects of this compound on antigen-stimulated degranulation and the constitutive production of TGF␤, which are thought not to be regulated by MAP kinase (7,19). PD 098059 had only minimal effects on stimulated release of the granule constituent, hexosaminidase (Fig. 3A) and the production of TGF␤ in unstimulated cells (Fig. 3B). The only significant effect was partial inhibition (Ͻ30%) of degranulation at 50 M PD 098059 (Fig. 3A).
The p38 MAP Kinase Inhibitor, SB 203580, Enhances Activation of p42 MAP Kinase, Release of Arachidonic Acid, and Production of TNF␣ in Antigen-stimulated Cells-Antigen stimulation also resulted in increased activity of p38 MAP kinase (Fig. 4A, compare lanes 1 and 2). The p38 kinase inhibitor, SB 203580, inhibited this activation (Fig. 4A, lanes 3 and  4). Interestingly, antigen activation of p42 MAP kinase was enhanced significantly by SB 203580. This enhancement was apparent when cells were stimulated with 20 or 200 ng/ml antigen (Fig. 4B). The latter concentration of antigen was known to elicit maximal activation of p42 MAP kinase. 2 These results suggested that p38 MAP kinase negatively regulates p42 MAP kinase and that this regulation is alleviated by SB 203580.
The enhanced activation of p42 MAP kinase in the presence of SB 203580 was associated with increased release of arachidonic acid (Fig. 5A) and production of TNF␣ (Fig. 5B). In the experiment shown in Fig. 5B, cells were stimulated with a low concentration of antigen (6 ng/ml) to maximize enhancement of the TNF␣ response (250% increase in Fig. 5B). At optimal doses of antigen enhancement of TNF␣ production was less (40 -80% increase) but still significant (data not shown).
Because pyridinyl imidazoles that are closely related to SB 203580 have cyclooxygenase-inhibitory activity (25,26), experiments were conducted to determine whether blockade of cyclooxygenase activity with indomethacin (27) altered accumulation of radiolabel in the medium by suppressing metabolism [ 14 C]arachidonate via this enzyme. Unlike SB 203580, 10 M indomethacin did not significantly alter release of radiolabel from antigen-stimulated cells (7.8 Ϯ 0.4% release over 15 min versus 7.2 Ϯ 0.2% release in the absence of indomethacin; mean Ϯ S.E. in eight cultures from two experiments). It seemed probable, therefore, that SB 203580 enhanced release rather than the accumulation of [ 14 C]arachidonic acid in the medium.
In contrast to the increased release of arachidonic acid and production of TNF␣, SB 203580 had no significant effect on antigen-induced degranulation (Fig. 5C) or constitutive production of TGF␤ (Fig. 5D). Collectively, these results provided further evidence for the notion that release of arachidonic acid and TNF␣ production are both regulated by p42 MAP kinase. In addition, the results suggested that p38 MAP kinase negatively modulates these responses through p42 MAP kinase.
The above results suggested that release of arachidonic acid, as well as production of TNF␣, was regulated by p42 MAP kinase. As in other systems (25,28), the phosphorylation of cPLA 2 in stimulated RBL-2H3 cells leads to decreased electrophoretic mobility of the enzyme (7). The connection between p42 MAP kinase and the release of arachidonic acid via cPLA 2 was further demonstrated by the finding that the antigeninduced retardation of electrophoretic migration of cPLA 2 (25) was suppressed by PD 098059 but not by SB 203580 (Fig. 6).
Activation of p38 MAP Kinase by Anisomycin Partially Suppresses Activation of p42 MAP Kinase and Release of Arachidonic Acid-The p38 MAP kinase activator, anisomycin, markedly activated this kinase (Fig. 7A) but much less so p42 MAP kinase (Fig. 7B). The combination of anisomycin and antigen revealed inhibitory communication between these two kinase. For example, the combination of stimulants resulted in less activation of p38 MAP kinase (Fig. 7C, lane 3) than that induced by antigen (Fig. 7C, lane 2) or anisomycin (Fig. 7A, lane  2) alone. The combination also caused less activation of p42 MAP kinase (Fig. 7D, lane 3) than that by antigen alone (Fig.  7D, lane 2). Thus, stimulation of p42 MAP kinase by antigen appeared to block activation of p38 MAP kinase by anisomycin, and conversely stimulation of p38 MAP kinase by anisomycin appeared to partially block activation of p42 MAP kinase by antigen. Consistent with the latter situation, anisomycin partially suppressed antigen-induced release of arachidonic acid (25 Ϯ 4% reduction, mean of three experiments). This reduction corresponded to an approximately 25% reduction in p42 MAP kinase activation as determined by densitometric analysis of the blots shown in Fig. 7D and two other experiments. Anisomycin almost totally blocked (by 83 Ϯ 4%) antigen-induced production of TNF␣, probably as a consequence, however, of its known inhibitory actions on protein synthesis at the translation step (29). Presumably, de novo synthesis of TNF␣ would be especially sensitive to inhibitors of protein synthesis.

DISCUSSION
Past studies have shown that the responses evoked by antigen in RBL-2H3 cells were dependent on calcium and signals generated through PKC or MAP kinase. These studies indicated, for example, that PKC regulated degranulation (6) as well as the production and secretion of TNF␣, although it appeared likely that additional Fc⑀RI-mediated signals facilitated TNF␣ production (13). Activation of p42 MAP kinase, in contrast, was associated with phosphorylation of cPLA 2 and release of arachidonic acid (2,7). These studies, however, did not address the issue of whether other MAP kinases, such as p38 MAP kinase, regulated cPLA 2 .
The present results demonstrate that both p38 and p42 MAP kinases are activated in antigen-stimulated cells. Activation of the latter kinase appears to be most closely related to release of arachidonic acid and production of TNF␣. All three events are inhibited by the MEK inhibitor, PD 098059 (Fig. 2), and enhanced by the p38 MAP kinase inhibitor, SB 203580 (Figs. 4 and 5). Both compounds are reported to be selective inhibitors of MEK (i.e. PD 098059) and p38 MAP kinase (i.e. SB 203580) when tested against a wide range of kinases (15)(16)(17). In addition, the enhancement of responses in the presence of SB 203580, in contrast to the attenuation of p42 MAP kinase activation by the p38 MAP kinase activator, anisomycin (Fig.  7), suggest that p38 MAP kinase negatively regulates activation of p42 MAP kinase and its associated responses. Antigenstimulated degranulation and the constitutive production of TGF␤ in RBL-2H3 cells are minimally affected by the inhibitors (Figs. 3 and 5). Collectively, the results support the notion that p42 MAP kinase regulates release of arachidonic acid and promotes an additional signal for stimulating TNF␣ production but does not regulate degranulation. Interestingly, the p38 MAP kinase inhibitor, SB 203580, was first identified as an inhibitor of cytokine biosynthesis in lipopolysaccharide-stimulated human monocytes (17) and was subsequently shown to suppress TNF␣ production in lipopolysaccharide-injected mice (30). The compound also possessed anti-inflammatory activity in mouse models of arthritis (collagen-and adjuvant-induced), whereas cellular immune responses measured ex vivo were unaffected (30). It is possible, therefore, that different MAP kinase pathways are utilized for activating gene transcription for cytokine synthesis when synthesis is induced by inflammatory agents or through multimeric immunologic receptors such as Fc⑀RI.
The question has been raised whether p38 rather than p42 MAP kinase is responsible for the activation of cPLA 2 (18,25). cPLA 2 is phosphorylated by both kinases in thrombin-stimulated platelets, although the phosphorylation by p38 MAP kinase does not appear to activate cPLA 2 (25). Our results indicate that p42 MAP kinase regulates phosphorylation of cPLA 2 and release of arachidonic acid and suggest, therefore, that this kinase is the activator of cPLA 2 at least in RBL-2H3 cells.
Mast cells, including RBL-2H3 cells, also contain a low molecular weight secreted form (type II) of PLA 2 (31) in secretory granules, and this form is presumably released along with other granule constituents in activated cells (32). A role for secreted PLA 2 is unlikely, however, because total suppression of degranulation by selective inhibitors of PKC, such as Ro31-7549 and calphostin C, minimally affects release of arachidonic acid in RBL-2H3 cells (Refs. 7 and 33 and see below). 3 In addition to the correlations between activation of the p42 MAP kinase/cPLA 2 pathway and release of arachidonic acid noted here with PD 098059 and SB 203580, similar correlations have been noted in previous studies with less specific MAP kinase inhibitors. Activation of the entire Raf/MEK/p42 MAP kinase pathway and release of arachidonic acid were suppressed equally by the glucorticoid, dexamethasone, and the kinase inhibitor, quercetin, while effects on degranulation were apparent only at relatively high concentrations of these agents (7,34). Correlations were noted with the PKC inhibitor, Ro31-7549. This inhibitor transiently delayed activation of p42 MAP kinase in antigen-stimulated RBL-2H3 cells. There was a corresponding transient delay in the release of arachidonic acid, although the cumulative release eventually equaled that observed in the absence of Ro31-7549 (14). On the basis of these and other results, we have suggested that the p42 MAP kinase/ cPLA 2 pathway, although transiently activated by PKC, was predominantly activated by Ras through recruitment of Shc/ Grb2/Sos or Vav by Fc⑀RI (14). Others have reported that fatty acids, particularly arachidonic acid, activate p42 and p44 MAP kinases through PKC (35). This scenario is unlikely in antigenstimulated RBL-2H3 cells, however, because of the predominance of the PKC-independent (i.e. Ro31-7549-resistant) pathway in RBL-2H3 cells.
The present data extend previous findings on the regulation of TNF␣ synthesis and release. This cytokine is synthesized de novo and subsequently secreted via Golgi in a PKC-and calciumdependent manner (13). The PKC inhibitor, Ro31-7549, blocks secretion of TNF␣ but only partially suppresses synthesis of TNF␣ (6,13,19). Thus, additional signals may be necessary for optimal stimulation of TNF␣ synthesis. Antigen is a particularly potent stimulant of TNF␣ production when compared with the combination of calcium ionophore and PKC agonist, phorbol 12-myristate 13-acetate (13,19). These observations and the present studies with the MAP kinase inhibitors suggest that optimal production of TNF␣ is achieved through activation of both PKC and p42 MAP kinase.
The present findings may explain why antigen-induced activation of p42 MAP kinase (34), release of arachidonic acid (22,34) and production of TNF␣ (36) exhibit similar sensitivity to dexamethasone. All three responses are totally suppressed in RBL-2H3 cells that have been treated with 10 nM dexamethasone, whereas antigen-stimulated hydrolysis of phosphoinositides, increase in [Ca 2ϩ ] i , and degranulation (22,34) are only partially suppressed by treatment of cells with 100 nM dexamethasone. Dexamethasone, as noted above, inhibits the entire Raf/MEK/p42 MAP kinase/cPLA 2 pathway at nanomolar concentrations (34). Therefore, if p42 MAP kinase is the common regulator of TNF␣ production and arachidonic acid release, the similar sensitivities to dexamethasone would be expected.
The connections between the p42 MAP kinase pathway and cytokine production are unknown for mast cells, but recent reports indicate the following connections in other types of cells. The overexpression of Raf1 (37,38) or p42 MAP kinase (39) results in enhanced expression of a variety of cytokine genes in T cells and macrophages (37,39), the inactivation of IB (38), and the enhanced binding activity of cytokine transcription factors such as NF-B and AP-1 (39).
In conclusion, the above results provide the first indication that p42 MAP kinase regulates antigen-mediated production of TNF␣ in a mast cell line and that p38 MAP kinase may negatively regulate the p42 MAP kinase/cytokine pathway. We can, for the first time, broadly define the regulatory pathways for all three functional responses of mast cells to antigen as follows. Along with elevated [Ca 2ϩ ] i , the additional primary signals are as follows: for degranulation (6) and secretion of newly formed TNF␣ (13), activation of PKC (6); for cPLA 2 -mediated release of arachidonic acid, activation of p42 MAP kinase (Refs. 2, 7, and 34, and this paper); and for production of TNF␣, the coactivation of PKC and p42 MAP kinase (Ref. 13 and this paper).