Secretory Phospholipase A 2 Mediates Cooperative Prostaglandin Generation by Growth Factor and Cytokine Independently of Preceding Cytosolic Phospholipase A 2 Expression in Rat Gastric Epithelial Cells*

Transforming growth factor (TGF)- a and interleukin (IL)-1 b are responsible for the healing of gastric lesions through, in part, prostaglandin (PG) generation. We examined the contribution of cytosolic and secretory phospholipase A 2 s (cPLA 2 and sPLA 2 ) to the PG generation by rat gastric epithelial cells in response to both stimuli. 2 2 2 2 2 2 2 2 2 but 2 IL-1 b -stimulated cells, in which p11, a putative cPLA 2 inhibitory molecule, was also increased and co-immuno-precipitated with cPLA 2 . These results suggest that syn- ergistic stimulation of sPLA 2 and COX-2 expression by TGF- a and IL-1 b results in an increase in PGE 2 . Presumably, the preceding cPLA 2 expression is not involved in the PGE 2 generation, because of impairment of its hy- drolytic activity in the stimulated cells.

Activation of phospholipase A 2 (PLA 2 ) 1 in a variety of cells including inflammatory cells leads to the liberation of arachidonic acid (AA), which is, in turn, converted to several types of prostaglandins (PGs) by the sequential action of cyclooxygenase (COX) and PG synthases, this being the initial step in the generation of PG (1). Among numerous types of PLA 2 s identified in mammalian cells and tissues, Ca 2ϩ -dependent cytosolic PLA 2 (cPLA 2 , type IV) and non-pancreatic secretory PLA 2 (sPLA 2 , types IIA, V, and X) are responsible for stimulusinduced AA liberation and subsequent PG generation (2)(3)(4)(5)(6)(7). Recent reports showed that activation of cPLA 2 is required for the onset of sPLA 2 -catalyzed hydrolysis of membrane phospholipids (8) or for induction of sPLA 2 expression (9, 10) in mouse P388D 1 macrophages and rat 3Y1 fibroblasts stimulated with proinflammatory stimuli. Conversely, exogenous sPLA 2 stimulates cPLA 2 activation through a receptor-mediated mechanism in human 1321N1 astrocytoma cells (11) or lysophospholipid formation resulting from hydrolysis of membrane phospholipids in rat mesangial cells (12). Thus, the existence of cross-talk between the two PLA 2 s has been proposed. On the other hand, stimulus-induced AA liberation in platelets having the two PLA 2 s is mainly mediated by cPLA 2 (13), because sPLA 2 is unable to hydrolyze membrane phospholipids of platelets unless the membrane undergoes certain modifications, such as lipid peroxidation (14). Therefore, the isozyme(s) of PLA 2 s contributing to the generation of PG varies with the types of cells and/or signaling pathways elicited by agonists.
It is generally accepted that PG plays an important role in the homeostasis or progression of a variety of diseases with inflammation. The generation of PG in gastric cells including epithelial cells and fibroblasts is involved in the healing of gastric lesions and in the maintenance of gastric mucosal integrity (15,16). In an animal model of stress-induced gastric ulcers, generation of PGE 2 and expression of two COX isoforms, COX-1 and COX-2, are accelerated during the healing of the lesions in parallel with increases in transforming growth factor (TGF)-␣ and epidermal growth factor (EGF) (17). TGF-␣ (18) and EGF (19) have been shown to stimulate PGE 2 generation and COX-2 expression in rat gastric epithelial RGM1 cells and guinea pig gastric mucosal cells. TGF-␣, which is produced by gastric epithelial cells (20), exhibits numerous biological activities including inhibition of gastric acid secretion and stimulation of migration and proliferation of gastric epithelial cells (21,22) through binding to EGF receptor (23,24). A recent study showed that TGF-␣ induces cell proliferation and morphogenesis through a COX-2-dependent mechanism in RGM1 cells (25). On the other hand, the expression of interleukin (IL)-1␤, a proinflammatory cytokine, as well as COX-2 is also accelerated during the healing of ischemia/reperfusion-induced gastric lesions in an animal model (26). In hu-man gastric fibroblasts, stimulation with IL-1␤ induces COX-2-dependent PGE 2 generation (27), leading to the expression of hepatocyte growth factor (27,28), which stimulates proliferation of rabbit gastric epithelial cells (28,29). Furthermore, cytoprotection by IL-1␤ against ethanol-induced gastric injury was shown to be dependent on the generation of PGE 2 in an animal model (30). Consequently, it is conceivable that TGF-␣ and IL-1␤ are implicated in the healing of gastric lesions through, at least in part, COX-2-dependent PGE 2 generation. However, the role of TGF-␣ and IL-1␤ in the activation of the PLA 2 isozyme(s) contributing to the generation during the healing remains to be elucidated.
Recently, we demonstrated that TGF-␣ induces cPLA 2 expression as well as PGE 2 generation with no change in sPLA 2 activity in RGM1 cells, although the increase in PGE 2 was slight (18). Furthermore, TGF-␣ was found to inhibit Ca 2ϩ ionophore-induced cPLA 2 activation in parallel with an increase in p11 (18), known as a putative cPLA 2 inhibitory molecule (31,32). These findings suggest that TGF-␣-stimulated PGE 2 generation occurs under the control of inducible p11, which negatively regulates the hydrolytic activity of cPLA 2 . Considering the possibility that IL-1␤ is also involved in the healing of gastric lesions through generation of PG, TGF-␣ may cooperate with IL-1␤ to stimulate efficiently the generation in gastric epithelial cells. This appears likely, because a combination of EGF and IL-1␤ has been shown to induce synergistically COX-2 expression and PGE 2 generation in human gingival fibroblasts (33). IL-1␤ is well known to induce sPLA 2 expression in a variety of cells (34 -36). However, several growth factors including EGF have been reported to inhibit cytokineinduced sPLA 2 release (37, 38) but enhance PG generation (37) in rat calvarial osteoblasts and astrocytes. Thus, little is known about the PLA 2 isozymes acting upon stimulation with both growth factors and cytokines.
The present study was undertaken to identify the isozyme(s) of PLA 2 s contributing to the generation of PGE 2 in rat gastric epithelial RGM1 cells stimulated with a combination of TGF-␣ and IL-1␤. For this purpose, we examined changes in the two PLA 2 isozymes, cPLA 2 and sPLA 2 , in comparison to an increase in PGE 2 upon the stimulation, taking into account the possible existence of cross-talk between the PLA 2 s and selective cooperation with certain COXs.
PGE 2 Generation-The [ 3 H]AA-labeled RGM1 cells were treated with various reagents and then stimulated with TGF-␣, IL-1␤, or both as described in the figure legends. Lipids in the medium and cells were extracted and separated by thin layer chromatography on a Silica Gel G plate using an upper phase of ethyl acetate/isooctane/acetic acid/H 2 O (9:5:2:10, v/v) as the development system. The area corresponding to PGE 2 was scraped off, and the radioactivity was determined by liquid scintillation counting.
AA Liberation Induced by A23187-The [ 3 H]AA-labeled RGM1 cells were stimulated with TGF-␣, IL-1␤, or both for 24 h, washed, and placed in 1 ml of fresh DMEM/F-12. The cells were treated with various reagents in the presence of 10 M BW755C (a COX and lipoxygenase inhibitor) for 30 min and further stimulated with 1 M A23187 for 10 min. Lipids in the medium and cells were extracted and separated by thin layer chromatography on a Silica Gel G plate using petroleum ether/diethyl ether/acetic acid (40:40:1, v/v) as the development system (40). The area corresponding to free fatty acid was scraped off, and the radioactivity was determined by liquid scintillation counting.
Conversion of Exogenous AA to PGE 2 -RGM1 cells were stimulated with TGF-␣ and IL-1␤ in the presence of indoxam as described in the figure legends, washed, and placed in 0.5 ml of fresh DMEM/F-12. The cells were further treated with indoxam for 30 min and incubated with a mixture of [ 3 H]AA and the unlabeled compound (50 Ci/mol, 2 M) for 30 min. After lipids in the medium and cells were extracted, the radioactivity of [ 3 H]PGE 2 was determined as described above.
Immunoblot Analysis for COXs, cPLA 2 , and p11-After stimulation of RGM1 cells with TGF-␣, IL-1␤, or both as described in the figure legends, the cells were washed and scraped off in buffer A (100 mM NaCl, 2 mM EGTA, 100 M p-(amidinophenyl)methanesulfonyl fluoride, 100 M leupeptin, 20 mM ␤-glycerophosphate, 1 mM Na 3 VO 4 , and 10 mM Tris-HCl, pH 7.4) containing 0.05% Triton X-100. Samples (10 g of protein for p11 and 20 g of protein for cPLA 2 and COXs) were solubilized and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 7.5 (for cPLA 2 and COXs) or 15% (for p11) gel. The proteins were transferred to a nitrocellulose membrane, and then antibodies against COXs, cPLA 2 , or p11 were applied. The bound antibodies were visualized using peroxidase-conjugated secondary antibodies and enhanced chemiluminescence Western blotting detection reagents (Amersham Pharmacia Biotech).
Assay for cPLA 2 Activity-RGM1 cells were treated with or without MAFP or AACOCF 3 and stimulated with TGF-␣, IL-1␤, or both, as described in the figure legends. After the medium was removed, the cells were scraped off and sonicated in buffer A containing 0.05% Triton X-100. The protein concentrations in the lysate were adjusted to 1 mg/ml, and the lysate was treated with 5 mM dithiothreitol at 37°C for 15 min to inhibit sPLA 2 activity. The sample containing 10 l of the lysate was incubated with a mixture of 1-stearoyl-2-[ 3 H]arachidonoylsn-glycero-3-phosphocholine and the unlabeled compound (250 Ci/mol, 2 M) at 37°C for 1 h in the presence of 5 mM CaCl 2 and 50 mM Tris-HCl, pH 8.5, in a final volume of 200 l. After lipid extraction, the [ 3 H]AA liberated was analyzed as described above, and the enzyme activity was calculated.
Assay for sPLA 2 Activity-RGM1 cells, placed in 0.5 ml of DMEM/ F-12 containing 0.1 mg/ml heparin, were stimulated with TGF-␣, IL-1␤, or both as described in the figure legends. After the medium was centrifuged, the supernatant (20 l), as an enzyme source, was incubated with 1-palmitoyl-2-[ 14 C]linoleoyl-sn-glycero-3-phosphoethanolamine (2 M) at 37°C for 15 min in the presence of 5 mM CaCl 2 , 1 mg/ml bovine serum albumin, and 50 mM Tris-HCl, pH 8.5, in a final volume of 200 l. After lipid extraction, the [ 14 C]linoleic acid liberated was analyzed as described above, and the enzyme activity was calculated.
RNA Blotting-RGM1 cells (8 ϫ 10 6 ) in 100-mm dishes were stimulated with TGF-␣, IL-1␤, or both for 24 h. Total RNA (30 g) was extracted from the cells using TRIzol reagent, subjected to formaldehyde-agarose gel electrophoresis on a 1% gel, and transferred to a nylon membrane. The membrane was hybridized with probes for rat type IIA (5Ј-GTG CCA CAT CCA CGT TTC TCC AGA CGG TTG), V (5Ј-TCA TGG ACT TCA GTT CTA GCA AGC CCC CTG), or X (5Ј-TAT CGG TAT AGC TTG GGG CTG CAG CCG GCA) sPLA 2 , which had been labeled with alkaline phosphatase for 12 h using a commercial labeling kit (Amersham Pharmacia Biotech). The bound probes were visualized using CDP-Star detection reagent (Amersham Pharmacia Biotech). After the probes were stripped off, the membrane was rehybridized with alkaline phosphatase-labeled probe for glyceraldehyde-3-phosphate dehydrogenase (5Ј-GAG GGA GTT GTC ATA TTT CTC GTG GTT CAC).
Immunoprecipitation-RGM1 cells (8 ϫ 10 6 ) in 100-mm dishes were stimulated with TGF-␣ and IL-1␤ for 24 h, sonicated in buffer B (100 mM KCl, 1 mM EGTA, 100 M leupeptin, 100 M p-(amidinophenyl)methanesulfonyl fluoride, 1.1 mM CaCl 2 , and 20 mM Tris-HCl, pH 7.4), and subjected to immunoprecipitation of cPLA 2 . Briefly, the lysate (0.4 mg of protein) was incubated with protein A-agarose at 4°C for 30 min, as a pre-clearing step (41), and centrifuged at 10,000 ϫ g for 3 min. The supernatant was incubated with anti-cPLA 2 antibodies overnight at 4°C and further with protein A-agarose for 2 h. After centrifugation, the pellet obtained was washed three times with buffer B and solubilized. The sample was subjected to immunoblot analysis with antibodies against cPLA 2 , phosphoserine, or p11 as described above.
Statistical Analysis-Values are expressed as the mean Ϯ S.E. of three or four separate experiments. Data were analyzed by one-way analysis of variance followed by Bonferroni's test. p Ͻ 0.05 was considered statistically significant. Fig. 1A, stimulation of RGM1 cells with 50 ng/ml TGF-␣ for 24 h slightly increased PGE 2 in a time-dependent manner, whereas 5 ng/ml IL-1␤ had no effect on the basal level of PGE 2 . However, a combination of 50 ng/ml TGF-␣ and 5 ng/ml IL-1␤ was found to stimulate synergistically the generation of PGE 2 with a marked increase observed 18 h after the stimulation. The synergistic effect on PGE 2 generation was dose-dependent of IL-1␤ or TGF-␣ (Fig. 1,  B and C). Under similar experimental conditions, we determined changes in COX-1 and COX-2 proteins (Fig. 2). Fig. 2A shows that 50 ng/ml TGF-␣ but not 5 or 10 ng/ml IL-1␤ significantly increased COX-2, whereas in combination they synergistically induced COX-2 expression. However, in contrast to the time course of PGE 2 generation (Fig. 1A), Fig. 2B revealed that COX-2 expression induced by the combination was apparently increased 3-6 h after the stimulation. On the other hand, COX-1 protein was not detectable in RGM1 cells even when the cells were stimulated with 50 ng/ml TGF-␣ or 5 ng/ml IL-1␤ under our experimental conditions ( Fig. 2A). However, the combination of TGF-␣ and IL-1␤ was found to increase markedly COX-1 protein. The increase in COX-1 induced by the combination was also observed 3-6 h after the stimulation, as shown in Fig. 2B. To examine the involvement of the two COXs in the synergistic effect on PGE 2 generation, the effects of COX inhibitors on the generation of PGE 2 were determined. The results shown in Fig. 2C indicate that generation of PGE 2 in cPLA 2 Expression and Delayed sPLA 2 Release in Response to TGF-␣ and IL-1␤-To identify the PLA 2 isozyme(s) contributing to the synergistic effect of TGF-␣ and IL-1␤ on PGE 2 generation, the change in cPLA 2 upon stimulation was examined (Fig. 3). Stimulation with 50 ng/ml TGF-␣ for 24 h increased cPLA 2 protein and activity, whereas 10 ng/ml IL-1␤ did not affect the basal levels (Fig. 3, A and B). In contrast to the stimulatory effect of IL-1␤ on TGF-␣-induced PGE 2 generation (Fig. 1B), 50 ng/ml TGF-␣-increased cPLA 2 protein or activity was not augmented in the presence of 2-10 ng/ml IL-1␤ (Fig. 3,  A and B). Furthermore, the combination of 50 ng/ml TGF-␣ and 5 ng/ml IL-1␤ time-dependently increased cPLA 2 protein with an apparent increase observed 6 h after the stimulation (Fig. 3C), indicating that the time-dependent expression of cPLA 2 preceded the onset of the synergistic increase in PGE 2 (Fig. 1A).

Increase in PGE 2 Preceded by Expression of COXs in Response to TGF-␣ and IL-1␤-As shown in
We further determined the change in sPLA 2 activity in the extracellular medium. The result shown in Fig. 4A indicates that IL-1␤ (2-10 ng/ml) dose-dependently increased sPLA 2 activity, which was enhanced in the presence of 50 ng/ml TGF-␣, although TGF-␣ did not increase the activity by itself. As shown in Fig. 4B, a marked increase in sPLA 2 activity was observed 18 h after stimulation with 5 ng/ml IL-1␤ alone or in combination with 50 ng/ml TGF-␣, indicating that the synergistic effect on PGE 2 generation (Fig. 1A) occurred in parallel with an increase in sPLA 2 activity. Pretreatment of cells with 1 M cycloheximide or 1 M actinomycin D inhibited the increase in sPLA 2 activity induced by both TGF-␣ and IL-1␤ (data not shown). We further determined the change in mRNA for types IIA, V, and X sPLA 2 s, as shown in Fig. 4C. Stimulation with 5 ng/ml IL-1␤ for 24 h induced type IIA sPLA 2 mRNA expression, whereas 50 ng/ml TGF-␣ did not affect the mRNA, which was undetectable in unstimulated cells. Consistent with the change in sPLA 2 activity, TGF-␣ augmented the mRNA expression of type IIA sPLA 2 induced by IL-1␤, whereas mRNA for type V or X sPLA 2 was not detectable under our experimental conditions. Recently, we reported the possible involvement of type VI Ca 2ϩ -independent PLA 2 in zymosan-stimulated liberation of AA in P388D 1 cells, in which the PLA 2 translocates to membranes upon the stimulation (44). We further examined the effect of TGF-␣ and IL-1␤ on type VI PLA 2 activity in RGM1 cells. However, the activity in the lysate or membrane fraction of unstimulated RGM1 cells was not affected by stimulation with 50 ng/ml TGF-␣ and/or 5 ng/ml IL-1␤ for 24 h (data not shown).
Effects of PLA 2 Inhibitors on TGF-␣/IL-1␤-stimulated PGE 2 Generation-To examine the involvement of cPLA 2 in the synergistic PGE 2 generation by TGF-␣ and IL-1␤, we tested the effects of MAFP and AACOCF 3 , cPLA 2 inhibitors, on cPLA 2 activity and PGE 2 generation upon the stimulation. As shown in Fig. 5A, when RGM1 cells pretreated with 2 M MAFP or 5 M AACOCF 3 were stimulated with both 50 ng/ml TGF-␣ and 5 ng/ml IL-1␤, the cPLA 2 activity in lysate of the stimulated cells as well as basal activity in unstimulated cells were significantly suppressed by the inhibitors. Under these conditions, PGE 2 generation induced by TGF-␣ alone was also inhibited, whereas synergistic PGE 2 generation induced by both TGF-␣ and IL-1␤ was not affected (Fig. 5B), indicating that the cPLA 2 inhibitors had no effect on the augmentation by IL-1␤ of TGF-␣-induced PGE 2 generation. Recently, MAFP and AACOCF 3 were reported to inhibit stimulus-induced sPLA 2 release (9, 10, 36), suggesting a requirement of cPLA 2 for sPLA 2 expression. As shown in Fig. 5A, however, the cPLA 2 inhibitors did not affect an increase in sPLA 2 activity induced by both TGF-␣ and IL-1␤. Under our experimental conditions, 2 M MAFP or 5 M AACOCF 3 had no effect on cell viability (92 or 90%, respectively; control, 96%; the mean of two determinations), but more than 10 M MAFP or AACOCF 3 decreased it to less than 70 or 80%, respectively, as estimated by trypan blue dye exclusion. Fig. 6 illustrates the effects of indoxam, a specific sPLA 2 inhibitor (45, 46), on increases in sPLA 2 activity and PGE 2 in response to TGF-␣ and IL-1␤. When RGM1 cells were stimulated with both 50 ng/ml TGF-␣ and 5 ng/ml IL-1␤ for 12 h and further incubated for 12 h in the presence of indoxam (5-100 nM), the increased sPLA 2 activity in the medium was inhibited by indoxam in a dose-dependent manner (Fig. 6A). Under these conditions, the sPLA 2 inhibitor significantly suppressed TGF-␣/IL-1␤-stimulated PGE 2 generation at the same concentration ranges (Fig. 6B). However, 100 or 200 nM indoxam did not affect cPLA 2 activity or conversion of exogenous AA to PGE 2 in TGF-␣/IL-1␤-stimulated cells (Fig. 6C).
Effects of TGF-␣ and IL-1␤ on A23187-induced AA Liberation-It was reported that p11, known as annexin II light chain, directly inhibits cPLA 2 activity through binding to the enzyme (31), and its overexpression results in suppression of Ca 2ϩ ionophore A23187-induced AA liberation (32). Recently, we showed that incubation of RGM1 cells with TGF-␣ for 3-24 h inhibited A23187-induced AA liberation in parallel with p11 expression (18). In the present study, we further examined whether TGF-␣ exhibits similar effects in the presence of IL-1␤. The results shown in Fig. 7A indicate that when RGM1 cells, which had been stimulated with 50 ng/ml TGF-␣ for 24 h in the presence or absence of 5 ng/ml IL-1␤, were washed and further stimulated with 1 M A23187 for 10 min, TGF-␣ diminished A23187-induced AA liberation as compared with TGF-␣unstimulated cells even in the presence of IL-1␤. Furthermore, 50 ng/ml TGF-␣ induced an increase in p11 irrespective of the presence of 5 ng/ml IL-1␤ (Fig. 7B). However, IL-1␤ alone did not inhibit, but rather enhanced, A23187-induced AA liberation without influencing the amount of p11. Moreover, as shown in Fig. 7C, when cPLA 2 in RGM1 cells was immunoprecipitated using anti-cPLA 2 antibodies, amounts of cPLA 2 protein including its phosphorylated form (phosphoserine) in the precipitate were increased by stimulation with 50 ng/ml TGF-␣ alone or in combination with 5 ng/ml IL-1␤. Under these conditions, p11 protein was also precipitated; its amounts in the precipitate were increased with an increase in cPLA 2 protein (Fig. 7C). These results suggest that the inhibitory effect of TGF-␣ on A23187-induced AA liberation occurs in parallel with p11 expression but independently of IL-1␤-elicited signaling.
It has been shown that overexpression of sPLA 2 enhances A23187-induced AA liberation, and the augmented liberation is suppressed by the cPLA 2 inhibitor MAFP (47), suggesting that cPLA 2 is involved in the sPLA 2 -catalyzed liberation of AA. The present study showed that 1 M A23187-induced AA liberation was enhanced by preincubation with 5 ng/ml IL-1␤ (Fig. 7A), which increased sPLA 2 (Fig. 4) in RGM1 cells having constitutive cPLA 2 (Fig. 3). To examine the contribution of cPLA 2 and sPLA 2 to the IL-1␤-enhanced AA liberation, the effects of MAFP and indoxam were determined (Fig. 7D). Treatment of IL-1␤-stimulated cells with 10 M MAFP markedly inhibited the IL-1␤/A23187-induced AA liberation, whereas 50 nM indoxam diminished the enhanced liberation to the level of AA liberation induced by A23187 alone. We confirmed that MAFP but not indoxam almost completely suppressed AA liberation induced by A23187 alone (Fig. 7D). Incubation with 10 M MAFP within 1 h had no effect on cell viability (data not shown). These findings suggest that cPLA 2 is involved in the hydrolytic action of sPLA 2 when IL-1␤-primed RGM1 cells are stimulated with Ca 2ϩ -mobilizing agonists.
Comparison of the Effects of TGF-␣ with Those of Dibutyryl cAMP-It has been suggested that the EGF receptor mediates the activation of type I protein kinase A through a direct binding of the Grb2 adaptor protein to the regulatory subunit of the kinase in EGF-stimulated human mammary epithelial MCF-10A cells (reviewed in Ref. 48). To examine the role of protein kinase A in the sPLA 2 release in TGF-␣/IL-1␤-stimulated RGM1 cells, effects of dibutyryl cAMP, a protein kinase A activator, were compared with those of TGF-␣. As shown in Fig.  8A, 0.5 mM dibutyryl cAMP as well as 50 ng/ml TGF-␣ augmented the increase in sPLA 2 activity induced by 5 ng/ml IL-1␤, although it did not increase sPLA 2 activity by itself. The stimulatory effect of dibutyryl cAMP or TGF-␣ was almost completely suppressed by pretreatment with 100 nM H-89, a protein kinase A inhibitor (data not shown). These results suggest that protein kinase A may be involved in the augmentation by TGF-␣ of IL-1␤-induced sPLA 2 release. However, unlike TGF-␣, the combination of 0.5 mM dibutyryl cAMP and 5 ng/ml IL-1␤ did not sufficiently induce PGE 2 generation or COX-2 expression as compared with those by the combination of TGF-␣ and IL-1␤ (Fig. 8, B and C). DISCUSSION Several growth factors and proinflammatory cytokines including TGF-␣ and IL-1␤ are implicated in the healing of gastric lesions. TGF-␣ and IL-1␤ have been shown to stimulate the generation of PG by rat gastric epithelial RGM1 cells (18,25) and human gastric fibroblasts (27), respectively. Furthermore, a COX inhibitor prevents TGF-␣-induced proliferation and morphogenesis of RGM1 cells (25) and IL-1␤-induced expression of hepatocyte growth factor in human gastric fibroblasts (27), which stimulates proliferation of rabbit gastric epithelial cells (28,29). These observations suggest that acceleration by TGF-␣ and IL-1␤ of the repair of gastric injury is mediated through, at least in part, generation of PG. We demonstrated here that whereas TGF-␣ or IL-1␤ alone had little or no effect, in combination they synergistically stimulated the generation of PGE 2 by RGM1 cells (Fig. 1), suggesting that the cooperative action of TGF-␣ and IL-1␤ also is involved in PG generation during the healing of gastric mucosal lesions. In the present study, we further examined the contribution of cPLA 2 and sPLA 2 to the generation of PGE 2 via a certain COX isoform(s), and the possible interaction between the two PLA 2 s in the TGF-␣/IL-1␤-stimulated RGM1 cells.
The onset of the synergistic increase in PGE 2 induced by both TGF-␣ and IL-1␤ was observed 18 h after the stimulation (Fig. 1A). Under these conditions, both stimuli synergistically increased levels of COX-1 and COX-2 proteins with a significant change observed 3-6 h after the stimulation (Fig. 2, A and  B), indicating that the increases in the two COXs preceded the synergistic effect on PGE 2 generation. Although COX-1 is known as a constitutive enzyme (49), our result showing an increase in COX-1 is consistent with the recent finding that the expression of COX-1 as well as COX-2 is stimulated during the healing of gastric ulcers in an animal model (17). However, the synergistic PGE 2 generation was suppressed by the COX-2 inhibitor NS-398 but not the COX-1 inhibitor valeryl salicylate (Fig. 2C), suggesting that PGE 2 generation in the stimulated cells is mediated by COX-2 rather than COX-1. The preference for COX-2 over COX-1 may be due to a difference in ability to utilize AA (50) and/or selective coupling of COX-2 with PGE 2 synthase (51). Collectively, our findings suggest that an increase in COX-2 is required for the synergistic PGE 2 generation induced by both TGF-␣ and IL-1␤ but is not critical for the onset of the generation.
The generation of PG in the stomach plays an important role in the healing of gastric lesions and in the maintenance of gastric mucosal integrity (15,16). Therefore, a number of studies have focused on the regulation of COX isoforms, as shown in gastric epithelial cells (19,25,52,53), gastric fibroblasts (27), and animal models of gastric ulcers (17,26,54). However, little is known about the regulation of PLA 2 isozyme(s) in gastric cells, despite an important role for PLA 2 as a key enzyme providing AA, a precursor for PGs. Recently, we reported that stimulation of RGM1 cells with TGF-␣ alone induces an increase in cPLA 2 but not sPLA 2 in parallel with a slight and gradual increment of PGE 2 (18), suggesting that cPLA 2 is involved in the generation of PGE 2 . In the present study, although IL-1␤ dose-dependently enhanced TGF-␣-induced PGE 2 generation (Fig. 1B), it did not augment increases in cPLA 2 protein and activity in response to TGF-␣ at the same concentration ranges (Fig. 3, A and B). The time-dependent increase in cPLA 2 induced by both TGF-␣ and IL-1␤ apparently preceded the onset of the synergistic PGE 2 generation (Figs. 1A and 3C). Furthermore, the augmentation by IL-1␤ of TGF-␣stimulated PGE 2 generation was not affected by the cPLA 2 inhibitors MAFP or AACOCF 3 (Fig. 5B). These results suggest, therefore, that the preceding increase in cPLA 2 is not involved in the synergistic PGE 2 generation induced by both IL-1␤ and TGF-␣.
Among several types of sPLA 2 s identified in mammalian cells and tissues (2), types IIA, V, and X sPLA 2 s are implicated in stimulus-induced PG generation in a variety of cells (5-12). However, the role of sPLA 2 in PG generation by gastric cells remains unknown. In the present study, we demonstrated that a dose-and time-dependent increase in sPLA 2 activity evoked by IL-1␤ in the presence of TGF-␣ was well consistent with the augmentation by IL-1␤ of the TGF-␣-induced PGE 2 generation ( Figs. 1 and 4). Furthermore, suppression of the augmented PGE 2 generation by the specific sPLA 2 inhibitor indoxam was observed in parallel with inhibition of the activity of sPLA 2 released at the same concentration ranges (Fig. 6). These findings suggest that an increase in sPLA 2 is involved in the synergistic PGE 2 generation induced by both TGF-␣ and IL-1␤. Considering the result that mRNA for type IIA sPLA 2 but not for types V or X sPLA 2 was increased upon the stimulation (Fig. 4C), the increased sPLA 2 activity may be due to, at least in part, induction of type IIA sPLA 2 expression. Although type IIA sPLA 2 has been suggested to stimulate liberation of AA through hydrolysis of membrane phospholipids (12,55,56), certain changes in membrane properties are required for an increase in the susceptibility of membrane phospholipids to sPLA 2 (7,14). At present, the mechanism(s) responsible for the change in cellular susceptibility to sPLA 2 in TGF-␣/IL-1␤-stimulated RGM1 cells is unclear.
A number of studies have suggested the existence of interaction between sPLA 2 and cPLA 2 . As the mechanism responsible for the interaction, exogenous sPLA 2 has been shown to activate p42 mitogen-activated protein kinase, inducing activation of cPLA 2 and liberation of AA, in human 1321N1 astrocytoma cells (11) and rat mesangial cells (12). Conversely, stimulus-induced cPLA 2 activation accelerates sPLA 2 -catalyzed AA liberation in P388D 1 macrophages (8) and sPLA 2overexpressing cells (47). Furthermore, cPLA 2 inhibitors (MAFP and AACOCF 3 ) suppress expression of sPLA 2 (9,10,36) and subsequent sPLA 2 -mediated generation of PG (9,10) induced by cytokines or lipopolysaccharide in several types of cells. Thus, the mechanisms underlying the cross-talk between the two PLA 2 s vary with the types of cells and/or stimulation systems. In the present study, we showed that A23187-induced AA liberation was potentiated in IL-1␤-stimulated RGM1 cells (Fig. 7A), in which constitutive cPLA 2 and inducible sPLA 2 co-existed (Figs. 3 and 4). The augmentation was markedly prevented by MAFP in addition to the sPLA 2 inhibitor indoxam (Fig. 7D). These findings also suggest that cPLA 2 may contribute to sPLA 2 -catalyzed AA liberation when IL-1␤-primed RGM1 cells are activated by Ca 2ϩ -mobilizing agonists. However, TGF-␣ alone did not stimulate sPLA 2 expression despite an increase in cPLA 2 (Figs. 3 and 4). Furthermore, under conditions where TGF-␣ and IL-1␤ synergistically stimulated PGE 2 generation with preceding cPLA 2 expression and concomitant sPLA 2 release (Figs. 1, 3 and 4), indoxam suppressed the PGE 2 generation, whereas cPLA 2 inhibitors had no effect on the increase in sPLA 2 or PGE 2 (Figs. 5 and 6). Accordingly, our findings suggest that cPLA 2 is not involved in the sPLA 2 release and subsequent PGE 2 generation in the TGF-␣/IL-1␤stimulated cells.
It has been shown that p11, known as annexin II light chain, directly inhibits cPLA 2 activity through binding to the enzyme (31), and further that a decrease in constitutive p11 results in a further increase in basal levels of free AA as well as A23187induced AA liberation (32), suggesting that the hydrolytic activity of cPLA 2 is ordinarily regulated by p11 in cells. We recently reported that priming of RGM1 cells with TGF-␣ alone for 3-24 h inhibited A23187-induced, cPLA 2 -catalyzed AA liberation, and concurrently increased p11 but not annexin II heavy chain (18). Suppression of the p11 expression by a tyrosine kinase inhibitor restored the inhibited AA liberation in response to A23187 (18). The present study showed that priming with both TGF-␣ and IL-1␤ also resulted in inhibition of cPLA 2 -catalyzed AA liberation despite an increase in cPLA 2 protein, including its phosphorylated form, and further that even in the presence of IL-1␤, TGF-␣ increased p11, which was co-immunoprecipitated with cPLA 2 (Fig. 7). Based on these observations, we speculate that p11 expression might be involved in the impairment of the hydrolytic activity of cPLA 2 , although further study is needed to clarify whether p11 actually associates with cPLA 2 within the intracellular environment. It is possible, therefore, that because of the inhibitory effect on cPLA 2 in the TGF-␣/IL-1␤-stimulated cells, cPLA 2 may be unable to contribute to the sPLA 2 -mediated PGE 2 generation.
IL-1␤ has been shown to stimulate the expression of sPLA 2 (34,35) and COX-2 (27,57) as well as the generation of PG in a variety of cells. In RGM1 cells stimulated with IL-1␤ alone, however, PGE 2 generation was not observed despite an increase in sPLA 2 (Figs. 1 and 4). This inability of IL-1␤ may be due to the small increase in COX-2 in response to IL-1␤ alone ( Fig. 2A). Indeed, under conditions where COX-2 was increased by TGF-␣, IL-1␤ was able to induce sPLA 2 -mediated PGE 2 generation (Figs. 1, 2, and 6). Although TGF-␣ increased COX-2, it augmented IL-1␤-induced sPLA 2 release (Fig. 4). Furthermore, dibutyryl cAMP, a protein kinase A activator, also exhibited similar stimulatory effect on the sPLA 2 release (Fig. 8A). However, the combination of IL-1␤ and dibutyryl cAMP did not sufficiently increase PGE 2 or COX-2 as compared with those by the combination of IL-1␤ and TGF-␣ (Fig. 8, B  and C). Taken together, these observations suggest that the synergistic PGE 2 generation induced by IL-1␤ and TGF-␣ is mediated through cooperative action of sPLA 2 with COX-2.
In summary, based on the results obtained here, we suggest that TGF-␣ and IL-1␤ synergistically stimulate the expression of sPLA 2 and COX-2, therefore inducing an increase in PGE 2 generation, which is independent of the expression of cPLA 2 and COX-1, in rat gastric epithelial RGM1 cells. Furthermore, no contribution of cPLA 2 to the PGE 2 generation may be ascribed in part to the inhibitory effect of TGF-␣ on the hydrolytic activity of cPLA 2 , the negative regulation of which presumably occurs in parallel with p11 expression.