Role of Cyclooxygenase-1 and -2, Phospholipase C, and Protein Kinase C in Prostaglandin-Mediated Gastroprotection

Orally administered 16,16-dimethyl-prostaglandin (PG) E₂ as well as endogenous PG released by 20% ethanol have been shown to be gastroprotective against 70% or absolute ethanol in rats (Robert et al., 1983, 1985; Gretzer et al., 1998). The gastroprotection by various agents, including 20% ethanol, is not only inhibited by indomethacin but also by the blocker of ATP-sensitive potassium (K_ATP) channels glibenclamide (Peskar et al., 2002). These findings suggest that the mechanism of action of endogenous PG involves the activation of K_ATP channels. Similarly, the gastroprotective effect of exogenous PG is antagonized by glibenclamide (Peskar et al., 2002). The specificity of the antagonism has been demonstrated by the use of the K_ATP channel activator cromakalim, which reverses the glibenclamide effects (Standen et al., 1989; Quayle et al., 1995). We have now further investigated the isoenzyme specificity and mechanisms of action of PG in gastroprotection. For this purpose, we have studied the effects of the nonselective cyclooxygenase (COX) inhibitor indomethacin, the COX-1 selective inhibitor SC-560 (Smith et al., 1998), and the COX-2 selective inhibitor rofecoxib (Chan et al., 1999) on gastroprotection conferred by 16,16-dimethyl-PGE₂ as well as endogenous PG, which mediate adaptive gastroprotection. Because it has been demonstrated that indomethacin-induced gastric mucosal damage can be prevented by cromakalim (Akar et al., 1999), we have investigated the interaction of cromakalim with indomethacin as well as the COX isoenzyme-selective inhibitors. Furthermore, in an attempt to further elucidate the mechanism of action of exogenous and endogenous PG we have used the phosphatidylinositol-specific inhibitor of phospholipase C U-73122 and its inactive analog U-73343 (Bleasdale et al., 1989). Finally, we have used the inhibitors of protein kinase C chelerythrine and staurosporine, both of which have been shown to be active in rats in vivo (Speechly-Dick et al., 1994; Kozak et al., 1997; Zacharowski et al., 1999).

Materials and Methods

Drugs. 5-(4-Chlorophenyl)-1-(4-methoxyphenyl)-3-trifluoromethylpyrazole (SC-560) was kindly provided by Dr. R. A. Marks (Searle, Skokie, IL). Rofecoxib (Vioxx) was purchased at the pharmacy. 1-(6-((17β-3-Methoxyestra-1,3,5(10)-trien-17-yl)amino)hexyl)-1H-pyrrole-2,5-dione (U-73122) and 1-(6-((17β-3-methoxyestra-1,3,5(10)-trien-17-yl)amino)hexyl)-2,5-pyrrolidine-dione (U-73343) were purchased from BIOMOL GmbH (Hamburg, Germany). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Animals. Male Wistar rats (180–220 g) were fasted overnight with free access to tap water. The studies reported in this article have been carried out in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the National Institutes of Health.

Protective Effect of 16,16-Dimethyl-PGE₂. All rats received 1 ml of 70% ethanol by oral intubation and were killed by cervical dislocation 5 min later. The stomach was removed and gross mucosal damage was assessed in a blinded manner by calculation of a lesion index by use of a 0 to 3 scoring system based on the number and severity factor of lesions as described previously (Stroff et al., 1996). The severity factor was defined according to the length of the lesions: 0, no lesions visible; I, lesions <1 mm; II, lesions 2 to 4 mm; and III, lesions >4 mm. The lesion index was calculated as the total number of lesions multiplied by their respective severity factor.

Rats received 16,16-dimethyl-PGE₂ (75 ng/kg p.o.) 30 min before instillation of 70% ethanol. Pilot experiments had shown that this dose is the lowest dose of the PG that confers full protection. Groups of six rats were treated with indomethacin (20 mg/kg p.o.), SC-560 (20 mg/kg p.o.), or rofecoxib (1 or 20 mg/kg p.o.) 30 min before administration of 16,16-dimethyl-PGE₂. In further experiments, rats pretreated with indomethacin (20 mg/kg), SC-560 (20 mg/kg), or rofecoxib (1 mg/kg) received additional oral treatment with cromakalim 30 min before administration of the COX inhibitors. The dose of cromakalim necessary to induce the maximal effect differed between the various treatment groups and ranged from 0.3 to 0.5 mg/kg. To investigate whether the attenuation of 16,16-dimethyl-PGE₂-induced gastroprotection by indomethacin depends on the dose of the PG used, groups of five to six rats received graded doses of the PG (75–300 ng/kg p.o.) 30 min after pretreatment with indomethacin (20 mg/kg p.o.), and mucosal damage was induced by 70% ethanol 30 min after administration of the PG.

To define the role of phospholipase C and protein kinase C in the protective effect of 16,16-dimethyl-PGE₂, groups of five to nine rats were treated with the phospholipase C inhibitor U-73122 (1 mg/kg i.v.) or its inactive analog U-73343 (1 mg/kg i.v.) or the protein kinase C inhibitors chelerythrine (0.7 mg/kg i.v.) or staurosporine (3 μg/kg i.v.), respectively. U-73122 and U-73343 were administered 15 min and chelerythrine and staurosporine 10 min before administration of 16,16-dimethyl-PGE₂. Pilot experiments had shown that the doses and treatment periods used induced maximal responses. As in the experiments with COX inhibitors, additional groups of rats were pretreated with cromakalim (0.3 mg/kg p.o.) 30 min before administration of U-73122, chelerythrine, or staurosporine.

Protective Effect of 20% Ethanol. All rats received 1 ml of 70% ethanol by oral intubation and gross mucosal damage was assessed 5 min later as described above. As protective agent, 1 ml of 20% ethanol was administered p.o. 30 min before instillation of 70% ethanol. Groups of six rats were pretreated with indomethacin (10 mg/kg p.o.), SC-560 (20 mg/kg p.o.), or rofecoxib (0.2 mg/kg p.o.) 30 min before instillation of 20% ethanol. Other rats (n = 6 in each group) received additional treatment with cromakalim (0.3–0.5 mg/kg p.o.) 30 min before oral administration of indomethacin (10 mg/kg), SC-560 (20 mg/kg), or rofecoxib (0.2 mg/kg). To evaluate whether the effects of SC-560 and rofecoxib on the protective activity of 20% ethanol are additive, rats (n = 4–6 in each group) were treated with doses of SC-560 (0.2 mg/kg) and rofecoxib (0.02 mg/kg) that only partially attenuated the protective action of 20% ethanol, and the effect of concurrent treatment with these doses of SC-560 and rofecoxib was assessed. In further experiments, groups of five to nine rats received intravenous injections of U-73122 (1 mg/kg) or U-73343 (1 mg/kg), respectively, 15 min before or chelerythrine (0.7 mg/kg) or staurosporine (3 μg/kg), respectively, 10 min before instillation of 20% ethanol without or with concurrent treatment with cromakalim (0.3 mg/kg p.o. 30 min before enzyme inhibitors).

To exclude the possibility that the phospholipase C inhibitor and the protein kinase C inhibitors have effects on gastric mucosal integrity by their own, groups of four rats were treated with intravenous injections of U-73122 (1 mg/kg), chelerythrine (0.7 mg/kg), or staurosporine (3 μg/kg) without receiving additional treatments and the stomachs were examined for damage 60 min later. Furthermore, U-73122 (1 mg/kg), chelerythrine (0.7 mg/kg), or staurosporine (3 μg/kg) were injected intravenously 60 min before instillation of 70% ethanol and gastric mucosal damage was assessed 5 min later (n = 4). In addition, indomethacin (20 mg/kg p.o.), SC-560 (20 mg/kg p.o.) and rofecoxib (1 mg/kg p.o.) were tested for a possible ulcerogenic activity or interference with the damaging effect of 70% ethanol (n = 5). To ascertain the selectivity of COX inhibitors rats (n = 5 in each group) were treated orally with SC-560 (20 mg/kg) or rofecoxib (1 mg/kg) and challenged with 70% ethanol 60 min later. After further 5 min, blood was collected by cardiac puncture and incubated for 60 min at 37°C. Release of thromboxane (TX) B₂ into serum was assessed using radioimmunoassay as described previously (Gretzer et al., 2001).

16,16-Dimethyl-PGE₂ was dissolved in 70% ethanol (1 mg/ml). All drugs administered orally were suspended in 0.25% methylcellulose and were given in a volume of 2.5 ml/kg. U-73122, U-73343, chelerythrine, and staurosporine were dissolved in 500 μl of 1% bovine serum albumin in saline containing 6 μl of dimethyl sulfoxide and injected intravenously. Cromakalim was dissolved in 30% ethanol and further diluted with distilled water. Cromakalim was administered orally in a volume of 2.5 ml/kg, and the final concentration of ethanol was 6%. All rats received either active agents or the corresponding vehicle so that the background volumes and solvents were identical in all groups. Vehicles for all drugs and their combinations were tested in groups of four to six rats for a possible interference with the damaging effect of 70% ethanol. The doses of the drugs used were selected from pilot experiments performed to establish the maximal effects. A flow sheet of the experimental procedures used is shown in Fig. 1.

Statistical Analysis. Results are expressed as mean ± S.E.M. of n values. Comparisons between groups were made using the Wilcoxon rank test for nonparametric data. A p value of <0.05 was considered significant.

Results

Control Experiments. Sixty minutes after oral administration, SC-560 (20 mg/kg) or rofecoxib (1 mg/kg) did not cause injury to the gastric mucosa in the absence of 70% ethanol, whereas indomethacin (20 mg/kg) had negligible effects (lesion index 4 ± 1.4). Pretreatment (60 min) with indomethacin (20 mg/kg) was previously found to increase 70% ethanol-induced damage (lesion index 45 ± 0.8 versus 39 ± 0.9 in vehicle-treated rats; p < 0.001) (Peskar et al., 2002). Oral pretreatment with SC-560 (20 mg/kg) or rofecoxib (1 mg/kg) did not aggravate the injurious effect of 70% ethanol (lesion index 38 ± 1.7 and 40 ± 3.2, respectively, versus 39 ± 1.5 in vehicle-treated control rats). We have previously shown that neither the solvent for cromakalim nor the solvents for the COX inhibitors and the protective agents interfere with the damaging effect of 70% ethanol (Peskar et al., 2002). Treatment with cromakalim (0.3 and 0.5 mg/kg p.o., 90 min before 70% ethanol) did not modify the injury caused by 70% ethanol (Peskar et al., 2002). Neither the phospholipase C inhibitor U-73122 (1 mg/kg i.v.) nor the protein kinase C inhibitors chelerythrine (0.7 mg/kg i.v.) and staurosporine (3 μg/kg i.v.) damaged the gastric mucosa in rats not treated with 70% ethanol. Furthermore, neither intravenous injections of U-73122 (1 mg/kg) nor of chelerythrine (0.7 mg/kg) nor staurosporine (3 μg/kg) modified the injury caused by instillation of 70% ethanol (lesion index 30 ± 2.8, 35 ± 0.2, and 34 ± 0.6, respectively, versus 35 ± 4.2 in rats treated with 70% ethanol in the absence of drugs).

Treatment with rofecoxib (1 mg/kg p.o.) before challenge with 70% ethanol did not inhibit platelet TXB₂ formation (46 ± 24 versus 39 ± 10 ng/ml in vehicle-treated rats), indicating lack of interference with COX-1 under the experimental conditions used. In contrast, platelet TXB₂ formation was reduced to undetectable levels after oral administration of SC-560 (20 mg/kg) and challenge with 70% ethanol.

Effects of Cyclooxygenase, Phospholipase C, and Protein Kinase C Inhibitors on 16,16-Dimethyl-PGE₂-Induced Protection and the Influence of Cromakalim. As shown in Fig. 2, oral administration of 16,16-dimethyl-PGE₂ significantly reduced gastric mucosal damage caused by subsequent challenge with 1 ml of 70% ethanol. Oral pretreatment with the nonselective COX inhibitor indomethacin (20 mg/kg), the selective COX-1 inhibitor SC-560 (20 mg/kg), or the selective COX-2 inhibitor rofecoxib (1 mg/kg) significantly antagonized the protective effect of 16,16-dimethyl-PGE₂. Pretreatment with cromakalim (0.3–0.5 mg/kg p.o.) reversed the attenuation of 16,16-dimethyl-PGE₂-induced protection by indomethacin, SC-560, and rofecoxib (Fig. 2).

The maximal effect of rofecoxib was observed with the dose of 1 mg/kg (lesion index 27 ± 3 compared with 5 ± 1 in rats treated with 16,16-dimethyl-PGE₂ alone; p < 0.001), whereas the high dose of 20 mg/kg caused much less, but still significant (p < 0.01) reversal of PG-induced protection (lesion index 10 ± 1).

The inhibition of the protective effect of 16,16-dimethyl-PGE₂ elicited by indomethacin (20 mg/kg) depended on the dose of the PG used. Thus, whereas substantial attenuation of protection was found when 16,16-dimethyl-PGE₂ was administered at doses of 75 or 150 ng/kg, indomethacin had no significant effect on the protection induced by 300 ng/kg PG (Fig. 3).

As shown in Fig. 4, pretreatment with the phospholipase C inhibitor U-73122 (1 mg/kg i.v.) significantly antagonized the protective effect of 16,16-dimethyl-PGE₂. The inactive analog U-73343 (1 mg/kg i.v.) did not affect the protection observed after treatment with the PG. The protein kinase C inhibitors chelerythrine (0.7 mg/kg i.v.) and staurosporine (1 mg/kg i.v.) likewise inhibited the protective effect of 16,16-dimethyl-PGE₂. Pretreatment with cromakalim (0.3 mg/kg p.o.) did not reverse the inhibitory effect of U-73122, chelerythrine, and staurosporine on the protection elicited by 16,16-dimethyl-PGE₂.

Effects of Cyclooxygenase, Phospholipase C, and Protein Kinase C Inhibitors on 20% Ethanol-Induced Protection and the Influence of Cromakalim. As shown in Fig. 5, instillation of 20% ethanol significantly reduced the lesion index observed in rats challenged with 70% ethanol. The protective effect of 20% ethanol was reversed by oral pretreatment with indomethacin (10 mg/kg), SC-560 (20 mg/kg), or rofecoxib (0.2 mg/kg). Concurrent treatment with cromakalim (0.3–0.5 mg/kg p.o.) significantly attenuated the inhibition of 20% ethanol-induced protection by indomethacin, SC-560, and rofecoxib.

Additional experiments demonstrated that the attenuation of 20% ethanol-induced protection after inhibition of COX-1 and COX-2 is additive. As shown in Fig. 6, oral pretreatment with submaximally effective doses of SC-560 (0.2 mg/kg) and rofecoxib (0.02 mg/kg) resulted in a lesion index of 20 ± 0.7 and 21 ± 1.8, respectively. Concurrent pretreatment with these doses of SC-560 and rofecoxib increased the lesion index to 39 ± 1.7.

Figure 7 demonstrates that pretreatment with the phospholipase C inhibitor U-73122 (1 mg/kg i.v.) reverses the protective effect of 20% ethanol. In contrast, the inactive analog U-73343 (1 mg/kg i.v.) did not modify the protective effect of 20% ethanol. The protein kinase C inhibitors chelerythrine (0.7 mg/kg i.v.) or staurosporine (1 mg/kg i.v.) inhibited the protective effect of 20% ethanol to the same extent. Pretreatment with cromakalim (0.3 mg/kg p.o.) did not reverse the inhibitory effect of U-73122, chelerythrine, or staurosporine on 20% ethanol-induced protection.

Discussion

The present results show for the first time that indomethacin not only inhibits adaptive gastroprotection induced by pretreatment with 20% ethanol (Robert et al., 1983; Gretzer et al., 1998) but also decreases protection against 70% ethanol conferred by exogenous PG. Obviously, endogenous PG released by 70% ethanol via the COX-1 and COX-2 pathway add to the gastroprotection observed under these conditions. This view is supported by the fact that indomethacin significantly increases gastric mucosal damage induced by 70% ethanol (Peskar et al., 2002), as well as by the protective effect of a high dose of exogenous PG, which overcomes the indomethacin effect.

The reversal of the protective activity of exogenous PG evoked by the COX-isoenzyme specific inhibitors SC-560 (Smith et al., 1998) and rofecoxib (Chan et al., 1999) suggests that PG released by 70% ethanol are generated by both COX-1 and COX-2. Similarly, adaptive gastroprotection induced by 20% ethanol can be inhibited by both SC-560 and rofecoxib.

Although rofecoxib had no effect on COX-1-mediated platelet TXB₂ release, previous studies in rats have shown that this compound reduces carrageenan-induced paw edema (ID₅₀ = 1.5 mg/kg) (Chan et al., 1999). Furthermore, rofecoxib (3 mg/kg) inhibited COX-2-mediated inflammatory PGE₂ release into s.c. carrageenan-soaked sponges by 88% (data not shown). In contrast, platelet TXB₂ formation was abolished after oral administration of SC-560 (20 mg/kg) and challenge with 70% ethanol as consequence of COX-1 inhibition. The results of these experiments confirm previous findings observed in rats not challenged with 70% ethanol (Gretzer et al., 2001).

Doses of indomethacin and SC-560 below 20 mg/kg showed reduced inhibitory potency on the protective effect of exogenous and endogenous PG (data not shown). In contrast, a low dose (1 mg/kg) of rofecoxib inhibited the protection conferred by 16,16-dimethyl-PGE₂ more effectively than a high dose (20 mg/kg). This is in contrast to the injurious action of rofecoxib in rats in the absence of protective agents where 20 mg/kg of the drug induced a maximal effect (Gretzer et al., 2001). Further studies are necessary to clarify the mechanisms underlying this discrepancy.

The COX-2 involved in adaptive gastroprotection seems to be constitutive, because pretreatment with dexamethasone has no effect under these conditions (Gretzer et al., 1998). Because under basal conditions COX-2 contributes only minimally to gastric mucosal PG formation (Gretzer et al., 1998) and COX-1 and COX-2 are most probably localized unevenly within the gastric mucosa, the high potency and efficacy of COX-2 inhibitors against gastroprotection induced by exogenous or endogenous PG is remarkable. However, the possibility cannot be completely excluded that rofecoxib acts directly or indirectly, e.g., by modulation of other protective systems such as nitric oxide or afferent neurons, on K_ATP channels. Further investigations are necessary to elucidate the exact mechanism of action of the isoenzyme-specific COX inhibitors. It should be pointed out, however, that under the conditions used PG generated by one COX isoenzyme obviously cannot compensate the injurious effects of an inhibitor of the other COX isoenzyme. Interestingly, it has been demonstrated that intact COX-1 cannot compensate COX-2 deficiency or long-term inhibition as to small intestinal pathology in mice (Sigthorsson et al., 2002). Our present results are in contrast to the damaging effect of COX-1 and COX-2 inhibitors in rats not challenged with 70% ethanol. In this experimental setting with longer observation periods, simultaneous inhibition of both COX isoenzymes is required to induce gastric mucosal damage (Wallace et al., 2000; Gretzer et al., 2001; Tanaka et al., 2001).

We have previously shown that the activator of K_ATP channels, cromakalim, counteracts the inhibitory effect of the K_ATP channel blocker glibenclamide on the protective activity of 16,16-dimethyl-PGE₂ and 20% ethanol (Peskar et al., 2002). The data of the present study demonstrate that cromakalim in addition reverses the antagonistic effect of COX inhibitors on the protection induced by exogenous and endogenous PG, although in the dosage used it does not have a protective action on its own in the absence of COX inhibition (Peskar et al., 2002). This result further supports the view that the mechanism of action of PG involves the activation of K_ATP channels (Peskar et al., 2002). Interestingly, it has been demonstrated that the potency of cromakalim is significantly increased by COX inhibition (Armstead, 2001; Peskar et al., 2002), possibly because oxygen radicals generated by the enzyme interfere with the cromakalim effect. It is remarkable, however, that cromakalim not only antagonizes the effects of indomethacin and SC-560, which significantly inhibit ex vivo PG biosynthesis by gastric mucosal tissues, but also the effect of a selective COX-2 inhibitor, which does not cause measurable effects on PG biosynthesis (Gretzer et al., 1998). Clearly, cromakalim does not antagonize the effects of the COX inhibitors by stimulation of PG biosynthesis (Peskar et al., 2002), but obviously can replace PG as activators of K_ATP channels.

Previous studies have shown that the protective effect of PG on gastric mucosa is independent of the cAMP system in vivo (Moron et al., 1982) and in vitro (Terano et al., 1984, 1987; Konda et al., 1990). On the other hand, it has been demonstrated that guinea pig chief cells in vitro are protected against ethanol and taurocholate injury by PG presumably through the activation of the diacylglycerol/protein kinase C signaling pathway (Konda et al., 1990). The importance of this intracellular pathway for PG-mediated gastroprotection under in vivo conditions has, however, so far not been investigated. Our data show that inhibition of phosphatidylinositol-specific phospholipase C or of protein kinase C significantly, although not completely, antagonizes the protective effects of exogenous and endogenous PG. Although staurosporine is a potent inhibitor of protein kinase C and in addition inhibits the activity of a variety of other protein kinases (Herbert et al., 1990), chelerythrine has been described as highly specific inhibitor of protein kinase C without affecting other protein kinases (Herbert et al., 1990). The finding that chelerythrine interacts with cyclic nucleotide phosphodiesterases (Eckly-Michel et al., 1997) cannot explain the antagonism of protection conferred by 16,16-dimethyl-PGE₂ and 20% ethanol because PG-induced protection is independent of the cAMP system (Moron et al., 1982; Terano et al., 1984, 1987; Konda et al., 1990). Furthermore, the effects of inhibitors of phospholipase C or protein kinase C are not due to impairment of PG biosynthesis, because administration of U-73122, chelerythrine, or staurosporine did not reduce PG release from gastric tissue ex vivo (data not shown). It has been demonstrated that the maleimide moiety of the phospholipase C inhibitor U-73122 has the capacity to inhibit K_ATP channels and thus to reduce the vasodilator response to cromakalim in vitro (Fulton et al., 1996). However, under our experimental in vivo conditions no functional interaction between U-73122 and cromakalim could be demonstrated. This result excludes the possibility that U-73122 acts like glibenclamide, the inhibitory effects of which on PG-mediated gastroprotection have been demonstrated previously (Peskar et al., 2002). These glibenclamide effects are completely antagonized by cromakalim (Peskar et al., 2002). As to the importance of protein kinase C for protective PG effects it has been demonstrated previously that this enzyme mediates a major part of PG-induced cardioprotection (Thiemermann and Zacharowski, 2000).

In summary, endogenous PG obviously contribute to the protective activity of exogenous PG. Gastroprotection by PG involves phospholipase C, protein kinase C, and K_ATP channels. Activation of K_ATP channels does not exert protection when the activity of phospholipase C or protein kinase C is suppressed.