Induction of cyclooxygenase-2 expression by peroxisome proliferators and non-tetradecanoylphorbol 12,13-myristate-type tumor promoters in immortalized mouse liver cells.

Increased expression of cyclooxygenase-2 (COX-2), the rate-limiting enzyme in prostaglandin synthesis, has been associated with growth regulation and carcinogenesis in several systems. COX-2 is known to be induced by cytokines and the skin tumor promoter 12-tetradecanoylphorbol-13-myristate (TPA). In the present study, we investigated the effects of several non-TPA-type tumor promoters on COX-2 expression in immortalized mouse liver cells. Specifically, we tested peroxisome proliferators (PPs), which are rodent liver tumor promoters that cause gross alterations in cellular lipid metabolism, the rodent liver tumor promoter phenobarbital, and the skin tumor promoters okadaic acid and thapsigargin. The PPs Wy-14643, mono-ethylhexyl phthalate, clofibrate, ciprofibrate ethyl ester, and eicosatetraynoic acid each caused large increases in COX-2 mRNA and protein, with maximal expression seen approximately 10 h after treatment of quiescent cells. COX-2 expression was also induced by thapsigargin, okadaic acid, and calcium ionophore A23187, but not by phenobarbital or the steroid PP dehydroepiandrosterone sulfate. Induction of COX-2 expression generally resulted in increased synthesis of prostaglandin E2 (PGE2). However, the PPs caused little or no increase in PGE2 levels, and they inhibited serum-induced PGE2 synthesis. Unlike non-steroidal anti-inflammatory drugs, the PPs do not directly inhibit cyclooxygenase enzyme activity in vitro Thus, PPs regulate prostaglandin metabolism via both positive (COX-2 induction) and inhibitory mechanisms. In summary, the strong induction of COX-2 expression by PPs, thapsigargin, and okadaic acid suggests a possible role for COX-2 in the growth regulatory activity of these non-TPA-type tumor promoters.

Cyclooxygenases catalyze the first, rate-limiting steps in the conversion of arachidonic acid into prostaglandins and thromboxanes. Two distinct isoforms are known: cyclooxygenase-1 (COX-1), 1 which is constitutively expressed in a wide variety of tissues and is thought to be involved in so-called "housekeeping" functions, and cyclooxygenase-2 (COX-2), which is undetectable in most normal tissues but is strongly induced at sites of inflammation (1). In addition to inflammation, elevated COX-2 expression has been associated with cell growth regulation and carcinogenesis. COX-2 is an immediate-early gene inducible by cytokines, growth factors, and the tumor promoter 12-tetradecanoylphorbol-13-myristate (TPA) (2)(3)(4)(5). COX-2 expression and prostaglandin E 2 (PGE 2 ) levels are increased in colon (1,6) and skin (7) tumors, and non-steroidal anti-inflammatory drugs (NSAIDS) inhibit carcinogenesis in several systems (1,(7)(8). Transformation of epithelial cells with the ras or src oncogenes causes elevated COX-2 expression and increased PGE 2 synthesis (9). Overexpression of COX-2 in epithelial cells causes growth arrest (10) and inhibits apoptosis (11), directly implicating COX-2 in cell growth regulation.
Despite the observations that COX-2 expression is increased in certain tumors and that it is inducible by phorbol ester TPA, the effects of non-TPA-type tumor promoters on COX-2 have not been explored. Previous observations suggest that COX-2 might be regulated by thapsigargin, a Ca 2ϩ -ATPase inhibitor, and okadaic acid, a phosphatase inhibitor, which are both potent skin tumor promoters in rodents. COX-2 expression is increased in skin tumors promoted by TPA, and skin tumorigenesis can be inhibited by NSAIDS (7). Furthermore, both thapsigargin and okadaic acid cause increased prostaglandin synthesis in vitro (12)(13)(14)(15).
COX-2 might also be regulated by a diverse group of rodent liver tumor promoters known as peroxisome proliferators (PPs). PPs cause an increase in the size and number of hepatic peroxisomes and enhanced expression of enzymes involved in fatty acid catabolism. Most PPs, or their direct metabolites, are structurally similar to fatty acids in being amphipathic carboxylic acids (16). Regulation of lipid metabolism and peroxisome proliferation by PPs occurs via activation of peroxisome-proliferator activated receptors (PPARs), members of the steroid receptor superfamily (17,18). PPs might activate PPARs indirectly by increasing the level of an endogenous ligand. PPARs are activated by fatty acids (17,19), including the eicosanoids arachidonic acid, its synthetic analog eicosatetraynoic acid (ETYA), 8(S)-hydroxyeicosatetraenoic acid, and several prostaglandins (20 -22). Recently, 15-deoxy-⌬ 12,14 -prostaglandin J2 was identified as a direct-binding ligand for PPAR␥ (21,22).
The mechanism of rodent liver tumor promotion by PPs is unclear, and it is likely that peroxisome proliferation is either incidental or not the sole causative factor (23)(24)(25). Mounting evidence suggests that PPs have growth regulatory activities that are independent of peroxisome proliferation (26,27) and that differentially affect preneoplastic liver cells versus normal hepatocytes (28 -30). For example, higher doses of the PP Wy-14643 are required for stimulation of liver cell proliferation than for induction of peroxisome proliferation in vivo (27). PPs have been shown to enhance proto-oncogene expression (31)(32)(33) and protein kinase C activity (34) in vivo. In mouse liver cell lines in vitro, PPs are potent activators of signal transduction pathways involving mitogen-activated protein kinases, 2 they strongly induce the expression of immediate-early proto-oncogenes (35,36), and they regulate cell cycle progression during both early G 1 and late G 1 (36). These growth regulatory effects of PPs in vitro are independent of peroxisome proliferation (35,36).
Fatty acids and other lipids play a major role in growth regulatory signal transduction (37)(38)(39), and it is possible that PPs function as tumor promoters by either mimicking, or modulating the metabolism of, growth regulatory lipids. Since COX-2 is an immediate-early gene involved in growth regulatory fatty acid metabolism, we investigated the effects of PPs on COX-2 gene expression and PGE 2 synthesis in immortalized mouse liver cells. The PPs were compared with another liver tumor promoter, phenobarbital, which is thought to promote via a different mechanism than PPs. In addition, we examined thapsigargin and okadaic acid, since the effects of these non-TPA-type tumor promoters on COX-2 gene expression have not been previously investigated.
Cell Culture-Two immortalized liver cell lines, BNL-CL.2 and ML-457, were used. ML-457 was derived from a CD-1 mouse liver tumor but is not tumorigenic in nude mice. 3 BNL-CL.2 is a non-tumorigenic cell line derived from the liver of a Balb/c mouse embryo (40). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 50 g/ml gentamycin and 10% fetal bovine serum. Before stimulation with peroxisome proliferators or other agents, confluent cultures were made quiescent by incubating them for 3 days in medium containing 0.1% serum. The medium was removed, and the quiescent cells were then treated with various agents diluted in fresh medium containing 0.1% serum and 5 mM HEPES, pH 7.1.
RNA Isolation and Northern Blot Analysis-At several time points after addition of PPs or other agents, cells were washed with cold phosphate-buffered saline and lysed by the addition of a 4 M guanidinium isothiocyanate solution, and total RNA was purified by phenolchloroform extraction and isopropyl alcohol precipitation as described by Chomczynski and Sacchi (41). Total RNA preparations were highly pure, with 260 nm/280 nm absorbance ratios Ն2.0. RNA was quantitated by absorbance at 260 nm, and exactly 10 g of each sample was electrophoresed in 1.4% agarose, 6% formaldehyde gels. After electrophoresis, we ensured that equal amounts of RNA were loaded into each lane by staining the RNA with ethidium bromide and photographing the gels on a UV transilluminator. The RNA was then transferred to a nylon-supported nitrocellulose filter by capillary blotting. After prehybridization for 2-4 h without radioactive probe, the blots were hybridized overnight at 42°C with a 32 P-nick translated COX-2 cDNA probe (5 ϫ 10 5 cpm/ml) in 50% formamide, 1 M NaCl, 1% sodium dodecyl sulfate, 10% dextran sulfate, and 100 g/ml denatured salmon sperm DNA. The blots were washed in 0.1 ϫ SSC, 0.5% sodium dodecyl sulfate and then exposed to Kodak XAR-5 film. All significant results were confirmed in two or more separate cell culture experiments.
Western Blot Analysis-At several time points after addition of PPs or other agents, cells were washed with cold phosphate-buffered saline and lysed in SDS lysis buffer (1% sodium dodecyl sulfate, 1 mM EDTA, 1 mM EGTA, 20 mM Tris-HCl, pH 7.5). Protein concentration was determined using the DC Protein Assay, a commercial modification of the Lowry method (Bio-Rad). Equal amounts of each sample (usually 40 g) were loaded onto a denaturing 4 -20% polyacrylamide gel (Integrated Separation Systems, Natick, MA). Following electrophoresis, proteins were transferred to a polyvinylidene difluoride membrane by semi-dry electroblotting. Blots were blocked in TBST solution (100 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 0.1% Tween 20) containing 5% nonfat milk, incubated for 1 h at room temperature in a 1:1000 dilution of rabbit anti-COX-2 (mouse) polyclonal antiserum (Cayman Chemical Co., Ann Arbor, MI) in TBST containing 0.25% nonfat milk, rinsed in TBST, incubated for 1 h at room temperature in a 1:5000 dilution of anti-rabbit IgG/horseradish peroxidase conjugate (Amersham Corp.) in TBST containing 0.25% nonfat milk, and then rinsed well in TBST. The COX-2 protein signal was then visualized using a chemiluminescent detection system for horseradish peroxidase activity (Amersham Corp.). To control for any possible loading or transfer artifacts during the Western blot procedure, each sample was processed on at least two duplicate gels. In addition, all significant results were confirmed in two or more separate cell culture experiments. Western analysis of COX-1 was carried out as above except using a rabbit polyclonal antiserum against COX-1, supplied by Merck Frosst Canada, Inc. (Quebec, Canada).
Radioimmunoassay for PGE 2 Synthesis-Cells were grown in 6-well or 12-well culture dishes and treated as described above. At several time points after refeeding with PPs or other agents, the conditioned medium above the cells was carefully collected and frozen at Ϫ70°C until the PGE 2 assay could be performed. To measure total cellular protein, the cell monolayer was washed with phosphate-buffered saline, lysed with SDS lysis buffer, and assayed by the DC Protein Assay (Bio-Rad). PGE 2 levels in the conditioned media were measured using an 125 I-radioimmunoassay (RIA) kit (DuPont NEN). The medium was diluted at least 1:5 in RIA assay buffer prior to analysis. All assays were performed in duplicate. The results were expressed as picograms of PGE 2 per g of protein. None of the agents tested caused significant toxicity, and total protein did not vary more than 5-10% between wells within an experiment.
For each treatment, an aliquot of medium that was not exposed to cells was tested for interference in the RIA in two ways: alone, to determine if the agent caused a higher apparent PGE 2 signal, and spiked with PGE 2 , to determine if the agent interfered with PGE 2 detection. Arachidonic acid caused severe background and thus was not tested for its effects on PGE 2 synthesis. None of the PPs or other tumor promoters tested caused significant background or interference in the RIA at the concentrations used. Medium containing serum had a low level of background signal which did not significantly affect the results (it represented 0 -5% of the induced level of PGE 2 ).
Prostaglandin G/H Synthase Assay-Prostaglandin G/H synthase activity was measured in vitro using purified COX-1 or COX-2 protein (Cayman Chemical Co.), stannous chloride to stimulate chemical conversion of prostaglandin H 2 (PGH 2 ) to prostaglandin F 2␣ (PGF 2␣ ), and an enzyme immunoassay for PGF 2␣ (Cayman Chemical Co.), essentially as recommended by the manufacturer. Assays were carried out in 2-ml volumes containing 0.1 M Tris-HCl, pH 8, 0.2 mM phenol, 1 M hematin, 25 units of COX-1 or 12.5 units of COX-2, and 20 l of test article in dimethyl sulfoxide. Following a 1-min preincubation at 37°C, prostaglandin synthesis was initiated by the addition of arachidonic acid substrate to a final concentration of 1 M. After 30 s, the reactions were quenched by the addition of 50 l of 100 mg/ml stannous chloride in 0.1 N HCl. Samples were incubated for an additional 10 min at 37°C to convert PGH 2 to PGF 2␣ and then centrifuged for 10 min at 4°C. A 20-l aliquot of the supernatant was evaporated to dryness, resuspended in enzyme immunoassay buffer, and assayed by enzyme immunoassay for PGF 2␣ (Cayman Chemical Co.).

Induction of COX-2 mRNA Expression by PPs and Other
Tumor Promoters-We first tested two well known PPs, Wy-14643, a hypolipidemic agent, and MEHP, a metabolite of a plasticizing agent, for their effects on COX-2 expression. Quiescent ML-457 cells were treated with 1 mM Wy-14643 or 1 mM MEHP, in medium containing 0.1% serum, and COX-2 expression was examined by Northern analysis at various time points. COX-2 expression was nearly undetectable in unstimulated quiescent cells and was not significantly induced by refeeding the cells with 0.1% serum (see "NA" lanes in Fig. 3).
As shown in Fig. 1, treatment with Wy-14643 and MEHP caused a strong induction of COX-2 gene expression, which peaked 8 -12 h after dosing. The time course of COX-2 gene expression was delayed and prolonged compared with other immediate-early genes, which generally peaked 1-2 h postdosing (see Ref. 36; c-jun is shown in Fig. 1 for comparison).
COX-2 gene induction by both Wy-14643 and MEHP was strongly inhibited by the presence of 1 M dexamethasone (Fig.  1), a nearly universal inhibitor of COX-2 expression which acts at the transcriptional level (26,27). Dexamethasone does not globally inhibit all immediate-early gene expression (26,27), and it did not cause a marked inhibition of c-jun expression by PPs (Fig. 1).
A large subgroup of PPs are hypolipidemic compounds known as fibrates. Like Wy-14643 and MEHP, the PPs clofibrate and ciprofibrate ethyl ester were potent inducers of COX-2 gene expression at 10 h (Fig. 2). COX-2 induction by the PPs exhibited a consistent dose-response: little expression at 100 M, moderate to strong expression at 500 M, and strong expression at 1 mM (only the 1 mM dose is shown).
Arachidonic acid (0.5 mM), the substrate for COX-2, and ETYA, a more stable synthetic analog of arachidonic acid, also act as peroxisome proliferators. ETYA (0.1 mM) induced a strong induction of COX-2 mRNA (Fig. 2), with a similar time course and intensity to the induction seen with PPs. Arachidonic caused only a weak induction of COX-2 gene expression, possibly because it is highly unstable compared to PPs and ETYA. The steroid DHEA-S, which is unusual among PPs in that it does not resemble a fatty acid (42), did not induce COX-2 expression (not shown).
Unlike the PPs, TPA (100 ng/ml) caused a rapid and transient induction of COX-2 mRNA, peaking 1 h after dosing (Fig.  3). Stimulation with 20% serum induced strong COX-2 expression, which peaked at about 1-4 h and declined by 10 h. Okadaic acid (50 ng/ml) caused a moderate induction of COX-2 mRNA at 10 h. Calcium ionophore A23187 (1 g/ml) caused a strong and prolonged induction of COX-2 mRNA, that was maximal at the 4-h time point. Thapsigargin (1 g/ml) was a strong inducer of COX-2 and, like the PPs, caused maximal expression at about 10 h. Phenobarbital, at doses up to 5 mM, had no effect on COX-2 expression.
The induction of COX-2 expression by PPs, thapsigargin, and okadaic acid was slower and more prolonged than the induction of other immediate-early genes (36) (see c-jun in Fig. 1). However, COX-2 expression was an immediate-early response, not requiring protein synthesis, since it was not inhibited by the protein synthesis inhibitor cycloheximide (not shown). Other known inducers of COX-2, including interleukin-1␣, exhibit a similar prolonged time course of COX-2 expression (4, 43). COX-2 expression is regulated at both the transcriptional and post-transcriptional level, and mRNA stabilization has been shown to contribute to the prolonged expression (43).

Induction of COX-2 Protein
Expression-To determine whether the observed increases in COX-2 mRNA corresponded with increased expression of COX-2 protein, Western blot analyses were performed using a polyclonal antiserum specific to COX-2. Fig. 4 shows COX-2 protein expression at 10 h postdosing. The PPs Wy-14643, MEHP, clofibrate, and ciprofibrate ethyl ester all induced COX-2 protein, with strongest expression seen at the 10-h time point shown. Thapsigargin, A23187, and okadaic acid also induced strong COX-2 protein expression at 10 h. Weak induction of COX-2 protein expression was observed with serum and TPA at the 10-h time point shown, and although the mRNA expression peaked much earlier, the protein levels were only slightly greater at 1-or 4-h time points (not shown). Arachidonic acid, which was only a weak inducer of COX-2 mRNA, did not significantly increase COX-2 protein in these experiments. Phenobarbital did not induce COX-2 mRNA or protein. Qualitatively, all agents that induced strong COX-2 mRNA expression also induced COX-2 protein expression, although the relative levels of protein versus mRNA varied between inducers.
For comparison, COX-1 protein expression is also shown in Fig. 4. As expected, COX-1 was constitutively expressed in ML-457 cells and its expression was not significantly increased by any of the agents tested.
Effects on PGE 2 Synthesis-Cyclooxygenase is the rate-limiting enzyme in prostaglandin synthesis. Prostaglandin levels, particularly PGE 2 , are often used as a measurement of cyclooxygenase activity. To investigate whether the increase in COX-2 mRNA and protein expression corresponded with increased cyclooxygenase activity, we measured PGE 2 levels in the medium following treatment (refeeding with test agents) of quiescent cells for 10 h. Serum, thapsigargin, okadaic acid, and A23187 consistently caused large increases in PGE 2 levels, often more than 10-fold greater than in control cells (Fig. 5). Note that serum, a relatively weak inducer of COX-2 protein, was one of the most potent inducers of PGE 2 synthesis. TPA consistently caused a modest increase in PGE 2 levels. The PPs caused little or no increase in PGE 2 synthesis, and a consistent dose-response was not obtained (see below). DHEA-S and phenobarbital had no effect on PGE 2 synthesis (not shown).
Since the PPs were strong inducers of COX-2 expression but relatively weak inducers of PGE 2 synthesis, and since no doseresponse was observed for PGE 2 synthesis, it appeared that the PPs may have a second, inhibitory activity that partially counteracts the increase in COX-2 protein expression. Therefore, we tested the effects of PPs on serum-induced PGE 2 synthesis. Serum-induced PGE 2 synthesis can be almost completely inhibited by 10 M indomethacin, an inhibitor of cyclooxygenases FIG. 3. Induction of COX-2 mRNA expression by other tumor promoters and growth regulators. Quiescent ML-457 cells were treated with no addition (NA), 20% serum, 100 ng/ml TPA, 5 mM phenobarbital, 50 ng/ml okadaic acid, 1 g/ml A23187, or 1 g/ml thapsigargin, for 1, 4, or 10 h. Ten-g portions of total RNA were analyzed by Northern analysis as described under "Experimental Procedures." Equal loading of total RNA between lanes was confirmed by ethidium bromide staining.

FIG. 4. Induction of COX-2 protein expression by PPs and other growth regulators.
Quiescent ML-457 cells were treated for 10 h with no addition (NA), 20% serum, 100 ng/ml TPA, 5 mM phenobarbital, 50 ng/ml okadaic acid, 1 g/ml thapsigargin, 1 g/ml A23187, 0.5 mM arachidonic acid, or 1 mM of each PP, as indicated on the figure. Western analysis was performed using an anti-COX-2 polyclonal antiserum, as described under "Experimental Procedures." For comparison, COX-1 protein levels (lower panel) were also examined.  (Fig. 6). Wy-14643, MEHP, clofibrate, and ciprofibrate-ethyl ester each caused a marked inhibition in PGE 2 synthesis when added simultaneously with 20% serum to quiescent cells. For these PPs, concentrations between 100 M and 1 mM were required for near-maximal inhibition of PGE 2 synthesis (Fig.  6). Cells treated with serum plus PPs had high levels of COX-2 expression (not shown), as expected for an additive response, and thus inhibition of serum-induced PGE 2 synthesis was not due to reduced COX-2 expression. None of the other agents tested, including thapsigargin, okadaic acid, A23187, TPA, phenobarbital, or DHEA-S exhibited this inhibitory activity on PGE 2 synthesis.
Induction of COX-2 Expression by Indomethacin-Since the concentrations of PPs which caused inhibition of serum-induced PGE 2 synthesis approached those required for strong COX-2 expression (500 M to 1 mM), it was possible that COX-2 induction was a consequence of inhibited cyclooxygenase activity. To test this possibility, we investigated the effect of indomethacin on COX-2 expression. Indomethacin caused almost complete inhibition of serum-induced PGE 2 synthesis at a concentration of 10 M (Fig. 6). However, a 10-fold higher dose, 100 M, had little effect on COX-2 mRNA or protein expression (Fig.  7). Indomethacin did induce COX-2 expression at 500 M to 1 mM, exhibiting a similar time course and dose-response as that induced by PPs. These concentrations of indomethacin are 50 -100-fold greater than required for cyclooxygenase inhibition, indicating that induction of COX-2 expression is not a consequence of inhibited cyclooxygenase activity. The fact that indomethacin at high doses behaved like the PPs in inducing COX-2 expression is not surprising given the structural similarity of many NSAIDS to peroxisome proliferators (44,45).
Effect of Wy-14643 on Prostaglandin G/H Synthase Activity in Vitro-To test whether the inhibition of serum-induced PGE 2 synthesis by PPs was due to direct inhibition of cyclooxygenase enzyme activity, we investigated the effect of Wy-14643 on the activity of purified COX-1 and COX-2 proteins in vitro. COX-1 and COX-2 catalyze the conversion of arachidonic acid into PGG 2 (via cyclooxygenase activity) and the subsequent conversion of PGG 2 into PGH 2 (via peroxidase activity). Since PGG 2 and PGH 2 are highly unstable, cyclooxygenase enzyme activity was measured by chemical conversion of PGH 2 into PGF 2␣ using stannous chloride, followed by an enzyme immunoassay for PGF 2␣ . As shown in Fig. 8, Wy-14643 did not inhibit either COX-1 or COX-2 enzyme activity in vitro, in fact a slight enhancement was observed. Under identical conditions, ibuprofen caused nearly complete inhibition of cyclooxygenase enzyme activity (not shown). These results indicate that the inhibitory effect of PPs on serum-induced prostaglandin synthesis is not due to direct inhibition of the cyclooxygenase enzyme activity, as is the case for NSAIDS.
Induction of COX-2 Expression in Other Cells-To determine if the induction of COX-2 expression by PPs was idiosyncratic to the ML-457 liver cell line, we also tested BNL-CL.2 cells, an immortalized liver cell line isolated from a mouse embryo (40). Nearly identical results were obtained in the BNL-CL.2 cells: Wy-14643, clofibrate, MEHP, thapsigargin, A23187, okadaic acid, and indomethacin all induced COX-2 expression, with similar time courses and dose-responses to those observed in ML457 cells (data not shown). In contrast to the results obtained in these immortalized mouse liver cells, no COX-2 expression could be detected in primary rat hepatocytes in vitro or in mouse liver in vivo, even after treatment with PPs (not shown). This was expected since COX-2 is generally not expressed in normal, differentiated cells. DISCUSSION COX-2, an immediate-early gene induced by cytokines and phorbol ester TPA, is an important factor in inflammation and the key target of anti-inflammatory drugs. Increasing evidence suggests that COX-2 also functions in cell growth regulation and carcinogenesis. In the present study, we found that PPs, thapsigargin, and okadaic acid act as potent inducers of COX-2 in immortalized mouse liver cells, suggesting that COX-2 might play a role in growth regulation by these non-TPA-type tumor promoters.
Stimulation of COX-2 gene expression by thapsigargin and okadaic acid is consistent with previous reports of increased prostaglandin production in response to these agents (12)(13)(14)(15). COX-2 expression has been shown to be increased in rodent skin papillomas and carcinomas promoted by TPA, and rodent skin tumorigenesis can be inhibited by NSAIDS (7). Thus, the ability of thapsigargin and okadaic acid to induce COX-2 expression could be relevant to their tumor promoting activity.
The induction of COX-2 expression by PPs is consistent with their ability to induce the expression of other immediate-early genes (36). The steroid DHEA-S is unusual among PPs in that it does not resemble a fatty acid, activate PPARs (42), or cause the characteristic burst of hepatocellular proliferation following short-term treatment in vivo (46). Consistent with these differences from other PPs, DHEA-S does not induce immediate-early gene expression (36) and does not induce COX-2. Phenobarbital, a liver tumor promoter and microsomal enzyme inducer in rodents, also does not induce COX-2 gene expression, consistent with the idea that phenobarbital promotes liver tumors by a different mechanism than PPs (47,48).
COX-2 is unusual among immediate-early genes in that its expression can be induced for prolonged times by certain agents. For example, while the induction of COX-2 expression by TPA generally peaks at about 1 h after dosing (4), the induction of COX-2 expression by interleukin-1␣ peaks 6 -24 h after dosing (43). In both cases, however, COX-2 is an immediate-early response in that it does not require protein synthesis. In our study, the PPs, thapsigargin, and okadaic acid each caused a prolonged induction of COX-2, and the induction was not inhibited by the protein synthesis inhibitor cycloheximide. Ristimäki et al. (43) previously demonstrated that sustained induction of COX-2 involves post-transcriptional mRNA stabilization.
The regulation of prostaglandin synthesis is complex, as evidenced by serum causing only a small increase in COX-2 protein expression but a pronounced stimulation of PGE 2 synthesis. Conversely, PPs cause a large increase in COX-2 expression but only a modest increase in PGE 2 synthesis. Furthermore, the PPs inhibit serum-induced PGE 2 synthesis. Thus, while PPs strongly induce COX-2 expression, their effects on prostaglandin levels are complicated by other activities. Unlike NSAIDS, Wy-14643 does not directly inhibit COX-1 or COX-2 enzyme activity in vitro. Thus, PPs might indirectly inhibit serum-induced PGE 2 synthesis by modulating other enzymes or signal transduction pathways that regulate prostaglandin metabolism.
The induction of COX-2 expression by PPs is unlikely to be a feedback response to their inhibitory activity on prostaglandin synthesis. The NSAID indomethacin, itself a peroxisome proliferator (44,45), completely inhibited PGE 2 synthesis at a dose of 10 M but required 500-1000 M to induce strong COX-2 expression. Induction of COX-2 by NSAIDS has been reported previously (49) and is consistent with the structural similarity between PPs and NSAIDS (44,45). Thus, COX-2 induction by PPs is probably not a consequence of inhibited prostaglandin synthesis, but rather is probably due to the activation of immediate-early signaling pathways (36).
It is unlikely that COX-2 plays a major role in peroxisome proliferation. COX-2 is generally not expressed in normal tissue, whereas peroxisome proliferation is a phenotype of normal differentiated hepatocytes (16). Note that dexamethasone induces differentiation in many cell types and stimulates PPAR expression in primary hepatocytes (50), whereas it is a global inhibitor of COX-2 expression. Lipid metabolism by COX-2 produces mediators of cell growth and inflammation, whereas lipid metabolism during peroxisome proliferation is predominantly catabolic. The preneoplastic cells which respond to the growth stimulatory activity of PPs are probably quite different from the normal differentiated hepatocytes which undergo peroxisome proliferation (28 -30). In fact, immortalized liver cell lines, including the cells used here, generally exhibit diminished peroxisomal gene expression (51). In addition, the doses of PPs required for stimulation of COX-2 expression in ML-457 cells are about 10-fold higher than those required for peroxisomal gene expression in hepatocytes. Therefore, it is unlikely that COX-2 would play a role in peroxisome proliferation, and it is not surprising that we were unable to detect COX-2 expression in primary hepatocytes or liver.
While COX-2 is probably not involved in peroxisome proliferation, the strong induction of COX-2 expression in immortalized liver cells raises the possibility that it could be involved in cell growth regulation by PPs. At doses comparable to those required for COX-2 expression, PPs induce other immediateearly genes, including the proto-oncogenes c-fos, c-jun, junB, and egr-1 (35,36), they mobilize intracellular Ca 2ϩ (52), and they regulate cell-cycle progression (36). Recently, we found that PPs activate the Erk1 and Erk2 mitogen-activated protein kinases with a similar dose-response. 2 Thus, COX-2 induction may be an integral part of the growth regulatory response to PPs. The relatively high doses of PPs required for growth regulation in vitro are consistent with the observation in vivo that a much higher dose of Wy-14643 is required for stimulation of hepatocellular proliferation than is necessary for peroxisome proliferation (27).
Although involvement of COX-2 in liver tumor promotion is highly speculative at present, it is supported by the ability of NSAIDS to inhibit 2-acetylaminofluorene-induced hepatocarcinogenesis (53). PPs have been found to cause a decrease in hepatic prostaglandin concentrations (54), but their effect on prostaglandin synthesis and COX-2 expression in preneoplastic cells and liver tumors remains to be determined.
In summary, we found that COX-2 gene expression is strongly induced by peroxisome-proliferating liver tumor promoters and by the skin tumor promoters thapsigargin and okadaic acid. Elevated COX-2 gene expression and prostaglandin levels have been associated with several types of carcinogenesis, and COX-2 has been directly implicated in the regulation of cell proliferation and apoptosis. Thus, COX-2 could play an important role in growth regulation by these non-TPA-type tumor promoters.