Novel Regulation of Cyclooxygenase-2 Expression and Prostaglandin E₂ Production by IFN-γ in Human Macrophages¹

Prostaglandins are potent mediators of inflammation and immunomodulators. PGE₂ is one of the main PGs released in high amounts by macrophages (1) in inflamed tissues (2). It inhibits the production of Th1-type cytokines (IFN-γ and IL-2), favoring the production of Th2-type cytokines (IL-4, IL-5) by human lymphocytes (3, 4). PGE₂ acts in an autocrine manner to inhibit macrophage accessory and effector cell functions, such as the expression of MHC molecules (5) or the production of TNF-α (6).

The rate-limiting step in the synthesis of PGs is the expression of the enzyme cyclooxygenase (COX).³ This enzyme catalyzes the conversion of arachidonic acid to PGH₂, the precursor of biologically active PGs. However, the enzyme is rapidly inactivated after catalytic activity. Therefore, to achieve a sustained production of PGs, new enzyme must be synthesized (7). Two isoforms of COX have been described: COX-1, generally referred to as the “constitutive” isoform and COX-2, which is readily induced after stimulation with cytokines, mitogens, and phorbol esters (8, 9, 10). COX-2 is predominantly expressed in cells involved in inflammatory reactions, such as macrophages, endothelial cells, and fibroblasts (8, 11, 12). COX isoforms are encoded by two different genes, thus being independently regulated. Several proinflammatory cytokines stimulate the expression of COX-2. IL-1α has been shown to induce a transient transcription of the COX-2 gene in human endothelial cells (9). IL-1β enhances COX-2 mRNA expression in human monocytes (8) and stabilizes COX-2 transcripts in renal mesangial cells (13). TNF-α induces COX-2 expression in osteoblasts and articular chondrocytes (14, 15). Interestingly, TNF-α appeared to be a less potent COX-2 inducer than IL-1β in osteoblasts and mesangial cells (14, 16). Similarly, human monocytes failed to express COX-2 protein upon stimulation with TNF-α (17). We have shown that only IFN-γ-primed human macrophages express COX-2 mRNA when TNF-α is the triggering signal (18). TGF-β is able to moderately induce COX-2 expression in human lung fibroblasts (12) and it can enhance the expression of COX-2 by phorbol esters in murine fibroblasts (19). This effect is apparently cell type dependent, since it was shown that TGF-β attenuates the expression of COX-2 induced by endotoxin in murine macrophages (20). Down-regulatory activities were also observed with IL-4 and IL-10, which inhibited PG production and COX-2 expression in human monocytes stimulated with LPS as a result of transcriptional and posttranscriptional regulation (21).

IFN-γ produced by T lymphocytes and NK cells plays a major role in the biology of monocytes/macrophages as the primary macrophage-activating factor (22). IFN-γ stimulates tumor cell cytotoxicity (23), antimicrobial activity (24), and Ag processing and presentation through increased expression of MHC molecules (25). However, the effect of this cytokine on COX-2 expression is still controversial. Studies done before the description of COX-2 have shown that IFN-γ could inhibit PGE₂ production by human monocytes and murine peritoneal macrophages stimulated with IL-1β (26) or zymosan (27). On the other hand, it was recently shown that IFN-γ has no effect on COX activity and PGE₂ production by human monocytes exposed to LPS (28).

Given the opposite effects that PGE₂ and IFN-γ have on macrophage functions, we hypothesized that IFN-γ-primed macrophages would show a differential expression of COX-2 when stimulated by IL-1β or LPS. In this study we investigated the effect of IFN-γ-priming on COX-2 expression mediated by IL-1β in human macrophages. Our data indicate that IFN-γ priming attenuates the expression of COX-2 mRNA, protein, and PGE₂ production induced by IL-1β but not by LPS. The inhibitory effect of IFN-γ priming on COX-2 mRNA and protein expression occurred in a dose- and time-dependent manner. IFN-γ priming down-regulated COX-2gene transcription in response to IL-1β but not to LPS and had no effect on the rate of COX-2 mRNA decay.

RPMI 1640 with l-glutamine was obtained from Life Technologies (Burlington, Ontario, Canada) and was completed with 100 U/ml of penicillin, 100 μg/ml streptomycin sulfate, 20 mM HEPES (Sigma, St. Louis MO), and 10% heat-inactivated FCS (HyClone Laboratories, Logan, UT). The following reagents were purchased from Sigma: LPS from Escherichia coli, serotype 0111, 4% phenol extract; PMA, dexamethasone (DEX); 3-(4,5-dimethylthiazol-2-yl-)-2,5-diphenyltetrazolium bromide (MTT). Actinomycin D (AD) was obtained from Life Technologies. IL-1β was from R&D Systems (Minneapolis, MN) and IFN-γ was purchased from Boehringer Mannheim (Laval, Quebec, Canada).

The human macrophage cell line U937 was maintained and subcultured in complete RPMI 1640. Cells were kept at 37°C in 5% CO₂-95% air and harvested at log phase of growth. The cells were adjusted to 1 × 10⁶ cells/ml in the presence of 10 nM PMA, placed in 24-well plates (1 × 10⁶ cells/well), and incubated for 3 days to differentiate into adherent macrophages. The cells were quiesced in fresh complete RPMI 1640 without PMA for 24 h before stimulation with cytokines or LPS. Cell viability was determined by the MTT method (29) and was 90% after 12 h of incubation with AD.

Total RNA was extracted with TRIZOL (Life Technologies) following the manufacturer’s instructions. Ten micrograms of RNA were denatured for 1 h at 50°C with glyoxal. Ethidium bromide was added to each sample before electrophoresis in a 1% agarose gel. The RNA was transferred to Hybond nylon membranes (Amersham Canada, Oakville, Ont., Canada) and UV cross-linked. After 3 h of prehybridization in a buffer containing 50% formamide (Life Technologies), 0.5% SDS, 5× SSPE (20× SSPE is 3 M NaCl, 0.2 M NaH₂PO₄ · H₂O, 20 mM EDTA, pH 7.4), 5× Denhardt’s mixture (50× Denhardt is 1% Ficoll, 1% polyvinylpyrrolidone, 1% BSA pentax fraction V) and 200 μg/ml of denatured salmon sperm DNA, the membranes were hybridized at 42°C overnight with [α-³²P]dCTP (10 mCi/ml, ICN Biochemicals, Montreal, Quebec, Canada)-labeled probes by nick translation (Amersham kit). The membrane was washed once at room temperature and twice at 55°C with 0.5× SSC (20× SSC is 3 M NaCl and 0.3 M sodium citrate, pH 7) for 30 min each and exposed for 24 h to an XAR-5 film (Kodak) with an intensifying screen. The probe for human COX-1 was the 1.8-kb HindIII and NotI fragment from the 5.8-kb pcDNAhCOX-1 plasmid. The probe for human COX-2 was the 1.8 kb EcoRI and ApaI fragment from the pcDNACOX-2 plasmid, a generous gift from Dr. T. Hla, American Red Cross (11). To ensure that equal amounts of RNA were analyzed, the blots were stripped (45 min in boiling 0.1% SDS) and probed for actin with the 1.25-kb PstI fragment of pBA-1 plasmid (30). The blots were scanned and the densitometric results analyzed with the National Institutes of Health Image program 1.59 (available from the Internet by anonymous FTP from zippy.nim.nih.gov or floppy disk from the National Technical Information service, Springfield, VA; part no. PB95-500195GEI).

One million cells were lysed with 100 μl of sample buffer (64 mM Tris-HCl, pH 6.8, 10.25% glycerol, 2% SDS, and 5.12% 2-ME) containing protease inhibitors (1.5 mM EDTA, 23 μM leupeptin, 14.5 μM pepstatin, 1.53 μM aprotinin, 30 μM Nα-p-tosyl-l-lysine chloromethyl ketone (TLCK), 1 mM PMSF). The whole cell lysates (equivalent to 1 × 10⁵ cells) were separated in a 10% SDS-PAGE and transferred to a nitrocellulose membrane (Hoeffer System). Polyclonal Abs prepared against the complete ovine COX-1 and COX-2 proteins (a gift from Dr. G. O’Neill, Merck Frosst, Kirkland, Quebec, Canada) were used at a 1:5000 dilution. The secondary Ab (donkey anti-rabbit IgG-horseradish peroxidase) was diluted 1:3000 and peroxidase activity was visualized by enhanced chemiluminiescence system (ECL, Amersham Canada) using Kodak XAR-5 film. Purified isoforms of COX (Cayman Chemicals, Ann Arbor, MI) were used to determine the specificity of the Abs.

Adherent macrophages were washed twice with cold Dulbecco’s PBS, scraped, and resuspended in Dulbecco’s PBS. Lysis buffer was added (320 mM sucrose, 5 mM MgCl₂, 10 mM Tris-HCl, 1% Triton X-100, pH 7.5) and the cells (5 × 10⁷) were kept on ice for 10 min and centrifuged for 15 min at 1300 × g at 4°C. The nuclear pellets were washed once in lysis buffer and the nuclei were resuspended in 100 μl of storage buffer (50 mM Tris-HCl, pH 8.0, 40% glycerol, EDTA 0.1 mM, 5 mM MgCl₂, and 1 mM DTT) and frozen in liquid nitrogen until used. Elongation of the nascent RNA was done by adding 100 μl of transcription buffer (20 mM Tris-HCl, pH 8.0, 300 mM KCl, 10 mM MgCl₂, 200 mM sucrose, 48 μM EDTA, 1 mM DTT, 1 mM of each ribonucleotides of adenosine, guanosine, and cytidine), and 10 μl of [α-³²P]uridine triphosphate (10 mCi/ml, ICN Biochemicals, Montreal, Quebec) to 100 μl of nuclei and incubating at 30°C for 30 min. The reaction was stopped with 1 ml of TRIZOL and the RNA was extracted and dissolved in 50 μl of water. The amount of radioactivity in each sample was adjusted to 5 × 10⁶ cpm, and the samples were hybridized onto nitrocellulose membranes containing 5 μg of pcDNA COX-1, pcDNACOX-2, pBA-1, pcDNA1Amp, and pBR322 linearized plasmids. The membranes were prehybridized for 3 h at 65°C in hybridization buffer (5× SSC, 5× Denhardt, 0.05% SDS, 0.5 mM EDTA, and 0.1 μg/μl of denatured salmon sperm DNA) and hybridized for 48 h. Blots were washed three times with 2× SSC and 0.1% SDS at room temperature for 10 min each time and twice with 0.2× SSC and 0.1% SDS at 60°C for 15 min and exposed for autoradiography for 15 days. The membranes were scanned and the densitometric analysis was performed as indicated for Northern blot.

PGE₂ production was measured by enzyme immunoassay (EIA, Cayman Chemicals) after different times of incubation with the cytokines.

We have previously shown that TNF-α plus IFN-γ can moderately enhance COX-2 mRNA levels in PMA-differentiated U937 human macrophages (18). To determine whether IL-1β, another potent proinflammatory cytokine, can modulate COX-2 expression, PMA-differentiated U937 human macrophages were exposed to increasing concentrations of IL-1β during 6 h and the levels of COX-2 transcripts were evaluated afterward. To evaluate the effect of macrophage priming on COX-2 expression, cells were preincubated for 24 h with 100 U/ml of IFN-γ and then stimulated with IL-1β for 6 h. Figure 1,A shows a dose-dependent increase on the steady state levels of COX-2 mRNA in both naive and IFN-γ-primed macrophages. The levels of COX-1 mRNA showed no alteration regardless of the IL-1β concentration used. Interestingly, primed macrophages consistently showed lower levels of COX-2 mRNA expression (40–60%) irrespective of the IL-1β concentration used as compared with naive macrophages. LPS (100 ng/ml), the positive control, stimulated high levels of COX-2 but not COX-1 mRNA expression in both naive and primed macrophages. Dexamethasone (DEX) completely inhibited COX-2 expression in response to IL-1β (10 ng/ml) stimulation but had no effect on constitutively-expressed COX-1 mRNA levels. As observed by immunoblot analysis, COX-2 protein was also inhibited in IFN-γ-primed cells as compared with naive macrophages (Fig. 1 B). However, the levels of COX-2 protein were similar in both naive and IFN-γ-primed cells in response to LPS stimulation. These results indicate that IL-1β is a strong stimulus to enhance COX-2 mRNA levels in human macrophages. Furthermore, they suggest that IFN-γ, at standard concentrations used for macrophage priming, can inhibit the levels of COX-2 transcripts and the expression of the COX-2 enzyme when IL-1β is the triggering signal.

To determine whether the lower levels of COX-2 mRNA observed above in primed macrophages were due only to a delay in the transcription, we studied the kinetics of COX-2 mRNA expression in response to a suboptimal dose of IL-1β (10 ng/ml). Figure 2,A shows that COX-2 mRNA was detected as early as 2 h after IL-1β stimulation, reaching maximum levels after 12 h of incubation in naive macrophages. Although a similar kinetics of COX-2 mRNA induction was observed in primed macrophages, the mRNA levels were consistently lower and never reached the same as in naive macrophages, even after 24 h of incubation (data not shown). The expression of COX-2 protein was detected after 6 h in naive macrophages whereas significantly reduced levels were observed in primed cells throughout the 12-h incubation period (Fig. 2 B). These data indicate a time dependency for COX-2 mRNA expression, a lag period for the synthesis of COX-2 protein, and clearly demonstrate the inhibitory effect of IFN-γ on COX-2 mRNA and protein expression stimulated by IL-1β.

Enzyme activity was indirectly measured as the secretion of PGE₂ in the supernatants of naive and primed macrophages exposed to IL-1β for 2, 6, and 12 h. As shown in Figure 2 B, PGE₂ production was enhanced (2.8-fold) above control levels only after 12 h of IL-1β stimulation. In IFN-γ-primed macrophages, however, 50% lower levels of PGE₂ were produced as compared with naive cells. The amount of PGE₂ released by nonstimulated cells was quantified at the end of the incubation (12 h) and represents enzyme activity of COX-1 protein. These results suggest that despite the presence of COX-2 protein at 6 h of stimulation with IL-1β, PGE₂ synthesis is augmented only after 12 h and indicate that IFN-γ is able to modulate the levels of PGE₂ secreted by macrophages.

We next determined if the time of preexposure with IFN-γ was critical for inhibiting COX-2 mRNA expression. In this experiment, macrophages were preincubated with IFN-γ at different times and then stimulated with IL-1β for 6 h. COX-2 mRNA expression was inhibited up to 47% in macrophages stimulated simultaneously with IFN-γ and IL-1β as compared with IL-1β-stimulated naive cells (Fig. 3,A). Short preexposure (2 and 6 h) with IFN-γ did not further affect COX-2 mRNA levels; however, following 12 and 24 h preexposure, COX-2 mRNA expression was significantly decreased (80%). A direct correlation was observed between COX-2 mRNA levels and protein expression (Fig. 3,B). Based on these results, we then determined whether the concentration of IFN-γ was crucial for regulating the expression of COX-2 mRNA and protein. Accordingly, macrophages were preincubated for 24 h with increasing concentrations of IFN-γ and then stimulated with IL-1β for 6 h. As shown in Figure 4, A and B, there was a dose-dependent effect of IFN-γ on the expression of COX-2 mRNA and protein stimulated by IL-1β. With 500 U/ml of IFN-γ, COX-2 mRNA levels were inhibited by 97% (Fig. 4 A). These data suggest that IFN-γ can modulate the expression of COX-2 mRNA and protein levels in a time- and dose-dependent manner.

The steady state levels of COX-2 mRNA in cells are the result of a balance between the rate of gene transcription and the rate of mRNA degradation. To determine whether IL-1β and IFN-γ were able to modulate the rate of COX-2 gene transcription, nuclear run-on analysis was performed as described in Materials and Methods. Naive and primed macrophages were stimulated for 1 h in the presence of IL-1β or LPS. Figure 5 shows that nonstimulated cells exhibited a basal level of COX-2 gene transcription, which was significantly enhanced by IL-1β (2.5-fold) and LPS (4.7-fold). However, in IFN-γ-primed macrophages stimulated with IL-1β, the relative rate of COX-2 gene transcription was inhibited by 37%. IFN-γ priming did not affect the rate of COX-2 gene transcription in response to LPS stimulation and is consistent with the data presented in Figure 1 A. These results indicate that the inhibitory effect of IFN-γ on COX-2 expression occurs at the transcriptional level.

Previous studies (13) have demonstrated that IL-1β can stabilize COX-2 mRNA in renal mesangial cells, showing that posttranscriptional mechanisms modulate COX-2 expression. A similar effect was observed with IL-1α in human endothelial cells (9). In contrast, IL-4 and IL-10 inhibit COX-2 expression by down-regulating the rate of gene transcription, but also by increasing the rate of COX-2 mRNA degradation (21). To investigate whether IFN-γ was also inhibiting the steady state levels of COX-2 mRNA by destabilizing the transcripts and therefore increasing their rate of decay, the following experiment was performed. COX-2 mRNA expression was maximally induced by IL-1β for 12 h, transcription was inhibited with AD, and the fate of the COX-2 mRNA determined after various times. Figure 6,A shows that the levels of COX-2 mRNA remained consistently high in both naive and IFN-γ-primed macrophages even after 12 h of inhibiting transcription. We confirmed that AD was inhibiting transcription as naive and primed cells simultaneously incubated with IL-1β and AD for 12 h demonstrated no expression of COX-2 mRNA (Fig. 6,A). In the absence of AD, the levels of COX-2 mRNA also remained elevated even after 24 h of stimulation with IL-1β, and COX-2 transcripts in primed macrophages never reached the same levels as in naive cells (Fig. 6,B). Similar experiments performed in the absence of IL-1β during the chase period with and without AD indicated that active transcription, together with the removal of IL-1β, was necessary for the rapid degradation of COX-2 mRNA (data not shown). Accordingly, to efficiently compare the rate of degradation between the COX-2 transcripts in naive and primed macrophages, the t_1/2 of the mRNA was measured in the presence of DEX, which has been shown to promote the degradation of COX-2 mRNA (31). The results presented in Figure 7 indicate that the half lives of COX-2 transcripts in naive and primed macrophages were similar (3.2 h), suggesting that IFN-γ did not affect the expression of COX-2 at the posttranscriptional level.

To elucidate whether de novo protein synthesis was required for the inhibitory effect of IFN-γ on COX-2 expression, naive and primed macrophages were incubated with IL-1β in the presence or absence of cycloheximide (CHX) for 12 h. Results in Figure 8 show that CHX enhanced COX-2 mRNA expression equally in naive and primed cells. Moreover, superinduction of COX-2 mRNA expression was observed in both naive and primed cells simultaneously incubated with CHX and IL-1β for 12 h. The abrogation of the down-regulatory effect of IFN-γ on COX-2 expression by CHX suggests the requirement of newly synthesized proteins (repressors) for such effect. In addition, newly synthesized factors may also be involved in the degradation of COX-2 mRNA and, in consequence, CHX can promote COX-2 accumulation.

A high output of PGs is the hallmark of inflammatory conditions and immune responses. COX-2, the inducible isoform, is considered responsible for high production of PGs (32). Several cytokines, such as IL-1α (9), TGF-β (19), IL-4, and IL-10 (21), have been described as modulators of COX-2. However, the combined effect of the network of cytokines in the milieu of inflamed sites on COX-2 is not completely understood. The objective of this study was to determine the effect of IFN-γ, a prominent modulator of macrophage functions, on the expression of COX-2 elicited by IL-1β, a proinflammatory cytokine. We demonstrated that IFN-γ-primed macrophages expressed lower levels of COX-2 mRNA and COX-2 protein, and produced lower amounts of PGE₂. IFN-γ exerted its inhibitory effect by decreasing the rate of COX-2 gene transcription and had no effect on the turnover rate of COX-2 mRNA. The inhibitory effect of IFN-γ was specific for IL-1β but not LPS-induced COX-2 mRNA expression.

IL-1β strongly enhanced the expression of COX-2 mRNA in a dose- and time-dependent manner in both naive and IFN-γ-primed macrophages, but had no effect on the expression of COX-1. A similar pattern of COX-2 induction was described in naive human monocytes and synoviocytes (8, 33). We detected COX-2 mRNA as early as 2 h after stimulation with IL-1β whereas COX-2 protein was observed after 6 h of incubation. Furthermore, the amount of PGE₂ released only increased above control levels after 12 h of incubation. This delay in PGE₂ production has been observed in LPS-stimulated macrophages (34) and is probably due to the unavailability of arachidonic acid, the substrate for COX enzymes. The activation of cytosolic phospholipase A₂ by platelet-derived growth factor has been shown to be crucial for the synthesis of PGE₂ in murine osteoblasts stimulated with IL-1α (35).

Our results clearly show a down-regulatory effect of IFN-γ on COX-2 mRNA and protein expression and PGE₂ production by macrophages. Previous studies have shown a blocking effect of IFN-γ on PGE₂ production by human monocytes when stimulated with either LPS or IL-1β (26). However, in the present study and those by Endo et al. (28), IFN-γ did not inhibit COX-2 when stimulated with LPS. In contrast, IFN-γ synergizes with LPS to enhance the expression of COX-2 and the release of PGE₂ in murine macrophages (36). We have also reported a synergistic effect of IFN-γ with TNF-α for the expression of COX-2 mRNA in human macrophages (18). These findings indicate that the cell source and the triggering signal determine the final outcome of COX-2 expression and PGE₂ production.

IFN-γ down-regulated the expression of COX-2 mRNA and protein in a dose- and time-dependent manner. Maximal inhibition was observed after 24 h of preincubation with this cytokine although simultaneous incubation with IL-1β also caused a significant decrease in COX-2 mRNA levels (47%). We then explored the possible role of IFN-γ on regulating COX-2 expression at both transcriptional and posttranscriptional levels. The steady state levels of COX-2 transcripts are the result of a balance between the rate of gene transcription and the rate of degradation of the mRNA produced. Studies in renal mesangial cells suggest that cytosolic factors activated after IL-1β stimulation can bind to the 3′-untranslated region of the COX-2 transcripts and prolong their t_1/2 (13). In our case, a possible mechanism for the down-regulatory effect of IFN-γ was that this cytokine could be promoting a faster degradation of the IL-1β-induced COX-2 mRNA and, therefore, reducing the levels detected by Northern blot, analagous to what has been reported for c-fos mRNA (37). Experiments with AD showed that transcription inhibition did not promote the rapid decay of COX-2 transcripts in either naive or primed macrophages. A rapid decay of COX-2 transcripts occurred only when IL-1β was removed after 12 h of stimulation (our unpublished observations). These observations, and the results with DEX, suggest that active transcription of certain genes or the removal of the stimulus (IL-1β) was necessary for the rapid decay of COX-2 mRNA. The ability of CHX to promote the accumulation of transcripts also indicates that COX-2 mRNA expression is strongly regulated at a posttranscriptional level. This protein synthesis inhibitor may be preventing the production of factors involved in the rapid degradation of COX-2 mRNA as it was suggested for endothelial cells (9).

Ristimakii et al. (31) have recently reported that DEX, a synthetic glucocorticoid known to inhibit the expression of COX-2 mRNA, acts by destabilizing the COX-2 transcripts. DEX does not affect the expression of COX-1 or actin, therefore we used it to specifically promote the decay of COX-2 mRNA in naive and primed macrophages to compare their half-lives. With this strategy, we demonstrated a similar t_1/2 of COX-2 transcripts in naive and primed macrophages, which ruled out a posttranscriptional effect of IFN-γ on COX-2 expression.

Interferon-γ is known to modulate the expression of a variety of genes (reviewed in Ref. 38; note that the list of genes regulated by IFN-γ are available at http://www.annurev.org./sup/material/htm). Although this cytokine acts mostly as a positive regulator, several constitutive and inducible genes such as IL-8 (39), insulin-like growth factor (40), platelet endothelial cell adhesion molecule (PECAM) (41), and low density lipoprotein (LDL)-receptor related protein (42) are susceptible to the inhibitory effect of IFN-γ. The transcriptional influence of IFN-γ on susceptible genes is mediated through some members of the Jak-STAT signaling pathway (43) and the IFN regulatory factor (IRF) family (44). By nuclear run-on analysis, we clearly demonstrated that IFN-γ-primed macrophages exhibit a lower rate of COX-2 gene transcription when the cells were stimulated with IL-1β, but not with LPS. A maximal promoter activity has been reported from −1838 to +99 bp in the 5′-flanking region of the human COX-2 gene (45). Although a variety of transcription factor elements have been identified (45, 46), no IFN-γ activation site (GAS) or IFN-stimulated response element sites have been reported. We did a computer analysis on the sequence reported by Kosaka et al. (46) of the 1.7-kb 5′-flanking region of the COX-2 gene that revealed a putative GAS sequence that expands from −1391 to −1383. An identical sequence has been described in the promoter region of the murine Ly-6A/E gene (−1219 to −1211) that can bind STAT1 (47). GAS sequences are mainly related to positive regulation of transcription (48). However, it has been shown that GAS elements in the promoter region of the human IgE Ig can also mediate repressor activity (49). Inhibition of protein synthesis with CHX indicated that newly synthesized proteins are required for the down-regulatory effect of IFN-γ on COX-2 expression. Although STAT proteins are constitutively expressed, it has been shown that STAT1 expression can be up-regulated by IFN-γ in human macrophages (50). It remains to be determined whether this transcription factor binds to the distal GAS sequence on the COX-2 promoter region or to other sequences and factors to mediate IFN-γ repression.

In summary, we have presented evidence for the differential expression of COX-2 mRNA, protein, and PGE₂ production in naive and IFN-γ-primed macrophages stimulated with IL-1β. Moreover, we showed that IFN-γ down-regulated COX-2 expression at the transcriptional level and had no effect on the stability of the transcripts. A clear understanding of the molecular mechanisms that regulate COX-2 expression may lead to the development of novel treatments for inflammation, colon cancer, and autoimmune diseases.

We thank Kathy Keller for excellent technical assistance and Denis Gaucher for critical review of the manuscript.