Two Opposing Effects of Non-steroidal Anti-inflammatory Drugs on the Expression of the Inducible Cyclooxygenase MEDIATION THROUGH DIFFERENT SIGNALING PATHWAYS*

The efficacy of non-steroidal anti-inflammatory drugs (NSAIDs) is considered to be a result of their inhibitory effect on cyclooxygenase (COX) activity. Here, we report that flufenamic acid shows two opposing effects on COX-2 expression; it induces COX-2 expression in the colon cancer cell line (HT-29) and macrophage cell line (RAW 264.7); conversely, it inhibits tumor necrosis factor a (TNFa)or lipopolysaccharide (LPS)-induced COX-2 expression. This inhibition correlates with the suppression of TNFaor LPS-induced NFkB activation by flufenamic acid. The inhibitor of extracellular signalregulated protein kinase, p38, or NFkB does not affect the NSAID-induced COX-2 expression. These results suggest that the NSAID-induced COX-2 expression is not mediated through activation of NFkB and mitogen-activated protein kinases. An activator of peroxisome proliferator-activated receptor g, 15-deoxy-D-prostaglandin J2, also induces COX-2 expression and inhibits TNFa-induced NFkB activation and COX-2 expression. Flufenamic acid and 15-deoxy-D-prostaglandin J2 also inhibit LPS-induced expression of inducible form of nitric-oxide synthase and interleukin-1a in RAW 264.7 cells. Together, these results indicate that the NSAIDs inhibit mitogen-induced COX-2 expression while they induce COX-2 expression. Furthermore, the results suggest that the anti-inflammatory effects of flufenamic acid and some other NSAIDs are due to their inhibitory action on the mitogen-induced expression of COX-2 and downstream markers of inflammation in addition to their inhibitory effect on COX enzyme activity.

Many epidemiological studies have revealed that the use of aspirin or other non-steroidal anti-inflammatory drugs (NSAIDs) 1 can reduce the risk of colon cancer. Since the well documented pharmacological action of aspirin and other NSAIDs is the inhibition of cyclooxygenase (COX, the ratelimiting enzyme in prostaglandin biosynthesis), it can be inferred that the beneficial effect of NSAIDs may be mediated through the inhibition of prostaglandin biosynthesis. However, experimental evidence to support this hypothesis has not been conclusively demonstrated. Several lines of experimental observations imply that the beneficial effects of NSAIDs may be mediated through both COX-dependent and COX-independent pathways.
Two isoforms of COX have been identified: constitutively expressed COX-1 (1)(2)(3)(4)(5) and mitogen-inducible COX-2 (6 -11). Evidence supporting the hypothesis that the beneficial effect of NSAIDs in reducing the risk of colon cancer is mediated by the inhibition of COX activity includes the fact that cross-breeding of APC ⌬716 knockout mice with COX-2 knockout mice, or the administration of the COX-2 specific inhibitor to APC ⌬716 knockouts, resulted in a dramatic reduction in the numbers and size of the intestinal polyps (12). In addition, it has been demonstrated that the overexpression of COX-2 in intestinal epithelial cells leads to enhanced tumorigenic phenotypes, metastatic potential, and angiogenesis (13)(14)(15).
It has been shown that NSAIDs have pharmacological effects other than the inhibition of COX activity. Sodium salicylate and aspirin were shown to inhibit the transcription factor NFB (16). NSAIDs can also activate peroxisome proliferatoractivated receptors (PPAR) ␣ and ␥, and induce differentiation of pre-adipocytes to adipocytes (17). Results from recent studies by Meade et al. (18) demonstrated that various NSAIDs, as PPAR activators, induce the expression of COX-2 in epithelial cells. However, Xu et al. (19) showed that aspirin and sodium salicylate suppress COX-2 expression induced by IL-1 in endothelial cells. In addition, the activation of PPAR␣ by Wy 14643 leads to the inhibition of IL-1-induced COX-2 expression in smooth muscle cells (20).
To clarify these seemingly opposing results, we studied the effects of NSAIDs on COX-2 expression in the presence or absence of a known inducer of COX-2 expression. COX-2 expression is induced by various mitogenic stimuli in different cell types (7,9,11,22). The cis-acting NFB element is present in the 5Ј-flanking regions of COX-2 genes of different species (23,24). Results from our previous studies demonstrated that the activation of NFB is required to induce the expression of COX-2 in the lipopolysaccharide (LPS)-stimulated macrophage cell line (25). The activation of mitogen-activated protein kinases (MAPKs, ERK-1 and -2, and p38) alone is not sufficient to induce the expression of COX-2, but the inhibition of ERK-1 and -2 or p38 results in partial suppression of COX-2 expression (25). Pro-inflammatory cytokines, such as TNF␣ and IL-1, also activate NFB and MAPKs, and induce the expression of COX-2 in many cell types (26,27).
Thus, we investigated signaling pathways through which NSAIDs modulate the expression of COX-2 in a colon tumor cell line (HT-29) treated with TNF␣ and a macrophage-like cell line (RAW 264.7) stimulated by LPS. If NSAIDs can modulate mitogen-induced expression of COX-2 in addition to inhibiting the enzyme activity of COX, this may represent a new mechanism of anti-inflammatory and anti-neoplastic actions of NSAIDs.
Cell Culture-HT-29 cells (a human colon adenocarcinoma cell line, ATCC HTB-38) or RAW 264.7 cells (a murine macrophage-like cell line, ATCC TIB-71) were cultured in Dulbecco's modified Eagle's medium containing 10% (v/v) heat-inactivated fetal bovine serum (FBS, Intergen) and 100 units/ml penicillin and 100 g/ml streptomycin (Life Technologies, Inc.) at 37°C in a 5% CO 2 /air environment. Cells (2 ϫ 10 6 ) were plated in 60-mm dishes (Falcon) and cultured for an additional 18 h to allow the number of cells to approximately double. Cells were maintained in the serum-poor (0.25% FBS) medium for another 18 h prior to the treatment with indicated reagents.
Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis and Immunoblotting-These were performed essentially the same as described previously (25,30,31).
Preparation of Nuclear Extracts-Cells (4 ϫ 10 6 ) were plated in 100-mm dishes and cultured in medium containing 0.25% FBS for 18 h. Fifteen minutes after the TNF␣ stimulation with or without various NSAIDs, cells were scraped in ice-cold PBS, pelleted, and washed with PBS one more time. Pellets were resuspended in 400 l of buffer A (10 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 10 mM KCl, 0.3 mM EGTA, 0.3 M sucrose, 10 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 0.5% (v/v) Nonidet P-40, 1 g/ml leupeptin) and incubated on ice for 20 min. Nuclei were pelleted by centrifugation at 7,200 ϫ g for 10 min at 4°C, followed by washing with buffer A. Nuclei were resuspended in 100 l of high salt buffer B (20 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 320 mM KCl, 0.2 mM EGTA, 0.5 mM dithiothreitol, 1 g/ml aprotinin, 1 g/ml leupeptin) and incubated on ice for 15 min. After centrifugation at 13,500 ϫ g for 15 min at 4°C, the supernatants were kept at Ϫ70°C until analyzed. The protein content of the supernatant was determined by the Bradford method.
Electrophoretic Mobility Shift Assay (EMSA)-A double-stranded oligonucleotide containing a tandem repeat of the consensus sequence for the NFB binding site was used: 5Ј-GATCCAAGGGGACTTTCCATG-GATCCAAGGGGACTTTCCATG-3Ј, 3Ј-GTTCCCCTGAAAGGTACCT-AGGTTCCCCGAAAGGTACCTAG-5Ј. Five nanograms of doublestranded oligonucleotide were end-labeled in polynucleotide kinase buffer (60 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 15 mM ␤-mercaptoethanol, 0.33 M ATP) using T4 polynucleotide kinase in the presence of 100 Ci of [␥-32 P]ATP. The labeled oligonucleotides were purified by G-50 Sephadex ® spin columns. Five micrograms of nuclear extract were mixed with incubation buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 4% (v/v) glycerol, 0.08 mg/ml sonicated salmon sperm DNA) and incubated at 4°C for 15 min. The labeled oligonucleotides (40,000 -100,000 cpm) were added to the preincubated mixture and the incubation continued at room temperature for 20 min. Reaction mixtures were run on a 6% non-denaturing polyacrylamide gel at 150 V until the front dye reached 2-3 cm of the bottom of the gel. Following completion of running, the gel was transferred to blotter paper and dried under vacuum. The dried gel was placed in the Phos-phorImager screen and exposed overnight.
Luciferase Reporter Gene Assay-Cells were plated in six-well plates (4 ϫ 10 5 cells/well) and transfected with the reporter gene plasmids using SuperFect transfect reagent (Quiagen, Valencia, CA) according to the manufacturer's instruction. HT-29 cells or RAW 264.7 cells were transfected with 2.5 g of NFB response element-driven pGL2 luciferase reporter plasmid, and 0.5 g of HSP70-lacZ as an internal control (kindly provided by Robert L. Modlin, UCLA, Los Angeles, CA). For COX-2 promoter assay, 2.5 g of murine COX-2 gene promoter (nt-1,017/ϩ93)-driven pGL2 luciferase reporter plasmid (gift from David Dewitt, Michigan State University, East Lansing, MI) was used. After 3 h, the medium was changed and further incubated for 6 h. The cells were further incubated in medium containing 0.25% FBS for an additional 15 h. The cells were treated with flufenamic acid alone or a combination of flufenamic acid with TNF␣ (50 ng/ml, Sigma) or LPS in the serum-poor (0.25% FBS) medium. The luciferase activity was determined using the Luciferase Assay System (Promega, Madison, WI) according to the manufacturer's recommended protocol. Luciferase activity was normalized to the internal control plasmid HSP70-lacZ by measuring ␤-galactosidase activity.
Ligand Binding Assay for PPAR␥-HT-29 cells or RAW 264.7 cells were transfected with 1 g of the chimeric receptor expression construct, pcDNA3-hPPAR␥/GAL4 containing the ligand binding domain of hPPAR␥ and the yeast GAL4 transcription factor DNA binding domain. Ligand binding activity was measured by co-transfecting 1.5 g of the reporter gene construct, pUAS(5x)-tk-luc, which contains five copies of GAL4 response element (kindly provided by Joel Berger, Merck Research Laboratory), and 0.5 g of HSP70-lacZ as an internal control. After 3 h, the medium was changed and further incubated for 6 h. The cells were further incubated in the serum-poor (0.25% FBS) medium for an additional 15 h. The medium was removed and fresh medium containing various NSAIDs or 1 M of 15-deoxy-⌬ 12,14 -prostaglandin J 2 (15d-PGJ 2 ) was added to each well and incubated for another 24 h. The luciferase activity was determined as described above.

NSAIDs, Flufenamic Acid, and Sulindac Sulfide Inhibit TNF␣-induced NFB Activation: This Inhibition Leads to the Suppression of COX-2
Expression-Among various NSAIDs tested, flufenamic acid and sulindac sulfide were the most potent inhibitors of TNF␣-induced NFB activation determined by IB␣ degradation in HT-29 cells (data not shown). Thus, we investigated these two NSAIDs in this report.
Pretreatment of HT-29 cells with flufenamic acid shows a biphasic effect on TNF␣-induced COX-2 expression: enhancement at concentrations below 200 M but inhibition at concentrations above 200 M (Fig. 1A). However, flufenamic acid suppressed the TNF␣-induced activation of NFB in a dose-dependent fashion without showing the biphasic effect as demonstrated by both EMSA and NFB reporter gene assay (Fig. 1, B and C). Similar to flufenamic acid, sulindac sulfide inhibits TNF␣induced COX-2 expression and this inhibition correlates with the suppression of TNF␣-induced NFB activation by sulindac sulfide (Fig. 2, A and B). In addition, TNF␣-induced COX-2 expression is suppressed by the inhibitor of NFB (data not shown). These results suggest that the inhibition of TNF␣induced COX-2 expression by flufenamic acid or sulindac sulfide is at least in part due to its inhibitory effect on TNF␣induced NFB activation.

Flufenamic Acid in the Absence of Other COX-2 Inducers in the Medium Induces the Expression of COX-2 in HT-29 Cells:
This Induction Was Not Inhibited by the Inhibitors of MAPKs or NFB-Next, to determine whether enhancement of TNF␣induced COX-2 expression by flufenamic acid at concentrations below 200 M (Fig. 1A) is due to COX-2 expression induced by flufenamic acid itself, cells were treated with flufenamic acid alone in the absence of other COX-2 inducers. Flufenamic acid alone induces COX-2 expression in a dose-dependent fashion (Fig. 3A). This induction was not inhibited by the pretreatment of cells with SB203580, a specific inhibitor of p38, or a mixture of inhibitors, PD98059 and TPCK, for MEK1 and NFB, respectively (Fig. 3, B and C). Sulindac sulfide also induces COX-2 in the serum-poor medium and this induction was not inhibited by inhibitors of MAPKs or NFB (Fig. 4, A-C). Flufenamic acid and other NSAIDs alone do not induce the degradation of IB␣ (data not shown). Flufenamic acid and other NSAIDs do not affect COX-1 expression in HT-29 cells (data not shown). These results suggest that the expression of COX-2 induced by NSAIDs such as flufenamic acid or sulindac sulfide is not mediated through the activation of MAPKs and NFB signaling pathway.
Results from recent studies demonstrated that some NSAIDs including flufenamic acid can bind and activate PPAR␥ and PPAR␣ (17) and induce the expression of COX-2 in epithelial cells and fibroblasts (18,32). Our immunoblot analyses demonstrated that PPAR␥, but not PPAR␣, was detected in HT-29 cells (data not shown). Thus, we determined the effects of another known activator of PPAR␥ on COX-2 expression in

FIG. 2. Inhibitory effects of sulindac sulfide on TNF␣-induced COX-2 expression (A) and activation of NFB (B).
Cells maintained in serum-poor medium were treated with sulindac sulfide (Si) for 3 h and then stimulated with TNF␣ (20 ng/ml) in the presence of sulindac sulfide. EMSA for NFB and COX-2 immunoblot analyses were performed as described in Fig. 1. HT-29 cells. Results show that, similar to flufenamic acid, other known PPAR␥ activators troglitazone, indomethacin, and 15d-PGJ 2 induce COX-2 expression (Fig. 5). In addition, pretreatment with 15d-PGJ 2 results in the inhibition of TNF␣induced IB␣ degradation and COX-2 expression (Fig. 6, A and  B). However, indomethacin, although it induces COX-2 expression, does not inhibit TNF␣-induced COX-2 expression (data not shown).

FIG. 3. COX-2 expression induced by flufenamic acid (Flu) in HT-29 cells is not inhibited by the inhibitor of p38 (SB203580) or the mixture of the inhibitors of NFB (TPCK) and MEK1 (PD98059). A, cells maintained in
To determine whether NSAIDs bind PPAR␥, HT-29 cells were transfected with the chimeric receptor expression construct, pcDNA3-hPPAR␥/GAL4 and the reporter gene construct, pUAS(5x)-tk-luc as described elsewhere (33). Treatment of HT-29 cells with flufenamic acid, sulindac sulfide, or 15d-PGJ 2 resulted in a significantly increased ligand binding activity to hPPAR␥ (Fig. 7A). The same pattern of results was shown in RAW 264.7 cells (Fig. 7B). These results imply that flufenamic acid-and sulindac sulfide-induced COX-2 expression is mediated through the activation of PPAR␥ both in HT-29 cells and RAW 264.7 cells.

Flufenamic Acid Induces COX-2 Expression and Also Inhibits the LPS-induced Activation of NFB and COX-2 Expression in the Murine Macrophage-like Cell Line (RAW 264.7)-We
next determined in a cell type other than HT-29 cells whether flufenamic acid induces COX-2 expression and also inhibits activation of NFB and COX-2 expression induced by mitogenic stimulation. Pretreatment of RAW 264.7 cells with flufenamic acid leads to a dose-dependent inhibition of LPS-stimulated transcriptional activity of COX-2 promoter and NFB reporter genes (Fig. 8, A and C, respectively).
Flufenamic acid, in the absence of other inducers of COX-2 expression, enhances transcriptional activity of COX-2 promoter-reporter gene in RAW 264.7 cells (Fig. 8B). However, flufenamic acid alone does not affect the basal promoter activity of NFB (Fig. 8D), indicating that flufenamic-induced COX-2 expression is not mediated through the activation of NFB. This result corroborates with the results, obtained by Western blot analyses of endogenous COX-2 protein in HT-29 cells (Fig. 3), demonstrating that the inhibitor of NFB does not suppress flufenamic-induced COX-2 expression.

Flufenamic Acid and 15d-PGJ 2 Inhibit the LPS-induced Expression of Other Pro-inflammatory Marker Gene Products
Such as iNOS and IL-1␣ in RAW 264.7 Cells-Pretreatment of RAW 264.7 cells with flufenamic acid or 15d-PGJ 2 leads to a dose-dependent inhibition of LPS-induced expression of iNOS and IL-1␣ as determined by Western blot analyses (Fig. 9, A  and B). These results suggest that NSAIDs, which inhibit mitogen-induced activation of NFB, can suppress the expression of many genes whose induction is mediated in part through activation of NFB. DISCUSSION Our results demonstrate that NSAIDs have two opposing effects on COX-2 expression; NSAIDs inhibit cytokine-induced COX-2 expression, while NSAIDs alone can induce COX-2 ex-

FIG. 4. COX-2 expression induced by sulindac sulfide (Si) in HT-29 cells is not inhibited by the inhibitor of p38 (SB203580) or the mixture of the inhibitors of NFB (TPCK) and MEK1
(PD98059). Cells in serum-poor medium were treated and analyzed as described in Fig. 3. The panels are representative immunoblots from more than three different analyses.

FIG. 6. Inhibitory effects of the PPAR␥ activator, 15d-PGJ 2 on the TNF␣-induced degradation of IB␣ and COX-2 expression in HT-29 cells.
A, cells maintained in serum-poor medium were treated with 15d-PGJ 2 for 3 h and then stimulated with TNF␣ (20 ng/ml) for 15 min in the presence of 15d-PGJ 2 . Cell lysates were analyzed by IB␣ immunoblot. B, cells were treated with 15d-PGJ 2 for 3 h and then stimulated with TNF␣ (20 ng/ml) for 8 h in the presence of 15d-PGJ 2 . Cell lysates were analyzed by COX-2 and ␤-actin immunoblot. The panels are representative immunoblots from more than three different analyses.
pression. Results from promoter-reporter assays demonstrate that flufenamic acid inhibits LPS-induced COX-2 expression and NFB activation in RAW 264.7 cells (Fig. 8, A and C), whereas it induces COX-2 expression in the absence of LPS (Fig. 8B). The concentrations of flufenamic acid required to inhibit LPS-induced COX-2 expression and to induce COX-2 expression are in a similar range. Thus, the magnitude of the inhibition of LPS-induced COX-2 expression by flufenamic acid might have been even greater if there was no simultaneous induction of COX-2. In HT-29 cells, enhancement of TNF␣induced COX-2 expression by flufenamic acid at 200 M or lower is likely due to the fact that the additive induction of COX-2 expression by flufenamic acid is greater than its inhibitory effect on TNF␣-induced COX-2 expression at these concentrations. However, the inhibitory effect of flufenamic acid on TNF␣-induced COX-2 expression at higher concentrations may far exceed the additive induction of COX-2 expression by flufenamic acid. Flufenamic acid does not cause cell death at concentrations up to 200 M; however, it induces cell death at concentrations above 200 M. It has been well documented that activation of NFB suppresses apoptotic signals in many cell types (36 -39); conversely, inhibition of NFB can induce apoptosis. Thus, it is likely that induction of apoptosis and inhibition of TNF␣-induced COX-2 expression by flufenamic acid are mediated through a common signaling pathway, i.e. inhibition of NFB.
Flufenamic acid does not have an opposing effect on NFB activation. In both cell types, pretreatment with flufenamic acid leads to a dose-dependent inhibition of TNF␣-or LPSinduced NFB activation. Flufenamic acid alone does not cause NFB activation.
Many NSAIDs bind and activate PPARs and some PPAR activators have been shown to inhibit NFB activity (17,20). TNF␣-induced COX-2 expression in HT-29 cells was inhibited by flufenamic acid, sulindac sulfide, or 15d-PGJ 2 , all of which bind PPAR␥ (Fig. 7). However, indomethacin, a known activator of PPAR␥, does not inhibit TNF␣-induced COX-2 expression and NFB activation (data not shown). It was previously shown that, unlike sulindac sulfide, indomethacin does not inhibit IB kinase (40,41). Recently, it was demonstrated that not all PPAR activators inhibit NFB activation (21) and cytokine production in macrophages (42). Together, these results suggest that the inhibition of TNF␣-induced COX-2 expression or NFB activation by flufenamic acid is not mediated through the activation of PPARs. Furthermore, these results suggest that NSAIDs such as indomethacin, which do not inhibit mitogen-induced activation of NFB, are unable to inhibit the mitogen-induced COX-2 expression.
The molecular target through which flufenamic acid inhibits TNF␣-or LPS-induced NFB activation is not known. Recent studies have demonstrated that aspirin and sodium salicylate suppress NFB activation by inhibition of IB kinase ␤ (16,40,43). Another study has demonstrated that 15d-PGJ 2 inhibits NFB by a covalent modification of a cysteine residue within its activation loop of IB kinase ␤ (21). This irreversible modification is rendered by the formation of Michael adducts between a reactive ␣,␤-unsaturated carbonyl group in the cyclopentane ring of 15d-PGJ 2 and cellular nucleophiles such as compounds containing free SH group. Salicylate and flufenamic acid do not appear to possess such a reactive group for nucleophilic attack in their structure. Thus, the mode of action in inhibiting NFB by flufenamic acid is likely different from that of 15d-PGJ 2 .
The flufenamic acid-or sulindac sulfide-induced COX-2 expression was not affected by either the inhibitor of p38 or by inhibitors of NFB and MEK1 (Figs. 3 and 4). These results indicate that, unlike the TNF␣-induced COX-2 expression, the NSAID-induced COX-2 expression is mediated through signaling pathways that do not require the activation of MAPKs and NFB. Some NSAIDs and 15d-PGJ 2 , which are known to activate PPAR␥, induce COX-2 expression in the absence of other inducers of COX-2 expression (Figs. [3][4][5]. Aspirin and sodium salicylate which do not activate PPARs (17) were unable to induce COX-2 expression (data not shown). Together, these results suggest, but do not prove, that flufenamic acid-and sulindac sulfide-induced COX-2 expression is mediated through the activation of PPAR␥. The pharmacological significance of the induction of COX-2 expression by NSAIDs is not known. Since NSAIDs inhibit the activity of COX-2 expressed in tissues in response to NSAIDs, the inhibitory effect of NSAIDs on cytokine-induced COX-2 expression would be a more important net effect.
It has been a prevailing belief that the efficacy of NSAIDs is due to their inhibitory effect on COX activity. However, the therapeutic benefit of NSAIDs is observed at plasma concentrations substantially higher than those required to inhibit COX (34). Emerging evidence now suggests that NSAIDs can also exert their anti-inflammatory and possible anti-tumor effects through COX-independent pathways (35). Our results demonstrating that NSAIDs inhibit TNF␣-induced activation of NFB signaling pathways suggest that NSAIDs can inhibit the cellular responses to pro-inflammatory cytokines by inhibiting the downstream signaling pathways derived from the activation of cytokine receptors. Furthermore, flufenamic acid inhibits not only COX-2 expression (Fig. 8) but also the expression of other inflammatory marker gene products such as iNOS and IL-1␣ induced by LPS in RAW 264.7 cells (Fig. 9). These results suggest that NSAIDs inhibit not only downstream signaling pathways derived from the activation of pro-inflammatory cytokine receptors, but also the expression of pro-inflammatory marker gene products in response to inflammatory stimuli.
Macrophages, important components of stromal cells in tumor tissues, can release cytokines, which in turn stimulate tumor cells and other stromal cells to induce the expression of COX-2. Our results suggest that NSAIDs can inhibit both the production of cytokines by macrophages, and the induction of COX-2 by tumor cells in response to the cytokines. These effects may represent an additional mechanism by which NSAIDs exert their anti-inflammatory and possible anti-neoplastic effects.
In summary, our results presented here suggest that the pharmacological effects of NSAIDs are mediated not only through the inhibition of COX activity but also the modulation of the expression of COX-2 and other pro-inflammatory marker gene products.