Volume 74 • Number 12

1754

Cyclooxygenase-2 Inhibitors DecreaseInterleukin-1ββ–Stimulated ProstaglandinE2 and IL-6 Production by Human GingivalFibroblastsDavid A. Tipton,*† Jon C. Flynn,* Sidney H. Stein,* and Mustafa Kh. Dabbous*†‡

Background: Previous work showed that normal and aggres-sive periodontitis (AgP) gingival fibroblasts produce the bone-resorbing cytokine IL-6. PGE2 is important in regulating IL-6production. Non-steroidal anti-inflammatory drugs inhibit PG syn-thesis via COX-1 and/or COX-2 isoenzymes and may inhibit peri-odontal destruction. COX-2 is induced after cellular activation(i.e., by inflammatory cytokines such as IL-1β). Little is knownabout IL-1β-stimulated AgP fibroblast IL-6 and PGE2 productionand their regulation by COX inhibitors. The objective of this studywas to determine the effects of COX-2 inhibitors on amounts ofPGE2 and IL-6 made by IL-1β-stimulated gingival fibroblasts.

Methods: Gingival fibroblasts (2.5 × 104) from healthy orsevere periodontitis patients were cultured in serum-free medium,with or without IL-1β (10−11M) for 24 hours, with or without theCOX-1/2 inhibitor indomethacin or the selective COX-2 inhibitorsNS-398, celecoxib, or rofecoxib. PGE2 and IL-6 in culture super-natants were determined by specific enzyme-linked immunosor-bent assay (ELISA)s.

Results: All of the COX inhibitors caused dose-dependentdecreases in IL-1β-stimulated PGE2, to a maximum of >90%in all cell lines (P ≤0.0001). The selective COX-2 inhibitors, butnot indomethacin, caused partial (generally up to approximately60%), dose-dependent decreases in IL-1β-stimulated IL-6 in allcell lines (P ≤0.003). When exogenous PGE2 was added con-currently with COX-2 inhibitors before addition of IL-1β, IL-6 pro-duction returned to levels at or approaching that produced bycells exposed only to IL-1β (P ≤0.04).

Conclusion: The results suggest that COX-2 inhibition may beuseful in helping to control fibroblast production of IL-6 in patientswith severe periodontitis. J Periodontol 2003;74:1754-1763.

KEY WORDSBone resorption; COX-2 inhibitors; fibroblasts, periodontal;interleukin-6; periodontitis, aggressive; prostaglandin E2.

* Department of Periodontology, The University of Tennessee Health Science Center,Memphis, TN.

† Dental Research Center, The University of Tennessee Health Science Center.‡ Department of Molecular Sciences, The University of Tennessee Health Science Center.

Aggressive periodontitis (AgP) is agroup of diseases characterizedby rapid and extreme periodontal

destruction. Severe periodontitis can alsobe a manifestation of chronic neutrope-nia. Many factors contribute to bony andsoft tissue destruction in these disorders,including the matrix metalloproteinases(MMPs), a family of enzymes producedby fibroblasts and other cells resident inthe periodontal tissues. MMPs togethercan degrade all of the components of theextracellular matrix of the periodontium,including bone.1 Work in this laboratorysuggested that AP gingival fibroblastsmay contribute to tissue destruction viaincreased production of stromelysin-1(MMP-3).

Cytokines produced by immune cellsand a variety of mesenchymal cells regu-late MMP expression and also participatein the initiation and progression of peri-odontal disease.2-4 Gingival fibroblastsare regulated by cytokines produced byimmune cells, but also produce cyto-kines themselves, making them part of acytokine network which participates inand maintains local inflammation. Inter-leukin-6 (IL-6) is a cytokine important inthe pathogenesis of periodontal diseasedue to its bone-resorbing activity, possi-bly through stimulating osteoclast pre-cursor recruitment and differentiation.5-8

Patients with refractory periodontitis haveincreased levels of IL-6 in gingival crevic-ular fluid (GCF),7 and IL-6 mRNA isincreased at periodontal disease sites.9,10

3029.qxd 1/9/04 11:02 AM Page 1754

1755

J Periodontol • December 2003 Tipton, Flynn, Stein, Dabbous

Takahashi et al.11 also suggested that endothelial cells,fibroblasts and macrophages contribute to periodontaldestruction via their production of IL-6. Inflammatorycytokines such as interleukin-1β (IL-1β), as well as Acti-nobacillus actinomycetemcomitans (Aa) and E. colilipopolysaccharide (LPS), stimulate the expression ofIL-6 by normal and adult periodontitis human gingivalfibroblasts.12,13

Prostaglandin E2 (PGE2) also is important in thepathogenesis of periodontal disease, both for its directeffects on the periodontium (i.e., enhancement of MMPsecretion, increased osteoclastic bone resorption, andincreased vasopermeability and vasodilation), and itsindirect effects via regulation of cytokines such as fibro-blast IL-6.4,14,15 PGE2 is associated with attachmentloss in periodontitis and PGE2 levels in GCF are sub-stantially higher in high-risk patients with refractoryAgP.16,17 PGE2 increases IL-6 production in synovialfibroblasts18 and in gingival fibroblasts from patientswith severe periodontitis.19 The non-steroidal anti-inflammatory drug (NSAID) indomethacin, an inhibitorof PGE2 production, can inhibit fibroblast IL-6 produc-tion,19 suggesting that PGE2 is an important endogenousmodulator of IL-6. NSAIDs that inhibit PG synthesis canprevent periodontal destruction, but these drugs canhave substantial gastric toxicity.20-23

A key enzyme in PG production is prostaglandinendoperoxidase H synthase (cyclooxygenase or COX),which exists in two isomeric forms, COX-1 and COX-2.24

The COX gene appears to be one genetic marker forAgP.25 COX-1 is expressed constitutively in many tis-sues and may be involved in homeostatic prostanoidbiosynthesis.26,27 COX-2 is involved in inflammation,28-33

induced following cellular activation by LPS or inflam-matory cytokines such as IL-1β. IL-1β is a potent stim-ulator of PGE2 and IL-6 production by human gingivalfibroblasts34,35 and is a risk factor for periodontitis.36

Gingival fibroblasts stimulated with IL-1β and LPS pro-duce PGE2 via de novo synthesis of COX-2.37-39 COX-2appears to have a major role in production of PGE2 inperiodontitis, and therefore a selective COX-2 inhibitormay be useful in treating it. NSAIDs inhibit PG synthe-sis via inhibition of COX-1 and COX-2 enzyme activity.40

Their ability to inhibit COX-2 may explain the thera-peutic effects of NSAIDs as anti-inflammatory drugs,and the inhibition of COX-1 may account for some oftheir unwanted side effects.27,41 Accordingly, much of therecent research on anti-inflammatory drugs has focusedon targeting the production of PGs by the inducibleCOX-2 enzyme. In this study we have determined theeffects of non-selective and selective COX-2 inhibitorson IL-1β-stimulated PGE2 and IL-6 production by humangingival fibroblasts from normal patients and from indi-viduals with severe forms of periodontitis. Understand-ing the mechanisms of the IL-6 production by fibroblastsmay be important in developing more effective treat-

ments for severe forms of periodontitis, including aggres-sive periodontitis.

MATERIALS AND METHODSHuman Gingival FibroblastsFibroblast cell lines derived from two patients with severeperiodontitis and matched normal gingival fibroblast celllines derived from healthy patients with non-inflamedgingiva were used in this study. The periodontitis patientswere African American, one male and one female, aged11 and 15 years old, respectively. The female patientwas diagnosed with AgP. The male patient had con-genital, severe, chronic neutropenia and was receivingsubcutaneous granulocyte colony stimulating factor(G-CSF). G-CSF is a specific stimulator of the growthand differentiation of hematopoietic progenitor cellscommitted to the neutrophil lineage and it aids in move-ment of mature neutrophils from the bone marrow intocirculation. This patient was diagnosed with periodon-titis as a manifestation of systemic disease, associatedwith a hematological disorder. Both patients presentedwith greater than 80% generalized alveolar bone lossand generalized edematous and erythematous tissue.

Fibroblast cell lines were derived from gingival explantsusing standard techniques42 after obtaining informedconsent, as approved by the UTHSC Institutional ReviewBoard. The cells were maintained in Dulbecco’s modifiedeagle medium (DMEM) supplemented with 10% (v/v)newborn calf serum (NCS);§ and 100 µg/ml gentamicini

(complete medium). The cells were grown at 37°C in ahumidified atmosphere of 5% CO2 in air and passagedby brief treatment with trypsin (0.25%).§ Fibroblastsbetween passages 2 and 10 were used in the experi-ments. Normal gingival fibroblasts were designated “GN”;GN 46, GN 60, and GN 62 were used in this study.Gingival fibroblasts from periodontitis patients weredesignated “GF” (GF 2 and GF 3, from the AgP patient)or GN/CN (from the patient with chronic neutropenia).

COX InhibitorsCOX inhibitors used in this study were indomethacin,¶

NS-398,¶ celecoxib,# and rofecoxib.** Indomethacinis a non-selective COX inhibitor (1-[4-Chlorobenzoyl]-5-methoxy-2-methyl-1H-indole-3-acetic acid; MW357.8). NS-398 is a sulfonamide class drug (N-[2-(cyclohexyloxyl)-4-nitrophenyl]-methanesulfonamide;MW 314.4) that is a selective COX-2 inhibitor. Cele-coxib (MW 381.38) and rofecoxib (MW 314.36) aremodifications of DuP 697, a tricyclic compound, andboth show an even greater selectivity for COX-2 inhi-bition than NS-398. Stock solutions (100×) of each

§ Gibco, Grand Island, NY.i Sigma Chemical Co., St. Louis, MO.¶ Cayman Chemical, Ann Arbor, MI.# Celebrex, Pharmacia Corp., St. Louis, MO.** Vioxx, Merck and Co., Inc., Rahway, NJ.

3029.qxd 1/9/04 11:02 AM Page 1755

1756

Downregulation of PGE2 and IL-6 by COX-2 Inhibitors Volume 74 • Number 12

COX inhibitor were prepared in dimethylsulfoxide(DMSO)iand were diluted 1:100 in serum-free DMEMcontaining 100 µg/ml gentamicin (DMEM-gent) beforeaddition to the cells.

Determination of Constitutive and IL-1ββ–InducedFibroblast Production of PGE2 and IL-6Constitutive and IL-1β-induced production of PGE2 andIL-6 were measured in fibroblast-conditioned mediausing an IL-6 ELISA development system†† or a PGE2ELISA kit.¶ Fibroblasts were seeded at 2.5 × 104 cells perwell in 24-well plates in complete medium and allowedto adhere overnight at 37°C. The next day, the mediumwas removed, the wells washed once with DPBS, andDMEM-gent, with or without IL-1⇇ (10−11M),18 wasadded. After incubation for 24 hours at 37°C, the mediawere harvested and assayed according to the ven-dor’s protocol for the particular assay. The results wereexpressed as ng PGE2 or IL-6 per mg protein in the cellmonolayer, determined by a modification of the methodof Lowry et al.43

Effect of COX Inhibitors on Constitutive and IL-1ββ–Induced Fibroblast Production of PGE2 and IL-6To determine the effect of COX inhibitors on constitu-tive and IL-1β (1 × 10−11M)-stimulated IL-6 and PGE2production, the experiments were carried out as describedin the previous section, using the representative AgPfibroblast cell line GF 3, except that the cells were cul-tured with or without indomethacin (1-500 µM),44 NS-398(1-1000 nM),45 celecoxib (0.25-25 µM),44 or rofecoxib(1-50 µM). Control medium was DMEM-gent containing1% (v/v) DMSO (DMEM-gent/DMSO) ± IL-1β.

Effect of DMSO and COX Inhibitors on CellViabilityThe effect of the COX inhibitors and control media(DMEM-gent or DMEM-gent/DMSO) on fibroblast viabilitywas determined by their effect on the ability of the cellsto reduce the tetrazolium salt (3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyl tetrazolium bromide) (MTT) to a for-mazan dye (Cell Proliferation Kit I [MTT]).§§ Individualwells of 96-well microtiter tissue culture plates�� wereseeded with 2.5 × 104 cells in 0.2 ml complete mediumand incubated at 37°C. The medium was then removedand replaced with DMEM-gent or DMEM-gent/DMSOor DMEM-gent supplemented with NS-398, celecoxib,rofecoxib, or indomethacin. The COX inhibitors wereprepared in DMSO; a 1% DMSO concentration was inworking solutions added to cells. After 24 hours, viabil-ity was assessed using the MTT assay. MTT was addedto the cells at a final concentration of 0.5 mg/ml andincubated for 4 hours at 37°C. Purple formazan crystalswere produced from the MTT by metabolically activecells. The crystals were solubilized by incubating themwith a solubilization solution provided by the manufac-

turer. Absorbance was read at 540 nm with a microtiterplate spectrophotometer.

Effect of Exogenous PGE2 on IL-1ββ–Stimulated IL-6Production by Fibroblasts Exposed to COXInhibitorsThese experiments were carried out as described abovefor determination of the effect of COX inhibitors onIL-1β-stimulated IL-6 and PGE2 production, except thatin some wells exogenous PGE2

¶ (5-1000 nM) wasadded, to determine if any inhibitory effects of theCOX inhibitors on IL-6 production could be reversed.The cells were pre-incubated for 30 minutes with COXinhibitors ± PGE2 before activation with IL-1β.44

Statistical AnalysisAll experiments were performed in duplicate or triplicateand repeated at least three times. The data wereexpressed as mean ± standard deviation and were ana-lyzed using a one-way analysis of variance (ANOVA).

RESULTSIn preliminary experiments, the effects of cell number,incubation time, and use of serum on consitiutive andIL-1β-stimulated IL-6 production were determined. Eas-ily detectable amounts of IL-6 in conditioned media wereproduced by 2.5 × 104 cells after an incubation periodof 24 hours10,12,46-48 (data not shown). Amounts of con-stitutive and IL-1β-stimulated IL-6 produced were usu-ally greater in the presence of 1% NCS than in serum-freemedium (data not shown). Others have found that bovineserum affects the production of IL-6 by human gingivalfibroblasts.46 In order to avoid factors in serum thatincrease IL-6 production and to clearly focus on theeffect on activation of the fibroblasts with IL-1β, subse-quent experiments were performed under serum-freeconditions. Other studies examining IL-6 productionby human gingival fibroblasts have used serum-freemedium.10,49 The effects of a range of IL-1β concen-trations (1 × 10−13 to 1 × 10−9 M) on IL-6 and PGE2production by all fibroblast cell lines were also deter-mined in preliminary work. In definitive experiments, 1 ×10−11 M IL-1β was used for fibroblast activation becauseit caused large increases in both IL-6 and PGE2 produc-tion by all fibroblast cell lines, approaching maximumstimulation (data not shown). Other studies investigat-ing the effects of IL-1β on human gingival fibroblast pro-duction of cytokines including IL-6 have activated thecells with 5 × 10−11 to 5 × 10−10 M IL-1β.47-50

Effect of DMSO and COX Inhibitors on CellViabilityThe effects of COX-2 inhibitors, at the concentrationschosen for further experimentation as described below,

†† Human DuoSet, R&D Systems, Minneapolis, MN.‡‡ R&D Systems.§§ Boehringer Mannheim Corp., Indianapolis, IN.�� Becton Dickinson, Lincoln Park, NJ.

3029.qxd 1/9/04 11:02 AM Page 1756

1757

J Periodontol • December 2003 Tipton, Flynn, Stein, Dabbous

Table 1.

Effect of DMSO and COX Inhibitors onCell Viability

Treatment Viability ± SD

DMEM-gent 100 0

DMEM-gent/DMSO 103 8

NS-398 (10 nM) 111 8

Celecoxib (1 µM) 93* 3

Rofecoxib (5 µM) 102 11

Indomethacin (1 µM) 93 9

Indomethacin (2.5 µM) 99 10

Indomethacin (5 µM) 98 2

Indomethacin (10 µM) 110 11

Indomethacin (25 µM) 103 11

Indomethacin (50 µM) 108 7

Indomethacin (100 µM) 109† 5

Indomethacin (250 µM) 106 4

Indomethacin (500 µM) 105 7

Fibroblasts (2.5 or 5 × 104) were incubated for 24 hours in DMEM-gent, DMEM-gent/DMSO, or DMEM-gent/DMSO containing the indicated concentrations ofCOX inhibitors. DMSO was 1% in all cases. Cell viability was determined by theMTT assay, which measures the ability of the cells to reduce the tetrazoliumsalt (3-[4, 5-dimethylthiazol-2-yl]-2, 5-diphenyl tetrazolium bromide) (MTT) toa formazan dye. Results are expressed as % control (A540 of cells exposed toDMEM-gent only).* P = 0.01.† P = 0.04.

a range of indomethacin concentrations, as well as con-trol media (DMEM-gent and DMEM-gent/DMSO [1%])on fibroblast viability was determined by measuringthe cells’ ability to reduce MTT (Table 1). DMEM-gent/DMSO, as well as NS-398 (10 nM) or rofecoxib (5 µM),had no significant effect on cell viability, compared toDMEM-gent. There was a slight decrease in viabilitycaused by 1 µM celecoxib (7 ± 3%; P ≤0.01). Indo-methacin at all concentrations tested generally hadno significant effect on fibroblast viability, with theexception of 100 µM (increase of 9 ± 5%; P = 0.04).The number of cells that were used in these experi-ments was within a range of cells in which there is alinear relationship between cell number and absorbancevalues. In preliminary experiments, we observed micro-scopically that ≥2% DMSO in DMEM-gent appeared todamage many cells, causing morphological changesand release from the tissue culture wells into the med-ium, whereas concentrations below that did not (datanot shown).

Effects of COX Inhibition on Constitutive andIL-1ββ–Stimulated PGE2 and IL-6 ProductionThe effects of indomethacin (1-500 µM), NS-398(1-1000 nM), celecoxib (0.25-5 µM) or rofecoxib (1-50µM) on IL-1β (1 × 10−11 M)-stimulated PGE2 and IL-6production by one representative cell line, GF 3, werefirst determined. Unstimulated fibroblasts produced rel-atively low levels of IL-6 and PGE2, and the COXinhibitors had inconsistent but relatively minor effectson this (data not shown). Others have found thatindomethacin had minimal effects on constitutive fibrob-last (synovial) production of cytokines, including IL-6.18 This finding was not unexpected, particularly in thecase of the COX-2 inhibitors. Nearly all normal tissuesexpress COX-1, but only at low to undetectable levelsof COX-2.51 COX-2 is expressed in most tissues underinflammatory conditions. Therefore, in subsequentexperiments, we focused on the effect of the COX-inhibitors on IL-1β-stimulated PGE2 and IL-6 production.

Figures 1 through 4 show the effects of COX inhibitorson IL-1β-stimulated IL-6 and PGE2 production by GF 3.Results were calculated as ng IL-6 or PGE2/mg protein,and converted to a percentage of the amount in controlmedium (DMEM-gent + IL-1β). Indomethacin (1-500µM) did not inhibit IL-6 production, and at some con-centrations (25, 50, and 500 µM) significantly (P ≤0.03)increased it (Fig. 1). PGE2 production was almost com-

Figure 1.Effect of indomethacin on IL-1β-stimulated IL-6 and PGE2 production byGF 3 fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±indomethacin (1 to 500 µM) in serum-free DMEM for 24 hours.Amounts of IL-6 and PGE2 in culture supernatants were measured byELISA.The mean values and standard deviations of triplicatemeasurements in representative experiments are indicated, calculatedas ng IL-6 or PGE2/mg protein in the cell monolayer, and converted to apercentage of the amount in control medium (DMEM-gent + IL-1β).

3029.qxd 1/9/04 11:02 AM Page 1757

1758

Downregulation of PGE2 and IL-6 by COX-2 Inhibitors Volume 74 • Number 12

Figure 2.Effect of NS-398 on IL-1β-stimulated IL-6 and PGE2 production by GF3 fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±NS-398 (1 to 1000 nM) in serum-free DMEM for 24 hours.Amountsof IL-6 and PGE2 in culture supernatants were measured by ELISA.Themean values and standard deviations of triplicate measurements inrepresentative experiments are indicated, calculated as ng IL-6 orPGE2/mg protein in the cell monolayer, and converted to a percentageof the amount in control medium (DMEM-gent + IL-1β).

Figure 3.Effect of celecoxib on IL-1β-stimulated IL-6 and PGE2 production byGF 3 fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±celecoxib (0.25 to 5 µM) in serum-free DMEM for 24 hours.Amountsof IL-6 and PGE2 in culture supernatants were measured by ELISA.Themean values and standard deviations of triplicate measurements inrepresentative experiments are indicated, calculated as ng IL-6 orPGE2/mg protein in the cell monolayer, and converted to a percentageof the amount in control medium (DMEM-gent + IL-1β).

Figure 4.Effect of rofecoxib on IL-1β-stimulated IL-6 and PGE2 production byGF 3 fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±rofecoxib (1 to 50 µM) in serum-free DMEM for 24 hours.Amounts ofIL-6 and PGE2 in culture supernatants were measured by ELISA.Themean values and standard deviations of triplicate measurements inrepresentative experiments are indicated, calculated as ng IL-6 orPGE2/mg protein in the cell monolayer, and converted to a percentageof the amount in control medium (DMEM-gent + IL-1β).

pletely inhibited by indomethacin concentrations ≥1 µM(P = 0.0001) (Fig. 1). NS-398 (1-1000 nM) reduced IL-6 production by a maximum of approximately 60% (P≤0.001) while PGE2 production was reduced at all con-centrations tested (P = 0.0001) (≥90% at concentrations≥10 nM) (Fig. 2). Celecoxib also decreased productionof IL-6 by a maximum of approximately 60% (P ≤0.003)and PGE2 production by ≥99% at concentrations ≥0.25µM (P = 0.0001) (Fig. 3). Rofecoxib decreased IL-6 pro-duction generally by 50% to 70% (P = 0.0001), anddecreased PGE2 production by >98% at concentrations≥1 µM (P = 0.0001) (Fig. 4). Because indomethacin didnot inhibit IL-6 production, subsequent experimentsfocused on the COX-2 inhibitors, which all inhibited bothPGE2 and IL-6 production. Concentrations chosen forfurther experimentation were 10 nM NS-398, 1 µM cele-coxib, and 5 µM rofecoxib.

All fibroblast cell lines were then tested for the effectof 10 nM NS-398, 1 µM celecoxib, or 5 µM rofecoxib onIL-1β-stimulated IL-6 and PGE2 production. IL-6 produc-tion by all cell lines was significantly inhibited by NS-398(25% to 60%; P ≤0.02), celecoxib (20% to 60%; P ≤0.001)or rofecoxib (10% to 70%; P ≤0.03) (Fig. 5). PGE2 pro-duction by all cell lines was also inhibited, but to agreater extent than IL-6, by NS-398 (70% to 90%; P≤0.004), celecoxib (70% to 95%; P ≤0.0001) or rofe-coxib (65% to 95%; P ≤0.0001) (Fig. 6).

3029.qxd 1/9/04 11:02 AM Page 1758

1759

J Periodontol • December 2003 Tipton, Flynn, Stein, Dabbous

Effects of COX Inhibitors Plus Exogenous PGE2 onIL-1ββ–Stimulated Fibroblast IL-6 ProductionThe ability of exogenous PGE2 to reverse the inhibitionof IL-1β-stimulated IL-6 production caused by the COXinhibitors was determined. In these experiments, all

fibroblast cell lines were incubated with spe-cific COX-2 inhibitors ± exogenous PGE2 (1µM) 30 minutes prior to IL-1β (1 × 10−11 M)stimulation. In initial trials with the represen-tative cell line GF 3, exogenous PGE2 (5 nM,50 nM, 500 nM, and 1 µM) caused dose-dependent increases in IL-6 production by cellsexposed to COX-inhibitors and activated withIL-1β (data not shown). Because 1 µM PGE2returned levels of IL-6 to control levels orabove, this concentration was used in subse-quent experiments. This amount of PGE2 wassimilar to that detected in culture supernatantsof fibroblasts stimulated with IL-1β alone inour and other work,52 and was also in therange of exogenous PGE2 added to cell cul-tures in similar studies.18,19,53

The effects of NS-398 (10 nM), celecoxib(1 µM) or rofecoxib (5 µM) ± exogenous PGE2on IL-1β-stimulated IL-6 production by allfibroblast strains are shown in Table 2. Therewas a decrease in IL-6 production by the AgPfibroblasts ranging from 15% to 40% for thethree COX-2 inhibitors. These decreases werestatistically significant compared to control (IL-1β-stimulated cells), except for GF 3/celecoxiband GN/CN/NS-398. When exogenous PGE2was added concurrently with the COX-2inhibitors before the addition of IL-1β, IL-6 pro-duction returned to levels at or approachingthat produced by cells exposed only to IL-1β.The increases in IL-6 production caused bythe exogenous PGE2 were statisticallysignificant compared to respective controlcultures exposed to the COX inhibitors only(i.e., N versus N + PGE2; C versus C + PGE2;R versus R + PGE2), except in the case of GF2 and GF 3 and celecoxib (which was nearstatistical significance at P = 0.07). There werealso similar trends for the normal fibroblast celllines. In these cell lines, decreases caused bythe COX-2 inhibitors in IL-6 production rangedfrom 10% to 60% and were statistically signif-icant in all cases. Addition of exogenous PGE2restored IL-6 production to control levels, againstatistically significant when compared to cul-tures exposed to the COX inhibitor only.

DISCUSSIONIn this study the relationship between PGE2and IL-6 production in normal and severe peri-

odontitis gingival fibroblasts was examined. BecausePGE2 elevates IL-6 production by human gingival fibro-blasts, the ability of selective (COX-2) or non-selectiveCOX-inhibitors to reduce fibroblast production of PGE2and in turn, IL-6 was determined. The results demon-

Figure 5.Effect of COX-2 inhibitors on IL-1β-stimulated IL-6 production by normal and severeperiodontitis gingival fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±NS-398 (10 nM), celecoxib (1 µM), or rofecoxib (5 µM) in serum-free DMEM for 24hours.Amounts of IL-6 in culture supernatants were measured by ELISA.The meanvalues and standard deviations of triplicate measurements in representativeexperiments are indicated, calculated as ng IL-6/mg protein in the cell monolayer, andconverted to a percentage of the amount in control medium (DMEM-gent + IL-1β).

Figure 6.Effect of COX-2 inhibitors on IL-1β-stimulated PGE2 production by normal and severeperiodontitis gingival fibroblasts. 2.5 × 104 cells were exposed to 1 × 10−11 M IL-1β ±NS-398 (10 nM), celecoxib (1 µM), or rofecoxib (5 µM) in serum-free DMEM for24 hours.Amounts of PGE2 in culture supernatants were measured by ELISA.Themean values and standard deviations of triplicate measurements in representativeexperiments are indicated, calculated as ng PGE2/mg protein in the cell monolayer, andconverted to a percentage of the amount in control medium (DMEM-gent + IL-1β).

3029.qxd 1/9/04 11:02 AM Page 1759

1760

Downregulation of PGE2 and IL-6 by COX-2 Inhibitors Volume 74 • Number 12

strate that the production of IL-1β-stimulated PGE2 andIL-6 by normal and severe periodontitis gingival fibro-blasts is significantly decreased by treating them withspecific COX-2 inhibitors. The MTT assay showed thatDMSO or the COX inhibitors themselves did notadversely affect cell viability and that the observeddecreases in IL-6 and PGE2 were not due to toxicity ofthe solutions used. The addition of exogenous PGE2restored the IL-1β stimulated IL-6 production by fibro-blasts treated with COX-2 inhibitors to control levels(amounts of IL-6 produced by fibroblasts exposed toIL-1β only), suggesting that inhibition of PGE2 and IL-6 were related.

The two periodontitis cell lines were derived frompatients with different forms of severe periodontitis and,therefore, could exhibit phenotypic differences and dif-fer in their response to certain exogenous agents. It isalso well known that cytokines produced by inflamma-tory cells infiltrating a tissue or organ can activate resi-dent mesenchymal cells such as fibroblasts.42,54,55 Theconstitutive functions and the responses of these two celllines, and indeed the normal gingival fibroblast cell lines,to the drugs and other exogenous agents used in thisstudy, appear to be similar. However, inclusion of fibro-blasts from inflamed gingiva, without attachment loss, ina study of this type would offer the advantage of morefully assessing the activation state of gingival fibroblastsderived from types of severe periodontitis gingiva andhow this contributes to the extreme attachment loss char-acteristic of these diseases.

IL-1β was used as a stimulus because it increases

both PGE2 and IL-6 production by human gin-gival fibroblasts,12,34,35 and because the rela-tionship between IL-1β and the severity ofperiodontal disease is well documented.36,56,57

IL-1β increases PGE2 production by stimu-lating release of arachadonic acid and induc-ing COX-2 expression.26,58 Other investigatorshave found that increases in IL-6 productionby IL-1β-stimulated human gingival fibro-blasts occur in conjunction with enhancedCOX-2 expression and subsequent PGE2 pro-duction.50 The work described in the presentstudy is consistent with these results, in thatIL-1β increased the production of PGE2 andIL-6 by all the fibroblast cell lines. The severeperiodontal destruction characteristic of AgPhas been linked to increased levels of PGE2and tissue destructive cytokines including IL-2, IL-4 and IL-6.16,59 Dongari-Bagtzoglou andEbersole found that AgP and normal fibro-blasts constitutively produced similar amountsof IL-6, but IL-1β-stimulated AgP fibroblastsproduced greater amounts of IL-6 than IL-1β-stimulated normal fibroblasts.60 Work in ourlaboratory has suggested that Aa LPS stimu-

lation has similar effects on IL-6 production by normaland AgP gingival fibroblasts.

Numerous studies have evaluated effects of non-selective COX inhibitors on PGE2 and IL-6 productionby a variety of cells. In the present study, IL-1β-stimulatedPGE2 production was reduced by nearly 90% or moreupon treatment of all cell lines with indomethacin. Con-sistent with these results, other investigators found thatindomethacin inhibited PGE2 in periodontal ligamentcells, Th lymphocytes, astrocytes, and IL-1β-stimulatednormal and chronic periodontitis human gingival fibro-blasts.10,19,44,61-63 However, while indomethacin inhib-ited fibroblast PGE2 production in our study, it did notinhibit IL-6 production, and at some concentrationsincreased it. There are conflicting reports of the effectsthat non-selective COX inhibitors, including indometha-cin, have on IL-6 production by a number of cell types.Indomethacin decreased IL-1β-stimulated IL-6 produc-tion by 20% in chronic periodontitis human gingivalfibroblasts,19 but other investigators found that indo-methacin did not inhibit IL-6 production by cells such asmononuclear cells, chondrocytes, and porcine andhuman synovial fibroblasts.63-65 In the present study,some concentrations of indomethacin increased IL-6production by the GF 3 cell line, and Takigawa et al.48

also noted similar increases in IL-6 production whentreating IL-1β-stimulated normal human gingival fibro-blasts with indomethacin. Different cell types, fibroblastsfrom different tissues, and gingival fibroblasts from agiven individual are heterogeneous in many respects,including cytokine production and response to drugs.

Table 2.

Effects of COX Inhibitors ±± Exogenous PGE2 onIL-1ββ–Stimulated Fibroblast IL-6 Production

Treatment

Cell Line‡ N* C* R* N + PG† C + PG† R + PG†

GF 2 74 ± 9 73 ± 12 67 ± 12 112 ± 18 89 ± 12 105 ± 9

GF 3 79 ± 10 86 ± 13 82 ± 8 115 ± 10 112 ± 13 104 ± 9

GN/CN 75 ± 23 72 ± 15 57 ± 25 124 ± 18 130 ± 6 154 ± 20

GN 46 84 ± 5 88 ± 4 82 ± 9 126 ± 21 102 ± 7 101 ± 6

GN 60 71 ± 15 82 ± 7 67 ± 14 101 ± 8 122 ± 18 101 ± 5

GN 62 69 ± 8 62 ± 6 40 ± 3 97 ± 15 107 ± 3 84 ± 3

Fibroblasts (2.5 × 104) were incubated for 24 hours in DMEM-gent/DMSO containing IL-1β(10−11M), with or without NS-398 (N) (10 nM), celecoxib (C) (1 µM), or rofecoxib (R) (5 µM).In some wells exogenous PGE2 (1 µM) was added. The cells were pre-incubated for 30 min-utes with COX inhibitors ± PGE2 before activation with IL-1β. The media were harvested andassayed, and results were calculated as ng IL-6 per mg protein and then expressed as %control (amount in control medium [DMEM-gent + IL-1β]).* P ≤0.04 except GF 3 + C and GN/CN + N† P ≤0.04 except GF 2 + C and GF 3 + C (P = 0.07)‡ GF 2, GF 3, and GN/CN are gingival fibroblast cell lines from severe periodontitis patients;GN 46, GN 60 and GN 62 are normal human gingival fibroblast cell lines.

3029.qxd 1/9/04 11:02 AM Page 1760

1761

J Periodontol • December 2003 Tipton, Flynn, Stein, Dabbous

One explanation for the inability of indomethacin todecrease IL-6 production by the fibroblast cell lines usedin this study may be related to its effect on regulationof nuclear factor (NF)-κB and IL-6 expression.

NF-κB is a transcription factor that regulates theexpression of over 150 genes for cytokines, chemokines,growth factors, pro-inflammatory enzymes, immunore-ceptors, and cell adhesion molecules, in response tostimuli such as IL-1β, bacterial LPS, or viruses.66-68

Chemotherapeutics can modulate NF-κB activity. Forexample, aspirin and the COX-2 inhibitors NS-398 andcelecoxib, but not indomethacin, inhibit NF-κB in lym-phocytes and dendritic cells.45,68,69 COX-2 gene acti-vation in macrophages and human gingival fibroblastsappears to be dependent upon NF-κB activation.70,71 Weare currently investigating the effects of indomethacinor NS-398 on NF-κB in AP and normal gingival fibro-blasts, to determine if decreased PGE2 production anddecreased NF-κB activities are both necessary for thepartial decreases in IL-6 that we observed. The inhibi-tion of only PGE2 by indomethacin that we observed,coupled with no effect by this drug on NF-κB and itssubsequent contribution to IL-6 production, may resultin no observed decrease in IL-6.

COX inhibitors decrease IL-6 production in humancells, to different extents depending on the type ofcell,18,53,64 and our results are consistent with those ofseveral other investigators who have studied the effectsof COX-2 inhibitors on IL-6 and PGE2 production.44,72,73

Inhibition of PGE2 by selective COX-2 inhibitors maybe only part of the mechanism by which these drugsdecrease IL-6 production, via disruption of this autocrinestimulatory loop. Matsui et al.74 found that endogenousPGE2 is an autocrine stimulus for production of IL-11(a member of the IL-6-type cytokine family) by acti-vated human lung fibroblasts. Blom et al.61 describeddecreases in PGE2 and IL-6 that were similar to thosein our study (maximum IL-6 decrease of 47%) in IL-1β-stimulated astrocytes treated with the COX-2 inhibitorBF389. Recent investigations have suggested that COXinhibitors have targets other than COX that can accountfor some of their actions.45

The results of this study and those of other investi-gators suggest that it is likely that PGE2 produced viaCOX-2 by IL-1β-stimulated gingival fibroblasts con-tributes to their production of IL-6, and that both PGE2and IL-6 are involved in the destruction of hard and softperiodontal tissue in AgP and other forms of severe peri-odontitis. NSAIDS that inhibit PGE2 production are effec-tive in preventing periodontal breakdown in humans.27,28

Our observed partial reduction in IL-6 coupled with thenearly 90% + reduction in PGE2 suggests that COX-2inhibitors may help reduce inflammation and boneresorption when treating severe periodontitis. The con-centrations of celecoxib used in this study ranged from0.25 to 5 µM (1 µM in most experiments). Peak plasma

levels of celecoxib (approximately 1 to 4.6 µM) are lin-early related to dose within the therapeutic range of 100to 600 mg, and 70% to 80% COX-2 inhibition is attainedby doses of 100 to 800 mg, corresponding to peakplasma levels of 1 to 7.2 µM.75 The concentrations ofrofecoxib used in this study ranged from 1 to 50 µM,and 5 µM reofecoxib was used in most experiments.Peak plasma levels of rofecoxib after therapeutic dosesof 12.5 to 50 mg can reach approximately 1 µM andcan inhibit PGE2 >90%.76 Peak plasma concentrationsof indomethacin range from approximately 3 µM to 12µM for doses of 25, 50, or 75 mg, respectively,77 and inthis study we used from 1 to 500 µM indomethacin.Therefore, the experimental concentrations of the COXinhibitors studied in this work, which are used in humans,are similar to their plasma levels achieved in vivo. Futureinvestigations into the role of NF-κB and its subsequentinhibition with chemotherapeutics could provide a ration-ale for combination drug therapy to further decrease thehard and soft tissue destruction by IL-6 in aggressiveperiodontitis.

ACKNOWLEDGMENTSThe authors thank Pharmacia Corporation, St. Louis, Mis-souri for its gift of celecoxib, and Merck and Company,Inc., Rahway, New Jersey for its gift of rofecoxib. Thiswork was supported by a grant from the Alumni Endow-ment Fund of the University of Tennessee College ofDentistry and was done in partial fulfillment of the require-ments for the degree of Master of Dental Science (JF).Dr. Flynn was the recipient of the 2001 Pennel ResearchCompetition award from the Southern Academy of Peri-odontology and thanks the Academy for its support.

REFERENCES1. Birkedal-Hansen H. Role of matrix metalloproteinases

in human periodontal disease. J Periodontol 1993;64:474-484.

2. Page RC. The role of inflammatory mediators in the patho-genesis of periodontal disease. J Periodont Res 1999;26:230-242.

3. Page RC, Kjeldsen M, Holmstrup M, Bendtzen K. Marginalperiodontitis and cytokines: A review of literature. J Peri-odontol 1993;64:1013-1022.

4. Birkedal-Hansen H. Role of cytokines and inflammatorymediators in tissue destruction. J Periodont Res 1993;28:500-510.

5. Irwin CR, Myrillas TT. The role of IL-6 in the pathogen-esis of periodontal disease. Oral Dis 1998;4:43-47.

6. Ishimi Y, Miyaura C, Jin CH, et al. IL-6 is produced byosteoblasts and induces bone resorption. J Immunol1990;145:3297-3303.

7. Geivelis M, Tuner DW, Pederson ED, Lamberts BL. Mea-surements of interleukin-6 in gingival crevicular fluidfrom adults with destructive periodontal disease. J Peri-odontol 1993;63:980-983.

8. Tamura T, Udagawa N, Takahashi N, et al. Solubleinterleukin-6 receptor triggers osteoclast formation byinterleukin 1. Proc Natl Acad Sci (USA) 1993;90:11924-11928.

3029.qxd 1/9/04 11:02 AM Page 1761

1762

Downregulation of PGE2 and IL-6 by COX-2 Inhibitors Volume 74 • Number 12

9. Guillot JL, Pollock SM, Johnson RB. Gingival interleukin-6concentration following phase I therapy. J Periodontol1995;66:667-672.

10. Kent LW, Rahemtulla F, Michalek SM. Interleukin (IL)-1and Porphyromonas gingivalis lipopolysaccharide stim-ulation of IL-6 production by fibroblasts derived fromhealthy or periodontally diseased human gingival tissue.J Periodontol 1999;70:274-282.

11. Takahashi K, Takashiba S, Nagai A, et al. Assessmentof interleukin-6 in the pathogenesis of periodontal dis-ease. J Periodontol 1994;65:147-153.

12. Dongari-Bagtzoglou AI, Ebersole JL. Increased pres-ence of interleukin-6 (IL-6) and IL-8 secreting sub-populations in adult periodontitis. J Periodontol 1998;69:899-910.

13. Agarwal S, Baran C, Piesco NP, et al. Synthesis of proin-flammatory cytokines by human gingival fibroblasts inresponse to lipopolysaccharides and interleukin-1 beta.J Periodontal Res 1995;30:382-389.

14. Offenbacher S, Heasman PA, Collins JG. Modulation ofhost PGE2 secretion as a determinant of periodontal dis-ease statement. J Periodontol 1993;64:432-444.

15. Offenbacher S. Periodontal diseases: Pathogenesis. AnnPeriodontol 1996;1:821-878.

16. Offenbacher S, Odle BM, Gray RC, Van Dyke TE. Crevic-ular fluid prostaglandin E levels as a measure of theperiodontal disease status of adult and juvenile peri-odontitis patients. J Periodontal Res 1984;19:1-13.

17. Offenbacher S, Collins JG, Yalda B, Haradon G. Role ofprostaglandins in high-risk periodontitis patients. In: GencoRJ, Hamada S, Lehner T, Mergenhagen S, eds. MolecularPathogenesis of Periodontal Disease. Washington DC: ASMPress; 1994:203-214.

18. Agro A, Langdon C, Smith F, Richards CD. ProstaglandinE2 enhances interleukin 8 (IL-8) and IL-6 but inhibitsGMCSF production by IL-1 stimulated human gingivalfibroblasts. J Rheumatol 1996;23:862-868.

19. Czuszak CA, Sutherland DE, Billman MA, Stein SH.Prostaglandin E2 potentiates interleukin-1β induced inter-leukin-6 production by human gingival fibroblasts. J ClinPeriodontol 1996;23:635-640.

20. Waite IM, Saxon CA, Young A, Wagg BJ, Corbett M. Theperiodontal status of subjects receiving non-steroidal anti-inflammatory drugs. J Periodontal Res 1981;16:100-108.

21. Williams RC, Jeffcoat MK, Howell TH, et al. Altering theprogression of human alveolar bone loss with the non-steroidal anti-inflammatory drug flurbiprofen. J Peri-odontol 1989;60:485-490.

22. Vane JR. Towards a better aspirin. Nature (London)1984;367:215-216.

23. Eberhart CE, Dubois RN. Eicosanoids and the gastro-intestinal tract. Gastroenterol 1995;109:285-301.

24. Smith WL. Prostanoid biosynthesis and mechanism ofaction. Am J Physiol 1992;263:F181-F191.

25. Wang S, Sun C, Gillanders. Genome scan for suscepti-bility to the complex disorder early onset periodontitis.Am J Hum Genet 1996;59(Suppl.):A240(Abstr. 1386).

26. Smith WL, DeWitt DL. Prostaglandin endoperoxide Hsynthase-1 and -2. Adv Immunol 1996;62:167-215.

27. Griswold DE, Adam JE. Constitutive cyclooxygenase(COX-1) and inducible cyclooxygenase (COX-2): Ration-ale for selective inhibition and progress to date. MedRes Rev 1996;16:181-206.

28. Kreamer SA, Meada EA, DeWitt DL. Prostaglandinendoperoxide synthase gene structure: Identification of thetranscriptional start site and the 5’-flanking regulatorysequences. Arch Biochem Biophys 1992;293:391-400.

29. Kujubu DA, Hershman HR. Dexamethasone inhibitsmitogen induction of the TIS10 prostaglandin synthase/cyclooxygenase gene. J Biol Chem 1992;267:7991-7994.

30. O’Neill GP, Ford-Hutchison AW. Statement of mRNA forcyclooxygenase-1 and cyclooxygenase-2 in human tis-sues. FEBS Lett 1993;330:156-160.

31. Hla T, Neilson K. Human cyclooxygenase-2 cDNA. ProcNatl Acad Sci (USA) 1992;89:7384-7388.

32. Jones DA, Carlton DP, McIntyre TM, Zimmerman GA,Prescott SM. Molecular cloning of human prostaglandinendoperoxide synthase type II and demonstration of state-ment in response to cytokines. J Biol Chem 1993;268:9049-9054.

33. Lee SH, Soyoola E, Chanmugam P, et al. Selective state-ment of mitogen-inducible cyclooxygenase in macrophagestimulated with lipopolysaccharide. J Biol Chem 1992;267:25934-25938.

34. Morton RS, Dongari-Bagtzoglou AI. Regulation of gingi-val fibroblast interleukin-6 secretion by cyclosporine A.J Periodontol 1999;70:1464-1471.

35. Noguchi K, Shitashige M, Endo H, Kondo H, Ishikawa I.Binary regulation of interleukin (IL)-6 production by EP1and EP2/EP4 subtypes of PGE2 receptors in IL-1beta-stimulated human gingival fibroblasts. J Periodontal Res2002;37:29-36.

36. Kornman KS, Crane A, Wang HY, et al. The interleukin-1genotype as a severity factor in adult periodontal disease.J Clin Periodontol 1997;24:72-77.

37. Hempel SL, Martha H, Monick M, Hunninghake GW.Lipopolysaccharide induces prostaglandin H synthase-2protein and mRNA in human alveolar macrophages andblood monocytes. J Clin Invest 1994;93:391-396.

38. Yucel-Lindberg T, Ahola H, Nilsson S, Carlstedt-Duke J,Modeer T. Interleukin-1β induces statement of cyclooxy-genase-2 mRNA in human gingival fibroblasts. Inflam-mation 1995;19:49-56.

39. Noguchi K, Shitashige M, Yanai M, et al. Prostaglandinproduction via induction of cyclooxygenase-2 by humangingival fibroblasts stimulated with lipopolysaccharide.Inflammation 1996;20:555-568.

40. Mitchell JA, Akaraseereenont P, Thiemermann C, Flower RJ,Vane JR. Selectivity of nonsteroidal anti-inflammatory drugssuch as inhibitors of constitutive and inducible cyclooxy-genase. Proc Natl Acad Sci (USA) 1994;90:11693-11697.

41. Jouzeau J-Y, Terlain B, Abid A, Nedelec E, Netter P.Cyclooxygenase isoenzymes; how recent findings affectthinking about non steroidal anti-inflammatory drugs.Drugs 1997;53:563-582.

42. Tipton DA, Pabst MJ, Dabbous MKh. Interleukin-1ß- andtumor necrosis factor α-independent monocyte stimu-lation of fibroblast collagenase activity. J Cell Biochem1990;44:253-264.

43. Lowry O, Rosebrough N, Farr A, Randall R. Protein mea-surement with the folin-phenol reagent. J Biol Chem1951;193:265-275.

44. Iniguez MA, Punzon C, Fresno M. Induction of cyclooxy-genase-2 on activated T lymphocytes: Regulation of T cellactivation by cyclogenase-2 inhibitors. J Immunol 1999;163:111-119.

45. Noguchi K, Shitashige M, Ishikawa I. Involvement ofcyclooxygenase-2 in interleukin-1α-induced prostaglandinproduction by human gingival periodontal ligament cells.J Periodontol 1999;70:902-908.

46. Dongari-Bagtzoglou AI, Ebersole JL. Production of inflam-matory mediators and cytokines by human gingivalfibroblasts following bacterial challenge. J PeriodontalRes 1996;31:90-98.

3029.qxd 1/9/04 11:02 AM Page 1762

1763

J Periodontol • December 2003 Tipton, Flynn, Stein, Dabbous

47. Myrillas TT, Linden GJ, Marley JJ, Irwin CR. CyclosporinA regulates interleukin-1beta and interleukin-6 expres-sion in gingiva: Implications for gingival overgrowth. JPeriodontol 1999;70:294-300.

48. Takigawa M, Tahashiba S,Takahashi K, Arai H, KuriharaH, Murayama Y. Prostaglandin E2 inhibits interleukin-6release but not its transcription in human gingival fibro-blasts stimulated with interleukin-1 beta or tumor necro-sis factor-alpha. J Periodontol 1994;65:1122-1127.

49. Kent LW, Rahemtulla F, Hockett RD, Gilleland RC,Michalek SM. Effect of lipopolysaccharide and inflam-matory cytokines on interleukin-6 production by healthyhuman gingival fibroblasts. Inf Immun 1998;66:608-614.

50. Yucel-Lindberg T, Nilsson S, Modeer T. Signal trans-duction pathways involved in the synergistic stimulationof prostaglandin production by interleukin-1beta andtumor necrosis factor alpha in human gingival fibroblasts.J Dent Res 1999;78:61-68.

51. Fosslien E. Biochemistry of cyclooxygenase (COX-2)-2 inhibitors and molecular physiology of COX-2 in neo-plasia. Crit Rev Clin Lab Sci 2000;37:431-502.

52. Noguchi K, Iwasaki K, Shitashige M, Endo H, Kondo H,Ishikawa I. Cyclooxygenase-2-dependent prostaglandinE2 down-regulates intercellular adhesion molecule-1expression via EP2/EP4 receptors in interleukin-1beta-stimulated human gingival fibroblasts. J Dent Res 2000;79:1955-1961.

53. Takigawa M, Takashiba S, Myokai F, et al. Cytokine-dependent synergistic regulation of interleukin-8 produc-tion from human gingival fibroblasts. J Periodontol 1994;65:1002-1007.

54. Narayanan A, Voss T, Page R. Transcriptional regulationof fibroblast collagen synthesis by activated mononu-clear cell supernatants. Biochem Biophys Res Commun1985;131;1028-1032.

55. Postlethwaite A, Smith G, Mainardi C, Seyer J, Kang A.Lymphocyte modulation of fibroblast function in vitro:Stimulation and inhibition of collagen production bydifferent effector molecules. J Immunol 1984;132:2470-2477.

56. Cullinan MP, Westerman B, Hamlet SM, et al. A longi-tudinal study of interleukin-1 gene polymorphisms andperiodontal disease in a general adult population. J ClinPeriodontol 2001;28:1137-1144.

57. Thomson WM, Edwards SJ, Dobson-Le DP, et al. IL-1genotype and adult periodontitis among young NewZealanders. J Dent Res 2001;80:1700-1703.

58. Modeer T, Yucel-Lindberg T, Iinuma M, Lerner UH,Andersson G. Epidermal growth factor potentiates inter-leukin 1 and tumor necrosis factor-induced prostaglandinbiosynthesis in human gingival fibroblasts. Cytokine1993;5:198-204.

59. Manhart SS, Reinhardt RA, Payne JB, et al. Gingival cellIL-2 and IL-4 in early-onset periodontitis. J Periodontol1994;65:807-813.

60. Dongari-Bagtzoglou AI, Ebersole JL. Gingival fibroblastcytokine profiles in Actinobacillus actinomycetemcomitans-associated periodontitis. J Periodontol 1996;67:871-878.

61. Blom MA, van Twillert MG, de Vries SC, et al. NSAIDSinhibit the IL-1 beta-induced IL-6 release from human post-mortem astrocytes: The involvement of prostaglandin E2.Brain Res 1997;777:210-218.

62. Tsuboi I, Tanaka H, Nakao M, Shichijo S, Itoh K. Non-steroidal anti-inflammatory drugs differentially regulate

cytokine production in human lymphocytes and upregu-lation of TNFα, INFγ and IL-2 in contrast to downregu-lation of IL-6 production. Cytokine 1995;7:372-379.

63. Rainsford KD, Ying C, Smith FC. Effects of meloxicam,compared with other NSAIDs, on cartilage proteoglycanmetabolism, synovial prostaglandin E2 and production ofinterleukin 1, 6, and 8, in human and porcine explantsin organ culture. J Pharm Pharmacol 1997;49:991-998.

64. Sirota L, Punsky I, Bessler H. Effect of indomethacin onIL-1beta, IL-6 and TNF alpha production by mononuclearcells of preterm newborns and adults. Acta Paediatr2000;89:331-335.

65. Clausen PA, Flechtenmacher J, Haeuselmann HJ, KuettnerKE, Aydelotte MB, Iyer AP. Evidence of an eicosanoidcontribution to IL-1 induction of IL-6 in human articularchondrocytes. Am J Ther 1996;3:101-108.

66. Pahl HL. Activators and target genes of Rel/NF-kappaBtranscription factors. Oncogene 1999;18:6853-6866.

67. Baeuerle PA, Baltimore D. NFκβ: Ten years after. Cell1996;87:13-20.

68. Matasic R, Dietz AB, Vuk-Pavlovic S. Cyclooxygenase-independent inhibition of dendritic cell maturation byaspirin. Immunology 2000;101:53-60.

69. Cianferoni A, Schroeder JT, Kim J et al. Selective inhi-bition of interleukin-4 gene expression in human T cellsby aspirin. Blood 2001;97:1742-1749.

70. Lo CJ, Cryer HG, Fu M, Lo F. Regulation of macrophageeicosanoid generation is dependent on nuclear factor κβ.J Trauma Inj Inf Crit Care 1998;45:19-23.

71. Nakao S, Ogata Y, Shimizu-Sasaki E, Yamazaki M,Furuyama S, Sugiya H. Activation of NK-κβ is also nec-essary for IL-1β-induced COX-2 expression in humangingival fibroblasts (healthy). Mol Cell Biochem 2000;209:113-121.

72. Morton RS, Dongari-Bagtzoglou AI. Cyclooxygenase-2is upregulated in inflamed gingival tissues. J Periodon-tol 2001;72:461-469.

73. Eisengart CA, Mestre JR, Naama HA, et al. Prostaglandinsregulate melanoma-induced cytokine production inmacrophages. Cell Immunol 2000;204:143-149.

74. Matsui S, Yamashita N, Mino T, et al. Role of the endoge-nous prostaglandin E2 in human lung fibroblast inter-leukin-11 production. Respir Med 1999;93:637-642.

75. McAdam BF, Catella-Lawson F, Mardini IA, Kapoor S,Lawson JA, Fitzgerald GA. Systemic bioysynthesis ofprostacyclin by cyclooxygenase (COX)-2: The humanpharmacology of a selective inhibitor of COX-2. ProcNatl Acad Sci (USA) 1999;96:272-277.

76. Depré M, Ehrich E, Van Hecken A, et al. Pharmacoki-netics, COX-2 specificity, and tolerability of suprathera-peutic doses of rofecoxib in humans. Eur J Clin Pharmacol2000;56:167-174.

77. Bruguerolle B, Barbeau G, Belanger PM, Labrecque G.Pharmacokinetics of a sustained-release product of indo-methacin in the elderly. Gerontology 1986;32:277-285.

Correspondence: Dr. David A. Tipton, Department of Peri-odontology and Dental Research Center, College of Dentistry,University of Tennessee Health Science Center, 894 UnionAvenue, Memphis, TN 38163. Fax: 901/448-7860; e-mail:dtipton@utmem.edu.

Accepted for publication May 2, 2003.

3029.qxd 1/9/04 11:02 AM Page 1763