ARTHRITIS & RHEUMATISMVol. 46, No. 7, July 2002, pp 18041812DOI 10.1002/art.10357 2002, American College of Rheumatology

Study of the Role of Leukotriene B4 in Abnormal Function ofHuman Subchondral Osteoarthritis Osteoblasts

Effects of Cyclooxygenase and/or 5-Lipoxygenase Inhibition

Yosabeth Paredes,1 Frederic Massicotte,1 Jean-Pierre Pelletier,1 Johanne Martel-Pelletier,1

Stefan Laufer,2 and Daniel Lajeunesse1

Objective. To compare the effect of licofelone,NS-398 (an inhibitor of cyclooxygenase 2 [COX-2]), andBayX-1005 (an inhibitor of 5-lipoxygenase activatingprotein) on the production of leukotriene B4 (LTB4) andprostaglandin E2 (PGE2), and on cell biomarkers byhuman osteoarthritis (OA) subchondral osteoblasts.

Methods. Primary in vitro osteoblasts were pre-pared from subchondral bone specimens obtained fromOA patients and autopsy subjects. LTB4 and PGE2levels were measured by enzyme-linked immunosorbentassay in conditioned media of osteoblasts incubated inthe presence or absence of licofelone, NS-398, or BayX-1005. The effect of these drugs or of the addition of LTB4on alkaline phosphatase (AP) activity and osteocalcinrelease by OA and normal osteoblasts was determined.The presence of LTB4 receptors in normal and OAosteoblasts was evaluated by Western blot analysis.

Results. OA osteoblasts produced variable levelsof PGE2 and LTB4 compared with normal osteoblasts.Licofelone, at the maximal dose used, inhibited pro-duction of PGE2 and LTB4 by OA osteoblasts by a

mean SEM of 61.2 6.4% and 67.0 7.6%, respec-tively. NS-398 reduced PGE2 production by 75.8 5.3%. BayX-1005 inhibited LTB4 production in OAosteoblasts by 38.7 14.5% and marginally affectedPGE2 levels (reduction of 14.8 5.3%). Licofelonedose-dependently stimulated 1,25-dihydroxyvitaminDinduced AP activity while inhibiting osteocalcin re-lease. BayX-1005 partly reproduced these effects, butNS-398 failed to affect them. LTB4 dose-dependentlyinhibited AP activity in OA osteoblasts, while its effecton osteocalcin depended on endogenous LTB4 levels inthese cells. In normal osteoblasts, LTB4 dose-dependently stimulated osteocalcin, whereas it failed toinfluence AP. LTB4 receptors BLT1 and BLT2 werepresent in normal and OA osteoblasts.

Conclusion. Licofelone inhibits the production ofPGE2 and LTB4. Selective effects of licofelone on AP andosteocalcin occur via its role on LTB4 production.Because LTB4 can modify cell biomarkers in OA andnormal osteoblasts, our results suggest licofelone couldmodify abnormal bone remodeling in OA.

Osteoarthritis (OA) is the leading cause of dis-ability among the elderly population (1), yet the etiol-ogy, pathogenesis, and progression of this disease arestill not fully understood (2,3). OA has a multifactorialorigin and is slowly progressive. The disease process canbe described as degradation and loss of articular carti-lage accompanied by hypertrophic bone changes, withosteophyte formation and subchondral plate thickening(4,5). The process includes changes in articular cartilageand surrounding bone, an imbalance in loss of cartilage(due to matrix degradation), and an attempt to repairthis matrix (4,5). Specific interactions between bone

Supported in part by a grant from Merckle GmbH, Ulm,Germany, and the Fonds de la Recherche en Sante du Quebec, EquipePrioritaire en Arthrose. Dr. Lajeunesse is a senior scholar whose workwas supported by the Fonds de la Recherche en Sante du Quebec.

1Yosabeth Paredes, Frederic Massicotte, Jean-Pierre Pelle-tier, MD, Johanne Martel-Pelletier, PhD, Daniel Lajeunesse, PhD:Centre Hospitalier de lUniversite de Montreal, Hopital Notre-Dame,Montreal, Quebec, Canada; 2Stefan Laufer, PhD: Eberhard-KarlsUniversity Tuebingen, Tuebingen, Germany.

Address correspondence and reprint requests to Daniel La-jeunesse, PhD, Unite de Recherche en Arthrose, CHUM, HopitalNotre-Dame, 1560 rue Sherbrooke Est, Montreal, Quebec H2L 4M1,Canada. E-mail: lajeunda@jonction.net.

Submitted for publication June 18, 2001; accepted in revisedform March 1, 2002.

1804

and cartilage in OA have not been clearly defined;however, mounting evidence indicates a direct role ofthe bone compartment in the initiation/progression ofOA (68)

Arachidonic acid is released from membranephospholipids following activation of phospholipase A2.Several enzymatic complexes can further metabolizearachidonic acid into a number of prostanoids by specificsyntheses in different cells (9), and osteoblasts mainlyproduce prostaglandin E2 (PGE2) (10). The enzyme5-lipoxygenase (5-LOX) catalyzes the formation of leu-kotrienes (LTs) from arachidonic acid. The first com-pound formed is LTA4, which rapidly converts intoLTB4 or LTC4. LTC4 can be further catalyzed into LTD4and LTE4 (11,12).

Conventional nonsteroidal antiinflammatorydrugs (NSAIDs) inhibit cyclooxygenase 1 (COX-1)and/or COX-2, the key enzymes that metabolize arachi-donic acid into prostaglandins and thromboxanes(13,14). Reduction of prostaglandins and thromboxaneis probably the basis for the antiinflammatory andanalgesic activity of NSAIDs, which are widely used forthe treatment of OA. However, side effects have limitedthe utility of these drugs. The most common side effectsare gastrointestinal and range from mild symptoms suchas dyspepsia and abdominal discomfort to more seriousevents such as peptic ulcers and life-threatening gastric/duodenal bleeding and perforation (15). Indeed, long-term inhibition of COX could result in a shunt to the5-LOX pathway, leading to the formation of leuko-trienes, which can induce gastric lesions and ulceration(16,17). Therefore, NSAIDs targeting both the COXand 5-LOX pathways may control the symptoms of OAwithout causing serious gastrointestinal side effects(16,17). Moreover, whether a shunt to the 5-LOX path-way and local production of leukotrienes in joint tissueare detrimental to tissue such as the subchondral bonecompartment remains unknown.

Osteoblasts produce prostaglandins via bothCOX-1 and COX-2 activities (18,19). Prostaglandinsstimulate bone resorption by increasing the number andactivity of osteoclasts, and PGE2 is the most potentagonist (20). The roles of a number of stimulators offormation of tartrate-resistant acid phosphatasepositive giant cells with osteoclast features are blockedby inhibiting endogenous prostaglandin synthesis (2123). Prostaglandins also enhance bone formation bystimulating the replication and differentiation of osteo-blasts along with an increase in the production of growthfactors (24). In fully differentiated osteoblasts, highconcentrations of prostaglandins can inhibit collagen

synthesis (25). Prostaglandins may also mediate theresponse to mechanical forces in bone, because boneformation stimulated by impact loading can be blockedby NSAIDs (26).

Osteoblasts also synthesize leukotrienes in vivo,although their in vitro production has not been studied.Moreover, the exact levels of PGE2 and leukotrienesobserved in vivo in OA bone tissue are controversial(27,28), and the levels of leukotrienes produced in vitroby OA osteoblasts have not been evaluated. Finally,whether leukotrienes may modulate the activity of OAosteoblasts and/or be involved in OA pathogenesis re-mains to be determined.

The aim of this study is to explore the effect ofdual inhibition of COX-1/COX-2 and 5-LOX usingclinically relevant concentrations of licofelone comparedwith specific inhibition of COX-2 or 5-LOX on themetabolism of OA subchondral osteoblasts. Licofeloneinhibits both COX-1, COX-2, and 5-LOX (29,30). It alsoinhibits shunting to leukotrienes and leukocyte adher-ence and shows better gastrointestinal tolerability (3133). We tested the effect of licofelone on the productionof LTB4 and PGE2, as well as on alkaline phosphatase(AP) activity and osteocalcin release by OA subchondralosteoblasts, 2 key biomarkers of normal osteoblastfunction.

PATIENTS AND METHODS

Patients and clinical parameters. OA specimens wereobtained from 34 patients (15 men, 19 women; mean SD age71.4 11.5 years) undergoing total knee replacement surgeryand classified as having OA according to American College ofRheumatology criteria (34). The subchondral bone plate wasdissected from the tibial plateau under sterile conditions, asdescribed previously (35,36), and the specimens representedmoderate-to-severe OA as defined by macroscopic criteria(3.0 0.5 on a scale of 04). Osteophytes were removed priorto processing. For 6 months before surgery, no patient hadreceived medication, including corticosteroids, which wouldinterfere with bone metabolism. Normal subchondral bonespecimens from tibial plateaus were collected from 8 normalsubjects at autopsy (7 men, 1 woman; mean SD age 57.6 15.9 years). Before using an autopsy specimen, we ensured thatthe donor had not received any medication that could interferewith bone metabolism, did not have any metabolic bonedisease, and had macroscopically normal cartilage. Thesesamples were used to determine the levels of LTB4 and PGE2produced by normal osteoblasts and to determine the regula-tion of biomarkers by addition of exogenous LTB4.

Subchondral bone primary osteoblast cell cultures.Isolation of the subchondral bone plate was performed undera magnifying microscope to insure complete removal of carti-lage and trabecular bone. Subchondral bone cell cultures wereprepared as previously described (37,38), using 3 sequential

ROLE OF LEUKOTRIENE B4 IN HUMAN OA OSTEOBLASTS 1805

digestions in the presence of 1 mg/ml type I collagenase(Sigma-Aldrich, St. Louis, MO). The digested bone pieceswere cultured in BGJb media (Sigma-Aldrich) containing 20%fetal bovine serum (FBS; Wisent, St. Bruno, Quebec, Canada).This medium was replaced every 2 days until cells wereobserved in the petri dishes. At this point, the culture mediumwas replaced with fresh media containing 10% FBS. Atconfluence, cells were passaged once at 25,000 cells/cm2 andgrown for 5 days before assays. We previously showed that cellsobtained under these culture conditions show an osteoblast-like cell phenotype (35,37,38).

Determination of PGE2 and leukotrienes. OA sub-chondral osteoblasts were grown in T25 flasks, and condi-tioned media were obtained from the last 2 days of culture inHams F-12/Dulbeccos modified Eagles medium (DMEM;Sigma-Aldrich) containing 0.5% FBS in the presence of 10 nMionomycin for the last 4 hours of culture to promote LTB4synthesis. Cells were incubated with or without 10 M NS-398(inhibitor of PGE2 synthesis; Sigma), 10 M BayX-1005 (in-hibitor of 5-LOX activating protein; Merckle GmbH, Ulm,Germany), or with licofelone (0.8, 2.6, or 8 M; MerckleGmbH). Supernatants were recuperated for the determinationof both PGE2 and LTB4, while the cells were solubilized andprepared for protein determination by the bicinchoninic acidmethod (39). The levels of PGE2 and LTB4 were determinedusing specific enzyme immunoassays (EIAs; Cayman Chemi-cal, Ann Arbor, MI).

Phenotypic characterization of osteoblasts: osteocal-cin and AP. For the determination of biomarkers, cells weretreated with 50 ng/ml of 1,25-dihydroxyvitamin D(1,25[OH]2D3; a generous gift from Dr. Uskokovic, Hoffman-LaRoche, Nutley, NJ), and were incubated in the presence orabsence of therapeutic concentrations of licofelone (0.88M), NS-398 (10 M), or BayX-1005 (10 M). Media werecollected at the end of the incubation period and frozen at80C prior to assays. Cells were then washed twice withphosphate buffered saline (pH 7.4) and solubilized in APbuffer consisting of 100 mM glycine, 1 mM MgCl2, 1 mMZnCl2, and 1% Triton X-100 (pH 10.5). Nascent osteocalcinwas determined by a specific EIA (Biomedical Technologies,Stoughton, MA) and AP activity was determined as the releaseof p-nitrophenol hydrolyzed from p-nitrophenyl phosphate aspreviously described (35,37,38). In another series of experi-ments, the effect of LTB4 on osteocalcin secretion and AP byboth normal and OA osteoblasts was determined. LTB4 atdoses ranging from 1014M to 109M was applied to confluentcells for the last 48 hours of culture in Hams F-12/DMEMcontaining 0.5% FBS, and with 10 nM ionomycin (Sigma-Aldrich) for the last 4 hours of culture.

Western blot analysis of leukotriene receptors BLT1and BLT2. Although 2 LTB4 receptors, BLT1 and BLT2, havebeen previously described (40,41), their presence in bonetissues was never assessed. Therefore, osteoblasts were ex-tracted in the radioimmunoprecipitation assay buffer (50 mMTris HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1 mMphenylmethylsulfonyl fluoride, 10 g/ml each of aprotinin,leupeptin, and pepstatin, 1% Nonidet P40, 1 mM sodiumorathovanadate, and 1 mM NaF), and the protein levels weredetermined. Twenty micrograms of cellular protein extract wasseparated on 10% sodium dodecyl sulfatepolyacrylamide gelsand transferred to polyvinylidene difluoride membrane. After

blocking with 5% skim milk in Tris buffered salineTween (20mM Tris HCl, 150 mM NaCl [pH 7.5], 0.1% Tween 20) andwashing, the membranes were sequentially incubated at 4Covernight with the primary antibody (rabbit polyclonal anti-human BLT1 or BLT2, 1:10,000 dilution; Cayman Chemicals,Ann Arbor, MI) and with horseradish peroxidase goat anti-rabbit (1:20,000 dilution; Pierce) and visualized using anEnhanced Chemiluminescence Plus kit (Pierce) according tothe manufacturers directions. BLT1 should give a band at5865 kd with possibly a doublet, whereas BLT2 gives a bandat 4752 kd.

Statistical analysis. Results are expressed as themean SEM. Assays were performed in duplicate. Dose-response data were analyzed statistically using analysis ofvariance (ANOVA). When ANOVA comparisons reachedsignificance, subtests were performed using Fishers protectedt-test. In some cases, a Students t-test was used. P values lessthan 0.05 were considered significant.

RESULTS

We first tested the capacity of normal and OAsubchondral osteoblasts to produce LTB4. Data indicatethat the endogenous production of LTB4 and PGE2 byOA subchondral osteoblasts is generally more variablethan that by normal osteoblasts (Figure 1). Basal levelsof LTB4 from all OA subchondral osteoblasts averaged61.3 10.3 pg/mg protein (n 16), whereas LTB4 levelsin normal osteoblasts were 29.8 2.8 pg/mg protein(n 5; P 0.01 by Students t-test). Interestingly, thedistribution of LTB4 and PGE2 levels varied from oneOA patient to another, yet the variance was in oppositedirections (Figure 1). Therefore, 2 groups of OA pa-tients were observed: those whose osteoblasts showedhigh PGE2 levels (4,000 pg/mg protein or 2 SD abovethe mean values for normal osteoblasts) and low LTB4levels, and those whose osteoblasts showed low PGE2levels and the highest LTB4 levels (Figure 1). Because ofthis inherent variability in LTB4 levels between OApatients, data for the effect of licofelone on OA osteo-blasts are presented as the percentage compared withcontrol (Figure 2).

At clinically relevant concentrations, licofelonehad a dose-dependent inhibitory effect on endogenousLTB4 production by OA osteoblasts (P 0.0001 byANOVA), reaching a maximum inhibition of 67.0 7.6% at 8 M licofelone (Figure 2). Inhibition bylicofelone varied according to initial endogenous LTB4levels in individual OA osteoblasts and at maximalconcentration of licofelone; these LTB4 values weresimilar to those in normal osteoblasts. Short-term (48-hour) treatment with NS-398 (10 M ) failed to signifi-cantly modify LTB4 values (Figure 2), although it re-duced PGE2 levels (Figure 2). However, long-term

1806 PAREDES ET AL

(5-day) inhibition with NS-398 increased LTB4 values4-fold (Figure 2) but did not reduce PGE2 levels morethan control levels following 2 days of treatment (datanot shown). BayX-1005 (10 M) inhibited LTB4 produc-tion by OA osteoblasts (38.7 14.5%; P 0.05), but itseffect was not as strong as that with licofelone (Figure 2).

OA osteoblasts also produced high levelsof PGE2 (3,974.7 459.6 pg/mg protein; n 16). Thisproduction of PGE2 was also inhibited by clinicallyrelevant concentrations of licofelone in OA subchon-dral osteoblasts (P 0.0001 by ANOVA) (Figure 2),reaching an average inhibition of 61.2 6.4% at 8 Mlicofelone. Initial PGE2 levels in OA osteoblasts werealso variable, as were LTB4 levels, but licofelone re-duced PGE2 to similar levels in all OA osteoblasts.NS-398 (10 M), which selectively inhibits COX-2 andtherefore PGE2 production, inhibited this production inOA subchondral osteoblasts by 75.8 5.3% (P 0.01),whereas BayX-1005 (10 M), a specific inhibitor of5-LOX activating protein (and therefore of leukotrieneproduction) only marginally inhibited PGE2 production(14.8 5.3%; P 0.05).

Two markers of the osteoblast phenotype,namely, AP activity and osteocalcin release, were notsimilarly affected by licofelone. Indeed, basal AP activitywas not significantly affected by increasing concentra-tions of licofelone up to 8 M (data not shown).

Conversely, under 1,25(OH)2D3 stimulation, which in-creases AP activity 2-fold (35,38), licofelone dose-dependently stimulated this activity (P 0.0009 byANOVA) (Figure 3). NS-398 (10 M), which inhibitsPGE2 production (Figure 2), failed to influence1,25(OH)2D3-stimulated AP activity, and BayX-1005(10 M) caused a slight but significant increase (P 0.05; Figure 3). In contrast to its effects on AP activity,osteocalcin secretion by OA subchondral osteoblastswas dose-dependently inhibited by licofelone (P 0.015by ANOVA), and a maximal inhibition of 36.7 15.0%

Figure 1. Relationship between levels of leukotriene B4 (LTB4) andprostaglandin E2 (PGE2) in human osteoarthritis (OA) and normalsubchondral osteoblasts (Ob). Data points represent each individualspecimen (5 normal, 16 OA). Bars show the mean SD. Low OAOb low PGE2 levels and the highest LTB4 levels; high OA Ob high PGE2 levels and low LTB4 levels.

Figure 2. Effect of licofelone on LTB4 and PGE2 production byhuman OA subchondral osteoblasts. Values are expressed as thepercentage of control without licofelone and are the mean and SEMfrom 10 patients for LTB4 and 16 patients for PGE2. P 0.0001 byanalysis of variance (ANOVA) for each of the different treatments forboth LTB4 and PGE2. P values shown are versus control and wereobtained by Fishers protected t-test following ANOVA. N.S. notsignificant; NS-398 Ch long-term treatment with NS-398. See Figure1 for other definitions.

ROLE OF LEUKOTRIENE B4 IN HUMAN OA OSTEOBLASTS 1807

was obtained at 8 M licofelone (Figure 3). NS-398 didnot affect osteocalcin secretion, a situation similar tothat of AP activity, while BayX-1005 inhibited osteo-calcin secretion by 28.2 5.2% (P 0.05). The effectof licofelone on osteocalcin secretion was not relatedto regulation of the expression of osteocalcin, becauseNorthern blot analysis (n 4 OA cases) failed toshow any significant effects of this drug on messengerRNA (mRNA) levels; osteocalcin:GAPDH ratios were1.05 0.06, 1.02 0.06, 0.97 0.06, and 0.97 0.06

(mean SEM) for 0, 0.8, 2.6, and 8 M licofelone,respectively (data not shown). Licofelone also did notaffect cell proliferation or cell protein levels in OAosteoblasts (data not shown).

Because licofelone inhibits both LTB4 and PGE2production in OA subchondral osteoblasts, but onlyinhibition of LTB4 production seemed to modulateexpression of osteoblast biomarkers, we next evaluatedwhether the addition of exogenous LTB4 could affect AP

Figure 3. Effect of licofelone on alkaline phosphatase activity of andosteocalcin secretion by osteoarthritis subchondral osteoblasts (n 5).Results are the mean and SEM. P 0.0009 and P 0.015 by analysisof variance (ANOVA) for alkaline phosphatase activity and osteocal-cin secretion, respectively. P values obtained for specific subtests (allversus control unless indicated otherwise) were obtained by Fishersprotected t-test following ANOVA.

Figure 4. Effect of exogenous addition of leukotriene B4 (LTB4) onalkaline phosphatase activity and osteocalcin secretion by osteoarthri-tis (OA) subchondral osteoblasts. Values are expressed as the percent-age of control and are the mean SEM from 9 patients for alkalinephosphatase, and 7 patients for osteocalcin (4 with low endogenousLTB4 levels [OA low] and 3 with high endogenous LTB4 levels [OAhigh]).

1808 PAREDES ET AL

activity and osteocalcin release in both normal and OAosteoblasts. Increasing doses of LTB4 (10

13M to1011M) significantly inhibited 1,25(OH)2D3-inducedAP activity in OA osteoblasts (Figure 4), whereas higherdoses (1010M) almost reverted this activity to its initialcontrol value. Data were similar for AP activity in

response to LTB4, regardless of endogenous levels inthese cells. In contrast, under similar conditions, exoge-nous LTB4 had 2 opposing effects on osteocalcin releasedepending on initial endogenous LTB4 production bythese cells. Indeed, when OA osteoblasts had higherinitial basal LTB4 levels, addition of exogenous LTB4inhibited osteocalcin production, whereas when cellshad lower basal levels, the addition of LTB4 stimulatedosteocalcin release (Figure 4). LTB4 addition (10

13Mto 1010M) to normal osteoblasts did not significantlymodulate 1,25(OH)2D3-induced AP activity (Figure 5).Interestingly, in normal osteoblasts, LTB4 dose-dependently stimulated osteocalcin release (P 0.02 byANOVA) (Figure 5).

To investigate by which pathway LTB4 affectsosteoblasts, we evaluated the presence of leukotrienereceptors BLT1 and BLT2 in normal and OA osteo-blasts by Western blot analysis. Using polyclonal anti-bodies against BLT1 receptors showed a strong responsein both normal and OA osteoblasts (Figure 6). Twomajor bands were observed at molecular weights of 53kd and 58 kd, which correspond to BLT2 and BLT1,respectively, suggesting that there was a cross-reactivitybetween the 2 receptors and the antibodies used. In-deed, using polyclonal anti-BLT2 antibodies showed asimilar pattern of expression (results not shown). Ofnote, in 1 normal osteoblast cell culture, the 53-kd bandrepresenting BLT2 was undetectable. However, densito-metric analysis of the BLT1 58-kd bands failed to showany significant differences between normal and OAosteoblasts.

DISCUSSION

Increasing evidence suggests that the subchon-dral bone compartment is intimately linked with theinitiation and/or progression of OA (68). We previ-ously observed abnormal levels of bone cell biomarkers,

Figure 5. Effect of exogenous addition of leukotriene B4 (LTB4) onalkaline phosphatase activity and osteocalcin secretion by normalhuman subchondral osteoblasts. Values are expressed as the percent-age of control and are the mean and SEM from 5 patients for alkalinephosphatase activity and 3 patients for osteocalcin secretion. P 0.02by analysis of variance (ANOVA) for osteocalcin. P value is versuscontrol and was obtained by Fishers protected t-test followingANOVA.

Figure 6. Western blot analysis of leukotriene receptors BLT1 innormal (n 3) and osteoarthritis (OA; n 3) osteoblasts (Ob).Confluent cells were lysed and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Western blot analysis of BLT1levels was performed using a polyclonal antibody and peroxidase-labeled second antibody.

ROLE OF LEUKOTRIENE B4 IN HUMAN OA OSTEOBLASTS 1809

cytokine, and prostaglandin levels in in vitro OA sub-chondral osteoblasts (28,35,36). The present data fur-ther indicate that in vitro OA osteoblasts can alsosynthesize LTB4. Two groups of OA osteoblasts, onewith PGE2 levels similar to those of normal osteoblastsand high LTB4 levels and the other with higher-than-normal PGE2 and low LTB4 levels, were observed; wepreviously noted this situation for the production ofcytokines and PGE2 (28). Licofelone inhibited endoge-nous in vitro production of both LTB4 and PGE2 by OAosteoblasts to similar levels. Because PGE2 levels werehigh and LTB4 levels were low or vice versa, this couldsuggest that both pathways are actively involved in OApathogenesis.

Synthesis of both LTB4 and PGE2 requires thepresence of arachidonic acid. Therefore, it is plausiblethat one or the other pathway is used under conditionsof high production of arachidonic acid. Leukotrienes arepotentially more harmful than PGE2 for the inflamma-tory process, because the former are potent chemotacticagents and can increase microvascular permeability (4244). The fact that PGE2 and LTB4 levels varied inopposite directions in OA osteoblasts raises the questionof whether long-term inhibition of COX-2 in OA pa-tients, possibly leading to a shunt to the 5-LOX pathway,would be more detrimental to osteoblasts and thereforeto joints in OA patients. Interestingly, we were able todemonstrate this shunt in OA osteoblasts followinglong-term inhibition of COX-2 with NS-398. This shuntagainst leukotriene production has been previouslyshown in other tissues (16,17,31), but, to our knowledge,the results of our study are the first indication of thisoccurring in osteoblasts. Therefore, a therapeutic inter-vention aimed at reducing both pathways may poten-tially be very attractive for the treatment of OA patients.

Licofelone reduced 1,25(OH)2D3-induced osteo-calcin secretion via its inhibition of LTB4 production inOA osteoblasts, not via inhibition of PGE2 synthesis,because NS-398 treatments could not reproduce thiseffect and BayX-1005 could. This inhibition of osteocal-cin secretion is posttranscriptional, because licofelonefailed to directly modify osteocalcin mRNA levels (datanot shown). In contrast, licofelone or BayX-1005 stimu-lated 1,25(OH)2D3-induced AP activity, while additionof exogenous LTB4 inhibited AP activity in OA osteo-blasts, and NS-398 was without effect. These resultssuggest that licofelone does not have any general inhib-itory effect on 1,25(OH)2D3-dependent pathways, andfurther suggest that its effects are mediated, at least inpart, by leukotrienes.

In contrast, LTB4 had mixed effects on cell

biomarkers in normal osteoblasts, stimulating osteocal-cin release while failing to significantly alter AP activity.This could indicate that LTB4 more specifically targetsthe secretion of osteocalcin and has more limited effectson AP activity under normal conditions. However, underabnormal conditions, such as in OA subchondral osteo-blasts in which AP activity is already increased (35), itcould have an effect as shown here. However, abnormallevels or expression of LTB4 receptors could not accountfor these limited differential effects of LTB4 in normaland OA osteoblasts. Indeed we observed similar levelsfor BLT1 receptors in normal and OA osteoblasts andvariable levels for BLT2 receptors in normal comparedwith OA osteoblasts. This is also peculiar, because BLT2receptors were believed to be ubiquitous (41), whereasBLT1 receptors are expressed almost exclusively inperipheral leukocytes (40); however, bone tissues werenot tested in these studies.

Osteocalcin levels have been shown to be ele-vated in bones of OA patients in vivo, even at nonweight-bearing sites (45), and in in vitro subchondralOA osteoblasts (35). Because osteocalcin may retardnormal mineralization in vivo, its high levels in OApatients could explain the abnormally low bone miner-alization in these individuals. Indeed, it was first be-lieved that OA bones were hypermineralized using min-eral density quantification by backscattered electricimage analysis (46) and a density fractionation tech-nique (47). However, the acceleration of bone turnoverin OA results in hypomineralized subchondral bone andreduces its stiffness for a given apparent density butincreases stiffness if this is offset by increased bonevolume (48). The stiffness of trabecular bone is in-creased in OA, because more trabeculae are present(49); however, material stiffness measured by an ultra-sonic method that excluded the contribution of in-creased subchondral bone thickness revealed that OAbones were less stiff (50).

Mansell and Bailey previously showed that sub-chondral trabecular OA bone explants from the femoralhead were hypomineralized, because molar calcium-to-collagen ratios were reduced in OA compared withnormal bone explants (51), whereas bone far away fromthe joint, such as the iliac crest, is hypermineralized inOA (52). Finally, urinary parameters of bone turnoverare slightly lower in patients with spinal osteoarthrosiscompared with normal subjects (53). In contrast, theseparameters are increased in both RA and OA patients(54), suggesting that generalized OA is more likelylinked to an increase in bone turnover. Therefore,

1810 PAREDES ET AL

reducing osteocalcin release from OA osteoblasts maypromote improved mineralization in these individuals.

Recent studies in our laboratory showed thatNSAIDs that affect subchondral osteoblasts preventedcartilage lesions in an OA dog model, whereas those thatdid not affect osteoblasts also did not significantlyprevent cartilage lesions in this model (55). It is there-fore possible that some NSAIDs may be promoting aprotective effect via a mechanism other than simplyreducing PGE2 production. Therefore, a compound withinhibitory activity against COX-1, COX-2, and 5-LOXmay be very desirable for the treatment of OA, becauseit would combine efficiency and gastric protection, andwould possibly impact an important mechanism involvedin the onset and/or progression of OA. Moreover,prolonged therapy with different NSAIDs that contrib-ute to a reduction in prostaglandin production may notnecessarily protect against an increase in leukotrienes.Indeed, the possible shunt from the COX to the 5-LOXpathway after prolonged inhibition of COX activityand/or a negative retro-feedback of PGE2 on COXcould contribute significantly to the inflammatory pro-cess observed in patients with OA.

In conclusion, OA subchondral osteoblasts pos-sess a functional 5-LOX activity that can synthesizeLTB4. Licofelone specifically inhibits LTB4 and PGE2synthesis, inhibits osteocalcin synthesis, and promotesAP activity via its effect on LTB4 synthesis rather thanPGE2 synthesis. Because licofelone modifies selectivepathways in osteoblasts which contributes to the alter-ation of bone remodeling in OA patients, this therapeu-tic intervention could protect against progression of OA.

ACKNOWLEDGMENTS

We wish to thank Aline Delalandre for her experttechnical assistance and Sophie Langevin, Colleen Byrne, andSanta Fiori for their assistance in manuscript preparation.

REFERENCES

1. Centers for Disease Control. Arthritis prevalence and activitylimitations: United States. MMWR Morb Mortal Wkly Rep1994;43:4338.

2. Dieppe P. Osteoarthritis: Clinical and research perspective. Br JRheumatol 1991;30 Suppl 1:14.

3. Davis MA. Epidemiology of osteoarthritis. Clin Geriatr Med1988;4:24155.

4. Pelletier JP, Martel-Pelletier J, Howell DS. Etiopathogenesis ofosteoarthritis. In: Koopman WJ, editor. Arthritis and allied con-ditions: a textbook of rheumatology. 13th ed. Baltimore: Williams& Wilkins; 1997. p. 196984.

5. Hough AJ. Pathology of osteoarthritis. Koopman WJ, editor.Arthritis and allied conditions. 13th ed. Baltimore: Williams &Wilkins; 1997. p. 194568.

6. Dequeker J, Luyten FP. Bone mass and osteoarthritis. Clin ExpRheumatol 2000;18:S216.

7. Burr DB, Schaffler MB. The involvement of subchondral miner-alized tissues in osteoarthrosis: quantitative microscopic evidence.Microsc Res Tech 1997;37:34357.

8. Bailey AJ, Mansell JP. Do subchondral bone changes exacerbateor precede articular cartilage destruction in osteoarthritis of theelderly? Gerontology 1997;43:296304.

9. Smith WL, Garavito RM, DeWitt DL. Prostaglandin endoperox-ide H synthases (cyclooxygenases)-1 and -2. J Biol Chem 1996;271:3315760.

10. Kawaguchi H, Pilbeam CC, Harrison JR, Raisz LG. The role ofprostaglandins in the regulation of bone metabolism. Clin Orthop1995;3646.

11. Radmark OP. The molecular biology and regulation of 5-lipoxy-genase. Am J Respir Crit Care Med 2000;161:S115.

12. Silverman ES, Drazen JM. The biology of 5-lipoxygenase: func-tion, structure, and regulatory mechanisms. Proc Assoc Am Phy-sicians 1999;111:52536.

13. Vane JR. Inhibition of prostaglandin synthesis as a mechanism ofaction for aspirin-like drugs. Nat New Biol 1971;231:2325.

14. Simon LS. Actions and toxicity of nonsteroidal anti-inflammatorydrugs. Curr Opin Rheumatol 1996;8:16975.

15. Wolfe MM, Lichtenstein DR, Singh G. Gastrointestinal toxicity ofnonsteroidal antiinflammatory drugs. N Engl J Med 1999;340:188899.

16. Dyer RD, Connor DT. Dual inhibitors of prostaglandin andleukotriene biosynthesis. Curr Pharm Des 1997;3:46372.

17. Rainsford KD. Leukotrienes in the pathogenesis of NSAID-induced gastric and intestinal mucosal damage. Agents Actions1993;39 Spec No:C246.

18. Mundy GR. Cytokines and growth factors in the regulation ofbone remodeling. J Bone Miner Res 1993;8 Suppl 2:S50510.

19. Boyce BF, Hughes DE, Wright KR, Xing L, Dai A. Recentadvances in bone biology provide insight into the pathogenesis ofbone diseases. Lab Invest 1999;79:8394.

20. Raisz LG. Prostaglandins and bone: physiology and pathophysiol-ogy. Osteoarthritis Cartilage 1999;7:41921.

21. Tai H, Miyaura C, Pilbeam CC, Tamura T, Ohsugi Y, KoishiharaY, et al. Transcriptional induction of cyclooxygenase-2 in osteo-blasts is involved in interleukin-6-induced osteoclast formation.Endocrinology 1997;138:23729.

22. Hurley MM, Lee SK, Raisz LG, Bernecker P, Lorenzo J. Basicfibroblast growth factor induces osteoclast formation in murinebone marrow cultures. Bone 1998;22:30916.

23. Lader CS, Flanagan AM. Prostaglandin E2, interleukin 1, andtumor necrosis factor- increase human osteoclast formation andbone resorption in vitro. Endocrinology 1998;139:315764.

24. Woodiel FN, Fall PM, Raisz LG. Anabolic effects of prostaglan-dins in cultured fetal rat calvariae: structure-activity relations andsignal transduction pathway. J Bone Miner Res 1996;11:124955.

25. Fall PM, Breault DT, Raisz LG. Inhibition of collagen synthesis byprostaglandins in the immortalized rat osteoblastic cell line Py1a:structure-activity relations and signal transduction mechanisms.J Bone Miner Res 1994;9:193543.

26. Chow JW, Chambers TJ. Indomethacin has distinct early and lateactions on bone formation induced by mechanical stimulation.Am J Physiol 1994;267:E28792.

27. Wittenberg RH, Willburger RE, Kleemeyer KS, Peskar BA. Invitro release of prostaglandins and leukotrienes from synovialtissue, cartilage, and bone in degenerative joint diseases. ArthritisRheum 1993;36:144450.

28. Benderdour M, Hilal G, Lajeunesse D, Pelletier JP, Duval N,Martel-Pelletier J. Osteoarthritic osteoblasts show variable levelsof cytokines production despite similar phenotypic expression[abstract]. Arthritis Rheum 1999;42:S251.

29. Laufer S, Tries S, Augustin J, Dannhardt G. Pharmacological

ROLE OF LEUKOTRIENE B4 IN HUMAN OA OSTEOBLASTS 1811

profile of a new pyrrolizine derivative inhibiting the enzymescyclo-oxygenase and 5-lipoxygenase. Arzneimittelforschung 1994;44:62936.

30. Laufer SA, Augustin J, Dannhardt G, Kiefer W. (6,7-Diaryldihy-dropyrrolizin-5-yl)acetic acids, a novel class of potent dual inhib-itors of both cyclooxygenase and 5-lipoxygenase. J Med Chem1994;37:18947.

31. Laufer S. Discovery and development of ML 3000. Inflammo-pharmacology 2001;9:10112.

32. Gay RE, Neidhart M, Pataky F, Tries S, Laufer S, Gay S. Dualinhibition of 5-lipoxygenase and cyclooxygenases 1 and 2 byML3000 reduces joint destruction in adjuvant arthritis. J Rheuma-tol 2001;28:20605.

33. Celotti F, Laufer S. Anti-inflammatory drugs: new multitargetcompounds to face an old problem: the dual inhibition concept.Pharmacol Res 2001;43:42936.

34. Altman RD, Asch E, Bloch DA, Bole G, Borenstein D, BrandtKD, et al. Development of criteria for the classification andreporting of osteoarthritis: classification of osteoarthritis of theknee. Arthritis Rheum 1986;29:103949.

35. Hilal G, Martel-Pelletier J, Pelletier JP, Ranger P, Lajeunesse D.Osteoblast-like cells from human subchondral osteoarthritic bonedemonstrate an altered phenotype in vitro: possible role in sub-chondral bone sclerosis. Arthritis Rheum 1998;41:8919.

36. Hilal G, Martel-Pelletier J, Pelletier JP, Duval N, Lajeunesse D.Abnormal regulation of urokinase plasminogen activator by insu-lin-like growth factor 1 in human osteoarthritic subchondralosteoblasts. Arthritis Rheum 1999;42:211222.

37. Lajeunesse D, Busque L, Menard P, Brunette MG, Bonny Y.Demonstration of an osteoblast defect in two cases of humanmalignant osteopetrosis: correction of the phenotype after bonemarrow transplant. J Clin Invest 1996;98:183542.

38. Lajeunesse D, Kiebzak GM, Frondoza C, Sacktor B. Regulation ofosteocalcin secretion by human primary bone cells and by thehuman osteosarcoma cell line MG-63. J Bone Miner Res 1991;14:23750.

39. Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH,Provenzano MD, et al. Measurement of protein using bicincho-ninic acid. Anal Biochem 1985;150:7685.

40. Yokomizo T, Izumi T, Chang K, Takuwa Y, Shimizu T. AG-protein-coupled receptor for leukotriene B4 that mediateschemotaxis. Nature 1997;387:6204.

41. Yokomizo T, Kato K, Terawaki K, Izumi T, Shimizu T. A secondleukotriene B(4) receptor, BLT2: a new therapeutic target ininflammation and immunological disorders. J Exp Med 2000;192:42132.

42. Penrose JF, Austen KF. The biochemical, molecular, and genomic

aspects of leukotriene C4 synthase. Proc Assoc Am Physicians1999;111:53746.

43. Los M, Schenk H, Hexel K, Baeuerle PA, Droge W, Schulze-Osthoff K. IL-2 gene expression and NF-kappa B activationthrough CD28 requires reactive oxygen production by 5-lipoxygen-ase. EMBO J 1995;14:373140.

44. Gok S, Ulker S, Huseyinov A, Hatip FB, Cinar MG, Evinc A. Roleof leukotrienes on coronary vasoconstriction in isolated hearts ofarthritic rats: effect of in vivo treatment with CI-986, a dualinhibitor of cyclooxygenase and lipoxygenase. Pharmacology 2000;60:416.

45. Gevers G, Dequeker J. Collagen and non-collagenous proteincontent (osteocalcin, sialoprotein, proteoglycan) in the iliac crestbone and serum osteocalcin in women with and without handosteoarthritis. Coll Relat Res 1987;7:43542.

46. Boyde A, Jones SJ, Aerssens J, Dequeker J. Mineral densityquantitation of the human cortical iliac crest by backscatteredelectron image analysis: variations with age, sex, and degree ofosteoarthritis. Bone 1995;16:61927.

47. Raymaekers G, Aerssens J, Van den Eynde R, Peeters J, GeusensP, Devos P, et al. Alterations of the mineralization profile andosteocalcin concentrations in osteoarthritic cortical iliac crestbone. Calcif Tissue Int 1992;51:26975.

48. Burr DB. The importance of subchondral bone in osteoarthrosis.Curr Opin Rheumatol 1998;10:256562.

49. Li B, Aspden RM. Mechanical and material properties of thesubchondral bone plate from the femoral head of patients withosteoarthritis or osteoporosis. Ann Rheum Dis 1997;56:24754.

50. Li B, Aspden RM. Composition and mechanical properties ofcancellous bone from the femoral head of patients with osteo-porosis or osteoarthritis. J Bone Miner Res 1997;12:64151.

51. Mansell JP, Bailey AJ. Abnormal cancellous bone collagen me-tabolism in osteoarthritis. J Clin Invest 1998;101:1596603.

52. Gevers G, Dequeker J, Martens M, Van Audekercke R, Nyssen-Behets C, Dhem A. Biomechanical characteristics of iliac crestbone in elderly women according to osteoarthritis grade at thehand joints. J Rheumatol 1989;16:6603.

53. Peel NF, Barrington NA, Blumsohn A, Colwell A, Hannon R,Eastell R. Bone mineral density and bone turnover in spinalosteoarthrosis. Ann Rheum Dis 1995;54:86771.

54. Seibel MJ, Duncan A, Robins SP. Urinary hydroxy-pyridiniumcrosslinks provide indices of cartilage and bone involvement inarthritic diseases. J Rheumatol 1989;16:96470.

55. Pelletier JP, Lajeunesse D, Jovanovic DV, Lascau-Coman V,Jolicoeur FC, Hilal G, et al. Carprofen simultaneously reducesprogression of morphological changes in cartilage and subchondralbone in experimental dog osteoarthritis. J Rheumatol 2000;27:2893902.

1812 PAREDES ET AL