ARTHRITIS & RHEUMATISMVol. 56, No. 10, October 2007, pp 3326–3335DOI 10.1002/art.22879© 2007, American College of Rheumatology

Neutrophil Gelatinase–Associated Lipocalin Is Expressed inOsteoarthritis and Forms a Complex With

Matrix Metalloproteinase 9

Kalpana Gupta,1 Meenakshi Shukla,1 Jack B. Cowland,2 Charles J. Malemud,1

and Tariq M. Haqqi1

Objective. Expression of matrix metalloproteinase9 (MMP-9) is up-regulated in osteoarthritis (OA) andusually presents as multiple bands when synovial fluid(SF) from OA patients is analyzed by zymography.Among these bands is an �125–130–kd band for highmolecular weight (HMW) gelatinase, which has notbeen characterized. This study was undertaken to char-acterize the HMW MMP activity in OA SF.

Methods. MMP activity in OA SF was determinedby gelatin zymography. Recombinant MMPs were usedto identify MMP activity on the zymogram. Westernimmunoblotting, immunoprecipitation, and immuno-depletion analyses were performed using antibodiesspecific for human MMP-9 and human neutrophilgelatinase–associated lipocalin (NGAL). Human carti-lage matrix degradation was determined by dimethyl-methylene blue assay.

Results. Zymographic analysis showed that theHMW gelatinase in OA SF comigrated with a purifiedNGAL–MMP-9 complex. Results of Western immuno-blotting showed that the HMW gelatinase was alsorecognized by antibodies specific for human NGAL orhuman MMP-9. These same antibodies also immuno-precipitated the HMW gelatinase activity from OA SF.The NGAL–MMP-9 complex was reconstituted in vitroin gelatinase buffer. In the presence of NGAL, MMP-9

activity was stabilized; in the absence of NGAL, rapidloss of MMP-9 activity occurred. MMP-9–mediatedrelease of cartilage matrix proteoglycans was signifi-cantly higher in the presence of NGAL (P < 0.05).

Conclusion. Our findings demonstrate that theHMW gelatinase activity in OA SF represents a complexof NGAL and MMP-9. The ability of NGAL to protectMMP-9 activity is relevant to cartilage matrix degrada-tion in OA and may represent an important mechanismby which NGAL may contribute to the loss of cartilagematrix proteins in OA.

The developed world’s aging population has ex-perienced a dramatic increase in the incidence of jointdysfunction and osteoarthritis (OA), leading to a com-promised quality of life. Aging, biomechanical stress,and oxidative stress, in addition to genetic factors andtrauma, all appear to be associated with degenerativeand progressive cartilage and bone changes in OA,particularly in the weight-bearing joints. Although OA isnot considered an inflammatory disease by traditionalstandards, elevated levels of proinflammatory cytokines,such as interleukin-1� (IL-1�) and tumor necrosis factor�, have been observed in OA synovial fluid (SF), sup-porting the notion that there is an inflammatory com-ponent associated with the pathogenesis of OA (forreview, see refs. 1–3).

The only cell type present in articular cartilage isthe chondrocyte, which is embedded within an extracel-lular matrix (ECM) primarily composed of type IIcollagen and aggrecan. Chondrocytes arise from a mes-enchymal progenitor cell and are responsible for thebiosynthesis of ECM proteins and their transport intothe space occupied by the ECM. Chondrocytes play animportant role in cartilage homeostasis by maintainingthe proper balance between anabolic pathways, which

Supported in part by the NIH (grants AR-48782 and AT-002258).

1Kalpana Gupta, PhD, Meenakshi Shukla, MS, Charles J.Malemud, PhD, Tariq M. Haqqi, PhD: Case Western Reserve Uni-versity, and University Hospitals of Cleveland, Cleveland, Ohio; 2JackB. Cowland, PhD: Righospitalet, University of Copenhagen, Copen-hagen, Denmark.

Address correspondence and reprint requests to Tariq M.Haqqi, MD, Case Western Reserve University, Division of RheumaticDiseases, Department of Medicine, 2109 Adelbert Road, Cleveland,OH 44106. E-mail: txh5@case.edu.

Submitted for publication December 18, 2006; accepted inrevised form June 12, 2007.

3326

result in the biosynthesis of ECM proteins, and catabolicpathways, which result in ECM protein turnover. Previ-ous studies have established that in degenerative jointdiseases such as OA, a metabolic imbalance betweenanabolic and catabolic pathways occurs (4).

Articular cartilage chondrocytes demonstrateunique patterns of gene expression under a variety ofconditions in vitro that may be relevant to understandingdisease progression in OA. For example, articular chon-drocytes treated with cytokines show both suppressedexpression of messenger RNA (mRNA) coding forcartilage matrix proteins (5) and increased expression ofmRNA coding for matrix metalloproteinases (MMPs)(6). As collagen synthesis is activated in OA cartilage(7), any failure to incorporate collagen into the ECMmay result from activation of chondrocyte or synoviocytecollagenases and/or gelatinases, which degrade thenewly synthesized collagen (for review, see ref. 8).

MMPs are a family of Zn2�-dependent extracel-lular enzymes that share amino acid sequences andstructural domains and are generally produced as inac-tive zymogens. They are activated by proteolytic cleav-age of the zymogen, and after activation MMPs arecapable of degrading the components of the cartilageECM. They are classified into the following 5 subgroups,based on structure and substrate specificity: collage-nases, gelatinases, stromelysins (1,2), membrane-typeMMPs, and other MMPs (such as matrilysin, stromelysin3, metalloelastase, MMP-19, enamelysin, and MMP-23[9]). MMPs play an important role in many physiologicprocesses, such as embryonic development and growth,tissue remodeling, and tissue repair.

When MMPs are activated, cartilage ECM degra-dation ensues, apparently because levels of endogenouscartilage MMP inhibitors cannot regulate MMP activity.The hypothesis that MMPs are the primary enzymesinvolved in cartilage collagen digestion is supported byevidence that indicates that one or more MMPs candigest all of the ECM components in vitro, the enzymesare expressed in OA cartilage at the site of cartilagedestruction, specific digestion products of MMP arepresent in OA SF, aggrecan neoepitopes found in hu-man SF result from ADAMTS-4 or ADAMTS-5 activity,and experimental strategies that alter the expressionand/or the activity of MMP alter the progression ofcartilage destruction in OA. These important observa-tions suggest that MMPs in general, and collagenolyticMMPs in particular, are promising targets for the treat-ment of OA (10).

Neutrophil gelatinase–associated lipocalin (NGAL)is a siderophore binding protein belonging to the lipoca-

lin family of proteins, a diverse family of �20 solubleand often secreted small proteins (11). These proteinsare termed “lipocalins” largely because of their barrel-shaped structure, and they are thought to bind differentlow molecular mass molecules, including retinoids, ara-chidonic acid, various steroids, and iron (for review, seerefs. 12 and 13). Members of the lipocalin family have alarge degree of diversity at the primary sequence level,but most of them share 3 conserved motifs (14).

Human NGAL consists of a single disulfide-bridged polypeptide chain of 178 amino acid residueswith a calculated molecular mass of 22 kd (11), butglycosylation increases its apparent molecular mass to 25kd. Originally identified as a protein produced by acti-vated neutrophils, it generally occurs as a monomer,with a small percentage occurring as dimers and trimers,and also in a complex with 92-kd human neutrophilgelatinase B, or MMP-9 (10,15). NGAL is also expressedat low levels in other human tissues, including thekidney, prostate, and epithelia of the respiratory andalimentary tracts (16,17).

Expression of NGAL has been observed in epi-thelial cells, where it is induced during inflammationand, more specifically, by IL-1� (18). The up-regulationof NGAL in involuting tissue has led to the postulationthat it may play a role in apoptosis, but it appears morelikely that NGAL is associated with a survival response(14). This appears to be the case in the kidney, where theNGAL–siderophore–iron complex rescues the mousekidney from ischemia-reperfusion injury (19). However,in studies using NGAL-knockout mice, no differenceswere noted in ischemic injury to the kidney betweenNGAL�/� mice and their wild-type counterparts (20).More recently, expression of NGAL was evaluated as abiomarker for nephritis in childhood-onset systemiclupus erythematosus, and was found to correlate withrenal disease activity and renal damage (21). In theepidermis, expression of NGAL has been found to belocalized to the hair follicle compartments in normaltissue, particularly to the inner root sheath and infun-dibulum, and to be up-regulated in psoriatic skin lesions(22).

Only a few studies limited to immunohistochem-ical analysis of the expression of NGAL in cartilage havebeen published. In developing rat embryos, expressionof NGAL has been found to be localized to the hyper-trophic region of growth plate cartilage and to beparticularly enriched in prehypertrophic chondrocytes(23). In non-OA cartilage, expression of NGAL has notbeen detected, while high levels of expression have beenfound in OA cartilage (23). Previous reports have doc-

NGAL IN OSTEOARTHRITIS 3327

umented the expression of MMP-9 and the presence of�125-kd gelatinase activity in OA SF (refs. 24 and 25;for review, see ref. 26), but this high molecular weight(HMW) gelatinase activity has not been characterized.Since NGAL has been shown to bind with MMP-9 (15),the purpose of this study was to examine the expressionof NGAL in OA joints, using a combination of molecu-lar, immunologic, and biochemical approaches to deter-mine whether NGAL forms a complex with MMP-9,giving rise to the HMW gelatinase activity in OA SF(24).

MATERIALS AND METHODS

Synovial fluid collection. The study was approved bythe Institutional Review Board of the University HospitalsCase Medical Center. Banked SF from 5 OA patients, storedat �80°C, was used in the analyses.

Zymography and image analysis. SF (10 �l) was mixedwith 90 �l of phosphate buffered saline (PBS) (pH 7.5), andcentrifuged at 10,000g for 5 minutes at 4°C, using Biofuge(Sorvall, Wilmington, DE) to remove any flocculate material.Gelatin zymography was performed essentially as previouslydescribed (15). Briefly, 10 �l of diluted SF was mixed withnonreducing sample buffer (4% sodium dodecyl sulfate [SDS],0.15M Tris [pH 6.8], and 20% [volume/volume] glycerol con-taining 0.05% [weight/volume] bromphenol blue) and resolvedon a 7.5% polyacrylamide gel containing copolymerized 0.2%gelatin (Bio-Rad, Hercules, CA). After electrophoresis, gelswere washed twice, for 15 minutes each time, with 2.5% TritonX-100. Digestion was carried out by incubating the gel in thegelatinase buffer (50 mM Tris HCl [pH 7.6], 10 mM CaCl2, 50mM NaCl, and 0.05% Brij-35) at 37°C for 16 hours. The gelwas stained with 0.1% Coomassie brilliant blue R350 (GEHealthcare, Piscataway, NJ), and the locations of gelatinolyticactivity were revealed as clear bands on a background ofuniform light blue staining.

Electrophoretic migration of MMPs in the SF wascompared with known gelatinase molecular weight standards,as previously described (27). After development, gel imageswere captured using AlphaInnotech Imaging System (AlphaInnotech, San Leandro, CA), and the clear bands were ana-lyzed using UnScan-It software (Silk Scientific, Orem, UT).Each band was scanned 3 times, and the mean band intensity(pixels per band) was used as a measure of enzymatic activityin the band.

Protein electrophoresis and Western immunoblotting.The protein concentration of SF samples was determined usingthe Lowry method (28) (Bio-Rad). In each well of a minigel, 30�g of total protein was loaded and resolved by SDS–polyacrylamide gel electrophoresis (PAGE) under nonreduc-ing conditions. Resolved proteins were electrophoreticallytransferred to nitrocellulose membranes (Bio-Rad). The mem-branes were blocked with 5% lowfat dry milk in Tris bufferedsaline–Tween (TBST) for 2 hours at room temperature, fol-lowed by incubation with primary antibody (anti–MMP-9[MAB911; R&D Systems, Minneapolis, MN] or anti-NGAL[MAB1757; R&D Systems]) at 4°C for 16 hours. Blots were

washed with TBST for 10 minutes and incubated with horse-radish peroxidase–conjugated secondary anti-mouse Ig(Pierce, Rockford, IL) or anti-rat Ig antibody (Southern Bio-technology, Birmingham, AL) diluted 1:5,000 (for NGAL) and1:2,000 (for MMP-9) in TBST for 2 hours at room tempera-ture. Immunoreactive proteins were visualized using enhancedchemiluminescence (Pierce).

Immunoprecipitation (IP) and immunodepletion. Analiquot of SF (200 �g protein) was mixed with 250 �l of lysisbuffer (10 mM Tris [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1mM EGTA, 1% Triton X-100) and 5 �l protein A/G, stirredfor 15 minutes, and centrifuged for 1 minute at 10,000 revolu-tions per minute. The clear supernatant was transferred to afresh tube, 5 �g of anti-human MMP-9 antibody (MAB911)was added, and the mixture was incubated overnight at 4°C ona rotor. Immune complexes that formed were mixed with 10 �lprotein A/G, incubated for 2 hours at 4°C on a rotor, andcollected by centrifugation (1 minute at 10,000 rpm). Thepellet was washed with cold lysis buffer, recentrifuged asdescribed above, and resuspended in 50 �l of lysis buffer.Aliquots of the immunoprecipitated proteins were resolved bySDS-PAGE, and the presence of MMP-9 or NGAL wasdetermined by Western immunoblotting. For immunodeple-tion analysis, an aliquot of SF (200 �g protein) was subjectedto 2 rounds of IP with either anti-NGAL or anti–MMP-9antibody, and the supernatants were used for gelatin zymog-raphy.

Quantitation of glycosaminoglycans (GAGs). Full-thickness OA cartilage slices approximately equal in size wereprepared from femoral head cartilage, which was macroscop-ically unaffected as determined by staining with India ink.Cartilage slices were washed with sterile PBS, and 2 cartilagepieces (�150 mg wet weight) were transferred to each well ofa 24-well flat-bottomed plate in Dulbecco’s modified Eagle’smedium/F-12 (Cellgro; Mediatech, Herndon, VA) containing10% fetal bovine serum and antibiotics, and cultured for 24hours in a tissue culture incubator at 37°C (5% CO2 and 95%air). Cartilage explants were cultured overnight without serumand then treated for 48 hours in fresh medium without serum,with either 10 �g/liter IL-1�, 100 �g/liter proMMP-9, 4mg/liter NGAL, 100 �g/liter proMMP-9 and 4 mg/liter NGAL,or 100 �g/liter proMMP-9, 4 mg/liter NGAL, and 10 �g/literIL-1�. Untreated cartilage was used as a control. At the end ofthe incubation, conditioned medium was collected from eachgroup and a 100-�l aliquot was used to estimate the total GAGrelease into the medium, as previously described (29). Valueswere calculated from a standard curve prepared using whalechondroitin 4-sulfate (Sigma, St. Louis, MO) and were ex-pressed as �g of GAG released per 150 mg of cartilage.

Modulation of MMP-9 autodegradation by NGAL.Recombinant human MMP-9 (R&D Systems) was diluted ingelatinase buffer (50 mM Tris, 10 mM CaCl2, 150 mM NaCl,and 0.05% Brij-35 [pH 7.5]) to a concentration of 2 �g/liter. Todetermine whether the HMW gelatinase activity comprisedNGAL and MMP-9, human recombinant MMP-9 and NGALwere mixed at a 1:60 ratio in gelatinase buffer, incubated at37°C for 1 hour, and analyzed by gelatin zymography. Todetermine whether the interaction of NGAL with MMP-9stabilized gelatinase activity, crosslinked recombinant humanNGAL and MMP-9 were incubated for up to 2 hours in 1�gelatinase buffer at 37°C. Aliquots of the NGAL–MMP-9

3328 GUPTA ET AL

mixture were collected at different time points, kept on ice,and subjected to gelatin zymography. The band intensitieswere analyzed as described above.

Statistical analysis. Experiments were repeated toensure reproducibility of the data. Data were analyzed byanalysis of variance, followed by Tukey’s post hoc test, usingSigmaPlot, version 9.0 or SAS/STAT, version 8.2. Values arepresented as the mean � SD. P values less than 0.05 wereconsidered significant.

RESULTS

MMP activity in OA SF. Although HMWMMP-9 enzymatic activity has been recognized in OASF in earlier studies (30), the 125-kd MMP activity(24,25) has not previously been characterized. Since theapparent size of the HMW synovial fluid MMP activitynoted in these and other studies was similar to the sizereported for the NGAL–MMP-9 complex purified fromhuman neutrophils (31) and also shown to be present inbreast cancer patients (15), we investigated the possibil-ity that this HMW (�125 kd) MMP activity in OA SFmight represent an NGAL–MMP-9 complex.

Electrophoretic analysis showed that the humanNGAL–MMP-9 complex purified from neutrophils andthe 125-kd SF MMP activity in OA SF comigrated onthe gels (Figure 1), suggesting that the 125-kd SF MMPcould indicate an NGAL–MMP-9 complex. Further-more, results of gelatin zymography revealed that SFsamples obtained from patients with OA were rich in72-kd, 92-kd, and 125-kd MMP activity (Figure 1). MMPactivity at 92 kd and 72 kd has previously been deter-mined to represent MMP-9 and MMP-2, respectively(9,24,31). All of the SF samples also contained thepartially proteolyzed forms of MMP-9 (�92 kd), indi-

cating the presence of active MMP-9 in the OA SFsamples.

Specificities of the anti-NGAL and anti–MMP-9antibodies. Because confirmation of the formation ofthe NGAL–MMP-9 complex, and of its presence in SFas HMW collagenolytic activity, was dependent on theuse of highly specific antibodies against the 2 proteins,we first ascertained the specificity of the antibodiesselected for these analyses. Western blots were preparedusing recombinant human NGAL, recombinant humanMMP-9, and recombinant human MMP-13 proteins.Blots were probed with anti-NGAL, anti–MMP-9, andanti–MMP-13 antibodies, and the results are shown inFigure 2. No cross-reactivity was noted; thus, thesefindings established the specificity of the antibodies andindicated that they were suitable for determining thepresence of the target proteins in OA SF.

Western blot analysis of OA SF. We next inves-tigated whether the HMW (�130 kd) collagenolyticactivity detected in OA SF (Figure 1) was a complex ofMMP-9 and NGAL. For these analyses SF samples weresubjected to Western immunoblotting, using the anti-

Figure 1. Comigration of the neutrophil gelatinase–associated lipoca-lin (NGAL)–matrix metalloproteinase 9 (MMP-9) complex and thehigh molecular weight gelatinase activity present in synovial fluid (SF).SF samples from 3 osteoarthritis patients (lanes 1–3) and NGAL–MMP-9 complex purified from human neutrophils (lane 4) wereanalyzed by gelatin zymography. The positions of MMP-2, MMP-9,and the NGAL–MMP-9 complex are shown. Precision Plus ProteinStandards (Bio-Rad) were used as molecular size markers.

Figure 2. Specificity of the antibodies used for immunoprecipitationand Western blot (WB) analysis of synovial fluid from osteoarthritispatients. Immunoblots of recombinant human NGAL (lane 1), humanMMP-13 (lane 2), and human MMP-9 (lane 3) were probed withantibodies against the respective proteins. Each antibody recognizedonly a single band, indicating high specificity of the antibodies. SeeFigure 1 for other definitions.

NGAL IN OSTEOARTHRITIS 3329

bodies shown to be specific for human MMP-9 andhuman NGAL (Figure 2). Purified neutrophil MMP-9–NGAL complex (EMD Biosciences, Darmstadt, Ger-many), along with SF samples from different OA pa-tients, was resolved by SDS-PAGE under nonreducingconditions and transferred to nitrocellulose membranes.When this blot was probed with either the anti–MMP-9antibody (Figure 3A) or the anti-NGAL antibody (Fig-ure 3B), each antibody recognized the HMW proteincomplex in the SF as well as the purified NGAL–MMP-9complex used as a positive control in these experiments.

Importantly, the anti–MMP-9 antibody reactedwith the �125-kd protein complex in the SF and thepurified NGAL–MMP-9 complex and also detected thefree MMP-9 of �92-kd with no cross-reactivity withMMP-2 (Figure 3A). Interestingly, when a similarlyprepared Western blot was probed using the anti-humanNGAL antibody, the anti-NGAL antibody recognizedthe purified neutrophil NGAL–MMP-9 complex as wellas the NGAL dimers present in the SF and the recom-binant human NGAL run on the same gel as a positivecontrol (Figure 3B), again showing the specificity of theantibody used. These results suggested that the HMWcollagenolytic activity in OA SF (Figure 1) was a com-plex of NGAL and MMP-9.

Findings of IP and immunodepletion analysis.The source of the �125-kd SF MMP activity was furtherverified by IP and by immunodepletion analysis. The

association of NGAL and MMP-9 in OA SF becameevident when proteins immunoprecipitated with anti-bodies specific for human NGAL were resolved bySDS-PAGE and the blot probed with antibodies specificfor human MMP-9 (Figure 4A). In other experiments,when SF was subjected to 2 rounds of immune complexformation using anti–MMP-9 antibody, anti–MMP-9antibody specifically immunodepleted the �125-kd SFMMP activity as well as proMMP-9 and activatedMMP-9 activity (Figure 4B). When anti-NGAL antibodywas used for immunodepletion analysis, the HMW com-plex with gelatinase activity was depleted without havingany effect on any other MMP activity (results notshown). These results further supported the notion thatthe �125-kd MMP activity in OA SF was a complex ofMMP-9 and NGAL.

Reconstitution of the NGAL–MMP-9 complexand protection of MMP-9 activity by NGAL in vitro.Once the �125-kd gelatinase activity in OA SF wasidentified as representing a complex of NGAL andMMP-9, we performed in vitro reconstitution experi-ments. Recombinant human MMP-9 and NGAL weremixed at a molar ratio of 1:60 in a gelatinase buffer witha pH similar to that of SF (pH 7.5). After incubation at37°C for 1 hour, formation of the NGAL–MMP-9complex was monitored by gelatin zymography. Asshown in Figure 5A, mixing of NGAL and MMP-9generated the HMW gelatinase activity present in OA

Figure 3. A, Western blot analysis of SF from different osteoarthritis (OA) patients, using anti-human MMP-9 antibody. SF proteins were resolvedby denaturing sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and the blot was probed with anti–MMP-9 antibody. Lanes1 and 2, SF samples from 2 different OA patients; lane 3, recombinant NGAL; lane 4, NGAL–MMP-9 complex purified from human neutrophils;lane 5, recombinant MMP-9 (positive control). These results further confirmed the presence of MMP-9 and the NGAL–MMP-9 complex in OA SF.B, Western blot analysis of SF from different OA patients, using anti-human NGAL antibody. SF proteins were resolved by nondenaturingSDS-PAGE, and the blot was probed with anti-NGAL antibody. Lanes 1–3, SF samples from 3 different OA patients; lane 4, recombinant NGAL;lane 5, NGAL–MMP-9 complex purified from human neutrophils. Immunoreactive proteins at �50 kd are NGAL dimers, which are detected undernonreducing conditions. See Figure 1 for other definitions.

3330 GUPTA ET AL

SF. No complex formation was detected in the controlexperiment, in which recombinant MMP-9 alone wassubjected to the same reconstitution conditions (resultsnot shown). These findings demonstrated that MMP-9and NGAL formed a complex similar to that present inOA SF, and that this complex retained the gelatinaseactivity.

To determine whether association of MMP-9with NGAL stabilizes MMP-9 and protects MMP-9against autocatalysis, recombinant human MMP-9 andrecombinant human NGAL were crosslinked and eithernot subjected to incubation or incubated at 37°C for 1 or2 hours. When human MMP-9 was incubated alone, asignificant decrease in MMP-9 activity over time wasobserved (Figure 5B). The loss of MMP activity due toenzymatic degradation has also been reported by other

investigators (15). In the present study, the loss ofenzymatic activity was blocked by an MMP inhibitor(results not shown). This indicated that the loss ofMMP-9 activity was due to autocleavage, supporting theview that MMP-9 loses activity as a result of autodegra-dation. In contrast, after incubation of crosslinkedNGAL–MMP-9, levels of MMP-9 activity remaining at 2hours were significantly higher than those observed withMMP-9 incubated alone (P � 0.05) (Figure 5C). Takentogether, these results further support the finding thatNGAL is capable of protecting MMP-9 against autode-gradation and therefore contributes to the preservationof MMP-9 activity in vitro, and possibly in vivo as well.

Enhancement of MMP-9–mediated degradationof human cartilage matrix in the presence of NGAL invitro. The functional consequences of protection ofMMP-9 activity by NGAL in relation to cartilage degra-dation in OA were also investigated. For these analyses,human OA cartilage explants were incubated for 48hours in the presence of either IL-1� alone, recombinantNGAL alone, proMMP-9 alone, proMMP-9 in the pres-ence of NGAL, or proMMP-9 in the presence of NGALand IL-1� (Figure 6). Cartilage matrix degradation wasmeasured by assaying for the release of proteoglycansinto conditioned medium. Cartilage explants incubatedalone showed some release of GAG into the conditionedmedium, which did not increase over time (data notshown). This value (�20 �g/150 mg cartilage) was takento be the basal level of GAG release. Cartilage explantsincubated with recombinant NGAL alone also showedno increase in GAG release over the 48-hour period(P � 0.05). Cartilage explants stimulated with IL-1�showed significantly higher levels of GAG released intothe culture medium compared with control or withexplants incubated with recombinant NGAL alone (P �0.05).

The level of GAG released into the medium wasincreased when cartilage explants were incubated withproMMP-9, an unexpected finding since the proMMPsare not active. To determine whether the commercialproMMP-9 contained some contaminating MMP activ-ity, we performed zymography and Western immuno-blotting using 2 ng of proMMP-9 and discovered thatthis commercial preparation contained both proMMP-9(92 kd) and active MMP-9 (85 kd) but no other contam-inating MMP activity (results not shown). Two othercommercial preparations (from Calbiochem, San Diego,CA) showed a similar pattern of activity.

When cartilage explants were incubated in me-dium containing proMMP-9, IL-1�, and NGAL, higherlevels of proteoglycan release occurred (P � 0.05)

Figure 4. A, Western blot (WB) analysis of proteins immunoprecipi-tated from SF samples from different osteoarthritis (OA) patients,using anti-human NGAL antibody. Immunoprecipitated proteins wereresolved by nondenaturing sodium dodecyl sulfate–polyacrylamide gelelectrophoresis, and the blot was probed with anti-human MMP-9antibody. Lane 1, Immunoprecipitation (IP) with irrelevant mouseIgG; lanes 2–4, IP with anti-NGAL antibody using SF from differentpatients with OA; lane 5, NGAL–MMP-9 complex purified fromhuman neutrophils (positive control). B, Results of immunodepletionanalysis of SF using anti–MMP-9 antibody. SF protein (200 �g) wasmixed with a pretitered amount of anti–MMP-9 antibody or a nonim-mune control antibody and subjected to 2 rounds of IP. The remaininggelatinase activity in the final supernatant was detected by gelatinzymography. The �125-kd gelatinase activity was depleted by theanti–MMP-9 antibody (immunodepleted) but not by the controlantibody (nonimmunodepleted). See Figure 1 for other definitions.

NGAL IN OSTEOARTHRITIS 3331

(Figure 6). Since IL-1�–stimulated chondrocytes pro-duce MMP-9 and enzymes that can activate MMP-9, itwas expected that exogenously added MMP-9 would beactivated and MMP-9 activity stabilized by NGAL,resulting in enhanced cartilage degradation. The resultssupported this hypothesis. Our results demonstratedthat NGAL alone was not a catabolic factor for cartilagebut likely contributed to cartilage degradation by pro-tecting MMP-9 against degradation, resulting in pro-longed MMP-9 enzymatic activity.

DISCUSSION

This is the first study to demonstrate the presenceof NGAL in OA SF and to characterize the HMW MMPactivity seen in OA SF as a complex of NGAL andMMP-9. Our findings are supported by several indepen-dent lines of evidence. First, the HMW MMP activity inSF comigrated with the 125-kd purified human neutro-phil NGAL–MMP-9 complex. Second, antibodies spe-cific for NGAL and human MMP-9 consistently re-

Figure 5. A, Reconstitution of NGAL–MMP-9 complex in vitro. Recombinant human MMP-9 and NGAL were mixed at a ratio of 1:60 in agelatinase buffer with a pH similar to that of human SF (pH 7.5) and incubated at 37°C for 1 hour. An aliquot of the mixture (lane 1) and 2 ng ofNGAL–MMP-9 complex purified from human neutrophils (positive control) (lane 2) were subjected to gelatin zymography. Mixing of MMP-9 andNGAL resulted in the generation of �125-kd gelatinase activity. B, Protection of MMP-9 activity by NGAL in vitro. Crosslinked NGAL–MMP-9complex (lanes 2, 4, and 6) and recombinant human MMP-9 alone (lanes 1, 3, and 5) were incubated at 37°C for different time periods. Enzymaticactivity remaining at each time point was determined by gelatin zymography. C, Activity levels of NGAL–MMP-9 compared with activity levels ofMMP-9 alone. Bands were scanned 3 times, and the intensity of each band was recorded as pixels per band. Bars show the mean and SD. At 1 hourand at 2 hours, NGAL–MMP-9 showed significantly higher levels of activity than MMP-9 alone (P � 0.05). See Figure 1 for definitions.

3332 GUPTA ET AL

vealed the 125-kd protein band with HMW MMPactivity in OA SF. Third, these same antibodies specifi-cally immunoprecipitated the HMW MMP complex andsuccessfully immunodepleted the �125-kd MMP activityfrom OA SF. Finally, the HMW MMP activity wasreconstituted in vitro by mixing recombinant MMP-9and recombinant NGAL.

Previous studies have documented the presenceof HMW MMP activity in the SF of arthritis patients(9,24,25), but these HMW gelatinases were not charac-terized. The NGAL–MMP-9 complex detected in OASF may be from joint-infiltrating cells, or possibly fromcytokine-stimulated chondrocytes and synovial cells. Hu-man chondrocytes are known to produce MMP-9 whenstimulated with IL-1�, but whether they also produceNGAL has not yet been determined. Our results alsodemonstrated that association of NGAL with MMP-9protected MMP-9 against autodegradation, consistentwith the findings of previous studies (15). In addition,the findings of the present study indicated that thepreservation of MMP-9 activity in OA SF by NGAL mayhave a significant role in OA pathogenesis. This issupported by our results showing that significantlyhigher amounts of GAG were released from humancartilage explants when they were stimulated with IL-1�in the presence of recombinant MMP-9 and NGAL, as

compared with cartilage explants incubated with recom-binant MMP-9 alone or with IL-1� alone (Figure 6).

Taken together, these results and the results ofprevious studies (9,24,25) suggest that MMP-9 activitymay be more damaging to cartilage matrix when it isprotected by NGAL against autodegradation. The de-tection of the NGAL–MMP-9 complex in OA SF sug-gests that protection of MMP-9 also occurs in vivo. Thus,induction and expression of NGAL may be an importantcontributing factor to the pathogenesis and progressionof cartilage matrix degradation in OA.

NGAL was first identified as a matrix proteinsince it was copurified with human neutrophil gelatinaseB, and association of NGAL and MMP-9 was shown togive rise to an �135-kd gelatinase (31,32). Previousstudies have shown that NGAL and MMP-9 are storedin specific granules in neutrophils, but MMP-9 has alsobeen detected in gelatinase granules (33,34).

High-level expression of NGAL can be inducedin the epithelial and endothelial cells of patients withinflammatory diseases, such as Kawasaki disease (35),classic antineutrophil cytoplasmic antibody–positive sys-temic vasculitis (36), and atherosclerosis (37). Bacterialferric siderophores are ligands for NGAL, and theability of NGAL to sequester iron is important for itsbacteriostatic effect in vivo (38–40). NGAL is alsothought to mediate inflammatory responses by seques-tering neutrophil chemoattractants, particularlyN-formylated tripeptides, and possibly leukotriene B4, aswell as platelet-activating factor (21). Expression ofNGAL in epithelial cells has been shown to be depen-dent on the activation of master transcription factorNF-�B and the expression of cofactor ��B-�,which isinduced only by IL-1� (18,41). However, whether such amechanism occurs in other tissue, including cartilage, isnot known.

Normal expression of human NGAL is restrictedto breast, lung, trachea, and bone marrow (16), and onlya limited number of studies have been published on theexpression of NGAL in cartilage and OA. Although itsexpression has been noted in OA cartilage (23), theexpression and regulation of NGAL in chondrocytes hasnot yet been elucidated. In addition, information regard-ing the function of NGAL in arthritis is lacking. To ourknowledge, this is the first study to demonstrate apossible role of NGAL in OA pathogenesis.

MMPs are the primary enzymes involved incartilage collagen digestion (1), and expression ofMMP-9 in OA SF has previously been reported (20).The important role of MMP-9 in long bone developmentis well established (42), and it is known that multiple

Figure 6. Enhancement of cartilage degradation in the presence ofNGAL in vitro. Human cartilage explants were incubated for 48 hourswithout treatment (control), or with interleukin-1� (IL-1�) alone,proMMP-9 alone, NGAL alone, proMMP-9 in the presence of NGAL,or proMMP-9 in the presence of NGAL and IL-1�. Glycosaminogly-can (GAG) release was quantified by dimethylmethylene blue assayand used as a measure of cartilage matrix degradation. Bars show themean and SD from 3 experiments performed in triplicate. Except forthe comparison of proMMP-9 alone versus NGAL alone, significantdifferences (P � 0.05) were found for each treatment versus controland for each treatment versus every other treatment. See Figure 1 forother definitions.

NGAL IN OSTEOARTHRITIS 3333

collagenase forms are expressed in resorbing cartilage,but recent studies have also shown that MMP-9 is thepredominant gelatinase in diseased cartilage (43). Aprevious study showed that MMP-9 acted on the pro-teoglycan aggrecan in the interglobular domain close tothe hyaluronan binding region (44). MMP-9 may alsohave cleaved cartilage collagens, releasing GAG in theprocess. Taken together, these findings and the results ofthe present study indicate that, because of its protectiveeffect on MMP-9, NGAL likely contributes to OApathogenesis and progression and thus may be an im-portant target for therapy.

In conclusion, we have identified the HMWgelatinase activity in OA SF as a complex of MMP-9 andNGAL. We have also shown that NGAL protectedMMP-9 by preventing the degradation of MMP-9 due toits own enzymatic activity, in vitro and possibly in vivo.The mechanism of this protection has yet to be identi-fied.

AUTHOR CONTRIBUTIONS

Dr. Haqqi had full access to all of the data in the study andtakes responsibility for the integrity of the data and the accuracy of thedata analysis.Study design. Haqqi.Acquisition of data. Gupta, Shukla.Analysis and interpretation of data. Gupta, Cowland, Malemud,Haqqi.Manuscript preparation. Gupta, Cowland, Malemud, Haqqi.Statistical analysis. Gupta, Shukla, Malemud.Culture set-up. Shukla.Provision of recombinant NGAL. Cowland.

REFERENCES

1. Malemud CJ, Islam N, Haqqi TM. Pathophysiological mechanismsin osteoarthritis lead to novel therapeutic strategies. Cells TissuesOrgans 2003;174:34–48.

2. Westacott CI, Sharif M. Cytokines in osteoarthritis: mediators ormarkers of joint destruction. Semin Arthritis Rheum 1996;25:254–72.

3. Goldring MB. The role of cytokines as inflammatory mediators inosteoarthritis: lessons from animal models. Connect Tissue Res1999;40:1–11.

4. Martin JA, Buckwalter JA. Aging, articular cartilage chondrocytesenescence and osteoarthritis. Biogerontology 2002;3:257–64.

5. Chandrasekhar S, Harvey AK, Higginbotham JD, Horton WE.Interleukin-1-induced suppression of type II collagen gene tran-scription involves DNA regulatory elements. Exp Cell Res 1990;191:105–14.

6. Horton WE Jr, Udo I, Precht P, Balakir R, Hasty K. Cytokineinducible matrix metalloproteinase expression in immortalized ratchondrocytes is independent of nitric oxide stimulation. In VitroCell Dev Biol Anim 1998;34:378–84.

7. Aigner T, Stoss H, Weseloh G, Zeiler G, von der Mark K.Activation of collagen type-II expression in osteoarthritic andrheumatoid cartilage. Virchows Arch B Cell Pathol Incl MolPathol 1992;62:337–45.

8. Lotz M. Cytokines in cartilage injury and repair. Clin OrthopRelat Res 2001;391 Suppl:S108–15.

9. Yoshihara Y, Nakamura H, Obata K, Yamada H, Hayakawa T,Fujikawa K, et al. Matrix metalloproteinases and tissue inhibitorsof metalloproteinases in synovial fluids from patients with rheu-matoid arthritis or osteoarthritis. Ann Rheum Dis 2000;59:455–61.

10. Smith GN Jr. The role of collagenolytic matrix metalloproteinasesin the loss of articular cartilage in osteoarthritis. Front Biosci2006;3081–95.

11. Pervais S, Brew K. Homology and structure-function correlationsbetween � 1-acid glycoprotein and serum retinol-binding proteinand its relatives. FASEB J 1987;1:209–14.

12. Flower DR. The lipocalin protein family: structure and function.Biochem J 1996;318:1–14.

13. Yang J, Goetz D, Li JY, Wang W, Mori K, Settik D, et al. An irondelivery pathway mediated by a lipocalin. Mol Cell 2002;10:1045–56.

14. Tong Z, Wu X, Ovcharenko D, Zhu J, Chen C, Kehrer JP.Neutrophil gelatinase-associated lipocalin as a survival factor.Biochem J 2005;391:441–8.

15. Yan L, Borregaard N, Kjeldsen L, Moses MA. The high molecularweight urinary matrix metalloproteinase (MMP) activity is acomplex of gelatinase B/MMP-9 and neutrophil gelatinase-associ-ated lipocalin (NGAL): modulation of MMP-9 activity by NGAL.J Biol Chem 2001;276:37258–65.

16. Cowland JB, Borregaard N. Molecular characterization and pat-tern of tissue expression of the gene for neutrophil gelatinase-associated lipocalin from humans. Genomics 1997;45:17–23.

17. Friedl A, Stoesz SP, Buckley P, Gould MN. Neutrophil gelatinase-associated lipocalin in normal and neoplastic human tissues: celltype-specific pattern of expression. Histochem J 1999;31:433–41.

18. Cowland JB, Sorensen OE, Sehested M, Borregaard N. Neutro-phil gelatinase-associated lipocalin is up-regulated in human epi-thelial cells by IL-1� but not by TNF-�. J Immunol 2003;171:6630–9.

19. Mori K, Lee HT, Rapoport D, Drexler IR, Foster K, Yang J, et al.Endocytic delivery of lipocalin-siderophore-iron complex rescuesthe kidney from ischemia-reperfusion injury. J Clin Invest 2005;115:610–21.

20. Berger T, Togawa A, Duncan GS, Elia AJ, You-Ten A, WakehamA, et al. Lipocalin 2-deficient mice exhibit increased sensitivity toEscherichia coli infection but not to ischemia-reperfusion injury.Proc Natl Acad Sci U S A 2006;103:1834–9.

21. Brunner HI, Mueller M, Rutherford C, Passo MH, Witte D, GromA, et al. Urinary neutrophil gelatinase–associated lipocalin as abiomarker of nephritis in childhood-onset systemic lupus erythem-atosus. Arthritis Rheum 2006;54:2577–84.

22. Seo SJ, Ahn JY, Hong CK, Seo EY, Kye KC, Lee WH, et al.Expression of neutrophil gelatinase-associated lipocalin in skinepidermis. J Invest Dermatol 2006;126:510–2.

23. Zerega B, Cermelli S, Michelis B, Cancedda R, Cancedda FD.Expression of NRL/NGAL (neu-related lipocalin/neutrophil gela-tinase-associated lipocalin) during mammalian embryonic devel-opment and in inflammation. Eur J Cell Biol 2000;79:165–72.

24. Makowski GS, Ramsby ML. Zymographic analysis of latent andactivated forms of matrix metalloproteinase-2 and -9 in synovialfluid: correlation to polymorphonuclear leukocyte infiltration andin response to infection. Clin Chim Acta 2003;329:77–81.

25. Freemont AJ, Hampson V, Tilman R, Goupille P, Taiwo Y,Hoyland JA. Gene expression of matrix metalloproteinases 1, 3,and 9 by chondrocytes in osteoarthritic human knee articularcartilage is zone and grade specific. Ann Rheum Dis 1997;56:542–9.

26. Burrage PS, Mix KS, Brinckerhoff CE. Matrix metalloproteinases:role in arthritis. Front Biosci 2006;11:529–43.

27. Makkowski GS, Ramsby ML. Calibrating gelatin zymograms withhuman gelatinase standards. Anal Biochem 1996;236:353–6.

3334 GUPTA ET AL

28. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein mea-surement with the Folin phenol reagent. J Biol Chem 1951;193:265–75.

29. Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation anddiscrimination of sulphated glycosaminoglycans by use of dimeth-ylmethylene blue. Biochim Biophys Acta 1986;883:173–7.

30. Hibbs MS, Hasty KA, Seyer JM, Kang AH, Mainardi CL. Bio-chemical and immunological characterization of the secretedforms of human neutrophil gelatinase. J Biol Chem 1985;260:2493–500.

31. Kjeldsen L, Johnsen AH, Sengelov H, Borregaard N. Isolation andprimary structure of NGAL, a novel protein associated withhuman neutrophil gelatinase. J Biol Chem 1993;268:10425–32.

32. Tchetverikov I, Ronday HK, van El B, Kiers GH, Verzijl N,TeKoppele JM, et al. MMP profile in paired serum and synovialfluid samples of patients with rheumatoid arthritis. Ann RheumDis 2004;63:881–3.

33. Kjeldsen L, Sengelov H, Lollike K, Nielsen MH, Borregaard N.Isolation and characterization of gelatinase granules from humanneutrophils. Blood 1994;83:1640–9.

34. Borregaard N, Cowland JB. Granules of the human neutrophilicpolymorphonuclear leukocyte. Blood 1997;89:3503–21.

35. Biezeveld MH, van Mierlo G, Lutter R, Kuipers IM, Dekker T,Hack CE, et al. Sustained activation of neutrophils in the course ofKawasaki disease: an association with matrix metalloproteinases.Clin Exp Immunol 2005;141:183–8.

36. Ohlsson S, Hellmark T, Pieters K, Sturfelt G, Wieslander J,Segelmark M. Increased monocyte transcription of the proteinase3 gene in small vessel vasculitis. Clin Exp Immunol 2005;141:174–82.

37. Elneihoum AM, Falke P, Hedblad B, Lindgarde F, Ohlsson K.Leukocyte activation in atherosclerosis: correlation with risk fac-tors. Atherosclerosis 1997;131:79–84.

38. Goetz DH, Holmes MA, Borregaard N, Bluhm ME, RaymondKN, Strong RK. The neutrophil lipocalin NGAL is a bacteriostaticagent that interferes with siderophore-mediated iron acquisition.Mol Cell 2002;10:1033–43.

39. Holmes MA, Paulsene W, Jide X, Ratledge C, Strong RK.Siderocalin (Lcn 2) also binds carboxymycobactins, potentiallydefending against mycobacterial infections through iron seques-tration. Structure 2005;13:29–41.

40. Flo TH, Smith KD, Sato S, Rodriguez DJ, Holmes MA, StrongRK, et al. Lipocalin 2 mediates an innate immune response tobacterial infection by sequestrating iron. Nature 2004;432:917–21.

41. Cowland JB, Muta T, Borregaard N. IL-1�-specific up-regulationof neutrophil gelatinase-associated lipocalin is controlled by I�B-�.J Immunol 2006;176:5559–66.

42. Colnot C, Thompson Z, Miclau T, Werb Z, Helms JA. Alteredfracture repair in the absence of MMP9. Development 2003;130:4123–33.

43. Milner JM, Rowan AD, Cawston TE, Young DA. Metalloprotein-ase and inhibitor expression profiling of resorbing cartilage revealspro-collagenase activation as a critical step for collagenolysis.Arthritis Res Ther 2006;8:R142.

44. Fosang AJ, Neame PJ, Last K, Hardingham TE, Murphy G,Hamilton JA. The interglobular domain of cartilage aggrecan iscleaved by PUMP, gelatinases, and cathepsin B. J Biol Chem1992;267:19470–4.

NGAL IN OSTEOARTHRITIS 3335