Trefoil factor 3 is induced during degenerative and inflammatory joint disease, activates matrix metalloproteinases, and enhances apoptosis of articular cartilage chondrocytes

ARTHRITIS & RHEUMATISMVol. 62, No. 3, March 2010, pp 815–825DOI 10.1002/art.27295© 2010, American College of Rheumatology

Trefoil Factor 3 Is Induced During Degenerativeand Inflammatory Joint Disease,

Activates Matrix Metalloproteinases, and EnhancesApoptosis of Articular Cartilage Chondrocytes

Sophie Rosler,1 Tobias Haase,1 Horst Claassen,1 Ute Schulze,1 Martin Schicht,1

Dagmar Riemann,1 Jorg Brandt,1 David Wohlrab,1 Brigitte Muller-Hilke,2 Mary B. Goldring,3

Saadettin Sel,1 Deike Varoga,4 Fabian Garreis,1 and Friedrich P. Paulsen1

Objective. Trefoil factor 3 (TFF3, also known asintestinal trefoil factor) is a member of a family ofprotease-resistant peptides containing a highly con-served motif with 6 cysteine residues. Recent studieshave shown that TFF3 is expressed in injured cornea,where it plays a role in corneal wound healing, but notin healthy cornea. Since cartilage and cornea havesimilar matrix properties, we undertook the presentstudy to investigate whether TFF3 could induce anabolicfunctions in diseased articular cartilage.

Methods. We used reverse transcriptase–polymerase chain reaction, Western blot analysis, andimmunohistochemistry to measure the expression ofTFF3 in healthy articular cartilage, osteoarthritis

(OA)–affected articular cartilage, and septic arthritis–affected articular cartilage and to assess the effects ofcytokines, bacterial products, and bacterial superna-tants on TFF3 production. The effects of TFF3 onmatrix metalloproteinase (MMP) production were mea-sured by enzyme-linked immunosorbent assay, and ef-fects on chondrocyte apoptosis were studied by caspaseassay and annexin V assay.

Results. Trefoil factors were not expressed inhealthy human articular cartilage, but expression ofTFF3 was highly up-regulated in the cartilage of pa-tients with OA. These findings were confirmed in animalmodels of OA and septic arthritis, as well as in tumornecrosis factor �– and interleukin-1�–treated primaryhuman articular chondrocytes, revealing induction ofTff3/TFF3 under inflammatory conditions. Applicationof the recombinant TFF3 protein to cultured chondro-cytes resulted in increased production of cartilage-degrading MMPs and increased chondrocyte apoptosis.

Conclusion. In this study using articular cartilageas a model, we demonstrated that TFF3 supports cata-bolic functions in diseased articular cartilage. Thesefindings widen our knowledge of the functional spec-trum of TFF peptides and demonstrate that TFF3 is amultifunctional trefoil factor with the ability to linkinflammation with tissue remodeling processes in artic-ular cartilage. Moreover, our data suggest that TFF3 isa factor in the pathogenesis of OA and septic arthritis.

The mammalian trefoil factor (TFF) family com-prises 3 protease-resistant peptides of 7–12 kd (TFF1[formerly called pS2], TFF2 [formerly called hSP], andTFF3 [formerly called hP1.B/hITF or ITF]) that have incommon a distinct highly conserved motif of 6 cysteine

Supported in part by the GI Company. Dr. Goldring’s workwas supported by the NIH (grant R01-AG-022021). Dr. Paulsen’s workwas supported by the DFG (grants PA 738/9-1, PA 738/9-2, and PA738/6-1) and the Bundesministerium fur Bildung und Forschung–Wilhelm Roux Program, Halle, Germany (grants FKZ 09/17, FKZ14/24, and FKZ 14/25).

1Sophie Rosler, Tobias Haase, DiplBiochem, Horst Claassen,MD, Ute Schulze, MSc, Martin Schicht, MSc, Dagmar Riemann, MD,Jorg Brandt, MD, David Wohlrab, MD, Saadettin Sel, MD, FabianGarreis, MSc, Friedrich P. Paulsen, MD: Martin-Luther UniversityHalle-Wittenberg, Halle (Saale), Germany; 2Brigitte Muller-Hilke,PhD: University of Rostock, Rostock, Germany; 3Mary B. Goldring,PhD: Hospital for Special Surgery, Weill College of Medicine ofCornell University, New York, New York; 4Deike Varoga, MD:Christian Albrecht University, Kiel, Germany.

Ms Rosler and Mr. Haase contributed equally to this work.Dr. Wohlrab has received speaking fees from Smith &

Nephew (less than $10,000).Address correspondence and reprint requests to Friedrich P.

Paulsen, MD, Department of Anatomy and Cell Biology, MartinLuther University Halle-Wittenberg, Faculty of Medicine, GrosseSteinstrasse 52, D-06097 Halle (Saale), Germany. E-mail:[email protected].

Submitted for publication July 4, 2009; accepted in revisedform November 17, 2009.

815

residues, the so-called trefoil domain (1). These peptidescan interact with mucins and influence mucus viscosity(2), promote migration of epithelial cells in vitro (3), arelinked to antiapoptosis (1), induce cell scattering (4),participate in immune response (5), trigger chemotaxis(6), and have further functions (7). To date, a putativeTFF receptor has not been identified. Therapeutic ef-fects of the TFFs have been shown in several animalmodels of gastrointestinal (GI) damage (7). In a recentstudy, our group demonstrated that TFFs are not presentin healthy cornea. In contrast, in certain corneal diseasestates, TFF3 is produced (8). Functional experimentsrevealed a pivotal role of Tff3 in corneal wound healingmechanisms and indicated broad implications for TFF3as a possible novel strategy to treat corneal disease (9).

Since articular cartilage is, like cornea, an avas-cular tissue and also has other similarities to cornealtissue, such as certain extracellular matrix (ECM) pro-teins, we hypothesized that TFF peptides might havesimilar roles in diseased cartilage, as antiapoptotic andcartilage repair–inducing factors. Indeed, we found that,similar to corneal tissue, TFF peptides are absent inhealthy articular cartilage. However, in certain diseasestates such as osteoarthritis (OA) or septic arthritis,TFF3 is expressed in articular cartilage. Treatment ofcultured chondrocytes with tumor necrosis factor �(TNF�) or interleukin-1� (IL-1�) increased TFF3 ex-pression, and incubation with recombinant human TFF3(rhTFF3) led to a concentration-dependent induction ofmatrix metalloproteinases (MMPs) 1, 3, and 13. Unex-pectedly, rhTFF3 increased caspase 3/7 activity, indicat-ing a proapoptotic effect on articular cartilage. Thesefindings suggest that TFF3 is induced during jointdisease and supports joint degeneration.

MATERIALS AND METHODS

Human tissue. OA cartilage (n � 20 samples) wasobtained, with institutional review board approval, from pa-tients ages 38–76 years who were undergoing knee jointreplacement surgery at the orthopedic surgery department ofMartin-Luther University Halle-Wittenberg or Christian Al-brecht University. Tissue samples were graded according tothe Mankin scale (10), and only samples with moderate tosevere OA (Mankin score 6–14) were included. Healthy carti-lage (n � 10 samples) was obtained from patients ages 21–52years who were undergoing resection arthroplasty because ofan extraarticular tumor (n � 5) or from autopsy cases (ages22–44 years at time of death; n � 5) at the Institute of LegalMedicine, University Hospital Hamburg Eppendorf, Ham-burg, Germany.

Primary chondrocytes and cartilage cell line. Primaryhuman chondrocytes were prepared from knee OA cartilage

(n � 20) and used to examine the expression and regulation ofTFF3 as well as the influence of rhTFF3 in vitro. The sameexperiments were performed with the human chondrocyte cellline C28/I2. This cell line is immortalized with SV40 large Tantigen and expresses proteins such as aggrecan, type IIcollagen, and other markers typical of the differentiatedphenotype (11) that have been used to study the regulation ofgene expression and signaling in response to cytokines andother factors (12). Cells (1 � 106) were seeded in 25-cm2 flasksand cultivated in Dulbecco’s modified Eagle’s medium(DMEM) with 10% fetal calf serum (FCS). When 80%confluence was reached, the medium was changed to serum-free DMEM containing 0.05% bovine serum albumin. IL-1�(10 ng/ml; Immunotools, Friesoythe, Germany) and/or TNF�(10 ng/ml; Immunotools), lipopolysaccharide (LPS) (500 ng/ml; Santa Cruz Biotechnology, Heidelberg, Germany), pepti-doglycan (10 ng/ml; Santa Cruz Biotechnology), or superna-tants of Pseudomonas aeruginosa (1:100) or Staphylococcusaureus (1:15) were added 3 hours later, and incubation wascontinued for 24 hours. Bacteria were produced as recentlydescribed (13).

Primary chondrocytes and the C28/I2 cell line werealso seeded on glass cover slides in 6-well culture plates andincubated with the above-mentioned stimulants. At the end ofeach experiment, the cells on the glass cover slides were fixedfor 5 minutes in 4% paraformaldehyde and processed forimmunohistochemistry. Immunostaining was performed withantibodies to TFF1, TFF2, and TFF3 as recently described indetail (8).

In addition, primary cells and C28/I2 cells were cul-tured and incubated with various amounts of rhTTF3 peptide(glycosylated TFF3, dimeric molecule; kindly provided by theGI Company, Framingham, MA) and with IL-1� and TNF�alone or in combination. Conditioned medium was removedand aliquots were assayed using commercial enzyme-linkedimmunosorbent assay (ELISA) kits for annexin V and caspase.

RNA preparation, complementary DNA (cDNA) syn-thesis, and reverse transcriptase–polymerase chain reaction(RT-PCR). Total RNA was isolated from OA samples (n �20), healthy control samples (n � 10), stimulated primarychondrocytes (from 5 different OA samples), or C28/I2 cells(106 per 25-cm2 flask) (n � 5) and reverse transcribed using astandard protocol. As an internal control for the integrity ofthe translated cDNA, �-actin was used. Semiquantitative PCRwas performed with the following primer sequences: TFF1forward TTT-GGA-GCA-CAG-AGG-AGG, reverse TTG-AGT-CAA-AGT-CAG-AGC-AG (438 bp); TFF2 forwardAGT-CAG-AAA-CCC-TCC-CCC-C, reverse AAC-ACC-CGG-TGA-GCC-AG (366 bp); TFF3 forward CTG-CCA-GCC-AAG-GAC-AG, reverse CGT-TAA-GAC-ATC-AGG-CTC-CAG (303 bp); �-actin forward CAA-CAG-ATG-GCC-ACG-GCT-GCT, reverse TCC-TTC-TGC-ATC-CTG-TCG-GCA (275 bp). PCR conditions were as follows: preheating for3 minutes at 94°C, 35 cycles of heat denaturation for 40seconds at 95°C, annealing for 40 seconds at 59°C for TFF1and TFF2 or 56°C for TFF3, extension for 40 seconds at 72°C,followed by a final extension for 7 minutes at 72°C. All primerswere synthesized at MWG-Biotech (Ebersberg, Germany).Basepair values were compared with gene bank data. PCRproducts were also confirmed by sequencing with a BigDyeTerminator Cycle Sequencing Kit (Applied Biosystems, Foster

816 ROSLER ET AL

City, CA). To estimate the amount of amplified PCR product,we performed �-actin PCR with specific primers for eachinvestigated cell culture. This additional PCR was performedunder the conditions described above.

Real-time RT-PCR. Quantitative real-time RT-PCRwith SYBR Green Master Mix (Applied Biosystems, Darm-stadt, Germany) as a double-stranded DNA–specific fluores-cent dye was performed using an Opticon 2 system (MJResearch, Waltham, MA). Primer sequences were as follows:TFF3 forward CAG-GAC-TGT-TCA-TCT-CAG-CTT, re-verse CAT-CAG-GCT-CCA-GAT-ATG-AAC-TTT-C; 18Sribosomal RNA (rRNA) forward ACT-CAA-CAC-GGG-AAA-CCT-CAC-C, reverse CAA-GAG-ATG-GCC-ACG-GCT-GCT (18S rRNA was used because its expression was notinfluenced by any of the mediators investigated). PCR wasinitiated for 2 minutes at 50°C, followed by 1.5 minutes at 94°C.The program continued with 30 cycles of 20 seconds at 94°Cand 60 seconds at 62°C. Each assay included duplicates of eachcDNA sample and a “no-template” control for TFF3. Relativeamounts of expression of messenger RNA (mRNA) for TFF3and 18S rRNA were calculated using the ��Ct method. Theexpression of 18S rRNA was used to normalize samples for theamount of cDNA used per reaction. To confirm the amplifi-cation, the resulting real-time PCR products were analyzedwith dissociation curves, visualized in an agarose gel (TFF3 64bp; 18S rRNA, 111 bp), and sequenced by BigDye sequencing.

Western blot analysis. Western blotting was performedaccording to a recently described protocol (14). Anti-TFF3antibody (9) (1:200) was kindly provided by Dr. Daniel K.Podolsky (Boston, MA). For Western blotting, human articu-lar cartilage (20 OA samples, 10 healthy articular cartilagesamples) or cultured cells (OA samples from 5 differentpatients) (standardized ratio 100 mg wet weight to 400 �lbuffer containing 1% sodium dodecyl sulfate and 4%2-mercaptoethanol) was extracted and the protein contentmeasured with a protein assay based on the Bradford dye-binding procedure (Bio-Rad, Hercules, CA). Conjunctivaltissue was used as a positive control.

Animals. Twenty male STR/Ort mice (ages 22–45weeks) were used in the present study. STR/Ort mice aregenetically predisposed to develop an OA-like lesion of themedial tibial cartilage. More than 85% of male STR/Ort miceshow signs of degenerative joint disease in the tibia by the ageof 6 months (15). After the mice were killed the knee jointswere removed and fixed in 4% paraformaldehyde.

Eight male inbred BALB/c mice (ages 8–12 weeks)were studied in experiments using a model of septic arthritis.These mice received intraarticular injections at the frontalaspect of the knee, just below the patella in the intercondylarregion. A volume of 10 �l, containing 107 colony-forming unitsof S aureus/ml, was injected into the knee joint. Animals in thecontrol group (n � 8) were injected with phosphate bufferedsaline (PBS) in the same volume. Six hours after injection ofthe bacteria, the mice were killed and the knee joints wereremoved and fixed in 4% paraformaldehyde.

Joint histology. After fixation, knee joints from theSTR/Ort mice and the BALB/c mice were demineralized,according to a procedure that has been described in detailpreviously (16,17). To study the in vivo expression of Tff3 inOA cartilage of STR/Ort mice and in mice with septic arthritis,we immunostained 6-�m serial histologic sections of joints

affected with OA at different stages (grades 1–3) or septicarthritis with polyclonal anti-rat Tff3 antibody (9) (diluted1:50). The OA-like lesions were classified as described previ-ously (18). Sections of mouse jejunum were used as positivecontrols. Age- and sex-matched BALB/c mice, which do notdevelop OA, were used as control animals for these experi-ments. The animal studies were approved by the institutionalreview board. For immunohistochemical analysis of primarychondrocytes and C28/I2 chondrocytes, cells were seeded onglass cover slides in 6-well culture plates and incubated withstimulants. Immunostaining was performed with antibodies toTFF1 (1:5,000), TFF2 (1:1,000), and TFF3 (1:1,000), as previ-ously described (8).

ELISA. For ELISA, aliquots of supernatant of culturedchondrocytes were analyzed. Primary chondrocytes (1 � 106)were seeded into fresh dishes and cultivated for 72 hours inDMEM/Ham’s F-12 medium (1/1 [volume/volume]) plus 10%FCS. The cultures were changed to serum-free medium, andafter 3 hours of equilibration, the cells were treated for 24hours with rhTFF3 at a concentration of 0.1, 0.5, 1.0, or 5.0mg/ml. Recombinant human glycosylated TFF3 (dimeric mol-ecule) was produced as previously described (19,20). ForELISA, conditioned medium was withdrawn and aliquots wereassayed using the following commercial kits (all from Amer-sham Biosciences, Roosendaal, The Netherlands): RPN 2610(for activated MMP-1), RPN 2613 (for MMP-3), RPN 2614(for MMP-9), RPN 2621 (for MMP-13), RPN 2611 (for tissueinhibitor of metalloproteinases 1 [TIMP-1]), and RPN 2618(for TIMP-2). Signals were detected by chemiluminescencereaction (ECL-Plus; Amersham Biosciences). Data from 3independent experiments performed in triplicate were as-sessed. For ELISA experiments, 3 wells were analyzed for eachsample.

Caspase assay. Caspase 3/7 activities were measuredusing the Apo-ONE homogeneous caspase 3/7 assay kit,according to the instructions of the manufacturer (Promega,Mannheim, Germany). Primary chondrocytes (from 5 differentOA patients) were plated in 96-well cell culture plates (Costar3603; Corning, Corning, NY). After stimulation for 24 hours(each stimulation performed in triplicate with 10 ng/ml IL-1�,10 ng/ml TNF�, and 0.3 mg/ml rhTFF3), cells were lysed withlysis buffer containing caspase substrate Z-DEVD-R100, andincubated at room temperature until analyzed. Assay resultswere measured by detection with a fluorescence microplatereader (Labvision system, BMG Lab Technologies, Offenburg,Germany), and fluorescence was measured at an excitation/emission wavelength of 485/535 nm. Means and SDs of thetriplicate results obtained in the independent experimentswere calculated.

Annexin V assay. Apoptotic cells were analyzed usingthe Annexin-V-FITC apoptosis detection kit (BioVision,Brugg, Switzerland), in which annexin V labels the membraneof early apoptotic cells. After incubation of OA chondrocyteswith stimulants and/or rhTFF3 for 24 hours, floating cells wereharvested from the supernatant, and adherent cells werescratched from the plate with disposable cell lifters and washedwith PBS. Floating and adherent cells were combined prior toincubation with fluorescein isothiocyanate–labeled annexin V.Labeled cells were quantified by 2-dimensional fluorescence-activated cell sorter analysis. A positive control for apoptosis

TFF3 IN DISEASED ARTICULAR CARTILAGE 817

was generated by incubating cells for 14 hours with 100 ng/mlanti-Fas (MBL, Nagoya, Japan).

Statistical analysis. Statistical analysis was performedwith SPSS 11.5 for Windows. P values less than 0.05 wereconsidered significant.

RESULTS

TFFs 1, 2, and 3 are absent in healthy humanarticular cartilage, but TFF3 is expressed in OA carti-lage. To test the hypothesis that TFFs are absent inhealthy articular cartilage but are expressed duringcertain disease states, as seen for TFF3 at the ocularsurface (8), we compared healthy human articular car-tilage with articular cartilage from patients with OA.Indeed, healthy human articular cartilage did not ex-press TFF3 at the mRNA or protein level, whereasTFF3 was expressed in OA cartilage (Figures 1A and B).Immunohistochemical analysis confirmed the presenceof TFF3 peptide in cartilage with OA at all gradesinvestigated (Figure 1C). In contrast, TFFs 1 and 2 werenot detected in either healthy or OA cartilage (resultsnot shown).

Tff3 is not expressed in healthy mouse cartilage,but is expressed in OA and septic inflamed cartilage. Aswas observed in human cartilage, healthy mouse carti-lage did not react with an antibody against Tff3 (Figure2A). To further evaluate the expression of Tff3 peptide,we used 2 mouse models. First, we studied the STR/Ortmouse model of OA. When the earliest signs of OAoccurred in STR/Ort mice, clear cytoplasmic reactivitywith Tff3 peptide was visible in superficial chondrocytesunderlying areas of early damage (Figure 2A). Duringthe middle stage of OA, chondrocytes near the tidemarkexpressed Tff3 in areas of cartilage damage (Figure 2A).Late-stage OA in STR/Ort mice was characterized byloss of cartilage, and thus also absence of Tff3 expressionin these areas (Figure 2A). However, chondrocytes neardenuded bone where cartilage was still present, as, forexample, near the insertion zone of the cruciate liga-ments, reacted strongly with the antibody against Tff3(Figure 2B).

Next, we investigated Tff3 induction in articularchondrocytes in a mouse model of septic arthritis. Noimmunohistochemical staining for Tff3 was detectable inarticular cartilage in the control group that had beeninjected intraarticularly with PBS (Figure 2C). However,6 hours after intraarticular injection of S aureus, Tff3peptide was expressed in articular cartilage from themurine knee joints (Figure 2C). Notably, almost all

chondrocytes of the articular cartilage showed consistentcytoplasmic labeling.

Proinflammatory cytokines, bacterial products,and bacterial supernatants induce endogenous TFF3production. Primary human chondrocytes isolated fromknee OA cartilage and the immortalized chondrocytecell line C28/I2 were used to examine the expression andregulation of TFF3. Cultured chondrocytes from pa-tients, as well as the C28/I2 cells, did not express TFF1or TFF2, but, with no stimulation, TFF3 expression wasdetectable at both the mRNA and the protein levels(Figures 3A and B).

We next investigated whether the inflammatorycytokines IL-1� and TNF� could influence TFF3 geneexpression. Real-time PCR revealed induction of TFF3

Figure 1. Expression of trefoil factor 3 (TFF3) in osteoarthritic (OA)cartilage but not in healthy articular cartilage. A, TFF3 mRNA wasdetected, by reverse transcriptase–polymerase chain reaction, in con-trol tissue (conjunctiva) (lane 1) and OA cartilage (lanes 4 and 5), butnot in healthy articular cartilage (lanes 2 and 3). M � marker. B, TFF3(13 kd) was detected, by Western blot analysis, in control tissue (lane1) and OA cartilage (lane 2), but not in healthy articular cartilage (lane3). C, Reactivity with TFF3 (red staining) was visible, by immunohis-tochemistry staining, in chondrocytes and chondrocyte clusters in OAcartilage from all stages of disease (arrows), whereas healthy cartilageexhibited no reactivity. Bars � 43 �m.

818 ROSLER ET AL

mRNA after stimulation of primary chondrocytes for 6hours with 10 ng/ml of IL-1� or TNF� alone or incombination at physiologically relevant concentrations(Figure 3C), but not in C28/I2 chondrocytes (data notshown). Six hours after exposure to IL-1� (this time

Figure 2. Induction of Tff3 during development of OA in STR/Ortmice and during septic joint inflammation in BALB/c mice. A, Sagittalsections through the tibial articular joint area of STR/Ort mice wereimmunohistochemically stained for Tff3. Sections obtained during theearly and middle stages of OA (grades 1 and 2) exhibited positivestaining of chondrocytes near the joint cavity (red staining andarrows). Macroscopically healthy tissue (grade 0) and late-stage OAtissue (grade 3) showed no reactivity. (In grade 3 OA, articularcartilage is completely lost, and reactivity with subchondral bone wasnot observed.) TM � tidemark. B, Joints with grade 3 OA wereimmunohistochemically stained for Tff3. Cells near abraded articularareas at the insertion of the cruciate ligaments exhibited stronglypositive staining (arrows). C, Sagittal sections through the knee joint ofBALB/c mice (above the femur, below the tibia, in the joint cavity [jc])without treatment (healthy) or after injection of supernatants ofStaphylococcus aureus (SA) were immunohistochemically stained forTff3. Septic joint conditions led to induction of Tff3 in most articularcartilage chondrocytes (arrows). Bars in A and B � 43 �m; bars in C �86 �m. See Figure 1 for other definitions.

Figure 3. Cultured primary articular chondrocytes from patients withosteoarthritis (OA) constitutively express trefoil factor 3 (TFF3), withfurther induction by proinflammatory cytokines, bacterial products, andbacterial supernatants. A, TFF3 mRNA was detected, by reversetranscriptase–polymerase chain reaction (RT-PCR), in control tissue(conjunctiva) (lane 1) and cultured primary articular chondrocytes fromdifferent OA patients (lanes 2 and 3). M � marker. B, TFF3 (13 kd) wasdetected, by Western blot analysis, in control tissue (lane 1) and culturedprimary articular chondrocytes from 2 different patients (lanes 2 and 3).C, Relative TFF3 induction by proinflammatory cytokines (10 ng/mlinterleukin-1� [IL-1�], 10 ng/ml tumor necrosis factor � [TNF�]) wasassessed and quantified by real-time RT-PCR. Values are the mean andSD. � � P � 0.05 versus controls, by Student’s t-test. D, Induction ofTFF3 (red staining and arrows) by proinflammatory cytokines (10 ng/mlIL-1�, 10 ng/ml TNF�) was visualized by immunohistochemistry. Bars �15 �m. E, Induction of TFF3 mRNA by bacterial products (500 ng/mllipopolysaccharide [LPS], 10 ng/ml peptidoglycan [PGN]) and bacterialsupernatants (Pseudomonas aeruginosa [PA; 1:100], Staphylococcus aureus[SA; 1:15]) was assessed by semiquantitative RT-PCR. All of the bacterialproducts and supernatants tested induced TFF3 mRNA. c � control(conjunctiva).

TFF3 IN DISEASED ARTICULAR CARTILAGE 819

point was chosen after performing a time course exper-iment), basal TFF3 mRNA levels in cultured primarychondrocytes increased 3-fold compared with those inunstimulated cells. Addition of TNF� increased expres-sion of TFF3 by nearly 5-fold. Addition of a combinationof IL-1� and TNF� led to TFF3 transcript levels com-parable with those obtained with TNF� alone (Figure3C).

To investigate the effects of IL-1� and TNF� atthe protein level, primary and C28/I2 chondrocytes wereseeded on glass cover slides, stimulated with eitherproinflammatory cytokine or a combination of both atrelevant levels, and analyzed by immunohistochemistry.Very weak reactivity with TFF3 peptide was found injust a few primary chondrocytes under basal conditionswithout stimulation (Figure 3D). Stimulation with IL-1�or TNF�, alone or in combination, led to clear inductionof TFF3, which was visible as strong cytoplasmic stainingin all cultured chondrocytes (Figure 3D). In addition,several cells that were treated with TNF� alone or witha combination of TNF� and IL-1� exhibited signs ofnuclear pycnosis, a typical morphologic sign of apoptosisor necrosis (Figure 3D). Cultured C28/I2 chondrocytesexhibited positive immunoreactivity for TFF3 peptide inlarge cytoplasmically located vacuoles in several cells, inthe absence or presence of proinflammatory cytokines(results not shown).

Primary human chondrocytes were challengedwith LPS, peptidoglycan, P aeruginosa, or S aureus.Semiquantitative RT-PCR revealed induction of TFF3mRNA after challenge with each of these agents (Figure3E).

Recombinant human TFF3 induces MMPs 1, 3,and 13 in a concentration-dependent manner. Sincechondrocyte responses to inflammatory cytokines aregenerally associated with up-regulation of cartilage-degrading enzymes, we investigated the influence ofrhTFF3 on the expression of MMPs, as well as the majorendogenous regulators of their activity, TIMPs 1 and 2.MMP-1, MMP-3, MMP-9, MMP-13, TIMP-1, andTIMP-2 protein could all be detected in cultured humanprimary chondrocytes and in C28/I2 chondrocyte super-natants, by specific ELISA. Exposure for 24 hours torhTFF3 protein at various concentrations (0.1, 0.5, 1,and 5 mg/ml) resulted in a clear concentration-dependent up-regulation of MMPs 1, 3, and 13 inprimary chondrocytes but not in C28/I2 cells, with theexception that MMP-13 levels were slightly increased inC28/I2 chondrocytes (Figure 4). Compared with basalexpression levels in primary chondrocytes, treatmentwith rhTFF3 for 24 hours resulted in up to a 10-fold

increase in MMP-1 (Figure 4A) and a nearly 17-foldincrease in MMP-13 (Figure 4C). The stromelysinMMP-3 was expressed at much higher levels (250 ng/ml)after rhTFF3 treatment (Figure 4B). In contrast,MMP-9 levels in the culture medium were unchanged

Figure 4. Concentration-dependent stimulation of matrix metallopro-teinase (MMP) production by recombinant human trefoil factor 3(rhTTF3) in primary cultured articular chondrocytes, but not in C28/I2chondrocytes. Cultured chondrocytes were exposed to rhTFF3 (0.1,0.5, 1, or 5 mg/ml) for 24 hours. At least 3 individual cultures werestimulated, and at least 3 measurements were obtained with eachstimulation. A, Concentration-dependent up-regulation of MMP-1 (upto 10-fold) in primary chondrocytes treated with rhTTF3, but not inC28/I2 chondrocytes, compared with untreated controls. B,Concentration-dependent up-regulation of MMP-3 (up to 250-fold) inprimary chondrocytes treated with rhTTF3, but not in C28/I2 chon-drocytes, compared with untreated controls. C, Slight concentration-dependent up-regulation of MMP-13 in primary chondrocytes (up to0.2-fold) and in C28/I2 chondrocytes (up to 0.05-fold) treated withrhTTF3, compared with untreated controls. Values are the mean andSD. � � P � 0.01 versus controls, by Student’s t-test.

820 ROSLER ET AL

after exposure of either primary chondrocytes or C28/I2cells to rhTFF3 (data not shown).

A specific ELISA for TIMP-1 revealed down-regulation following treatment with rhTFF3 only at aconcentration of 5 mg/ml (from a basal level of 6 ng/mlto 5 ng/ml in primary chondrocytes and from 6.5 ng/mlto 4.5 ng/ml in C28/I2 chondrocytes) (data not shown).TIMP-2 was not affected by rhTFF3 at any concentra-tion investigated in primary chondrocytes, but was in-

duced with rhTFF3 at a concentration of 5 mg/ml inC28/I2 cells (from a basal level of 0 to 9 ng/ml) (data notshown).

Recombinant human TFF3 acts as a proapop-totic effector on cultured human articular chondrocytes.Because TNF� promoted strong induction of TFF3 incultured primary chondrocytes and this was associatedwith the morphologic aspect of apoptosis (Figure 3D),we assessed a possible proapoptotic role of TFF3 bymeasuring caspase 3/7 activity and quantifying the cellapoptosis rate by annexin V assay. With no stimulation,little apoptosis was detected. Treatment with rhTFF3induced a significant (1.9-fold) increase in caspase 3/7activity, treatment with TNF� or IL-1� alone led to a1.6-fold increase, and treatment with the combination ofTNF� and rhTFF3 significantly increased caspase 3/7activity (2.8-fold) (Figure 5A). IL-1� in combinationwith rhTFF3 also caused an increase in caspase 3/7activity; however, this was not significant. These findingswere confirmed by measuring the cell death rate. Thecell death rate in the presence of rhTFF3 was nearly 3times that of control cells, �5 times that of cellsincubated with TNF� alone (results not shown) or IL-1�alone, and 5.3 times that of cells incubated with IL-1�and TNF� (Figure 5B).

DISCUSSION

Numerous studies have documented that TFFpeptides have many positive in vivo and in vitro effectson epithelial restitution, wound healing, cell motility,and apoptosis, among others (21). In avascular tissuesuch as the cornea, similar effects have been observed.Goke et al (22) demonstrated that rhTFF3 enhances therestitution of primary rabbit corneal epithelial cells invitro. In previous studies, our group showed that TFF3 isinduced in the cornea under certain pathologic condi-tions (8) and has a pivotal role in the mechanism ofcorneal wound healing, suggesting broad implicationsfor development of novel strategies for treating nonheal-ing wounds (9). In the first report of such treatmentstrategies, data from a phase II multicenter, randomized,placebo-controlled trial of the prophylactic effects ofrhTFF3 for the treatment of chemotherapy-induced oralmucositis indicated that rhTFF3 in oral spray formula-tion is a safe and well-tolerated drug when given con-currently with chemotherapy. Prophylactic use ofrhTFF3 at either high or low dose was associated with asignificant reduction (�80%) in the occurrence of oralmucositis in patients at high risk for development of thelesion (23). Because articular cartilage, like cornea, is an

Figure 5. Proapoptotic effect of rhTFF3 on cultured human articularchondrocytes. A, Relative caspase units (RLU) were determined bymeasuring the activity of caspases 3 and 7 using the fluorometricsubstrate Z-DEVD–rhodamine 110. Assays on samples from each of 5different patients were performed separately, in triplicate. Values arethe mean and SD. � � P � 0.05 versus controls, by Student’s t-test. B,Rates of chondrocyte apoptosis were quantified by flow cytometry.Chondrocytes were incubated with 10 ng/ml IL-1� for 24 hours, with 10ng/ml IL-1� for 4 hours followed by addition of 0.3 mg/ml rhTFF3 andincubation for an additional 20 hours, with 10 ng/ml IL-1� and 10ng/ml TNF� for 24 hours, or with 10 ng/ml IL-1� and 10 ng/ml TNF�for 4 hours followed by addition of 0.3 mg/ml rhTFF3 and incubationfor an additional 20 hours. There were a total of 10,000 events. Cellswere labeled with annexin V (indicator of apoptotic cells) before flowcytometry. X-axes represent annexin V labeling (R2); percentages ofpositive cells are shown in the lower right corners. See Figure 3 forother definitions.

TFF3 IN DISEASED ARTICULAR CARTILAGE 821

avascular tissue, these results prompted us to undertakea detailed functional analysis of TFF3 in articular carti-lage.

After evaluating the expression pattern of TFFpeptides in healthy human OA cartilage, we assessed theexpression pattern of Tff3 in 2 mouse models. Ourresults demonstrate that healthy articular cartilage doesnot express TFF peptides, but under conditions of OAor sepsis, TFF3 is induced at the mRNA and proteinlevels in this mesenchymal-derived tissue. These findingsare largely in accordance with results obtained in humancornea as well as in wild-type mice after corneal wound-ing (8,9).

The proinflammatory cytokines IL-1� and TNF�are thought to be central agitators of the catabolic eventsseen in OA. Both cytokines are significantly up-regulated both in human OA (24) and in STR/Ort mice(25). In septic arthritis, a severe disease involving rapiddestruction of the affected joint that is associated with ahigh rate of mortality, S aureus has been shown to be theresponsible pathogen in �50% of all cases (26).

To analyze the effects of mediators of OA andseptic arthritis with regard to TFF3 expression in artic-ular cartilage, we cultured primary articular chondro-cytes from OA patients, as well as C28/I2 chondrocytes(in order to have 2 different models and to study theusefulness of C28/I2 chondrocytes) and analyzed theexpression patterns of TFF3 under basal conditions andafter challenge with IL-1�, TNF�, bacterial products,and bacterial supernatants. The expression of TFF3 atboth the mRNA and protein levels in cultured chondro-cytes may be explainable by the fact that the culturedprimary chondrocytes were derived from patients withOA and that the C28/I2 cell line is transfected with thelarge T antigen. After stimulation with proinflammatorycytokines, bacterial components, or bacterial superna-tants, TFF3 expression was further induced. Interest-ingly, such induction of TFF peptides by a bacterium hasbeen demonstrated recently with Helicobacter pylori, abacterium that acts mainly in the human stomach (27).In this context, very recent data from models of murineand human goblet cells reveal that Toll-like receptor 2activation selectively induces synthesis of TFF3 (28), andit may be interesting to investigate this mechanism inarticular cartilage in the future.

MMPs play a key role in the destruction ofarticular cartilage during the development of OA (29).Expression of MMPs in chondrocytes is activated byIL-1� and TNF� (30) and inactivated by their endoge-nous inhibitors (TIMPs). To elucidate the possible in-volvement of TFF3 in cartilage destruction or repair, we

assessed the effects of recombinant proteins in vitro.Thus, we analyzed the influence of rhTFF3 on the majorECM-degrading MMPs and TIMPs. The ECM of artic-ular cartilage consists mainly of type II collagen, aggre-can and other proteoglycans, minor collagens (types V,VI, IX, X, and XI), and other noncollagenous matrixproteins. In primary chondrocytes, MMP-1, MMP-3,and MMP-13 were up-regulated in a concentration-dependent manner after stimulation with rhTFF3.MMP-1 and MMP-13 (collagenases 1 and 3, respec-tively) are able to cleave the triple-helical domain ofcollagens, including types II and X collagen (31), andthey therefore play a decisive role in cartilage degrada-tion. Effects of rhTFF3 on both MMP-1 and MMP-13were weak, and were obvious only with high concentra-tions of rhTFF3 (1–5 mg/ml).

However, strong effects on the induction ofMMP-3 were observed, with significant up-regulationoccurring upon treatment with rhTFF3 even at a lowconcentration (0.1 mg/ml). MMP-3 (stromelysin) cleavesthe telopeptide regions or noncollagen domains of typesIX and XI collagen. As a consequence, incubation ofcartilage explants with MMP-3 results in breakdown ofthe collagen network (32). The interaction betweenMMP-3 and TFF3 is of great interest, since induction ofMMP-3 is a very early feature of OA (32) and our resultsindicate that induction of TFF3 also occurs particularlyin the early stages of OA. Hypothetically, this findingcould also suggest an anabolic function of TFF3, sincestudies have revealed actions of TNF� and IL-1� be-yond their catabolic functions. Apart from their destruc-tive abilities, they are able to stimulate growth factorssuch as bone morphogenetic protein 2 (BMP-2), osteo-genic protein 1 (BMP-7), and transforming growth fac-tor �, resulting in increased synthesis of aggrecan andtype II collagen in articular cartilage (33–36). Regard-less of their ultimately deleterious effects on articularcartilage, TNF� and IL-1� could initiate the repairresponse displayed by injured cartilage in early stages ofOA through their ability to enhance anabolic pathwaysin chondrocytes. Because both also induce TFF3, it isreasonable to speculate that TFF3 is involved in therepair process.

TIMPs 1 and 2 were not affected by rhTFF3, withthe exception that rhTFF3 at very high concentrations (5mg/ml) led to a significant down-regulation of TIMP-1.Stimulation of C28/I2 chondrocytes with rhTFF3 had noeffect with regard to MMP or TIMP induction, nor didstimulation of C28/I2 chondrocytes with IL-1�, TNF�,or a combination of both, indicating that this cell line is

822 ROSLER ET AL

not useful for the investigation of TFF3 function inarticular cartilage.

Septic arthritis has been found to be associatedwith induction of MMPs 2 and 9 (37), MMP-7 (38),MMPs 10 and 11 (39), and MMPs 1, 3, and 13 (40).Because our results clearly indicate an important role ofTFF3 in septic joint arthritis, the effects of TFF3 onMMPs in septic arthritis will be of interest and requirefurther elucidation.

There is a direct positive correlation between theseverity of OA and the extent of chondrocyte apoptosis(41). To analyze possible effects of TFF3 on chondrocyteapoptosis, primary cultured human chondrocytes werestimulated with IL-1� or TNF�. Both cytokines arepresent at increased levels during OA and have beenshown to exert proapoptotic actions (42,43). Our resultsdemonstrate that rhTFF3 increased annexin V andcaspase 3/7 activity, indicating a proapoptotic effect onarticular cartilage. Evidence in support of the notion ofa proapoptotic function of TFF3 was reported morethan 10 years ago, from a study in which rTFF3 wasdemonstrated to induce DNA fragmentation and mor-phologic changes characteristic of apoptosis in a colo-rectal carcinoma cell line (44). In addition, as-yet-unpublished data indicate expression and proapoptoticeffects of TFF2 in murine retina (Dunker N: personalcommunication). The proapoptotic characteristics ofTFFs require further elucidation.

In contrast, antiapoptotic and protective effectsof TFF3 at mucosal sides of the GI tract are well known(1,45). TFF3 has been shown to be essential for woundhealing of mucous membrane of the colon (1). Aftermucosal injury in the GI tract, immediately up-regulatedTFF peptides enhance wound healing by promoting cellmigration (3,46). Perhaps TFF3 has a similar influenceon cell migration in articular cartilage. During thecourse of OA, chondrocytes form clusters due to earlyproliferative activity and are eventually reduced in num-ber. TFF3 could enhance this step, probably by destroy-ing the collagen network through MMP-3; this would bein accordance with findings in the cornea (9), which doesnot express TFF3 under healthy conditions but inducesTFF3 after epithelial injury. As in the cornea, TFF3promotes wound healing in the respiratory tract byenhancing re-epithelialization (47).

Many of the multiple biologic functions of TFFsare thought to be triggered by receptor activation.However, to date, most attempts to isolate TFF bindingproteins with characteristics of typical receptors havefailed. Of great interest, during preparation of thepresent manuscript, a report was published in which the

chemokine receptor CXCR4 was described as a low-affinity receptor for TFF2 (48); the authors state that itmight also be a receptor for TFF3. Expression ofCXCR4 has been demonstrated on chondrocytes (49–51) and is induced during OA (49,52). Its activationresults in activation of MMP-3 (49) as well as MMP-13(51) and leads to chondrocyte apoptosis (50). Given thehypothesis that TFF3, in addition to TFF2, is also aligand for CXCR4, this would strongly support ourpresent results that TFF3 induces MMP-3 and MMP-13and has proapoptotic effects. This requires further elu-cidation and will be of great future interest.

In conclusion, our results demonstrate that TFF3is not expressed in healthy articular cartilage but isinduced in chondrocytes during OA and under septicconditions. Further, our findings indicate—completelycontrary to our expectations—that the well-known pro-tective functions of TFF3 do not occur in OA cartilage.In contrast, in OA cartilage, TFF3 has a previouslynearly unrecognized proapoptotic function and en-hances degradation of ECM by inducing MMPs, espe-cially MMP-3. However, this step could also be initiatedto protect cartilage. Finally, the results indicate thatTFF3 is also involved in the pathogenesis of septicarthritis. More work is needed to further characterizethe functions of TFF3 in OA, septic arthritis, andprobably other cartilage diseases as well.

ACKNOWLEDGMENTS

We are grateful to Ute Beyer, Susann Moschter,Michaela Risch, and Karin Stengel for technical assistance andvaluable discussions. We thank Drs. Nicholas P. Barker (GICompany, Framingham, MA) and Daniel K. Podolsky (Mas-sachusetts General Hospital, Boston, MA) for providing anti-Tff3 and Dr. Michael Tsokos (Institute of Legal Medicine,Charite, Germany) for providing cartilage tissue.

AUTHOR CONTRIBUTIONSAll authors were involved in drafting the article or revising it

critically for important intellectual content, and all authors approvedthe final version to be published. Dr. Paulsen had full access to all ofthe data in the study and takes responsibility for the integrity of thedata and the accuracy of the data analysis.Study conception and design. Rosler, Claassen, Goldring, Sel, Varoga,Paulsen.Acquisition of data. Rosler, Haase, Claassen, Schulze, Schicht, Rie-mann, Brandt, Wohlrab, Muller-Hilke, Sel, Varoga, Garreis, Paulsen.Analysis and interpretation of data. Rosler, Haase, Claassen, Schulze,Schicht, Riemann, Brandt, Wohlrab, Sel, Varoga, Garreis, Paulsen.

REFERENCES

1. Taupin DR, Kinoshita K, Podolsky DK. Intestinal trefoil factorconfers colonic epithelial resistance to apoptosis. Proc Natl AcadSci U S A 2000;97:799–804.

TFF3 IN DISEASED ARTICULAR CARTILAGE 823

2. Thim L, Madsen F, Poulsen SS. Effect of trefoil factors on theviscoelastic properties of mucus gels. Eur J Clin Invest 2002;32:519–27.

3. Dignass A, Lynch-Devaney K, Kindon H, Thim L, Podolsky DK.Trefoil peptides promote epithelial migration through a trans-forming growth factor �-independent pathway. J Clin Invest1994;94:376–83.

4. Emami S, Le Floch N, Bruyneel E, Thim L, May F, Westley B, etal. Induction of scattering and cellular invasion by trefoil peptidesin src- and RhoA-transformed kidney and colonic epithelial cells.FASEB J 2001;15:351–61.

5. Cook GA, Familari M, Thim L, Giraud AS. The trefoil peptidesTFF2 and TFF3 are expressed in rat lymphoid tissues and partic-ipate in the immune response. FEBS Lett 1999;456:155–9

6. Chwieralski CE, Schnurra I, Thim L, Hoffmann W. Epidermalgrowth factor and trefoil factor family 2 synergistically triggerchemotaxis on BEAS-2B cells via different signaling cascades.Am J Respir Cell Mol Biol 2004;31:528–37.

7. Kjellev S. The trefoil factor family—small peptides with multiplefunctionalities. Cell Mol Life Sci 2009;66:1350–69.

8. Steven P, Schafer G, Nolle B, Hinz M, Hoffmann W, Paulsen F.Distribution of TFF-peptides in corneal disease and pterygium.Peptides 2004;25:819–25.

9. Paulsen FP, Woon CW, Varoga D, Jansen A, Garreis F, Jager K,et al. Intestinal trefoil factor/TFF3 promotes re-epithelialization ofcorneal wounds. J Biol Chem 2008;283:13418–27.

10. Mankin HJ, Dorfmaan H, Lipiello L, Zarins A. Biochemical andmetabolic abnormalities in articular cartilage from osteoarthritichips: correlation of morphology with biochemical and metabolicdata. J Bone Joint Surg Am 1971;53:523–37.

11. Goldring MB, Birkhead JR, Suen LF, Yamin R, Mizuno S,Glowacki J, et al. Interleukin-1 �-modulated gene expression inimmortalized human chondrocytes. J Clin Invest 1994;94:2307–16.

12. Grall F, Gu X, Tan L, Cho JY, Inan MS, Pettit AR, et al.Responses to the proinflammatory cytokines interleukin-1 andtumor necrosis factor � in cells derived from rheumatoid synoviumand other joint tissues involve nuclear factor �B–mediated induc-tion of the Ets transcription factor ESE-1. Arthritis Rheum2003;48:1249–60.

13. Jager K, Bonisch U, Risch M, Worlitzsch D, Paulsen F. Detectionand regulation of cationic amino acid transporters in healthy anddiseased ocular surface. Invest Ophthalmol Vis Sci 2009;50:1112–21.

14. Paulsen F, Varoga D, Paulsen A, Tsokos M. Trefoil factor family(TFF) peptides of normal human Vater’s ampulla. Cell Tissue Res2005;321:67–74.

15. Mason RM, Chambers MG, Flannelly J, Gaffen JD, Dudhia J,Bayliss JT. The STR/ort mouse and its use as a model ofosteoarthritis. Osteoarthritis Cartilage 2001;9:85–91.

16. Deng GM, Nilsson IM, Verdrengh M, Tarkowski A. Intraarticu-larly localized bacterial DNA containing CpG motifs inducesarthritis. Nat Med 1999;5:702–5.

17. Varoga D, Pufe T, Harder J, Schroder JM, Mentlein R, Meyer-Hoffert U, et al. Human �-defensin 3 mediates tissue remodelingprocesses in articular cartilage by increasing levels of metallo-proteinases and reducing levels of their endogenous inhibitors.Arthritis Rheum 2005;52:1736–45.

18. Kurz B, Jost B, Schunke M. Dietary vitamins and seleniumdiminish the develoment of mechanically induced osteoarthritisand increase the expression of antioxidative enzymes in the kneejoint of STR/1N mice. Osteoarthritis Cartilage 2002;10:119–26.

19. Dignass A, Lynch-Devaney K, Kindon H, Thim L, PodolskyDK.Trefoil peptides promote epithelial migration though a trans-forming growth factor �-independent pathway. J Clin Invest1994;94:376–83.

20. Babyatsky MW, deBeaumont M, Thim L, Podolsky DK. Oral

trefoil peptides protect against ethanol- and indomethacin-in-duced gastric injury in rats. Gastroenterology 1996;110:489–97.

21. Otto WR, Thim L. Trefoil factor family-interacting proteins. CellMol Life Sci 2005;62:2939–46.

22. Goke MN, Cook JR, Kunert KS, Fini ME, Gipson IK, PodolskyDK. Trefoil peptides promote restitution of wounded cornealepithelial cells. Exp Cell Res 2001;264:337–44.

23. Barker NP, Peterson DE, Akhmadullina LI, Rodionova I,Sherman NZ, Gertner JM, et al. Prophylaxis of recurrentchemotherapy-induced oral mucositis: a phase II multicenter,randomized, placebo-controlled trial of recombinant human intes-tinal trefoil factor (rhITF) [abstract]. Am Soc Clin Oncol 2008;26Suppl:9514.

24. Berenbaum F. Signal transduction: target in osteoarthritis. CurrOpin Rheumatol 2004;16:616–22.

25. Chambers MG, Bayliss MT, Mason RM. Chondrocyte cytokineand growth factor expression in murine osteoarthritis. Osteoarthri-tis Cartilage 1997;5:301–8.

26. Ryan MJ, Kavanagh R, Wall PG, Hazleman BL. Bacterial jointinfections in England and Wales: analysis of bacterial isolates overa four year period. Br J Rheumatol 1997;36:370–3.

27. Matsuda K, Yamauchi K, Matsumoto T, Sano K, Yamaoka Y, OtaH. Quantitative analysis of the effect of Helicobacter pylori on theexpressions of SOX2, CDX2, MUC2, MUC5AC, MUC6, TFF1,TFF2, and TFF3 mRNAs in human gastric carcinoma cells. ScandJ Gastroenterol 2008;43:25–33.

28. Podolsky DK, Gerken G, Eyking A, Cario E. Colitis-associatedvariant of TLR2 causes impaired mucosal repair because of TFF3deficiency. Gastroenterology 2009;137:209–20.

29. Monfort J, Garcia-Giralt N, Lopez-Armada M, Monllau JC,Bonilla A, Benito P, et al. Decreased metalloproteinase produc-tion as a response to mechanical pressure in human cartilage: amechanism for homeostatic regulation. Arthritis Res Ther 2006;8:R149.

30. Fernandes JC, Martel-Pelletier J, Pelletier JP. The role of cyto-kines in osteoarthritis pathophysiology. Biorheology 2002;39:237–46.

31. Mitchell PJ, Magna HA, Reeves LM, Lopresti-Morrow LL, Yo-cum SA, Rosner PJ. Cloning, expression, and type II collagenolyticactivity of matrix metalloproteinase-13 from human osteoarthriticcartilage. J Clin Invest 1996;97:761–8.

32. Thibault M, Poole AR, Buschmann MD. Cyclic compression ofcartilage/bone explants in vitro leads to physical weakening, me-chanical breakdown of collagen and release of matrix fragments.J Orthop Res 2002;20:1265–73.

33. Merrihew C, Soeder S, Rueger DC, Kuettner KE, ChubinskayaSE. Modulation of endogenous osteogenic protein-1 (OP-1) byinterleukin-1 in adult human articular cartilage. J Bone Joint SurgAm 2003;85-A Suppl 3:67–74.

34. Kaiser M, Haag J, Soder S, Bau B, Aigner T. Bone morphogeneticprotein and transforming growth factor � inhibitory Smads 6 and7 are expressed in human adult normal and osteoarthritic cartilagein vivo and are differentially regulated in vitro by interleukin-1�.Arthritis Rheum 2004;50:3535–40.

35. Fukui N, Ikeda Y, Ohnuki T, Hikita A, Tanaka S, Yamane S, et al.Pro-inflammatory cytokine tumor necrosis factor-� induces bonemorphogenetic protein-2 in chondrocytes via mRNA stabilizationand transcriptional up-regulation. J Biol Chem 2006;281:27229–41.

36. Hayashi M, Muneta T, Ju YJ, Mochizuki T, Sekiya I. Weeklyintra-articular injections of bone morphogenetic protein-7 inhibitosteoarthritis progression. Arthitis Res Ther 2008;10:R118.

37. Chu SC, Yang SF, Lue KH, Hsieh YS, Lin ZI, Lu KH. Clinicalsignificance of gelatinases in septic arthritis of native and replacedknees. Clin Orthop Relat Res 2004;427:179–83.

38. Gjertsson I, Innocenti M, Matrisian LM, Tarkowski A. Metallo-

824 ROSLER ET AL

proteinase-7 contributes to joint destruction in Staphylococcusaureus induced arthritis. Microb Pathog 2005;38:97–105.

39. Kanangat S, Postlethwaite A, Hasty K, Kang A, Smeltzer M,Appling W, et al. Induction of multiple matrix metalloproteinasesin human dermal and synovial fibroblasts by Staphylococcusaureus: implications in the pathogenesis of septic arthritis andother soft tissue infections. Arthritis Res Ther 2006;8:R176.

40. Kyburz D, Rethage J, Seibl R, Lauener R, Gay RE, Carson DA, etal. Bacterial peptidoglycans but not CpG oligodeoxynucleotidesactivate synovial fibroblasts by Toll-like receptor signaling. Arthri-tis Rheum 2003;48:642–50.

41. Thomas CM, Fuller CJ, Whittles CE, Sharif M. Chondrocytedeath by apoptosis is associated with cartilage matrix degradation.Osteoarthritis Cartilage 2007;15:27–34.

42. Carames B, Lopez-Armada MJ, Cillero-Pastor B, Lires-Dean M,Vaemonde C, Galdo F, et al. Differential effects of tumor necrosisfactor-� and interleukin-1� on cell death in human articularchondrocytes. Osteoarthritis Cartilage 2008;16:715–22.

43. Roman-Blas JA, Stokes DG, Jimenez SA. Modulation of TFG-�signaling by proinflammatory cytokines in articular chondrocytes.Osteoarthritis Cartilage 2007;15:1367–77.

44. Efstathiou JA, Noda M, Rowan A, Dixon C, Chinery R, JawhariA, et al. Intestinal trefoil factor controls the expression of theadenomatous polyposis coli-catenin and the E-cadherin-catenincomplexes in human colon carcinoma cells. Proc Natl Acad SciU S A 1998;95:3122–7.

45. Kinoshita T, Taupin DR, Itoh H, Podolsky DK. Distinct pathwaysof cell migration and antiapoptotic response to epithelial injury:

structure-function analysis of human intestinal trefoil factor. MolCell Biol 2000;20;4680–90.

46. Sands BE, Podolsky DK. The trefoil peptide family. Annu RevPhysiol 1996;58:253–73.

47. Oertel M, Graness A, Thim L, Buhling F, Kalbacher H, HoffmannW. Trefoil factor family-peptides promote migration of humanbronchial epithelial cells: synergistic effect with epidermal growthfactor. Am J Respir Cell Mol Biol 2001;25:418–24.

48. Dubeykovskaya Z, Dubeykovskaya A, Solal-Cohen J, Wang TC.Secreted trefoil factor 2 activates the CXCR4 receptor in epithe-lial and lymphocytic cancer cell lines. J Biol Chem 2009;284:3650–62.

49. Kanbe K, Takagishi K, Chen Q. Stimulation of matrix metallopro-tease 3 release from human chondrocytes by the interaction ofstromal cell–derived factor 1 and CXC chemokine receptor 4.Arthritis Rheum 2002;46:130–7.

50. Wei L, Sun X, Kanbe K, Wang Z, Sun C, Terek R, et al.Chondrocyte death induced by pathological concentration ofchemokine stromal cell-derived factor-1. J Rheumatol 2006;33:1818–26.

51. Chiu YC, Yang RS, Hsieh KH, Fong YC, Way TD, Lee TS, et al.Stromal cell-derived factor-1 induces matrix metalloprotease-13expression in human chondrocytes. Mol Pharmacol 2007;72:695–703.

52. Appleton CT, Pitelka V, Henry J, Beier F. Global analyses of geneexpression in early experimental osteoarthritis. Arthritis Rheum2007;56:1854–68.

TFF3 IN DISEASED ARTICULAR CARTILAGE 825