Differentially regulated expression of growth differentiation factor 5 and bone morphogenetic protein 7 in articular cartilage and synovium in murine chronic arthritis: Potential importance for cartilage breakdown and synovial hypertrophy




To examine whether the endogenous expression of growth differentiation factor 5 (GDF-5) and bone morphogenetic protein 7 (BMP-7) is altered in the cartilage and synovium of human tumor necrosis factor α (TNFα)–transgenic (hTNFtg) mice with chronic arthritis, and to investigate the response of hTNFtg chondrocytes as well as fibroblast-like synoviocytes (FLS) to these morphogens in vitro.


Analyses were performed in hTNFtg mice with chronic destructive arthritis and in wild-type (WT) mice as controls. Expression of GDF-5 and BMP-7 in the articular cartilage and synovium was examined by real-time polymerase chain reaction and immunohistochemistry. Human TNFtg cartilage explants, chondrocytes, and FLS monolayer cultures were assessed for basal matrix biosynthesis as well as growth factor responsiveness, using 35S-sulfate incorporation assays. In addition, the DNA content/cell proliferation rate was measured.


The expression of GDF-5 and BMP-7 was decreased in articular cartilage from hTNFtg mice, whereas expression of both morphogens was increased in arthritic synovium from hTNFtg mice, as compared with the levels in WT controls. Isotope incorporation revealed a marked reduction of matrix synthesis in hTNFtg cartilage as well as a decrease in responsiveness to GDF-5 and BMP-7. The DNA content did not change in arthritic cartilage as compared with WT cartilage. In hTNFtg FLS, growth factor stimulation increased the rate of cell proliferation and the production of extracellular matrix.


In this murine model of TNFα-mediated arthritis, the expression of GDF-5 and BMP-7 is regulated differentially in articular cartilage and synovium. In articular cartilage, the down-regulation of GDF-5 and BMP-7, which function to maintain matrix integrity, could potentially compromise tissue repair, whereas in synovium, the increased expression of GDF-5 and BMP-7 might contribute to synovial hypertrophy.

A hallmark of rheumatoid arthritis (RA) is persistent inflammation of the synovial membrane. Subsequent pannus formation is considered a key trigger of bone and cartilage destruction. Despite recent advances in understanding the pathways of bone erosion in arthritis (1–3), the mechanisms leading to cartilage destruction have not been fully elucidated. A common hypothesis suggests that cartilage destruction is the result of a disequilibrium between matrix degradation and matrix formation, with predominance of the degradative process.

Indeed, catabolic factors leading to matrix degradation are abundantly expressed in the synovial tissue of patients with RA. Macrophage- and fibroblast-like synoviocytes (FLS) synthesize large amounts of metalloproteinases, cathepsins, and proinflammatory cytokines. In addition to the direct effects of various proteases on articular cartilage, cytokines induce a change in the functional expression pattern of articular chondrocytes, diverting the pattern away from production of matrix molecules to release of proteases and nitric oxide (4–9).

Although an imbalance between matrix degradation and matrix formation is hypothesized as a cause of cartilage destruction in arthritis, it is not known how the mechanisms that counteract matrix loss are regulated in conditions of chronic inflammatory joint disease. These anabolic factors apparently fail to compensate for matrix degradation in arthritis. This could be attributable to either failure of up-regulation or failure of down-regulation of these factors.

Members of the bone morphogenetic protein (BMP) family are likely candidate factors that play a role in cartilage remodeling in arthritic joints. They belong to the transforming growth factor β (TGFβ) superfamily of growth factors, a large group of structurally related polypeptide cytokines known to elicit various functions in the development, homeostasis, and repair of different tissues (10). These molecules were originally characterized by their unique ability to induce cartilage and bone formation when applied at ectopic sites (11, 12). However, they are now recognized as important players in developmental patterning (13) and in late morphogenesis in different organs, including the skeleton and joints (14). Furthermore, the role of distinct members of the BMP family in maintaining cartilage homeostasis and repair in osteoarthritis (OA) is well established (15–17). In contrast, little is known about the functions of BMPs in chronic inflammatory joint diseases. Since it has been reported that BMPs have the ability to overcome the catabolic effects of proinflammatory cytokines (18–21), any alteration in the synthesis and expression of BMPs may predispose the cartilage to destruction.

To elucidate these issues, we investigated whether the endogenous expression of BMPs is altered in the cartilage as well as synovial membrane of chronically inflamed joints. Based on our group's previous studies in articular cartilage (22, 23), we analyzed the expression of growth differentiation factor 5 (GDF-5; also known as cartilage-derived morphogenetic protein 1 or BMP-14) and BMP-7 (also known as osteogenic protein 1) in human tumor necrosis factor α (TNFα)–transgenic (hTNFtg) mice, an established model of chronic destructive arthritis. In addition, we investigated the response of hTNFtg articular chondrocytes and FLS to these morphogens, in terms of their effects on extracellular matrix synthesis.


TNFα-transgenic animals.

The heterozygous hTNFtg mice (strain Tg197) have been described elsewhere (24). Briefly, mice were made transgenic for the hTNF gene construct with an unmodified 5′ promoter region, but without modifying the deregulated hTNF gene expression in vivo. Preparation of the hTNF gene constructs has been described in detail previously (24). These transgenic mice develop a chronic inflammatory and destructive polyarthritis within 6 weeks after birth. Most importantly, severe cartilage damage with rapid loss of normal proteoglycan content occurs in the mice (3). All mice, both the hTNFtg mice and their wild-type (WT) littermates, were inbred on a C57BL/6 genetic background. Mice were fed a normal diet with water ad libitum, and were killed by cervical dislocation at 12 weeks after birth. By age 12 weeks, all hTNFtg mice had developed clinical signs of arthritis that had persisted for 7–8 weeks and was characterized by progressive joint swelling and loss of grip strength, as described previously (3). The local ethics committee approved all animal procedures.


Hind paws were fixed in 4.0% formalin overnight and then decalcified in 14% EDTA (Sigma, St. Louis, MO) at 4°C (pH adjusted to 7.2 by the addition of ammonium hydroxide [Sigma]) until the bones were pliable. For immunohistochemistry, deparaffined, ethanol-dehydrated tissue sections (2 μm) were boiled for 2 minutes in 10 mM sodium citrate buffer (pH 6.0) using a 700W microwave oven, and then allowed to cool to room temperature and rinsed in detergent solution (0.5% Tween in phosphate buffered saline [PBS]) for 10 minutes.

Tissue sections were blocked for 20 minutes in PBS containing 20% rabbit serum, followed by incubation for 1 hour at room temperature with goat anti-mouse GDF-5 and BMP-7 polyclonal antibodies (diluted 1:100; both from Santa Cruz Biotechnology, Santa Cruz, CA). After rinsing, endogenous peroxidase was blocked with 0.3% hydrogen peroxide in Tris buffered saline (10 mM Tris HCl, 140 mM NaCl, pH 7.4) for 10 minutes. This was followed by 30 minutes' incubation with a biotinylated species-specific anti-IgG secondary antibody (Vector, Burlingame, CA). Sections were then incubated with the appropriate Vectastain ABC reagent (Vector) for another 30 minutes, using 3,3′-diaminobenzidine (Sigma) for the color reaction, with results visualized as brown staining of antigen-expressing cells. Sections in which the primary antibody was omitted served as negative controls.

Cell culture.

For explant cultures, knee joints of WT and hTNFtg mice were opened aseptically and articular cartilage was carefully obtained. Explants were either prepared for RNA extraction or subjected to 35S-sulfate incorporation, hydroxyproline, and DNA assays.

For monolayer cultures, chondrocytes were released from the knee cartilage of WT and hTNFtg mice by 5-hour digestion in 0.2% collagenase B (Boehringer Mannheim, Mannheim, Germany) and filtered through a cell strainer (Falcon; Becton Dickinson Labware, Lincoln Park, NJ) to remove debris. The cell filtrate was then centrifuged at 500g for 10 minutes. Pellets were resuspended in a 1:1 mixture of Dulbecco's modified Eagle's medium (DMEM) (25 mM HEPES plus 4,500 mg/liter glucose plus pyridoxine, without sodium pyruvate; Life Technologies, Gaithersburg, MD) and Ham's F-12 medium (Ham's F-12 plus L-glutamine; Life Technologies) containing 10% fetal bovine serum (PAA Laboratories, Linz, Austria) and antibiotics/antimycotics (100 units/ml penicillin G, 100 mg/ml streptomycin, and 0.25 μg/ml amphotericin B; Life Technologies). The isolated cells were grown as monolayer cultures in 48-well plates (Costar, Cambridge, MA) at a density of 5 × 104 cells/well until subconfluence was reached in the serum-containing medium. Thereafter, the medium was switched to a chemically defined, serum-free basal medium (25) and the cultures were incubated with either recombinant GDF-5 or BMP-7 (both from R&D Systems, Minneapolis, MN) at different concentrations (100, 250, and 500 ng/ml) over a period of 5 days.

Murine FLS were obtained by dispase II digestion of the tarsus derived from 3 hTNFtg mice. The tarsus was isolated from the hind paws, minced into small pieces, and digested in 0.15% dispase II (Roche Diagnostics, Mannheim, Germany) for 2 hours at 37°C. The supernatant was filtered through a 70-μm cell strainer (Falcon; Becton Dickinson Labware) and the cell filtrate was then centrifuged at 500g for 10 minutes. The supernatant was removed and the pellet was washed with PBS (Life Technologies) before being centrifuged again. After resuspension of the pellet in DMEM, the cells were distributed onto a T25 tissue culture flask (Costar). FLS were used after passage 3 for cell culture experiments.

For 35S-sulfate incorporation and DNA assays, FLS were plated as monolayer cultures, in quadruplicate, at a density of 1 × 105 cells/well in 24-well plates (Costar). Stimulation with either GDF-5 or BMP-7 was performed at a final concentration of 100 ng/ml over a period of 5 days. For the 3H-thymidine incorporation assay, FLS were cultured, in quadruplicate, in 96-well plates (Packard BioScience, Meriden, CT) at a density of 5,000 cells per well. For total RNA preparation, quadruplicate cultures of 1 × 106 cells were seeded in 100-mm tissue culture dishes (Costar) and cultured with or without the addition of the different growth factors (100 ng/ml) for 5 days. Basal medium was added as needed, and culture medium was replaced every other day. Cell cultures maintained in basal medium alone served as negative controls. The plates were maintained at 37°C in a humidified atmosphere of 5% CO2.

Assays for biosynthesis of macromolecules.

For 35S-sulfate incorporation assays, cartilage explants were distributed in 24-well plates and washed 3 times with PBS. The explants were then labeled in identical aliquots of 1 ml of a sulfate-free, serum-free basal medium containing 20 μCi/ml of 35S-sulfate (carrier-free; Amersham, Buckinghamshire, UK) for 5 hours at 37°C. After radiolabeling, the medium was discarded and explants were washed 3 times with ice-cold buffer (10 nM EDTA, 0.1M sodium phosphate, pH 6.5), followed by an overnight digestion in 1 ml sodium phosphate wash buffer containing proteinase K (1 mg/ml) at 80°C.

Labeling of the monolayer cultures, either chondrocytes or FLS, with radioactivity was done by incubation with 20 μCi/ml of 35S-sulfate (Amersham) for 5 hours at 37°C in basal medium. Cells were extracted in guanidine-HCl buffer (4M guanidine-HCl, 50 mM sodium acetate buffered at pH 7.2) in the presence of protease inhibitors. Unincorporated isotope was removed by using Sephadex G-25 (PD-10 columns; Pharmacia Biotech, Piscataway, NJ) gel chromatography. Results were obtained by liquid scintillation counting (1410 liquid scintillation counter; Wallac Oy, Turku, Finland) of aliquots from void-volume fractions, with results normalized to the hydroxyproline or DNA content. The hydroxyproline content for each proteinase K–digested sample was determined using a previously described method (26).

Determination of DNA content/cell proliferation.

The DNA content of either cartilage explants or monolayer cultures was assessed by staining with bisbenzimide (Hoechst 33258; Sigma) (27). FLS were preincubated with serum-free medium for 24 hours. GDF-5 or BMP-7 was added for an additional 24 hours. Cells were pulsed with 1 μCi/ml of methyl-3H-thymidine (Amersham) for the last 6 hours of culture. The cells were washed extensively with PBS, and thymidine incorporation was measured directly in the culture plate, after the addition of scintillant in a liquid scintillation counter (Packard BioScience).

RNA isolation and quantitative real-time polymerase chain reaction (PCR).

Total RNA from ex vivo cartilage tissue was obtained using the RNAgent total RNA isolation system (Promega BioSciences, San Luis Obispo, CA), while chondrocytes and FLS in monolayers were extracted using a different commercially available kit (RNeasy kit; Qiagen, Valencia, CA). Real-time PCR conditions were optimized for maximal PCR efficiency by concentration adjustment of the PCR primers used: for mouse type I collagen (Col1a1), forward TGACGCATGGCCAAGAAGA and reverse ATTGCACGTCATCGCACACA; for mouse type II collagen (Col2a1), forward AAGACCCAGACTGCCTCAA and reverse ATCACCTCTGGGTCCTTGTT; for mouse GDF-5, forward TGCTGACAGAAAGGGAGGTA and reverse TCCAAGGCACTGATGTCAA; for mouse BMP-7, forward TTCCTCACTGACGCCGACAT and reverse TCCCGGATGTAGTCCTTATA; and for mouse GAPDH, forward ACTGAGGACCAGGTTGTC and reverse TGCTGTAGCCGTATTCATTG. Four different combinations of forward/reverse primers were tested: 100 nM/100 nM, 100 nM/300 nM, 300 nM/100 nM, and 300 nM/300 nM. The optimum concentrations for each primer were as follows: for Col1a1, 100 nM/100 nM; for Col2a1, 300 nM/100 nM; for GDF-5, 100 nM/100 nM; for BMP-7, 100 nM/300 nM; and for GAPDH, 100 nM/100 nM.

PCR was performed using 25 μl of 1× PCR buffer containing 50 μM KCl, 3 μM MgCl2, 200 nM dNTPs, primers, 1× SYBR Green, 0.5 units Taq DNA polymerase (Qiagen), and 0.25 μl complementary DNA (cDNA). Real-time PCR was performed using the Icycler iQ detection system (BioRad, Hercules, CA). The PCR reaction was carried out at 95°C for 3 minutes, followed by 50 cycles of 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds, with a final extension at 72°C for 4 minutes. At the end of the PCR cycles, a melting curve, using a temperature range between 55°C and 95°C with intervals of 0.5°C, was generated to test the specificity of the PCR product. The cDNA in each experiment was tested in triplicate. GAPDH was used as the internal control.

Statistical analysis.

Results between groups were compared using Student's t-test. Statistically significant differences were defined as P values less than 0.05.


Expression of GDF-5 and BMP-7 in hTNFtg articular cartilage.

To address the possibility of a decrease in GDF-5 and BMP-7 gene expression in the course of TNFα-mediated arthritis, articular chondrocytes derived from WT and hTNFtg mice were evaluated for the expression of GDF-5 and BMP-7 messenger RNA (mRNA), using quantitative real-time PCR. A marked reduction was observed in the relative mRNA expression of both GDF-5 and BMP-7 in hTNFtg chondrocytes, by 75% and 65%, respectively, as compared with WT chondrocytes (Figures 1A and B), suggesting that TNFα directly or indirectly down-regulates the mRNA expression of GDF-5 and BMP-7.

Figure 1.

Endogenous expression of growth differentiation factor 5 (GDF-5) (A) and bone morphogenetic protein 7 (BMP-7) (B) in articular chondrocytes derived from human tumor necrosis factor α–transgenic (hTNFtg) and wild-type (WT) mice. Articular cartilage from the knee joints of 3 animals in each group was obtained directly after removal of the tissue, as described in Materials and Methods. The tissue was pooled and subjected to quantitative real-time polymerase chain reaction. The mean relative expression of GDF-5 (A) and BMP-7 (B) was determined after normalization to GAPDH.

Assessment of the expression of GDF-5 and BMP-7 by immunohistochemistry in the paw sections from WT and hTNFtg mice also revealed a decrease in the protein levels of each factor in the articular cartilage of hTNFtg mice, which confirmed, in a semiquantitative manner, the real-time PCR data (Figures 2A–D). In WT mice, a mean ± SEM 33 ± 3.1% of articular chondrocytes per total cell number per joint were GDF-5–positive, whereas chondrocyte expression of GDF-5 was decreased to 12 ± 1.8% per total cell number per joint in the articular cartilage of hTNFtg mice (P < 0.004). BMP-7–positive articular chondrocytes accounted for 50 ± 6.9% per total cell number per joint in the WT articular cartilage, whereas in hTNFtg mice, the articular cartilage contained 28 ± 4.9% BMP-7–positive cells (P < 0.02).

Figure 2.

Expression of GDF-5 and BMP-7 on the protein level in articular cartilage derived from hTNFtg and WT mice. Immunohistochemistry using specific goat anti-mouse polyclonal antibodies against GDF-5 (A and B) and BMP-7 (C and D) was performed on the hind paw sections of WT or hTNFtg animals (n = 3 per group). Results from 1 representative experiment are shown. Insets show higher-magnification views of the boxed areas in A–D (original magnification × 50; × 400 in the insets). See Figure 1 for definitions.

Neosynthesis of matrix proteoglycans and DNA content in hTNFtg cartilage.

The decreased expression of GDF-5 and BMP-7 in the articular cartilage of hTNFtg mice might hinder tissue regeneration, and consequently might play a role in cartilage destruction during chronic joint inflammation. To determine whether the state of chronic inflammation due to overexpression of TNFα affects chondrocyte biosynthetic activity, we assessed sulfated glycosaminoglycan (sGAG) synthesis in the knee cartilage explants derived from WT and hTNFtg mice. Using the 35S-sulfate incorporation rate as a surrogate marker for sGAG synthesis, we observed a significant decrease in isotope uptake in the cartilage from hTNFtg mice (mean ± SEM 592.2 ± 110.8 counts per minute/μg hydroxyproline) compared with that in WT controls (1,183.4 ± 173.8 cpm/μg hydroxyproline) (P < 0.02) (Figure 3A).

Figure 3.

Synthesis of matrix macromolecules and DNA content in cartilage explants obtained from hTNFtg mice. A, Cartilage explants were obtained from the knee joints of hTNFtg (n = 6) and WT (n = 7) mice, and specimens were labeled with 35S-sulfate right after removal from the joint, followed by proteinase K digestion. The incorporated radiolabel into newly synthesized glycosaminoglycans was measured; values were normalized to hydroxyproline content. ∗ = P < 0.02 versus hTNFtg mice. B, DNA content of explants derived from hTNFtg or WT cartilage was determined; values were normalized to hydroxyproline content. C, Dose-response experiments were conducted using different concentrations (100, 250, and 500 ng/ml) of either GDF-5 or BMP-7 in hTNFtg and WT chondrocyte monolayer cultures; basal medium (BM) alone was used as the control. ∗ = P < 0.05; ∗∗ = P < 0.001, versus the respective BM control. Bars show the mean and SEM. See Figure 1 for other definitions.

In addition, we measured the DNA content of the cartilage explant cultures. The results revealed no significant differences in the DNA content between WT mice (mean ± SEM 41.9 ± 7.4 μg DNA/μg hydroxyproline) and hTNFtg mice (39.9 ± 7.8 μg DNA/μg hydroxyproline) (P = 0.86) (Figure 3B), indicating that there was no significant chondrocyte loss in the hTNFtg cartilage.

Response of hTNFtg articular chondrocytes to GDF-5 and BMP-7 stimulation in vitro.

To evaluate whether hTNFtg chondrocytes respond to stimulation with GDF-5 and BMP-7 in vitro, we investigated the responsiveness of WT and hTNFtg articular chondrocytes to GDF-5 and BMP-7 in monolayer cultures. Interestingly, cells derived from hTNFtg cartilage required about twice the time needed to grow to subconfluence as that required by chondrocytes from WT cartilage. The subsequent assessment of newly synthesized sGAG revealed a very low basal synthesis rate in chondrocyte cultures from hTNFtg mice (mean ± SEM 7.3 ± 0.9 cpm/μg DNA) compared with that in WT controls (31.4 ± 2.1 cpm/μg DNA) (Figure 3C).

Furthermore, stimulation with GDF-5 and BMP-7 increased sGAG neosynthesis in hTNFtg chondrocytes only after very high concentrations of the growth factors and only to an extent that was, at best, comparable with the basal sGAG synthesis levels in WT controls (sGAG neosynthesis in hTNFtg cells stimulated with GDF-5 at 100 ng/ml, 11.5 ± 2.4 cpm/μg DNA [P = 0.1 versus basal medium controls]; at 250 ng/ml, 15.7 ± 1 cpm/μg DNA [P < 0.004 versus controls]; at 500 ng/ml, 14.5 ± 3.4 cpm/μg DNA [P = 0.03 versus controls]; sGAG neosynthesis in hTNFtg cells stimulated with BMP-7 at 100 ng/ml, 7.6 ± 1 cpm/μg DNA [P = 0.85 versus basal medium controls]; at 250 ng/ml, 28.8 ± 4.9 cpm/μg DNA [P < 0.001 versus controls]; at 500 ng/ml, 32.7 ± 8.9 cpm/μg DNA [P = 0.004 versus controls]) (Figure 3C). In contrast to the results in chondrocytes derived from hTNFtg mice, specimens obtained from WT mice showed a strong increase in isotope uptake after stimulation with the growth factors at all concentrations used (sGAG neosynthesis in WT cells stimulated with GDF-5 at 100 ng/ml, 76.9 ± 4.4 cpm/μg DNA [P < 0.001 versus basal medium controls]; at 250 ng/ml, 232.8 ± 22.1 cpm/μg DNA [P < 0.001 versus controls]; at 500 ng/ml, 292.5 ± 13.7 cpm/μg DNA [P < 0.001 versus controls]; sGAG neosynthesis in WT cells stimulated with BMP-7 at 100 ng/ml, 57.9 ± 1.7 cpm/μg DNA [P < 0.001 versus basal medium controls]; at 250 ng/ml, 473.6 ± 32.2 cpm/μg DNA [P < 0.001 versus controls]; at 500 ng/ml, 486.1 ± 51.6 cpm/μg DNA [P < 0.001 versus controls]) (Figure 3C).

Overexpression of BMP-7 in the synovial membrane.

Since the growth factor expression was decreased in the articular cartilage of hTNFtg mice, we wondered whether this would also be the case in the synovium of hTNFtg mice. Interestingly, we found that in the hyperplastic synovial membrane of hTNFtg mice, the majority of the cells were BMP-7 positive, whereas only a few scattered cells displayed positive staining in the synovium of WT animals (Figures 4C and D). With regard to expression of GDF-5, a small increase in GDF-5–expressing cells could be observed in the synovial lining layer, as compared with that in WT controls (Figures 4A and B).

Figure 4.

Expression of growth differentiation factor 5 (GDF-5) and bone morphogenetic protein 7 (BMP-7) in the synovium. Histologic sections of wild-type (WT) and human tumor necrosis factor α–transgenic (hTNFtg) mice were analyzed by immunohistochemistry for the expression of GDF-5 (A and B) and BMP-7 (C and D) (n = 3 per group). Results from 1 representative experiment are shown (original magnification × 400).

Cell proliferation rate and extracellular matrix biosynthesis in hTNFtg FLS.

The increased expression of GDF-5 and, in particular, BMP-7 led us to investigate the effects of these growth factors on the cell proliferation rate and the synthesis of extracellular matrix in hTNFtg FLS.

To elucidate their stimulatory effects on the collagenous matrix components, real-time PCR analysis was performed to quantify the expression of the Col1a1 and Col2a1 genes. As shown in Figure 5A, Col1a1 mRNA expression was increased almost 4-fold after GDF-5 or BMP-7 stimulation, while Col2a1 expression was decreased by 50% with either growth factor.

Figure 5.

Effects of stimulation with GDF-5 and BMP-7 on hTNFtg fibroblast-like synoviocytes (FLS). A, Endogenous expression of types I and II collagen in hTNFtg FLS after stimulation with GDF-5 and BMP-7 was determined by quantitative real-time polymerase chain reaction. Subconfluent FLS monolayers were cultured in the presence or absence of GDF-5 (100 ng/ml) and BMP-7 (100 ng/ml) under serum-free conditions; basal medium (BM) alone served as the unstimulated control. Total RNA was isolated on day 5. Results are the mean relative gene expression after normalization to GAPDH. B, Synthesis of sulfated glycosaminoglycans in hTNFtg FLS after GDF-5 and BMP-7 stimulation was determined by incubating FLS in BM alone or in BM with the addition of GDF-5 (100 ng/ml) or BMP-7 (100 ng/ml). On day 5 of stimulation, cell cultures were labeled with 35S-sulfate for 5 hours. Isotope uptake was measured by gel chromatography and normalized to cell DNA content. ∗ = P < 0.006; ∗∗ = P < 0.001 versus BM control. C, The rate of proliferation of hTNFtg FLS after GDF-5 and BMP-7 stimulation was determined in serum-starved cultures of subconfluent FLS incubated with GDF-5 or BMP-7 (each at a final concentration of 100 ng/ml) for 24 hours; cultures maintained in serum-free BM alone served as the negative control. Cells were labeled with 1 μCi/ml of 3H-thymidine for 6 hours. Results in B and C are the mean and SEM. ∗ = P < 0.03 versus BM control. See Figure 4 for other definitions.

Moreover, both GDF-5 and BMP-7 markedly induced the neosynthesis of sGAG in hTNFtg FLS. The rate of newly synthesized matrix proteoglycans was found to be a mean ± SEM 23.4 ± 0.3 cpm/μg DNA when cells were incubated with GDF-5, as compared with a rate of 15.8 ± 0.4 cpm/μg DNA in basal medium controls (P < 0.001). After BMP-7 stimulation, the rate of matrix proteoglycan neosynthesis was 26.9 ± 1.3 cpm/μg DNA (P < 0.006 versus basal medium controls) (Figure 5B).

The 3H-thymidine incorporation experiments demonstrated a significant increase in the cell proliferation rate in BMP-7–treated cultures (mean ± SEM 3,037.8 ± 244.9 cpm versus 1,552.9 ± 212.8 cpm in basal medium controls; P < 0.03). In contrast, stimulation with GDF-5 did not lead to a significant increase in the cell proliferation rate (2,779.2 ± 587.6 cpm; P = 0.16 versus basal medium controls), although there was a trend toward significance (Figure 5C).


TNFα-mediated arthritis is characterized by the destruction of bone and cartilage, both in humans and in mice (24, 28, 29). While the mechanisms leading to bone erosions have been well defined in recent years (1–3), little is known about the molecular mechanisms involved in cartilage degradation. Inflammation-driven activation of cells residing in the synovial membrane is believed to be involved, at least in part, in the process of cartilage destruction (30, 31). In the present study, however, we found that the expression of GDF-5 and BMP-7 as well as synthesis of sGAG were decreased in the articular cartilage derived from an animal model of chronic destructive arthritis, while the FLS displayed an increase in growth factor expression and matrix biosynthesis. These events possibly contribute to the impairment of cartilage repair and synovial hypertrophy in arthritis.

Degradation of articular cartilage is a severe consequence of chronic arthritis and a central feature of RA. In this condition, the balance between anabolic and catabolic factors is disturbed (32, 33), presumably due to overexpression of proinflammatory cytokines. In fact, TNFα is recognized as a central molecule in RA (34); it is not only a key player in perpetuating joint inflammation and joint destruction (35) but also a major catabolic effector molecule in cartilage, due to its stimulatory effects on matrix degradation by metalloproteinases and inhibitory effects on proteoglycan synthesis (8, 36). Whether TNFα-driven arthritis affects anabolic signals, and thus enhances degradation of cartilage and also prevents tissue repair, has not yet been addressed in an in vivo situation.

Taking into account the catabolic potential of TNFα and the anabolic potential of BMPs in articular cartilage, we evaluated the expression of and response to GDF-5 and BMP-7 in hTNFtg chondrocytes. Herein we report a decrease in GDF-5 and BMP-7 expression as well as a loss in growth factor responsiveness in hTNFtg chondrocytes. These data suggest that a state of chronic inflammation, as induced by TNFα in our model, might lead to and be responsible for a down-regulation of growth factor expression and loss of responsiveness to these proteins. Our findings are in accordance with those from studies by Glansbeek et al and Verschure et al demonstrating the failure of various growth factors, including BMP-2, to increase proteoglycan synthesis and replenish depleted cartilage matrix in zymosan-induced arthritis in mice (37, 38). Intriguingly, Fukui and coworkers reported an up-regulation of BMP-2 by TNFα stimulation in primary chondrocytes derived from normal and OA cartilage, as well as in differentiated chondroprogenitor-like cells (39, 40). It must be assumed that in nonchronic conditions (i.e., short-term exposure to TNFα), the presence of proinflammatory cytokines may result in a differentially regulated response compared with that in conditions of chronic inflammation, and might therefore reflect only a transient increase in distinct gene expression. Thus, the data presented indicate that impairment of cartilage extracellular matrix synthesis in chronic arthritis might be due to down-regulation and/or loss of responsiveness to BMPs.

Results of previous studies on BMP expression and the stimulatory response in OA cartilage, however, suggested that cartilage breakdown and the resulting failure to promote matrix synthesis in OA are not governed by a decrease in the expression of or response to BMPs (22, 23, 41, 42). These findings in OA cartilage are in contrast to our data in hTNFtg chondrocytes. Thus, there appears to be a fundamental difference between OA and chronic arthritis (such as RA) in the contributions of the various processes involved in cartilage remodeling. While both disorders are characterized by excessive extracellular matrix catabolism and breakdown as well as insufficient tissue repair, the inflammation-driven decrease in cartilage matrix synthesis appears to be, at least in part, mediated by an impaired response to, or decreased expression of, anabolic factors, such as GDF-5 or BMP-7.

A different explanation for the reduced sGAG synthesis in cartilage of arthritic joints could be a loss of chondrocytes due to cell death, as proposed earlier (43, 44). However, our data do not support this concept. In fact, TNFα has been shown to potentially protect articular chondrocytes against apoptosis (45). Moreover, as shown in the analyses of hTNFtg mice, the decrease in cartilage matrix synthesis was apparently not primarily due to a loss of chondrocytes, since there was no difference in DNA content in the hTNFtg cartilage tissue explants compared with the WT controls.

Interestingly, in contrast to BMP down-regulation in articular chondrocytes, we found that the BMPs, especially BMP-7, were overexpressed in the arthritic synovium of hTNFtg mice. Consistent with this observation, Lories and coworkers recently reported an up-regulation of BMPs, including BMP-7, in synovial samples derived from human RA joints (46). Moreover, that study demonstrated an increased BMP expression by FLS cultures after incubation with TNFα. Given the presence of BMP receptors in FLS from RA patients (47), we extended these findings by investigating the presence of GDF-5 and BMP-7 in, and their stimulatory effects on, hTNFtg FLS and were able to show an overexpression of BMPs in the synovial membrane. This was accompanied by a significant increase in the cell proliferation rate as well as in the matrix synthesis after growth factor stimulation. These findings inversely mirror the data obtained in hTNFtg cartilage, and suggest a possible role of BMPs in the synovial hyperplasia of chronic arthritis.

In contrast, TGFβ, but not BMPs 2 and 6, stimulate cell proliferation in FLS (46). Another study even revealed a decrease in the expression of BMPs 4 and 5 in human RA synovium (48). These discrepancies should not be overly surprising, since specificity and diversity in BMP signaling is dependent on a variety of factors (49) and could be indicative of the differences in biologic function and regulation of individual members of the BMP family.

In conclusion, the presented data reveal a decrease in GDF-5 and BMP-7 expression in the articular cartilage in contrast to an up-regulation of these growth factors in the synovial membrane in TNFα-mediated arthritis. Whether TNFα affects chondrocytes and FLS directly or indirectly has not yet been elucidated, and further studies will be needed to address this question. However, the results obtained shed new light on the alterations that can occur in the articular cartilage and synovial membrane in a state of chronic arthritis. On the one hand, the down-regulation of GDF-5 and BMP-7 suggests that a lack of anabolic factors might play an important role in arthritic cartilage breakdown. On the other hand, increased BMP expression could contribute to synovial hypertrophy.


Dr. Bobacz had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Bobacz, Smolen, Schett.

Acquisition of data. Bobacz, Sunk, Hayer, Amoyo, Makiyeh, Kollias.

Analysis and interpretation of data. Bobacz, Sunk.

Manuscript preparation. Bobacz, Smolen, Schett.

Statistical analysis. Bobacz.


This investigation has been performed under the auspices of the Joint and Bone Center (Schwerpunkt muskuloskelettale Erkrankungen) of the Medical University of Vienna.