A novel role for suppressor of cytokine signaling 3 in cartilage destruction via induction of chondrocyte desensitization toward insulin-like growth factor

Abstract

Objective

An important mechanism contributing to cartilage destruction in arthritis is chondrocyte desensitization toward its main anabolic factor, insulin-like growth factor 1 (IGF-1). In this study, we sought to determine the role of suppressor of cytokine signaling 3 (SOCS-3) in the induction of IGF-1 desensitization of murine chondrocytes.

Methods

Chondrocyte responsiveness to IGF-1 was assessed by ³⁵S-sulfate incorporation into proteoglycans (PGs), via aggrecan messenger RNA expression, using quantitative real-time polymerase chain reaction or insulin receptor substrate 1 (IRS-1) tyrosine phosphorylation (Western blot analysis). IGF-1 desensitization of patellar chondrocytes was studied in zymosan-induced arthritis. IGF-1 desensitization was induced in patellar cartilage explants or the H4 chondrocyte cell line, exposed to interleukin-1α (IL-1α). SOCS-3 protein expression was assessed by immunohistochemistry or by Western blot analysis of protein extracts. The role of SOCS-3 in IGF-1 signaling was elucidated by adenoviral overexpression.

Results

Exposure of murine articular cartilage to IL-1 caused a significant decrease in IGF-1–induced PG synthesis. This effect also occurred in inducible nitric oxide synthase–knockout mice, revealing the involvement of a secondary IL-1–induced factor other than nitric oxide. We showed that IL-1 significantly up-regulated SOCS-3 transcription and protein synthesis in H4 chondrocytes. In contrast, IL-18 was unable to induce SOCS-3 expression and failed to induce chondrocyte IGF-1 desensitization. Histologic analysis of samples from arthritic knee joints revealed high expression of SOCS-3 in chondrocytes. Through adenoviral overexpression of SOCS-3, we obtained direct evidence that SOCS-3 inhibits IGF-1–mediated cell signaling, since IRS-1 phosphorylation was reduced.

Conclusion

This study demonstrates that IL-1–induced SOCS-3 expression is a novel mechanism of IGF-1 desensitization in chondrocytes; in conjunction with nitric oxide it can contribute to cartilage damage during arthritis.

Destruction of articular cartilage is a prominent feature of inflammatory joint diseases such as rheumatoid arthritis (RA). Cartilage degradation is characterized by initial loss of proteoglycans (PGs) from the cartilage, followed by irreversible destruction of the collagen matrix and changes in the subchondral bone. Although the precise mechanism underlying cartilage damage remains largely unknown, a role for interleukin-1 (IL-1) appears evident (1, 2).

Chondrocytes are the sole residents of the cartilage, and are therefore believed to maintain the equilibrium between synthesis and degradation of the cartilage matrix. This cartilage homeostasis is tightly controlled by catabolic and anabolic factors. Important anabolic proteins, e.g., insulin-like growth factor 1 (IGF-1), control cartilage homeostasis by inducing the production of extracellular matrix components, such as aggrecan (3, 4). This characteristic makes IGF-1 an essential signal for monitoring cartilage integrity.

A principal tenet holds that cartilage damage results from the disparity between anabolic and catabolic factors. Essential catabolic factors, such as IL-1 and tumor necrosis factor α (TNFα), are well-known inhibitors of chondrocyte PG synthesis, with IL-1 being more potent than TNFα (5). These proinflammatory cytokines also play a crucial role in the pathogenesis of arthritis. While these cytokines help to regulate the secretion and activation of the main contributors to cartilage destruction, matrix metalloproteinases, they also directly affect chondrocyte metabolic activity, influencing cartilage homeostasis. IGF-1 can stimulate PG synthesis to superoptimal levels in chondrocytes, and IGF-1 nonresponsiveness might compromise cartilage repair after an arthritic insult.

Nonresponsiveness toward IGF-1 supplementation can be observed in patellar cartilage explants from mice with experimental arthritis, which is not caused by general impairment of chondrocytes (6–8). It has been demonstrated that addition of IL-1 to patellar explants can lead to loss of anabolic signaling by IGF-1. Previously, we showed that IGF receptor I (IGFRI) is abundantly expressed in articular cartilage, and that IGF-1 desensitization, as can be observed during arthritis or osteoarthritis, was not the result of decreased IGFRI expression on chondrocytes (9–11). IL-1–induced nitric oxide (NO) can cause cartilage dysfunction in arthritis (12, 13) by the inhibition of PG (14) and collagen (15) synthesis in chondrocytes. There is evidence that NO is also involved in the induction of chondrocyte IGF-1 nonresponsiveness. However, blocking of NO using L-arginine–based inhibitors led to only partial recovery of the IGF-1 response (16, 17), thereby revealing that NO is not the only factor involved in the induction of IGF-1 desensitization.

The purpose of the present study was to elucidate the NO-independent factors that contribute to IGF-1 nonresponsiveness in chondrocytes. Since suppressor of cytokine signaling 3 (SOCS-3) works as a negative feedback regulator for cytokine and growth factor signaling (18, 19), we investigated the involvement of SOCS-3 in IGF-1 desensitization. We demonstrated SOCS-3 expression in chondrocytes during experimental arthritis and after IL-1 exposure. Furthermore, IL-1 could induce IGF-1 desensitization in chondrocytes under conditions of NO deprivation. We provide evidence that SOCS-3 can desensitize chondrocytes to IGF-1 by adenoviral overexpression of SOCS-3, which can prevent IGF-1 cell signaling by inhibiting insulin receptor substrate 1 (IRS-1) phosphorylation. Therefore, IL-1–induced SOCS-3 expression could be a novel mediator of cartilage destruction by interfering with the IGF-1 signaling that causes chondrocyte desensitization.

MATERIALS AND METHODS

Animals.

Male C57BL/6 mice ages 10–12 weeks were obtained from Charles River Wiga (Sulzfeld, Germany). The homozygous inducible NO synthase (iNOS)–deficient mice and N/N (C57BL/6 × 129/SV) wild-type mice were described previously (1) and bred in our own animal housing facilities. The mice were housed in filtertop cages and were fed a standard diet, with water and food ad libitum. All in vivo studies complied with national legislation and were approved by local authorities for care and use of animals with related codes of practice.

Materials.

Bovine serum albumin (BSA) was purchased from Sigma (St. Louis, MO), and RPMI 1640 and Dulbecco's modified Eagle's medium (DMEM)–Ham's F-12, containing L-glutamine and 15 mM HEPES buffered saline, were obtained from Invitrogen (Breda, The Netherlands). Recombinant murine IL-1α and IL-18 were purchased from R&D Systems (Abingdon, UK). Recombinant human IGF-1 was purchased from Preprotech (Rocky Hill, NJ). Radioactive ³⁵S-sulfate was obtained from NEN Life Science Products (Boston, MA). NO scavengers L-N⁵(1-iminoethyl)ornithine (L-NIO) and N^G-monomethyl-L-arginine (L-NMMA) were obtained from Biomol (Plymouth Meeting, PA) and were used in cell culture at a concentration of 1 mM.

Construction of adenoviral vector.

The first step in construction of a first-generation E1/E3–deleted serotype 5 adenovirus was the cloning of the gene of interest. The SOCS-3 was cloned from complementary DNA (cDNA) obtained from the murine B9 hybridoma cell line (European Collection of Cell Cultures). Viral vectors were E1A, B, and E3 deleted, and produced according to the method described by Chartier et al (20). Briefly, SOCS-3 cDNA was inserted into the E1-deleted region of the serotype 5 adenoviral plasmid, and was regulated by a cytomegalovirus promoter. A purified recombinant adenoviral vector DNA was linearized through digestion with Pac I and transfected into viral packaging 293 cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Ad5SOCS-3 and the control vector Ad5 luciferase were purified using 2× CsCl gradient purification and stored in small aliquots at −80°C. The viral titer of the purified viral vectors was determined in human embryonic retinoblastoma 911 indicator cells by immunohistochemical detection of viral capsid protein 20 hours after transfection (21).

Chondrocyte proteoglycan synthesis.

Patellae with minimal surrounding synovial tissue were dissected from the knee joint of naive C57BL/6 mice (22) and were cultured for 24 hours in RPMI 1640 medium containing Glutamax and gentamicin (50 μg/ml) in the presence or absence of IGF-1 (250 ng/ml), IL-1α (10 ng/ml), or IL-18 (100 ng/ml). After 24 hours, the culture medium was replaced by either medium containing 500 ng/ml IGF-1 or control medium, and patellae were cultured for an additional 24 hours. Thereafter, patellae were placed in RPMI 1640 containing Glutamax, gentamicin, and ³⁵S-sulfate (0.74 mBq/ml) and incubated for 3 hours (37°C, 5% CO₂). Patellae were washed 3 times in saline, fixed in 4% formaldehyde, and subsequently decalcified for 4 hours in 5% formic acid.

Remaining adjacent synovial tissue was removed and the patellae were dissolved at 65°C in 0.25 ml Lumasolve (Omnilabo, Breda, The Netherlands). After addition of 1 ml Lipoluma (Omnilabo), ³⁵S-sulfate incorporation was determined by liquid scintillation measurement (Trilux 1450 microbeta; EG&G Wallac, Turku, Finland).

Zymosan-induced arthritis (ZIA).

ZIA was elicited through intraarticular injection of 180 μg zymosan A (Saccharomyces cerevisiae) as described previously (23).

Intraarticular overexpression of IL-1.

Human IL-1β was overexpressed by intraarticular injection of 3 × 10⁶ plaque-forming units of adenoviral vector (Ad5hIL-1β). Seven days after injection, knee joints were dissected and processed for further analysis of SOCS-3 expression in the articular cartilage.

Western blotting.

For protein detection using Western blotting, 30 μg of a protein sample was loaded on a sodium dodecyl sulfate–10% polyacrylamide gel. Proteins were electrophoretically transferred onto polyvinylidene difluoride membranes (Hybond P; Amersham Pharmacia Biotech, Buckinghamshire, UK). Nonspecific protein binding was blocked using 0.05% Tris buffered saline–Tween 20 containing 1% BSA and 1% nonfat dry milk (Campina, Eindhoven, The Netherlands). After incubation with the first (1:1,000) and secondary (1:1,500) antibodies, the membranes were developed using an enhanced chemiluminescence detection system (ECL Plus; Amersham Pharmacia Biotech). For reprobing, membranes were stripped using a solution of 0.2M glycine/0.05% Tween 20 (pH 2.5) at 80°C for 20 minutes. Subsequently, quantitative Western blot analysis was performed by calculating the relative density of the immunoreactive bands after acquisition of the blot image with a JVC 3-CCD video camera (Victor Company of Japan, Tokyo, Japan) and analysis with the Qwin image analysis system (Leica Imaging Systems, Cambridge, UK).

Antibodies.

Rabbit anti-mouse STAT-1, STAT-3, phospho–STAT-1 (Tyr⁷⁰¹), and phospho–STAT-3 (Tyr⁷⁰⁵) were obtained from Cell Signaling Technology (Beverly, MA). Goat anti-mouse SOCS-3 and phospho–IRS-1 (Tyr⁹⁸⁹) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). As a control for protein loading, goat anti-mouse β-actin was purchased from Santa Cruz Biotechnology. Horseradish peroxidase–labeled and biotinylated secondary antibodies were obtained from Dako (Glostrup, Denmark). For immunofluorescence, streptavidin fluorescein isothiocyanate (1:100 in phosphate buffered saline [PBS]; Dako) and rhodamine avidin D (1:500 in PBS; Vector, Burlingame, CA) were used. For nuclear counterstaining, 4′,6-diamidino-2-phenylindole (DAPI) was used (400 ng/ml in PBS; Molecular Probes, Eugene, OR).

Cell culture.

SV40 large T antigen–immortalized H4 chondrocytes (25) derived from hip articular cartilage of C57BL/6 mice were cultured in DMEM–Ham's F-12 medium (Gibco BRL, s'-Hertogenbosch, The Netherlands) supplemented with 5% fetal calf serum (FCS; Life Technologies, Breda, The Netherlands) and 40 μg/ml gentamicin (Centrafarm, Etten-Leur, The Netherlands) at 37°C. One day prior to the experiments, cells were seeded in 24-well plates (2 × 10⁵/well) or 6-well plates (1 × 10⁶/well). Sixteen hours before stimulation, cells were cultured in DMEM without FCS. In studies of adenoviral gene delivery in H4 chondrocytes, cells were stimulated 1 day posttransfection, with either IL-1α (10 ng/ml), IL-18 (100 ng/ml), or IGF-1 (250 ng/ml).

Immunofluorescence.

H4 chondrocytes were cultured in 4-well chamber slides (Nalge Nunc International, Naperville, IL) and transfected with adenoviruses at multiplicity of infection (MOI) of 50. Twenty-four hours later, cells were incubated with 250 ng/ml IGF-1 for 10 minutes, and thereafter fixed in methanol for 20 minutes at −80°C. Cells were permeabilized with 1× PBS/0.2% Triton X-100. Nonspecific binding was blocked by incubation with blocking buffer consisting of 4% nonfat dry milk and 1% BSA in PBS for 1 hour at room temperature. Cells were incubated for 1 hour at room temperature with the first antibody diluted 1:250 in blocking buffer. Unbound antibodies were removed by extensive washing with a 0.2% Tween solution. Thereafter, cells were incubated for 1 hour with the conjugated second antibody. After washing, the chamber slides were washed with H₂O and methanol for 5 minutes, air dried, and enclosed in Mowiol (EMD Biosciences, La Jolla, CA) for fluorescence microscope analysis.

Assessment of nitrite levels.

The concentration of NO (a stable end product of NO) was determined by Griess reaction, using NaNO₂ as a standard as described previously (1).

RNA isolation and quantitative polymerase chain reaction (PCR) analysis.

H4 chondrocytes were cultured for 24 hours in the presence of the stimuli, unless stated otherwise. Thereafter, cells were washed with saline. Total RNA was extracted from these samples using TRIzol reagent, as described by Chomczynski and Sacchi (24). Isolated RNA was treated with DNase before being reverse-transcribed into cDNA with Moloney murine leukemia virus reverse transcriptase using oligo(dT) primers. Quantitative real-time PCR was performed using the ABI Prism 7000 sequence detection system (Applied Biosystems, Foster City, CA). The PCR protocol was as follows: 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. All PCRs were performed using SYBR Green master mix (Applied Biosystems), 10 ng of cDNA, and a primer concentration of 300 nmoles/liter in a total reaction volume of 25 μl.

PCR signals were quantified by comparing the cycle threshold of the gene of interest of each sample with the cycle threshold value of the reference gene GAPDH. Primers were designed over intron–exon boundaries using Primer Express (Applied Biosystems). Primer sequences for gene expression analysis were as follows: for mouse GAPDH forward 5′-GGCAAATTCAACGGCACA-3′, reverse 5′-GTTAGTGGGGTCTCGCTCCTG-3′, for mouse SOCS-3 forward 5′-CTGGTACTGAGCCGACCTCTCT-3′, reverse 5′-CCGTTGACAGTCTTCCGACAA-3′, and for mouse aggrecan forward 5′-TCTACCCCAACCAAACCGG-3′, reverse 5′-AGGCATGGTGCTTTGACAGTG-3′.

Statistical analysis.

Significance of differences was determined using Student's t-test (GraphPad Software, San Diego CA). P values less than or equal to 0.05 were considered significant.

RESULTS

IGF-1 desensitization of articular cartilage independent of nitric oxide production.

Culturing patellar cartilage explants with IGF-1 enhanced chondrocyte PG synthesis, as determined by ³⁵S-sulfate incorporation. PG synthesis was suppressed in cartilage obtained from joints with ZIA, and IGF-1 supplementation had no anabolic effect (Figure 1A). Arthritis-induced inhibition of PG synthesis was completely prevented by iNOS deficiency, whereas the IGF-1–induced synthesis did not reach its maximal stimulation (Figure 1B). Short-term (48 hour) incubation of patellar cartilage with IL-1α (10 ng/ml) induced pronounced inhibition of PG synthesis and complete IGF-1 desensitization (Figure 1C). In the presence of the NO scavengers L-NIO and L-NMMA, the IL-1–induced inhibition of PG synthesis was completely prevented. Notably, the chondrocytes remained IGF-1 nonresponsive (Figure 1C). Both iNOS inhibitors completely blocked the IL-1–induced NO generation, and had no effect on basal or IGF-1–stimulated chondrocyte PG synthesis (Figures 1C and D). This suggests that chondrocyte IGF-1 desensitization can occur via an alternative mechanism, independent of NO.

Figure 1. Involvement of nitric oxide (NO) in insulin-like growth factor 1 (IGF-1) nonresponsiveness of chondrocytes. A and B, Patellae (n = 6 per group) were dissected 1 day after induction of zymosan-induced arthritis (ZIA). Proteoglycan (PG) synthesis was determined after 24-hour incubation; samples were labeled by incubation with ³⁵S-sulfate for an additional 3 hours. Addition of IGF-1 (250 ng/ml) significantly stimulated PG synthesis in chondrocytes of naive mice. A, Chondrocytes in cartilage derived from ZIA joints had reduced PG synthesis, and were not responsive to IGF-1. ∗ = P < 0.05 versus controls without IGF-1 stimulation. B, IGF-1 responsiveness of articular chondrocytes in arthritic joints was partially restored in inducible NO synthase (iNOS)–deficient mice. C and D, Patellae of naive mice (n = 6 per group) were cultured for 48 hours in the presence of NOS inhibitors N-iminoethyl L-ornithine (L-NIO) (1 mM) and N^G-monomethyl-L-arginine (L-NMMA) (1 mM). C, Under culture conditions without NO production, interleukin-1 (IL-1)–induced inhibition of IGF-1 stimulation of PG synthesis in chondrocytes was decreased. D, Under culture conditions without NO production, IL-1–induced NO production by chondrocytes was decreased. Values are the mean and SD, representative of 2 experiments.

SOCS-3 expression in articular chondrocytes during joint inflammation.

SOCS-3 plays an important intermediate role in insulin insensitivity of patients with diabetes, and in parallel could be a prime candidate in chondrocyte IGF-1 desensitization. By immunohistochemical analysis, we found that chondrocytes expressed SOCS-3 protein in inflamed knee joints that were either affected by ZIA or had undergone long-term (≥7 days) exposure to IL-1 in a model of forced intraarticular expression, in which we used an adenoviral vector containing the human IL-1β gene (Figure 2). The presence of SOCS-3 in chondrocytes, under conditions that also showed IGF-1 desensitization, suggested an association, which was investigated in more detail using our SV40 large T antigen–immortalized murine H4 articular chondrocyte cell line (25).

Figure 2. Expression of suppressor of cytokine signaling 3 (SOCS-3) protein in articular cartilage from mouse knee joints. A, Normal joint. B, Zymosan-injected joint, 1 day after ZIA induction. C, Intraarticular overexpression of IL-1 in C57BL/6 mouse knee joint. Tissue sections were stained for immunohistochemical analysis of SOCS-3 expression and counterstained using hematoxylin. SOCS-3 protein expression could not be detected in articular chondrocytes of naive mice, and expression was already markedly increased 1 day after induction of ZIA. After intraarticular overexpression of IL-1 for 7 days, SOCS-3 expression in chondrocytes was dramatically increased. Arrowheads indicate the main expression of SOCS-3, located in chondrocytes. P = patella; F = femur; I = inflammatory cell mass located in the patellofemoral joint space. (Original magnification × 200 in A and C; × 400 in B.) See Figure 1 for other definitions.

High levels of SOCS-3 expression in chondrocytes that are IGF-1–nonresponsive after IL-1 exposure.

Incubation of H4 chondrocytes with IL-1 significantly reduced aggrecan messenger RNA (mRNA) expression for 24 hours thereafter (Figure 3A). In contrast, IL-18 incubation did not suppress levels of aggrecan mRNA (Figure 3A), and in comparison with IL-1, only marginally enhanced NO synthesis (data not shown). IL-18 failed to inhibit PG synthesis on patellar cartilage explants, and IL-18–stimulated explants also remained IGF-1 responsive (Figure 3B), confirming that the lack of response of H4 chondrocytes to IL-18 was not an artifact. Since both IL-1 and IL-18 induce NF-κB on chondrocytes (data not shown), we analyzed additional differences in the signaling pathways. We found that IL-1 induced phosphorylation of both STAT-1 and STAT-3 in chondrocytes, whereas IL-18 only induced STAT-3 phosphorylation. This could explain the difference between the 2 cytokines with regard to PG synthesis. In addition, the incubation with IL-1 led to a significant increase of SOCS-3 mRNA levels (∼700% compared with control levels), whereas IL-18 only marginally induced SOCS-3 transcription (Figure 3D).

Figure 3. Inhibition of chondrocyte PG synthesis by IL-1, but not by IL-18. H4 chondrocyte cells were cultured in the presence or absence of IL-1α (10 ng/ml) or IL-18 (100 ng/ml). Thereafter, cells were either lysed and total protein was obtained, or total RNA was extracted using TRIzol. A and D, Expression of mRNA for both aggrecan (A) and suppressor of cytokine signaling 3 (SOCS-3) (D) was determined using quantitative reverse transcriptase–polymerase chain reaction (PCR) analysis. Levels of mRNA were normalized against GAPDH mRNA levels, quantitative analysis was performed after 35 PCR cycles, and results presented as a percentage of control (see Materials and Methods for further details). Values are the mean and SEM of 3 experiments. ∗ = P < 0.01 versus control treatment. B, Patellae from naive C57BL/6 mice (n = 5 per group) were cultured for 24 hours with or without IL-1 (10 ng/ml) or IL-18 (100 ng/ml) in the presence of IGF-1 (250 ng/ml), followed by an additional incubation in either the presence or absence of 500 ng/ml IGF-1 for 24 hours. Thereafter, patellae were pulse labeled for 3 hours with ³⁵S-sulfate. Stimulation of articular chondrocytes with IL-1 resulted in significant inhibition of IGF-1 stimulation, whereas stimulation with IL-18 did not. Values are the mean and SD (n = 5 patellae per group). C, Western blot analysis was used to study STAT-1 and -3 phosphorylation 2 hours after stimulation with IL-1, IL-18, IGF-1, or combinations of the selected cytokines. Cell extracts from HeLa cells stimulated with interferon-α were used as a positive control (+), whereas unstimulated cell extracts from HeLa cells served as negative controls (−). D, SOCS-3 protein expression was studied after 8 or 24 hours of stimulation with IL-1 or IL-18. Values are the mean and SD. ∗ = P < 0.01 versus control treatment. See Figure 1 for other definitions.

Consistent with increased levels of SOCS-3 mRNA in chondrocytes after IL-1 stimulation, Western blot analysis revealed increased SOCS-3 protein levels after IL-1, but not IL-18 treatment (Figure 3D). Neither incubation with IL-1 nor incubation with IL-18 led to an increase of SOCS-1 mRNA and protein in cultured H4 chondrocytes (data not shown). Similar to the observation in cartilage explants, the H4 chondrocytes became IGF-1–insensitive with IL-1 treatment, and at the same time showed enhanced SOCS-3 expression (Figure 4).

Figure 4. Relationship between suppressor of cytokine signaling 3 (SOCS-3) and aggrecan expression. H4 chondrocytes were cultured with or without IL-1 (10 ng/ml) for 48 hours and then for 24 hours in the presence of IGF (250 ng/ml). Thereafter, total RNA was extracted using TRIzol; expression levels of SOCS-3 and aggrecan mRNA were determined using quantitative reverse transcriptase–polymerase chain reaction (PCR). The mRNA levels were normalized against GAPDH mRNA levels, quantitative analysis was performed after 35 PCR cycles, and results presented as a percentage of control (see Materials and Methods for further details). Values are the mean and SD, representative of 2 experiments, with similar outcome. ∗ = P < 0.01 versus control. See Figure 1 for other definitions.

SOCS-3–induced chondrocyte IGF-1 desensitization.

To provide direct evidence that SOCS-3 is able to induce chondrocyte IGF-1 desensitization, we overexpressed the murine SOCS-3 gene in H4 chondrocytes. For this, the cells were infected with an adenovirus encoding the murine SOCS-3 gene or as a control, the luciferase gene. Forced expression of SOCS-3 protein abrogated the IGF-1–induced expression of aggrecan mRNA expression in H4 chondrocytes (Figure 5). Furthermore, IGF-1 desensitization of chondrocytes by SOCS-3 was comparable with IL-1–induced desensitization. These data reveal the involvement of SOCS-3 in IGF-1 signaling, and furthermore, suggest that the IL-1–induced desensitization is likely to be mediated through SOCS-3.

Figure 5. Suppressor of cytokine signaling 3 (SOCS-3)–induced IGF-1 nonresponsiveness of chondrocytes. H4 chondrocytes were infected at a multiplicity of infection of 50 with either Ad5 luciferase or the adenoviral vector containing the murine SOCS-3 gene. SOCS-3 overexpression significantly inhibited IGF-1–induced stimulation of aggrecan mRNA. Aggrecan mRNA expression levels were corrected for the household gene GAPDH and normalized for the basal aggrecan mRNA expression levels. SOCS-3 overexpression followed by addition of IL-1 resulted in an almost complete inhibition of aggrecan expression by H4 articular chondrocytes. Values are the mean and SD; the experiments were performed twice with similar results. ∗ = P < 0.01 versus Ad5 luciferase stimulation, by Student's t-test. See Figure 1 for other definitions.

SOCS-3 inhibition of IGF-1–induced STAT-3 and IRS-1 phosphorylation.

As shown in Figure 3C, stimulation of H4 chondrocytes for 2 hours with either IL-1 or IL-18 led to the phosphorylation of STAT-3, whereas only IL-1 induced STAT-1 phosphorylation. Forced expression of SOCS-3 in H4 chondrocytes inhibited phosphorylation of both STAT-1 and STAT-3, revealing antagonistic activity of the SOCS-3 protein (Figure 6A). Because IGF-1 signaling is initiated upon binding of IGF-1 to the IGFR, which directly leads to phosphorylation of IRS-1, we wanted to evaluate whether SOCS-3 had a direct effect on IRS-1 phosphorylation. Immunofluorescence studies of cells infected at an MOI of 50 with Ad5SOCS-3 revealed increased cytoplasmic expression of SOCS-3. Stimulation of articular chondrocytes with IGF-1 (250 ng/ml) induced IRS-1 phosphorylation, and this was clearly reduced in cells overexpressing SOCS-3 (Figure 6B). Control transfection did not affect IRS-1 phosphorylation in chondrocytes (results not shown). Western blot analysis showed a clear-cut quantitative reduction of IGF-1 signaling by SOCS-3, resulting in decreased amounts of phosphorylated IRS-1 as determined by densitometry (Figures 6C and D).

Figure 6. Suppressor of cytokine signaling 3 (SOCS-3) overexpression functionally antagonizes STAT activation and inhibits IGF-1 signaling by directly affecting insulin receptor substrate 1 (IRS-1) phosphorylation. H4 chondrocytes were infected with either Ad5 luciferase or the adenoviral vector containing the murine SOCS-3 gene at a multiplicity of infection (MOI) of 50, 24 hours before cytokine or growth factor stimulation, and cultured in the absence of fetal calf serum. A, Transduced H4 chondrocytes were stimulated for 2 hours with either IL-1 (10 ng/ml), IL-18 (100 ng/ml), or IGF-1 (250 ng/ml), and the effect of forced SOCS-3 expression on STAT-1 and -3 phosphorylation was determined by Western blot analysis. B, Immunofluorescence (rhodamine) on H4 cells infected at a MOI of 50 with Ad5SOCS-3, revealing increased cytoplasmic expression of SOCS-3 after 24 hours (solid arrowheads). Stimulation of articular chondrocytes with IGF-1 (250 ng/ml) for 10 minutes induced IRS-1 phophorylation (open arrowheads), which was reduced in cells overexpressing SOCS-3 (original magnification × 600). C, Control or SOCS-3–transfected H4 chondrocytes were stimulated with IGF-1 (250 ng/ml) for 30 minutes or 1 hour. Thereafter, cells were lysed and IGF-1–induced IRS-1 phosphorylation was determined by Western blot analysis. Similar results were obtained in at least 2 identical experiments. D, Quantitative densitometry of Western blot analysis depicting the relative ratios of phosphorylated IRS-1 to actin (solid bars) and phosphorylated IRS-1 to total IRS-1 (shaded bars). IGF-1 stimulated IRS-1 phosphorylation within 60 minutes, and this could be prevented by forced SOCS-3 expression. See Figure 1 for other definitions.

DISCUSSION

This study shows that IL-1, which is released during induction of experimental arthritis, is a potent inducer of SOCS-3 expression in articular chondrocytes. SOCS-3 has been suggested to be a protein with therapeutic potential in arthritis (26), since previous studies have shown that STAT activation is an important feature of initiation and perpetuation of arthritis in mice (27). Herein we show a novel role for SOCS-3 in the regulation of cartilage damage, i.e., direct inhibition of IGF-1–induced chondrocyte PG synthesis.

Cartilage damage can be the result of various pathologic mechanisms, including decreased PG synthesis, increased PG degradation, and increased chondrocyte apoptosis. In this study, we focused mainly on the pathologic mechanisms controlling PG synthesis. Chondrocyte PG synthesis in healthy cartilage is controlled by the anabolic growth factor IGF-1. However, during experimental arthritis, nonresponsiveness of articular cartilage toward IGF-1 is a naturally occurring but undesirable phenomenon, which results in reduced PG synthesis by articular chondrocytes (7, 8) (Figure 1A). This desensitization of chondrocytes toward IGF-1 has also been noted in cartilage of patients with RA. Unfortunately, the specific mechanism underlying the induction of IGF-1 desensitization of articular chondrocytes is far from resolved.

Chondrocytes exposed to IL-1, a cytokine that is elevated in arthritic synovial fluid, express iNOS and produce high levels of NO. Recent studies demonstrate that part of the IGF-1 insensitivity observed in chondrocytes is caused by the induction of NO (17, 28, 29). Our initial experiments using iNOS-knockout mice with ZIA confirmed that cartilage maintained the ability to respond to IGF-1, although to a lesser extent (1) (Figure 1B).

We observed that stimulation of chondrocytes with IL-18 only marginally induced iNOS mRNA and protein, whereas incubation of chondrocytes with IL-1 induced high levels of iNOS mRNA and protein (data not shown). We suggest that the difference between IL-1 and IL-18 is probably dependent on differences in the signaling cascade leading to iNOS induction. In contrast to IL-1, IL-18 stimulation of articular chondrocytes does not lead to activation of STAT-1. Taken together, these results imply that the inability of IL-18 to inhibit cartilage PG synthesis is dependent on the absence of STAT-1 activity. Interestingly, supplementation of STAT-1 activation using interferon-γ (IFNγ) stimulation augmented IL-18–induced iNOS expression in chondrocytes, generating high levels of NO (Smeets RL, et al: unpublished observations). In contrast to IL-1, IL-18 was unable to inhibit IGF-1–induced chondrocyte PG synthesis. Whether simultaneous stimulation of articular chondrocytes with IL-18 and IFNγ will also lead to IGF-1 desensitization in chondrocytes is currently under investigation.

Earlier studies showed that high concentrations of NO directly affect IGF-1 signaling by reducing IGF-1 receptor tyrosine phosphorylation in response to IGF in chondrocytes (17). However, culturing of chondrocytes in the presence of high concentrations of NO donors results in markedly decreased metabolic activity and activated caspases, which ultimately leads to the activation of the apoptotic pathway (30, 31). Therefore, the effects of NO donors on IGF-1–induced PG synthesis by articular chondrocytes should be studied with caution, since part of the effects of NO on PG synthesis can also be due to NO-induced inhibition of metabolic function or even chondrocyte death.

Interestingly, on cartilage explants cultured with NOS inhibitors, IL-1 still potently antagonized IGF-1–induced PG synthesis by articular chondrocytes, revealing that the process of IGF-1 desensitization was in part independent of NO (Figures 1C and D). Our data imply that another IL-1–induced factor is required for the regulation of IGF-1 responsiveness in chondrocytes. In diabetes, IL-1 plays an important role in impairing pancreatic β cell survival and functioning, and it has been recently shown that IL-1 stimulation of β cells reduces insulin signaling, independent of NO (32–34). Stimulation of pancreatic β cells with IL-1 leads to increased expression of SOCS-3, which reduced insulin signaling by deceasing IRS-1 phosphorylation. SOCS-3 could also be an important factor controlling signal transduction of growth factors and cytokines in chondrocytes. We observed that 1 day after induction of ZIA, IGF-1 desensitization of articular chondrocytes occurred, which increased expression of SOCS-3 in the articular chondrocytes of the femur and patella. Furthermore, direct stimulation of cultured chondrocytes with IL-1 resulted in high expression of SOCS-3 protein, in vivo and in vitro.

Our study demonstrates that SOCS-3 is a functional inhibitor of IGF-1 signaling, resulting in reduced PG synthesis in chondrocytes. Apparently, inhibition of IRS-1 phosphorylation by SOCS-3 is also conserved in articular chondrocytes. The mechanism by which SOCS-3, and probably other SOCS proteins, can inhibit IGF-1 or insulin signaling is most likely dependent on 2 characteristic domains of the SOCS proteins. The src homology 2 domain present in SOCS-3 can directly bind phosphotyrosine residues present on activated IGFR, thereby inhibiting IRS-1 phosphorylation. Furthermore, overexpression of SOCS-3 could also lead to increased degradation of signal transducing proteins, due to the presence of a SOCS box in SOCS-3, which is thought to regulate proteosomal degradation of interacting proteins (18, 35, 36).

Recent studies by Hauselmann et al revealed that chondrocytes harvested from the superficial layer of articular cartilage produced significantly more NO after IL-1 challenge. In addition, the IL-1–induced inhibition of PG synthesis was NO independent, whereas IL-1–induced inhibition in the deeper cartilage layers appeared to be NO dependent (14, 16). These findings, along with our data, suggest that IGF-1 desensitization could occur in the superficial cartilage layer, dependent on IL-1–induced SOCS-3 expression, whereas IGF-1 desensitization in the deeper layers of articular cartilage remains NO dependent.

Due to the importance of SOCS-3 in directing IGF-1 sensitivity in chondrocytes, SOCS-3 protein expression has to be tightly regulated at the promoter level. Generally, SOCS proteins are not highly expressed in unstimulated tissues. Cytokine-inducible genes rapidly generate these proteins, often in a STAT-dependent manner (37). We showed that stimulation of chondrocytes with IL-1 leads to the phosphorylation of STAT-1 and -3, thereby leading to induction of SOCS-3. Interestingly, IL-18–induced STAT-3 activation was insufficient to induce SOCS-3. Additional data have revealed that SOCS-3 protein expression is tightly regulated not only by STAT activation, but also by other transcription factors such as Sp-3 (17). Interestingly, stimulation of articular chondrocytes using IL-1 directly induces Sp-3 protein expression (38).

It is known that S-nitrosylation of nuclear regulatory proteins by NO can directly affect transcriptional activity, in both a positive and a negative manner. The nuclear regulatory function of S-nitrosylation appears to be dose-dependent for Sp-3; low concentrations of NO increased binding of Sp-3 to its promoter DNA binding sites, whereas high concentrations of NO decreased Sp-3 binding (39). Therefore, initial induction of iNOS by IL-1, which would produce low, non–stress-inducing levels of NO, could amplify the DNA binding affinity of transcription factors, thereby enhancing the IL-1–orchestrated SOCS-3 expression.

During arthritis, STAT-1 and -3 activation can be observed in the inflamed synovial tissue. Interestingly, STAT-1 deficiency resulted in exacerbation of chronic joint inflammation, revealing reduced SOCS-1 expression in the synovium of mice (27). These data suggested a beneficial role for SOCS expression in the regulation of inflammation of the synovial tissue during arthritis. In contrast to the deleterious effects of increased SOCS-3 expression on cartilage PG synthesis in articular chondrocytes, adenoviral-mediated overexpression of SOCS-3 in the synovium appears to be a beneficial strategy for treating experimental arthritis (26). Controlling SOCS-3 expression in cartilage, without affecting its expression in synovium, will be a challenging effort, and worthy of further research.

In summary, this study demonstrates that IL-1–induced SOCS-3 expression is a novel regulator of IGF-1 signaling in chondrocytes. We show that SOCS-3 is highly expressed in articular chondrocytes in ZIA and after intraarticular overexpression of IL-1. Furthermore, SOCS-3 overexpression in chondrocytes inhibits IGF-1–induced aggrecan expression, by directly inhibiting IGF-1 signaling through antagonizing IRS-1 phosphorylation. Since impaired IGF-1 signaling in chondrocytes can contribute to the progression and pathogenesis of RA, controlling the expression of SOCS-3 in articular chondrocytes could be a promising new strategy leading to restoration of IGF-1 responsiveness in articular chondrocytes of patients with arthritis.

Acknowledgements

We thank Dr. H. van Beuningen for kindly providing the immortalized H4 chondrocyte cell line. The NOS^−/− breeding pairs were a kind gift from Dr. J. S. Mudgett (Merck Laboratories, Rahway, NJ). The 911 human embryonic retinoblastoma cell line was provided by Prof. Dr. R. C. Hoeben (Leiden University Medical Center, Leiden, The Netherlands). The Ad5hIL-1β was a generous gift from Dr. C. Richards (McMaster University, Hamilton, Ontario, Canada).