Shift from toll-like receptor 2 (TLR-2) toward TLR-4 dependency in the erosive stage of chronic streptococcal cell wall arthritis coincident with TLR-4–mediated interleukin-17 production




Toll-like receptors (TLRs) may activate innate and adaptive immune responses in rheumatoid arthritis (RA) through recognition of microbial as well as endogenous ligands that have repeatedly been found in arthritic joints. The objective of this study was to investigate the involvement of TLR-2 and TLR-4 in the development of chronic destructive streptococcal cell wall (SCW)–induced arthritis, in which interleukin-1 (IL-1)/IL-17–dependent T cell–driven pathologic changes replace the macrophage-driven acute phase.


Chronic SCW arthritis was induced by 4 repeated intraarticular injections of SCW fragments in wild-type, TLR-2−/−, and TLR-4−/− mice. Clinical, histopathologic, and immunologic parameters of arthritis were evaluated.


The TLR-2 dependency of joint swelling during the acute phase was shifted to TLR-4 dependency during the chronic phase. Persistent joint inflammation in the latter phase of the model was significantly suppressed in TLR-4−/− mice. In the chronic phase, TLR-4 actively contributed to matrix metalloproteinase (MMP)–mediated cartilage destruction and to osteoclast formation, since the expression of the MMP-specific aggrecan neoepitope VDIPEN and the osteoclast marker cathepsin K was significantly reduced in TLR-4−/− mice. Furthermore, TLR-4−/− mice expressed less IL-1β, tumor necrosis factor α, IL-6, and IL-23, cytokines that are implicated in IL-17 production. Accordingly, SCW-specific IL-17 production was found to be dependent on TLR-4 activation, since T cells from arthritic TLR-4−/− mice produced markedly less IL-17 upon SCW stimulation, whereas interferon-γ production remained unaffected.


These data indicate the involvement of TLR-4 in the chronicity and erosive character of arthritis coincident with the antigen-specific IL-17 response. The high position of TLR-4 in the hierarchy of erosive arthritis provides an interesting therapeutic target for RA.

Rheumatoid arthritis (RA) is a systemic autoimmune disease manifested by chronic inflammation and cartilage and bone destruction in multiple joints. Although the cause of RA remains unclear, it is known that a complex interplay between proinflammatory cytokines drives its progression. Insight into the crucial role of proinflammatory cytokines in RA has led to the development of new treatments, the biologic agents, which are intended to inhibit tumor necrosis factor (TNFα), interleukin-1 (IL-1), and IL-6. Other new therapies that inhibit B cell or T cell activation are being applied as well. Successful application of biologic agents has caused major improvements in clinical practice; however, a considerable number of patients (30–50%) remain unresponsive and some others become resistant to a particular drug after several years of responsiveness (1). Therefore, unraveling the mechanisms involved in the production of proinflammatory cytokines and B cell/T cell activation is highly desirable. The discovery of Toll-like receptors (TLRs) during the last decade has introduced new relevant mechanisms that are potentially involved in these processes.

TLRs are a family of pattern-recognition receptors that evolved to recognize conserved pathogen-associated molecular patterns (PAMPs) (2). TLR ligands are potent inducers of a variety of proinflammatory cytokines, including TNFα, IL-1, and IL-6, as well as the matrix metalloproteinases (MMPs). In addition, they promote the interaction between antigen-presenting cells and T cells through up-regulation of costimulatory molecules in the immunologic synapse (3, 4). It is known that direct TLR activation occurs in certain self-limiting types of arthritis, such as Lyme disease and Chlamydia-induced arthritis, and contributes to bacterial clearance (5, 6).

Bacterial and viral infections have also long been associated with the pathogenesis and the occurrence of flare reactions in RA, and this idea has been used in experimental research on arthritis. In this context, bacterial-derived TLR ligands such as double-stranded RNA, lipopolysaccharide (LPS) and CpG-containing DNA, which signal through TLR-3, TLR-4, and TLR-9, respectively, have frequently been used to provoke or accelerate experimental arthritis (7–9). LPS circumvents the IL-1 dependence of the K/BxN serum–transfer model of arthritis (10) and intraarticular injection of cell wall fragments of Streptococcus pyogenes has recently been reported to induce arthritis through the TLR-2/myeloid differentiation factor 88 (MyD88) pathway (11).

TLR activation is also involved in the chronic phase of other experimental models of arthritis in which PAMPs are not directly applied. It has recently been demonstrated that specific inhibition of TLR-4 using a naturally occurring antagonist strongly suppresses clinical as well as histopathologic features of collagen-induced arthritis and arthritis in the IL-1Ra−/− mouse (12, 13). Furthermore, TLR-4 gene deficiency protects IL-1Ra−/− mice from severe arthritis by reducing the number of pathogenic Th17 cells and the production of IL-17 (14). The contribution of TLRs to the severity of arthritis in these models might be explained by their ability to recognize endogenous ligands released from stressed cells or damaged tissue, such as heat-shock proteins, high mobility group box chromosomal protein 1 (HMGB-1), and breakdown products of heparan sulfate and hyaluronic acid. These ligands predominantly activate TLR-2, TLR-4, or both (15–18). By means of recognition of endogenous ligands, TLRs alert the immune system in case of injury and play a critical role in tissue repair and clearance of cellular debris (16, 19); however, failure to appropriately regulate the TLR response to these self antigens might contribute to autoimmunity in the context of certain environmental or genetic factors. For example, systemic lupus erythematosus–associated autoantigens containing self RNA or DNA have been shown to induce autoantibody production by B cells through sequential engagement of B cell receptor together with TLR-7 or TLR-9, respectively (20, 21).

In the context of RA, endogenous TLR ligands have repeatedly been found in the joints or serum of RA patients, and their levels correlate positively with disease activity scores (17, 22–25). In addition, anti–HMGB-1 autoantibodies have been detected in the serum of patients with juvenile rheumatoid arthritis (26). The clinical relevance of TLR activation in RA is supported by enhanced expression of TLRs 2, 3, 4, and 7 in the synovial lining and elevated TLR-2 expression in CD16+ peripheral blood monocytes and synovial macrophages from RA patients (25, 27–29). Furthermore, dendritic cells from RA patients produce significantly more proinflammatory cytokines upon TLR-2 and TLR-4 stimulation, but not TLR-3 and TLR-7 stimulation, as compared with control cells (25). There is increasing evidence that endogenous TLR-2 and TLR-4 ligands that are present in arthritic joints contribute to the spontaneous production of cytokines, chemokines, and MMPs by RA synovial tissue (14, 30). Taken together, our current knowledge indicates that microbial as well as endogenous activation of TLRs may contribute to the innate and adaptive immune responses during RA.

The aim of the present study was to investigate the involvement of TLR-2 and TLR-4 in the development of chronic streptococcal cell wall (SCW) arthritis as a model induced by PAMP-driven macrophage activation that ultimately leads to an antigen-specific adaptive immune response. In this model, the TNF-dependent macrophage-driven process during the acute phase turns into an IL-1/IL-17–dependent T cell–driven process during the chronic phase (31, 32). Here, we demonstrate a shift from TLR-2 dependency in the acute phase of arthritis toward TLR-4 dependency in the chronic phase, manifested by reduced expression of mediators and markers of cartilage and bone destruction and a lower antigen-specific IL-17 response in TLR-4−/− mice.



Male C57BL/6 mice were purchased from Janvier (Le Genest St. Isle, France). TLR-2−/− and TLR-4−/− mice on the C57BL/6 background were kindly provided by Prof. S. Akira (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan). The mice were housed in filter-top cages, and water and food were provided ad libitum. Sex-matched animals ages 10–12 weeks were used in all experiments. The animal studies were approved by the Institutional Review Board and were performed according to the relevant codes of practice.

Preparation of SCW fragments and induction of chronic SCW arthritis.

Streptococcus pyogenes T12 organisms were cultured overnight in Todd-Hewitt broth. Cell walls were prepared as described previously (33). The resulting supernatant obtained after centrifugation at 10,000g contained 11% muramic acid. Unilateral arthritis was induced by intraarticular injection of 25 μg of SCW fragments (rhamnose content) in 6 μl of phosphate buffered saline (PBS) into the right knee joint of naive mice. To induce chronic SCW arthritis, 4 intraarticular injections were performed, 1 each on days 0, 7, 14, and 21.

Measurement of joint swelling.

Joint swelling was assessed by measuring the accumulation of 99mTc in the inflamed joint due to increased blood flow and edema. To this end, 0.74 MBq of 99mTc in 200 μl of saline was injected subcutaneously, and after several minutes of distribution throughout the body, external gamma radiation in the knee joints was measured. Swelling is expressed as the ratio of gamma counts in the right (inflamed) knee joint to gamma counts in the left (control) knee joint. Values higher than 1.1 counts per minute were considered to represent inflammation.

Isolation of RNA from synovial biopsy tissues and patellar cartilage.

Synovial biopsy samples from the knee joints were isolated from the lateral and medial sides of patellae using a 3-mm punch (Stiefel, Wächtersbach, Germany) at the indicated time points. Six biopsy samples from 3 mice (2 from each mouse) were pooled to yield 2 samples per experimental group. Samples were stored in liquid nitrogen until RNA isolation. Patellae were decalcified overnight at 4°C in 5% EDTA, after which cartilage was stripped under a dissection microscope. Total RNA was isolated in 1 ml of TRIzol reagent (Sigma, St. Louis, MO), then precipitated with isopropanol, washed with 70% ethanol, and dissolved in water. RNA was treated with DNase and, subsequently, was reverse transcribed into complementary DNA using oligo(dT) primers and Moloney murine leukemia virus reverse transcriptase.

Real-time quantitative polymerase chain reaction (PCR).

Real-time quantitative PCR was performed using an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) for quantification with SYBR Green and melting curve analysis. Primer sequences (forward and reverse, respectively) were as follows: for GAPDH (housekeeping gene), 5′-GGC-AAA-TTC-AAC-GGC-ACA-3′ and 5′-GTT-AGT-GGG-GTC-TCG-CTC-TG-3′; for TLR-2, 5′-AAC-CTC-AGA-CAA-AGC-GTC-AAA-TC-3′ and 5′-ACC-AAG-ATC-CAG-AAG-AGC-CAA-A-3′; for TLR-4, 5′-TTC-CTT-CTT-CAA-CCA-AGA-ACA-TAG-ATC-3′ and 5′-TTG-TTT-CAA-TTT-CAC-ACC-TGG-ATA-A-3′; for MyD88, 5′-GTG-GCC-AGA-GTG-GAA-AGC-A-3′ and 5′-AAG-TTC-CGG-CGT-TTG-TCC-TA-3′; for TRIF, 5′-TTC-TCA-AGA-TTC-AGT-AAG-GAG-CAG-TAA-T-3′ and 5′-TAG-GAT-GCC-CAG-AAG-AAC-TTG-TAT-C-3′; for IL-1β, 5′-GGA-CAG-AAT-ATC-AAC-CAA-CAA-GTG-ATA-3′ and 5′-GTG-TGC-CGT-CTT-TCA-TTA-CAC-AG-3′; for TNFα, 5′-CAG-ACC-CTC-ACA-CTC-AGA-TCA-TCT-3′ and 5′-CCT-CCA-CTT-GGT-GGT-TTG-CTA-3′; for matrix metallo- proteinase 3 (MMP-3), 5′-TGG-AGC-TGA-TGC-ATA-AGC-CC-3′ and 5′-TGA-AGC-CAC-CAA-CAT-CAG-GA-3′; for MMP-13, 5′-AGA-CCT-TGT-GTT-TGC-AGA-GCA-CTA-C-3′ and 5′-CTT-CAG-GAT-TCC-CGC-AAG-AG-3′; for inducible nitric oxide synthase (iNOS), 5′-GGG-CAG-CCT-GTG-AGA-CCT-T-3′ and 5′-CGT-TTC-GGG-ATC-TGA-ATG-TGA-3′; for IL-6, 5′-CAA-GTC-GGA-GGC-TTA-ATT-ACA-CAT-G-3′ and 5′-ATT-GCC-ATT-GCA-CAA-CTC-TTT-TCT-3′; for IL-23 p19, 5′-CCA-GCG-GGA-CAT-ATG-AAT-CTA-CT-3′ and 5′-CTT-GTG-GGT-CAC-AAC-CAT-CTT-C-3′; for IL-17A, 5′-CAG-GAC-GCG-CAA-ACA-TGA-3′ and 5′-GCA-ACA-GCA-TCA-GAG-ACA-CAG- AT-3′.

PCR conditions were 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, with data collection during the last 30 seconds. For all PCRs, SYBR Green Master Mix was used in the reaction. Primer concentrations were 300 nM. The threshold cycle (Ct) value of the gene of interest was corrected for the Ct of the reference gene GAPDH to obtain the ΔCt, then the ΔΔCt was calculated in comparison with the levels in naive or wild-type (WT) mice. Quantitative PCR analysis of each sample was performed in duplicate, and melting curves were run for each PCR.

Assessment of inflammation.

Joint inflammation was scored both macroscopically and histologically by 2 observers (SA-R and either LABJ, MMH, or BW) in a blinded manner using a 0–3-point scale. For histologic analysis, total knee joints were isolated on day 28 of chronic SCW arthritis, fixed for 4 days in 4% formaldehyde, decalcified in 5% formic acid, and embedded in paraffin. Tissue sections (7 μm) were stained with hematoxylin and eosin.

Stimulation of lymphocytes, preparation of patella washouts, and measurement of cytokines.

Spleens were isolated and disrupted, and erythrocytes were lysed in 0.16M NH4Cl (pH 7.2). The cell suspension was washed with saline and enriched for lymphocytes by allowing antigen-presenting cells to adhere to plastic culture flasks for 45 minutes. Cells (2 × 105/well) were cultured for 72 hours at 37°C in an atmosphere of 5% CO2, in RPMI 1640 (Gibco-Invitrogen, Paisley, UK) supplemented with 5% fetal calf serum, 1 mM pyruvate, and 50 mg/liter of gentamicin in the presence of SCW fragments (6 μg/ml) or plate-coated anti-CD3 (2 μg/ml; R&D Systems, Abingdon, UK)/soluble anti-CD28 (2 μg/ml; BD Biosciences, Oxford, UK). Patella washouts were prepared by culturing patellae with the surrounding tissue in RPMI 1640 containing 0.1% bovine serum albumin (BSA) for 1 hour at room temperature. Cytokine concentrations in cell culture supernatants, patella washouts, and sera were determined using Bioplex cytokine assays from Bio-Rad (Hercules, CA) according to the manufacturer's instructions.


Local expression of IL-1β and cathepsin K was evaluated on paraffin sections of the knee joints on day 28. Sections were deparaffinized in xylol and rehydrated in serial dilutions of ethanol. Endogenous peroxidase was blocked using 1% hydrogen peroxide for 15 minutes. Sections were incubated for 1 hour with rabbit anti-mouse IL-1β (7.5 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA) or cathepsin K (200 μg/ml; a kind gift of Dr. E. Sakai, Nagasaki University School of Dentistry, Nagasaki, Japan) antibodies or normal rabbit IgG (Santa Cruz Biotechnology) and then were incubated with biotinylated swine anti-rabbit antibodies and peroxidase-labeled streptavidin. Color was developed with diaminobenzidine, and tissues were counterstained with hematoxylin.

IL-1β expression on articular chondrocytes (scored 0–2) and on synovial tissue around the patella, tibia, and femur (each scored 0–2, then averaged to obtain overall expression in synovium) was scored. Cathepsin K expression was evaluated by scoring the area of stained multinuclear cells along the bone surface (0–2-point scale for each region) of the patella, femur, and tibia. Total expression in these bones is illustrated below.

Irreversible proteoglycan damage by MMP activity in cartilage was assessed by immunostaining for the neoepitope VDIPEN, as described previously (34).

Determination of chondrocyte proteoglycan synthesis.

Patellae with minimal surrounding tissue were isolated from C57BL/6 WT mice and collected in RPMI 1640 containing Glutamax, penicillin/streptomycin (100 IU/100 μg/ml), and recombinant insulin growth factor (250 ng/ml; PeproTech, Rocky Hill, NJ). Patellae were incubated for 24 hours with IL-1β (10 ng/ml; R&D Systems), SCW fragments (10 μg/ml), and the TLR-4 ligand LPS (1 μg/ml; Sigma), with or without IL-1 receptor antagonist (IL-1Ra) (10 μg/ml; Amgen, Thousand Oaks, CA), followed by a 3-hour incubation with 35S (0.74 MBq/ml) at 37°C in an atmosphere of 5% CO2. Patellae were washed with saline, fixed in 4% formaldehyde, and decalcified in 5% formic acid for 4 hours, then punched out of the adjacent tissue and dissolved in 0.5 ml of LumaSolve (Omnilabo, Breda, The Netherlands) at 65°C. The 35S content was measured by liquid scintillation counting after the addition of 10 ml of Lipoluma (Omnilabo). Values are presented as the percentage of 35S incorporation as compared with that in the medium control.

Measurement of anti-SCW antibodies.

Levels of anti-SCW antibodies in sera (day 28) were analyzed by enzyme-linked immunosorbent assay. Briefly, 10 ng of SCW fragments was coated onto 96-well plates overnight. Plates were washed, and nonspecific binding sites were blocked with 1% BSA in PBS–Tween 80 (0.05%). Serial 2× dilutions of sera, starting with an initial dilution of 20×, were incubated in the plates for 1 hour. The plates were then washed, and isotype-specific horseradish peroxidase–labeled goat anti-mouse Ig (1:1,000) was added for 1 hour at room temperature; 5-aminosalicylic acid was used as substrate. Absorbance was measured at 450 nm.

Statistical analysis.

Group measures are expressed as the mean ± SEM. Statistical significance was assessed using the Mann-Whitney U test to compare 2 experimental groups or the Kruskal-Wallis test to compare 3 groups and was performed using GraphPad Prism 4.0 software (GraphPad Software, San Diego, CA). P values less than or equal to 0.05 were considered significant.


Sustained up-regulation of TLR-2 and TLR-4 expression in synovial tissue during reactivation of SCW arthritis.

Expression of messenger RNA (mRNA) for TLR-2, TLR-4, MyD88, and TRIF was determined in synovial biopsy samples from the knee joints collected at various time points during acute or chronic SCW arthritis and compared with that in synovial biopsy samples from naive mice. Quantitative PCR analysis showed that a single injection of SCW fragments into the knee joints resulted in up-regulation of mRNA for both TLR-2 and TLR-4 (Figure 1). This mRNA level decreased slightly over time up to day 7, but was strongly up-regulated by every SCW injection during the reactivation phase. The expression of TLR-2 and TLR-4 mRNA during the chronic phase (day 28) remained as high as that during the acute phase. The main TLR adaptor molecules MyD88 and TRIF were also up-regulated and exhibited an expression pattern similar to that of TLR-2 and TLR-4, although TRIF was less regulated compared with MyD88. Figure 1 shows that repeated injection of SCW fragments into the joints did not lead to a compensatory down-regulation of mRNA transcript expression of TLRs and their adaptor molecules in the synovium.

Figure 1.

Up-regulation of mRNA for Toll-like receptor 2 (TLR-2), TLR-4, myeloid differentiation factor 88 (MyD88), and TRIF in synovial biopsy samples from the knee joints of wild-type mice during the chronic phase of streptococcal cell wall (SCW)–induced arthritis. Expression levels were measured by quantitative real-time polymerase chain reaction analysis. The threshold cycle (Ct) value of the gene of interest was corrected for the Ct of the housekeeping gene GAPDH to obtain the ΔCt, then the ΔΔCt was calculated in comparison with levels in the joints of naive mice.

Shift from TLR-2 dependency toward TLR-4 dependency of chronic SCW arthritis during the chronic phase.

Acute SCW arthritis has previously been shown to be dependent on TLR-2 and MyD88 signaling (11). We investigated the TLR-2 and TLR-4 dependency of chronic SCW arthritis by administering 4 injections (1 each week) of SCW fragments to TLR-2−/− and TLR-4−/− mice. As we expected, TLR-2−/− mice showed significantly less joint swelling during the acute phase of arthritis, i.e., after the first 2 SCW injections (Figure 2A). During this phase, TLR-4 deficiency had no influence on the severity of joint swelling. Interestingly, the chronic phase of arthritis became independent of TLR-2 and, instead, dependent on TLR-4 activation, since TLR-4−/− mice clearly had less severe joint swelling after the final 2 injections (Figure 2A). Significant suppression of joint swelling in TLR-4−/− mice was sustained until day 28. Accordingly, the macroscopic inflammation score in the knee joints was significantly reduced in TLR-2−/− and TLR-4−/− mice on day 22, but was significantly reduced in only TLR-4−/− mice on day 28 (Figure 2B). Histologic examination of the knee joints revealed less synovial inflammation in both the TLR-2−/− and the TLR-4−/− mice (Figures 2C and D).

Figure 2.

A, The Toll-like receptor 2 (TLR-2) dependency of joint swelling during the acute phase of streptococcal cell wall (SCW)–induced arthritis shifted to TLR-4 dependency during the chronic phase, as measured by 99mTc uptake in the right (swollen) joint compared with the left (control) joint of wild-type (WT), TLR-2−/−, and TLR-4−/− mice. Underlined days are the days on which SCW fragments were injected. B, Reduction in the macroscopic inflammation score in the knee joints of both TLR-2−/− and TLR-4−/− mice as compared with WT mice 1 day after the last injection of SCW fragments. This reduction was sustained until day 28 in only the TLR-4−/− mice. C and D, Suppression of microscopic synovial inflammation, as determined histologically, in the joints of TLR-2−/− and TLR-4−/− mice as compared with WT mice on day 28. Representative images of hematoxylin and eosin–stained knee joint sections are shown in D. JS = joint space; P = patella; S = synovium; F = femur. (Original magnification × 50.) Values in A–C are the mean and SEM of at least 6 mice per group. = P < 0.05; ∗∗ = P < 0.01 versus WT mice, by Mann-Whitney U test.

Diminished expression of markers of early cartilage and bone destruction in the TLR-4−/− mouse joint.

TLR-4 activation has been shown to play a critical role in driving cartilage and bone destruction during the chronic phase of disease in experimental models of arthritis (12, 14). Therefore, we examined the expression of markers of early-stage cartilage and bone destruction on day 28 using immunohistochemistry. Staining of the knee joints revealed the expression of the MMP-specific aggrecan neoepitope VDIPEN in cartilage at patellofemoral junction (data not shown) as well as the femorotibial junction (Figure 3A). The osteoclast marker cathepsin K was also highly expressed in cells adjacent to the bone surface through the entire joint (Figure 3C). Quantitative analysis revealed that expression of both VDIPEN and cathepsin K was significantly reduced in TLR-4−/−, but not TLR-2−/−, mouse knee joints on day 28 as compared with the WT animals (Figures 3B and D). Attenuation of joint inflammation and cartilage and bone destruction in TLR-4−/− mouse joints led us to focus on the expression of molecules involved in these processes, including proinflammatory cytokines and MMPs, as potential underlying mechanisms.

Figure 3.

A, Representative images of knee joint sections from wild-type (WT), TLR-2−/−, and TLR-4−/− mice, showing immunohistochemical staining for VDIPEN, the aggrecan neoepitope expressed upon matrix metalloproteinase–specific cartilage breakdown. F = femur; T = tibia. (Original magnification × 100.) B, Quantitative measurement of VDIPEN expression (percentage of positively stained area) in cartilage from the 3 groups of mice. Expression was scored using Leica Qwin software (Leica Microsystems, Rijswijk, The Netherlands). Horizontal bars show the mean. = P < 0.05 versus WT mice, by Mann-Whitney U test. C, Representative images of knee joint sections from the 3 groups of mice, showing immunohistochemical staining for cathepsin K, the osteoclast marker involved in bone resorption. Positive cells can be seen along the femur (F). S = synovium. (Original magnification × 100.) D, Scores for cathepsin K expression along different bones in the knee joints of the 3 groups of mice on day 28. Horizontal bars show the mean. ∗∗ = P < 0.01 versus WT mice, by Mann-Whitney U test.

Reduced local mRNA expression of proinflammatory genes in the TLR-4−/− mouse joint.

Determination of the expression of several proinflammatory genes by quantitative PCR analysis revealed diminished expression of IL-1β, TNFα, MMP-3, and MMP-13 in patellar cartilage from both TLR-2−/− and TLR-4−/− mice as compared with WT mice on day 28 (Figure 4A). Furthermore, the expression of iNOS, an important mediator of inflammation and cartilage degradation, was strongly decreased in the cartilage of TLR-4−/− mice, while showing only a slight reduction in TLR-2−/− animals (Figure 4A). Quantitative PCR analysis of synovial tissue showed lower levels of expression of IL-1β, TNFα, IL-6, and in particular, IL-23p19 and IL-17, which are predominantly expressed during the chronic phase of SCW arthritis, in the synovium of TLR-4−/− mice, whereas synovium from TLR-2−/− mice showed almost no difference compared with synovium from WT mice (Figure 4B).

Figure 4.

A, Expression of mRNA for inflammatory genes involved in cartilage degradation and VDIPEN expression in patellar cartilage from TLR-2−/− and TLR-4−/− mice on day 28 of chronic streptococcal cell wall (SCW)–induced arthritis. B, Expression of mRNA for inflammatory cytokines in synovial tissue from TLR-2−/− and TLR-4−/− mice on day 28 of chronic SCW-induced arthritis. Expression levels were measured by quantitative real-time polymerase chain reaction analysis. The threshold cycle (Ct) value of the gene of interest was corrected for the Ct of the housekeeping gene GAPDH to obtain the ΔCt, then the ΔΔCt was calculated in comparison with levels in the joints of wild-type (WT) mice on day 28 of arthritis. IL-1β = interleukin-1β; TNFα = tumor necrosis factor α; MMP-3 = matrix metalloproteinase 3; iNOS = inducible nitric oxide synthase.

Reduced levels of proinflammatory cytokine proteins in TLR-4−/− mice and IL-1–mediated effects of TLR stimulation on chondrocyte function.

Assessment of local concentrations of inflammatory cytokines showed a significant reduction in IL-1β, TNFα, and IL-6 concentrations in patella washouts from TLR-4−/− mice compared with WT and TLR-2−/− mice, confirming the involvement of TLR-4 activation during the chronic phase of disease (Figure 5A). Furthermore, systemic levels of IL-1β and IL-6 were significantly lower in serum samples from TLR-4−/− mice compared with WT and TLR-2−/− mice on day 28 (Figure 5B). Immunohistochemical staining of the knee joints revealed high levels of IL-1β expression in both articular cartilage and synovial tissue from WT and TLR-2−/− mice. Consistent with the IL-1β concentrations in patella washouts, TLR-4−/− mice showed clearly diminished expression of IL-1β in both cartilage and synovium (Figure 5C).

Figure 5.

A and B, Concentrations of cytokines that mediate joint inflammation and destruction in patella washouts obtained on day 22 (A) and in sera obtained on day 28 (B) from wild-type (WT), TLR-2−/−, and TLR-4−/− mice, as measured by Luminex bead array analysis. Concentrations of interleukin-1β (IL-1β), tumor necrosis factor α (TNFα), and IL-6 were significantly reduced in TLR-4−/− mice compared with WT mice. TLR-2−/− mice had similar concentrations of cytokines as the WT mice. C, IL-1β expression in synovium and articular chondrocytes from the 3 groups of mice. Expression of IL-1β was reduced in both synovial tissue and articular chondrocytes from TLR-4−/− mice as compared with WT mice. Horizontal bars show the mean. D, Chondrocyte proteoglycan (PG) synthesis in patellar cartilage explants from C57BL/6 mice. Chondrocyte PG synthesis was determined in patellar cartilage incubated for 24 hours with medium, IL-1 (10 ng/ml), lipopolysaccharide (LPS; 1 μg/ml), or streptococcal cell wall (SCW; 10 μg/ml) fragments, with or without IL-1 receptor antagonist (IL-1Ra; 10 μg/ml). Values are the percentage of 35S incorporation in synthesized PGs compared with medium control. Values in A, B, and D are the mean and SEM of at least 6 mice per group. = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001 versus WT controls or versus medium alone, by Mann-Whitney U test.

We have previously shown that TLR-4 inhibition exerts protective effects on cartilage in the collagen-induced arthritis model through down-regulation of IL-1 (12). Therefore, we investigated whether TLR stimulation is capable of affecting the anabolic function (i.e., proteoglycan synthesis) of chondrocytes and whether IL-1 plays a role in this process. To this end, patellar cartilage explants from C57BL/6 mice were incubated with IL-1β, SCW fragments, and the TLR-4 ligand LPS in combination with IL-1 receptor antagonist (IL-1Ra). Both SCW and LPS were able to inhibit proteoglycan synthesis by chondrocytes to the same extent as recombinant IL-1 (Figure 5D). Addition of IL-1Ra restored the function of chondrocytes to normal, indicating that the inhibitory effect of TLR ligands on chondrocyte function is mediated through IL-1 (Figure 5D).

Dependence of the antigen-specific adaptive immune response on TLR-4 during chronic SCW arthritis.

The antigen-specific antibody response in WT mice was developed after the fourth intraarticular injection of SCW fragments (Figure 6A). Anti-SCW antibodies of IgG1, IgG2b, and IgG3, but not IgG2a, classes were detected in mouse sera (Figure 6A). Although the adaptive B cell response was comparable between WT and TLR-2−/− mice, TLR-4−/− mice had lower levels of anti-SCW–specific IgG1 and IgG3 antibodies in their sera on day 28 (Figure 6B).

Figure 6.

A, Kinetics of the anti–streptococcal cell wall (anti-SCW)–specific antibody response during chronic SCW arthritis in wild-type (WT) mice, as determined by enzyme-linked immunosorbent assay. Mice were injected on days 0, 7, 14, and 21, and antibodies were determined at the indicated time points. Antibodies were detectable after the fourth injection of SCW fragments. B, Serum levels of anti-SCW–specific IgG1 and IgG3 antibodies on day 28 in WT, TLR-2−/−, and TLR-4−/− mice. Levels were significantly lower in sera from TLR-4−/− mice as compared with WT and TLR-2−/− mice. C and D, Production of interleukin-17 (IL-17) (C) and interferon-γ (IFNγ) (D) by nonadherent splenocytes obtained on day 28 from the 3 groups of mice. IL-17 and IFNγ production was measured after pan–T cell stimulation of splenocytes with anti-CD3/anti-CD28 (both at 2 μg/ml) or after antigen-specific stimulation with SCW fragments (6 μg/ml) for 72 hours. IL-17, but not IFNγ, production by nonadherent splenocytes from TLR-4−/− mice was significantly reduced upon SCW-specific stimulation. Values are the mean and SEM. = P < 0.05 for the indicated comparisons, by Kruskal-Wallis test. NS = not significant.

With regard to the anti-SCW T cell response, T cells from the spleens of WT and TLR-deficient mice proliferated to a similar extent upon stimulation with anti-CD3 or anti-SCW antibodies, indicating an unaltered proliferative response of T cells (data not shown). The T cell cytokine IL-17 has been implicated in cartilage destruction during chronic SCW arthritis (32), and TLR-4 activation has been shown to drive IL-17 production during experimental arthritis (14). Since in the present study, the expression of a number of cytokines closely related to the development and survival of Th17 cells was found to be reduced in TLR-4−/− mice (Figure 4B), we analyzed the T cell cytokine response to nonspecific, as well as antigen-specific, stimulation. Pan–T cell stimulation (anti-CD3/anti-CD28) of splenic T cells isolated on day 28 of chronic arthritis resulted in a high level of production of both interferon-γ (IFNγ) and IL-17 (Figures 6C and D), indicating the presence of systemic Th1 as well as the Th17 responses; however, the antigen specificity of these cells differed strikingly, in that SCW-specific stimulation resulted in the production of high concentrations of IL-17, but very low amounts of IFNγ (Figures 6C and D). Of great interest is our finding that T cells from TLR-4−/− mice produced markedly less IL-17 upon SCW-specific stimulation as compared with cells from WT and TLR-2−/− mice (Figure 6C). IFNγ production was not significantly different among the groups (Figure 6D). These data indicate the development of antigen-specific IL-17 production during the chronic phase of SCW arthritis, a process that appeared to be controlled by TLR-4 activation.


As crucial receptors in the initiation of the innate immune response and in the instruction of the adaptive immune response, TLRs may control multiple features of the immunopathology of RA. Therefore, study of the exact role of TLRs in various phases of arthritis is a matter of considerable interest and may lead to novel therapeutic approaches that complement existing therapies. In the present study, we examined the role of TLR-2 and TLR-4 in joint inflammation and erosive processes in chronic SCW-induced arthritis. This experimental model reproduces the repeated flare reactions of RA, a characteristic likely to be driven by the activation of TLRs upon infection. The acute and the chronic phase of SCW-induced arthritis reflect 2 different pathologic processes. The acute phase represents an innate immune response to SCW fragments that is driven by direct activation of macrophages via TLR-2 and another pattern-recognition receptor, nucleotide-binding oligomerization domain 2 (11, 35). This process is gradually replaced by the adaptive immune response during the chronic phase, as the SCW-specific B cell and T cell responses develop.

In the present study, we demonstrated that synovial infiltration upon each flare of arthritis is dependent on TLR-2; however, the erosive processes in the joint are independent of TLR-2, being dependent instead on TLR-4 during the chronic phase, when B cells and T cells become involved. Although joint swelling and expression of cytokines and markers of cartilage and bone destruction were unaffected in TLR-2−/− mice, synovial inflammation was still significantly reduced in these mice. This might be explained by the role of TLR-2 in the induction of chemokines, such as keratinocyte-derived chemokine and macrophage inflammatory protein 1α, by SCW fragments in vivo (11). In contrast, cartilage and bone erosion does not necessarily correspond to the severity of inflammation, since these processes are principally driven by other factors, such as IL-1, IL-17, and MMPs. The chronic phase of SCW arthritis has previously been shown to be dependent on the T cell cytokine IL-17, which is responsible for the expression of IL-1 and some MMPs, and for irreversible cartilage destruction, among other pathologic changes (31, 32).

The present data demonstrate that the SCW-specific antibody and IL-17 responses are partly dependent on TLR-4 activation (Figure 6). Features of cartilage and bone destruction are correspondingly dependent on TLR-4 (Figure 3). In this context, the shift from TLR-2 dependency toward TLR-4 dependency is consistent with the shift from innate immune involvement toward adaptive immune involvement. This shift was not due to a lack of TLR-2 expression or responsiveness after repeated exposure to TLR-2 ligands. The observation that TLR-2−/− mice continued to develop chronic arthritis indicates that the adaptive immune responses against SCW fragments are independent of TLR-2, as confirmed by the antibody and IL-17 measurements. Still, this does not exclude a role of TLR-2 in certain stages of TLR-2–dependent types of reactive (infectious) arthritis, such as Lyme disease (5).

During the chronic phase of arthritis, TLR-4 actively contributes to MMP-mediated cartilage destruction, as evidenced by VDIPEN neoepitope expression, and to osteoclast formation and activation, as manifested by cathepsin K expression (Figure 3). The cysteine protease cathepsin K plays an important role in osteoclast-mediated bone resorption under physiologic as well as pathologic conditions (36); however, its expression by RA synovial fibroblasts and its critical role in cartilage degradation have also been reported (37). Diminished cathepsin K expression in TLR-4−/− mice is consistent with a recent report on the osteoclastogenic capacity of TLRs, in which TLR-2 and, especially, TLR-4 stimulation of RA synovial fibroblasts was shown to promote the differentiation of cocultured monocytes into cathepsin K–expressing osteoclasts (38).

TLR-4−/− mice also expressed substantially lower levels of iNOS in the patellar cartilage (Figure 4). Inducible nitric oxide synthase is produced not only by chondrocytes, but also by synovial fibroblasts and macrophages, upon stimulation with LPS as well as proinflammatory cytokines such as IL-1 and TNF (39), and it catalyzes the inducible (inflammation-related) pathway of nitric oxide (NO) production. There is considerable evidence implicating NO in joint inflammation and cartilage degradation in RA (40). TLR-2 and TLR-4 strongly induce the production of NO and MMPs in first-passage osteoarthritic chondrocytes, and promote the degradation of proteoglycans and type II collagen in osteoarthritic cartilage explants (41). Since TLR-2 and TLR-4 mRNA are expressed at low levels on naive patellar chondrocytes and is up-regulated during the SCW arthritis (data not shown), the reduction in iNOS expression in cartilage from the TLR-4−/− mice might be a direct effect of TLR-4 activation on chondrocytes, or it may reflect the reduced inflammation in these mice. Since NO is also involved in the activation of MMPs (39), reduced iNOS expression might have contributed to lower levels of VDIPEN expression in TLR-4−/− mice.

Further studies of the mechanisms underlying reduced cartilage and bone destruction revealed that TLR-4 controls the production of IL-1β, TNFα, IL-6, and IL-23p19 (Figures 4 and 5). These cytokines fulfill multiple, and to some extent differential, functions in the arthritis process. Although both TLR-2−/− and TLR-4−/− mice showed reductions in IL-1β mRNA expression, only TLR-4−/− mice had a significant reduction in the IL-1β protein concentrations in patella washouts and sera. This might be related to differences in IL-1β processing by caspase 1 under the influence of TLR-4 activation. IL-1 has previously been described to play a dominant role in catabolic events in chondrocytes and to drive cartilage degradation ex vivo (42, 43). In SCW arthritis, IL-1β drives joint inflammation and cartilage destruction during both the acute and the chronic phases, whereas TNFα is mainly involved in joint swelling during the acute phase (31, 44). We found that SCW fragments do not contain bacterial TLR-4 ligands, since they were not able to activate TLR-4/myeloid differentiation protein 2/CD14 complex on HEK 293 cells (data not shown). Therefore, reduced levels of proinflammatory cytokines such as IL-1β and TNFα in the joints and sera of TLR-4−/− mice (Figure 5) indicate activation of TLR-4, probably by endogenous damage-associated ligands, during the latter phase of the disease. In this context, the shift from TLR-2 dependency toward TLR-4 dependency might also rely on the presence of the corresponding ligands in the inflamed joint.

IL-1β and TNFα also promote Th17 cell commitment and IL-17 production, a process principally driven by IL-6 and TGFβ (45). Furthermore, IL-1 participates in the induction of Th17-mediated autoimmune encephalomyelitis (46). IL-23, a factor involved in the survival and expansion of Th17 cells, is, like IL-17, up-regulated during the later stages of SCW arthritis and is involved in the chronic phase of the disease (31). Therefore, the remarkable suppression of IL-23 and IL-17 in TLR-4−/− mouse synovium on day 28 is relevant in terms of the contribution of TLR-4 to the chronicity of the disease.

IL-17 is highly pathogenic to cartilage and bone in various T cell–mediated experimental models of arthritis, including SCW arthritis, and has been shown to be able to take over the catabolic functions of IL-1 (32, 34). Importantly, recent evidence indicates that conditioned medium from TLR-4–activated dendritic cells is sufficient to induce Th17 differentiation when TGFβ is added (47). We have recently demonstrated the involvement of TLR-4 activation in the Th17/IL-17 pathway in a T cell–mediated autoimmune model of arthritis, the spontaneous IL-1rn−/− model (14). Since the exact antigen that drives the IL-1rn−/− arthritis is not known, the study of antigen specificity of Th17 cells and IL-17 production in the IL-1rn−/− model is difficult. We report here the production of high concentrations of IL-17 upon stimulation with SCW antigen. TLR-4−/− mice exhibited lower production of SCW-specific IL-17, indicating a role of TLR-4 in antigen-specific Th17 development. Since TLR-2 and TLR-4 deficiency has previously been reported not to affect the cellular immune response to antigens, a disruption of the adaptive immunity in these mice is unlikely (48). Furthermore, the normal proliferation of TLR-deficient T cells upon SCW stimulation in our experiments supports this point (data not shown). In the present study, although IFNγ-producing T cells were also present in the spleen, they did not exhibit SCW specificity. In this context, the cartilage and bone changes in chronic SCW arthritis are probably attributable to IL-17 rather than IFNγ.

A lack of TLR-4 signaling also led to a lower SCW-specific antibody response. Although B cell–intrinsic TLR signals are not required for antibody production or for B cell memory responses, coadministration of LPS enhances antigen-specific antibody production (49). LPS can also cause polyclonal B cell proliferation in the absence of antigen and trigger immunoglobulin secretion and IgG class switching (50). This might explain the lower amounts of anti-SCW IgG antibodies in the sera of TLR-4−/− mice.

It is of great interest that the attenuation of cartilage and bone destruction in chronic SCW arthritis was associated with the suppression of antigen-specific IL-17 and antibody responses. The exact nature of the TLR-4 ligands that contribute to these processes remains unclear; however, considering the essential roles of the cytokines that are regulated by TLR-4 activation in joint pathology, TLR-4 appears to be a promising target in the treatment of RA.


Dr. Abdollahi-Roodsaz 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. Abdollahi-Roodsaz, Joosten, Koenders, van den Berg.

Acquisition of data. Abdollahi-Roodsaz, Joosten, Helsen, Walgreen, van den Bersselaar.

Analysis and interpretation of data. Abdollahi-Roodsaz, Joosten, van Lent, Koenders, van den Berg.

Manuscript preparation. Abdollahi-Roodsaz, Joosten, van Lent, van den Berg.

Statistical analysis. Abdollahi-Roodsaz.


We are grateful to Prof. S. Akira (Osaka, Japan) for providing the TLR-2−/− and TLR-4−/− mice.