To investigate whether alarmins S100A8 and S100A9 are involved in mediating cartilage destruction during murine and human osteoarthritis (OA).
To investigate whether alarmins S100A8 and S100A9 are involved in mediating cartilage destruction during murine and human osteoarthritis (OA).
Two different murine models of OA that differed in terms of synovial activation were compared. Cartilage destruction was measured histologically. Synovial biopsy and serum samples from OA patients were derived from the Cohort Hip and Cohort Knee (CHECK) patients with symptomatic early OA. Expression of mediators in the synovium was measured by reverse transcription–polymerase chain reaction analysis and immunolocalization.
In collagenase-induced OA, which showed marked synovial activation, interleukin-1β was expressed at significant levels only during the early stages of disease, whereas S100A8 and S100A9 expression remained high for a prolonged period of time (up to day 21 after induction). In S100A9-knockout mice, we found a major impact of S100A8 and S100A9 on synovial activation (62% inhibition) and OA cartilage destruction (45–73% inhibition) as compared to wild-type controls. In contrast, in the surgically induced destabilized medial meniscus model, in which synovial involvement is scant, we found no role of S100A8 and S100A9 in the focal OA cartilage destruction. Examination of arthroscopic synovial biopsy samples from patients in the early symptomatic OA CHECK cohort revealed substantial levels of S100A8 and S100A9 messenger RNA and protein, which correlated significantly with synovial lining thickness, cellularity in the subintima, and joint destruction. Levels of S100A8/A9 serum protein were significantly enhanced (19%) at baseline in patients who had pronounced progression of joint destruction after 2 years.
Our data suggest that the S100A8 and S100A9 proteins are crucially involved in synovial activation and cartilage destruction during OA and that high levels may predict joint destruction in humans with OA.
Breakdown of the cartilage matrix is one of the hallmarks of osteoarthritis (OA). Cartilage destruction is predominantly mediated by cytokines and enzymes (1–4). During OA, a clear pericellular activation of chondrocytes is observed, which express large amounts of neoepitopes induced by metalloproteinases that drive pericellular breakdown of the matrix, eventually culminating in erosion (3). Pathologic changes that develop during OA are driven by at least 3 different tissues: cartilage, subchondral bone, and synovium (5, 6). Although OA has generally been described as a disease of the cartilage, the synovium is considered to contribute to the pathology during progression of the disease (7–11). A substantial group of OA patients (up to 50%) show activation of the synovium, not only during late phases, but also during early stages of the disease (8). In previous studies, we showed that activated synovial lining macrophages are crucial in regulating synovial inflammation and subsequent cartilage destruction during experimental OA. Selective elimination of synovial lining macrophages prior to induction of collagenase-induced OA diminished synovial activation and almost completely inhibited cartilage destruction (12).
Activated macrophages, which cover the inside of diarthrodial joints, produce a plethora of mediators, among them, cytokines such as interleukin-1β (IL-1β) (2). Although IL-1β has been considered important in mediating cartilage destruction (2, 13), its role in OA is still a matter of debate. The most prominent proteins released by activated macrophages are the calgranulins: myeloid-related protein 8 (MRP-8; also known as S100A8) and MRP-14 (also known as S100A9) (14, 15). These proteins belong to the group of damage-associated molecular patterns (DAMPs), which are crucial in innate immunity. Both proteins belong to the S100 family of calcium binding proteins, which comprises 24 members. They are expressed as homodimers and heterodimer assemblies. S100A8 is generally coexpressed with S100A9, and the heterodimer S100A8/A9 is translocated to membrane and cytoskeletal structures upon activation (16). In mice, S100A8 forms the active part, whereas S100A9 forms the regulating unit, which binds to S100A8 and thereby prevents its degradation. S100A8 stimulates macrophages via Toll-like receptor 4 (TLR-4) signaling (17, 18). When secreted, S100A8 exhibits proinflammatory functions, leading to activation of endothelial cells and phagocytes.
The homodimers S100A8 and S100A9 and the heterodimer S100A8/A9 accumulate in inflammatory fluids during arthritis (18, 19), and their levels correlate significantly with the severity of arthritis and were shown to predict a 10-year radiographic progression in RA patients (20). In previous studies, we found that S100A8/S100A9 is crucial in mediating cartilage destruction during experimental arthritis (21). S100A8 and S100A9 have recently also been described in experimental OA (22), and S100A8 appeared to be a potent stimulator of murine chondrocytes, thereby inducing a catabolic phenotype (23).
In the present study, we investigated whether these proteins are involved in synovial activation and cartilage destruction in experimental OA using 2 murine models that differ in the degree of synovial activation. In addition, we explored the clinical relevance of S100A8/A9 in samples of synovial biopsy tissues and sera from participants in the Cohort Hip and Cohort Knee (CHECK) early OA symptomatic cohort.
Experimental OA was elicited in male C57BL/6 mice and in S100A9–/– mice backcrossed to the C57BL/6 background for 12 generations. Myeloid cells of S100A9–/– mice also lack S100A8 at the protein level (24).
Collagenase osteoarthritis was induced by injecting 5 μg of bacterial collagenase (Sigma-Aldrich) 2 times on alternate days into knee joints of mice (C57BL/6), which causes disruption of the ligaments and local instability of the knee joint (25). Destabilized medial meniscus OA was induced by transection of the medial anterior meniscotibial ligament (26). Experimental arthritis was induced by injection of 25 μg of streptococcal cell wall (SCW) fragments into the murine knee joint (27).
Total knee joint sections of the mice were fixed, dehydrated, and embedded in paraffin. Sections (7 μm) were cut and stained with hematoxylin and eosin (H&E) or Safranin O. H&E staining was used to score synovial activation, which was identified as thickening of the intima lining layer (number of cell layers). Safranin O staining was used to examine and score OA-like cartilage changes; both the severity and the extent of the lesions were incorporated into the score. The scoring method described by Pritzker et al (28) was modified to make it more suitable for measuring pathologic changes in murine cartilage. Briefly, the OA score is the assessment of the combined OA grade and OA stage. The OA grade represents OA depth progression into cartilage and includes 6 grades: grades 1–4 involve cartilage changes only; grades 5 and 6 involve bone as well. The stage represents the horizontal extent of cartilage involvement irrespective of the underlying grade. A total of 4 stages are distinguished: stage 1 = <10%, 2 = 10% to <25%, 3 = 25–50%, and 4 >50%.
Whole knee joint sections were incubated for 2 hours with 10 mM citrate (pH 6.0) and then incubated for 1 hour with the primary antibody directed against S100A8 or S100A9 (made in our own facilities). Rabbit IgG antibody was used as a control. After rinsing, sections were incubated for 30 minutes with biotinylated horseradish peroxidase–conjugated goat anti-rabbit IgG (Dako). Diaminobenzidine was used to develop the peroxidase staining. Counterstaining was done with Mayer's hematoxylin (Biochemica).
At various time points after induction of OA, synovial specimens were isolated as described previously (29). Briefly, joint capsule specimens were isolated on the medial and lateral sides of the patella with a biopsy punch. Synovial specimens were snap-frozen in liquid nitrogen and stored for RNA isolation.
RNA was isolated from synovium using TRIzol reagent according to the manufacturer's protocol (Invitrogen). Five micrograms of total RNA was reverse transcribed, and complementary DNA aliquots were subjected to PCR. RT-PCR was normalized according to the transcription levels of GAPDH. Levels of messenger RNA (mRNA) for various members of the S100 family (S100B, A2, A4, A6, A7, A8, A9, and A11) and cytokines/chemokines (IL-1β, IL-6, tumor necrosis factor α [TNFα], IL-10, keratinocyte-derived chemokine (CXCL1), and macrophage inflammatory protein 1α [MIP-1α]) were measured in synovial specimens obtained at various time points after induction of collagenase-induced OA or destabilized medial meniscus OA and were quantified using an ABI Prism 7000 sequence detection system.
Synovial tissue was obtained from patients with OA (n = 10) who were undergoing arthroplasty. Furthermore biopsy samples were taken arthroscopically from CHECK cohort participants with knee pain (n = 19). Biopsy samples were taken from various areas of the synovium to minimize bias related to sampling of inflamed tissue rather than global sampling. Patients undergoing arthroplasty met the American College of Rheumatology classification criteria for OA (30).
CHECK is a prospective cohort study of 1,002 individuals with early symptomatic OA of the knee or hip (31). The CHECK cohort was initiated by the Dutch Arthritis Association and performed at Erasmus Medical Centre Rotterdam, Kennemer Gasthuis Haarlem, Leiden University Medical Centre, Maastricht University Medical Centre, Martini Hospital Groningen/Allied Health Care Centre for Rheumatology and Rehabilitation Groningen, Medical Spectrum Twente Enschede/Ziekenhuisgroep Twente Almelo, Reade/Jan van Breemen Research Institute and VU Medical Center Amsterdam, St. Maartenskliniek Nijmegen, University Medical Centre Utrecht, and Wilhelmina Hospital Assen.
On entry into the CHECK study, all participants had pain or stiffness of the knee or hip, and all of them were ages 45–65 years. They had not yet consulted their physician for these symptoms, or the first consultation occurred within 6 months before cohort entry. Participants with any other pathologic condition that could explain the symptoms were excluded (e.g., other rheumatic disease, previous hip or knee joint replacement, congenital dysplasia, osteochondritis dissecans, intraarticular fractures, septic arthritis, Perthes disease, ligament or meniscus damage, plica syndrome, Baker's cyst). Joint destruction was determined at baseline and 2 years after the start of the study, using the Kellgren/Lawrence system to score radiographs. The sum of the Kellgren/Lawrence score for both hips and knee joints was calculated. Two patient groups were defined, nonprogressors and progressors. Nonprogressors were those who showed no joint abnormalities (Kellgren/Lawrence score 0) at baseline or after 2 years. Progressors were those who showed no joint abnormalities at baseline, but after 2 years, did have joint damage (Kellgren/Lawrence score ≥1).
Tissues were fixed in 10% buffered formalin. Five-micron sections were stained with rabbit anti-human S100A8 or S100A9 (prepared in our own facilities) or with isotype control and then incubated with biotinylated rabbit anti-goat antibody (Dako). Sections were treated with streptavidin–peroxidase conjugate (Millipore), and color was visualized with hematoxylin.
Serum was kept at –20°C, and S100A8/A9 serum concentrations were determined by a sandwich enzyme-linked immunosorbent assay (ELISA) as described previously (32). For calibration, different amounts (0.25–250 ng/ml) of the native complex of human S100A8/A9 were used, which were isolated from human granulocytes. The assay has a sensitivity of <0.5 ng/ml and a linear range of 1–30 ng/ml. S100A8 and S100A9 form noncovalently associated complexes, which are detected by the sandwich ELISA. Therefore, the ELISA is calibrated with the native S100A8/A9 complex, and the data are expressed as ng/ml of S100A8/A9. The serum samples were coded, and the results represent the mean of duplicate determinations of each of 3 dilutions within the linear range.
All animal experiments were approved by the local ethics commission of the Radboud University Nijmegen Medical Centre (RU-DEC 2008-162). The human study was approved by the medical ethics committees of all participating centers, and all participants gave their written informed consent before entering the study.
Data were statistically evaluated using the Mann-Whitney U test and Spearman's correlation. Confounding by body mass index was excluded by multiple regression analysis.
In a considerable number of OA patients, synovial activation is observed, characterized by a thickened intima containing hypertrophic macrophages (33). We evaluated the course and degree of synovial activation at various time points after induction of 2 models of instability-related murine OA. In control murine knee joints, the synovial intima is only 1 layer thick, comprising type A macrophage-like and type B fibroblast-like cells (Figures 1B and C). In collagenase-induced OA, a clear thickening of the intima was seen (Figures 1D and E), mainly consisting of hypertrophic F4/80 vesicular type A macrophage-like cells (compare Figure 1H with control in Figure 1I). Thickening of the intima lining layer containing 3–12 layers was observed throughout the 42-day observation period, reaching its maximum on day 14 (Figure 1A). In contrast, in the destabilized medial meniscus model, synovial activation was much lower than in the collagenase-induced OA model and reached its maximum on day 7, expressing an intima layer of a maximum of 2–3 layers consisting of type A macrophages, which showed a less-pronounced hypertrophic character (Figures 1F and G).
We next investigated whether the synovium produces S100 proteins. RNA was isolated from synovial samples derived from collagenase-induced knee joints at several time points (7, 21, and 42 days after induction). Levels of mRNA for most S100 proteins were strongly up-regulated in the synovium of mice with collagenase-induced OA as compared to control synovium, with S100A8 and S100A9 being the most up-regulated. On days 7 and 21, expression of S100A8 was increased 24- and 32-fold, respectively, and for S100A9, 16- and 64-fold (Figure 2A). Interestingly, levels of mRNA for S100A9 increased during the course of OA up until day 21, whereas levels of mRNA for S100A8 remained high from day 7 to day 21. In contrast, all other members of the S100 family had returned to low levels on day 21.
In addition, mRNA levels of S100 family members were measured in the synovium during the course of destabilized medial meniscus OA, during which, less synovial activation was observed. In contrast to the levels found in collagenase-induced OA, destabilized medial meniscus OA showed only minor increases, except for S100A8 and S100A9 on day 7 (Figure 2B).
Expression of S100A8 and S100A9 protein in the synovium of mice from both OA models was further investigated by immunolocalization. Strong expression of S100A9 (and, to a lesser extent, S100A8 [data not shown]) was found in the intimal layers of collagenase-induced OA on day 7 (Figure 2C) and was still detectable on day 42 (Figure 2E). No staining of S100A8 or S100A9 was found in the intimal layer of OA knee joints of S100A9–/– mice on day 7 after induction (Figure 2D). In contrast to the prominent expression in knee joints of mice with collagenase-induced OA, in the destabilized medial meniscus OA model, expression of S100A8 and S100A9 was almost absent within the thin intimal lining layer on day 7 after induction (Figure 2F). Staining of S100A8 and S100A9 protein was negative in the synovial lining of normal mouse knee joints (data not shown).
Next, the expression of S100A8/A9 in the synovium of mice with collagenase-induced OA was compared with that in synovium from mice with an acutely inflamed joint. Arthritis was induced by injection of 25 μg of SCW fragments into the mouse knee joint. Total knee joint sections showed that shortly after injection, the cellularity of the synovium was significantly higher in mice with SCW-induced arthritis as compared to those with OA (data not shown). On day 7 after induction, the thickness of the intimal layer was comparable between the 2 models (Figures 3A and B). In knee joints with SCW-induced arthritis, S100A8/A9 mRNA levels were high shortly after injection but rapidly normalized and were no longer detected by day 7 after injection (Figure 3C). In contrast, in the OA model, although showing comparable intimal thickness on day 7, S100A8/A9 levels remained high and even increased between days 7 and 21. This suggests that in this OA model, synovial intimal lining cells are activated for prolonged periods of time, probably through factors released from the affected cartilage during the course of OA.
Apart from S100A8/A9, we determined the expression of other proinflammatory cytokines/chemokines that might be involved in driving synovial activation. Proinflammatory cytokines/chemokines were hardly expressed during collagenase-induced OA (IL-6, TNFα, MIP-1α, and KC increased 16-, 11-, 7-, and 5-fold, respectively), with the exception of IL-1β and its antagonist IL-1 receptor antagonist (IL-1Ra), which were up-regulated 65- and 99-fold, respectively. In contrast to S100A8 and S100A9, expression of IL-1β was already markedly reduced on day 21. (IL-1β increased 7–9-fold and S100A8 and S100A9 increased 30–35-fold and 66–69-fold, respectively, compared to normal synovium) (Figure 4A).
We next investigated the effect of S100A8/A9 in the development of synovial lining thickness in S100A9–/– mice. S100A9–/– mice have a normal phenotype, whereas S100A8–/– mice are not viable (34). S100A9–/– mice also lack S100A8 protein in their myeloid cells. S100A9–/– is thus functionally a double knockout, probably due to a high turnover of S100A8 in the absence of its binding partner S100A9 (17). When collagenase-induced OA was elicited in the knee joints of S100A9–/– mice, significantly less thickening of the synovial lining was observed as compared to that in wild-type mice. Synovial thickness was 62% lower on day 42 as compared to OA induced in wild-type controls (Figure 4B and compare Figure 4D with 4C). Synovial lining cells express enzymes, such as ADAMTS, which cleave proteoglycans, thereby leaving NITEGE neoepitopes. On day 42 after OA induction, synovial intima lining cells stained strongly positive for NITEGE neoepitopes (Figure 4E), whereas no staining was observed in the intimal lining of S100A9–/– mice (Figure 4F). In contrast, in the destabilized medial meniscus model of OA, in which only low levels of staining of S100A9 in the synovium were found, only minimal synovial thickening was observed, which was not different between S100A9–/– mice and wild-type mice (data not shown).
Subsequently, we explored by histologic analysis whether S100A8/A9 mediates severe cartilage destruction during collagenase-induced OA. Cartilage destruction was measured using an arbitrary scale developed by Pritzker et al (28), which we adapted for use in mice. Cartilage destruction was measured in various cartilage surfaces (medial and lateral femur and tibia) of the knee joint. Cartilage destruction in all surfaces was significantly lower in S100A9–/– mice (Figure 5A) and ranged from a 45% reduction in the lateral tibia to a 73% reduction in the medial femur, as compared to OA in wild-type mice (Figure 5B and compare Figure 5C with Figure 5B).
When destabilized medial meniscus OA was induced in the knee joints of wild-type mice, most of the cartilage matrix destruction on day 56 was observed in the medial tibia and femur, whereas less destruction was observed on the lateral side (Figure 5D and compare Figures 5F and 5E). Interestingly, no significant differences were measured between S100A9–/– mice and wild-type mice, suggesting that S100A8/A9 does not mediate cartilage destruction in this model.
In synovial biopsy samples from patients with early OA (n = 19), high levels of mRNA for S100A8 and S100A9 were noted as compared to normal synovium obtained from patients in whom acute joint trauma was suspected. S100A8 and S100A9 levels were increased 11-fold and 10-fold, respectively (Figure 6A). Pathologic changes in the knee joints of the CHECK patients was scored using the Kellgren/Lawrence scale. Interestingly, levels of S100A8 mRNA were significantly enhanced (by 4.5-fold) in patients with a Kellgren/Lawrence score >0 at the time of biopsy (Figure 6B) as compared to those with a Kellgren/Lawrence score of 0 (P = 0.046).
Expression of S100A8 and S100A9 proteins was determined using immunohistochemistry. S100A8 was abundantly expressed in biopsy samples from patients with early OA (n = 19) (Figure 6C) as well as in synovium of patients with late-stage OA (n = 10) (Figure 6D). Control irrelevant IgG was negative (data not shown). Similar expression was observed for S100A9 (data not shown), suggesting that these proteins are produced during prolonged periods throughout the OA process. High levels of S100A8 were particularly found within synovia containing a thickened intima. Positive staining was mainly found within CD68-positive macrophages in the lining layer, subintimal lining layer, and surrounding the blood vessels (Figure 6D). The number of S100A8- and S100A9-positive cells varied between patients: 45–60% of total intima cells in synovial samples from those with early OA and 50–60% of total intima cells in samples from those with late OA. The intima lining thickness and subintima cellularity were determined and were found to correlate with the expression of S100A8 and S100A9 proteins. A significant positive correlation was found for S100A8 expression and the intima lining thickness and the subintima cellularity (Figure 6E). Similar correlations were found for S100A9.
To examine whether S100A8/A9 levels are associated with the progression of joint damage, baseline levels of S100A8/A9 in sera from patients with early symptomatic OA (from the CHECK cohort) were measured by ELISA. Joint destruction was determined at baseline and 2 years after the start of the study, using the Kellgren/Lawrence radiographic scoring system. Two patient groups were defined: nonprogressors and progressors. No significant differences in the demographic data, such as age or body mass index, were identified between the two groups. Strikingly, significantly higher serum levels of S100A8/A9 were found in those with OA progression as compared to those without progression (Figure 6F). The mean serum level of S100A8/A9 was 487 ng/ml in nonprogressors (n = 95) and 581 ng/ml in progressors (n = 82), indicating the potential value of assessing serum levels of S100A8/A9 in predicting OA progression.
In the present study, we found that S100A8 and S100A9 are pivotal proteins involved in mediating cartilage destruction during experimental collagenase-induced OA, in which synovial activation is important for the induction of joint pathology, but not in the destabilized medial meniscus model of OA, in which synovial activation is absent. Arthroscopic studies performed in the joints of OA patients suggest that localized proliferation and inflammatory changes in the synovium occur in ∼50% of OA patients (7, 8). Macrophages are the dominant cell types present in activated synovium of humans with OA, although B cells and T cells have also been previously described (11). As in human OA, macrophages were the predominant cell type in the inflamed synovium of mice with collagenase-induced OA, and no polymorphonuclear leukocytes were observed during the chronic phase of disease (between day 7 and day 42).
S100A8/S100A9 appeared to be important in sustaining synovial activation during collagenase-induced OA, since S100A9–/– mice, whose peripheral myeloid cells also lack the S100A8 protein (24), showed reduced synovial activation and cartilage destruction. Both the mRNA and protein levels of S100A8 and S100A9 remained high for extended periods (up to 21 days after injection) in the synovium of mice with collagenase-induced OA. S100A9 mRNA expression even increased between day 7 and day 21. This finding is consistent with the data from the human studies showing that biopsy samples from patients with early symptomatic OA as well as from patients with end-stage OA expressed high levels of S100A8 and S100A9 protein, suggesting a prolonged expression throughout the OA process. S100A8 and S100A9 have been shown to induce a specific inflammatory response in human microvascular endothelial cells, macrophages, osteoblasts, and chondroblasts (35). Synovial activated macrophages are the source of S100A8/S100A9 in collagenase-induced OA.
Apart from regulating synovial inflammation, macrophages have also been shown to mediate cartilage destruction. In previous studies, we found that removal of synovial macrophages prior to induction of experimental OA almost completely blocked metalloproteinase-induced cartilage destruction and osteophyte formation (12, 36). Synovial inflammation has been considered a factor that most likely contributes to dysregulation of chondrocyte function, favoring an imbalance between the catabolic and anabolic activities of the chondrocyte in remodeling the cartilage extracellular matrix (37). Earlier studies showed that intraarticular injection of S100A8 into murine knee joints strongly stimulate both cytokine and matrix metalloproteinase (MMP)/ADAMTS expression in the intima layer (23). Fibroblast-like type B cells lying within the intimal layer produce proteoglycans and at the same time are high producers of ADAMTS (38), which may explain the presence of large amounts of NITEGE neoepitopes within the intimal layer of mice with collagenase-induced OA. In contrast, stimulation of macrophages with S100A8 up-regulated only cytokines, but not MMP/ADAMTS, suggesting that the contribution of macrophages to cartilage destruction may be through the abundant release of catabolic cytokines and S100 species, thereby further stimulating fibroblast-like type B cells to produce ADAMTS.
IL-1β, IL-6, and TNFα are all capable of inducing MMPs, aggrecanases, and other catabolic factors (39). TNFα and IL-1β colocalized with MMPs 1, 3, 9, and 13 in regions of matrix depletion in OA cartilage (40). There is a strong relationship between the increased levels of catabolic enzymes and inflammatory mediators such as IL-1β. IL-1β was previously shown to be crucial in mediating cartilage destruction in experimental murine arthritis (41). The role of IL-1β in driving the pathologic changes of OA is, however, still a matter of debate (14). We found a 2-fold down-regulation of IL-1β in the synovium of the knee joint after a short initial phase of up-regulation. IL-1β and its effects are probably inhibited by cytokines such as IL-10 and IL-1Ra, which were highly expressed shortly after OA induction. IL-1β drives cartilage destruction by stimulating MMP production in chondrocytes. However, MMPs are secreted in a latent form and need an activation step in order to become activated. S100A8 protein may be involved in this by inducing MMP-activating factors, such as oxygen radicals (23).
Interestingly, when proinflammatory triggers such as SCW fragments were injected into a normal mouse knee joint, only a brief enhancement of S100A8 and S100A9 expression in the synovium was noted, indicating that the expression of both proteins is tightly controlled during acute inflammation. Expression of S100A8 and S100A9 in the synovium for prolonged periods during collagenase-induced OA implies that the expression becomes reactivated, which may be driven by cartilage degradation products released during OA. Interestingly, whereas S100A8 and S100A9 levels remained high up until day 21, expression of such cytokines as IL-1β,TNFα, and IL-6 were already strongly decreased at this time point.
Although the pathogenic triggers responsible for prolonged and specific up-regulation of S100A8/S100A9 in OA are not known, recent data point to a proinflammatory crosstalk between extracellular matrix metabolites and phagocyte-specific danger molecules. Elevated catabolic breakdown of the matrix by metalloproteinases leads to additional release of matrix products, such as biglycan, decorin, and aspirin (42). Biglycan has recently been shown to activate the inflammasome within macrophages (43) and may reactivate macrophages, forming a positive feedback loop with S100A8/S100A9. Interestingly, both molecules stimulate macrophages via TLR-4 (44).
Apart from having an effect on synovial inflammation, S100 proteins may have a direct effect on chondrocyte metabolism (23). Although homodimers and heterodimers of S100A8 and S100A9 are negatively charged, they may penetrate cartilage layers due to their small size (20 kd) (45). In a recent study of human chondrocytes isolated from OA cartilage that is reported elsewhere in this issue of Arthritis & Rheumatism (46), we found that not only S100A8, but also S100A9 was a potent stimulator of a catabolic phenotype, which is reflected by high levels of MMP-1, MMP-3, and MMP-9 release and inhibition of matrix molecules, type II collagen, and aggrecan. MMP-3 is a crucial enzyme involved in cartilage destruction during experimental arthritis (47), as well as during collagenase-induced OA (13).
The mechanism by which S100A8 activates chondrocytes is not known. Various receptors have been suggested to be involved in S100A8 signaling, such as TLR-4 (17, 48), RAGE (49), and N-glycans (50). In previous studies, it was shown that S100A8 signaling in macrophages occurs via TLR-4 (17). Using primary chondrocytes obtained at arthroplasty from patients with OA, we recently found that stimulation by S100A8 and S100A9 was significantly inhibited by the TLR-4 inhibitor TAK-242, but not by inhibitors of RAGE or N-glycans (46).
In synovial fluid samples from OA patients, abundant amounts of S100A8/S100A9 (up to 5–7 μg/ml) have been detected (51). Significantly elevated concentrations of S100, probably overflow from the joints, were also measured in the sera of patients in whom OA was suspected, as compared to sera from healthy controls. Interestingly, significantly higher levels were measured at baseline in OA patients who had developed clear cartilage destruction 2 years later. When a cutoff value of 600 ng/ml was used, above which the S100A8/A9 level was defined as increased, and only patients with severe OA progression (change in Kellgren/Lawrence score ≥3) were included in the analysis, a significantly higher number of patients with severe progression showed increased levels of S100A8/A9 (by chi-square test). The odds ratio was 7.5, indicating a highly increased risk of severe progression when S100A8/A9 levels are increased (results not shown). However, the numbers of strong progressors were very low (n = 8), and further research at later followup points are planned in order to determine the value of these proteins as markers of progression.
Synovial activation, which is clearly present in a subpopulation of OA patients, may explain the high levels of S100A8 and S100A9 found in synovial fluid, as well as in the blood, where they can easily be measured. Since S100A8 and S100A9 levels remain high for prolonged periods and since they are important stimulators of cartilage destruction, these proteins may be effective biomarkers for predicting progressive cartilage destruction in OA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. van Lent 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 conception and design. Van Lent, Blom, Schelbergen, Cats, Vogl, Roth, van den Berg.
Acquisition of data. Van Lent, Blom, Schelbergen, Slöetjes, Lafeber, Lems, Cats.
Analysis and interpretation of data. Van Lent, Blom, Schelbergen, Slöetjes, Cats, Vogl, Roth.
The authors would like to thank Laura van Hulst (Department of Rheumatology, Radboud University Nijmegen Medical Centre) for statistical advice.