The multiligand receptor for advanced glycation end products (RAGE) mediates certain chronic vascular and neurologic degenerative diseases accompanied by low-grade inflammation. RAGE ligands include S100/calgranulins, a class of low-molecular-mass, calcium-binding polypeptides, several of which are chondrocyte expressed. Here, we tested the hypothesis that S100A11 and RAGE signaling modulate osteoarthritis (OA) pathogenesis by regulating a shift in chondrocyte differentiation to hypertrophy. We analyzed human cartilages and cultured human articular chondrocytes, and used recombinant human S100A11, soluble RAGE, and previously characterized RAGE-specific blocking Abs. Normal human knee cartilages demonstrated constitutive RAGE and S100A11 expression, and RAGE and S100A11 expression were up-regulated in OA cartilages studied by immunohistochemistry. CXCL8 and TNF-α induced S100A11 expression and release in cultured chondrocytes. Moreover, S100A11 induced cell size increase and expression of type X collagen consistent with chondrocyte hypertrophy in vitro. CXCL8-induced, IL-8-induced, and TNF-α-induced but not retinoic acid-induced chondrocyte hypertrophy were suppressed by treatment with soluble RAGE or RAGE-specific blocking Abs. Last, via transfection of dominant-negative RAGE and dominant-negative MAPK kinase 3, we demonstrated that S100A11-induced chondrocyte type X collagen expression was dependent on RAGE-mediated p38 MAPK pathway activation. We conclude that up-regulated chondrocyte expression of the RAGE ligand S100A11 in OA cartilage, and RAGE signaling through the p38 MAPK pathway, promote inflammation-associated chondrocyte hypertrophy. RAGE signaling thereby has the potential to contribute to the progression of OA.
Low-grade chronic inflammation, mediated partly by the effects on chondrocytes of cartilage-expressed and synovium-expressed cytokines including IL-1β, TNF-α, (IL-8)/CXCL8, and (growth-related oncogene α)/CXCL1 appear to contribute to the progression of osteoarthritis (OA)3 (1, 2, 3, 4, 5, 6, 7, 8). Recently, we linked the induction of altered chondrocyte differentiation to chemokine-induced inflammation, by demonstrating that CXCL8 and CXCL1, both of which are up-regulated in OA cartilage (6), stimulated chondrocyte hypertrophic differentiation in vitro (8). Chondrocyte hypertrophy can contribute to the progression of OA via effects including dysregulation of matrix repair through reduced expression of collagen II and aggrecan, increased expression of type collagen X, up-regulation of matrix metalloproteinase 13 (MMP-13), and promotion of pathologic calcification (8, 9, 10). OA cartilages typically develop foci of maturation of cells to hypertrophic differentiation (10).
A growing body of evidence has implicated the receptor for advanced glycation end products (RAGE) in certain chronic arterial, renal, and neurologic degenerative conditions associated with low-grade tissue inflammation (11, 12, 13, 14). RAGE is a broadly expressed 45-kDa transmembrane protein bearing a 43-aa cytosolic tail, and three extracellular domains (i.e., the ligand-binding V (V′) domain, and two C domains) that confer RAGE membership in the Ig superfamily (11, 12, 13, 14). RAGE is a cognate receptor for four distinct classes of ligands (11, 12, 13, 14), including S100/calgranulins, a family of >20 low-molecular-mass (∼10–14 kDa), acidic proteins that form homodimers, heterodimers, and larger multimers, and bind calcium via two internal EF-hand motifs in each monomer (14, 15, 16). Certain individual S100/calgranulins (S100A1, S100A2, S100A4, S100B) have been reported to be expressed by chondrocytes (17, 18, 19), and intense S100 immunoreactivity is a marker of chondrogenic differentiation (20). In addition, concentrations of certain calgranulins in diseased joint fluids reach nanomolar concentrations (21, 22).
Recently, RAGE expression was demonstrated in human articular cartilage and observed to be more robust in aging and OA cartilages (18). Furthermore, IL-1 and fibronectin fragments induced chondrocyte RAGE expression and micromolar S100B induced MMP-13 in chondrocytes mediated by RAGE in vitro (18). In this study, we examined the potential role in articular chondrocyte hypertrophy of RAGE, with particular focus on RAGE interaction with S100A11 (S100C, calgizzarin) (15, 16). Our results implicate up-regulated S100A11 expression, and RAGE-dependent and p38 MAPK-dependent signaling, in inflammatory regulation of chondrocyte hypertrophic differentiation in OA cartilage.
Materials and Methods
All-trans retinoic acid (ATRA), and human recombinant CXCL8 and TNF-α were from R&D Systems, and rabbit polyclonal Abs to type X collagen were from Calbiochem. mAb to RAGE was from Chemicon International. Soluble human RAGE (sRAGE) and rabbit polyclonal RAGE-specific functional blocking Abs were as previously characterized (23). Human RAGE and dominant-negative (DN) RAGE cDNA in the vector pCDNA4-V5His (Invitrogen Life Technologies) were as previously described (23, 24). DN-MAPK kinase 3 (MKK3) (25) was generously provided by Dr. J. Han (The Scripps Research Institute, La Jolla, CA). Unless otherwise indicated, all other reagents were obtained from Sigma-Aldrich.
Immunohistochemistry for RAGE and S100A11
Mouse polyclonal Abs recognizing the human S100A11-specific peptide CHDSFLKAVPSQKRT, which is expressed by S100A11 but not by other S100/calgranulins, were generated by Zymed Laboratories. All research involving human subjects was performed under institutionally approved protocols. As previously described (26), defined specimens of normal and OA human articular cartilage (age ranges, 22–32 and 66–83, respectively) were taken as full-thickness blocks. For immunohistologic analyses of RAGE and S100A11, frozen sections (5 μm) were fixed using 4% paraformaldehyde for 20 min. After washing with PBS, the sections were incubated with 1% Triton X-100 and microwaved for 1 min, blocked with 10% goat serum for 20 min, and incubated for 24 h at 4°C with mouse anti-RAGE or anti-S100A11 Abs. The primary Ab was detected via the avidin-biotin conjugate method applied according to manufacturer’s instructions using the Sigma Rabbit ExtrAvidin Peroxidase Staining kit. Peroxidase activity was detected using the Sigma Fast DAB staining kit, according to manufacturer’s instructions.
Chondrocyte isolation, culture, transfection, stimulation and agonists used, and SDS-PAGE/Western blotting
Chondrocytes were obtained from the medial and lateral condyles, the patellar groove and the tibial plateau of normal human knee articular cartilages (donor age range, 22–62), by previously described methods (26). For the studies described, chondrocytes were carried in monolayer culture in DMEM/high glucose supplemented with 10% FCS, 1% l-glutamine, 100 U/ml penicillin, and 50 μg/ml streptomycin at 37°C with 5% CO2. For experiments in which first passage was stimulated with agonist serum supplementation in the medium was changed to 1% from 10%, and unless otherwise indicated, cells were treated with 10 ng/ml CXCL8 and TNF-α, 10 nM ATRA, and 100 ng/ml S100A11.
In transfection studies, aliquots of 1.0 × 106 human chondrocytes in 100-mm culture dishes were grown for 18 h in DMEM/high glucose supplemented with 10% FCS. Then, transfections were performed using an electroporator from Amaxa, according to the manufacturer’s protocol for chondrocytes. Transfection efficiency, assessed as a control in each experiment via β-galactosidase transfection and staining, was consistently between 70 and 80% in human chondrocytes. Forced expression of RAGE and DN-RAGE were further verified by SDS-PAGE/Western blotting analysis (26), using the aforementioned murine monoclonal anti-RAGE Ab.
Unless otherwise indicated, Western blotting in this study analyzed aliquots of 30-μg protein obtained from whole-cell lysates, or precipitated from conditioned medium using 15% TCA, followed by separation via 10% SDS-PAGE. The Western blots used HRP-conjugated secondary Abs, as described (26). Immunoreactive products were detected using ECL. Primary and secondary Ab dilutions were 1/2000. Abs to type X collagen were from Calbiochem, to p38 and phosphorylated p38 were from Cell Signaling, and to α-tubulin were from Sigma. HRP-conjugated goat anti-rabbit IgG and anti-mouse IgG were obtained from Santa Cruz Biotechnology.
Flow cytometric analyses
Human articular chondrocytes (1 × 106) were permeabilized using the BD Cytofix/Cytoperm kit (BD Pharmingen). Where indicated, the cells were incubated for 60 min at 4°C with murine monoclonal anti-RAGE or murine IgG isotype control. Washed cells were incubated with FITC-conjugated goat F(ab′)2 anti-mouse IgG for 60 min at 4°C. Fluorescence was detected using a FACSCalibur apparatus (BD Biosciences) with data analyzed using CellQuest software (Purdue University, West Lafayette, IN).
Preparation of recombinant soluble S100A11 and “dot-blot” immunoblotting studies
Human S100A11 cDNA was purchased from American Type Culture Collection and was subcloned into pcDNA4/HisMax (Invitrogen Life Technologies). Aliquots of 6 × 107 HEK 293 cells were transfected with N-terminal His-tagged S100A11 cDNA using Superfect (Qiagen), according to manufacturer’s protocol. S100A11 protein was isolated using the ProBond Purification System (Invitrogen Life Technologies) under native conditions. S100A11 protein was then dialyzed using the Slide-A-Lyzer 3,500 MWCO dialysis cassette (Pierce Biotechnology) in PBS overnight at 4°C. SDS-PAGE/Western blotting was performed to confirm purity. LPS was undetectable (<0.025 endotoxin U/ml) in rS100A11 by the Limulus amebocyte lysate assay (BioWhittaker). For “dot-blot” immunoblotting, 250 ng of S100A11, human recombinant S100B, and S100A12 were directly spotted onto nitrocellulose and allowed to dry for 1 h. The blots were blocked in 5% milk for 2 h, then probed with the anti-peptide Ab to S100A11 for 18 h, and with secondary Ab as described above for Western blots. Where indicated, we mixed the Abs to S100A11 with a 5-fold molar excess of the immunogenic S100A11 polypeptide and incubated for 18 h at 4°C before applying this mixture to the blot.
RAGE and S100A11 expression in OA cartilage in situ
S100/calgranulin sequences contain several highly conserved elements (16). Hence, for these studies, we generated murine polyclonal Abs to the C-terminal 15-aa peptide of the calgranulin RAGE ligand S100A11, a region C-terminal to the canonical EF-hand domain (Fig. 1,A) and that was determined to have a unique sequence by BLAST analysis, including comparison aligned C termini of other S100/calgranulins (Table I). “Dot-blot” immunoblotting confirmed specificity of the Ab for S100A11 (Fig. 1 B).
Alignment of C-terminal amino acid sequences of the indicated S100/calgranulins by BLAST analysis is shown. The C-terminal 15-aa peptide sequence of S100A11 (in bold italics) was verified to be unique not only to calgranulins but also other proteins by BLAST analysis, and this peptide was the immunogen in generating murine polyclonal S100A11-specific Abs as described in Materials and Methods.
Immunohistochemical analyses revealed low-level constitutive expression of RAGE and S100A11 in normal human knee articular cartilages (Fig. 2). We confirmed (18) that chondrocyte RAGE expression appeared up-regulated in OA knee cartilages, particularly so in the deep zone but also as detected in chondrocytes in the superficial zone (Fig. 2). S100A11 expression was markedly increased in all zones of OA cartilages (Fig. 2). S100A11 became particularly abundant in the pericellular matrix of OA chondrocytes (Fig. 2), suggesting that chondrocytes actively secreted S100A11 in the course of OA.
Induction of S100A11 in cultured human articular chondrocytes
Cultured normal human knee articular chondrocytes were stimulated for 24 h with the known inducers of chondrocyte hypertrophy CXCL8 and ATRA (8, 9, 26). We also stimulated chondrocytes with TNF-α, a modulator of OA pathogenesis (1, 2, 7). As assessed by flow cytometry, RAGE was expressed by 59 ± 9% of normal articular chondrocytes (n = 6), with no detectable, significant effects on RAGE expression of CXCL8, ATRA, or TNF-α (Fig. 3,A). Concurrent analysis of the normal articular chondrocytes by SDS-PAGE/Western blotting revealed that S100A11, which remained below limits of detection in both unstimulated and stimulated cell lysates (not shown), was both robustly induced and released into the conditioned medium by chondrocytes in response to CXCL8, ATRA, and TNF-α (Fig. 3,B). S100A11, like certain other calgranulins, is known to form covalently bonded dimers and multimers and this process optimizes binding to RAGE (16, 27, 28). Whereas S100A11 isolated from 293 cells was detectable only as a monomer of ∼11 kDa, S100A11 in conditioned medium of chondrocytes treated with exogenous S100A11, or with CXCL8, ATRA, and TNF-α, appeared partially multimeric, with particularly large multimers detected in CXCL8-treated cells (Fig. 3 B).
S100A11-induced chondrocyte hypertrophy mediated by RAGE
Because the release of S100A11 was inducible in normal chondrocytes, we next tested for the potential of S100A11 to regulate chondrocyte differentiation. We observed that ∼10 nM S100A11 (100 ng/ml) induced type X collagen expression in human articular chondrocytes after 5 days in culture (Fig. 4,A). Dose-response studies indicated detectable induction of type X collagen by 100 ng/ml S100A11 but not 0.1–10 ng/ml S100A11 (data not shown). Chondrocytes coincubated with sRAGE, which lacks the transmembrane and cytosolic tail domains of RAGE (23), demonstrated attenuated type X collagen expression in response to 100 ng/ml S100A11 (Fig. 4,A). Because sRAGE has the potential to affect the binding of RAGE ligands to receptors other than RAGE, we additionally studied the effects of chondrocyte treatment with 20 μg/ml previously described rabbit polyclonal blocking Abs specific for RAGE (23). We used conditions under which 20 μg/ml nonimmune rabbit IgG was confirmed to not suppress the induction of type X collagen (data not shown). Under these conditions, the RAGE-specific blocking Abs were observed to suppress S100A11-induced chondrocyte type X collagen expression (Fig. 4 A).
Next, we validated induction of chondrocyte hypertrophy by examining the chondrocytes for S100A11-induced size increase. Within 24 h, >20% of chondrocytes increased detectably beyond the resting size range in response to S100A11, as assessed by increase in forward scatter via flow cytometry analysis (Fig. 4,B). These changes in chondrocyte size in response to S100A11 were also inhibited by both sRAGE and RAGE-specific blocking Abs (Fig. 4 B).
RAGE mediation of inflammatory cytokine-induced but not ATRA-induced chondrocyte hypertrophy
We confirmed (8, 9, 26) the induction of cultured articular chondrocyte hypertrophy by CXCL8 and ATRA, and we concurrently observed that type X collagen expression induction by CXCL8 but not ATRA was attenuated by coincubation with sRAGE or with RAGE-specific blocking Abs (Fig. 5, A and C, respectively). Concomitantly, we discovered that TNF-α induced chondrocyte type X collagen expression and that TNF-α did so in a manner attenuated by sRAGE or RAGE-specific blocking Abs (Fig. 5 B).
S100A11 signaling through RAGE to induce p38 MAPK kinase pathway activation and type X collagen
We previously demonstrated that p38 MAPK pathway activation was essential for transduction of chondrocyte hypertrophy in response to CXCL8 and CXCL1 (8). Hence, we concluded the study by assessing the specific role of RAGE signaling via the p38 MAPK pathway in S100A11-induced type X collagen expression. We observed that S100A11 stimulated rapid p38 phosphorylation in human chondrocytes (Fig. 6,A). S100A11-induced p38 activation was inhibited by transfection of chondrocytes with a DN construct of MKK3 (DN-MKK3), a specific inhibitor of the highly selective p38 pathway activation inducer MKK3 (25) (Fig. 6,B). DN-MKK3 also inhibited type X collagen expression induction by S100A11 (Fig. 6,B). Because RAGE is not the sole plasma membrane receptor for S100/calgranulins (16, 29, 30), we assessed the effects of specific inhibition of RAGE signal transduction using DN RAGE (DN-RAGE). Transient transfection of DN-RAGE attenuated both S100A11-induced p38 phosphorylation and type X collagen expression in human articular chondrocytes (Fig. 6 C).
In this study, we demonstrated marked up-regulation of expression of the RAGE ligand S100A11 within human OA cartilage. S100/calgranulins exert a variety of physiologic effects on cell function, mediated partly by calcium binding and calmodulin-like activities, complex formation with selected annexins and other proteins, and intracellular translocation (15, 16, 19, 22). For example, S100A11 has been implicated in regulatory effects of TGFβ on cell growth, via induction by TGFβ of protein kinase C-mediated S100A11 phosphorylation and nuclear S100A11 translocation (31). But, increasingly, certain S100 proteins have been observed to be secreted and to exert extracellular effects, illustrated by proinflammatory effects on leukocyte recruitment by S100A8, and S100A9 (12, 32), and by proinflammatory RAGE-dependent effects of S100A12 (extracellular newly identified RAGE binding protein) on endothelial cell and macrophage cytokine release and adhesion (14). In this study, we demonstrated that CXCL8, TNF-α, and ATRA induced chondrocytes to release S100A11. Moreover, we observed that exogenous nanomolar rS100A11 was sufficient to promote chondrocyte hypertrophy, as evidenced by both type X collagen expression and cell size increase.
In this study, we observed that RAGE was constitutively expressed by normal articular chondrocytes in situ, with the qualitative appearance of up-regulated chondrocyte expression of RAGE by chondrocytes in OA cartilage, confirming recently published observations (18). S100/calgranulins bind not only RAGE but also CD36 (30). S100/calgranulin-mediated cellular effects via binding to cell surface proteoglycans also have been demonstrated (30). However, using an established approach based on parallel results of treatment with sRAGE or with function-blocking Abs to block RAGE ligand-induced signaling (23), we demonstrated that both S100A11-induced chondrocyte size increase and expression of type X collagen were RAGE mediated.
The prior observation that the chemokines CXCL8 and CXCL1 induce cultured articular chondrocyte hypertrophy (8) suggested a novel mechanistic linkage between low-grade joint inflammation and dysregulated chondrocyte differentiation within OA cartilage (1, 2, 3, 4, 5, 6, 7). An unexpected finding in this study was that CXCL8- and TNF-α-induced type X collagen expression also were inhibited via sRAGE and RAGE-specific blocking Abs. This result suggested that S100A11, and likely other RAGE ligands induced by CXCL8 and TNF-α, were critical downstream mediators of chondrocyte hypertrophy in response to IL-8 and TNF-α, as schematized in Fig. 7. Importantly, under conditions in which RAGE mediated CXCL8- and TNF-α-induced chondrocyte hypertrophy and in which ATRA induced S100A11, we observed no suppression of ATRA-induced expression of type X collagen expression. We speculate that chondrocyte hypertrophy in response to the growth and differentiation mediator ATRA is driven by more redundant mechanisms (9) than is the hypertrophy response to the inflammatory cytokines CXCL8 and TNF-α.
Using transfection of exquisitely selective DN inhibitors of RAGE and of p38 MAPK pathway activation, we demonstrated that activation of the p38 MAPK pathway via RAGE-dependent signaling was essential for transduction of chondrocyte hypertrophy in response to S100A11. We previously observed that CXCL8-induced p38 MAPK pathway activation and chondrocyte hypertrophy (8) are critically mediated by stimulation of Pit-1 expression and increased sodium-dependent phosphate cotransporter-mediated Pi uptake (26). Hence, it will be of interest to test potential roles of Pit-1 expression, and Pi uptake in S100A11-induced and RAGE-mediated chondrocyte hypertrophy.
Significantly, RAGE ligands of the advanced glycation end products (AGE) class accumulate in tissues in association with both normal aging and diabetes mellitus (11, 12) and also form by posttranslational modification of collagen and other extracellular matrix proteins in association with cartilage aging and OA (33). AGE recognition by RAGE is a central mediator of diabetic microvascular, renal, and neuropathologic complications (11, 12). Moreover, AGE-induced effects on chondrocytes have been linked to decreased proteoglycans and collagen synthesis in vitro (33). Certain biological responses to the chromatin-binding protein amphoterin/high-mobility group box-1 (HMGB-1) are mediated directly by RAGE (34). HMGB-1 is expressed by articular chondrocytes (35), and secreted HMGB-1 can exert proinflammatory cytokine-like effects and has been implicated as an inflammatory mediator in rheumatoid arthritis (36). We have observed induction of type X collagen by AGEs, HMBG-1, and S100B in cultured chondrocytes (D. L. Cecil and R. Terkeltaub, unpublished observations), and it was recently reported that HMGB-1 and S100B induced MMP-13 expression in primary chondrocytes mediated by RAGE-dependent signaling (18). Therefore, in the model of inflammation-stimulated chondrocyte hypertrophy schematized in Fig. 7, we propose that RAGE ligands including S100A11 synergize in promoting altered chondrocyte differentiation and function when externalized by activated chondrocytes.
Limitations of this study included the use of first-passage as opposed to primary chondrocytes. We did not investigate potential effects on chondrocyte differentiation of intracellular S100A11, an inhibitor of cell growth, and regulator of cell senescence and apoptosis (37, 38). Because p38 pathway activation, but not activation of ERK1/2 or JNK, is central to CXCL8-induced chondrocyte hypertrophy (8), we limited mechanistic studies here to the involvement of p38 pathway activation in chondrocyte hypertrophic differentiation mediated by nanomolar S100A11. The current study did not assess potential effects of S100A11-induced chondrocyte activation of ERK1/ERK2 pathway and NF-κB, which were recently reported to modulate chondrocyte activation by S100B added at micromolar concentration (18). We did not monitor for potential constitutive or stimulated chondrocyte release of endogenous sRAGE (39) or for potential function-regulating effects of chondrocyte-derived glycosaminoglycans binding by the RAGE extracellular domain (40) or possible effects of chondrocyte MMPs on RAGE extracellular domain attachment to the plasma membrane (39). Additionally, S100A11, like several other calgranulins, is a substrate for transglutaminase-dependent covalent modification, with the reactive regions in S100A11 defined to be at function-specifying N- and C-terminal domains (41, 42). In this context, we observed S100A11 multimerization in the conditioned medium of cultured chondrocytes in this study, a process for S100A11 that is catalyzed by transglutaminase 2 (TG2) (27) and that promotes the capacity of certain S100/calgranulins to activate signaling via RAGE (16, 28). TG2 is externalized by activated chondrocytes (43) and plays a major role in mediating both CXCL8-induced and ATRA-induced chondrocyte hypertrophy (8, 9). Hence, it will be of interest to test the direct role of TG2 in S100A11-induced chondrocyte hypertrophy.
We conclude that there is accumulation in OA articular cartilages of the RAGE ligand S100A11. RAGE-dependent signaling in response to not only S100A11 but also IL-8 and TNF-α, promotes chondrocyte hypertrophic differentiation and thereby may factor into contributory effects of low-grade inflammation to OA progression.
We gratefully acknowledge Jacquie Quach (The Scripps Research Institute, La Jolla, CA) for technical help with collection of chondrocyte samples and also gratefully acknowledge the other staff of the Tissue Collection Core of the National Institutes of Health-sponsored Program Project on Cartilage and Aging for assistance with cartilage sample acquisition and handling.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a Merit Review grant from Department of Veterans Affairs (to R.T.) and by support from the National Institutes of Health (including Grants P01AGO7996, HL077360-01, and AR049366-01A2).
Abbreviations used in this paper: OA, osteoarthritis; MMP, matrix metalloproteinase; RAGE, receptor for advanced glycation end products; ATRA, all-trans retinoic acid; sRAGE, soluble RAGE; DN, dominant negative; MKK, MAPK kinase; AGE, advanced glycation end products; HMGB-1, high-mobility group box-1; TG2, transglutaminase 2.